Exploring Heterogeneity in Signaling Environments and Differentiation Trajectories in Migrating Mammalian Germ Cells
Advisor: Dr. Anuj Chaudhri
Internship Location: University of California, San Francisco (UCSF)
Onsite Mentor: Jay Zussman
At the Laird Lab at UCSF, I will be working with Jay Zussman, an MD-PhD candidate in the Laird and Wagner Labs, to understand the cellular and molecular details of germ cell development that will help us more broadly understand the determinants of ovarian fertility. According to the Center for Disease Control and Prevention (CDC), around 19% of women experience infertility (they are unable to get pregnant after one or more years of trying). Reduced size and quality of the ovarian reserve underlies ovarian infertility in cases of premature ovarian insufficiency and age-related infertility alike. Primordial germ cells (PGCs), the precursors of eggs and sperm, undergo a peculiar and challenging migration during embryonic development. We will investigate whether this migration contributes to differences in the likelihood of a germ cell surviving and contributing to the ovarian reserve. The Laird Lab has generated a single-cell RNA sequencing dataset of migrating PGCs from E9.5-E11.5 in mice. We hypothesize that anterior migrating germ cells will be more proliferative, will encounter survival and proliferation signals sooner (from their surrounding niches), and will be more developmentally mature than their posterior counterparts. Through this research project, we will use differential gene expression, pseudotime, receptor-ligand interaction, and gene regulatory network inference analyses to understand developmental heterogeneity among migratory and post-migratory mouse primordial germ cells, the findings of which will be compiled in a poster presentation. By understanding what alters germ cell developmental potential, we can help guide the selection of the best quality germ cells in fertility medicine procedures. This will provide greater autonomy to people in structuring their lives and building families.
Project Blog
Week 0: Project Introduction
Hi everyone! My name is Anjali, and welcome to my first senior project blog post! Here, I'll discuss the background of my project and what draws me to fertility-related research.
We all know about germ cells (eggs and sperm) and reproductive organs (ovaries and testes). The ovarian reserve (the egg capacity of the ovary) is established before birth and is finite. Understanding the cellular and molecular details of germ cell development will help us better understand the determinants of ovarian fertility. In my project, I seek to identify markers of germ cells that are most likely to contribute to the ovarian reserve. Primordial germ cells (PGCs), the precursors of eggs and sperm, undergo a peculiar and challenging migration during embryonic development. I will investigate whether the heterogeneity in this migration contributes to differences in the likelihood of a germ cell surviving and contributing to the ovarian reserve to help understand what eventually selects for an egg. Simultaneously, I will be trying to develop in-vitro models to help me observe this germ cell selection.
Why does understanding ovarian fertility and fertility matter? Fertility and reproductive health define our lives in many ways—in our society, many people organize their lives around building families. My study will help identify developmental determinants that shape the size and quality of the ovarian reserve. This may lead to future interventions to help preserve and maintain fertility, giving people more autonomy and more options to build families and structure their lives without being constrained by age or work. Furthermore, menopause — the end of the ovarian fertility period and egg release — has been linked to a higher risk of coronary heart disease and osteoporosis, as decreased estrogen levels accelerate fat buildup in the arteries and bone loss. Late menopause also increases the chances of breast cancer — the most common female cancer. Thus, understanding the balance between the proliferation and differentiation of PGCs is critical to understanding the general health of women and other people with ovaries.
Since my future blogs will reference a lot about the female germ cell development process, I think it’s handy to explain it here first if you are interested in learning more about my project. Buckle in!
At E7.25 (embryonic developmental day 7.25), primordial germ cells (PGCs), stem cells that are the precursors to eggs and sperm, are specified in the proximal epiblast.
From E9.5 to E11.5, the PGCs migrate through the hindgut epithelium toward the gonadal ridges (the developing sex organs). By E11.5, most PGCs have localized at the gonad.
From E11.5 to E13.5, the PGCs commit to becoming female germ cells (at this stage, they are called oogonia) as a result of bone morphogenic protein (BMP) and retinoic acid (RA) signaling. During this time, the PGCs proliferate and differentiate into oogonia in cysts.
From E13.5 to E15.5, the oogonial cysts start breaking down as the oogonia begin entering meiosis asynchronously, becoming oocytes. Complete oocyte formation involves two rounds of meiosis, but from E15.5 to E18.5, the oocytes are arrested in prophase of meiosis 1.
At E16.5, the pachytene checkpoint (a cell cycle checkpoint) causes the oogonial cysts to break down further as it kills off DNA-damaged and genomically unstable cells.
While the oocytes are still in meoitic arrest, the somatic cells (the non germ cells) in the ovary begin proliferating around the oocytes. These cells are known as granulosa cells, and with the layer they form around the oocyte, the full entity is called a primordial follicle. These primordial follicles are what we refer to when we say “ovarian reserve” - they have passed the necessary tests!
During puberty, additional layers of granulosa cells surround the primordial follicle as it is selected to mature into primary, secondary, and antral follicles. During this time, the oocytes resume meiosis until metaphase 2 of meiosis 2. During ovulation, the follicle releases what is now an almost fully developed egg into the fallopian tube, and it is only after fertilization that the oocytes complete meiosis 2.
So many words! Here are some visuals I hope will help. For future reference, anything “E” indicates a timepoint during embryonic development, and anything “P” indicates a timepoint during post-natal (after birth) development.
Source: https://europepmc.org/article/med/27329176
https://ibb.co/GQPsxp8
Source: ResearchGate
Week 1: Mice, mouse embryos, and mouse data visualizations!
Hi everyone, and welcome back to my project! I hope you are as excited as me to learn about the coolest wet lab and computational techniques in biology (well, at least developmental biology!). Thanks to my advisor, I had a blast learning and helping run different experiments. In addition to this, I picked up some super fun and handy skills in coding and visualization these past 2 weeks!
Week 0: The beginnings...wet-lab and mice
In the first week (Ski Week/Week 0) of this internship, I was able to observe a mouse dissection in which the embryos of a pregnant mouse were dissected out. Following this, the female embryos were further dissected for the embryonic ovaries to be extracted. We focused on the female emrbyos as my project pertains to ovarian as opposed to testicular fertility. The ovaries at this point contain germ cells — the oogonia, which are pre-meiotic cells that upon maturing and entering meiosis will become oocytes/eggs — in addition to somatic cells (non-germ cells). We need both the somatic and oogonia for later endeavors in this project. But since these cells were mixed together and in chunks, we needed to find a way to properly gather the cells we wanted. After gathering the controls and dissociating the ovaries, I was able to help run fluorescence-activated cell sorting (FACS), a type of flow cytometry that is able to separate the cells into streams based on the fluorescent signals from the cells (which are detected due to markers/tags we add to the cells before sorting!). Essentially, the FACS was able to separate for us the germ cells from the somatic cells in the dissociated ovaries.
Week 1: Diving into computations (+ more wet-lab and mice!)
This next week (Week 1), on the computational side of things, I learned about quality control analysis of single-cell RNA-sequencing (scRNA-seq) data using a previously analyzed dataset by the lab. Quality control and filtering gets rid of low-quality cells, which we identify through their mitochondrial reads (a characteristic that is provided for us in this dataset). It also gets rid of cells that might have accidentally been labeled by the same cell label (as each one of the cells in this dataset needs to have its own unique identifier). At home, I was later able to perform my own quality control analysis on another dataset using Python, which involved filtering the cells in the dataset, normalizing data and reducing variation, and identifying clusters of similar cells through clustering algorithms.
https://ibb.co/nj5mjVj
https://ibb.co/nsJtgxz
I also had so much fun building 2D and 3D visualizations/representations of the data — including ones that rotate (I’m talking about you, UMAP.gif!). I played around with the data some more to build an ML model to predict which cluster a cell belonged to (um let’s just say…it’s a major work-in-progress).
Check out the 3D UMAP here! https://drive.google.com/file/d/1v_F7_aGr7wiEOUionMfKGODM09uA6lAk/view?usp=sharing
My advisor later discussed my progress on the computational analysis with me. I also got to attend a (super fun) lab meeting, where I got the chance to learn so much more about developmental biology and hear about the cutting-edge research every lab member was doing!
The highlight of this week has to be the mice! My advisor showed me how to breed adult mice and plan genetic crosses to ensure we will have embryos with the right genotype in future experiments. The next day, we immediately went in to check the female mice for potential fertilization. This is known as plug-checking, as a plug is composed of a layer of leftover semen - a sign that the male and female mice engaged in reproductive behavior. This is important for us to keep track of so we can understand embryonic development and maturity — we assume the day the plug occurred, the female mouse's eggs were fertilized and embryos were formed. This is largely because the mice usually engage in reproductive behavior when the female mouse is ovulating. As a result, plugs have shown to be a good indicator of pregnancy around 70% of the time.
Additionally, since we are dissecting out oogonia, which form from E11.5 to E13.5 and peak at E13.5, we need a way of knowing what stage/timepoint in development the embryos we are dissecting are at. To do this, we start at the plug date and assume any and all embryos were conceived that day itself (as mentioned previously), and dissect the female mouse around 13 days later.
Following plug-checking, I was able to help run a DNA extraction and another FACS experiment!
Since then, I have begun learning how to use CellTypist to create annotations on scRNA-seq data — join me for my progress next week!
Week 2: Gene Expression Analysis and Mouse Breeding
Hi everyone! Welcome back to my Week 2 project update 🤩 Also side note: I finally have gained a skill I should have had all along - navigating public transport. The commute to UCSF is a ride, but now I can safely say I know what I’m doing. Yay!
Gene Expression Analysis and Cell Typist Annotations
Carrying forward from last week, I have been using CellTypist to create automated and manual cell type annotations on my scRNA-seq data. CellTypist allows me to train and test a model on a reference dataset of multiple primitive cell types to make predictions on the specific kinds of cells in the query/target dataset.
I have also been using pseudo-bulk-based differential expression analysis through the pyDESeq2 package, which essentially aggregates the gene expression of the cells in the dataset. This allows the analysis to be more robust to dropouts, or cases where the genes are expressed differently by a cell. Essentially, if the program is trained to see only one kind of gene expression due to using a small sample size (for instance, treating each cell as its own sample), it won’t pick up on the same gene being expressed in a different way. For this reason, we choose to create bigger samples (grouping all the cells into one big sample), so the program can see all the different ways a gene is being expressed!
So what did I analyze with pyDESeq2? I used this program to see differences in gene expression between migrating PGCs at different timepoints — specifically, at E9.5, E10.5, and E11.5 (timepoints in embryonic development). Next week, I will investigate the differences in gene expression between anterior and posterior migrating PGCs at the first two timepoints (as by E11.5, the PGCs have mostly reached the gonad already).
Wet-Lab Update
My advisor and I genotyped a mouse litter to identify mice to be used in future experiments. We took tiny samples of their tails to get DNA and then performed PCR on those samples, followed by gel electrophoresis.
But first…
What is genotyping?
Genotyping helps us determine…you guessed it…the genotype of the mice! We are looking for mice with very specific genetics, so we want to check that they are homozygous or heterozygous for the genes of interest. After plug checking the mice in the morning, getting tail samples (which we will dissociate to get DNA) from the mice that were to be genotyped, and attending the Wagner Lab meeting, I prepared the samples for the PCR. Polymerase Chain Reaction (PCR) is a DNA amplification technique, and DNA amplification is important because for the best readout, we want as many DNA fragments as possible. PCR involves 3 main steps: denaturation, annealing, and extension. During denaturation, the DNA strands are separated due to high temperatures, producing two single strands. In annealing, cooler temperatures allow primers (a very short stand of DNA that serves as a starting point for DNA synthesis) to bind to the complementary parts of the DNA. As a result, primers are highly specific and targeted, so we add primers that will specifically bind to the part of the DNA we want to amplify (the part of the DNA that contains the alleles for the gene of interest). In extension, DNA polymerase uses the primers as a guide to finish making the complementary sequence to make small double-stranded DNA again. Now, PCR repeats the cycle 25-40 times, allowing amplification again and again.
After running the PCR on my samples, I performed gel electrophoresis, which allows DNA fragments to separate based on size and charge. DNA is negatively charged (a property derived from the negative charges on the oxygen in the phosphate group) — the gel has a positive and negative end, so once the gel runs, the DNA moves to the positive end. When making the agarose gel, we use “combs” to create dents, known as “wells” in the gel where we can load our samples. Once the gel is loaded with the samples, we “run” it by applying an electric current. As the current is applied, the smaller and more negatively charged DNA fragments move further to the positive end.
Here is a picture of a gel I ran that was only for the genotyping of 3 genes:
https://ibb.co/sqTThjw
Since different alleles of a gene are different sizes, we can tell whether a mouse is homozygous for either of the genetic alleles or heterozygous for a gene based on the sizes of DNA fragments we observe. In the readout for the gel, this can be seen by “bands” at different vertical positions across the gel — with bands closer to the positive end having a smaller number of base pairs and the bands farther having a greater number of base pairs. If we see two bands, we know there must be DNA fragments of different sizes, so the mouse is heterozygous for that gene. If we see one band, we know it is homozygous, and we have to check the relative position of the band — aka the relative size of the DNA fragments — to check whether it is the wild type or mutant allele. This we do using a DNA ladder, which will tell us where fragments of certain base pair lengths, like 100, 200, 300, etc. base pairs, should land in the gel.
In this genotyping sample, you can only see one band amplified for each gene. That means all the mice were homozygous for all the genes!
The results of the gel were used to identify mice that were homozygous at 3 gene loci: at the CRISPR-associated protein 9 (Cas9), reverse tetracycline transactivator (rtTA), and octamer-binding transcription factor 4—green fluorescent protein (Oct4GFP) loci. Why do we care about Cas9, rtTA, and Oct4GFP? These 3 proteins combined are important for genomic editing (handy when we want to edit into each cell a barcode that will label it so we can track it) and the regulation/activation of gene marker expression that enables us to visualize germ cell migration and germ cell fate.
Remember FACS? FACS is able to separate the germ cells from the somatic cells because the germ cells express Oct4GFP, and the somatic cells don’t!
In addition to the above, I went with my advisor to attend a very enlightening talk on the Neuropsychiatry of Lupus by Dr. Betty Diamond from the Feinstein Institutes for Medical Research!
Week 3: Mouse Dissections, Imaging, and Presentations!
Hi everyone! Welcome back to another fun week of science in the lab. This week, I did less coding and focused more on honing some key developmental biology research techniques!
Last Friday, I attended the lab meeting for my advisor’s second lab since they are getting a joint M.D-Ph.D. degree from 2 labs. During the meeting, I learned about retinoic acid (RA) signaling and the rebirth of the yolk syncytial layer (YSL) in zebrafish embryos. The YSL contains yolk syncytial nuclei (YSN) that are the focus of the lab member’s project. They are trying to investigate heterogeneity in YSN to determine whether there are distinct nuclei populations (anterior, posterior, dorsal, or ventral). Their positioning may be indicative of later developments that happen in the embryo. I also got to learn about all the different ways in which RA affects nervous system development — yay neuro! Too much or too little RA can affect hindgut development in the zebrafish as well as cause neural crest defects.
Throughout the week, we checked in on the mice that we would eventually dissect for Friday’s experiment. We checked for pregnancies in another litter of mice (to be used for a future project). We had also hoped to perform more FACS of oogonia and somatic cells — however, to do that, we needed to collect oogonia and somatic cells. For those to be collected, recall from Week 1’s post that we would need to dissect female embryos at the E13.5 time point. For the embryo dissection to take place though, the female mouse carrying these embryos would have to be pregnant! While visual cues (such as size) may make it very clear when a mouse is pregnant, it is difficult to say when it is not, since it is possible for the female mouse to carry a small litter. Unfortunately, when we checked for mouse pregnancies (via dissection) on both Monday and Wednesday, none of them were pregnant, which meant we couldn’t gather cells for our FACS. However, I will be doing a FACS sort on Friday (using cells from a mouse we definitely know is pregnant), so join me for that in next week’s blog! For the mice that weren’t pregnant, I got to dissect them further to identify the different organs in the body and get a better understanding of mouse anatomy!
I also helped my advisor genotype a mouse litter using PCR, to determine the right pups to breed for their future experiments. Finally, I was able to see other grad students’ amazing images of mouse blastocysts and whole-mount stained mouse ovaries under the NSPARC microscope (a faster, higher resolution microscope that offers a larger field-of-view than traditional confocal microscopy)!
Week 4: St. Patrick's Donuts, Sorting, Statistics, and Some More DEG Analysis!
Hi everyone! I hope everyone enjoyed their March Break—or maybe tried out some of Krispy Kreme’s St. Patrick's-themed donuts (I certainly did, and they were delicious). This week, I only went to the lab for two days as I was mostly focused on coding, coding, and more coding!
On Friday, I attended the Wagner Lab meeting, where I heard a grad student’s amazing work on understanding molecular mechanisms during canalization in zebrafish embryo development, including how hypoxia and anoxia affect signaling pathways during development.
We also planned that day our experiment aimed to induce oogenesis in culture by isolating germ and somatic cells from the gonads of our embryos — the reconstituted embryonic mouse ovaries (rOvaries). Once again, that called for — you guessed it — a dissection! After dissecting the pregnant mouse to find her embryos, it turns out it's really hard to find developing gonads from those mouse embryos! Luckily, my advisor was there to help me throughout the process and help me retrieve the gonads! We gathered our controls and our ovaries. After more dissociation/preparing for the sort steps, we went down to the sorter, but as it turns out, we did not have enough oogonia for the main experiment :( We cannot bank oogonia and keep collecting over time until we have enough — they must be used fresh. This presents several problems as it means we have to hope that every time we check in on our mouse at the timepoint we want to test, she is a) pregnant, b) has enough embryos, c) has enough female embryos, and d) the ovaries from the embryos have enough oogonia. We are currently troubleshooting to find out ways to optimize the retrieval of the oogonia for better yield!
Tuesday, we repeated the sort, but this time, we dissected a mouse from a different mouse line as this was for another lineage tracing experiment (more on what that is next week) my advisor has been working on. I also took some time to practice my microscope dissection skills by trying to retrieve the gonads of the male embryos alone. After taking a very long time and nearly shredding the tiny sample, I was able to finally find the testes out of the 2 male embryos I was dissecting!! (There are a pair of testes in each embryo, so I was only able to find half the gonads, but still — progress!). We prepared for the sort, went down to the sorting machine again, and got the germ and somatic cells we needed.
On the computational side, since last time, I’ve been learning about and using differential gene expression analysis, specifically using the pyDESeq2 package and using the Wilcoxon Rank Sum statistical test to analyze more RNA-seq datasets. More recently, I’ve been using Gene Set Enrichment Analysis (GSEA) to check whether the differentially expressed genes can be found in pathways in already annotated gene sets — essentially, whether the genes are positively or negatively enriched among the gene sets. This shows genes that are likely being modulated at the time point we are performing the GSEA analysis.
Since I didn’t show it in Week 2, here is an example of some differential gene expression testing (using pyDESeq2 and another statistcal test called wilcoxon!) The first image does a comparison of all anterior vs posterior PGCs while the second image does a timepoint comparison of the PGCs.
https://ibb.co/k63QRsy
https://ibb.co/n16RsVM
Week 5: Genotyping, Gels, GSEA, and Gene Regulatory Network Analysis!
Guten Morgen allerseits and hello from Switzerland! This week I left a little early due to spring break travels to Europe, but I was still able to make progress on my project though and learn so much from my mentor!
Last Friday, I got to perform another genotyping experiment! Over the weekend, I observed my lab advisor perform an RNA extraction (which was perhaps one of the most careful and rigorous experiments to be performed). The RNA extraction actually starts with DNA from the somatic cells which are bust open to reveal the CARLIN-edited DNA (DNA from the cells of CARLIN mice). CARLIN mice are the CRISPR array repair lineage tracing (CARLIN) engineered mouse line. The CARLIN system uses CRISPR-Cas9 to induce genetic modifications to generate cell barcodes that we can track and read out later for lineage tracing. Lineage tracing involves labeling cells at a particular time in development and then tracking them as they progress to see their differentiation and the cells/tissues they gave rise to (we can map cell fate and cell clones to the cells they came from).
The DNA from the CARLIN-edited mice is transcribed into RNA to produce mRNAs (per the central dogma). In an RNA extraction experiment, the mRNAs are transcribed back into single-stranded cDNA (complementary DNA) using reverse transcriptase. Since the cDNAs were transcribed from functional, full processed mRNAs, they don’t contain the introns, or the noncoding regions of DNA. Looking at the cDNAs, we can now see the parts of the genome that actually contained the CARLIN barcodes!
On Monday, I watched my advisor plug-check and eartag mice again (eartagging just involved putting a numbered tag in a mouse’s ear so we have some way of identifying them!) while I helped them plan crosses and set up breeding cages. The rest of the day was pretty much all spent in coding! I’ve been performing Gene Set Enrichment Analysis (GSEA) and generating GSEA plots on the results from our differential gene expression (DGE) analysis on our single-cell RNA sequencing dataset. Essentially, we are taking annotated gene sets and pathway lists and trying to see how much of those genes and pathways are represented in our data, specifically in our list of genes that our DGE analysis ranked by differential expression metrics. We hope to identify how much of the genes in the gene sets that we pull are expressed by the cells we are analyzing, using statistical thresholds to understand enrichment significance.
I made a series of barplots and dotplots to visualize this enrichment! Here are a couple examples:
On Wednesday, I ran another genotyping experiment and since then, I’ve been working on pySCENIC analysis with my advisor. Python Single-Cell Regulatory Network Inference and Clustering (pySCENIC) involves checking the cluster of genes (known as regulons) that a list of transcription factors regulate. Knowing how and which transcription factors control gene expression helps us understand gene regulatory networks and interactions, and more broadly, and as a result, heterogeneity among the migrating PGCs.
Finally, I got to try my hand at zebrafish embryo dissections! (P.S. they're very hard.)
Auf Wiedersehen! That’s it for me! Super duper thankful again for all the learning I get to do and for my mentor’s support and guidance — more on all the things I get to learn and do next time!
Week 6: Building reconstituted embryonic mouse ovaries (rOvaries) in culture
Bonjour tout le monde! I’m back from my week in Europe, where I got to eat Swiss chocolates in Zurich, see the Eiffel Tower in Paris, and visit Platform 9 ¾ in London (unfortunately, I couldn’t go through the wall 💔)
Back into it! This week is probably the most important week of my project. Our goal was to build reconstituted ovaries of different sizes in culture. Or more specifically, reconstituted embryonic mouse ovaries (as we attempted in Week 5)! Creating ovaries in culture allows us to observe oocyte formation and attrition and follicle production, in addition to generating “an alternative source of gametes for reproduction” which will further fertility medicine and research (Hayashi et al., 2017). To do this, we needed to once again dissect the pregnant mouse, gather her embryos, and dissect the female embryos to retrieve the embryonic ovaries (as remember, my project deals with studying germ cell migration, proliferation, and selection during embryonic development!). After dissociating the ovaries, we put them through a FACS sorting machine so we can gather the germ and somatic cells! As a reminder, germ cells (which are at the oogonia stage at our dissection timepoint) express a green fluorescent protein (GFP) that the sorting machine can detect, which helps it distinguish between germ and somatic cells. We also unfroze banked somatic cells, since the ratio of somatic to germ cells is very high, and somatic cells can survive well being frozen and thawed.
Now that we have the oogonia and somatic cells that compose ovaries, we did a LOT of math to get them at the concentrations in media that we wanted (as in how many thousands of cells per mL). Then, using our fixed concentrations, we took out different amounts of our cell solutions for each oogonia and somatic cells and put them together in a 96-well plate to begin creating ovary replicates of different sizes. The protocol we referenced used a ratio of 5000 germ cells to 50000 somatic cells when putting cells on the plate. Our goal was to see how much we could scale this protocol down for the reconstitution to still work. We made a replicate of the same size as the protocol (5k), a replicate of 3000 germ cells to 30000 somatic cells (3k), and 3 replicates of 1000 germ cells to 10000 somatic cells (1k).
The next day we came in to image the rOvaries and check how they aggregated. After a lot of imaging, we were able to see that it seems to have worked! The somatic cells (which included pre-granulosa cells, cells that support the oocyte and provide nutrients) were aggregating and proliferating. The oocytes were also visible due to their fluorescence under the microscope! However, one of the 1k replicates did not seem to have aggregated much at all, so going forward we will only be using the 2 1k replicates that survived.
When imaging the rOvaries, I image them 4 ways - under 4x and 10x magnification, under each of two lasers: brightfield (standard) and the 488 laser. 488 nm is the wavelength for green light, so it will help illuminate the cells that Oct4GFP (remember GFP = green fluorescence protein, and the cells that express them are the oogonia!)
I’ve attached a sample of the images I have been generating! All the images show the rOvaries after 1 day in culture. The first image is one of the 1k rOvary replicates at 10x magnification, the second image is the 3k rOvary at 4x magnification, while the 3rd image is the 5k rOvary at 4x magnification. This week, I am attaching pictures of what the rOvaries look like under the 488 laser. Next week, I will show them under the brightfield laser! Notice the glowing dots on the images - those are collections of oogonia. As they mature into oocytes, the oogonia will express less and less Oct4GFP, but as of right now, we expect the strong GFP expression.
On Monday, I also got to watch another dissection - but this time, it was a mouse brain dissection done by another lab’s team member.
That’s it for me! Anjali out.
Week 7: Genotyping, Imaging, and Sorting!
Hi everyone!!
This week, I wasn’t able to do much coding at all, mostly because I was trying to get the R environment for CellChat — an “R package designed for inference, analysis, and visualization of cell-cell communication from single-cell and spatially resolved transcriptomics” — working. Unfortunately, it has not been working! However, I will be meeting today with my advisor to resolve that.
On Monday, my advisor and I did another FACS sort to get cells to make a few more replicates of our reconstituted embryonic mouse ovaries. P.S. My dissection skills have significantly improved! I’m able to dissect out the embryos to retrieve nearly all the gonads of the male embryos a lot quicker than before! As you might be able to tell, embryonic mouse ovaries are incredibly tiny, which means to observe and properly document their growth and the proliferation of the cells, we must do a lot of imaging. Monday, Tuesday, and Thursday, I imaged the ovaries using different lasers under the microscope and under different magnifications. So far, they look good! We also transferred the ovaries from the 96 well plate to the transwell after 4 days and made sure to change media in between as they progressed, more of which will be done early next week!
Images time!
Here is a picture of the 1k, 3k, and 5k replicates, all at 10x magnification and under the brightfield laser. These were on the 3rd day of being in the transwell, so they were in their 7th day of culture.
In addition to a lab meeting on Wednesday (and a very interesting paper discussion!), I mostly spent this week doing more genotyping and reconstituted ovary imaging (in addition to one more FACS on Thursday). There were a lot of new mice litters, so a lot of genotyping to do! I've attached a picture of one of the genotyping experiments I ran this week!
I also got the opportunity to watch my advisor do a library prep, where they took cDNA (generated from an RNA extraction - see Week 5’s post for more information on RNA extraction and genotyping), tagged it with UMIs (unique molecular identifiers), used size selection to ensure that only the cDNAs and not the other non-reverse transcribed RNAs were selected, and then performed PCR to extend the cDNAs. After all of this, they ensured that there were no cDNA duplicates (that the cDNAs that remain have unique UMIs). These cDNAs will then be sent to a sequencing lab, which will tell us the genetic composition of the cDNAs and where in our genome the genetic barcodes that the mice embryos have are!
More updates next week. Thank you for joining me! See you soon.
Week 8: Fixing and Blocking!
Hey everyone! This week was a little short as I was gone toward the end of the week for Admit Weekend at Stanford. However, on Monday, I was able to image the rOvaries once more before helping my advisor wash, “fix,” and “block” the rOvaries so we could stain and image them properly.
Fixing is a common technique in histology that involves adding reagents to the specimen (in our case, the ovary) to preserve the interactions between and organization of molecules within (it almost "glues" the molecules in place). Blocking usually involves adding blocking antibodies — but in our case, we use a compound called BSA (Bovine Serum Albumin) — that bind to an antigen to prevent the binding of other antibodies that are not specific to the antigen. The albumin binds to all the antigens. Only if an antibody is specific enough can it overcome the albumin binding and also bind to the antigen. Blocking helps to prevent unwanted reactions from occurring and with it, variability between experiments. Together, fixing and blocking help us produce clearer images!
I shadowed my advisor as they imaged a previous set of reconstituted ovaries using confocal microscopy, while also starting to run some CellChat analysis. My advisor also brought the most delicious cookies and creme ice cream cake to celebrate my birthday with the rest of the lab! I had such an amazing time and am so grateful to have had them as my mentor and to be a part of this amazing lab :D
Week 9: Designing immunofluorescence panels and MACS
Hi guys!
Last week, we took our reconstituted ovaries, fixed them in paraformaldehyde, washed in a wash buffer called PBST, and soaked them in a blocking buffer for a week till we got back.
On Monday, we incubated the reconstituted ovaries with the first set of primary antibodies that would bind to our antigens of interest, which are mostly fluorescent proteins like Oct4GFP, FOXL2, and VASA that help us detect the different kinds of cells in the reconstituted ovary (different stages of germ cells and somatic cells). In addition, we use stains like DAPI and Phalloidin. Also recall, the germ cells we added in to the reconstituted ovaries were oogonia, but we know that over time, they will have differentiated into becoming oocytes (oogonia that have entered meiosis).
Now, a little about what the antibodies and stains target. The anti-Oct4GFP antibody binds to the Oct4GFP, which is present in germ cells but most strongly expressed in oogonia (it is less expressed as the oogonia develops into oocytes and follicles). DAPI is a stain that targets all nuclear DNA, so it will target nearly every cell whose nucleus it can penetrate. FOXL2 is a protein expressed by pre-granulosa and granulosa cells, the cells that surround and support the oocyte. While VASA is also a germ cell marker, it is expressed by oocytes specifically (so no oogonia this time! That's why this marker is really important for the specificity of development we are trying to observe). Finally, phalloidin stains actin, and while actin isn't as related to observing oogenesis right now, it can certainly help us visualize the follicle morphology and structure.
Now when an antibody binds to the antigen, we can capture the signal during imaging so we can identify the location of the antigens/proteins based on where the antibody binds. The primary antibodies were added to the blocking buffer, and the next day, we washed away the blocking buffer and incubated the reconstituted ovaries in secondary antibodies. Why do we need secondary antibodies when we already have primary antibodies? A secondary antibody binds to a primary antibody. This is useful when the signal from the primary antibody isn’t strong enough, or when we are trying to be extra robust and ensure that we only capture the signals from the primary antibodies, as secondary antibodies are also highly specific. Then we put the reconstituted ovaries on a slide, added an antifade reagent which stabilizes the fluorescent signal from the proteins, and sealed the slide. (Guess what we sealed the slide with? Sally Hansen Top Coat! That's right, only the highest quality for our reconstituted ovaries). The reconstituted ovaries once again are placed in a room that blocks out light, as light can damage the ovaries.
On Wednesday, I also got to watch a MACS sort — Magnetic Activated Cell Sorting. You might recall up till this point we have been using FACS (Fluorescence Activated Cell Sorting) to sort our germ cells from the somatic cells each time we dissociate a mouse embryonic ovary after a dissection. FACS is extremely pure as it detects the fluorescent protein our germ cells express to give a high accuracy during sorting. However, it is extremely long, and MACS is useful for “lesser-stake” experiments. In our case, we used MACS just to get some somatic cells to bank, so even if we lost oogonia along the way, it doesn't make a big difference. MACS uses magnetic beads, and we used beads that target specific antigens found in germ cells, like SSEA-1 (Stage-Specific Embryonic Antigen). Thus, the germ cells would get stuck to the beads and the somatic cells flow through the magnetic column during sorting. With this, we were able to bank some more somatic cells, which is very helpful for the next time we want to make reconstituted ovaries!
I also genotyped another mouse litter, imaged the gel, and got to make a few more visits to the mouse house!
On the computational side of things, I am still working on the CellChat analysis from last week. Additionally, some of the code from our pySCENIC analysis took too much RAM to perform before so we are troubleshooting and finding a way to use the UCSF cluster to perform the job!
Thank you for joining me! See you next week.
Week 10: (Grand) Finale and Final Presentations!
Hi everyone! This week has mostly been focused on grinding out my final product (a poster) and beginning to make my Senior Project Presentation. This week at lab, I did some more genotyping of a female mouse and her pups, met with Dr. Laird (my advisor’s PI) to update her on my project’s progress, and then imaged the reconstituted ovaries under the confocal microscope.
We took the reconstituted ovaries that were sitting on the slides in the dark (we do this to reduce photobleaching, which decreases the fluorescence in the sample as a result of light exposure). The images we generated after putting the slides under the confocal microscope let us see the different proteins we stained for (FOXL2, VASA, Oct4GFP) in the reconstituted ovary along with the DNA and actin stains (DAPI and Phalloidin) we put in. The imager uses a laser to scans in horizontal cross-sections to quantitatively measure the fluorescence generated by the specific antibody binding (for the proteins) and the specific stains. If the reconstituted ovary replicate (like our 5k replicate) was too big to image in one go, the laser sectioned it into 4 and then stitched the respective images together at the end. The images were all so beautiful! We saw strong expression of the VASA and Oct4GFP (showing us we had oocytes as these are oocyte and germ cells specific) and FOXL2 in cells surrounding the oocytes (granulosa cells). These indicate we are seeing proliferation of the granulosa cells around the oocytes into what are now primordial follicles, which is amazing and shows us that our scaledown actually worked! Additionally, we see much fewer oocytes than the oogonia we started with, which indicates to us that oocyte selection is in fact happening. Now, we must figure out what made those surviving oocytes better or different than the rest. How to we do that? By examining the primordial germ cell (PGC) migration pathway and tracking heterogeneity, as I am doing in the computational analysis of this project! I also did some regular imaging with the tissue culture microscope of the new reconstituted ovary replicates we are testing. So far, we tested 5k, 3k, and 1k replicates. However, we started with 3 1k replicated on the 96 well plate, out of which only 2 aggregated and survived well — these were the ones we could see under the confocal microscope. As a result, to continue to optimize the scaledown we are testing a 2k replicate as well!
Here are some images! These are the colors in the images and what expression they are measuring:
DAPI for nuclear DNA staining (Blue)
FOXL2 for pre-granulosa and granulosa cell staining (Yellow)
VASA for oocyte staining (Magenta)
Oct4GFP for germ cell staining, although Oct4 downregulated in oocytes compared to oogonia (Green)
Phalloidin for actin staining (Red)
These are images of the 1k replicate under the confocal microscope. I’ve shown all the stains with VASA (magenta) so we can see where the oocytes are in every image.
Wednesday was a meeting day! I attended the lab meeting in the morning followed by a 4-hour workshop on genomic analysis and sequencing, which I didn't understand. But hey - I got a free lunch (and cookies) out of it! Sounds like a win to me.
To recap: The question I asked at the start of this project was how differences during the PGC migration impact germ cell selection later on during development. To do this, I wanted to first understand the heterogeneity that exists among the migrating PGCs. That includes answering questions like, how do the migrating PGCs interact with their somatic microenvironment? What signals or interactions affect the PGC migration and more specifically, their proliferation, survival, and development? Identifying these differences can help us make a sort of atlas of the molecular profiles of these PGCs spatially and temporally, which can help inform us more about this migration factors into germ cell developmental potential. Through the computational analysis we've done, we have been able to note heterogeneity through differences in gene expression, cell-cell communication, and the presence of gene sets, networks, and pathways among PGCs — analysis that we’ve done through timepoint and position comparisons.
PGC migration is awesome, but we also want to be able to observe germ cell selection as it happens. This is especially relevant during later stages of development as the germ cells commit to become female germ cells (oogenesis, beginning with the formation of oogonia) and eventually, only a few of these committed female germ cells are recruited to join the ovarian reserve. By this I mean that many of these cells will be killed to ensure that only those of the best quality get the go-ahead to become eggs!
To see selection, we want to see which descendants of the germ cells we add will survive all the rounds of oogenesis and development. Specifically, we are now interested in the E13.5 timepoint as it has the peak of all the germ cells we will ever make. Reminder! E13.5 germ cells have committed to becoming female germ cells in female embryos, and they are now called oogonia.
Tracking selection warrants a technique called “lineage tracing.” Lineage tracing involves labeling or “barcoding” cells of interest as a way of uniquely identifying them. Following barcoding, we can observe the descendants - or lineage - of our labeled cells. To observe this lineage tracing, we need to make - you guessed it - reconstituted ovaries! We used a protocol from Hayashi et al., 2017 as our reference, which directed us to make a relatively large reconstituted ovary. But since retrieving the cells necessary to make big replicates of these ovaries is very difficult (getting large amount of oogonia and somatic cells from dissections and FACS is complicated), we wanted to explore this model to see how much we could manipulate the sizes of the reconstituted ovaries. Specifically, we wanted to see how much we could scale this model down. Since this was more of a pilot experiment, the cells we used were not particularly labeled when they were put in, so we can not do any lineage tracing right now. We can, however, note from our results that this reconstituted ovary model does in fact works for observing germ cell selection (as we saw with the oocytes earlier in the blog). Once we optimize this, the next time we can label the germ cells and then turn this into a lineage-tracing experiment that will tell us all about the oocytes the labeled oogonia give rise to!
What a ride it’s been! While I won’t be blogging about this project anymore, I will still be genotyping, sorting, dissecting, analyzing, and imaging for a few more weeks as part of my internship. Super excited for what’s to come! Don't miss me too much :)
Finally, I'd like to give the biggest shout-out to my onsite advisor, Jay Zussman, for generously letting me be a part of their science these past few months, and for all the wonderful support, guidance, mentorship, and opportunity they have given me in this internship! Science is cool, but it's even cooler when you have a super duper cool advisor. Thank you to the Laird Lab for allowing me this experience and providing me with the resources for this project, and to Dr. Chaudhri and Ms. Nagpal for their supervision and direction in this process.
Wishing everyone the most wonderful and relaxing summer!!!
- Anjali