Towards Regeneration: Uncovering Molecular Players in Utricle Hair Bundle Development and Maturation
Advisor: Dr. Sharma
Internship Location: Stanford University Cheng Lab
Onsite Mentor: Tian Wang
At the Cheng Lab at Stanford University, I will be working with Dr. Tian Wang, a senior research scientist, to understand utricle development and regeneration.
The utricle, an inner ear organ critical for balance, relies on the mechanotransduction capabilities of its hair cells. These cells detect head motion via the displacement of calcium carbonate crystals, known as otoconia, which shift in response to movement. Damage to utricle hair cells, caused by aging, ototoxic drugs, or other factors, can lead to vestibular dysfunction, a condition that affects the quality of life of over 35% of individuals aged 40 and older in the US. Understanding more about utricle hair cells and their development, through systematically assessing hair bundle dimensions, can provide information to guide us towards the regeneration of hair cells.
Fascin-2 (FSCN2), an actin-bundling protein, has been implicated in hair cell function and morphology, with prior studies identifying its role in auditory hair cells and hearing loss in mice. However, its role in utricle hair cell development remains underexplored. My project focuses on investigating the dimensions and expression of FSCN2 in 3D-reconstructed utricle hair cells during developmental stages, from embryonic day 13.5 to six months postnatally, emphasizing on prenatally versus postnatally added cells. I will create a research poster to present my results.
Through this research project, we can understand more about characterizing utricle hairs’ development to provide information to push the field towards treatments, helping improve the quality of life for over 35% of individuals aged 40 and older in the US.
Project Blog
Week 0: Project Introduction
Hi everyone! My name is Katie and this is my first senior project blog post! My project’s title is a bit complicated but it boils down to trying to understand how to treat balance disorders.
Balance disorders include diseases like Benign Paroxysmal Positional Vertigo (BPPV) and Ménière’s Disease which can impact up to 40% of the US population aged 40 and older. That’s a huge percent of our population but we currently don’t have any ways of permanently curing many of these diseases. The baseline cause of balance disorders is often issues with the utricle, a balance organ that’s part of our inner ear. The utricle has a bunch of tiny hairs inside, often a couple micrometers long, that have a semi viscous fluid surrounding them. On top of the fluid, there are calcium crystals called otoconia which move depending on your head location. The movement of otoconia causes the hairs to shift and send signals to the brain to help you balance. When these hairs degrade from drugs, aging, or other factors, you can lose your sense of balance and develop disorders like BPPV.
My project focuses on a specific characteristic of utricles which is that around half of utricle hairs are added and develop postnatally or after a person is born. If we can understand how and why this works, we may be able to use the same method to regenerate hairs in human adults, curing their balance disorders! My project will involve looking at mice of embryonic age 13.5 (E13.5), which means 13.5 days after the mice embryo is created, to postnatal day 180 (P180), which means 180 days after the mice is born. I will collect hairs from these mice by, unfortunately, euthanizing them and looking at the hairs under a special microscope that lets us create 3D images. These hairs would be treated with proteins that make them fluorescent or glow. Once I get the pictures, I will use a computer program called Imaris to compare different aspects of them: their shape, size, and protein expression. Specifically, I will examine a protein called Fascin-2 (FSCN2). In a previous study, we found that FSCN2 expression increased precisely when hairs got older so we wanted to explore the protein further.
My final research poster will describe any differences or new characteristics that I discover between the prenatal or embryonic hairs compared to the postnatal hairs. Come read the rest of my posts to see how I get there!
Week 1: Mice, Genotyping, and Lots of Image Analysis
Hey everyone and welcome back to my project. This week I focused on learning basic procedures and beginning my project. I began with getting caught up on lab updates and discussed which images had already been analyzed and any gaps I had noticed. Looking at the images that we had, my mentor and I decided that I would begin by focusing on images stained with FSCN2 and traced with tdTomato. These images essentially allow me to see the difference in FSCN2 expression between the prenatally and postnatally added hair cells. The ages that are most important for these images are P15, P30, P60, and adult/P180. We already had good images for P60 and P180 so we decided to focus on gathering samples from P15 and P30 mice. An important update I discovered at the lab is that we discovered a new protein of interest called PCDH15, which, through UMAP analysis, seems to be the opposite of FSCN2. UMAP analysis is a special kind of graph that can tell us the protein distribution across different parts of the utricle, including immature versus mature cells and hair cells versus supporting cells. FSCN2 increases as the hair bundles get older, but PCDH15 seems to be higher in immature hair cells. If we are able to get Pcdh15 to stain properly, the difference in the two proteins could give us interesting results on how hair cells mature. After I finalized the details for my project and caught up on lab updates, I began to work on P60 traced analysis.
Throughout the next couple of days, I finished my analysis of P60 and P180 traced FSCN2 hair cells using Imaris. I measured things like the tallest stereocilia, which is basically the longest hair, the shortest stereocilia, and the distance where FSCN2 begins expressing. I found some weird cells where FSCN2 didn’t express on all the hairs, which needs further analysis. Additionally, I began to do some wet lab work by completing genotyping of some mice. We genotype to see what kind of mice we have because these mice were made from a heterozygous-heterozygous cross. My mentor had previously done a PCR of the mouse DNA, so I ran an agarose gel off the PCR products. The gel lets us see the size of the DNA, allowing us to determine if they are heterozygous or homozygous dominant for the gene we want. We found some mice to fit our criteria, but a lot of the gel was inconclusive.
After I finished the P60 and P180 traced FSCN2 hair cell analysis, I started on the analysis of P30 traced hair cells. These cells were not stained with FSCN2, and instead, I focused on analyzing another part of the hair called the kinocilium, which gives us a better understanding of the overall structure of hair cells. I plan on focusing on this set of data for the next couple days. See you in two weeks!
Week 2: Images, Pictures, and Photos
Hello everyone and welcome back to my project. This week was very short because I just came back from our senior trip to Costa Rica and started my work on Thursday. I began by rereading my notes from before I left for my trip to remind myself of the tasks I have to complete. I remembered that I had to finish my analysis for P30 traced images. I looked back on my data and noticed that I had to polish the numbers by looking for more perfect cells. The cells I had been looking at before had very curled kinocilium, which are a specialized cilium found within hair cells. Instead of the normal cilium, which are found in numerous amounts on every cell, each cell only has one kinocilium, which is much longer than the other cilium and is stained separately. The cells I had found before break had very curled kinocilium, and with the staining, it is very difficult to measure how long these hairs are. Because of this, I spent my Thursday morning sorting through cells to find some that have relatively straight kinocilium for the most accurate measurements. I’ve attached some example pictures below of a curled kinocilium and normal kinocilium. The kinocilium are shown in purple and, as you can see, the bundles themselves are similar. However, image A has a much more curled kinocilium than image B and getting measurements from it would be much less accurate.
I then moved onto measuring a new P30 picture. This image is again of a P30 mouse that was stained with a kinocilium marker instead of a FSCN2 one. Even though my project is on FSCN2, images like this allow me to have a comparison or a control that makes my results more significant. On this image, I found a few bundle-less hair cells. These are especially interesting cells because, although the nucleus of the cells has formed, there are no stereocilia on top of them. An interesting side avenue to my project could be tracking these cells, especially since there seems to be a relationship to age. These cells imply that utricle hair cells develop through their nuclei first, and it would be fascinating to see how quickly stereocilia are added after the nuclei are formed. By just examining by eye, I have noticed that the embryonic ages and the early postnatal ages have some bundleless cells. It has been rare for them to occur in P30 and older mice, but I find fewer bundleless cells and the two that I did find were postnatally added, which follows our current understanding of utricle development.
Next week I hope to focus on compiling my data and making some graphs to uncover potential trends. I also hope to get images for P15 and P30 mice that are stained with FSCN2 to finish my data collection of all FSCN2 ages and begin further analysis.
Week 3: A New Protein Enters the Ring!
Hi again everyone. We’re already in week 3 of senior project! This week I prioritized finishing my up traced data and figuring out which parameters to look at for my PCDH15-related data. Just for some review, PCDH15 is a protein that we used an analysis tool called UMAP to investigate. Through UMAP analysis, we found that PCDH15 was inversely related to utricle bundle age so we wanted to look at the protein more to further understand utricle development. This protein is especially interesting because FSCN2 seems to be positively related to utricle bundle age so having two opposite proteins could make for some interesting analysis. I was confused over how to analyze it at first because the fluorescent staining for PCDH15 seemed to be bigger than Pha which stains for the cell itself. It doesn’t make sense that a protein on the hair would be expressed outside of the hair, right? But, my mentor told me that it is normal to see. She explained that PCDH15 seems to express even earlier than Pha and having low Pha expression and high PCDH15 expression is common on young cells. Right now, we guess that the overall trend for PCDH15 expression would be high in E18.5 cells because the hairs are young, a bit lower in P0 cells because the hairs got a little older but high again in P15 cells. We hypothesize that this is because P15 cells have postnatally added cells which are young that can express PCDH15 while the prenatally added cells don’t.
Also, PCDH15 has a unique expression pattern. Unlike FSCN2 which expresses on the top part of the bundle in most cases, PCDH15 seems to express throughout the entire bundle when it is young and spread to only expressing in the tips of the bundles when the cells get older. Along with looking at where PCDH15 is expressed on the cell, I also need to look at where the cells are in the sample. We can roughly divide the utricle into two parts: the extrastriola and striola. The extrastriola is more towards the middle of the samples of the striola are on the ring outside. These two regions have different compositions of cell types with the striola having more type i cells while the extrastriola has more of a mix of type i and type ii cells. Because of this difference, we also hypothesize that PCDH15 expression could vary across the two different regions. Understanding this aspect of utricle development could help us understand how the extrastriola and striola develop as a whole.
Next week I hope to get started on some graphs, especially for the traced data that I have finished so we can start seeing some patterns. Yay! See you then!
Week 4: A Big Setback
Hi again! It’s week 4 of senior project and I’m finally starting to see some patterns and results beyond my Excel sheet full of numbers. Here's one of the graphs I’ve been working on. There are some points that are unchecked and it's very preliminary overall but there seems to be something cool there. This graph analyzes PCDH15 by looking at the percentage of PCDH15-positive cells by each age. On each of my images I have a certain number of cells and, to get this graph, I counted all the ones that were shining bright red (PCDH15+) and counted the total of cells on the image to get the percentage. I only have one image for each age and region so these results aren’t super accurate yet but I am very happy to see an overall decline in PCDH15% as the cells get older. This makes sense with our theory that PCDH15 is a marker for utricle maturation and decreases in expression as utricles get older. Having more images for each age and adding some points from E15.5 and P30, P60, and P180 should hopefully help make this trend even more clear.
Also, this Wednesday one of the people in my project presented at our weekly lab meeting! There we got to discuss what we’ve been doing and get asked questions by other people in the lab. A person in the lab meeting mentioned that the way we’ve been measuring volumes could lead to some inaccurate results. After the full lab meeting, we sectioned off and discussed this issue further. The problem with our volume was some of the super small hairs were not captured by our normal method. In these cases, we made the volume bigger to make sure the small hairs were covered, but this volume overestimated the big bundles. We accounted for this by normalizing the values, however, this requires the volumes to have a linear relationship. We had never thought about that part before and just normalized by multiplying the volumes of a fully covered small bundle by a fully covered big bundle and dividing by the overestimated amount. However, while talking to the other lab member, we realized that there is more of an exponential relationship in the volumes due to how the images are formatted so we couldn’t use that method of normalizing. At the end of our discussion, we ended up deciding to test a new strategy where we crop out each cell to find its specific volume.
My project for next week is to try this new strategy and compare it to our old data. I hope to finish that and get some extra points for the graph I’ve shown above and another one I’m working on in order to have some cool results to share with you guys! Come back next week to hopefully see them!!
Week 5: Graphs...Yay!
Hi again everyone! It’s week 5 of senior project and there has been some new exciting stuff happening this week. First, I finished 1 set of each age and region so I finally have some graphs with full patterns! Of course, each age only has data from 1 image so the results are not completely accurate yet but I really like the trends I am seeing. For image A, which I showed you a little bit of last time, I added in all the ages and the trend is very promising. You can see that, generally, there is a steady decline in PCDH15 expression as the mice get older which is exactly what we expected to see. There’s a bit of an increase at P0 which may seem unexpected but this actually makes sense according to current information. At around P0, there’s another wave of hair cells being added which means that P0 utricles have a bunch of immature hair bundles that, according to our current understanding, would express more PCDH15.
For image B, you can see that PCDH15+ cells have shorter stereocilia than PCDH15- cells. This implies that PCDH15 expresses not only due to age but also the morphology of the bundle. Another interesting thing about this graph is that there are two groups of cells in the PCDH15- column (one grouped near the top and one grouped closer to the bottom). I checked all of those cells and found that, for the ones I can characterize, all the cells are type 2. However, a small number of type 2 cells are still in the top group which means they are generally bigger. I think this has to do with the time at which these type 2 cells were added which I hope to investigate more when we can look at cells with PCDH15 and tomato expression. It would be very interesting to see if the cells being added prenatally versus postnatally can account for the two groups.
Beyond the graphs, we also had a bunch of long lab meetings this week where I learned more about cochlea development and that microglia may have a lesser role in synaptic pruning than initially thought. I also had the exciting opportunity to review a paper that our lab is about to publish and give my insights on it. The paper is on how the cochlea regenerates hair cells and the two routes that it can take to regenerate at the different turns within the cochlea. The paper was very strong and had a lot of different pieces of evidence using different methodologies to support their conclusion. It was really informative to see how labs review papers and discuss various aspects that reviewers may ask for. To be honest, I didn’t understand the paper 100% but it was really interesting to sit in and add a few comments about things that were confusing or the few grammar mistakes that were left in.
In the next week I hope to finish off the PCDH15 analysis and maybe the FSCN2 one as well. I’m also going to discuss the volume issue we had from last week with another lab member who is an expert in the software I use. I’ll let you know what I find out. See you next week!
Week 6: Bit by Bit Progress
Hi again everyone! Happy “we are past halfway” week! Senior project has been flying by so fast and it's almost a month to our final presentation. This week has been a bit shorter and feels like I have made very little progress. I finished the PCDH15 data set, including a few repeats to verify results. Throughout the week, I focused on analyzing the data and noting if any of the data points did not make sense. The issues with volume are still ongoing though I believe we are coming closer to a solution that should make our overall data more accurate. I started with the youngest time point of E15.5 and noted that the volume I had used before likely underestimated a bunch of bundles. This led me to redo and very specifically check all of my measurements for E15.5 which was very tedious but I am now more confident about my data. I also had to redo parts of my E18.5 data and add more points. We have looked at all these ages before but had never stained for PCDH15. Nonetheless, the measurements from the two types of E18.5 data should be the same as the antibodies that allow us to track specific proteins don’t impact the hair bundles themselves. However, I noticed that I had a bunch more small cells in my data compared to the old data so I added in some more cells to see if it was a random error or if something was wrong with the image. Thankfully, it was just a random error and my data set seems more normal now!
Moving back to the volume issue, I have formed a method that I believe would yield us the most accurate results. The way measuring volume works is that the Imaris program selects parts of the image with a specific threshold of fluorescence and assumes that it is part of the bundle. Our previous method used a percentage of the max fluorescence in each image to form the volumes which was fine until we encountered small bundles in each image which were too light to be considered part of the bundle by Imaris. My working method would be to pick a threshold for both big cells and small cells but then work on each cell individually. When I look at each cell, I will see if either of those thresholds work and, if not, create a new volume specifically for that cell. Other members in the lab who have more experience have been trying something new but they are still testing. Hopefully one of our methods works!
Next week, I hope to finalize all of my PCDH15 data, finally finish my graphs, and hopefully figure out this volume issue. See you next week!
Week 7: Almost There!
Hi everyone! We are approaching the end of the senior project. Wow! This week we finally figured out the volume issue after spending a whole day testing out parameters. We figured out that the best way to go about it was through decreasing the background subtraction and increasing the amount of fluorescence we take. I’m still not entirely sure what background subtraction does, but it seems to make smaller sections, which allows us to better control the overall volume. This version is not perfect, but the numbers we get out of it make sense given how big the volumes appear to us by eye. I’m going to have a meeting with my mentor soon to better discuss this overall issue.
I also finished looking through and checking all of my PCDH15 data! I’m starting to finalize graph types and format for my final poster. I was thinking of doing one that looks at TS of PCDH15+ compared to negative cells for both ES and S and one with only ES. I also plan on adding the cell count bar graph that I showed you guys before. My poster will also include FSCN2 data, which I haven’t talked about for a while. The data for that is done, but I still have to check it, which I will prioritize for the rest of the week and next week. For FSCN2, I will similarly include a TS graph, but I also hope to do something about intensity. When I look at the FSCN2 images, it is very clear that the size of the hair bundle correlates with how intensely FSCN2 is expressed. Bigger bundles that express FSCN2 have a dimmer expression and vice versa. Some potential problems with this are the volume selected because we use the volume surface to get the intensity of FSCN2 expression on the bundle. I hope that our new volume parameters can help improve the accuracy of this, but I will need to test it on older data to make sure the numbers make sense.
Next week, I hope to come back with some finalized graphs and a finished introduction for my poster. See you then!
Week 8: Finalizing Figures
Hi everyone! This week I started working on gathering everything together to make the poster. I’ve attached two of my final graphs that I plan to add onto my figures. These two graphs will look familiar but I have refined them quite a bit. The first graph compares the tallest stereocilia in groups of PCDH15+ versus PCDH15- cells. This graph might not look that different compared to the one I showed you guys before but I’ve added a lot more values in the E18.5 and P0 ages. I also have started comparing the type of cell that I see. Overall, utricle hair cells start unclassified which means that we can’t really determine what they are when they are younger. As they get older, we can see the shape of the cell body change. At around P15, I can see some transitional cells which look like long and fat type 1 cells. As the utricle gets older, I can see more type 1 and type 2 hair cells which look like the shape shown in the diagram below. Overall, my graph shows that PCDH15+ cells are shorter on average than PCDH15+ cells and the older utricles hairs that have PCDH15 expression are either transitional or type 2. There doesn’t seem to be an older type 1 hair cell that has PCDH15 expression which is interesting.
Not much has changed about the PCDH15 cell count graph. I still want to include it because it will be an important graph on my final poster. For the overall structure of the PCDH15 graphs, I will do the TS graph, the TS graph with only the ES cells, the TS graph with only S cells, and then the cell count graph. I’m still figuring out my FSCN2 data and graphs but it will likely include something similar. However, the FSCN2 data is complicated by the fact that we included some data sets that track prenatally and postnatally added cells which will need to be considered in the analysis.
It’s been a very productive week. I will see you guys next week when I finish my FSCN2 data and start writing everything for the poster. See you then!
Week 9: Research Retreat and Drafting out Poster
Hi everyone! This week is a bit different because we had a research retreat on Monday. This retreat consisted of presentations from scientists in my department and talks by guest speakers Dr. Rosenthal and Dr. Tward. The presentations from scientists in my department made me realize how wide the field of ENT spans. There were the more expected presentations regarding hearing loss but also some unexpected ones. One of the coolest presentations was one where the scientist showed an augmented reality software for doing ear surgery. The demonstration videos for it seemed super promising and the mapping of the ear and the surrounding bones seemed very realistic and accurate. A very unexpected presentation was about cancer. It wasn’t specific about ENT related cancers but about using neutrophils to prevent the metastasis of cancer in general. There wasn’t a super direct application to ENT specifically but the new strides made by combining immunology, molecular biology, and otolaryngology were super interesting.
The guest speakers were also amazing and inspiring to listen to. Dr. Rosenthal spoke about how success can look like failure and went through his entire journey in becoming a surgeon-scientist. He was really funny and explained many scientific topics super clearly. His encouragement and excitement surrounding becoming an MD/PHD has inspired me to consider it a bit more. Dr. Tward was similarly inspiring. His path towards his current position as a surgeon-scientist and involvement in creating biotechnology companies showed me to try to take advantage of opportunities that come by.
After the research retreat on Monday, my mentor and I focused on figuring out how to measure intensity in my FSCN2 analysis. Just by eye, we noticed that FSNC2 became brighter as cells got bigger and wanted to measure this quantitatively. We tried a bunch of different parameters within Imaris but eventually figured that Imaris may not be able to measure what we wanted it to. We then switched to using ImageJ/Fiji which I was already familiar with and worked very well for getting intensity values quickly. I’m still working out some of the parameters we should follow for FSCN2 intensity analysis but I’m getting there and this method is far better than using Imaris.
During this week I also worked on drafting up my introduction and methods. My introduction was modified based on all the previous work and literature review I had done both for my project specifically and the requirements of the senior project like the proposal and abstract. My methods were written based on previous literature and the specific parameters that my lab uses.
See you next week with a more finalized poster and results!
Week 10: The End!
Hi everyone! It’s the last week of senior project. I’ve finished with data collection officially for PCDH15 but there are some more files I would like to add to my FSCN2 data if I have the time. The graphs I’ve shown you before were made in R but I’ve spent the last week trying to recreate them on Prism to make them more consistent with the other graphs my lab has made. Prism doesn’t seem super suited to making the specific graph I want so I’ve spent a lot of time tweaking things to try to get it to work. I’ve also been spending my time on writing my actual poster. My introduction and methods have been completed and I just need to finalize my graphs a bit more and write my results out.
My FSCN2 traced data may not get included in my final poster. There are some issues with the files because of the fluorescence. Basically, the marker that we use to track “tomato” or whether or not the file is prenatally or postnatally added has bled onto the tracker that we use to look at the bundle itself. This makes it very hard to tell what the bundle looks like and to get measurements off of it. However, I may be able to include data on if cells that are added prenatally vs postnatally are FSCN2+ or FSCN2-.
In summary, the overall content of my poster and presentation will include WT ES and S cells, traced WT ES cells, FSCN2 ES and S cells (untraced), and PCDH15 ES and S cells (untraced). The results I’m hoping to present include graphs about TS vs age, TS vs height of FSCN2 expression, and TS vs PCDH15 expression. I still need to organize all my data so there could be some on FSCN2 vs Pha volume and something about the intensity of FSCN2 expression though these could be difficult to finish due to the time constraints.
Thanks for following me through these 10 weeks! I hope to see you all at the symposium. See you!