Chromatin, the complex of DNA-wrapped histone octamers that make up our chromosomes, is decorated with post-translational histone modifications (PTMs) that either increase or decrease transcriptional accessibility. Most regions have either predominantly active or repressive modifications that shape chromatin into euchromatin or heterochromatin, respectively. In addition to euchromatin and heterochromatin, some cells have poised chromatin that is decorated with both permissive and repressive modifications. While much is still unknown about a poised chromatin state, it is thought to permit swift changes in gene expression, which is a feature common in stem cells and lymphoid memory cells. Ultimately, the Escobar lab aims to determine the epigenetic mechanisms involved in maintaining the poised chromatin state of memory CD8+ T cells, and in line with this aim, plans to use a CRISPR-Cas12a biotinylation system to tag and precipitate poised chromatin regions for protein analysis and mechanistic studies. This project details the development and proof of concept of this CRISPR-Cas12a biotinylation system. Using traditional cloning techniques and a one-pot strategy to assemble CRISPR arrays, we will express and purify dCas12a-BirA+gRNA ribonucleoproteins (RNPs), introduce these CRISPR RNPs to mouse embryonic stem cells (mESCs), and perform CUT&RUN to verify effective biotinylation at specific chromatin loci. Preliminary results have demonstrated the successful purification of 5 CRISPR RNPs and a dCas12a-BirA control, as well as verified the presence of these CRISPR RNPs and the biotinylation of H3 upon delivery of this system to mESCs. Upon CUT&RUN analysis, we expect to see biotinylation of H3 at our targeted loci of interest. The completion of this validation step will allow us to apply this technology to any cell of interest, particularly CD8+ T cells, which may support significant insights to the mechanistic determinants of poised chromatin.

By 2050, the global human population is expected to reach 9.7 billion. Supporting this rapid growth will challenge global food systems, increasing the demand for healthier affordable foods such as poultry-sourced products (meat and eggs). Improving the accuracy of genotype-to-phenotype predictions for farmed animals could enable better breeding strategies and management practices that are crucial to meeting this goal. By characterizing regulatory genomic regions within various tissues, epigenetic factors which dictate specific cellular phenotypes can be pinpointed: improving phenotype prediction accuracy. To this end, we genetically sequenced (whole-genome bisulfite sequencing) samples of reproductive tissues (magnum, shell gland, isthmus, and ovary) from farmed groups of chickens (G. gallus). Through bioinformatics, we functionally annotated regulatory patterns of DNA methylation to identify tissue-specific epigenetic variation across the female chicken reproductive system. To tackle this, we utilized CGmapTools software to comparatively analyze tissues in a pairwise manner. In order to quantify each pairs’ differentially methylated regions (DMRs), all methylated regions were intersected and filtered through a statistical t-test. We found the pairwise comparison analysis between isthmus and ovary tissues to show the highest number of significant DMRs, being 51% hypermethylated in the isthmus. On the other hand, a comparative analysis between isthmus and shell gland tissues showed the lowest number of significant DMRs, being 54% hypermethylated in the isthmus. Upon annotating and conducting enrichment analyses on the DMRs, we learned that genes related to ECM-receptor interactions and focal adhesion were most prominent. Results of this research will aid the FAANG consortium (an international effort to improve farmed animal production) in improving genotype-to-phenotype predictions, hopefully enabling more sustainable genomic selection practices and genome-enabled management in the future of agriculture.

A grand challenge in the field of mass spectrometry proteomics is the problem of peptide sequencing. The dominant approach to this problem uses a database search; however, with the use of machine learning, peptide sequencing can be solved without using a database. Casanovo is a recently-developed, state-of-the-art machine learning model that solves this problem. However, Casanovo is trained on biased data, because most mass spectrometry proteomics experiments digest proteins using trypsin, which preferentially cleaves after lysine and arginine. This can result in incorrect predictions for data that was not generated using trypsin. Hence, we hypothesized that using a non-tryptic dataset to refine an existing Casanovo model would produce more accurate predictions on non-tryptic data. I constructed an unbiased dataset by downsampling from preexisting data, yielding a set of peptides with a uniform distribution of n-terminal amino acids. After splitting the data and fine-tuning Casanovo on an unbiased training dataset, I used the model to predict on an unbiased validation set. I also applied Casanovo to two non-tryptic datasets: antibody sequencing data and immunopeptidomics sequencing data. For the latter dataset, an antibody binding affinity tool, NetMHCpan4.0, was used to determine the binding probability of the predictions to test plausibility. We demonstrate that the fine-tuned model, Casanovo_ne, increases performance when predicting on non-tryptic data. Applied to the uniformly distributed validation set, Casanovo_ne predicts more accurately than the original Casanovo model, and predicts more uniform terminal amino acid distributions. Additionally, Casanovo_ne predicts more peptides that are likely to be MHC binders than a database search strategy. Finally, Casanovo_ne accurately represents the digestion rules for most non-tryptic enzymes. As future work, we will modify Casanovo to take as input the identity of the digestion enzyme, alongside each spectrum. We hypothesize this approach will further improve Casanovo’s performance for samples prepared with alternative enzymes.

Genomic amplification of specific genes is a common mechanism of adaptation that also underpins many human disorders. We use yeast (Saccharomyces cerevisiae) to investigate the mechanism of one such gene amplification. When yeast are grown in sulfate-limited conditions for many generations, the population becomes dominated by cells possessing an inverted triplication of the SUL1 gene, which produces a sulfate transporter. Because of increased transporter levels, these cells have higher fitness in limited sulfate conditions. The Brewer Lab proposed a model — Origin Dependent Inverted Repeat Amplification (ODIRA) — where this gene amplification is initiated via a DNA replication error. In the ODIRA model, DNA replication fork regression at short inverted repeats leads to template switching of the replication machinery and the extrusion of a replication-competent hairpin molecule, which after replication, recombines at the original locus to produce an inverted triplication. An alternative explanation behind the amplification is that the hairpin molecule is generated by double-stranded DNA breaks (DSB). To distinguish between these possibilities, we used an engineered strain in which the selectable marker gene, URA3, is split into overlapping fragments (“ura” and “ra3”) on two different chromosomes. The complete URA3 gene is only present in yeast that undergo rare direct recombination between chromosomes or by recombination of the replicated hairpin formed by ODIRA or DSB. We used CRISPR-Cas9 to induce DSBs upstream of the ura fragment and identify the type of event that restores URA3 function with contour-clamped homogeneous electric field gels (CHEF gels), Southern blots, and polymerase chain reactions (PCR). If DSBs drive hairpin formation, cutting the chromosome upstream of the ura fragment should increase the frequency of URA3 assembly via hairpin intermediate. We demonstrate that double-stranded DNA breaks do not increase frequency of hairpin intermediates, providing further evidence that ODIRA is responsible for the inverted triplications of SUL1 in yeast.

BRCA1-associated RING domain protein 1 (BARD1) is a key interactor with tumor suppressor BRCA1. Due to this interaction, deleterious variants of BARD1 have been associated with breast and ovarian cancer. In recent years, the use of clinical sequencing technologies to inform and personalize patient care, precision medicine, has skyrocketed. Despite the increased prevalence of clinical sequencing, in many clinically relevant genes, like BARD1, most single-nucleotide variants (SNVs) are cataloged as variants of uncertain significance (VUS). These VUS effectively prevent clinicians from using this data to help patients as it is unknown if the observed variant is pathogenic or benign. Consequently, a strong need to functionally assess BARD1 SNVs exists. To help resolve BARD1 VUS, we are applying saturation genome editing (SGE). SGE is a multiplex assay for variant effect that functionally assesses all SNVs for genes, like BARD1, that are essential in the HAP1 cell line. SGE uses CRISPR-Cas9 gene editing to integrate a plasmid library containing all possible BARD1 SNVs into a HAP1 population. Due to BARD1’s essentiality, cells with deleterious variants become depleted from the population. These changes in cell viability are quantified through next-generation sequencing and bioinformatic analysis comparing the abundance of a variant in the original SNV library versus its abundance in the cell population at the end of the experiment. Functional scores are then calculated for each variant. To date, I have designed targeted SNV libraries for 34 regions that span the entire coding region of BARD1. These libraries are preparing to enter tissue culture as we complete final quality checks. Ultimately, we expect the functional scores for BARD1 SNVs to be bimodally distributed, showing strong separation between deleterious and benign variants. These scores will be directly used to reclassify current BARD1 VUS – allowing clinicians to better guide patient care with respect to BARD1 SNVs.

Diabetes is a disease causing hyperglycemia in millions due to insulin resistance and ultimately the impairment of insulin production. Diabetes is associated with dyslipidemia, including increased triglyceride-rich-lipoproteins (TRLs) in the bloodstream. TRL related dyslipidemia contributes to diabetes and its complications. Apolipoprotein C3 (APOC3) regulates TRL metabolism and appears to play a causative role in diabetes-related cardiovascular disease (CVD). We hypothesize that by reducing APOC3 levels, we can reduce TRLs in circulation and restore insulin-producing beta cell function. To test this, we used a type 2 diabetes mouse model (leptin deficiency; OB) that lacked the LDL receptor alongside non-diabetic controls. Mice treated with an APOC3 antisense oligonucleotide (ASO) that silences APOC3 had reduced APOC3 levels, triglycerides, and glucose levels compared to the control ASO treated mice. The pancreas was sectioned and stained for insulin and glucagon. Diabetic mice had about a 75% reduction in pancreatic islet insulin staining (n=6-9, p<0.0001), which was partially restored in diabetic mice treated with the APOC3 ASO (a 2.2-fold increase compared to control diabetic mice, n=9-12, p=0.0051). Glucagon staining was unaltered. To investigate the possibility of TRL related beta cell dysfunction, we treated mouse pancreas B-TC3 cells with very low-density lipoprotein (VLDL). VLDL induced a 1.8-fold increase in early apoptosis (n=6, p=0.002) and a 2.3-fold increase in cell death (n=6, p=0.002) as assessed by Annexin V-staining alone or with 7-AAD staining, respectively. This was associated with increased activation of p-JNK, and increased expression of ER stress markers such as Grp78 and Chop mRNA. These findings corroborate the idea that TRLs contribute to beta cell demise and that silencing APOC3 reduces TRLs and mitigates their detriments. Further research will investigate the possible mechanistic pathway whereby TRLs induce beta cell dysfunction.