Contents
Special Guest Speaker:
Vanessa Liu
L A U R E N M E I, K E N T D O N G
Characterization of a new target for tuberculosis therapy by identifying a gene responsible for resistance to two anti-tuberculosis compounds in Mycobacterium smegmatis
R A P H A E L K I R O U
A novel approach to using pseudospecies effects exhibited by dsDNA waterborne bacteriophages as an indicator for mutagenic contamination
P H I L I P K U T C H E R
CEP-1, p53 Homologue, Modulates Fat Metabolism and Longevity
N I N A U Z O I G W E
Noninvasive detection of reactive oxygen species using xanthine/uric acid to use as a quantitative measure of tumors
Y A S S M I N E L B A N N A
Crowding Vision
C L A I R E J I N
Identification of a Factor that Induces Prostate Specific Membrane Antigen (PSMA) Expression in Human Endothelial Cells
J U D Y L I U
Computational Analysis and Prediction of Post-Stroke Recovery from Baseline Brain Connectome Metrics
K I M B E R L Y H O
Usability Testing of a Web-Based Decision Aid for Breast Cancer Risk Assessment Among Multi-Ethnic Women
S H A Y A N H U D A C H O W D H U R Y
Regulation of Skeletal Fracture Angiogenesis by Immune-Derived Paracrine Secretions
C O N S T A N T I N E A T H A N I T I S
P H I L I P K U T C H E R
Special Guest Speaker:
Vanessa Liu
L A U R E N M E I, K E N T D O N G
Special Guest Speaker:
Vanessa Liu
HOW DID YOU GET INTERESTED IN SCIENCE AT STUY?
I was lucky to have a great mentor who worked me to improve my skills of asking questions and making inquiries about various scientific processes.
L A U R E N M E I, K E N T D O N G
CAN YOU ELABORATE ON HOW YOU BUILD UP A RELATIONSHIP WITH YOUR MENTOR?
Liu was a previous Intel Finalist and alumni of Stuyvesant High School, and later went on to attend Harvard University. She is currently focused on media and marketing.
Building up a relationship takes time and effort and I remember sitting down with my mentor and setting a goal for what I was going to do as well as asking if I could help him with his work.
CAN YOU DESCRIBE THE PROCESS OF FINDING YOUR LAB?
The main thing about finding a lab is to be broad in scope. This widens the possibility of getting into a lab that will be as close to your interest as possible. Highly specialized labs are uncommon and there may be only one or two in the entire country. Be persistent in your search for a lab, and take the initiative to build understanding with your potential mentor before them meeting them. Lastly, be genuinely interested and don’t give up if you get rejected from a few labs!
IF YOU CAN EVER REDO ANYTHING DURING YOUR RESEARCH EXPERIENCE, WHAT WILL YOU DO?
Surprisingly, there is nothing that I want to redo. Every mistakes I’ve made helped me to become a better researcher. It may hurt at first, but with determination, the harm can be reversed and you will only increase in understanding. However, one thing that I would like to take more advantage of during my research experience is talking to the grad students in my lab. I was too focused on my own work to pay attention to those of others around me -- who knows, maybe I would have found something interesting, or even better my own experiment!
IF YOU HAVE ALL THE MONEY IN THE WORLD TO ADVANCE A SCIENTIFIC FIELD, WHAT WILL IT BE?
I am fascinated by the human brain. Take for instance, the story of a young woman who predicted her own death during the pregnancy, and as a result was able to prevent herself from dying of blood-loss. (For the whole story, please read the book 37 Seconds by Stephanie Arnold -- 37 seconds refer to the time she was pronounced dead). There is so little we know about our own brain, and if we are to unlock its potential, we can do so much more.
Tuberculosis Therapy
Characterization of a new target for tuberculosis therapy by identifying a gene responsible for resistance to two anti-tuberculosis compounds in Mycobacterium smegmatis
R A P H A E L K I R O U
I N T R O D U C T I O N
Tuberculosis is the deadliest human infectious disease, killing 1.5 million people each year1. Although several drugs kill Mycobacterium tuberculosis (Mtb), resistant mutants have emerged as a global public health problem. In order to identify new drugs against tuberculosis, compound libraries have been screened for anti-tuberculosis activity. In this study, two similar anti-tuberculosis compounds, Compounds 1A and 1B, were tested against Mycobacterium smegmatis (Msm) as a proxy to Mtb. Strains resistant to the compounds were then studied for genetic mutations.
In this study, msm mycobacteria are used as a proxy to tuberculosis, due to it nonpathogenic nature and reproduction rate. Msm cultures that grew despite exposure to high concentrations of the drugs were then analyzed for gene mutations. Gene YMsm, was found to a carry a novel mutation in resistant Msm strains to Compound 1A. This data suggests that the two compounds target Gene YMsm in killing the cells and hopefully will facilitate the development of new, alternate, less harmful drugs targeting Gene YMtb.
Then, 420 μL isopropanol and 10 μL 3M sodium acetate were added to the samples. The content are centrifuged, supernatant discarded, and pellet lysed and air-dried. The DNA concentration was measured with a nanodrop machine and then diluted to 100 ng/μL.
Finally, the genes responsible for causing resistance to the compounds in mutants were targeted. Polymerase Chain Reaction (PCR) was used to amplify the two genes suspected of being mutated (Gene XMsm and Gene YMsm). The samples were split up into 5 PCR tubes so that a gradient of annealing temperatures could be tested.
A small amount of each reaction was loaded on an agarose gel (0.5 g agarose, 50 mL TAE, 6 μL ethidium bromide) to check if the PCR was successful. If it was, PCR purification was used on the remaining PCR product not used in the gel using a DNA purification kit. The concentration of the liquid in ng/μL was then measured for each solution using a nanodrop machine. Samples were sent to Macrogen USA along with primers for sequencing.
R E S U L T S
Possible further experiments to be done include testing of more Msm mutants for Gene YMsm, but also for Gene XMsm mutations. In addition, experiments could be done to confirm that the mutation in Gene YMsm is indeed the cause of resistance to Compound 1A. This study addresses a pressing need for new antibiotics against tuberculosis, a global public health problem. Combining drugs with anti-tuberculosis activity under both proliferating and stress conditions will hopefully lead to shorter and more efficient treatment regimens for tuberculosis.
M E T H O D S
R E S U L T S
The first part of the project entailed isolating a large number of Msm mutants resistant to two compounds (Compound 1A and Compound 1B), and Wild type Msm was plated onto 7H11 agar. The Minimum Inhibitory Concentration (MIC90) of wild type Msm is 0.75 μM for both Compounds 1A and 1B, based on previous experiments. This was used as a basis for making plates. Wild type Msm was streaked on all of the plates. After they were incubated for three days, the number of colonies that grew in each plate was counted to calculate the Frequency of Resistance (FOR).
56 candidate mutants were re-streaked onto new plates with the same concentrations as the parent plates to confirm their resistance. If there is resistance, they are grown in liquid culture containing glycerol, and a 0.02% concentration of tyloxapol. The mutants A3, B4, C1, D4, E1, F8, G4, and H2 were chosen for further testing based on abnormal colony morphology (Table 1). Next, the phenotypes of the eight principal mutants were tested on agar without any compound. Then, the small number of cells remaining on the pipette tip was punctured into the swarming agar (Figures 2A and 2B). Then, they were left to shake in the incubator. The OD580 of each culture was measured at several times between the start and end time (24 hours) (Figure 3).
Another phenotypic experiment entailed finding the Minimum Inhibitory Concentrations (MICs) of the eight principal mutants and wild type under Compound 1A. The MICs were found by using nine 96-well (8 X 12) plates for our mutants and wild type. The plates were shaken in the incubator for 48 hours. The results are shown in (Figures 4 and 5). In order to carry out genotypic analysis on our mutants, genomic DNA had to be extracted from the cells using a two-day chromosomal DNA preparation procedure.
The first step entailed making a TE Buffer solution.500 mL water, 0.788 g TRIS Hydrochloride, and 0.18612 g EDTA (Ethylenediaminetetraacetic acid disodium salt dihydrate). The solution was adjusted to a pH of 8. The solution is centrifuged, the supernatant discarded and pellet lysed using lysozyme, and then incubated at 37 celsius overnight. Cold chloroform were added to the cultures and vortexed for 2 times.
One of the mutants, B4, showed a clear difference in morphology compared to the other mutants and wild type when grown on two different 7H11 agar plates (Figures 1A and 1B). The colony looked flat; this stood in contrast to the other colonies, which looked ruffled (a characteristic stemming from the fact that Msm cells often grow on top of each other). This difference could signal slower growth and/or a cell wall defect. In liquid media, B4 did not show a significant difference in growth (Figure 3).
The mutant genotypes were analyzed through DNa extraction, PCR, and sequencing. One mutant, D4, was found to carry a mutation in Gene YMsm. The mutation was a single nucleotide substitution that resulted in a proline to leucine amino acid change (P76L) in the gene product (Figure 6). (Figure 8) shows differences in protein structures. In addition, mutant D4 showed complete resistance up to 100 μM of Compound 1A (Figure 4), while the other mutants were eventually inhibited.
In this study, 56 Msm mutants resistant to Compounds 1A and 1B were isolated. Based on colony morphology on 7H11 agar, eight mutants were chosen for further experiments. Out of these eight mutants, one of them (B4) showed a striking difference in colony morphology on 7H11 agar. Finally, the DNA of the eight mutants was extracted for use in PCR. PCR amplification and sequencing for Gene YMsm in Mutant D4 showed a nucleotide substitution.
The identification of the Gene YMsm mutation provides evidence that this gene’s product is probably the target of these anti-tuberculosis compounds. Furthermore, the predicted conformational change in the protein’s structure from the mutation in Gene YMsm demonstrates the potential functional significance of the mutation and supports the notion that this protein is the target.
A novel approach to using pseudospecies effects exhibited by dsDNA waterborne bacteriophages as an indicator for mutagenic contamination
P H I L I P K U T C H E R
CEP-1, p53 Homologue, Modulates Fat Metabolism and Longevity
P H I L I P K U T C H E R
A B S T R A C T
Although autophagy proteins and their use of lipids ensure quality control in metabolism, they are linked to aging and their loss is detrimental towards health and lifespan. However, lipid profiles of long-lived Caenorhabditis elegans suggest that lipids are beneficial towards lifespan and have been evolutionarily conserved. In this study, I sought to determine the role of autophagy genes associated with longevity, specifically the allele gk138 in the gene cep-1. Results indicate that cep-1 is not only a critical modulator of the lifespan of C. elegans, but the loss of function of cep-1 and atg-7 significantly increases the lipid droplet concentrations compared to wild-type animals.
Introduction
Autophagy is a pathway that maintains cellular homeostasis by degrading macromolecules and organelles in lysosomes in a process called macroautophagy. Lipids can also undergo degradation through the sequestration of autophagosomes and autophagy-related proteins, called ATGs, that fuse with lysosomes (3,5). This process, lipolysis, ensures control of metabolism and energy through the hydrolysis of triglycerides into free fatty acids. However, triglycerides have been linked to aging (Lapierre et al., 2011).
The lipids extracted from the offspring of the long-lived individuals contain a higher ratio of monounsaturated fatty acids (MUFAs) over polyunsaturated fatty acids (PUFAs) (4). These worms express increased levels of Δ9 enzymes. However, PUFAs are vulnerable to oxidation and promote aging, generating radical oxygen species (ROS).
In this study, I sought to further characterize the role of autophagy genes and their association with longevity. Both mammalian p53 and its nematode homolog cep-1 have been described as tonic inhibitors of autophagy (5). When deleted or inhibited, they have been known to regulate genotoxic stress to relieve metabolic stressors and organelle damage (6,7,8,9,10).
I decided to investigate the lipid droplet staining of C. elegans with this genotype to see the relationship between longevity and the induction of autophagy. I also conducted a lipid droplet staining of C. elegans with atg-7(bp411), a mutation in atg-7, an autophagy gene that encodes an E1 that is essential for lipid homeostasis. Results show that cep-1 is not only a critical modulator of the lifespan of C. elegans, but a loss of function in cep-1(gk138) resulted in a significant increase in lipid droplet concentrations when compared to the atg-7(bp411) and wild-type animals, showing that cep-1 plays a crucial role in regulating lipid droplet metabolism.
Figure 1.1: Staging of C. elegans. As seen to the left, this is the developmental staging progression of C. elegans. L4 hermaphrodites were used for all staining of the lipid droplets, and their lipid concentration was stained by Oil-Red-O. Once the worms were stained, they were imaged with an Axiocam at 16x magnification. To measure the Oil Red-O Staining intensity, ImageJ was used.
Results
Table 2.1 Oil Red-O Staining of Lipid Droplets (Representative Images of Each Strain)
Table 2.1 shows the Oil Red-O Staining of the lipid droplets of representatives from each of the strains. The animals with the allele atg-7 (bp411) have a visibly significant decrease in their lipid droplet concentrations in comparison to the presence of the droplets in the wild type.
Table 2.2: Relative Oil Red-O Intensity in cep-1, atg-7 and wild-type animals
Quantification reveals that cep-1(gk138) mutants have high lipid droplet storage, whereas atg-7(bp411) mutants have a decrease in Oil-Red-O staining.
Table 2.2 shows the means of the Oil Red-O Intensities found from 3 trials of staining. As seen above, the animals with the cep-1(gk138) mutation statistically had the higher amount of lipid droplet staining when compared to those with atg-7(bp411) and N2. Mutants with cep-1(gk138) were significantly different from those of atg-7(bp411) and wild type.
Conclusion
Mammalian p53 and its nematode homolog cep-1 act as inhibitors of autophagy, which means that their deletion stimulates a continuous state of it (5). We found that cep-1 mutants have an increase in lipids as determined by the staining, demonstrating a previously unknown interrelationship between autophagy and lipid metabolism in cep-1(gk138) mutants. In addition to the inhibition of atg-7, my results suggest that cep-1 normally regulates lipid droplet content, and that cep-1(gk138) mutants either synthesize more triglycerides or breakdown less.
We hypothesize that there may be an apparent decrease in lipophagy in these long-lived worms, raising the question of whether or not increased lipid droplet levels also contribute to or result from the longevity found in these mutants (5). The next step would be to create a double mutant that consists of a loss of function in both cep-1(gk138) and atg-7(bp411) and stain them with Oil Red-O to check whether the loss of cep-1 and atg-7 activity would impact the concentration of lipid droplets. With this information, our knowledge of p53 will be greatly enhanced. One can not only pinpoint what it does to impact longevity in C. elegans, but also what can be done to increase the lifespan of the human cell.
A C K N O W L E D G E M E N T S
I would first like to recognize the PI (principal investigator) of the laboratory, Dr. Alicia Melendez, an Associate Professor of Biology. She helped me to go in the right direction in terms of ideas for my research. I would also like to recognize Melissa Silvestrini and Nicholas Palmisano, who are PhD students in the Meléndez lab, pursuing their own research. They helped me to familiarize myself with the lab environment and know how to carry out the procedures.
Noninvasive detection of reactive oxygen species using xanthine/uric acid to use as a quantitative measure of tumors
Y A S S M I N E L B A N N A
A B S T R A C T
An emerging body of data indicates that cancer cells have more Reactive Oxygen Species, which would theoretically aid in the imaging of tumors to see which parts are more active and understand how quickly the tumor is growing. In this context, reactive oxygen species are a key focus of the research into the chemical mechanisms underlying the progression of the disease. The goal to detect reactive oxygen species has motivated the construction of a primary detection. This project utilized Xanthine Oxidase, an enzyme that generates reactive oxygen species, to see how quickly the rate at which the Xanthine is picked up by the cell and is converted in the Uric Acid and other reactive oxygen species. To summarize, we believe that xanthine oxidase is the best method for measuring the rate at which Xanthine is picked up by a cell and converted into Xanthine oxidase. Using ultraviolet visible spectroscopy, we determined that Xanthine was the most efficient metabolite and that cells do in fact pick up the Xanthine and change in their absorbency. We then believe that this would a useful indication of the metabolic qualities of cancer. This project's importance is that this would develop a more efficient and detailed measurement of cancer imaging.
I N T R O D U C T I O N
Magnetic resonance imaging is primarily a qualitative measure of tumors and the progression of cancer, yet it doesn’t show the activity of these tumor cells. Hyperpolarized Magnetic Resonance Imaging is unique in that it allows for imaging of metabolites, overstepping the flaw held by Magnetic Resonance Imaging. This new technology is based on the premise that cancer cells use and metabolize nutrients differently and often at a faster rate, and it is unique in that it can be used to decide whether to continue treatment or to try a new approach [1].
There are fundamental differences between Hyperpolarized Magnetic Resonance Imaging and radioactive contrast agents, although both provide similar images. Hyperpolarized Magnetic Resonance Imaging contrast agents do not use ionizing radiation, making them a useful tool in analyzing the progression of cancer without endangering the patient and allowing for simultaneous imaging of multiple metabolites as well. This is a clear benefit because they enable measurements of metabolism as opposed to the tracer uptake typically measured with PET scans. The advantages of Hyperpolarized Magnetic Resonance Imaging include its increasing sensitivity due to the amount of nuclear spin polarization, its imaging of hyperpolarized contrast agents while in a low magnetic field, opening up new opportunities for multi-modal imaging, and its low-cost due to there no longer being a need for high-field cryogenic magnets [2].
Using hyperpolarization techniques, the signal from a given number of nuclear spins can be raised more than 100,000 times, enabling the imaging of nuclei other than proteins and their molecular distribution in vivo to bevisualized in a relatively small window of time, and the mapping of molecular distribution a few seconds after injection, offering a more direct form of molecular imaging.
Despite these differences, hyperpolarized contrast agents have a significantly shorter lifetime, anywhere from 0.5-5 minutes in vivo. This makes it slightly more difficult to take the image; however, it also offers an opportunity to perform repeat scans within minutes because there won’t be any background signal [3].
Reactive Oxygen Species are byproducts of cellular metabolism produced during aerobic respiration in mitochondria (Figure 1) [4]. The superoxide produced during the aerobic respiration is converted to a second reactive oxygen species molecule, hydrogen peroxide, by superoxide dismutase. In the presence H2O2 ions, it can form hydroxyl radicals. These molecules can oxidize DNA, protein, and lipids, leading to negative effects and must be carefully controlled.
Cancer cells proliferate quickly in comparison to normal cells, and therefore require more energy to replicate DNA, resulting in more aerobic respiration, which in turn creates increased levels of reactive oxygen species, which have been observed during the replication of cancer cells [5]. These higher levels have been shown to inactivate tumor suppressors, such as phosphatase and tensin homolog (PTEN) [6]. Reactive oxygen species are a component of the killing response of immune cells to microbial invasion, ending up working to the benefits of cancers for multiple unintended reasons, specifically that reactive oxygen species interact with other things if there is too much DNA, leading to genome instability or the denature of proteins, furthering the progression of the cancer. An emerging body of data indicates that cancer cells have more reactive oxygen species, which would theoretically aid in the imaging of tumors to see which parts are more active and to understand how quickly the tumor is growing. In this context, reactive oxygen species are a key focus of the research into the chemical mechanisms underlying the progression of the disease.
C O N C L U S I O N
Our results demonstrated that Xanthine interacts best with superoxides in generating reactive oxygen species. These specific molecules would allow us to scavenge toxic species from cancer cells. This would generate a more effective metabolite imaging of sarcomas through hyperpolarized Magnetic Resonance Imaging, which would in turn be able to image the high concentrations of reactive oxygen species present within the cells.
The project’s implications can change the way cancer is assessed. By developing novel imaging probes for hyperpolarized Magnetic Resonance Imaging within cancer research, we can create a non-invasive, quantitative measure of tumors. This will increase our ability to provide more accurate prognoses and teach us more about cancer cells and their metabolic behavior.
Crowding Vision
C L A I R E J I N
ABSTRACT
The fovea is the centermost part of the retina, the area is responsible for our central, detailed vision. Crowding is the difficulty of perceiving an object surrounded by clutter. Foveal crowding can impact daily activities that involve identifying small objects in a cluttered background. It’s prominent in patients with amblyopia and may predict the slowness of reading of children with developmental dyslexia. Labs and clinics do not yet have the ability to measure foveal crowding. A new test is available to test it, however, involving identifying a digit.
INTRODUCTION
Crowding greatly limits visual perception. Due to crowding, a simple object, like a number, can only be recognized if clutter is a certain critical spacing away (Bouma, 1970; Levi, 2008; Pelli & Tillman, 2008).
METHODS
We utilized a new acuity test called Critical Spacing.m on the program MATLAB 2015b with Psychtoolbox 3.0.12 extensions on computers running OS X or Windows. There are 9 custom designed letters, in “Sloan” and “Pelli” fonts, which are used with flankers. The observer’s objective is to fixate at the cross hairs and identify it with flankers present. The letters only flash for a fraction of a second before they disappear. Correct answers are addressed with a “ding” sound. The acuity test measures the level of peripheral vision the observer has. The healthy level of critical spacing in the fovea is .05 degrees, but is higher in patients with amblyopia.
We needed a font that had a small width of the critical spacing we aimed to measure. The “Pelli” font has a 1:5 aspect ratio while the “Sloan” font has a 1:1 aspect ratio.
RESULTS
CONDITION 1
Spacing(deg) P fit P Trials
0.115 0.10 0.00 1
0.201 0.10 1.00 1
0.244 0.11 0.00 1
0.674 0.25 0.00 1
0.875 0.39 0.00 1
0.933 0.43 0.00 1
0.961 0.46 1.00 1
0.990 0.48 1.00 2
1.004 0.50 1.00 1
1.019 0.51 1.00 1
1.062 0.55 1.00 1
1.105 0.58 0.50 2
1.148 0.62 1.00 1
1.220 0.68 1.00 2
1.306 0.75 1.00 1
1.794 0.96 1.00 2
We uploaded data that we collected from two subjects into the KaleidaGraph and it output the polar graphs shown above.
DISCUSSION
Foveal crowding limits visual perception and more severe cases of foveal crowding are present in patients with amblyopia. Unlike visual acuity, crowding is measured in critical spacing. The two are independent of one another. However, it is still unknown what accounts for crowding. One study conducted shows that the difference in accuracy between spaced and condensed stimuli was significantly associated with lower grey matter volume in the right collateral sulcus (Yong). This raises the possibility of crowding being associated with neuronal density.
Identification of a Factor that Induces Prostate Specific Membrane Antigen (PSMA) Expression in Human Endothelial Cells
J U D Y L I U
A B S T R A C T
Prostate-specific membrane antigen (PSMA) is the most well-established, highly restricted cell surface prostate antigen. In addition to prostate cancer, PSMA is expressed in many other solid tumors. Importantly, PSMA expression is not found in adjacent normal endothelium, suggesting that a tumor-related factor may induce PSMA expression in neovessels. PSMA has become increasingly appealing to scientists as more research is conducted, specifically on its role in angiogenesis. Although it is now known that PSMA regulates angiogenesis, the specific factor(s) that induced PSMA expression in human umbilical vein endothelial cells (HUVEC) remains unknown. These previous studies paved the way for us to generate a monoclonal antibody that can block PSMA induction and use it to isolate the molecule that is responsible for the induction of PSMA expression in HUVECs. Clone 215 was produced using fraction 10 to 50 kDa from a kidney cancer cell line SK-RC-13. PSMA induction can be blocked by using clone 215 incubated with the conditioned medium from SK-RC-13 and MDA-MB-231. ELISA results showed that clone 215 binds to cancer secreted factor with molecular weight range 10 to 50 kDa. These results indicated that clone 215 binds the factor is the response for PSMA induction. We can use the monoclonal antibody from clone 215 to isolate and identify the factor that can induce PSMA expression in human endothelial cells. It will further expand our understanding in the role PSMA plays in tumor angiogenesis.
In this report, we used the in vitro model system from the study mentioned above to study PSMA expression and identify the factor that can induce PSMA expression in HUVEC. We generated a monoclonal antibody by using fractionated conditioned medium from SK-RC-13 to immunize the mice. This monoclonal antibody was able to block PSMA induction in HUVEC from conditioned medium. Figure 1 depicts the following: A. Tumor cells secrete factor(s) into the medium. Once released into the medium, the factor then cross-talks with HUVEC and transmits a signal that induces PSMA expression. B. In order to identify the factor (s), we decided to generate monoclonal antibody(s) that bind to the factor(s), which would therefore block PSMA induction. C. Isolation of the factor(s) by immunoprecipitation with monoclonal antibody(s) and identification of the factor(s) by mass spectrometry to determine the amino acid sequence. Succeeding to identify the factor(s) would contribute a great deal to the understanding of the role of PSMA in tumor angiogenesis.
Figure 1.
A. Tumor cell secretes the factor(s) into the medium. The factor then cross-talks with HUVEC and transmit a signal that induces PSMA expression.
B. Tumor cell secretes the factor into the medium. A generated antibody then binds to the molecule, preventing cross-talk with HUVEC, which blocks PSMA expression.
C. Isolation of the factor by binding it with the monoclonal antibody and immunoprecipitation with protein G beads.
D I S C U S S I O N
Screening of hybridoma clones: 846 clones were picked from the fusion by a scientist for us to test the ability of each clone to block the induction of PSMA expression in HUVEC. Clone 215 that showed strong blocking of induction of PSMA expression in HUVEC and positive staining of SK-RC-13 was selected for further study. Clone 547, which showed a positive staining of SK-RC-13, but did not block PSMA expression in HUVEC, was used as the experimental control.
I N T R O D U C T I O N
Prostate-specific membrane antigen, commonly referred to as PSMA, is a 100 kilodalton (kDa) type 2 integral membrane glycoprotein. It is mainly expressed in prostate epithelial cells. Its expression is increased in prostate cancer, and 95% of prostate cancers are PSMA positive (1). J591 is a monoclonal antibody that binds to the extracellular domain of PSMA. It has been extensively studied in both preclinical models and clinical trials. J591 showed high-affinity binding to viable PSMA-expressing prostate cancer cell line LNCaP cells in the tissue culture and was rapidly internalized (2,3). This ability to be engulfed by the prostate cancer cells allows the bound antibody to deliver any contents attached to that antibody, such as immunotoxins, drugs, and radioisotopes, into the prostate cancer cells, which can then be utilized to fight and kill the cancer cells. Radiolabeled J591 has successfully targeted the metastatic tumor sites of prostate cancer. These studies have demonstrated that PSMA is an ideal candidate for antibody targeting therapy for prostate cancer.
Recently, scientists have established a model system to study PSMA expression in human umbilical vein endothelial cells (HUVEC). In their publication (entry), the data showed that they were able to induce PSMA expression in HUVEC cultured on Matrigel using conditioned medium from a subset of human cancer cell lines in both levels of mRNA and protein. The data also indicated that J591 was internalized by HUVEC with induced PSMA. Furthermore, the scientists were able to induce PSMA expression in HUVEC co-implanted with a kidney cancer cell line (SK-RC-13) in the mouse model. While these results suggest that cancer cells and HUVEC crosstalk induces PSMA expression, they did not address what is responsible for inducing PSMA expression in HUVEC. That is the gap we aim to fill in our project.
M A T E R I A L S A N D M E T H O D S
-
Cell cultures
-
Preparation and fractionation of conditioned medium
-
PSMA induction in HUVEC
-
Immunofluorescence staining
-
Generation of monoclonal antibodies
-
Screening of hybridoma clones by blocking induction of PSMA expression in HUVEC
-
Enzyme-linked immunosorbent assay (ELISA)
R E S U L T S
To narrow the molecular weight range of the factor for the immunizations, the conditioned medium (SK-RC-13) was separated into three different fractions by using 50 kDa and 10 kDa filters. Fractions above 50 kDa, 50 to 10 kDa, and below 10 kDa were tested for the ability to induce PSMA expression in HUVEC and the unfractionated conditioned medium was used as the positive control for induction of PSMA expression. The results showed that of the three fractions, only the fraction 10 to 50 kDa highly induced PSMA expression. The unfractionated conditioned medium also strongly induced PSMA expression (Figure 2).
Figure 2.
Determining the fraction for immunizations from SK-RC-13. Unfractionated conditioned medium (CM), fractions above 50 kD, 50-10 kD, and below
10 kD, were tested for the ability to induce PSMA expression in HUVEC. Fraction 10-50 kD showed strong fluorescence intensity. Fractions above 50
kD and below 10 kD showed very weak fluorescence intensity. Unfractionated CM also showed strong fluorescence intensity.
PSMA is the most well-established, highly restricted cell surface prostate antigen. Beyond prostate cancer, PSMA expression is found in the tumor endothelium of a wide variety of solid tumors including renal cell carcinoma, breast carcinoma, colorectal, gastric adenocarcinoma, gynecological and head and neck cancers and glioblastoma, etc. [5–8]. Importantly, PSMA is not expressed in adjacent normal endothelium in these cancers, which suggest that tumor-related factors may induce PSMA expression in neovessels. A few studies were conducted in exploring the role of PSMA in tumor angiogenesis because absence of in vitro or in vivo experimental models of PSMA-expressing neo-vasculature. In this report, we used the in vitro endothelial model system that was developed by the scientists in the lab to study the regulation of PSMA expression in human endothelial cells. Our study was focused on identification of the factors that can induce PSMA expression in HUVEC. According to previous reports from the scientists in the lab, conditioned medium from 14 out of 19 human cancer cell lines were able to induce PSMA expression, which strongly suggested that the factor(s) was secreted into cultured medium by cancer cells. We should be able to collect and semi-purify it, and use it for immunizations. One of the advantages for using monoclonal antibodies to fish out the factor is the specificity of the monoclonal antibody, and another advantage is the functional assay in which induction of PSMA can be blocked by the monoclonal antibodies. For the reasons stated above, we generated monoclonal antibodies to bind to the molecules in the conditioned medium, and looked for the antibodies with blocking activity.
D I S C U S S I O N
In conclusion, in this report, our data demonstrated that we have generated a monoclonal antibody, and this monoclonal antibody was able to block the induction of PSMA expression in HUVEC from conditioned medium. This monoclonal antibody provides a valuable tool to isolate and identify the factor that is response for PSMA induction in human endothelial cells. It will help to understand the role of PSMA in tumor angiogenesis.
Clone 215 needs to be subcloned three times to make a stable clone. More supernatant and purified antibodies will be obtained after subcloning three times. We have limited supply of the supernatant from clone 215, preventing some key experiments from being done. In the future, a western blot should be performed against fraction 10 to 50 kDa and cell lysates from several cancer cell lines in order to determine the molecular weight of the factor.
A B S T R A C T
Stroke is the fifth leading cause of death in the United States. It is the result of problems with blood supply to the brain. Stroke victims suffer motor, cognitive, and speech disabilities. There are various methods of rehabilitation, but it is not well known who will recover and to what extent. My research aims to help improve the accuracy of prognoses after stroke, and possibly to choose the most effective treatments.
Machine-learning techniques called neural networks are designed to simulate a biological neural system. The goal of the Artificial Neural Network (ANN) is to predict output variables based on input data. ANNs can predict how a patient is expected to recover based on measures extracted from their imaging. The results of this research shows promising benefits.
Computational Analysis and Prediction of Post-Stroke Recovery from Baseline Brain Connectome Metrics
K I M B E R L Y H O
A stroke occurs when there is a blockage in the arteries that carry oxygen and nutrients to the brain.26 The region of the brain that the blood flow cannot reach will lead to motor, language and cognitive impairments. Many stroke victims participate in rehabilitation to help regain some of the disabled functions. For example, a patient might practice walking with a cane to strengthen motor skills and muscles in the quadriceps. However, research in the specific areas of stroke recovery is limited. This research aims to explore the brain and its diseases and to answer: how can technology be an application in stroke recovery?
The brain’s structural connectome39 is a map of all the major neural pathways in the brain.33 Current research relies on diffusion magnetic resonance imaging (dMRI) to assess the brain. Once the pathways are traced, structural connectomes can be created. Structural connectomes represent the amount of white matter traced between pairs of gray matter regions via tractography.
A network is used to summarize the structural connectome. The patient’s imaging and other measures are entered into the artificial neural network (ANN) so the network can generate its own predicted post-stroke datasets. This technique has the ability to decipher patterns between datasets and explain the wiring of brain.
Introduction
I. Subjects and Data
This research utilizes public datasets of 16 anonymous post-stroke patients, consisting of magnetic resonance imaging (MRI) and functional MRI values between 86 different brain regions. In this research, MATLAB model R2013a and the Neural Network Fitting Tool were used. The input datasets represent the amount of structural connectivity between the 86 regions in the brain to each of the other regions. The total number of structural connections in the brain is 86 x 86, thus 7396 values. The output datasets represent the NIHSS section scores for each of the 16 patients.
II. Training Algorithms and the Neural Network Fitting Tool
The neural network simulates brain activity and is tested for its ability to predict stroke recovery. The neural network used in this research is a multi-layer ANN, which has several levels called hidden layers37. The Neural Network Fitting Tool creates an optimized model, but is sensitive to outliers.
The forward phase of backpropagation computes the output value for each neuron based on weights from previous iteration. The next phase is the backward phase. This is the reverse of the forward phase. Weights at the current level would be updated before weights at the previous iteration. This computation allows the network to use error for neurons at the current level to estimate the error for neurons at the previous iteration.30
The accuracy of the ANN’s prediction is understood with regression analysis.35 In Equation 1, a linear regression finds the best-fit line for a dataset, which is assessed using a statistic called R2.
The Neural Network Fitting Tool also uses mean square error (MSE) to evaluate error.
Materials
and Methods
I. Relationship Between Training for Baseline Connectome and Baseline Function
It was hypothesized that because the signs of post-stroke in the first stage are easily visible, the measures determined by the NIHSS would match closely to results in the structural connectome. This hypothesis was consistent with the results after training the model. The training for this relationship shows moderate error and a R2 of 0.8726. In Figure 4, the MSE of the majority of the samples lie near 0.1998. In Figure 5, the regression for training, validation, and testing all show a regression close to 1. As the NIHSS measure increases, there are weaker connections, which demonstrates that this machine-learning technique is a faster and more convenient approach to measure the severity of the stroke.
II. Relationship Between Training for Baseline Connectome and 6-Months Follow-up Function
It is important to compare baseline connectivity to 6-months follow-up measures using ANNs to test how accurate the model’s predictions are. In Figure 6, MSE values are lower in comparison to the training for baseline connectivity and baseline functions. While there were at least two outliers, most of the error of the data trained resided at -0.083. This low MSE indicates the predictability power of neural networks. The decrease in MSE is a favorable indicator that the use of ANNs can be beneficial in this area. There is also a similar R2 0.863261, which is displayed in Figure 7. This value indicates the network predicted outcomes that were almost exactly the same as the target outputs.
III. Relationship Between Training for 6-Months Follow-up Connectome and 6-Months Follow-up Function
In six months, the patient has improved, meaning the NIHSS will not be as accurate an indicator in the later stages of recovery. As seen in Figure 9, the R2 is a low value of 0.15521. The test set resulted in a low R2 value because of the great error in this trial. As a result, the model’s predicted output was not closely related to the data from the NIHSS.
Results
D I S C U S S I O N
Limitations: Because the data did not account for any important medical information like age, gender, etc., this may have resulted in a higher MSE value. In addition, only datasets of 16 patients were used. In the future, it is best to expand the sample size. ANNs are also sensitive to outliers. Outliers show an effect on MSE, R2, and the overall performance of the model. Another limitation is the training time of the network. Backpropagation networks tend to train slower than other networks.
C O N C L U S I O N
This research successfully used MATLAB’s Neural Network Fitting Tool to predict stroke recovery from baseline neuroimaging measures. This technique can be used to create accurate predictions of stroke recovery and inform rehabilitation plans for stroke recovery.
F U T U R E W O R K S
Based on this research, ANNs are powerful tools in forecasting stroke recovery. The next step of this project is to test it clinically. In the next stage, it is important to account for the patient’s history. This can strongly supplement forecasting stroke recovery. The data gathered will be a qualitative measure from the NIHSS, similar to the datasets in this research. However, this trial work will not consist of a 6-month post-stroke dataset. This is where the success of ANNs becomes important. Based on the NIHSS measure of the stroke, physicians can use the ANN to predict the patient’s recovery. Medical professionals can then direct the patient’s rehabilitation procedures to meet his or her needs. To measure the success of these trials, data will be collected from each patient in 6-months post-stroke. It will compare the data acquired then to the predicted data at the beginning. If both datasets show a close relationship, or the recovery is progressing as expected, this research will be a major breakthrough for clinical practice. By then, it can be concluded that ANNs should be incorporated in clinical and rehabilitation stroke use. This tool essentially has the power to forecast how well the patient will recover and target the appropriate steps to take to recover effectively.
A C K N O W L E D G E M E N T S
This research was supported by Dr. Amy Kuceyeski (Ph.D. Assistant Professor of Mathematics and Neuroscience) of Weill Cornell Medicine’s Feil Family Brain and Mind Research Institute. I would like to thank Dr. Kuceyeski for giving insight and expertise in this area of neuroscience, assisting my research, and commenting on my report. I thank McGill University for permitting the use of the public and unidentified stroke subject data.
Discussion
And Conclusion
Usability Testing of a Web-Based Decision Aid for Breast Cancer Risk Assessment Among Multi-Ethnic Women
S H A Y A N H U D A C H O W D H U R Y
</background>
Chemoprevention with antiestrogens such as tamoxifen and raloxifene could decrease the incidence of invasive breast cancer among women with high-risk of breast cancer by a staggering 30-70%, similar to the risk reduction of heart diseases. And yet, the adoption of such treatment is abysmally low, to the tune of 5%, with a major reason being lack of awareness among the target beneficiaries about the benefits of chemoprevention, lack of routine breast cancer assessment and lack of knowledge about chemoprevention leading to concerns about its side effects, due to it being associated with those of chemotherapy.
</methods>
Dr. Katherine Crew’s team has developed RealRisks, a web-based patient decision aid (PtDA) to facilitate informed decision-making about chemoprevention by communicating eachpatient’s individual risk of breast cancer using their own personal information regarding family and medical history. The purpose of this research is to understand usability issues of RealRisks among patients of different ethnic, economic and educational backgrounds, ages, and English proficiencies. The hypothesis is that such factors will significantly affect the subject’s ability to use RealRisks and thus influence adoption of chemoprevention.
</results>
In-depth analysis of the study’s findings suggested that user behavior using surveys, usability recordings, ease of use and navigability of the tool, etc. may make improvements in the content and thus lead to increased adoption. Over 85% of the participants thought that RealRisks was useful and easy to use, while others alluded that it had limitations, which helped identify most of the various usability concerns regarding the tool.
</conclusions>
Most notable of these concerns were those relating to the childish undertones of the light and dense narratives and the confusion aroused in the study volunteers by the risk games and the comprehensive action plan, generated after all the seven modules of the decision aid are completed. Improvements were made and resulting findings suggested that such developments had a tremendous effect on usability for the target demographic. Larger implications of this research are that the study underscores importance of usability testing in creating culturally relevant and thus, targeted patient decision aids designed for diverse populations with varying health literacy and acculturation.
Regulation of Skeletal Fracture Angiogenesis by Immune-Derived Paracrine Secretions
C O N S T A N T I N E A T H A N I T I S
A B S T R A C T
Many mechanisms are activated during skeletal fracture repair, including angiogenesis, osteogenesis, and the inflammatory response. While the individual events are known, the interplay of the inflammatory responses and angiogenesis are not completely understood. We hypothesize that the early inflammatory response facilitates angiogenesis through the upregulation of cytokines. Due to the inherent flaws of purely in-vivo or in-vitro studies, we have developed a combination of the two, creating a 3D biomimetic angiogenesis model to study the interactions of the cytokines on C166-GFP endothelial cells within the engineered microenvironment. As a result of complex cytokine behaviors and effects, we tested different concentrations and times of TNF-alpha, IL-1beta, and IL-6. The angiogenic sprouting length and frequency increased in the presence of cytokine conditioned media (CM).
I N T R O D U C T I O N
For decades, there have been two main ways to study and analyze biological mechanisms: in-vivo and in-vitro. Each has its own advantages and disadvantages. In-vivo studies are those that are tested on living organisms and are advantageous because any effects observed in the study show how an actual organism’s body will respond to the experiment or treatment. The downside is having to address any moral and ethical dilemmas. For this reason, we used a new design for a two-layered 3-D microfluidic device. These devices integrate a 3-Dimensional collagen scaffold with two micro-channels (400 μm) to allow for not only the complete control of the cells’ physical environment, as in a 2-D in-vitro device, but also for the extraction of solutes, growth of endothelial cells, delivery of circulating cells, and control of mechanical factors such as wall tension and shear stress.1 Angiogenesis is the formation of new blood vessels from preexisting vessels, involving the migration, growth, and differentiation of endothelial cells. This process ensures that all metabolically active cells receive their necessary biochemical factors. Our project focuses on the immune-angiogenesis crosstalk during fracture repair. Angiogenesis is tied with bone repair, since nutrients such as oxygen, growth factors, cytokines, osteoblast and osteoclast precursors require a vascular route to reach the injury site for acute treatment. Within the first day of bone fracture, a hematoma (or a solid swelling of clotted blood) forms at the site to prevent the loss of blood and chemical factors.2 An acute inflammatory response arises to coagulate the hematoma in between the two fractured bone ends and to release proinflammatory cytokines, signaling molecules, adjacent to the injury site. It is known proinflammatory cytokines have degradative effects for bone tissue when exposed long-term.3 Smacrophages are Tumor Necrosis Factor (TNF-a), Interleukin-1, 6, 11, and 18, platelet-derived growth factor, vascular endothelial growth factor (VEGF), and bone morphogenetic protein.
R E S U L T S
Fig. 1 Schematic of Microfluidic Manufacturing Process.
Uncured 10:1 PDMS is poured over SU-8 Silicon Master wafer to create a thick and thin PDMS layer. After curing in an 80 degree Celsius oven, the PDMS is etched with the necessary channels via soft lithography. A two tiered microfluidic device was fabricated as a scaffold to mimic the three-dimensional geometry of blood vessels. A 10:1 solution of Polydimethylsiloxane (PDMS) and curing agent (Dow, Midland, MI) was made and poured over a photoresist SU-8 silicon wafer to create thick and thin layers. These layers were then finally bonded together via oxygen plasma treatment.
Fig. 2 Biomimicry of Bone Fracture Site.
One of the parallel channels mimics a blood vessel located at the fracture site, while the other channel mimics the source of cytokine factors from nearby cells.
Fig. 3 Capillary Tube Formation Assay to screen optimal concentrations of TNF-alpha, IL-1beta, and IL-6.
(A) Tubule formation for cells supplied with TNF-alpha 3000pg/mL.
(B) Tubule formation for cells supplied with IL-1beta 1000pg/mL.
(C)Tubule formation for cells supplied with IL-6 1000pg/mL.
(D)Compared to the negative control, all concentrations of TNF-alpha showed a relatively large increase in capillary area. 1000pg/mL and 3000pg/mL seem to be the most effective concentrations.
(E) While all concentrations of IL-1beta induced higher capillary areas than the negative control, 1000pg/ml had the biggest difference.
(F) The IL-6 concentrations appear to have a bell-curve effect on capillary area. As the concentration increases to 1000pg/mL, the area increases as well. After that concentration however, the area decreases. Data are represented as Mean+/-Standard Deviation. Scale Bar = 100um.
Fig. 4 CM-Induced Sprouting caused by IL-6, IL-1beta, and TNF-alpha.
(A) All concentrations of IL-6 produced sprouts longer than the negative control, but 5000 pg/mL was not better than the VEGF, or positive control, that we used. It seems that the lowest concentration, 10 pg/mL was the most effective
(B) In contrast to A, 10 pg/mL of IL-1beta was the only concentration to be more effective than the negative control. Even then, VEGF was better in both days 2 and 3.
(C) TNF-alpha also produced considerably long sprouts. Although 3000 pg/mL was the least effective out of the rest of the concentrations, 5000 pg/mL was still better than both positive and negative controls. Overall IL-6 produced the longest sprouting lengths, while TNF-alpha and IL-1beta came in second and last, respectively.
Fig. 5 Brightfield images of Day 0 and Day 3.
The images show the difference in sprouting lengths, on the left side of each vascular channel, for 10 ng/mL of positive control, and 10 pg/mL of the three types of conditioned media, which was the most effective concentration for all of them. Day 0 shows relatively no sprouting growth, just normal vascular vessel protrusions. Day 3 shows the maximum growth of individual sprouts. Scale bar = 100um. In respect to the three-dimensional sprouting formation, the most effective concentration of each cytokine was 10 pg/mL. For IL-1beta, VEGF was more effective in creating longer sprouts when compared to 10 pg/mL of the cytokine. It seems that for all of the cytokines, as the concentration increases, the sprouting length decreases. This must mean that the most effective range of all cytokines falls in from 5 pg/mL to 1000 pg/mL.
D I S C U S S I O N
The inflammatory response is an important modulator for signalling recruited cells into the bone fracture microenvironment. This study has shown that concentration-specific paracrine cytokines have a proangiogenic effect which may modulate the angiogenesis in bone repair.
From previous studies, it has been indicated that low concentrations of TNF-alpha promote angiogenesis, while high concentrations of TNF-alpha inhibit it.8 Due to this dual-nature of cytokine behavior, our study has screened that the optimal proangiogenic cytokine concentrations are closer to 10pg/mL for endothelial sprouting. However, our tubule formation assay indicates otherwise, resulting in an optimal concentration of 1000-3000pg/mL for all tested cytokines. This discrepancy may be a result of two different processes: angiogenesis and vasculogenesis.
From previous studies, it has been indicated that low concentrations of TNF-alpha promote angiogenesis, while high concentrations of TNF-alpha inhibit it.8 Due to this dual-nature of cytokine behavior, our study has screened that the optimal proangiogenic cytokine concentrations are closer to 10pg/mL for endothelial sprouting. However, our tubule formation assay indicates otherwise, resulting in an optimal concentration of 1000-3000pg/mL for all tested cytokines. This discrepancy may be a result of two different processes: angiogenesis and vasculogenesis.
C O N C L U S I O N A N D F U T U R E W O R K