Research

Anatomical Reconstruction of the Aorta from an Echocardiogram

• Presenting Author: Jared D. Rosen (Massachusetts Academy of Math and Science at WPI)
• Primary Investigator: Zhenglun “Alan” Wei, PhD 
• Co-Author: Biao Si
  

Introduction:  

Cardiovascular flow simulation involves using computational and mathematical models to replicate blood flow within the cardiovascular system. While the precision of these simulations is continually being refined, the accuracy of reconstructing three-dimensional models of blood vessels is equally vital. Although significant progress has been made in developing models for the aorta and similar cardiovascular structures, there remains a critical need for accurate, patient-specific models, especially in cases where the patient has an abnormal anatomical shape. This study introduces a method that allows medical professionals to efficiently and accurately create a personalized model of a patient’s aorta and extending arteries using a two-dimensional echocardiogram scan. 

Materials and Methods:  

Our method begins by developing a Python script that utilizes OpenCV, an artificial intelligence image recognition library, to detect the aorta’s contours against the background of the image. These contour images, along with the original echocardiogram scan as a reference, are then used by a medical professional in 3D Slicer, an open-access, multi-purpose software for three-dimensional medical image modeling. The professional uses 3D Slicer to identify a two-dimensional segment that matches the shape of the aorta in the MRI scan.  

With this segment, the medical professional can extract the coordinates and radii values at each point along the aorta’s centerline using the VMTK (Vascular Modeling Toolkit) extension within 3D Slicer. These coordinates and radii values are then exported as a CSV file. This file is subsequently re-imported into our Python script, which generates a three-dimensional representation of the aorta by simulating it as a series of circles with linearly varying radii, forming a sequence of obliquely truncated cones replicating the aorta’s shape. The resulting three-dimensional image is then re-imported into 3D Slicer to produce a virtual model as a stereolithographic (STL) file. 

Results, Conclusions, and Discussions:  

This process was applied to echocardiogram images from two healthy subjects. As the procedure was carried out on these images, the entire process was documented and compiled into a step-by-step tutorial as a presentation. To evaluate the method’s accuracy, the resulting models were compared to a reference model created by a medical expert. The high degree of similarity between the models provided strong evidence that this approach successfully reconstructs an accurate model of a patient’s cardiovascular anatomy from an echocardiogram.  

Although the process was tested on the patients’ aortae, its intended and most likely application is to model multiple vessels in addition to the aorta. This would allow for the creation of a complete model of the blood vessels comprising a patient’s central cardiovascular system, which could then be used in various simulated tests available through the wide range of cardiovascular and anatomical simulation software.  

In developing this process, an innovative code was created that converts any line represented as a list of coordinates with corresponding radii into a cylinder-like structure. This software has potential applications beyond cardiovascular flow simulation, where the anatomical reconstruction of cylindrical-shaped systems, tissues, and organs is valuable. It could also be useful in other fields such as mechanical, material, and aerospace engineering, where precision in modeling is crucial for system design.  

Overall, this process represents an important advancement in personalized treatment and medicine, as technology continues to enhance the ability of medical professionals to understand and model biological processes more accurately. 

Acknowledgements:  

The authors would like to acknowledge OpenCV, 3D Slicer, and the VMTK extension within 3D Slicer as free and open-source software utilized for the purposes of this study. 

Basal CD44 / Hyaluronic Acid and Shear Stress Dependent Endothelial Cell Remodeling 

• Presenting Author: Zoe Vittum
• Primary Investigator: Solomon Mensah, PhD
  

Introduction:  

The endothelial glycocalyx is a transmembrane grass-like structure that encompasses endothelial cells (ECs) serving to transduce extracellular signals, regulate vascular permeability, and inhibit the adhesion of leukocytes. Proteoglycans and glycoproteins serve as the surface receptors for various carbohydrate chains known as glycosaminoglycans (GAGs). The glycocalyx along the lumen of the vessel is referred to as the apical glycocalyx, which is directly exposed to shear due to blood flow while the basal glycocalyx is present at the smooth muscle cell interface. Historically, the apical and basal glycocalyx have been independently investigated due to their differences in their local environment and stimuli. The apical glycocalyx has been at the forefront of research efforts due to its direct role in the mechanotransduction of shear forces due to blood flow. However, shear forces due to flow are transmitted from the apical glycocalyx through the cell cytoskeleton to basal glycocalyx components promoting alternate signaling pathways and influencing cell migration and morphology. Yet, the basal glycocalyx remains largely unexplored. 

CD44 is a transmembrane glycoprotein serving as a surface receptor for hyaluronic acid (HA), a known fluid shear sensor in the apical glycocalyx shown to influence the development of cardiovascular disease. The role of basal CD44 and HA in mechanotransduction and EC remodeling is unclear. Here we aim to examine the relationship between shear stress and CD44 and HA-instigated EC remodeling by visualizing basal CD44 and HA expression and cytoskeletal alignment under normal and pathophysiological conditions. 

Materials and Methods:  

Human lung microvascular endothelial cells (HUVECs) were maintained in Microvascular Endothelial Cell Growth Medium at 37 °C in static culture. Passage five HLMVECs were plated at 100,000 cells/mL on a 24 mm x 60 mm fibronectin-coated (10 ug/mL) glass cover slip. Once confluent the slide was placed in a simple continuous flow loop parallel plate flow chamber. Media flowing through the chamber exposed the HLMVECs to 10 dynes/cm2 of shear stress for 30 minutes. This shear stress is within the normal reported range for microvascular endothelial cells. To determine shear stress rate and time dependence the effects of physiologically low and high shear rates as well as increased exposure time will be investigated. 

Slides were fixed with 4% paraformaldehyde and 2% sucrose for membrane stabilization. A blocking solution containing 10% goat serum and 3% BSA was supplemented with 0.1% saponin for permeabilization. Immunostaining of CD44 and HA was conducted using CD44 antibody (1:250, MA5 44271 Thermo) and hyaluronic acid binding protein (1:100, 50 ug/mL, 385911 Sigma) as primary antibodies and Rb IgG (H+L) Alexa-flour 488 (1:250) and Alexa-flour 488 conjugated anti-biotin as secondary antibodies respectively. The slides were mounted with Flouromount DAPI Mounting Media. Z-stack images were taken with a 0.1 µm slice thickness on a Leica STELLARIS 8 laser scanning confocal microscope. Sum projections of the apical and basal cell and Z-direction orthogonal views were generated to visualize the coverage and thickness of CD44 and HA in the apical and basal cell. 

Results, Conclusions, and Discussions:  

Visualization of CD44 (Fig. 1) and HA (Fig. 2) in the apical (A) and basal (C) HLMVEC after 30 minutes of exposure to physiologically normal fluid shear stress was as shown in Figure 1 and Figure 2 respectively. In Figure 1A and Figure 1C distributed expression of CD44 can be seen throughout the apical and basal cell and on-cell processes. CD44 signal is seen encompassing the nuclei, and cell boundaries can be seen in Figure 1B. Apical and basal HA signal is shown in Figure 2A and Figure 2C respectively. A distinct separation of the apical and basal cell HA signal can be seen in Figure 2B.  

The presence of CD44 and HA in both the apical and basal cell suggests that CD44 in the apical and basal glycocalyx cell, through their connection to the cytoskeleton, form a whole cell force transduction network that allows the cell to respond to complex extracellular forces such as shear forces due to blood flow as other models have suggested. Future experiments will examine how CD44 and HA expression and cytoskeletal alignment change with alterations in shear rates and exposure times. Changes in CD44 and HA expression (cell coverage and florescent intensity) and glycocalyx component thickness (orthogonal view) will be quantified. The individual roles of CD44 and HA will also be examined using enzyme degradation and cell knock-out models. The role of the cytoskeleton in CD44 and HA expression will also be investigated by blocking actin polymerization. The results of this experiment will expose the shear stress-dependent role of CD44 in regulating EC remodeling. 

Future studies will examine how basal CD44 and HA affect integrin clustering and modulate crucial mechanotransduction pathways such as YAP/TAZ and Rho which modulate EC proliferation and cytoskeletal remodeling. The cumulative knowledge from this study will be used to further complete existing models of EC shear force transduction which historically neglect the role of the basal glycocalyx. Finally, the characterization of basal CD44 will provide insight into basal glycocalyx function in vascular physiology and disease allowing for the identification and development of new therapeutic targets. 

Acknowledgements:  

This research was supported with startup funds from Worcester Polytechnic Institute awarded to Dr. Mensah. 

Can DEXA Predict Metatarsal Bone Quality in Runners Affected by the Female Athlete Triad? 

• Presenting Author: Amelia Adin – University of Florida (REU Participant) 
• Primary Investigator: Karen Troy, PhD
  
• Co-Author(s): Julia Nicolescu, Andrew Wilzman  

Introduction:   

Up to 74.7% of young female athletes are at moderate or high risk of the Female Athlete Triad [1]. The Triad is characterized by a low energy availability, menstrual dysfunction, and low bone mineral density (BMD) [2, 3], which leaves affected athletes vulnerable to bone stress injuries (BSIs). Sports that emphasize leanness and endurance, such as running, have a higher prevalence of the Triad [4]. Dual energy X-ray absorptiometry (DEXA) scans are used to confirm a Triad diagnosis by measuring BMD of the total body, at the hip, the lumbar spine, and the distal forearm. Although up to 38% of BSIs in college athletes occur in the metatarsals [5], this area is not assessed by a DEXA scan. Here, we examined relationships between metatarsal bone micro- and macrostructure and DEXA measures in female runners to determine the degree to which DEXA could predict metatarsal bone quality. Female Athlete Triad risk was also determined for each participant. We hypothesized that DEXA measures would be positively associated with measures of metatarsal quality, and that higher Triad risk (low, moderate, or high) would be associated lower with bone quality by both DEXA and in the metatarsals. 

Materials and Methods:   

Thirty-three female runners each with a weekly running volume of at least 10 miles for the last year and aged 17-41 years participated in this institutionally approved study. We collected high resolution peripheral quantitative computed tomography (HRpQCT) scans of the distal 2nd metatarsal (XtremeCT I, Scanco, Switzerland), DEXA scans, and Triad risk assessments. ¬ Low-resolution (voxel size 246 µm) CT scans of whole metatarsals were processed using Mimics (Version 24.0, Materialise, Belgium) software to generate a segmented mask of the left 2nd metatarsal. This mask was processed in MATLAB to extract measures of size, shape, and mineral content [6]. High-resolution (voxel size 82 microns) scans of the 10-20% distal region of the second metatarsal were evaluated with Scanco XtremeCT I software to extract measures of microstructure and density [7]. All measures obtained from low- and high-resolution scans are outlined in Table 1. DEXA scans assessed areal BMD (aBMD) of the total body, femoral neck, lumbar spine, and distal forearm, and reported lean and fat body mass. The Triad risk assessment questionnaire consisted of ten questions compiled from the 2014 Female Athlete Triad Coalition Consensus Statement [2]. Answers to this questionnaire and Z-scores from the DEXA scans were used to assign participants a risk score that classified them as low (0-1), moderate (2-5), or high risk (6-12) [2]. Ten participants were not scored for Triad risk due to incomplete questionnaire or DEXA data. A Pearson’s correlation test was performed between 26 metatarsal bone quality measures and 6 DEXA measures. 

Results, Conclusions, and Discussions:  

We observed a significant positive association (r=0.529, p=0.018) between lowest DEXA-reported Z-scores and metatarsal compressive strength index (CSI; Figure 1A). Z-scores and aBMD at the forearm and hip had weak to moderate correlations with metatarsal CSI, bone mineral content and volumetric BMD (r=0.282–0.429), though lowest Z-score had the highest correlations with these three measures (r=0.429-0.529). Measures of bone size and shape had very low associations with DEXA measures (r < 0.200), though some microstructure measures had weak correlations (r < 0.400) with DEXA aBMD and Z-scores (Figure 2). Of the DEXA measures, lowest Z-score was the best predictor of metatarsal bone health (Figure 2), which supports current Triad diagnosis procedures. It is worth noting that the average lowest Z-score in our group of female runners was -0.734 (SD=0.653), which is significantly (p < 0.001) below the expected average Z-score of 0.0 for healthy, age-matched adults. 

Body fat percentage was inversely related to metatarsal measures, with higher fat percentage being associated with poorer bone quality. For example, greater trabecular spacing had a significant correlation with body fat percentage (r=0.485, p=0.051) (Figure 1B). Although very low body fat is associated with low energy availability in the Female Athlete Triad, our cohort may not contain enough athletes at low body fat to show such associations. Within our sample, low body fat percentage may be associated with general fitness and better bone quality.  

Although DEXA measures were moderately predictive of metatarsal bone health, Triad risk scores showed no clear distributions among the data comparisons (Figure 1). In fact, risk scores seemed to be evenly scattered between high and low DEXA and metatarsal bone quality. However, most of our participants were in the moderate risk category. The Triad Coalition risk assessment accounts for a variety of past and present factors that affect bone metabolism, and not all these factors may influence our measures of metatarsal bone quality. Additionally, this cohort may not be representative of the greater population of female runners since participants were recruited for an ongoing study requiring them to either have a current metatarsal BSI or no history of BSI. 

Cathepsin Proteolytic Cleavage of Sars-cov-2 Spike Protein and Identification of Cleavage Sites 

• Presenting Author: Maya Evohr  
• Primary Investigator: Manu O. Platt, PhD (he/him/his) – NIH/NIBIB 

Introduction:  

Proteases are implicated in the infection process of SARS-CoV-2, the virus causing the COVID-19 pandemic. At the cell surface, furin and TMPRSS2 proteases can cleave the spike protein into S1 and S2 subunits, leading to the virion fusing to and entering the membrane of the target cell. After endocytosis, proteases (including cathepsins B and L) facilitate endolysosomal escape in acidic environments. While spike protein activation is primarily facilitated by these proteases, the identification of other proteases involved in SARS-Cov-2 entry into cells is important for the development of new therapeutic targets. Cathepsins K, S, and V all share 60-80% sequence homology with cathepsin L, implicating them as potentially being able to cleave the SARS-CoV-2 spike protein. In patients with preexisting conditions such as cardiovascular disease, sickle cell anemia, emphysema, and diabetes, all of which are associated with an upregulation in these cathepsins, there is an elevated risk of severe COVID-19 and mortality. This suggests that cathepsins K, S, and V may cleave the SARS-CoV-2 spike protein. Understanding cathepsin cleavage of the SARS-CoV-2 virus may provide new therapeutic targets for the treatment or prevention of SARS-CoV-2 infection. 

Materials and Methods:  

Recombinant proteins were produced at NCI’s Protein Expression Laboratory. SARS-CoV-2 spike proteins were expressed in human HEK293T cells, cathepsin S in Tni-FNL baculovirus, and cathepsin K in P pastoris yeast. Cathepsins and spike proteins were verified by SDS-PAGE and Western Blotting. To determine proteolytic activity of these recombinant cathepsins, DQ Gelatin was used as a fluorogenic substrate, incubated with increasing concentrations of cathepsins S and K, and fluorescence measured every five minutes using a BioTek Synergy plate reader for two hours. Western blotting and peptide mapping using LC/MS were employed to determine cleavage locations. Cathepsins were co-incubated with spike protein and aliquots taken from 0 to 60 minutes, prior to stopping reaction with E64 inhibition and boiling for 5 minutes. Proteins were then loaded for SDS-PAGE in identical gels where aliquots from reaction were either (1) transferred to membranes for Western blotting or (2) Coomassie-stained and bands extracted for trypsin digestion and LC/MS. For Western Blots, a multiplex assay was done with antibodies for the N-Terminal and the S2 domains to identify sizes of cleaved products. LC/MS bands were submitted to mass spectrometry and peptide mapping. 

Results, Conclusions, and Discussions:  

Gelatinase activity of cathepsins K, L, S, and V was measured using a DQ gelatin assay to determine the lowest effective concentration of each cathepsin. All recombinant cathepsins were able to cleave gelatin; in fact, very small micromolar amounts of the cathepsins were needed to work as a potent gelatinase (Figure 1). The ideal cathepsin amount for the DQ gelatin assay with a 125 µL volume was found to be 5 picomoles, which demonstrated that the cathepsin cleavage of gelatin increased over time for all cathepsins.  

Western blotting of co-incubated wild type spike protein with cathepsin K (Figure 2A) and cathepsin S (Figure 2B) demonstrated the locations of cathepsin cleavage of the spike protein, with the S2 domain highlighted in red, and the N-Terminal domain highlighted in green. These results demonstrate that cathepsins S and K cleave the SARS-CoV-2 spike protein. The disappearance of NTD and S2 signals over time suggests that the cathepsins are either cleaving at the antibody binding sites or that the fragments are too small to detect. Cathepsins K and S are both initially cleaving the NTD side of the protein into 2 pieces – an approximately 105 kDa fragment connected to the S2 domain and a 75 kDa fragment connected to the NTD (Figure 3). However, as the cleavages continue, separate cleavages are observed in the cathepsin K and cathepsin S incubations. This shows that cathepsin K and cathepsin S share some cleavage sites but also have some unique cleavage sites.  

Trypsin-digested cathepsins and spike proteins were analyzed using liquid chromatography mass spectrometry in order to define what peptides were present prior to cathepsin cleavage of spike proteins. Peptides were successfully mapped from SDS-PAGE gel band excisions with 65-92% sequence coverage. The peptide sequences present in each cathepsin and spike protein sample will then be compared to samples with cathepsin and spike protein co-incubated in order to determine where the cathepsins are cleaving spike protein.

A Cell Culture Model for Lactation Studies 

• Presenting Author: Diana Alatalo, PhD, CLC
  

Introduction:  

Exclusive breastfeeding for 6 months is the recommended feeding method for an infant with health benefits for both infant and breastfeeding parent. Despite approximately 90% of birthing parents in the U.S. initiating breastfeeding after birth, currently only 25% of infants are exclusively breastfed at 6 months. Challenges with breastfeeding that frequently lead to early weaning include concerns of milk production, but only an estimated 5% of birthing parents have a physiological barrier to producing sufficient milk. Clinically, a diagnosis of insufficient milk production is a diagnosis of exclusion and cannot be made until 1-2 weeks postpartum. Milk production occurs within the cell in response to the polypeptide hormone prolactin and regulated by the signal transducer and activator of transcription 5 (STAT5) protein. To develop an assay that can distinguish between perceived and real insufficient milk production requires a cell culture model that can vary lactation output. 

Materials and Methods:  

We performed a 2D culture of MCF7 mammary cells at 3 different temperatures (37°C, 38°C, and 39°C) with (lactation) and without (control) lactogenic culture media on 35 mm well plates with a glass slide cover on the bottom and seeded at 2 x 104 cells per cm2. The control media was prepared according to ATCC protocol with the addition of 1% (v/v) GlutaMAX supplement and 1% (v/v) Penicillin Streptomycin. The lactation media consisted of the control media with dexamethasone (1 ug/mL) and prolactin (3 ug/mL). Media was changed every 2 days until Day 8 when the glass slides with attached cells were removed, stained for the milk protein β-casein and nuclei, and imaged. Lactation production was quantified with a custom MATLAB code to count the number of pixels with β-casein and nuclei. 

Results, Conclusions, and Discussions:  

All cells cultured in the lactation media showed signs of lactation (Figure 1). The concentration of β-casein in the 37°C and 38°C cultures was significantly greater compared with the 39°C cultures. The ratio of β-casein-stained pixels to nuclei-stained pixels was 106.05%, 6.71%, and 1.281% for 37°C, 38°C, and 39°C cultures, respectively, indicating that more β-casein was released and moved outside the nuclei of the cells towards the cell wall. Prolactin caused hypertrophy of cells, so cell size was measured and evaluated with a one factor ANOVA test. While the lactating cells were statistically larger (p < 0.001) at 65.65 um2 than the control cells at 46.56 um2, no significance was found between the sizes of lactating cells at different cultures. This indicates that prolactin binds to the cells in less-than-ideal conditions. Clinically during pregnancy, an increase in breast size due to lactogenic hormones and cell hypertrophy is usually predictive of lactation success. However, this study shows increased cell size is not a good biomarker of lactation success. Similarly, all the lactating cells had long connections between cells compared to the cobblestone morphology of the control group. In conclusion, we successfully designed a cell culture protocol that can vary lactation output for human mammary cells by varying the temperature of the culture. This protocol can be modified for other mammary cell lines in both humans and other mammals to study insufficient milk production at a cellular level and develop an assay for clinical use. 

Acknowledgements:  

The author would like to acknowledge the assistance of Leithsa Dimanche, Jazmyn Ewing, Caroline Major, and Taina Quiñones for their assistance with this project. 

Characterization of Porous Fibrin Scaffolds with Tunable Anisotropic Features 

• Presenter: Bryanna L. Samolyk
• Primary Investigator: George Pins, PhD
  
• Co-Author: Jeannine Coburn, PhD
  

Introduction:  

Volumetric muscle loss (VML) impacts 65.8 million Americans annually, including a disproportionate number of wounded warriors injured in combat. VML is caused when significant skeletal muscle loss damages the biophysical and biochemical signaling cues that direct functional muscle regeneration. Tissue engineering strategies that use biomimetic scaffolds offer a promising solution for treating VML by providing anisotropic topographic features to guide myotube alignment which is important to produce uniaxial contractile forces. Recently, our lab created anisotropic fibrin scaffolds using directional lyophilization, where anisotropic scaffold architectures are tuned by altering fabrication parameters (Samolyk et al., Tissue Eng Part C, 2024). We showed that scaffolds fabricated with lower freezing temperatures and higher fibrin concentrations result in maximized myoblast orientation. Here, we aim to further characterize the properties of anisotropic scaffolds under physiologic conditions. The aim of this study is to determine how the fabrication parameters impact the hydration, porosity, and degradation properties of the scaffolds to predict how they will perform after implantation. 

Materials and Methods:  

Several fabrication parameters were assessed for the scaffolds, including fibrin concentrations of 10 and 20 mg/mL as well as freezing temperatures of -40, -80, and -195C. To assess hydrated scaffold architecture, scaffolds were soaked in PBS for 15 min, fixed, and paraffin embedded. Scaffold sections (8 μm thick) were cut in a plane parallel to the freezing direction, mounted, stained with eosin, and imaged in brightfield. Feature size analyses were performed by manually measuring the distance between parallel struts within the scaffolds using ImageJ. To characterize the porosity of the scaffolds, ethanol immersion was used on rectangular sections. Initial weight, length, width, and height were measured, then samples were submerged in 100% ethanol for 2 min, followed by a second weight measurement used to calculate percent porosity. To assess changes in scaffold morphology due to hydration, scaffold dimensions were quantified when dry and hydrated. Scaffolds were sectioned into 1 mm thick samples, and then 2 mm diameter biopsy punches were used to create samples. Brightfield images of the planar surfaces of scaffolds were taken before and after hydration in PBS for 1 h, then analyzed with ImageJ to quantify changes in scaffold area. To assess proteolytic degradation, scaffolds were incubated in 0.2 mL of 0.5 U/mL human plasmin in PBS at 37C. Scaffolds were imaged at 0, 3, 6, 9, 24, 48, and 72 hours after the addition of plasmin. Each timepoint was analyzed with ImageJ to quantify the reduction in scaffold surface area. 

Results, Conclusions, and Discussions:  

To quantify the effect of hydration on scaffold morphology, we measured changes in strut width as well as bulk changes in dimensional surface area. Average strut widths for hydrated directional scaffolds ranged from 3.42-21.91 μm depending on the experimental condition. Interestingly, although the strut widths were smaller on hydrated scaffolds compared to dry scaffolds, the same trends were followed. The -40C, 10 mg/mL scaffolds showed significantly increased strut width compared with all other scaffold conditions. When observing the bulk dimensional changes of the material, the average shrinkage ranged from 24.20-43.98%. The -40C, 10mg/mL scaffolds shrank significantly more than both -195C scaffold groups. There was also a general trend, where scaffolds with smaller strut widths displayed less shrinkage than those with larger strut widths. Scaffold porosity was quantified using ethanol immersion. Average porosities ranged from 79.00-94.16%; however, porosity was not significantly different for any of the scaffold groups. Proteolytic degradation of scaffolds was evaluated over 4 d, periodically measuring the decrease in initial area of samples. For all the scaffold groups, there was a sharp decrease in area after the first 3 h in plasmin, followed by a more gradual decrease in area. In contrast, the no plasmin control maintained a relatively flat curve, with no sharp decrease in area. None of the scaffold groups were significantly different from one another at each time point. However, the -80C, 10 mg/mL and the -195C, 10 mg/mL scaffolds did display significant decreases in area compared with no plasmin scaffold controls at 24, 48, and 72 hours. 

Here, we describe the creation of a novel porous fibrin scaffold with anisotropic microarchitectural features. Structural and morphological evaluations of our scaffolds were used to characterize these constructs under physiological conditions. We showed that the scaffolds all exhibited a degree of shrinkage upon hydration, which has been observed in other porous biologic scaffolds. We also demonstrated that porosity and proteolytic degradation are not altered by the fabrication parameters we explored. Future work will focus on myotube formation on these scaffolds. 

Acknowledgements:  

This work was funded in part by AHA Institutional Research Enhancement Award 953260 and NIH 1R15AR080988. 

Competency-Based Outcomes in a BME Laboratory Course: Effects on Student Learning and Satisfaction 

• Presenting Author: Karen Troy, PhD
  

Introduction:  

Traditional laboratory courses in engineering are often structured with data collection and analysis taking place in small groups. Emphasis is on producing lab reports that reinforce the scientific method. However, these reports can be time consuming to produce and grade, and group work allows some students to complete the course without fully understanding the material. Here, we describe outcomes from two sequential offerings of a Skeletal Biomechanics Laboratory course (BME 3503). The first offering was structured as a traditional laboratory class with group reports. In the second offering the course was restructured to be competency based, with individual student submissions. 

Materials and Methods: 

Offering 1 consisted of five lab modules, completed in small groups. Grades were determined based on lab reports and lecture quizzes. Offering 2 consisted of the same five modules with data collection completed in groups. Instead of requiring a lab report, each module had 4-5 competencies associated with it. For example, the center of mass lab had the following: (a) Use free-body-diagrams and statics concepts to calculate the center of mass; (b) Derive formulas needed to use a reaction board and explain their meaning; (c) Use the segment method and explain the concept. Labs were graded either based on the standard format (Offering 1) or only on the degree to which the student’s work met the competencies (Offering 2). Student evaluations and grades were assessed as outcomes. 

Results, Conclusions, and Discussions:  

Results:  

Thirty-seven students were enrolled in Offering 1, and 34 in Offering 2. Of those, 22 and 20 students completed course evaluations, respectively. In Offering 1 everyone earned an A grade, while in Offering 2 there were 20 A, 6 B and the rest were incomplete or no-record. In Offering 1, students rated the overall course a 3.7 (out of 5) and rated the amount they learned as a 4. They rated the assignment feedback helpfulness as a 3.7. In Offering 2 students rated the overall course a 4.5 the amount they learned as a 4.3, and the assignment feedback helpfulness as a 4.5. When specifically asked about the competency-based grading scheme, students reported, “I liked competency-based grading greatly. It encouraged me to learn how to do each aspect of the lab rather than only a portion by completing a lab report. I also feel like at WPI, work gets unevenly distributed, causing someone to not learn as much as they can solely based on the segment they are assigned. I feel like I got the most out of this course that I could based off the competency grading as an individual rather than as a group”. Some students felt it was too much work turning in individual assignments and that the grading scheme sometimes made it difficult to understand what was expected.  

Discussion:  

Competency-based grading can be implemented in many ways. We piloted one way here and the students reviewed it positively. We also felt that it was easier to assess individual student learning with this scheme, and that grading itself was much faster because of clearly defined rubrics that focused on the most important concepts. 

Cyclic Stretch Inhibits Cell Invasion in 3D Scaffolds 

• Presenting Author: Rozanne W. Mungai
• Primary Investigator: Kristen L. Billiar, PhD
  
• Co-Author(s): Grace E. Jolin, Ying Lei
  

Introduction:  

Cell invasion and migration are fundamental cellular behaviors that drive biological events including morphogenesis, wound healing, cancer metastasis, and infiltration of engineered extracellular matrices (ECM). While cells are known to migrate in response to chemical and physical cues, little is known about the role of dynamic stretch on cell invasion into tissue despite the prevalence of dynamic loading in many organs. Studies utilizing 2D substrates have demonstrated reduction of cell migration in response to cyclic stretch. Most cell types reorient away from stretch in these 2D systems. In contrast, in 3D hydrogels, cells embedded within scaffolds generally align with the direction of uniaxial stretch. This cell alignment could promote contact guidance and subsequent increased directional migration, yet cell invasion in 3D scaffolds has not been studied under stretch. The goal of this study is to determine the effect of cyclic stretch on cellular invasion into 3D ECM. We hypothesize that 10% uniaxial cyclic stretch will reduce cell invasion compared to static culture and that the effect will be most pronounced across the direction of stretch. 

Materials and Methods:  

We utilize a common in vitro multicellular spheroid invasion model embedded in a collagen hydrogel and plated into stretchable wells (Figure Panel A). The spheroids are made from valvular interstitial cells, dermal fibroblasts, or smooth muscle cells. The hydrogels are cultured for two days under static or uniaxial cyclic 10% stretch at 1Hz frequency before fixation and staining the cell nuclei with Hoechst dye. Z-stack images are captured through the thickness of the spheroids both at Day 0 and Day 2 and processed into maximum projection images. We couple this model with a custom MATLAB program that segments the boundary of the Day 0 image to identify the pixel locations of the stained cell nuclei outside of that boundary at Day 2 (Figure Panel B). The extent of invasion with reference to the boundary is analyzed using standard metrics (i.e., area change) as well as using our novel integrative metric, the area moment of inertia, which reflects how the nuclear pixels are distributed with respect to the stretch and cross-stretch axes and thus incorporates both the number of nuclear pixels and their distance from the boundary. 

Results, Conclusions, and Discussions:  

On Day 2, we observed reduced cell invasion under stretch compared to the static condition for the three cell types (Figure Panel C, representative smooth muscle cell spheroid image). Quantitatively, both the area change and the area moment of inertia metrics affirm that cyclic stretch reduces the extent of cell invasion for all three cell types (Figure Panel C, data for smooth muscle cells shown). However, the extent of the reduction varied by cell type, with valvular interstitial cells being the most sensitive to stretch followed by the smooth muscle cells. In addition, we found that, for the three cell types, the reduction of cell invasion was not dependent on the direction of stretch as shown by the area moment of inertia (Figure Panel C, data for smooth muscle cells shown). These results indicate that while cyclic stretch reduces cell invasion into 3D scaffolds, the extent of reduction may be cell-type dependent. Since cell proliferation and cell migration are the major mechanisms of cell invasion, in future studies we will explore if and how these mechanisms play a role in reducing cell invasion under cyclic stretch. Additionally, it is unclear why there is not a directional dependence of stretch on cell invasion, so future studies can explore the role of cell and matrix alignment on the directionality of cell invasion under cyclic stretch. This work has the potential to inform how exposure to cyclic stretch affects cell invasion for different tissues in vivo with applications such as wound healing and the repopulation of decellularized tissues. 

Design of a Feeding Pump for Premature Babies in Low to Middle Income Countries 

• Presenting Author: Audrey Tetteh (Celerion)
• Primary Investigator: Solomon Mensah, PhD
• Co-Author: Dirk Albrecht PhD, Joelle Hanley 
  

Introduction:   

Premature births are a significant global health challenge, with approximately 147 million births occurring annually, translating to about 402 thousand daily births worldwide. Among these, around 15 million are premature, and more than 3 million of these preterm infants die within their first month of life. The overwhelming majority of these deaths, over 90%, occur in low- and middle-income countries (LMICs). A major contributor to these high mortality rates is infant malnutrition and inadequate oral and respiratory control, which accounts for about 98% of these deaths. Premature infants in LMICs face a higher risk due to their underdeveloped organs, which hinder their ability to coordinate sucking, breathing, and swallowing mechanisms necessary for effective feeding. Data highlight this disparity, showing significantly higher neonatal preterm mortality rates in LMICs compared to high-income countries. 

In response to this urgent issue, we have designed a low-cost automatic feeding delivery pump specifically tailored for premature infants in LMICs. This device aims to provide a reliable, affordable, and efficient solution to improve the survival and health outcomes of these vulnerable infants by ensuring they receive the necessary nutrition safely and effectively. The pump leverages advanced yet cost-effective technologies to automate the feeding process, reducing the risk of feeding-related complications and promoting healthy growth and development in preterm infants. Our design prioritizes accessibility and ease of use to facilitate widespread adoption in resource-limited settings, ultimately aiming to reduce the high mortality rates associated with preterm births in these regions. 

Materials and Methods:   

The device was designed using a combination of hardware and software components to ensure accurate and reliable feeding for preterm babies. The microcontroller ATMEGA328p was used as the central processing unit, interfaced with various sensors and actuators. The sensors included the DHT22 for temperature and humidity measurements and the ACS712 for current sensing. The power supply unit utilized LM7805 voltage regulators to step down the input voltage and ensure stable operation of the microcontroller and sensors. Decoupling capacitors were used at the input and output of the LM7805 to suppress high-frequency noise and maintain voltage stability. The peristaltic pump, driven by a motor controller, was selected for its ability to provide precise and controlled fluid delivery. The system was designed using EasyEDA software for circuit design and Fritzing software for breadboard connections. The components were soldered onto a perforated board to create a compact and durable assembly. For the casing, 3D printing with HD PLA material was chosen due to its safety, ease of use, and environmental friendliness. The design was validated through a series of tests, including flow rate and voltage measurements, to ensure the system met the required specifications for neonatal feeding. 

Results, Conclusions, and Discussions: 

Results: 

The automated neonatal feeding system demonstrated reliable performance across various tests. The flow rate tests showed the device could consistently deliver breast milk at rates ranging from 5 ml to 15 ml per minute, with a maximum error margin of ±3 ml. Volume delivery tests indicated that the system accurately administered specified volumes between 20 ml and 60 ml. The temperature sensor-maintained milk within the optimal temperature range, ensuring the safety and nutritional integrity of the feed. The alert system effectively notified caregivers of any issues, such as deviations in flow rate or temperature. These results indicate that the system meets the essential requirements for neonatal feeding in low-resource settings. 

Conclusions: 

The development of an automated neonatal feeding system has the potential to significantly improve the care of premature infants in LMICs. By providing accurate and reliable delivery of breast milk, this system can alleviate the burden on healthcare professionals and ensure that preterm infants receive the necessary nutrition for healthy development. The successful implementation and testing of the device suggests it is a viable solution for addressing the high mortality rates associated with premature births in resource-limited environments. Further refinements and large-scale testing are recommended to optimize the system’s functionality and integration into existing healthcare practices. 

Discussion:  

The design and implementation of the automated neonatal feeding system addresses critical challenges faced by healthcare providers in LMICs. The system’s ability to deliver precise volumes and flow rates of breast milk can enhance the nutritional care of preterm infants, potentially reducing mortality rates. Additionally, the integration of temperature control and alert systems ensures the safety and reliability of the feeding process. While the current results are promising, future work should focus on scaling the system for widespread use, improving its user interface, and conducting longitudinal studies to assess its long-term impact on neonatal health outcomes. Collaboration with local healthcare providers will be crucial to adapt the system to specific regional needs and ensure its successful adoption. 

Acknowledgements:  

First and foremost, praise and thanks to God for His showers of blessings throughout our design project for successful completion. 

We would like to express our deep and sincere gratitude to our project supervisor, Prof. E. K. Tiburu for giving us the opportunity to do this design project and providing invaluable guidance throughout this project. We would also like to thank our external supervisors Prof. Dirk Albrecht and Prof. Solomon Mensah for their guidance, motivation, and time. We thank them for their immense support and also believing in us to work on this project together with our teammates at Worcester Polytechnic Institute (WPI). We are extending our heartfelt thanks to WPI, Biomedical Engineering Department. We say a big thank you to lab manager Lisa Wall for her help with our project. 

We express our special gratitude to Mr. Douglas Ayittey for his genuine support throughout this project. We would also like to say thanks to our international teammates, Akansha Deshpande, Joelle Hanley, Isabelle Ting, and Julie Webster for their help and contribution to this design project.  

Determining Suitable Leaf Species for Perfusion of Leaf-Derived Vascular Scaffolds  

• Presenting Author: Andrea Bartus– University of New Hampshire (REU Participant)
• Primary Investigator: George Pins, PhD
• Co-Author: Bryanna L. Samolyk 

Introduction:   

In the United States alone, nearly 500,000 Americans are treated annually for burns with 40,000 requiring hospitalization and advanced treatment. The healthcare costs associated with these burns are between $2-3 billion USD annually with many needing skin grafts or bioengineered skin substitutes. Treating full-thickness burns is vital for preventing infection, water loss, scarring, and death in severe cases. The standard of care is a split-thickness skin autograft, but due to limited tissue availability and donor site morbidity, a bioengineered skin substitute may be necessary. Current bioengineered skin substitutes lack vasculature necessary to enhance graft take. Since oxygen and nutrient diffusion is limited to 200 µm from capillaries, vasculature is necessary to support new skin ingrowth. Decellularized leaf scaffolds possess vasculature mimicking the structural architecture of human skin and they have the potential to function as inexpensive bioengineered skin substitutes, since cellulose is biocompatible and elicits a minimal inflammatory response. Previous work utilized spinach leaves as decellularized scaffolds, however different leaf types may possess characteristics for enhancing endothelial cell perfusion throughout the vasculature of the leaf. As such, the goal of this study is to evaluate several different leaf types for their decellularization ability, their perfusable vascular network morphology, as well as their xylem diameters. 

Materials and Methods:  

Leaves for decellularization analyses were selected based on criteria including dicot vasculature, the presence of a petiole, and the ability to withstand the decellularization process without degrading. Leaves were quantified for vascular density of perfusable vascular structures as well as xylem diameter. Leaves selected for decellularization include baby kale (Brassica oleracea), leatherleaf viburnum (Viburnum rhytidophyllum), sweet basil (Ocimum basilicum), Chicago hardy fig (Ficus carica), sage (Salvia officinalis), and spinach (Spinacia oleracea) as a control. To decellularize the leaves, a series of three rinses alternating between hexanes and deionized water were performed to remove the waxy cuticle of the leaf. Leaves were then placed in 50 mL conical tubes containing 1% w/v sodium dodecyl sulfate solution on a rocker for 5 days, with daily solution changes. Next, the leaves were switched to a 1% Tween 20, 3% bleach solution for 3-21 days with daily solution changes until colorless. Decellularized leaves were rinsed with deionized water before storing at 4 °C. To examine the vasculature of the decellularized leaves, samples were dried for a minimum of 90 minutes on a benchtop before submerging the petiole in red dye. Vessel Analysis (ImageJ plugin) was used to analyze vascular density. To determine average Feret diameters of the xylem in the petioles for each leaf type, a portion of each sample was embedded in paraffin, sectioned, and imaged via phase contrast microscopy followed by image analysis using a custom ImageJ macro to isolate xylem features from the surrounding tissue morphology. 

Results, Conclusions, and Discussions:  

All leaves were durable enough to withstand the decellularization process, but the basil leaves were the most fragile. Following decellularization, the dye perfusion test was performed, and vascular density of each leaf was analyzed. Spinach had a mean vascular density of 26.1 ± 2.4%; sweet basil, 20.3 ± 3.9%; Chicago hardy fig, 25.3 ± 2.8%; baby kale, 20.7 ± 3.5%; and sage, 16.0 ± 0.6% with n = 2 samples per leaf. Vascular density could not be analyzed for leatherleaf viburnum as the dye perfused uniformly as a front through the leaf instead of flowing only through the vasculature with sage exhibiting a similar perfusion pattern where vasculature was only visible in the tip. The spinach leaves had the greatest mean vascular density followed by Chicago hardy fig. Analysis of the petiole cross-sections revealed spinach had mean xylem Feret diameter of 13.0 ± 2.4 µm; sweet basil, 25.6 ± 0.7 µm; Chicago hardy fig, 20.5 ± 2.32 µm; baby kale, 17.4 ± 1.3 µm; leatherleaf viburnum 16.4 µm; and sage, 17.7 µm with n = 1- 3 samples per leaf. Of the leaves analyzed, basil had the largest mean xylem Feret diameter, followed by the fig leaves; all leaves had a greater mean xylem Feret diameter than spinach. Based on the results of this study, Chicago hardy fig and sweet basil possess vascular density closest to spinach, mean Feret xylem diameters greater than spinach, and have potential to function as decellularized scaffolds for seeding endothelial cells through the petiole. Additionally, leatherleaf viburnum and sage may be useful in drug delivery to a graft because of the uniform perfusion pattern observed during the dye perfusion test. 

Acknowledgements:  

This work was supported by NSF REU grant EEC2150076 and NIH grant 1R15AR080988. 

Developing a Low-Cost Particle Image Velocimetry System 

• Presenting Author: Katie Boeckman: Oklahoma State University (REU Participant) 
• Primary Investigator: Zhenglun “Alan” Wei, PhD 
• Co-Author: Sucheta Tamragouri
  

Introduction:   

Understanding cardiovascular flow structure is necessary in the diagnosis and treatment of anomalies. In vitro experiments provide this useful information without clinical intervention. However, most in vitro methods used to visualize cardiovascular flow are often invasive and disruptive. Particle image velocimetry (PIV) is a non-invasive way to visualize flow, measure velocity, and analyze flow structure. PIV uses a fluid containing particles, a laser to illuminate those particles, and a camera to capture images of the illuminated particles. The images captured with the camera are then analyzed through cross-correlation to calculate the distance and direction traveled by the particles between consecutive images. Velocity can be calculated using the estimated displacement and the known time between the images. Typical PIV systems require a synchronized high-powered laser and a high-speed camera, which make PIV systems expensive and inaccessible to most. Therefore, there is a need for an accessible, low-cost PIV system. 

Materials and Methods:  

The test section of the system used an acrylic pipe surrounded by an acrylic box to minimize the refraction of the laser caused by the curved surface of the pipe. Water was chosen as the testing fluid because it has approximately the same index of refraction as acrylic. Particle to water concentrations of 25, 50, and 100 µL of particles to 2000 mL of water were tested. In addition, the surrounding box was filled with water to further reduce the unwanted refraction of the laser. The length of the pipe was chosen to ensure fully developed flow at the region of interest. Air vents were added to the pipe and surrounding box to minimize the air bubbles in the system. The laser was placed at a height that minimized glare and maximized focus. A phone camera was used to record videos at 30, 60, 240, and 960 frames per second (fps) to determine the best settings to capture laser illumination without a synchronizer. Stands were used to stabilize the tubing and test section, and a tripod was used to stabilize the camera. A steady flow pump was used to power the flow. Clamps were used to optimize the flow rate for the chosen camera and laser settings. A flow sensor was used to measure the flow and provided a comparison for the PIV results. 

Results, Conclusions, and Discussions:  

An accurate PIV system was built for less than $200. The system was able to accurately quantify velocity – within 10% of the measured flow rate. Flow structure and velocity profiles were accurate compared to literature results. 50 µL of particles in 2000 mL of water paired with a camera speed of 60 fps provided the clearest images and the most accurate results. Decreased particle concentration did not provide enough particles per frame, and increased particle concentration overwhelmed the image. Increased frames per second captured images without laser illumination. In conclusion, though there are some limitations in frame rate and flow rate, a low-cost PIV system was designed and provided accurate results. In the future, this system can be used for calibration in future PIV testing and can be used to make PIV more accessible for in vitro flow visualization and analysis. 

Acknowledgements:   

This work was supported by NSF REU grant EEC2150076. 

Dynamic Heart Performance Assessment: Mathematical Models for Tackling Heart Transplantation Challenges in Real Time 

• Presenting Author: Farhad R. Nezami, Brigham and Women’s Hospital 
• Co-Author: Mostafa Asheghan, PhD (ECE)
  

Introduction:  

Cardiac transplantation, often considered the ultimate treatment for end-stage heart failure, encounters a significant obstacle in the form of organ availability mismatched with the high demand from recipients. Normothermic ex-vivo heart perfusion (EVHP) platforms have emerged as a promising solution to expand the heart transplantation pool by enhancing the functionality of discarded hearts. However, there exists a pressing need for standardized assessment of the efficacy of this approach. To address this gap, we propose leveraging well-established mathematical modeling techniques to facilitate such assessment. Our approach involves applying mathematical modeling to an ex vivo experiment comparing a set of Swine hearts at two different cold storage times, thereby demonstrating the practical application of this methodology in evaluating the effectiveness of EVHP. 

Materials and Methods:  

In collaboration with our clinical partners, we successfully restored the function of swine hearts following different cold storage times, as determined by their cold ischemia time (CIT). Our team developed an in-house ex-vivo heart perfusion system capable of assessing heart function in both Langendorff and loaded models. Various signals, including left ventricle pressure, aortic tube pressure, and flow rates, were measured during the heart runs and utilized to optimize our computational model (Fig.1). Leveraging MATLAB Simulink, we constructed a lumped parameter model (LPM) of the cardiovascular system, incorporating passive electrical elements and active voltage/current sources. Model parameters were fine-tuned to match experimental and clinical recordings, minimizing error using the sum squared error (SSE) cost function. 

Results, Conclusions, and Discussions:  

We have developed a comprehensive model that captures the dynamic behavior of the left heart, including the corresponding pressure and flow dynamics at various positions within the vascular tree. This robust model offers diverse waveforms representing functional variables across different pathological conditions in relevant anatomical segments. Through rigorous validation against experimental data and existing models, our approach demonstrates real-time computational efficiency using ordinary differential equations. Importantly, our model enables non-invasive extraction of functional information from multiple locations, eliminating the need for invasive measurements in clinical settings. Intriguingly, our findings challenge the conventional belief that longer CIT pose a greater risk to heart function upon reperfusion. Our ex-vivo reperfusion experiments revealed comparable heart function in hearts subjected to longer CIT durations compared to those with shorter or no CIT. The swine heart subject to a 0hr CIT showed an estimated ejection fraction of 32% whereas the 5hr CIT one showed an ejection fraction of 34% (Fig.2). This observation carries significant implications, potentially expanding the pool of viable donor hearts by allowing procurement from more distant sites. 

Our model accurately replicates the hemodynamic waveforms observed in the ex vivo experimental setup. Utilizing these outputs in conjunction with echocardiography measurements conducted during the experiments, we are able to successfully extract key metrics of heart function. Performance analysis, including the computation of pressure-volume loops from which stroke volume and ejection fraction are derived, indicates comparable function between the two experimental conditions with different CIT. The developed physiological lumped-parameter model (LPM) serves as a comprehensive tool that advances our understanding of cardiovascular function. This innovative approach has the potential to revolutionize transplantation research by streamlining the prompt evaluation of heart function and contractility, while also reducing the costs associated with ex vivo and in vivo experimentation in normothermic ex vivo heart perfusion (EVHP). 

Acknowledgements:  

The authors acknowledge the assistance provided by George Olverson and Doug Vincent in conducting the experiments. 

Effect of Fluid Shear Stress on Endothelial Cell Basal CD44 

• Presenting Author: Udaya Rattan 
• Primary Investigator: Solomon Mensah, PhD
• Co-Author: Jacqueline O’Donnell, Zoe Vittum
  

Introduction:   

The glycocalyx (GCX) is a grass-like structure surrounding endothelial cells (ECs) made of components known as core proteins and attached carbohydrate chains called glycosaminoglycans (GAGs). These components are responsible for transducing mechanical signals such as shear stress induced from exposure to blood flow, causing EC remodeling. The behavior of core proteins and GAGs has been studied primarily on the apical surface of ECs, while the basal surface remains unexplored. We chose to study the core protein, CD44, which is present on both the apical and basal surface of ECs along with its GAG, hyaluronic acid (HA). CD44 is a transmembrane protein that has no conjugates and only binds to HA. Studying these allows us to consistently investigate EC remodeling behavior affected by changes in these components’ expression. Our goal is to characterize the expression of basal CD44 and cytoskeletal remodeling when ECs are exposed to different physiologically relevant rates of apical shear stress. 

Materials and Methods:   

Human Lung Microvascular Endothelial Cells (HLMVECs) were seeded on fibronectin-coated glass slides. Blood flow was simulated inside a parallel plate flow chamber using HLMVEC growth media circulated by a peristaltic pump. Shear stress rates of 10 dynes/cm2, a physiologically normal rate, and 0.5 dynes/cm2, a pathophysiologically low rate, were investigated for 30 minutes and 12 hours. After flow slides were washed with 3% BSA in PBS and fixed with 4% paraformaldehyde and 2% sucrose for membrane stabilization. Slides were permeabilized with a solution of 10% Goat Serum, 3% BSA, and 0.1% Saponin in PBS. This was followed by immunostaining for CD44 using a primary solution of CD44 binding protein and 1% BSA in PBS, and a secondary solution. Additionally, immunostaining was done with phalloidin for cytoskeleton visualization using 1ug Phalloidin /1 ml of 1% BSA and 0.1% saponin in PBS. Slides were mounted with DAPI media for visualization of the CD44, cytoskeleton, and nuclei using a laser-scanning confocal microscope. 

Results, Conclusions, and Discussions: 

To isolate the apical and basal signatures within the Z-stacks collected, the centroid of the nucleus was identified and used to generate apical and basal sub-stacks. Sum projections as seen in Figure 1 (A, B, and C) were generated of apical and basal sub-stacks for analysis of coverage (Figure 1D and F) and integrated intensity (Figure 1E and G). Additionally, Marcotti, et al.’s method for quantifying cytoskeletal alignment was implemented (Figure 2E and H) (1). The presence of CD44 on both surfaces suggests that, through their connection to the cytoskeleton, they may form the whole cell force transduction network as suggested by other models (3). We found CD44 expression consistently increases at a pathophysiologically low shear rate, while at a normal shear rate, expression levels show an initial increase followed by a return to static levels at 12 hours. We noted that basal CD44 shows significantly higher expression levels after 12 hours at a low shear rate compared to a physiologically normal shear rate. Overall, we can conclude that basal CD44 may play a significant role in EC remodeling at pathophysiologically lower shear rates, possibly implicating CD44 and HA in vascular disease progression associated with low fluid shear rates such as bifurcated vessels (2). This is coupled with decreased cytoskeleton expression at a low shear rate compared to a physiologically normal shear rate. There is an overall lower degree of basal cytoskeletal organization compared to apical at both shear rates; however, at a low shear rate, basal cytoskeletal organization is higher (Figure 2H) than under physiologically normal conditions (Figure 2E). The increase seen in both basal cytoskeletal organization and basal CD44 expression at low shear rates may be due to the basal CD44 and cytoskeletal axis possibly implying a signaling pathway triggered in the diseased state altering basal remodeling. These findings help establish a start to understanding the role of basal cell components when exposed to apical shear forces. Future studies will investigate CD44 expression at a pathophysiologically high shear rate along with the effect of removing HA from CD44 on HLMVEC remodeling and CD44 expression. 

Effects of Modulating YAP Signaling on Lymphatic Endothelial Cell Morphology and Cell-Cell Junction Formation 

• Presenting Author: Zhyan Noble
• Primary Investigator: Catherine F. Whittington, PhD
• Co-Author: Brian Ruliffson
  

Introduction:   

Lymphatic vessels are a significant part of the immune system and includes lymphatic capillaries that are directly attached to the underlying extracellular matrix (ECM) and respond to ECM changes. Yes associated protein 1 (YAP) is a mechanosensitive transcription factor, and ECM stiffness and YAP activity can affect growth of lymphatic endothelial cells (LEC), the cells that make up lymphatic capillaries. ECM stiffness has been implicated in YAP activation and lymphatic vessel formation, and we have observed that increased ECM stiffness results in increased cell area and junction formation of VE-Cadherin, an adhesion molecule between LECs that helps control fluid flow across capillary walls (i.e., vessel function). Stressors can lead to changes in LEC actin cytoskeleton dynamic, which then affects YAP activity, cell morphology, and possibly VE-Cadherin. Yet, less is known about how YAP activity affects LEC morphology and capillary function. We aim to investigate whether cell morphology and VE-Cadherin expression change in response to mevalonic acid and simvastatin treatments that promote and inhibit activity downstream of YAP. This approach simulates stiffness-mediated changes in YAP activity. We hypothesize that mevalonic acid will increase cell area, perimeter, and VE-Cadherin, while simvastatin will decrease cell area, perimeter, and VE-Cadherin. 

Materials and Methods:   

YAP modulation treatments were applied to primary human dermal lymphatic endothelial cells (LEC; PromoCell) to evaluate effects on LEC cell morphology and function. Treatments include promotion using mevalonic acid and inhibition using simvastatin. LECs were cultured in MV2 endothelial cell supplemented media (PromoCell) and seeded in a 96-well plate at 20,000 cells/cm2 to analyze morphology and VE-cadherin. Cells were divided into 3 categories: Control, YAP promotion, and YAP inhibition, with each treatment category made up of a low (n=3; below recommended dosage), medium (n=3; recommended dosage), and high (n=3; above recommended) dosage. Mevalonic acid treatment groups were: 0.25 mM, 0.5 mM, and 1 mM within MV2 media. The simvastatin treatment groups were: 0.5 µM, 1 µM, and 2 µM in MV2 media. Treatments were applied 24 hours after initial cell seeding and treated for an additional 24 hours. Cells were fixed and stained with Phalloidin (F-Actin), DAPI (nucleus), and anti-VE-Cadherin antibody and imaged using the Keyence BZ800 Fluorescence Microscope. Fluorescence images were analyzed in CellProfilerTM on cell shape metrics (area, perimeter, form factor). Using GraphPad 10, Shapiro-Wilks tests were used to determine normality and a one-way Anova with a post-hoc Tukey’s test was used to determine differences between groups. 

Results, Conclusions, and Discussions:  

Qualitative evaluation of treated LECs did not show significant morphological differences (Figure 1). However, after analyzing samples, the mevalonic acid treatment (YAP promoter) showed a slight decrease in cell area and no significant change in cell perimeter compared to the no-treatment control (Figure 2). In the simvastatin treatment (YAP inhibitor), there was an increase in cell perimeter at a high dosage and decreased form factor, while cell area remained unchanged when compared to the no-treatment control. These shape trends combine to suggest that higher YAP inhibition made cells more irregularly shaped toward a more characteristic LEC oak leaf shape. These data contradict what previous studies have observed when YAP nuclear accumulation (i.e., activation) was modulated with substrate stiffness. The discrepancy of results highlight the need for further research into elucidating the mechanisms by which stiffness and YAP modulate cellular morphology. Our VE-Cadherin results were inconclusive due to a short culture time that did not allow for junction formation (Figure 1). In future, the pre-treatment culture time will be extended to allow junctions to form prior to YAP modulation. In those experiments, it is expected that promoter groups will result in VE-Cadherin junctions with tighter, zipper-like cell junctions. In the inhibition group, the expectation is to see discontinuous and button-like VE-Cadherin junctions, implying looser cell junctions. Additional studies will be designed to link YAP activity, ECM stiffness, and lymphatic capillary function. Those studies will help solidify knowledge on how YAP alters junction formation and subsequent capillary function, which will guide researchers in new lymphatic endothelial research toward understanding factors that affect function, as well as growth. 

Acknowledgements:   

I would like to thank Professor Whttington, Brian Ruliffson, Athenia Jones, and Stephen Larson for their assistance in this project. This work was supported by NSF Engineering Research Initiation Award (2138841), NIH/NCI R03 (5R03CA267449-02), Genentech Research Award, and NIH URISE@WPI (1T34GM153552-01). 

Efficacy of K-Nearest Neighbors Algorithm for Diabetes Onset Classification 

• Presenting Author(s): Milad Jaffari, Benjamin Breslov, Leah Johnson
• Last Author: Taimoor Afzal, PhD
• Co-Author: Brendan Dodier, Brandon Morales
  

Introduction:  

Building upon the current understanding of diabetes mellitus classification using machine learning, our research explores the efficacy of the K-Nearest Neighbors (KNN) classification method to enhance prediction accuracy for diabetes onset. Leveraging the studied Pima Indian Diabetes Dataset, which includes diverse physiological metrics from subjects, we aim to refine predictive models by implementing and tuning the KNN algorithm. This approach is motivated by the limitations identified in previous studies where traditional algorithms, while robust, often lack flexibility in handling real-world data variabilities. 

Our study particularly addresses the challenge of achieving high classification accuracy amidst the inherent complexities of medical datasets, characterized by imbalanced classes and varying degrees of feature dependencies. The KNN method is chosen for its simplicity and effectiveness in classification tasks where relationships between features significantly influence outcomes. By experimenting with different configurations and feature sets, we optimize our model to capture better the nuanced patterns that signify the progression to diabetes. 

Furthermore, our work contributes to the broader field of biomedical engineering by advancing diabetes prediction and enhancing the interpretability of machine learning models in medical diagnostics. In essence, this research aims to refine diabetes prediction models using KNN, offering insights into its applicability and effectiveness in medical diagnostics, and potentially setting a precedent for the adoption of similar techniques in other complex disease prediction scenarios. 

Materials and Methods:  

The dataset with 768 samples was obtained from the National Institute of Diabetes and Digestive and Kidney Diseases aimed to predict diabetes outcomes, utilizing KNN algorithm, a machine learning method classifying outcomes into binary values. The parameters that were provided in the dataset include Insulin and Glucose levels, Number of Pregnancies, Age, Blood Pressure, BMI, Skin Thickness, Diabetes Pedigree Function, and Diabetes Outcome. Our preliminary analysis showed that Insulin levels, Glucose levels and number of pregnancies were the most viable features for diabetes classification, so we decided to test different combinations of these features to predict diabetes. We tested the KNN classifier using two pairs of features: 1) Insulin levels and number of pregnancies and 2) Insulin levels and Glucose levels. Model training was crucial for the algorithm to discern correlations between variables and diabetes outcomes. Leave-one-out cross-validation method was used to evaluate the trained model’s performance. The algorithm’s scalar input determined the number of nearest neighbors considered when classifying new data points, assessing the relationship strengths between features like insulin levels and pregnancies. 

Cross-validation iteratively tested K values from 1 to100 to optimize model accuracy, sensitivity, and specificity. Contrastingly, other ML techniques, including Naive Bayes, random forest classifiers, and J48 decision trees, are constrained by sample size and assumptions such as axis-parallel decision boundaries, potentially leading to overfitting or underfitting. KNN’s adaptability to growing datasets and simplicity post-training sets it apart from these models, as it doesn’t require a training phase. 

Results, Conclusions, and Discussions:  

In case 1, the model achieved a classification accuracy of 0.707 at K=13 and specificity of 0.9 (figure 1). Sensitivity was at 0.3. For case 2, classification accuracy was 0.66 at K=59, sensitivity was at 0.95 and specificity was around 0.1 (figure 2). In both cases, accuracy, specificity, or sensitivity plateaued for K>50. The low specificity indicates high false positives showing that the current feature combinations gave us high true positives but less true negatives. 

We demonstrated the efficacy of the KNN classification method to predict diabetes. Through our prior analysis of the Pima Indian Diabetes Dataset, we found a significant correlation between physiological metrics and diabetes onset, with insulin levels being the crucial predictor. While KNN offers simplicity and adaptability, its shortcomings is its reliance on the tuning of the function, particularly regarding the number of neighbors. Our analysis shows that the value of K would vary based on the features being used to classify. This is an interesting finding highlighting how different features capture different aspects of the data.

We observed both insulin and glucose levels and insulin levels and number of pregnancies produced similar levels of accuracy. Potential benefits of utilizing this KNN classify method could expand diabetes research, in diagnosing Type 2 diabetes earlier based on factors such as BMI, blood glucose levels, pregnancies, and blood pressure. However, this same method has a level of simplicity that correlates these factors incorrectly when using a lower sensitivity value, like k=2 and misdiagnoses the outcome. Accuracy alone cannot be used to measure the performance of a classifier. From our results both cases showed a high level of sensitivity but a reduced level of specificity. Improvement in these numbers would require bringing in more features that may not necessarily show a correlation with diabetes onset when individually analyzed. For future research, we would like to perform the analysis for more than two features using the KNN classification method and expand this to other populations. Furthermore, this method could also be used to classify other chronic illnesses based on the features included in the data set. 

Elucidating the Role of Fibronectin Conformation on Synovial Fluid Film Formation Using a Simple Model of Articular Cartilage Surface  

• Presenting Author: Katelyn Lunny
• Primary Investigator: Roberto Andresen Eguiluz, PhD UC: Merced
  

Introduction:   

Synovial fluid (SF) is a lubricant found between articulated joints that reduces friction and surface damage during movement. The molecular mechanisms through which articular cartilage surface extracellular matrix (ECM) proteins regulate the adsorption of SF components remain largely unclear. Pathologies, like arthritis, can impact the concentration and conformation of such proteins, fibronectin (Fn) among others. This research investigates the hypothesis that unfolded Fn protein conformation facilitates the adsorption of SF components for the formation of a lubricating and protecting film. 

Materials and Methods:   

The adsorption of SF components is quantified using the Quartz Crystal Microbalance with Dissipation (QCM-D) technique, which records time-resolved frequency and dissipation changes in a resonating quartz crystal. The QCM-D crystal surface is functionalized in this model with Fn, to mimic the Fn at the articular cartilage surface. The precursor film of Fn is formed from 35-50 ug/ml of bulk concentration in pH 4 to induce conformational changes, or in pH 7 to maintain the conformation. Then, SF at 25% dilution is flowed over the crystal surface in the QCM-D fluidic chamber. The adsorbed mass of these films is quantified using the Sauerbrey equation, which communicates the linear relationship between the frequency changes of an oscillating quartz crystal and its mass changes. 

Results, Conclusions, and Discussions:  

Preliminary results show important differences in the amount of adsorption of SF on Fn films formed at pH 4 (unfolded) vs. pH 7 (compact) and the amount of Fn precursor film adsorption between the two concentrations tested, 35 ug/ml and 50 ug/ml. These results indicate more Fn precursor film adsorption in pH 4 at 50 ug/ml and more SF adsorption at 35 ug/ml in pH 4 and 7. Future studies will characterize unfolded and folded fractions of Fn due to changes in pH and ionic strength. We predict the Fn conformational changes impact the adsorption of SF components and therefore the lubricating and surface damage protecting performance function. 

Endothelial Glycocalyx-Dependent Expression of Endothelin-1 

• Presenting Author: Samantha F. Cocchiaro
• Primary Investigator: Solomon Mensah, PhD
  

Introduction:  

Endothelin-1 (ET-1) plays a significant role in many cardiovascular-related conditions such as hypertension, and atherosclerosis. ET-1 is produced by endothelial cells (EC) and secreted into the extracellular space to bind to either endothelial B receptor (ETB) on EC or the endothelial A receptor on smooth muscle cells. Reports suggest that the expression of ET-1 is shear stress magnitude and time-dependent and that the disruption of cytoskeletal structures could mediate the shear stress-induced production of ET-1. The endothelial glycocalyx (GCX) which is a mechanotransducer and is located in close proximity to ETB and is also firmly rooted in the EC membrane through its core proteins could co-participate in the release of ETI during disease. 

Materials and Methods:  

With the use of immunofluorescence staining, enzymes that target specific GCX components and micro-fluidic technology, we defined the role of shear forces and the GCX in the expression of ET-1 and its receptor ETB on the surface of human lung microvascular endothelial cells. Flow trials were 30 minutes long and were done with and without Heparanase III to emulate a degraded glycocalyx. 

Results, Conclusions, and Discussions:  

We found that the shear-dependent increase in the expression of GCX component heparan sulfate resulted in a significant increase in the expression of ETB. We also discovered that enzyme-induced removal of GCX heparan sulfate resulted in the decrease in expression of ETB and a significant increase in the release of ET-1. These results confirm that GCX HS could co-participate with ETB to release ET-1, in an alternative GCX-mediated pathway for modulating the expression of ET-1 during disease. 

Engineering an Ecosystem for Recycling Spent Cell Culture Media using Chlorella Sorokiniana 

• Presenting Author: Richard Thyden, PhD 
• Primary Investigator: Glenn Gaudette, PhD (Boston College)
• Co-Author: Pamela Weathers, PhD, Tanja Dominko, PhD
  

Introduction:  

Due to their costs and availability, many currently marketed advanced therapies, such as cell and antibody therapies and 

tissue engineered constructs are limited in their access to large populations. Additionally, the realization of whole-organ engineered constructs or cultured meat will depend on affordable manufacturing of large quantities of cells. As cells grow, metabolic waste, such as ammonia, accumulates in the cell culture media to the detriment of critical cellular functions, including proliferation. It is estimated that 300 million liters of cell culture media, a costly input, are discarded per year [1]. Spent media may be up-cyclable if metabolicwaste products are sequestered and key media components are replenished. We hypothesized that Chlorella Sorokiniana, a thermoresistant strain of unicellular algae, can grow in spent animal cell culture media at 37°C, in the absence of light, and remove ammonia waste to improve media use efficiency. 

Materials and Methods:  

Spent media stock was collected from QM7 cell cultures (a quail myoblast cell line) containing fresh media (GM) (DMEM/F12, 10% FBS, and 1% Pen/Strep) after 4 and 8 days (4D-SGM and 8D-SGM, respectively). C. sorokiniana was inoculated into fresh GM, 4D-SGM, and 8D-SGM at 37C, with or without light, and growth curves were generated via absorbance measurement. Ammonia and glucose consumption curves from 4D-SGM were generated via enzymatic assays. An MTT assay was used to assess QM7 metabolic activity in algal-treated, 4D-SGM, and fresh GM. 

Results, Conclusions, and Discussions:  

Results:

Sorokiniana exhibited the fastest growth in fresh GM when exposed to minimal or no light, indicating its capacity for metabolic use ofglucose independent from sunlight (Fig 1A). Greater algal growth was observed in both 4D-SGM and 8D-SGM than in fresh GM after7 days (Fig 1B). This is believed to be due to the increased ammonia concentrations in spent media. A high density of algae (10 8 algae/ml) removed nearly all ammonia from 4D-SGM within 24 hours of treatment, and due to light independent metabolism, consumed nearly all remaining glucose within 48 hours of treatment. Greater metabolic activity of QM7 cells was observed in algal treated 8D-SGM compared to untreated 8D-SGM.

Conclusions:  

We identified the rates at which C. sorokiniana can sequester ammonia waste and we compared the metabolic activity of animal cells grown in fresh, spent, and algal-treated spent media. By incorporating algae into mass-bioproduction systems for animal cells, aimed at reducing the cost of the most expensive input, cell culture media, we move closer to developing an engineered ecosystem for enhancing bioproduction capacity, reducing costs, and ultimately improving equity of next generation medicines. 

Acknowledgements:  

This research was partially funded by a New Harvest fellowship awarded to Richard Thyden 

Fluid Shear Stress-Dependent Remodeling of Endothelial Cell Basal Hyaluronic Acid 

• Presenting Author: Jacqueline O’Donnell
• Primary Investigator: Solomon Mensah, PhD
  
• Co-Author(s): Udaya Rattan, Zoe Vittum
  

Introduction:   

The endothelial glycocalyx (GCX) is a grass-like structure that surrounds the surface of endothelial cells (ECs) serving to transduce extracellular signals and perform barrier functions. As blood flows over ECs the GCX transduces fluid shear stress instigating EC remodeling. The apical GCX has been primarily studied due to its direct role in sensing fluid shear stress. However, the basal GCX which interfaces with the smooth muscle layer has also been shown to be involved in the mechanotransduction of fluid shear stress. Hyaluronic acid (HA) is one of the many carbohydrate chains that serves as mechanotransducer in the GCX. Our aim is to understand the impact of fluid shear and HA on vascular ECs with a particular interest in defining HA’s role in the basal GCX. 

Materials and Methods:   

Slides were coated with fibronectin and seeded with human lung microvascular endothelial cells (HLMVECs). The HLMVECs were then subjected to flow at 0.5 dynes/cm2, a pathophysiologically low flow rate, and 10 dynes/cm2, a physiologically healthy flow rate, for 30 minutes and 12 hours using a flow chamber and peristaltic pump. After flow slides were washed with 3% BSA, then cells were fixed in 4% paraformaldehyde and 2% sucrose in PBS. Samples were then immunofluorescent stained for HA and Phalloidin. HA samples were permeabilized using a solution of 10% goat serum, 3% BSA, and 0.1% saponin in PBS. HA was then exposed to a primary solution of HA binding protein in 1% BSA in PBS, then a secondary solution. Cytoskeleton samples are stained in a solution of 1ug Phalloidin / 1ml of 1% BSA and 0.1% saponin in PBS. All samples were mounted using DAPI mounting media to visualize the nuclei. The slides were then imaged using a laser scanning confocal microscope to collect Z-stacks. 

Results, Conclusions, and Discussions:  

To analyze the apical and basal cells separately, the average centroid of all nuclei in each image was used to divide the Z-stack collected to generate apical and basal sub-stacks. The apical and basal sub-stacks were used to generate sum projections for analysis of coverage (Figure 1D and Figure 2D,G) and integrated intensity (Figure 1E and Figure 2F,I). To analyze cytoskeletal alignment a previously published method [1] was implemented to generate order parameter (Figure 2E,H) measurements. Immunofluorescent sum projections of HA shown in Figure 1 (A,B,C) showed that HA was present in both the apical and basal GCX and coverage significantly increased after flow showing that HA is fluid shear stress dependent. HA expression trends upwards in apical and basal GCX however no significance was found. Only physiologially healthy data is shown for HA due to imaging limitations. Immunofluorescent sum projection of the cytoskeleton at a healthy shear rate, shown in Figure 2 (A,B,C), showed an initial decrease followed by an overall increase in cytoskeletal alignment over time (Figure 2E). However, a lower degree of cytoskeletal organization is seen in the basal cell which reflects literature findings [2]. At a pathophysiologically low shear rate apical and basal cytoskeletal organization steadily increases (Figure 2H) with basal organization higher than at a healthy shear rate. The presence of HA in both the apical and basal GCX may result in whole cell signaling responses to apical fluid shear stress through connections to the cytoskeleton. The response of the basal GCX to altered apical stimuli supports this idea and findings of other models that have investigated the mechanotransduction of apical fluid shear stress by the basal GCX [3]. Future work will investigate the behavior of ECs at a physiologically high shear rate and without HA present in the GCX. The investigation of pathophysiological shear rates will provide insight into important disease states such as surrounding atherosclerotic plaques (high) and bifurcated areas (low). Investigation of HA deficient cells will provide crucial insight on the role of HA as a mechanotransducer in both the apical and basal GCX. 

Functionalized Leaf-derived Vascular Structures (LeaVS) Modulate In Vitro Skin Regeneration 

• Presenting Author: Bryanna L. Samolyk 
• Primary Investigator: George Pins, PhD
• Co-Author: Tanja Dominko, PhD
  

Introduction:  

Every year, more than 500,000 Americans are treated for burn injuries and chronic wounds. These injuries often lead to extensive scarring and permanent loss of function. The current standard of care is a split-thickness skin autograft; however, more extensive burns may have insufficient donor sites, requiring an alternative treatment. Dermal analogs have demonstrated some success but require 2-3 weeks for vascular integration. The absence of robust internal vascular networks is typically the cause of graft failure. So, we need to develop implantable skin scaffolds that already contain vascular network structures. The project aims to develop Leaf-derived Vascular Structures (LeaVS) that support skin regeneration and provide provisional vascular support. Previous data showed that decellularized leaf scaffolds retain their vascular networks while maintaining structural integrity. We hypothesize that the surfaces of these LeaVS can be functionalized with fibronectin (FN) to promote keratinocyte attachment, epidermal stratification, and differentiation. 

Materials and Methods:  

To prepare LeaVS, spinach leaves were subject to cuticle removal, decellularization, and sterilization before being stored at 4C. Keratinocyte attachment assays were performed by seeding normal human epidermal keratinocytes (NHEKs) at 500,000 cells/cm2 on LeaVS or tissue culture polystyrene (TCP) controls treated with 0, 0.01, or 0.1% poly-L-lysine (PLL) as well as 0 or 50 g/mL human fibronectin (FN) for 2 hours. Samples were stained with Hoechst, imaged, and quantified for cell density. Epidermal morphology assessments were performed using leaves treated with 0 or 0.01% PLL as well as 0, 10, 50, or 100 g/mL FN. Neonatal human fibroblasts (NHFs) were seeded at 100,000 cells/cm2 on the underside of the leaves, followed by neonatal human keratinocytes (NHKs) seeded at 1 million cells/cm2 on the topside of the leaves. For decellularized dermis (DED) controls, only NHKs were seeded. Constructs were cultured submerged for 3 days, followed by culture at the air/liquid interface for 7 days. Histological analyses of hematoxylin and eosin (H&E) stained samples were performed to quantify the epidermal thicknesses. 

Results, Conclusions, and Discussions:  

Keratinocyte attachment assays showed that the TCP positive controls were not significantly enhanced by surface treatment. Interestingly, the LeaVS showed decreasing cell attachment as PLL concentration increased. However, the addition of FN to LeaVS resulted in a significant increase in cell density for PLL groups. Untreated LeaVS did not exhibit a significant difference in NHEK attachment compared to the untreated TCP controls. Epidermal morphology experiments demonstrated that LeaVS support the growth of a contiguous layer of keratinocytes with characteristic cobblestone morphology and progressive epithelial stratification, as well as fibroblast attachment. Interestingly, untreated LeaVS had the thickest epithelial layer (~20 m) compared to other treatment conditions and was comparable to the DED positive control (~20 m), although not as thick as native epithelial layers (~80 m), which is expected as it is a fully matured tissue (Figure 1). 

These studies describe the strategic functionalization of LeaVS to modulate keratinocyte attachment as well as epidermal stratification and differentiation. We showed that fibronectin conjugated LeaVS displayed promise for supporting cellular attachment and epidermal stratification. We anticipate that these scaffolds will enable the future design of a multifunctional engineered skin substitutes for the treatment of traumatic skin injuries by facilitating rapid tissue vascularization and functional regeneration. 

Acknowledgements:  

This work is supported by NIH 1R15AR080988. 

Image Guidance and Safe Insertion Region Localization for Lumbar Puncture using Ultrasound 

• Presenting Author: Aabha Tamhankar (RBE)
• Last Author: Christopher J. Nycz, PhD (RBE)
  

Introduction:  

Lumbar Puncture (LP), or Spinal Tap, is a medical procedure used to access cerebrospinal fluid (CSF) for diagnostic and therapeutic purposes. Approximately 363,000 LPs are performed each year among ED-treated patients in the United States, and every year approximately 400,000 diagnostic lumbar punctures are performed by neurologists in the United States. Traditionally, this procedure relies on manual needle insertion guided by ultrasound and palpation of anatomical landmarks. The insertion is into the interspinal spacing between two spinous processes of the lumbar vertebrae. This interspinal space typically ranges from about 1 to 3 centimeters (cm). This can make the procedure complex and increase the learning curve. Misalignment of the needle can cause traumatic lumbar puncture or leakage of CSF, which has side-effects include headaches, cranial neuropathies, prolonged backache, nerve root injury, and meningitis. We believe that effective visualization can simplify the learning process and enhance the workflow in lumbar puncture procedure. Our work proposes the integration of 2D ultrasound imaging with 3D MRI model, aiming to create a real-time visualization tool for the lumbar spine and entry locations which might help in bringing ease to the puncture procedure. 

Materials and Methods:  

The process was divided into two phases: the Registration Phase and the Dynamic Scanning Phase. In the registration phase, the Ultrasound Lumbar Puncture Simulator IIA (Kyoto Kagaku) was scanned using a GE Logiq E9 Ultrasound Machine with a Linear Array Probe, capturing 20-25 images over approximately one minute. An intensity based image processing model was developed for identifying the edges of the spinous processes in the ultrasound images. A VICON motion capture system tracked the probe poses, and using corresponding image points, a sparse ultrasound point cloud was constructed. This point cloud was initially aligned with a ground truth MRI point cloud using the spinous process tips of both modalities as 4-4 correspondences. This alignment was refined using the Iterative Closest Point Algorithm and a transformation matrix between MRI ground truth and ultrasound point cloud was obtained. In the dynamic scanning phase, the established transformation train is used to overlay MRI-defined safe regions onto real-time ultrasound images. To overcome errors in the registration due to inaccurate edge detection or motion of the body, a local optimization technique was introduced into the dynamic scanning pipeline. For every real-time ultrasound image captured, the probe pose is localized on the MRI ground truth, and an ultrasound equivalent slice of MRI is taken. In local optimization, the distances between edges of ultrasound image and MRI slices are minimized while updating the registration between them. 

Results, Conclusions, and Discussions:  

The registration, transformation, and local optimization is used for real-time overlay of predefined “safe regions” from the MRI model onto ultrasound images. The transformation between MRI and Ultrasound point clouds is continuously updated by local optimization and Intersection of Union (IoU) percentages are used to track alignment quality. Testing on 44 dynamically scanned ultrasound images showed an improvement from an average IoU of 72.15% before optimization to 85.35% after optimization. This system establishes a framework for merging ultrasound images with MRI ground truth to provide a visual guide to lumbar puncture procedures. Future developments will focus on defining needle entry and target points directly on MRI scans to facilitate precise, real-world insertion guidance. We can also explore the use of a generalized 3D spine model for broader applicability, alternative tracking technologies for clinical environments, and adjustments to accommodate multi-directional patient movements. These advancements can potentially reduce procedural learning curves and improve outcomes significantly. 

Improving Quality of Automated Image Analysis for High-throughput Neuromodulator Screens 

• Presenting Author: Ananya Sundararajan: Case Western Reserve University (REU Participant)
• Primary Investigator: Dirk Albrecht, PhD
• Co-Author: Fox Avery
  

Introduction:   

Chemical neuromodulators can be leveraged as potential therapies for neuropsychological disorders. We developed a high-throughput functional screening method that can analyze the effects of such neuromodulators on neural activity over time in living organisms [1]. This method yields large imaging data sets, which can increase the power and certainty of results. However, manual data processing is time consuming and often results in imprecise neuron detection [1]. A previously constructed automated data analysis system reduced manual processing time by 6-fold (22 hours of manual processing vs. 3.5 hours of automated processing), but resulted in ~10% increased variability, likely due to neuron selection and position tracking errors [2]. This work focuses on refining the automation of high-throughput data analysis to improve accuracy and validate neuron identification and quality. The increased accuracy and reduced manual processing time should allow for more sensitive and reliable analyses and identification of neuromodulators with therapeutic potential. 

Materials and Methods:  

Raw image data were obtained from prior experiments using Caenorhabditis elegans that co-expressed the fluorescent calcium sensor GCaMP and optogenetic excitatory ion channel Chrimson in AWA chemosensory neurons (Figure 1) [1]. Animals were immobilized through hydrogel fixation in 384-well plates and compounds from the Library of Pharmacologically Active Compounds (LOPAC, 1280 drugs) were added to the wells at 100 uM. 1% DMSO was used for negative controls and calcium channel blocker Nemadipine-A was used for positive controls (10 and 100 uM). Neurological responses to 5-second optogenetic activation (via 617 nm red light) were recorded every 6 hours for 18 hours [1]. 

Because active neurons increase fluorescence only during stimulation, individual candidate neurons were identified based on increased pixel brightness coincident with stimulation onset. A custom ImageJ script applied a Gaussian blur, minimum auto-threshold, and particle size filter. Candidate neurons were grouped by animal to accommodate minor movement between imaging intervals. A montage of the candidate neurons was created to validate chosen animals. Manual quality scoring was used to identify an automated algorithm that utilized max projections and thresholds to exclude candidates with excess movement, missing neurons, overlapping animals, and non-neurons. Pixel intensity data were recorded for each true neuron to generate a neural response trace. Mean peak neural response was averaged for each well and compared over time. Candidate hits were selected by a change in neural activity over 12 hours exceeding three standard deviations of negative vehicle controls (|Z-score| > 3). 

Results, Conclusions, and Discussions:  

While manual processing of this data required 22 hours of active user interaction per 384-well plate, the previous automated analysis reduced user interaction time 40-fold at the expense of a small increase in uncertainty in neuron selections. The refined algorithm described here processed data for a 384-well plate with equal user interaction time as the previous algorithm. Although the computation time increased when compared to the previous algorithm (10 hours vs. 3.5 hours), the refined algorithm increased certainty of the neurons that are tracked through assessment of a montage visualization of candidate regions (Figure 2). Candidate neurons were effectively identified using functional responses that increased signal intensity only during stimulation, excluding autofluorescence, dust, gut signals, and non-targeted neurons. Further, animals were 100% correctly identified and matched across time points, enabling assessment of drug effects in each animal. However, up to 40-50% of candidate regions were found to contain errors, including excess movement during the imaging trial, autofluorescent gut flashes, overlapping animals, rotation of animals, flipping of animals between cycles, and excess background noise. From manual categorical binning of data quality, further post-processing algorithms were developed that automatically classified neuron quality based on minimal movement and appropriate size, shape and location. Although higher quality data selection was expected to reduce variability among negative controls, it remained roughly the same with a coefficient of variation of 10% (Figure 3). This suggests that the variability seen in the data set is reflective of true biological variability. Additionally, felodipine was identified as a hit, consistent with previous manual and automated analyses (Figure 3). The improved algorithm and data quality strengthen the certainty of candidate neuroactive compound hits. Further work may involve refining threshold criteria for neuron quality. The analyzed data set targeted intracellular disruption of neural response, since optogenetic activation and activity output were in the same neuron. Future work will include analyzing the effects of pharmacological modulators on intracellular communication like glutamatergic synapses in the AFD-AIY neural circuit. This system can accelerate other studies of functional neural imaging, such as investigating neural activity after traumatic brain injury or dose response for neuroactive compounds. 

Indentation Measurement of Cell Mechanical Anisotropy Using a Toroidal Probe 

• Presenting Author: Juanyong Li 
• Primary Investigator: Kristen L. Billiar, PhD 
• Co-Author: Ying Lei 
  

Introduction:  

The mechanical properties of living cells reflect their physiological state and may serve as a hallmark for diseases such as cancer, calcification, and atherosclerosis. While many cell types are highly polarized in their structure, the mechanical anisotropy of the cells is often neglected in mechanical analysis. In most studies, an isotropic elastic modulus is used to report or describe the mechanical properties of the cell. This oversimplification may bias the interpretation in cell mechanics and mechanobiology.  

Understanding mechanical anisotropy of cells is hindered by a lack of accessible methods for measurement. Indentation is the most common method for measuring stiffness of cells, yet conventional indenters are equipped with spherical, conical, or pyramidal probes which do not yield information related to mechanical anisotropy. Alternative approaches such as magnetic twisting cytometry, magnetic tweezers, and micro biaxial stretch generally require custom-made devices or complex image analysis of the 3D deformation fields which are not accessible to most researchers. A straight-forward indentation method for the measurement of anisotropic stiffness of cells is needed.  

Here, we propose a method to measure anisotropic stiffness of cells by two nano-indentations in orthogonal directions using a probe mounted with a novel toroidal tip. We nano-3D printed probe tips with 10:1 aspect ratio and attached the tip to a commercial indenter probe, then indented aligned and isotropic cell monolayer and single polarized cells using a commercial nano-indenter equipped with this probe to demonstrate the applicability of this method. 

Materials and Methods:  

The toroidal probe tip was designed to be 10 µm in major radius and 1 µm in minor radius. The tip was 3D printed by two-photon polymerization in 50 nm resolution using a NanoScribe GT+ 3D printer. Then the tip was mounted to a tipless nanoindenter probe (Optics11). (Fig. 1).  

Anisotropic porcine aortic valvular interstitial cell (PAVIC) monolayers and isotropic PC-9 cell monolayers were cultured in 35-mm cell culture petri dishes with parallel scratches at the bottom (Fig. 2A). Single PAVICs were also cultured in 35-mm dishes. 

Indentation tests were performed using an Optics11 Chiaro nano-indenter equipped with the toroidal tip probe mounted onto an optical microscope. The cell monolayers were randomly indented at 15-25 spots in the direction parallel to and across the scratch. For single cells, the indentation was performed on both nucleus and cytosol along the directions parallel and perpendicular to the cell long axis. The indentation speed was set to 500 nm/s, with 500 nm maximum indentation depth. The order of indentation was randomized. The force-indentation curves were fit to the Hertzian model for elliptical contact, in which the effective Young’s modulus (E*) is calculated based on the force, indentation depth, and probe geometry. Effective Young’s moduli along and across the cell long axis were compared with a t-test with p< 0.05 considered significant. 

Results, Conclusions, and Discussions:  

Results: 

Mechanical anisotropy was found on all eight PAVIC monolayers, with an average effective Young’s modulus of 10.48±1.44 kPa in indentation parallel to the scratch and 24.47±3.87 kPa in indentation perpendicular to the scratch (p < 0.001). Anisotropic indentation on all three PC-9 monolayers did not find mechanical anisotropy, with an average effective Young’s modulus of 2.12±0.17 kPa in the parallel direction and 2.25 ±0.25 kPa in the perpendicular direction (p=0.63) (Fig. 3A). 

Anisotropic indentation on both the nucleus and cytosol of the individual PAVICs yielded a higher effective Young’s modulus when indenting along the perpendicular direction (p < 0.001) (Fig. 3B). Indentation on the cell nucleus yielded an effective Young’s modulus of 10.75±3.35 kPa in the parallel direction and 17.25±3.35 kPa in the perpendicular direction. The indentation on the cytosol yielded an effective Young’s modulus of 12.02±4.51 kPa in the parallel direction and 17.91±5.87 kPa in the perpendicular direction. 

Discussion & Conclusion: 

Despite the importance of cell polarity and cytoskeletal alignment, cell mechanical anisotropy is rarely measured experimentally due to technical difficulties. Here, we present a facile method to measure the mechanical anisotropy of cells by making two orthogonal indentations with a toroidal probe without the need for image analysis. Our methods can be implemented with most commercially available indenters and AFMs for a cost of ~$20 and the tip can be printed and mounted in less than four hours.  

In this study, we measured the mechanical anisotropy in aligned and non-aligned cell monolayers. As expected, we found the aligned cell monolayer to be mechanically anisotropic, while the non-aligned cell monolayer to be mechanically isotropic. The indentation on both cytosol and nucleus of PAVICs shows mechanical anisotropy with similar modulus. The mechanical anisotropy of the cells may be attributed to the aligned F-actin. 

One limitation of this study is that the modulus is calculated based on Hertzian contact theory which assumes material isotropy. In the next step of the study, we plan to apply an analytical model to directly extract the anisotropic mechanical parameters from the indentation curves. We expect this method will facilitate our understanding of cell mechanics and mechanobiology. 

Acknowledgements:  

This work was supported by the NSF (CMMI 1761432) and ARMI BiofabUSA (T0137).

Influence of Extracellular Matrix Stiffness on Exosome Production from PDAC Line Bxpc-3 

• Presenting Author: Pavlina Adhami (REU Participant) 
• Primary Investigator: Catherine F. Whittington, PhD
• Co-Author: Athenia E. Jones
  

Introduction:   

Pancreatic Ducal Adenocarcinoma Cancer (PDAC) remains one of the deadliest cancers in the United States due to poor early detection and high metastasis rates, resulting in a five-year survival rate of 12.8%. Therefore, developing new diagnostic biomarkers is critical for improving patients’ outcomes. As PDAC advances, the extracellular matrix (ECM) stiffens due to excess type I collagen deposition and increased ECM crosslinking that influences tumor cell proliferation and changes in intercellular communication, including exosome signaling.  

Exosomes are small extracellular nanovesicles that contain biomolecules (e.g. lipids, RNA, metabolites, etc.) that facilitate intercellular signaling under normal and disease conditions. Recently, exosomes have emerged as potential biomarkers for diagnosing PDAC and understanding how tumor and surrounding stromal cells interact. Therefore, it is important to understand how exosomes change as the tumor ECM stiffens during PDAC progression.  

This study aims to determine if ECM stiffness alters exosome yield from PDAC cells cultured in vitro. Epithelial BxPC-3 cells (PDAC cell line) were seeded onto methacrylated type I collagen hydrogels capable of on-demand stiffening via photo-crosslinking, and exosomes were isolated from cell media. We hypothesized that stiffer photo-crosslinked collagen gels would yield a higher exosome concentration than lower-stiffness collagen gels. 

Materials and Methods:   

Methacrylated type I collagen (PhotoCol®, Advanced Biomatrix) was combined with 8% v/v manufacturer-supplied neutralization solution and 2% v/v LAP (Lithium phenyl (2,4,6-trimethyl benzoyl phosphonate) photoinitiator and incubated at 37°C/5% CO2 for 30 minutes for gel formation. Post gelation, the high stiffness group was photo-crosslinked with 405 nm light for 3 minutes while low stiffness gels were left uncrosslinked. BxPC-3 cells (ATCC) were seeded on top of the hydrogel (30,000 cells/mL) and cultured for 3 days. At approximately 80% cell confluency, samples were washed and switched to fetal bovine serum (FBS) free media in preparation for exosome collection. After 24 hours, media containing exosomes was collected, mixed with Total Exosome Isolation Reagent (ThermoFisher Scientific), and incubated overnight at 4°C to promote precipitation and concentration of exosomes in cell culture media. Samples were centrifuged, and isolated exosomes were resuspended in PBS for analysis with a Bradford Assay to determine exosome concentration. Hydrogel samples seeded with cells were fixed with 4% formaldehyde and stained with phalloidin (F-actin), DAPI (4′,6-diamidino-2-phenylindole; nucleus), and Ki67 (proliferation). A study was also performed to compare exosome concentrations in samples collected from culture media with and without FBS to identify potential interference from exosomes already present in FBS. 

Results, Conclusions, and Discussions:  

Results from the Bradford Assay show a higher concentration of exosomes (M=843.75 ± 192.69 µg/mL) on stiffer photo-crosslinked gels (~3 kPa) compared to exosome concentration (605.7, SD ± 291.01 µg/mL) from BxPC-3 on low-stiffness gels (~0.75 kPa) (Figure 1A). Moreover, culture media supplemented with FBS exhibited increased concentrations over FBS-free media in the Bradford Assays (Figure 1B).  

We also observed that BxPC-3 cultured on high-stiffness collagen gels grew to confluency faster than cells on low-stiffness gels. Fluorescence staining of the actin cytoskeleton and cell nucleus shows that cells grown on high-stiffness collagen are spread out and grow as multicellular clusters while cells on low-stiffness gels are rounded and grow as individual cells (Figure 2A). The confluency on high-stiffness gels exceeded 100% after 5 days of growth. During the same time interval, low-stiffness collagen gels had a lower cell confluency of 75-80%. This finding also aligned with proliferation results using Ki-67 that that showed reduced Ki67 expression and cell proliferation in low-stiffness gels compared to high-stiffness gels (Figure 2B). 

Future studies will involve performing a Calcein-AM/ethidium homodimer viability assay on BxPC-3 cells grown on low and high-stiffness collagen gels and further characterizing exosome production from cells on these substrates (size and exosome content). Cancer is heterogeneous with different cell lines from the same pathology displaying different behaviors, therefore we will use our system to expose different PDAC cell lines to varying ECM stiffness to evaluate how exosome production may change from these cells. Additionally, we will evaluate exosome production from normal pancreatic epithelial cells, to determine if matrix stiffness alters exosome yield or cargo to promote cancer initiation. 

Acknowledgements: 

Thank you to Brian Ruliffson (PhD candidate) and Stephen Larson (PhD student) for their invaluable support.   This work was funded by the Pancreatic Cancer Action Network Career Development Award (831461).  This work was supported by NSF REU grant EEC2150076. 

Investigating the Cooperation Between Endothelin B Receptor and Heparan Sulfate in Regulating Endothelin-1 Expression Using qPCR 

• Presenting Author: Christina I. Kyriacou: University of Rochester (REU Participant) 
• Last Author: Solomon Mensah, PhD
• Co-Author: Camden Holm
  

Introduction:   

The endothelial glycocalyx (eGCX) is the luminal layer surrounding the endothelial cell membrane made up of glycoproteins and glycolipids. Its fundamental roles include modulating vascular permeability, mediating vascular tone, and mechanotransduction. Mechanotransduction is the conversion of mechanical stimuli, such as shear stress from blood flow, into biochemical signals by transmitting the force to the cytoskeleton where it can influence biochemical processes. Endothelin-1 (ET-1) is a signaling molecule that can cause vasoconstriction through its binding to the endothelin B receptor (ETB) found on the endothelial cell membrane. ET-1 expression is shear-dependent and may be regulated through cytoskeletal changes, implicating potential cooperation with HS. Heparan sulfate (HS) is a glycosaminoglycan that makes up 50% – 90% of the eGCX and is a known regulator of mechanotransduction in the cell. This work aims to understand the relationship between ETB and HS. We hypothesize that there is a cooperative relationship between ETB and HS that may affect ET-1 expression. 

Materials and Methods:   

To investigate this cooperation, human lung microvascular endothelial cells (HLMVECs) were subjected to several shear stress magnitudes, 5, 15, and 25 dynes/cm2, and the expression changes of ET-1, ETB, and HS were assessed using real time reverse transcription quantitative polymerase chain reaction (RT-qPCR). Cells were cultured and seeded on glass coverslips. The fluid flow experiments were conducted using a parallel-plate flow chamber in a sterile environment to impart shear stress to the luminal surface of the cells on the coverslip. In addition to a static sample that experienced no exposure to fluid flow, other HLMVECs were exposed to 5, 15, and 25 dynes/cm^2 of shear stress for 30 minutes. SYBR Green Fast Advanced Cells-toCT™ Kit from ThermoFisher Scientific was used to perform RT-qPCR for ET-1, ETB, and HS. Primers were designed using Primer-BLAST software from NCBI. Preliminary qPCR results were analyzed using the 2^-ΔΔCt method. 

Results, Conclusions, and Discussions:  

Primer design yielded few successful primers. Primers that functioned sufficiently well were used to gather preliminary data. Fold change analysis using the 2^-ΔΔCt method showed HS expression to be decreased for all the flow rates with the lowest at 15 dynes/cm^2. ET-1 expression increased significantly at the 5 dynes/cm^2 flow rate, and insignificantly increased at 15 dynes/cm^2 and 25 dynes/cm^2. ETB expression also increased significantly at 5 dynes/cm^2 and decreased at 15 dynes/cm^2. The ETB expression at 25 dynes/cm^2 was insignificant. It is important to note that ET-1 and ETB followed a similar pattern. These results are only from one round of experiments (n=1), and further experimentation is ongoing to obtain more results before any claims can be made.  

Challenges in RT-qPCR protocol primer design created several delays in collection gene expression data. The primers used for the preliminary results have at this time not been fully validated. While ET-1 gene expression after exposure to shear stress aligns with known expression profiles, HS gene expression contradicts what is currently known about HS expression under shear stress. This is likely due to insufficient RT-qPCR primer development and is the current focus of our research. Continued refinement of primer choices will allow for a proper understanding of changes in gene expression of ET-1, ETB, and HS, and will allow us to begin to understand the potential cooperation between ETB and HS causing changes in ET-1 expression. Defining this cooperation may lead to a greater understanding of cardiovascular diseases such as hypertension. 

Acknowledgements:   

This work was supported by NSF REU grant EEC2150076. 

IQGAP1 Regulates HeLa Cell Mechanosensing, Shape and Motility 

• Presenting Author: Alberto Salgado, University of Texas: Rio Grande Valley (REU Participant) 
• Last Author: Qi Wen, PhD
• Co-Author: Pengbo Wang
  

Introduction:   

The IQ Motif Containing GTPase Activating Protein 1 (IQGAP1) is a scaffold protein involved in cellular activities such as invasion, migration, and morphology. It is ubiquitously expressed in multiple cell lines across the human body and is a necessity for normal cell function. In this study, we used HeLa cells, an immortal human epithelial cancer cell line derived from cervical cancer, which on average, shows a high expression of IQGAP1 protein compared to other cells in the cervix. Cancer cells can sense the stiffness adjustments of their environment, in this case the PAA gel, and can adapt their behavior accordingly. By knocking down IQGAP1 in HeLa cells, we compared their migration speed and traction force with non-knockdown cells to understand cancer spread and aid in metastasis prevention research. Previous data showed a significant decrease in traction force with IQGAP1 knockdown, leading us to hypothesize a similar reduction in migration speed. 

Materials and Methods:   

HeLa cells were cultured in a controlled environment until it reached a confluency of 80-90%. Fluorescent beads of 0.1 um in diameter were placed on a coverslip. The cells were plated on polyacrylamide gels (PAA) of different stiffness to mimic human cervix tissue. The gel stiffness prepared were: 2, 7.5, 13, 20, and 40 kPa. The gel substrate was then placed on the bead coated slip before being placed on a glutaraldehyde slide inside a 60mm petri dish. After the gels polymerized, they were coated in Sulfo-SANPAH, a chemical used for cell-surface protein crosslinking that will allow the IQGAP1 protein to link together. The gel is also coated in collagen (9.37mg/mL) to support cell adhesion and proliferation. Before plating HeLa cells, they were subjected to IQGAP1 suppression by using Lipofectamine 3000. Phase contrast images were taken of single cells to compare cell area, aspect ratio, and circularity using free hand tool in ImageJ. Traction force microscopy was done by taking fluorescent images of beads underneath the HeLa cell attached to the gel. To detach the cells from the gel surface, trypsin is added. Fluorescent images were taken again in the same location to check the displacement of the beads. A MATLAB script was used to measure the displacement of the fluorescent beads before and after cells were released using trypsin. 

Results, Conclusions, and Discussions:  

Cell imaging shows that IQGAP1 knockdown has steady cell area across the 5 gel stiffness with low fluctuations. Compared to non-knocked IQGAP1 HeLa cells, their area increased until 13kPa when it reached a plateau and maintained a similar area for 20 and 40 kPa. For the aspect ratio, the opposite is shown. IQGAP1 knockdown starts low and increases as the stiffness of the gel increases. Non-knockdown HeLa cells maintain their average aspect ratio of ~2 with a slight decrease in the 13kPa gel. Circularity of IQGAP1 knockdown starts high at 2kPa and gradually decreases as gel stiffness increases. HeLa cells maintained a similar circularity of ~0.4. The maximum traction stress of HaLa cells increased as the gel stiffness increased. However, it was the opposite for IQGAP1 knockdown. IQGAP1 knockdown started with slightly greater traction stress than HeLa cells in the 2kpa gel but decreased as the stiffness increased.  

The HeLa cells shows that it can sense the stiffness of the gel and gradually increase their area while IQGAP1 knockdown stays at a similar area. This indicates that suppressing IQGAP1 expression decreases the cells mechanosensing abilities. The maximum traction stress of IQGAP1 knockdown is relatively low compared to that of HeLa cells with a normal expression of IQGAP1. Future research on IQGAP1 will be done on visualizing F-Acting staining Furthermore, IQGAP1 knockdown reduces the cells’ mechanosensing ability, significantly decreasing traction force. This highlights IQGAP1’s crucial role in cells’ response to substrate stiffness, providing insights for future cancer metastasis research. 

LHRH-Prodigiosin Conjugates as Targeted Therapeutic Agents for Triple-Negative Breast Cancer Therapy 

• Presenter: Ali Salifu, PhD (Boston College)
• Co-Author: John Obayemi, PhD
  

Introduction:  

Triple-negative breast cancer (TNBC) accounts for about 10-15% of all breast cancers, affecting about 13 in every 100,000 American females. It is aggressive and challenging to treat due to its lack of estrogen, progesterone, and human epidermal growth factor receptor 2 (HER2) receptors typically used to detect and treat breast cancer. Current TNBC treatments (chemotherapy, radiation therapy, surgery) lack specificity and are associated with high tumor recurrence rates. Recent efforts have been made to develop targeted therapies that interact with overexpressed receptors on TNBC cells, such as luteinizing hormone-releasing hormone (LHRH), to selectively kill cancer cells. In prior work, we have demonstrated the anticancer properties of prodigiosin, a bacterial pigment extracted from Serratia marcescens, towards TNBC cells. In this work, we present a combinatorial approach to TNBC treatment that (i) utilizes LHRH-prodigiosin conjugates as targeted chemotherapeutic agents, (ii) encapsulates LHRH-prodigiosin conjugates into poly(lactic-co-glycolic acid)/polyethylene glycol (PLGA/PEG) microparticles for sustained-release drug delivery, and (iii) incorporates LHRH-prodigiosin conjugates and gold nanoparticles into PLGA core-shell microparticles for chemo-photothermal TNBC therapy. The ultimate goals of this work are to potentiate the tumor-killing properties of prodigiosin toward TNBC cells and utilize laser-assisted plasmonic heating of gold nanoparticles to regulate the release of prodigiosin and deliver hyperthermia to suppress TNBC cell growth, kill residual cancer cells, and limit tumor recurrence. 

Materials and Methods:  

Prodigiosin was extracted from Serratia marcescens using our established protocols. It was then conjugated with [D-Lys6]LHRH peptide using EDC/NHS crosslinking and purified with a silica-loaded gel column chromatography system. Excess LHRH was removed using 3kDa Amicon Ultra-4 Centrifugal Filter Units. This was followed by characterization using Fourier-transform infrared spectroscopy (FTIR) and liquid chromatography/mass spectrometry (LC/MS). The LHRH-prodigiosin conjugates (LPC) were subsequently encapsulated into PLGA/PEG microparticles using a single emulsion process. Finally, the LPC drugs were incorporated into PLGA core-shell microparticles containing plasmonic gold nanoparticles. These drug delivery systems were characterized using scanning electron microscopy, transmission electron microscopy, dynamic light scattering, atomic force microscopy, and in vitro drug release studies in PBS at 37°C. The in vitro viability of a TNBC cell line (MDA-MB-231) was assessed against the drug formulations using alamar blue assay, whereas the in vivo effects of the released LPC drugs on TNBC xenograft tumors were studied in 4–6-week-old female athymic nude mice. The LPC and free drugs were administered intravenously through the tail vein of the mice induced with TNBC xenograft tumors at a dose of 10 mg/kg body weight. There were three treatment groups corresponding to 14-day, 21-day, and 28-day-old tumors and two drug injections at weekly intervals. Tumor volumes were monitored daily post-injections. 

Results, Conclusions, and Discussions:  

The goal was to develop a combinatorial approach to targeted TNBC therapy that utilizes LHRH-prodigiosin conjugates (LPC) in various drug formulations. The synthesized LPC drugs were confirmed through FTIR and LC/MS analyses. In vitro cell viability results indicated that LPC drugs inhibited TNBC cell growth more than the free drugs after 72 h (Fig 1 a,b). The LPC drugs were also more effective in shrinking tumors in vivo, where they eliminated 14-day-old tumors and shrunk 21-day and 28-day-old tumors by approx. 90% vs. the free drugs (Fig 2). Immunofluorescence staining and RT-qPCR analysis showed that LHRH receptors were initially expressed (Fig 3 b,d) and knocked-down after siRNA transfection (Fig 3 c,e,f). Afterward, there were no differences in drug-induced cytotoxicity between TNBC cells with or without knock-down receptors (Fig 1c), indicating that the LPC drug-LHRH receptor attachment allowed for targeted drug delivery to the cancer cells. This was also confirmed by AFM adhesion force measurements between drugs and xenograft tumors, where the LPC drugs exhibited greater than 2-fold adhesion forces than the free drugs. 

After showing that LPC drugs selectively target TNBC, we incorporated them into PLGA/PEG microparticles to prolong their release. Drug release studies indicated that ~60% of the drugs were released within 40 days (Fig 4), and the released LPC drugs inhibited TNBC cell growth more than the free drugs. The microparticles were then deployed in vivo to kill residual cancer cells, following the resection of a 4-week-old xenograft tumor, to suppress locoregional tumor recurrence in athymic nude mice. The microparticles were placed in the resected tumor region, and after 8 weeks, there was no tumor recurrence, whereas the control mice had tumor recurrence and metastasis. Finally, LPC and gold nanoparticle-loaded core-shell microparticles (Fig 5 a,b,c) released ~40% of drugs within 30 days, significantly inhibiting cell growth. Treatment with a laser beam (808 nm) generated photothermal heating that increased drug release by ~ 1.5 times (Fig 5d), which in turn increased TNBC inhibition. Ultimately, we anticipate that our combinatorial approach will create a new standard of care for TNBC therapy that combines targeted drug delivery and photothermal therapy. 

Lithium Carbonate Encapsulation in a Liposomal Delivery Vehicle for Bipolar Medication Usage During Pregnancy 

• Presenting Author: Jon Balyeat: University of Connecticut (REU Participant) 
• Last Author: Christina M. Bailey-Hytholt, PhD (Chem Eng)
• Co-Author: Diana Alatalo, PhD, CLC, Alexandra Harrison (Chem Eng)
  

Introduction:   

Bipolar disorder is defined as a mental illness which can lead to sudden mood shifts[1]. Lithium carbonate (Li2CO3) is a medication used for bipolar disorder shown to offset symptoms, prevent suicide, and improve mental wellbeing over time[2]. The exact neurobiological and chemical pathways of lithium treatment are still unknown; however, it is well understood that the lithium ions interfere with the metabolism of phosphatidylinositol[7]. Use of mood stabilizers, such as lithium, during pregnancy and postpartum have been shown to reduce relapse[2]; however, continuation also risks fetal malformations[3]. Lithium carbonate has been shown to cross the placental barrier between mother and fetus, which can result in various complications including cardiac malformations, Epstein Syndrome, or miscarriage[3]. For breastfeeding women, lithium ions can enter the breast milk due to the contents being highly dependent on blood and plasma composition[4]. There is a need to develop a controlled delivery vehicle for lithium treatment, and liposomes are one potential avenue. Liposomes are a simple nanoparticle which features 3 common components: Dipalmitoyl phosphatidylcholine (DOPC), cholesterol, and polyethylene glycol (PEG). In this work, liposomes were chosen due to their high loading efficiency and extensive use in previous drug delivery systems[5]. This study aims to test the viability of using liposomes to load and deliver lithium carbonate. 

Materials and Methods:   

Four liposome formulations and multilamellar variations were formulated to contain varying molar ratios of DOPC, PEG, and cholesterol (Figure 1A). Formulation 1 was based on a previous liposome study using lithium[6], and the additional formulations were iterated to investigate differences in the PEG and cholesterol content. A thin film hydration method followed by extrusion was used to develop the liposomes. Briefly, lipids stored in chloroform were dried under nitrogen gas and vacuumed to create a thin film. The thin lipid film was hydrated using nuclease-free water containing 60.7 mM of lithium carbonate solution. A set of 5 freeze-thaw-vortex cycles were conducted to establish multilamellar vesicles. Then, extrusion through a 100 nm polycarbonate membrane was performed. Finally, Amicon ultra centrifugal filters (100 kDa molecular weight cut-off) were used to separate unencapsulated lithium carbonate from the liposomes, and the flowthrough was collected for analysis. Physicochemical characterization of the extruded liposomes and multilamellar liposomes was performed using a Zetasizer to measure size, polydispersity index (PDI), and zeta potential. Encapsulation of lithium carbonate within the liposomes was measured using a conductivity meter. A standard curve was established with known lithium concentrations. Empty liposomes of formulations 1-3 were measured as a control to assess the lipid membrane effects on the conductivity readings. To assess rupturing the liposomes, Triton X-100 was added to the liposomes to destabilize the membrane and release any loaded lithium. 

Results, Conclusions, and Discussions:  

Multilamellar and extruded liposomes resulted in uniform size distributions for all conditions tested (Figure 1). All extruded formulations resulted in particles with smaller diameters and PDIs compared to multilamellar liposomes. After liposome characterization, conductivity was assessed to measure lithium loading. Using the conductivity meter, an initial standard curve with lithium carbonate (3.125 mg/L – 400 mg/L) resulted in an R2 value of 0.99. At a lithium carbonate concentration of 3.125 mg/L, the conductivity was determined to be 7-8 µS. Due to the dilutions necessary for the samples, this resulted in a need to produce a lower limit of detection. Next, the standard curve was produced with lithium carbonate at a lower concentration of 0.78125 mg/L, which resulted in a limit of detection of ~2.9 µS. The multilamellar liposome formulation 2 resulted in a conductivity reading ranging between 9.44 µS and 9.75 µS, which provides a proof-of-concept measurement of lithium loading within the conductivity standard curve . Using the multilamellar liposome formulation 2, a test to disrupt the liposomes using Triton X-100 was performed. Before the addition of Triton X-100, conductivity was measured at 9.76 µS, and after Triton, it was 12.64 µS, which indicates release of encapsulated lithium carbonate. Finally, an increase in conductivity was observed when measuring the lithium loaded liposomes kinetically over multiple days, which will be further investigated in future work. 

In this work, multilamellar and extruded liposomes were developed and characterized. These different liposome structures were assessed for their ability to load lithium carbonate. Conductivity measurements were optimized for lithium carbonate using this measurement technique. The multilamellar formulation 2 (DOPC:PEG:Chol at 62.5:5.0:32.5 mol%) resulted in a measurable conductivity within the standard curve produced, and results with Triton X-100 indicate initial promise of lithium carbonate loading. Further, increases in conductivity was observed over time, which motivated future assessment of the lithium carbonate encapsulation stability. In the future, increasing the concentration of lithium carbonate may provide additional insights into the loading and stability. Ultimately, this initial work provides a foundation for future testing to establish a lithium-loaded particle to improve controlled delivery for use during pregnancy and lactation. 

Mesoscale Modeling to Improve Brain Strain Accuracy: An Initial Study 

• Presenting Author: Nan Lin 
• Primary Investigator: Songbai Ji, PhD
• Co-Author: Wei Zhao
  

Introduction:  

Finite element (FE) models of the brain are widely used to study mechanisms of traumatic brain injury. While higher mesh density usually improves prediction accuracy, the simulation runtime would dramatically increase, which makes the model less feasible for application. In addition, it is also challenging to uniformly refine the mesh for a given brain injury model, due to the complex geometry. Instead of uniformly refining the global model mesh, a mesoscale model that refines mesh density only in a given region of interest (ROI) may provide a more practical solution. A few studies have developed mesoscale brain injury models. However, its accuracy relative to a finely meshed global model has not been carefully investigated. In this study, we aim to initially evaluate accuracy improvement of a mesoscale model originated from a coarse mesh model relative to a finely meshed global model. Experimental and simulation data from a simplified cylindrical gel physical model under axial rotation were used for model development (global and mesoscale), validation, and further parametric investigation. 

Materials and Methods:  

We developed an FE cylindrical model based on a previous study (Bayly et al., 2008) that used gel to simulate brain material. A rigid shell container was used to represent the skull. To first validate the FE model, the gel material was set to be that used in the reference to replicate a sudden z-axis rotational deceleration (Figure 1(c)).  

After successful model validation, the gel was evenly and horizontally separated into “gray” and “white” matter (Figure 1(a)). The gray matter material was the same as isotropic brain in WHIM V1 (Ji et al., 2015). The white matter was defined as 25% stiffer (Mao et al., 2013). A mesh convergence study was carried out, by using a 1 mm mesh resolution as the reference global model to compare peak maximal principal strain (MPS). A series of global models with mesh density ranging from 5 mm to 1.5 mm were created (Figure 1 (b)). The same validation rotational profile was simulated, and the model with 5 mm mesh resolution (>10% relative difference) was chosen as the coarse mesh size. A 20 mm3 cubic ROI was then re-meshed with a 1 mm mesh resolution to create the mesoscale model (Figure 1(a)) and simulated by the submodeling technique, where displacement obtained from coarse mesh model were used as input for the mesoscale model boundary nodes. A 10 mm3 central cube was chosen for strain comparisons using resampled MPS in the same region based on either the coarse or fine global, or mesoscale model. 

Results, Conclusions, and Discussions:  

Results and Discussion:  

The initial results show that at 40 ms when peak MPS occurred, the mesoscale model largely had similar MPS magnitude relative to the global model, which was considerably less than that of the finely meshed global model. However, it improved the MPS distribution similarity compared to the finely meshed global model, with the Pearson correlation coefficient, r, increased from 0.66 to 0.81 for the ROI.  

Compared to the coarse global model, the mesoscale model can improve the accuracy of strain distribution relative to fine mesh. However, because it uses the inherently less accurate displacement field from the coarse model as input, the accuracy of strain magnitude is limited, especially when there is substantial difference in mesh resolution between the coarse and fine global models. Future work should investigate how strain magnitude and distribution from the mesoscale model varies with a range of global model mesh density. This will help identify an optimal combination of coarse global model and mesoscale model to accurately approximate the fine global model in a given ROI. Ultimately, findings from the simplified cylindrical model will be extended to an actual brain injury model for the human brain. 

Conclusions:  

The study aims to investigate how a mesoscale model can be combined with a coarse global model for a more targeted simulation of mechanical responses for a given region of interest. When the global model mesh density is overly coarse, a mesoscale model can improve strain distribution pattern relative to a finely meshed global model. However, strain magnitude differences remain. 

Acknowledgements:  

Funding is provided by the National Science Foundation (NSF) under Grant No. 2114697. 

Musculoskeletal Modeling to Evaluate the Influence of Carbon Fiber Insoles on Gait Metabolic Cost 

• Presenting Author: Elizabeth K. Bowman 
• Last Author: Daniel J. Davis, PhD: University of Utah
  

Introduction:  

Humans frequently choose to walk at a speed that corresponds with the minimum metabolic energy expenditure per unit distance [1]. Factors such as age, muscular disorders, and neurological conditions, which affect approximately 10% of all U.S. adults [2], can increase the amount of energy necessary to complete daily activities. This increased energy expenditure can potentially lead to increased fatigue, injury risk, and healthcare costs [3,4,5], negatively affecting an individual’s quality of life. To address this, various assistive devices have been developed to reduce lower limb energy demands. Carbon fiber insoles are a low-cost assistive device that have the potential to reduce the whole-body metabolic cost of walking, but the contributions from individual muscles have yet to be elucidated [6,7]. Elucidating the effects of carbon fiber insoles on specific muscles would facilitate design improvements to better reduce injury and optimize lower limb energy expenditure. Musculoskeletal modeling and simulation informed by motion capture and direct muscle measurements can provide insights into individual-specific muscle mechanics which are not possible to assess in vivo [8]. The objective of this study was to generate computational musculoskeletal simulations from subject-specific models to examine the effects of footwear stiffness via carbon fiber insoles on lower-limb muscle energy expenditure. These simulations were compared with previous experimental data including lower-limb joint kinematics, soleus fascicle dynamics, and trends in whole-body energy expenditure. 

Materials and Methods:   

One participant (male, 1.69 m, 68.5 kg) who completed the experimental protocol by Ray & Takahashi [6] was chosen at random for initial pilot simulations. In brief, Ray & Takahashi [6] collected data from motion capture, cine B-mode ultrasound, indirect calorimetry, and electromyography data as participants walked on a treadmill at three different speeds (1.25, 1.75, and 2.0 m/s) with three different footwear stiffness conditions in a block randomized order. The footwear stiffness conditions consisted of a standardized shoe (Reebok RealFlex Train) with no carbon fiber insole (low stiffness), a 1.6 mm carbon fiber insole (medium stiffness), and a 3.2 mm carbon fiber insole (high stiffness). For this analysis, one speed (1.25 m/s; a typical walking speed) was examined. A lower extremity model (OpenSim’s gait2354 model [8]) was first scaled to align with participant anthropometry and inertial properties. To represent the stiffening effect of the carbon fiber insoles, a torsional spring was added to the model’s metatarsophalangeal (MTP) joint with values obtained from 3-point bending tests [6]. Inverse kinematics was then employed to estimate model joint angles from measured marker trajectories. Finally, residual reduction and computed muscle control were used to solve for muscle control values that produced consistent kinematic and kinetic results (Figure 1) [8,9]. Whole-body and muscle-specific metabolic costs were calculated and summed across each joint for the simulated stride (biarticular muscles were partitioned based on the ratio of the moment arms at each joint [10]). 

Results, Conclusions, and Discussions:  

The power fluctuations about the experimental distal foot and simulated MTP joint showed similar trends during the stance phase (Figure 1). The net work performed by the distal foot structures in vivo with increasing footwear stiffness varied in a similar pattern as the net work performed about the MTP joint with increasing torsional spring stiffness: -6.7, -7.6, and -0.9 J compared with -8.4, -13.9, and -2.7 J, respectively. The stance-averaged soleus fascicle shortening velocity decreased with increasing stiffness in both the experimental and simulated data: 23.52% (med) and 46.97% (high) relative to the low stiffness condition experimentally, and 36.07% (med) and 38.62% (high) relative to the low stiffness condition in the simulation, respectively. Additionally, the experimental and simulated metabolic cost of transport varied similarly, with greater relative decreases (1.29% and 2.63%, respectively) in the medium stiffness condition compared with the higher stiffness condition (0.32% and 2.44%, respectively) (Figure 2). Simulated muscles crossing the ankle joint consumed less energy with increasing torsional spring stiffness (Figure 3).  

Similarities across the distal foot mechanical energetics and soleus fascicle shortening velocity trends between experimental and simulated data provide preliminary validation for modeling the muscle-specific metabolic costs that are difficult to determine in vivo. Specifically, the decrease in energy consumed by the ankle joint musculature with increasing torsional spring stiffness agrees well with bioenergetic modeling of experimental data of footwear stiffness modifications [11]. The main limitation of the model is the lack of anatomical markers to define specific joints and body segments. For example, no trunk markers were used in the experimental protocol. Therefore, the model’s pelvis was constrained to produce an upright trunk posture which may have altered lower limb gait dynamics. Additionally, only one marker distal to the MTP joint was used experimentally, decreasing the confidence in that joint’s kinematics and kinetics. This initial modelling framework requires further refinement but has the potential to be used to better understand individual muscle contributions to whole-body metabolic cost. This improved understanding can help to design future assistive devices aimed at reducing the energy requirements of those with mobility impairment. 

Acknowledgements:  

This work was supported by SPUR from the Office of Undergraduate Research at the University of Utah, NIH R01AR081287 awarded to KZT, and NIH T32TR004394 awarded to DJD. The authors would like to acknowledge Samuel F. Ray for collecting the experimental data associated with this study.

PDAC Cell Lines Display Differential Responses to Obese Adipocyte Signaling Under Fibrotic Matrix Conditions In Vitro 

• Presenting Author: Athenia E. Jones
• Primary Investigator: Catherine F. Whittington, PhD
  

Introduction: 

Pancreatic ductal adenocarcinoma (PDAC) is set to become the third deadliest cancer by 2030, and obesity is a significant risk factor. Knowledge is limited on how obesity sustains PDAC progression but shared fibrotic and inflammatory features of each condition’s extracellular matrix (ECM) and microenvironment could provide insight into obesity’s influence. In both cases, increased collagen deposition and crosslinking raise ECM stiffness and contribute to dysregulated cell phenotypes. Fibrosis creates a complex microarchitecture that forces cells to alter their migration/invasion modes that can result in increased malignancy and chemoresistance, highlighting the need to include fibrotic elements in in vitro models.  

Our project objective is to investigate how a fibrotic microenvironment impacts PDAC and adipose cell phenotype and alters paracrine signaling that involves cytokines, growth factors, and exosomes. We hypothesize that initial cell phenotype will differentially influence PDAC cell migration mode in response to adipocyte signaling and ECM stiffness. To model fibrotic ECM conditions, we used a methacrylated type I collagen hydrogel capable of photo-crosslinking to expose PDAC cells to stiffness levels representing either normal or PDAC tissue. Encapsulated PDAC cells were exposed to adipocyte conditioned media and assessed for phenotypic changes to determine if ECM stiffness influences how PDAC cells respond to adipocyte signaling molecules. 

Materials and Methods:  

Adipocytes were differentiated from adipose-derived human mesenchymal stem cells with an adipocyte differentiation toolkit (ATCC). Lean adipocytes were differentiated on tissue culture polystyrene and obese phenotype was induced by differentiating cells in a 6 mg/ml type I collagen hydrogel (TeloCol-6, Advanced Biomatrix). Once adipocytes reached maturity (≥ 15 days), conditioned media (CM) was collected. PANC-1 and BxPC-3 PDAC cell lines (ATCC) were maintained in their standard growth media (DMEM or RPMI, 10% FBS, 1x Pen-Strep). Tunable stiffness hydrogels were created by combining 8 mg/ml methacrylated type I collagen (PhotoCol, Advanced Biomatrix) with 2% LAP (lithium phenyl-2,4,6-trimethylbenzoylphosphinate) photoinitiator. PANC-1 and BxPC-3 cells were encapsulated in hydrogels at low (~0.75kPa, uncrosslinked) or high (~3 kPa, crosslinked for 3 minutes with 405 nm light) stiffness that represent normal and pathological tissue stiffness Samples were maintained in standard growth media for 24 hours before exposure to 75% v/v adipocyte CM for an additional 48 hrs. Samples were fixed to visualize F-actin (Phalloidin), epithelial-to-mesenchymal transition (Vimentin), nuclei (DAPI). All samples were imaged on a Keyence BZ-X810 fluorescence microscope (20x) to obtain max projection of 3D z-stacks (100-200 µm stacks; 8 µm/slice). In parallel, adipocyte conditioned media was characterized with a Human Obesity Array C1 (RayBiotech) to detect the presence of key adipokines. Densiometry was used to extract spot intensity which was used to calculate relative protein expression using a manufacturer supplied program. A student’s t-test (p < 0.05) was used to test for significance. 

Results, Conclusions, and Discussions:  

We characterized conditioned media produced by lean and obese adipocytes (Figure 1). Obese adipocytes produced significantly higher levels of obesity-associated proteins (ENA-78, CRP, resistin) and cytokines IL-10 (1498±372.7) and IL-8 (17437±7491) compared to lean adipocytes (IL-10: 607.3±377.3; IL-8: 6689±1554). Both IL-8 and IL-10 increase EMT in PDAC cells.  

Exposure to obese CM increased expression of vimentin (mesenchymal marker) in PANC-1 cells (mesenchymal) cultured in low stiffness hydrogels (Figure 2A). These changes in vimentin expression were not seen in BxPC-3 cells (epithelial) under the same conditions (Figure 2B). Obese CM similarly increased vimentin expression in PANC-1 cells at high matrix stiffness and did not induce any changes in vimentin expression for BxPC-3 cells in high stiffness conditions. It is possible that IL-10 and IL-8 present in obese CM caused PANC-1 to increase their vimentin expression. PANC-1 cells are a mesenchymal PDAC cell line while BxPC-3 cells are epithelial. Our results suggest that initial cell phenotype influences response to obese adipocyte signaling. We additionally found that the same was true for PANC-1 in high stiffness gels without adipocyte signaling that exhibited amoeboid phenotype. This phenotype was not observed in BxPC-3 cells. Amoeboid morphology is part of the epithelial-to-mesenchymal transition cascade, wherein mesenchymal cells gain a more rounded morphology with small bleb formation in the actin cytoskeleton. Amoeboid migration is protease-independent and allows cells to navigate complex architecture. The crosslinks present in high stiffness PhotoCol® introduce degradation resistance, and the high collagen density of the gel creates cellular compaction. Both of which force mesenchymal cells into an amoeboid migration mode.  

These results contribute to the long-term goal of determining mechanisms of paracrine signaling for improved understanding of how PDAC risk factors contribute to disease progression. Our model contains human cell sources for both PDAC and obesity, to help address an existing gap in obesity-PDAC research, where studies traditionally use murine fat sources. Additionally, our model recapitulates complex environmental changes that occur during fibrosis and produces an amoeboid phenotype that warrants future study on the impact of migratory phenotype of progression and treatment response.

Acknowledgements:   

This work was supported by the Pancreatic Cancer Action Network Career Development Award (831461). 

Resilience of Brain Strain to Kinematic Filtering in Contact Sports 

• Presenting Author: Nan Lin
• Primary Investigator: Songbai Ji, PhD
  

Introduction:  

Head impact kinematics such as peak angular velocity (PAV) and peak angular acceleration (PAA) are widely used as metrics for brain injury risk. On-field head impact kinematics collected by sensors usually contain noise and artefacts that require signal filtering. However, PAV and PAA are sensitive to filtering methods, and different filters or filtering methods to the same signal could lead to disparate PAV or PAA values (Wu et al., 2016). This makes it challenging to compare different studies for consistent findings. Instead of applying a uniform filter, a recent study developed a novel filtering method called Head Exposure to Acceleration Database in Sport (HEADSport). HEADSport customizes filters for each impact kinematic profile based on its Power Spectrum Density (PSD) characteristics (Tierney et al., 2024). Compared to the other two commonly used filters, lowpass 200Hz and Channel Frequency Class (CFC) 180, HEADSport is more effective in suppressing artefacts. Nevertheless, few studies have investigated how data filtering affects brain strain most relevant to traumatic brain injury (TBI). The aim of this study, therefore, is to investigate how brain strain responses are influenced by different filtering methods. Specifically, we apply HEADSport, low-pass 200 Hz and CFC180 to a set of head impact kinematic profiles measured on the field and compare the resulting brain strains with those from unfiltered impacts. 

Materials and Methods:  

A total of 5694 on-field head impacts were previously collected to develop the HEADSport filtering method, and each impact was filtered by HEADSport, 200Hz and CFC 180. To quantify the relative difference in profile shape and magnitude, a given angular velocity profile and its filtered counterparts were normalized (relative to the largest component in the unfiltered profile) to compute piece-wise Euclidean distance. To limit computational cost, only impacts with an average relative difference exceeding the 90th percentile across the entire dataset were considered. They were further divided into three groups based on the piece-wise Euclidean distance: 90th – 95th, 95th – 99th, and above 99th percentile. For each group, 20 impacts were selected for simulation using the anisotropic Worcester Head Injury Model (WHIM) V1.0 (Zhao and Ji, 2019). For each selected impact, four simulations were conducted, using the unfiltered profile as well as those filtered by HEADSport, 200 Hz and CFC180 as model input. We compared the relative difference of PAA, PAV, as well as peak strain magnitude of the whole brain and that in the corpus callosum (CC) between filtered and the unfiltered counterparts. To further compare peak strain distributions in the whole brain and in CC, we also calculated linear regression slope, k, and Pearson correlation coefficient, r, between peak brain strains resulting from filtered impacts and those from the unfiltered counterpart. 

Results, Conclusions, and Discussions:  

Results:  

The current study found that peak brain strains, including relative magnitude and spatial distribution pattern, were much less sensitive for the whole brain and in CC than PAV and PAA. For most head impacts (e.g., average Euclidean distance < 95th percentile), peak brain strain, especially in the CC, were similar regardless of filter usage. Even for impacts >99th percentile, the difference in brain strain magnitude was much reduced than PAV and PAA (e.g., median relative difference of 9% and 3% for peak maximum principal strain of the whole brain and CC, respectively, vs. 47% and 90% for PAV and PAA, respectively, based on HEADSport filtering method). A larger relative difference in strain mostly occurred when significant artefact (i.e., aggressive sharp spike in angular velocity profile) existed in the unfiltered data. Figure 1 below shows two examples of peak brain strains and their corresponding impact resultant profiles. One led to similar brain strains from an impact that had an average Euclidean distance of 0.52 (90th – 95th percentile). The other one resulted in significant differences from the unfiltered impact had an average Euclidean distance of 1.25 (>99th percentile). The latter had a sharp spike in the angular velocity profile (duration < 3 ms; most likely due to unexpected movement (Tierney et al., 2024)). Peak strains in CC, nevertheless, largely maintained similarity.  

Discussion:  

The commonly used impact kinematics such as PAV and PAA are sensitive to filters. This study finds that peak brain strain, especially in the CC deep in the brain, is much less sensitive to filtering. This is due to the unique and inherent viscoelastic properties of the brain that acts as a low-pass filter, itself, to suppress high-frequency noise in impact kinematics. The effect is most evident when brain strain emanating from the surface reaches to the deep region.  

Conclusions:  

The study finds that brain strain is much less sensitive to filtering methods compared to typical impact kinematics commonly used for injury risk evaluation. Therefore, brain strain can serve as a common metric for TBI biomechanical studies to maximize relevance to the injury. 

Acknowledgements:  

Funding is provided by the National Science Foundation (NSF) under grant No. 2114697 (SJ). Impact data collection was funded by World Rugby. GT has received funding from Prevent Biometrics, Inc.  

The Role of Mechanical Properties in Graphene-pdms Based Nanocomposite Structure Opportunity for Enhancing Cell-surface Interactions 

• Presenting Author: John Obayemi, PhD
• Co-Author: Adianne Ramos-Delgado
  

Introduction:  

The integration of graphene-based nanocomposites with polydimethylsiloxane (PDMS) holds significant promise for enhancing cell-surface interactions in biomedical applications. This paper investigates the pivotal role of mechanical properties in graphene-PDMS nanocomposite structures and their potential to augment cell-surface interactions. Through a blend of experimental characterization and theoretical analysis, we delve into the mechanical behavior of these nanocomposites, focusing on how variations in graphene nanoparticle concentration, dispersion, and alignment impact their overall mechanical properties and structure. Our findings underscore the significance of these mechanical attributes in modulating cell interactions, adhesion, proliferation, and differentiation, thereby highlighting the potential of graphene-PDMS nanocomposites to create an optimal microenvironment for fostering enhanced cell-surface interactions. 

Materials and Methods:  

Through a combination of experimental characterization and theoretical analysis, the mechanical behavior and relative to the structure of graphene-PDMS nanocomposites is investigated, with a focus on how variations in graphene concentration, dispersion, and alignment influence the overall mechanical properties of the nanocomposites. Graphene-PDMS nanocomposites were prepared by dispersing graphene nanoparticles into PDMS using a unique mechanical mixing method – in situ shear exfoliation. The concentration, dispersion, and alignment of graphene nanoparticles were systematically varied to investigate their effects on structure and mechanical properties of the nanocomposite on disease and healthy breast cells. Experimental physicochemical and structural characterization of the Graphene-PDMS nanocomposite materials involved the use of FTIR, Raman spectroscopy, scanning electron microscopy (SEM), atomic force microscopy (AFM), while the mechanical properties was characterize using the Instron mechanical tester and nanoindentation in the presence of theoretical analysis employed composite models. Cell interaction, adhesion, proliferation, and differentiation were evaluated using cell culture assays on nanocomposite substrates. Statistical analysis was conducted to assess significance relative to the mechanical properties. 

Results, Conclusions, and Discussions: 

Experimental characterization revealed that variations in graphene concentration, dispersion, and alignment significantly influenced the mechanical properties of graphene-PDMS nanocomposites. Higher graphene content and improved dispersion led to enhanced mechanical strength and stiffness. Cell culture assays demonstrated improved cell interaction, adhesion, proliferation, and differentiation on graphene-PDMS nanocomposite substrates compared to pure PDMS materials. Our findings underscore the critical role of mechanical properties in modulating cell-surface interactions in graphene-PDMS nanocomposites. By optimizing graphene concentration and dispersion, tailored nanocomposite structures can be designed to create an optimal microenvironment for promoting cell interaction, adhesion, and proliferation. These insights provide valuable guidance for the development of graphene-PDMS nanocomposites tailored for specific biomedical applications, including tissue engineering, and the development of implantable drug delivery biomedical devices for the localized treatment of breast cancer.  

SmileScope: A Baby Bottle Device to Capture Intraoral Images of the Maxillary Palate 

• Presenting Author: Kenza Bezzat 
• Last Author: Diana Alatalo, PhD, CLC 
• Co-Author(s): Nicolas Loycano, Jacob McDonald, Samantha Turner, Haichong Zhang, PhD
  

Introduction:  

Cleft palate is a congenital malformation that results from the incomplete fusion of the hard and soft palate. Prior to operative reconstruction, infants undergo nasoalveolar molding (NAM) corrective therapy to non-surgically reposition the palatal tissue and reduce the severity of the cleft opening. To build the artificial palate, a dental impression of the cleft is obtained while holding the baby inverted and pressing the molding material into the maxillary palate and cleft. This process is uncomfortable, highly invasive, hazardous for an infant, and requires frequent visits over the treatment period. Currently NAM treatment is only accessible in higher-income nations where infants can routinely visit specialists.  

Modern visualization technologies like dental scanners offer less intrusive methods for intraoral examination, however these scanners primarily cater to adult patient care and at a high price point. The necessity for a pediatric oral assessment technique that is noninvasive, more cost-effective for underprivileged communities, and designed to ensure a positive patient experience inspired the development of SmileScope. 

Materials and Methods:  

SmileScope is comprised of 3 different subsystems and utilizes photogrammetry to digitize images into a 3D model: the delivery system, the imaging system, and the 3D modeling software. The delivery system consists of a modified bottle designed for feeding infants with a cleft palate. The elongated silicone nipple enables the camera to reach deeper into the oral cavity for imaging the soft palate while gently depressing the tongue. The use of a modified bottle nipple provides a familiar device for infant comfort, is designed to fit into the oral cavity of newborns, and is readily available in low-income communities. The imaging system utilizes a portable, water-resistant, dual lens borescope that can fit within the nipple. Multiple tests were conducted for image quality in dark and wet environments, image distance, field of view, and heat generation during use. A glass straw approved for food and drink consumption is used to stabilize the camera within the body of the bottle. The 3D modeling software Polycam was tested for accuracy, completeness, time, and accessibility.  

A preliminary design review was presented to Craniofacial Anomalies Clinic Team at UMass Memorial Children’s Medical Center, Worcester, Massachusetts, for feedback. The final design was tested on a healthy adult according to the WPI IRB protocol IRB-24-0380 (see Figures 1 & 2) and compared to a donated dental scan performed with a 3Shape Trios Scanner. 

Results, Conclusions, and Discussions:  

The SmileScope prototype successfully captured 130 photos in 5 minutes with a digital 3D model of the maxilla created in Polycam in 5 minutes (see Figure 3). The area imaged included the soft palate, hard palate, and posterior gums. The level of detail and structure size proved comparable to the 3Shape model. Pediatric and adult dental models were altered to simulate a cleft and then imaged with all relevant features captured. The size of the nipple remained the same which will allow for use in even premature infants.  

Clinician feedback highlighted the need for scanning technology in patient care even in high-income settings to facilitate transfer of care and accuracy of medical records. In terms of NAM creation, the gums also need imaging as a NAM attaches to the gums. Scanning of the gums was not included in the IRB-approved protocol, and therefore not tested at the time of abstract submission. SmileScope combines off-the-shelf hardware with a total price of $170 USD. A Polycam yearly subscription price of $100 USD. The total price is significantly lower than the closest competitor 3Shape Trios at $25,900 USD for the base model. 

Other features of the SmileScope include a rechargeable battery and a memory card. These features will enable scanning in low-resource areas to capture the images that can later be sent to trained clinicians for assembly with Polycam and review. Clinicians can then design the NAM that will either be shipped to the patient/local clinician or the design parameters can be sent electronically and built locally. Additionally, SmileScope can be used to assess NAM fit and progress between fittings remotely.  

In conclusion, SmileScope coupled with Polycam software can be used to image the palate of infants and create a 3D digital model thus replacing the need for traditional impression molding. The combined price is significantly lower than comparable devices for adults and images can be captured in low-resource settings by individuals with minimal training, which can expand access of care to low-income communities. 

Towards Wearable Forearm Ultrasound Based Gesture Recognition for Human-Robot Interfacing 

• Presenting Author: Alexis Murphy: University of Mount Union (REU Participant)
• Moderator: Haichong “Kai” Zhang, PhD 
  
• Co-Author(s): Keshav Bimbraw, MS, Alberty Enyedy, MS
  

Introduction:   

Our research focuses on developing a lightweight, cost-effective wearable device with non-invasive functionality that utilizes ultrasound probes to capture real-time forearm muscle images, enabling a robot to mimic precise and accurate hand gestures. The research focuses on the utilization of the WULPUS system (Wearable Ultra-Low-Power Ultrasound) as it is a cost-efficient and lightweight wearable device. In addition, the research utilized Verasonics Software for reliability of the WULPUS system. 

Materials and Methods:  

Two ways of studying hand gestures were used in this research. Both processes include placing the ultrasound probe on the forearm placed in anatomical position and studying the muscles that are used when performing two hand gestures: open and closed motions of the hand. The first method of study was the Verasonics Software. It used a single element ultrasound probe. The second method of study used the WULPUS system which utilized an ultrasound transducer.  

WULPUS consists of an Acquisition PCB designed for ultrasound measurement control and data management, complemented by a High-Voltage PCB for ultrasound transducer driving and signal multiplexing, thereby enhancing the devices modularity and functionality [1]. WULPUS not only has hardware components but also uses a Python library and Graphical User Interface (GUI) for logging and processing data then allowing the data to be visualized through graphical interfaces [2]. 

WULPUS allows for communication with robotic arms, enabling accurate replication of hand gestures such as performing open and closed movements. Post results show that the data acquired from Verasonics, established a baseline comparison of the muscle activity when performing different hand gestures. The baseline study used a single-element ultrasound probe that analyzed 500 frames per gesture. The WULPUS system provided a set of data called B-mode data for comparing the two types of hand gestures, open and closed. B-mode allowed for 2D imaging of the muscles in the forearm. This provided real-time imaging of the muscle in reference to the transducer. 

Results, Conclusions, and Discussions:  

In the WULPUS B-Mode data images, the closed hand gesture demonstrated that there is a higher intensity of signal within the forearm muscles. This is due to being positioned closer to the ultrasound probe. The open hand gesture reveals more gaps and fewer waveforms compared to the closed hand gesture. This indicates the further away the muscle activation is from the ultrasound probe, the less activation is shown. Using a neural network program through Verasonics, it showed a 90% accuracy of hand gesture identification. This confirms the data acquired from the WULPUS system. Future developments, within training the neural network to classify additional hand gestures, will refine the robot’s accuracy and precision when mimicking the controller’s arm movements. 

Acknowledgements: This work was supported and funded through the NSF (Grant: EEC2150076), ETH and WPI. I would also like to acknowledge Shang Gao, Yichuan Tang, and Keerthana Cheelamanthula for their support on the project. 

Tunable Autocrine Growth Factor Production in Skeletal Muscle Satellite Cells Using Lipid Nanoparticles for Sustainable Cellular Agriculture and Bioproduction 

• Presenting Author: Luke R. Perreault, PhD (Boston College)
• Last Author: Christina M. Bailey-Hytholt, PhD (Chem Eng)
• Co-Author: Daniel Zimmer (Chem Eng)
  

Introduction:  

The rising world population has induced unsustainable demand for growth on conventional animal protein production, already hampered by the proliferation of climate change. Cellular agriculture (CellAg) offers a method to alleviate this concern by shifting the burden of animal protein production to a cell-based bioreactor production model. However, muscle progenitor cells like bovine satellite cells (BSCs) must be cultured with growth factors such as fibroblast growth factor 2 (FGF2) to maintain proliferation, representing a major cost burden for CellAg scale-up. Recent research has established that genetically modifying BSCs to self-produce FGF2 could help reduce costs by eliminating the need for exogenous FGF2 in culture media.  

Here, we propose the use of lipid nanoparticle (LNP)-delivered plasmid DNA (pDNA) to induce transient expression of FGF2, as a method to control growth factor expression and production by satellite cells, and potentially enable tunable engineered autocrine signaling in unmodified primary cells. Specifically, we hypothesize that pDNA-induced FGF2 expression will maintain BSC stemness consistent with exogenous FGF2 supplementation in media. 

Materials and Methods:  

We produced two LNP formulations containing SM-102 and C12-200 ionizable lipids loaded with green fluorescent protein (GFP)-expressing pDNA. LNPs were analyzed for size, polydispersity index (PDI), zeta potential, encapsulation efficiency (EE). BSCs were cultured with one of the two LNP formulations (untreated cells were used as a control) within a 96-well plate format. Cells were imaged to assess GFP expression at 0, 8, 24, and 48 hours. BSCs in 6-well plates were incubated for 24 hours and evaluated for GFP expression using flow cytometry. 

Results, Conclusions, and Discussions:  

The two generated LNP formulations were produced as described above (N=3 technical replicates). SM-102 LNPs and C12-200 LNPs resulted in uniform average hydrodynamic diameters of 131 nm and 211 nm, respectively. Zeta potential values for SM-102 and C12-200 LNPs were 8.7 mV and 18 mV, respectively. Encapsulation efficiency of pDNA within the LNPs was ~100% for both formulations. We observed high GFP expression using both vehicles (>99% cell expression for both vehicles), but greater fluorescence intensity using SM-102 LNPs (Figure 1), which may suggest greater efficiency in transfection. Therefore, we concluded that pDNA can be effectively delivered into primary bovine satellite cells with LNPs, and that SM-102 may be the preferential vehicle for delivery of pDNA into BSCs. 

The present work shows that we are able to produce LNPs loaded with pDNA for use with BSCs, a notoriously difficult cell type to transfect. Flow cytometry evaluation indicates high transfection rates for both vehicles. Current work is targeting delivery of FGF2 pDNA to BSCs, and further refinement of the LNP formulations to decrease production cost and increase efficiency. Further, we are beginning work on a techno-economic analysis to evaluate the potential costs of this approach versus conventional growth factor use in culture media, and whether pDNA-induced autocrine signaling can cut costs for CellAg at scale. 

UVC Disinfection of a Low-Cost Bubble CPAP 

• Presenter: Kathryn Gurski: Saint Louis University (REU Participant) 
• Last Author: Dirk Albrecht, PhD
• Co-Author: Solomon Mensah, PhD
  

Introduction:  

Bubble continuous positive airway pressure (bCPAP) devices provide respiratory support to preterm infants to allow spontaneous breathing. However, many current bCPAP devices are expensive and unaffordable for low to middle income countries [1]. Therefore, there is a need for a low cost bCPAP device designed to be cost effective without compromising safety, effectiveness, or reliability. Therapeutic Innovations is working to fill this need with the AirBaby. Bubble CPAP devices like the AirBaby work by providing a heated, humidified air and oxygen blend using a humidifier water chamber. If non-sterile water is added to the humidifier chamber, the warm temperatures promote bacterial growth. Because the AirBaby humidifier is designed to be manually refilled and limited-resource hospitals may not have access to clean water, providing a way to disinfect the humidifier water while the device is running could help prevent infection and make the device safer for patients. Shortwave ultraviolet (UVC) light can be used for disinfection because it breaks the DNA and RNA strands of bacteria and viruses, preventing reproduction [2]. Many bulb-based UVC lights are used for disinfection in hospitals and other household devices. However, UVC LEDs provide an alternative for UVC light on a much smaller scale. This project investigated the intensity and bacterial disinfection capabilities of a UVC LED to determine the efficacy of integrating UVC light into the AirBaby’s humidifier chamber. 

Materials and Methods:  

To measure UVC intensity at different power levels and distances, a UVC LED was positioned 5 – 75 mm from a UVC meter and powered at 0.13 – 0.43W by limiting current and voltage. Once stable, intensity was recorded for each position. To verify the relationship between power and intensity, the LED was positioned 20 mm from the meter and power was adjusted from 0.05 to 0.4W. This procedure was repeated 3 times with 5 minutes of no power between trials. 

To assess bacterial disinfection by UVC light, 1.876mL of E. coli OP50 stock was added to 500mL of sterilized tap water and stirred at 330 rpm. After one minute, a 100uL control sample was collected on an agar plate. The 500mL solution was exposed to UVC light at 0.05 – 0.2W for 40 minutes, then a 100uL sample was collected. Plates were incubated at 37℃ for 48 hours then observed for bacterial colony formation. 

To assess disinfection by long UVC exposure, a humidifier filled with 500mL non-sterile tap water was connected to an air pump. After one minute of 4 L/min airflow, a 0.6 mL sample was collected on an agar plate. Then, the UVC LED was powered continuously at 0.28 – 0.3W or programmed to shine for 15 minutes per hour (25% duty cycle). Water samples were collected on agar plates at 2, 4, 6, and 24 hours of UVC exposure. After incubation at 37℃ for 2 – 3 days, bacterial colonies were counted. 

Results, Conclusions, and Discussions:  

UVC light intensity exponentially decreased with distance to the UVC LED regardless of power level, supporting the inverse square law. As power increased, UVC LED light intensity plateaued beginning at 0.2W. The iClevr Sanitizing Wand, a bulb-based UVC light marketed for household disinfection, has an intensity approximately 45x greater than a single UVC LED. A similar bulb-based UVC light used in hospitals has a reported disinfection time of 5 minutes [3]. Therefore, a single UVC LED would require approximately 4 hours to deliver the same dose of irradiation as its larger, bulb-based counterparts. However, disinfecting a liquid volume may impact the irradiation dose required compared to a rigid surface. When exposing E. coli in liquid to UVC, the number of visible colonies present after 40 minutes decreased in a dose dependent manner. Furthermore, continuously exposing the water in the AirBaby’s humidifier chamber to UVC reduced bacterial counts to zero by 24 hours exposure. Exposing the humidifier water to 15 minutes of UVC every hour also reduced the number of bacteria present at a slower rate, likely because any remaining bacteria is able to reproduce when the light is off. Ultimately, these results suggest that continuous, low-power UVC LED exposure of the CPAP humidifier chamber is effective in reducing bacterial growth. Further studies of other common, waterborne infectious organisms and viruses would help identify the full impact of exposing UVC light into the CPAP humidifier chamber.  

Vascular Endothelial Heparan Sulfate Cooperates with Endothelin B Receptor to Promote Endothelin-1 Synthesis 

• Presenting Author: Camden Holm 
• Primary Investigator: Solomon Mensah, PhD
  

Introduction:  

The endothelial glycocalyx (GCX), a layer of proteins coating the luminal surface of endothelial cells (ECs), plays a crucial role in vascular permeability and endothelial integrity. Its functions have only recently become the subject of in-depth research. Comprised of various transmembrane proteins connected to the cytoskeleton serving as mechanotransducers for cellular responses, the GCX can influence many biochemical activities. Diseases like atherosclerosis or hypertension can lead to damage to the GCX, adversely affecting vasculature.  

Endothelin-1 (ET-1) is a potent vasoconstrictor that plays a significant role in many cardiovascular-related conditions. It may function to regulate contraction and dilation of blood vessels in response to changes in blood flow. Elevated ET-1 has been noted in patients with moderate-to-severe hypertension and atherosclerosis, implicating ET-1 in the pathogenesis of these conditions. ET-1 is produced by ECs and secreted into the extracellular space to bind to either the endothelin B receptor (ETB) on ECs, stimulating vasodilation, or the endothelial A receptor on smooth muscle cells, stimulating vasoconstriction. Due to proximity and the related roles of GCX and ETB, it is hypothesized that ETB cooperates with the GCX to trigger the production and release of ET-1 via disruptions in the cell cytoskeleton. Reports suggest that the expression of ET-1 is shear stress magnitude and time-dependent and that the disruption of cytoskeletal structures could mediate the shear stress-induced production of ET-1. This cooperation between ETB, ET-1 and GCX is yet to be clarified, a question that needs further investigation, which is the goal of this project. 

Materials and Methods:  

To measure the expression of ETB on the endothelial cell surface, Human Lung Microvascular ECs (HLMVEC) were subjected to laminar flow in a parallel-plate flow chamber and expression changes of ET-1, ETB, and Heparan Sulfate (HS), a major component of the glycocalyx, were measured. The cells were exposed to fluid flow at controlled rates and durations to impart a shear stress of 5, 15, and 25 dynes/cm2 on the cells for 30 minutes. A static control not subjected to fluid flow was included. 

After exposure to flow, cells were fixed using 4% formaldehyde and immunostaining was performed to fluorescently tag ET-1, ETB, and HS. A laser-scanning confocal microscope was used to capture Z-stack images for quantification. The sum projection of these images was generated using ImageJ FIJI, and analysis of fluorescent intensity was performed in CellProfiler. Fluorescent intensity was calculated per-cell area, and a fold-change calculation was performed to compare the change in fluorescent intensity between the flow conditions and the static control. 

Results, Conclusions, and Discussions:  

Preliminary results show that ETB expression is reduced, and HS expression is slightly increased when exposed to fluid flow at 5, 15, and 25 dynes/cm2 for 30 minutes. ET-1 expression may be reduced in the presence of low shear stress (5 dynes/cm2), though there is variability in our preliminary data, but is increased under higher shear stress, 15 and 25 dynes/cm2. 

Long-term exposure to fluid flow is known to upregulate ETB, but little is known about ETB’s shear stress response to short-term exposure. HS is known to exhibit increased expression when exposed to shear stress – its natural condition in the vasculature. ET-1 expression has been shown to be increased during short-term exposure to low to high shear stress with a peak between 0.5-2 hours of exposure. Our preliminary results align with the literature, demonstrating the accuracy of our model. 

Ongoing experiments include further quantification of gene expression profiles using qPCR to confirm our results match the literature. The increased expression of HS demonstrates glycocalyx health under physiological conditions. We will begin to establish the relationship between ETB, ET-1 and the GCX by inducing GCX damage by cleaving HS with a Heparinase-III treatment prior to short-term flow exposure. This will clarify the relationship between the molecules by showing the response of ETB and ET-1 to GCX damage.  

During longer term exposure (6-12 hours), ET-1 expression has been demonstrated to decrease. Further experiments will include 12- and 24-hour flow exposure durations to study long-term shear stress exposure and western blotting for further gene expression profile quantification. 

The GCX is known to play a vital role in vascular health and function. Understanding this role is a recent development, and the scale of GCX influence in the body is not well understood. ET-1 and ETB also play a major role in vascular function. Based on our preliminary data on expression changes when exposed to shear stress, ET-1, ETB, and GCX may lead a cooperative response to changes in vascular condition.