Enhanced Antitumor Response to Immune Checkpoint Blockade Exerted by Cisplatin-Induced Mutagenesis in a Murine Melanoma Model

Sequencing data from different types of cancers including melanomas demonstrate that tumors with high mutational loads are more likely to respond to immune checkpoint blockade (ICB) therapies. We have previously shown that low-dose intratumoral injection of the chemotherapeutic DNA damaging drug cisplatin activates intrinsic mutagenic DNA damage tolerance pathway, and when combined with ICB regimen leads to tumor regression in the mouse YUMM1.7 melanoma model. We now report that tumors generated with an in vitro cisplatin-mutagenized YUMM1.7 clone (YUMM1.7-CM) regress in response to ICB, while an identical ICB regimen alone fails to suppress growth of tumors generated with the parental YUMM1.7 cells. Regressing YUMM1.7-CM tumors show greater infiltration of CD8 T lymphocytes, higher granzyme B expression, and higher tumoral cell death. Similarly, ex-vivo, immune cells isolated from YUMM1.7-CM tumors-draining lymph nodes (TDLNs) co-incubated with cultured YUMM1.7-CM cells, eliminate the tumor cells more efficiently than immune cells isolated from TDLNs of YUMM1.7 tumor-bearing mice. Collectively, our findings show that in vitro induced cisplatin mutations potentiate the antitumor immune response and ICB efficacy, akin to tumor regression achieved in the parental YUMM1.7 model by ICB administered in conjunction with intratumoral cisplatin injection. Hence, our data uphold the role of tumoral mutation burden in improving immune surveillance and response to ICB, suggesting a path for expanding the range of patients benefiting from ICB therapy.


INTRODUCTION
Differences in tumoral mutational burden have emerged as predictors of clinical outcomes of immune checkpoint blockade (ICB) therapies across different cancers (1,2). This notion has been informed by large data sets which demonstrated that mutations acquired during tumorigenesis can be translated into neoantigens that elicit antitumor immune responses, which subsequently are bolstered by ICB (3,4). In view of these findings we reasoned that intratumoral delivery of low-dose DNA damaging agent calibrated to activate the tumoral mutagenic DNA damage tolerance pathway (5,6), without causing extensive tumoral cell death, might be a productive route to increase tumoral neoantigen formation and reinvigorate the antitumor immune response (2,(7)(8)(9)(10). Activation of the DNA damage tolerance pathway involves recruitment of error-prone translesion synthesis (TLS) DNA polymerases (6,11). Using the YUMM1.7 melanoma mouse model (12), we have previously demonstrated that tumoral TLS polymerases are transiently elevated following intratumoral delivery of the DNA damaging chemotherapeutic drug, cisplatin (13) and that the anti-CTLA-4/anti-PD-1 ICB regimen (14,15), given in conjunction with intratumoral cisplatin, leads to complete tumor regression in the mouse (13). These findings are consistent with the notion that lowdose chemotherapeutic treatments that increase tumoral mutation burden contribute to formation of immunogenic neoantigens and thereby provide a plausible path for enhancing the host's antitumor immune response and improving immunotherapy outcomes. Notably, this targeted approach affords an important advantage over conventional chemotherapy, because tightly calibrated intratumoral delivery of a chemotherapeutic drug allows for significant dose reduction limiting debilitating generalized toxicity, while achieving durable desired antitumor effect via revitalized immune response.
To probe further the involvement of cisplatin-induced mutagenesis in achieving the desired outcomes of combination treatments, we asked in the current study whether this scenario could be recapitulated with the in vitro cisplatin-mutagenized YUMM1.7-CM clone-generated tumors. We found that the YUMM1.7-CM clone produced by in vitro exposure of YUMM1.7 cells to low concentration of cisplatin, acquired nonsynonymous mutations, as well as phenotypic changes, including increased cell dimensions and reduced doubling time. Notably, compared to parental YUMM1.7 cells, the mutagenized YUMM1.7-CM cells were more efficiently eliminated when cocultured ex-vivo with immune cells isolated from the YUMM1.7-CM tumors draining lymph nodes. We also detected higher levels of infiltrating CD8 T lymphocytes with higher granzyme B expression in the YUMM1.7-CM generated tumors and most importantly, suppression of tumor growth in response to anti-CTLA-4/anti-PD-1 checkpoint blockade regimen. Considered together, our findings show that the in vitro cisplatin-mutagenized YUMM1.7-CM melanoma clone-generated tumors respond to ICB regimen similarly to the parental YUMM1.7-generated tumors, subjected in vivo to intratumoral cisplatin injection, supporting a pivotal role for increases in tumoral mutational loads in augmentation of the host's antitumor immune response.

METHODS
Cell Culture and Cisplatin Exposure YUMM1.7 mouse melanoma cells (12) were purchased from ATCC (ATCC CRL-3362) and cultured in DMEM/F12 (Invitrogen #11320033) with 10% FBS (ATCC #30-2020), 1% non-essential amino acid (Gibco #11440-076), and 1% penicillin/streptomycin. Cells were maintained at confluence below 85% and were routinely inspected for Mycoplasma (Lonza MycoAlert Kit #LT07-118) as we described (16). YUMM1.7 cells exposure to 0.2 µM cisplatin was initiated 24 h post seeding; cells were cultured in the presence of cisplatin for 5 weeks with routine refeeding by replacement of half medium volume twice/week and splitting at 1:2 as needed. Extended cisplatin exposure caused cell flattening with progressive slowing of proliferation that came to near halt by 5 weeks. At that time, cells were harvested, seeded in cisplatin-free medium at~1 cell/well in 96-well plates and cultured for 12 days. Wells with single clones were identified, collected, and expanded. Selected clones were passaged, and doubling times determined. A single YUMM1.7-Cisplatin Mutagenized (YUMM1.7-CM) clone was chosen for subsequent studies.

Whole Exome Sequencing and Mutation Mapping
Library construction and sequencing: Genomic DNA was isolated from YUMM1.7 and YUMM1.7-CM melanoma cells using the Easy-DNA ™ kit (K1800-01, Invitrogen). DNA was fragmented to less than 500 base pairs with a Covaris S220 instrument. Approximately 50 ng was used to prepare sequencing libraries with the NEBNext Ultra II DNA library kit (NEB #E7103, New England BioLabs) following the manufacturer's protocol. Exome capture was performed with the Twist Biosciences (San Francisco, CA, USA) Mouse Exome Panel kit (#102035) following the suggested protocol. Briefly,~400 ng of each library was pooled, dried in a vacuum concentrator, and dissolved in hybridization mix including the biotinylated exome capture probes, denatured, and incubated at 70°C for 16 h. Paramagnetic streptavidin beads were used to capture the hybridized material, the beads were then washed and placed directly into a PCR reaction to amplify the libraries. Following purification, the libraries were quantified and sequenced on NextSeq 550 Illumina System with the High Output flow cell and a 40-base paired-end protocol. Reads were demultiplexed and converted to FASTQ format with Illumina's bcl2fast2 software. YUMM1.7 DNA had 242 million read pairs and YUMM1.7-CE 184 million (data are available at NCBI Bioproject PRJNA734588). Variant analysis: The Broad Institute Best Practices protocols for variant discovery using the tumor-normal methods, with the mutated cell line taking the place of the tumor sample, were followed (17). Reads were aligned to the mouse mm10 reference genome with the Burrows-Wheeler Aligner (BWA) alignment program (18), version 0.7.17, with the BWA-backtrack algorithm commands. The resulting BAM format files were pre-processed for variant discovery with the MarkDuplicates and BaseRecalibrator functions of GATK, version 4.1.9.0, with default parameters.
The processed BAM files were input into the Mutect2 software with YUMM1.7 as the normal sample. The output variant call format (VCF) file was filtered with the FilterMutectCalls function with parameters -min-allele-fraction 0.15 andunique-alt-read-count 3. Variants found in exon coding regions were inspected for amino acid changes.

Lymph Node Collection and Immune Cell Isolation for Co-Cultures
Inguinal lymph nodes (LNs) were excised, weighed, and processed on ice for cell isolation: LNs were placed in 70 µm nylon mesh cell strainers propped over collection tubes. Cells were released by applying gentle pressure with the flat end of a syringe plunger, passed through the mesh, and rinsed with 3 ml ice-cold PBS. The dispersed cells were counted based on trypan blue exclusion and typically yielded 10 6 intact cells/mg LN tissue. Isolated cells were collected by 300 g/8 min centrifugation at 4°C, mixed with freshly collected Yumm1.7 or YUMM1.7-CM cells at a ratio of 1:32 (target:effector) and seeded in black-wall 96-well plates. Typically, 10 4 YUMM1.7 or YUMM1.7-CM cells were mixed with 3.2 × 10 5 00 lymph node cells and seeded in triplicate in 50 µg/ml Poly-L-lysine coated wells for 18-h co-incubation.

EdU Incorporation and Detection
Tumoral cell proliferation was visualized by incorporation of the thymidine analog, 5-ethynyl-2'-deoxyuridine (EdU) (#146186, Abcam) into newly synthesized DNA. EdU was prepared in 0.9% NaCl and given IP at 50 mg/kg, 24 h prior to euthanasia. Tumors were excised and cryosections prepared as described above. For EdU fluorescence detection, the Click-IT ™ EdU Alexa Fluor ™azide 594 Imaging Kit (Invitrogen, #C10339) was used per manufacturer's protocol (21) and as we described (22). For double fluorescence detection, immunostaining was done ahead of applying the EdU detection solution (1:2,000) for 30 min/25°C in the dark. The number of positive cells/mm 2 was quantified for four randomly selected fields in each section using ImageJ and mean value was calculated from four non-consecutive sections per mouse (n = 3); data are reported as mean ± SEM. Fluorescence was captured on Olympus IX71 fluorescent with QIC-F-M-12-C cooled digital camera. For in vitro studies, prior to co-culturing, YUMM1.7 and YUMM1.7-CM cells were pre-incubated with 10 µM EdU for 24 h to allow for EdU incorporation. After incubation cells were harvested, washed, mixed with LNs cells at 1:32 ratio, and plated in 96-well black wall plates for 18-h co-incubation. To visualize EdU positive cells, the Click-IT ™ EdU Alexa Fluor ™ -azide 594 detection Kit (Invitrogen, #C10339) was used (as described above). Following three washes, in-well EdU fluorescence was read using the Tecan FL200 plate reader with Magellan ™ software (Tecan, San Jose, CA, USA). Bright field/fluorescence images of co-cultures were captured and EdU positive nuclei were counted as we previously described (22). EdU positive nuclei were counted in randomly selected four fields per well for three co-incubation experiments; data are reported as mean ± SEM.

TUNEL Assay
Tumoral cell death was detected in situ using the terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL) kit (#11684795910, Roche). Briefly, cryopreserved tissue sections were permeabilized with 0.1% TritonX-100 in 0.1% sodium citrate solution for 8 min, washed with PBS, and incubated with reaction solution in humidified chamber for 1 h in the dark and mounted with Prolong Diamond anti-fade with DAPI; images were captured with Olympus IX71 fluorescent microscope.

Real-Time (RT) qPCR Determination of mRNA Levels
The RNeasy plus mini kit (#74 134 Qiagen, Valencia, CA, USA) was used. Reverse transcription was with iScript RT supermix (#1708840 Bio-Rad) and RT-qPCR analyses with the CFX96 Real-Time System (Bio-Rad, USA). 18s mRNA was used as reference and relative expression levels were calculated using the formula: Relative expression = 2 [-(CT gene of interest − CT internal control)] according to Schmittigen & Litvak (23). PCR reactions were done in triplicate with SSO FAST EvaGreen ® supermix (#1725201 Bio-Rad). Cycling program was 95°C for 2 min, 40 cycles of two-step incubation, initially at 95°C 5 s then 15 s at 55°C followed by melting curve analysis. Primers are listed in Supplemental Table S1.

Statistical Analysis
Data are provided as mean ± SEM calculated from three to four independent biological experiments, as stated. Unpaired twotailed Student's t-test was used to compare the means between groups. P value <0.05 was considered statistically significant. MegaStat ® software for Excel was used.

Features of YUMM1.7-CM Clone Generated by In Vitro Exposure to Cisplatin
Phenotypic changes observed in cultured cisplatin-mutagenized YUMM1.7-CM cells included increased cell size and reduced doubling time ( Figure 1A and Table 1). The changes were visualized by nuclear DAPI stain and cytosolic immunofluorescence of the ubiquitous glycolytic enzyme, 6-phosphofructo-2-kinase/fructose-2,6-biphosphatase 3 (Pfkfb3) ( Figure 1A). An approximately 25% reduction in cell doubling time was detected ( Table 1). Interestingly, higher baseline expression of DNA repair proteins, including the base excision repair DNA polymerase beta, which is involved in oxidative DNA damage repair (24), the DNA damage recognition protein Xpa (25,26), as well as the transcriptionally regulated error-prone  translesion synthesis (TLS) DNA polymerase kappa (27), was observed ( Figure 1B). The mutational signature of in vitro cisplatin exposure was determined from whole exome sequencing of parental YUMM1.7 melanoma cells and the mutagenized YUMM1.7-CM clone ( Figure 1C). Comparison of the YUMM1.7-CM clone to parental cells revealed 599 cisplatin induced base substitutions (Supplemental Table S2), including 81 non-synonymous exonic mutations (Supplemental Table S3).
Mutational signatures of cisplatin have been widely documented in different cancer cell lines and human tumors post cisplatin chemotherapy (28,29).
Augmented Ex-Vivo Cytotoxicity of Immune Cells Isolated From the YUMM1.7-CM Tumor-Draining Lymph Nodes (TDLNs) In co-cultures, clustering of immune cells isolated from the tumor draining lymph nodes, was observed around YUMM1.7 or YUMM1.7-CM cells. In contrast, clustering was not seen with immune cells isolated from lymph nodes of naïve non-tumor bearing mice. Bright field images of co-cultures seeded at a 1:32 ratio of target:effector cells show clustering of TDLN-derived immune cells around cultured YUMM1.7 and YUMM1.7-CM cells (Figure 2A, examples demarcated in red). The immune system engagement was also reflected in three-fold weight increases of TDLNs compared to non-DLNs or to LNs collected from naïve non-tumor bearing mice ( Figure 2B). This is consistent with TDLN enlargement reported in patients (30) and recapitulated in mouse models (31,32). Interestingly, dependence of anti-PD-1/PD-L1 therapy success on extant tumor draining lymph nodes, was recently reported, suggestive of clinical benefits while performing ICB prior to TDLN resection (33,34). Here, we observed that immune cells isolated from the YUMM1.7-CM tumor-bearing mice TDLNs, were more effective at eliminating co-cultured target cells compared to cells from TDLNs of YUMM1.7 tumor-bearing mice. For quantitative assessments, prior to co-culturing, tumor cells were pre-incubated in the presence of the thymidine analog, EdU (10 µM) to allow for EdU incorporation into nuclear DNA and subsequent florescent detection of tumor cells. After 24-h incubation, cells were harvested and re-plated in EdU-free growth medium, either alone (control), or after mixing with TDLN derived cells for 18-h co-incubation ( Figure 3A). Next, the extent of EdU-positive cells elimination from co-cultures was assessed using two methods: 1) measurement of in-well EdU fluorescence and 2) counting EdU positive cells ( Figure 3B). While cell counting was slightly more sensitive than in-well fluorescence readings, results obtained with both methods, revealed greater elimination of YUMM1.7-CM compared to YUMM1.7 cells following co-incubation with the respective TDLNs-derived immune cells (44 ± 4% for YUMM1.7-CM cells versus 35 ± 3% for YUMM1.7 cells). These findings agree with immunofluorescent imaging that revealed greater presence of CD8 + T cells among lymphocytes surrounding the YUMM1.7-CM cells ( Figure 3C). Higher levels of granzyme B positivity in co-cultures with cells isolated from TDLNs of YUMM1.7-CM tumors-bearing mice were also observed ( Figure 4). Granzyme B protease is a key mediator of target cell elimination by cytotoxic CD8 T lymphocytes (35).

Curtailed Tumoral Cell Proliferation in YUMM1.7-CM-Compared to YUMM1.7-Generated Tumors
In vivo proliferation of tumoral cells was assessed by immunodetection of the proliferation biomarker Ki-67, which is expressed in cycling cells and identifies the proliferating fraction in a given cell population (36), and by detection of tumor cells that incorporate EdU into nascent DNA. Analyses of  tumoral cryosections revealed higher numbers of Ki-67 positive cells in the YUMM1.7 when compared to YUMM1.7-CM generated tumors (Figures 5A, B). Similarly, the number of EdU incorporating tumoral cells was higher in YUMM1.7 compared to YUMM1.7-CM tumors ( Figures 5A, B). The complex nuclear patterns of Ki-67 expression and EdU incorporation were observed at high magnification in merged images ( Figure 5C), reflecting heterogeneity of tumoral cells transitioning through the different stages of the cell cycle.
Higher Levels of Tumor Infiltrating CD8 T Lymphocytes and Granzyme B Expression in YUMM1.7-CM Compared to YUMM1.7-Generated Tumors Mice inoculated with either parental YUMM1.7 or mutagenized YUMM1.7-CM melanoma cells according to our standard protocol (13), were randomly assigned to receive IP, isogenic or anti-CTLA-4/anti-PD-1 sera 3×/week (n = 6-10). Tumors and lymph nodes were collected 20 days post inoculation. Analyses of tumoral cryosections revealed~3-fold higher baseline levels of infiltrating CD8 + T lymphocyte ( Figure 6) and granzyme B expression ( Figure 7) in YUMM1.7-CM when compared to YUMM1.7-generated tumors. Interestingly, ICB treatment of mice bearing the YUMM1.7 tumors, resulted in marked increases in tumor infiltrating CD8 lymphocytes and granzyme B expression, nearing the baseline observed in YUMM1.7-CM generated tumors (Figures 6 and 7, bottom panels). In contrast, when ICB was administered to the mutagenized YUMM1.7-CM tumors-bearing mice, the treatment resulted in only marginal changes in the number of infiltrating CD8 T lymphocytes ( Figure 6) and granzyme B expression levels (Figure 7).

YUMM1. 7-CM and YUMM1.7-Generated
Tumors Differ in Growth Profiles, the Extent of Tumoral Cell Death, and Responses to ICB Regimen The extent of tumoral cell death assessed by TUNEL assay, revealed significantly higher baseline levels of TUNEL positive cells in YUMM1.7-CM compared to YUMM1.7-generated tumors, as well as marked increases in TUNEL positivity following ICB regimen ( Figure 8A). Individual tumor growth curves for the different groups revealed slower growth rates of YUMM1.7-CM compared to YUMM1.7 tumors ( Figure 8B). In the fast-growing YUMM1.7 cohort, all tumors were collected for analyses 20 days post inoculation when tumor volumes reached about 500 mm 3 . For comparative analyses, half of the YUMM1.7-CM cohort, was similarly processed at 20 days post inoculation for tumor and lymph node collection, while for the remining mice, tumor growth rates continued to be monitored (n = 6). Importantly, in the YUMM1.7-CM tumor-bearing mice/ ICB treatment group, nearly all tumors regressed, while growth of YUMM1.7 tumors was not suppressed by the ICB regimen ( Figure 8B). This is consistent with our earlier findings, that demonstrated that a combination protocol that includes ICB in conjunction with intratumoral cisplatin delivery is required for eradication of YUMM1.7 melanoma tumors (13).

DISCUSSION
Our study demonstrates that tumors generated with in vitro mutagenized YUMM1.7-CM cells elicit a productive antitumor immune response in vivo, in the mouse ( Figure 9). Consistent with enhanced immune surveillance in immunocompetent mice, growth rates of the cisplatin-mutagenized YUMM1.7-CMgenerated tumors are reduced. Furthermore, immune cells isolated from YUMM1.7-CM tumors draining lymph nodes are more potent ex-vivo at eliminating co-cultured tumor cells, when compared to cells isolated from lymph nodes draining the parental YUMM1.7 tumors. Importantly, we show that the YUMM1.7-CM tumors readily regress in response to immune checkpoint blockade (ICB) regimen, similarly to complete response to ICB achieved in YUMM1.7 tumor bearing mice when ICB is administered in conjunction with low-dose intratumoral cisplatin delivery, as we previously reported (13). In the current study, we characterized the mutational signature of cisplatin and described phenotypic changes detected in cultured YUMM1.7-CM cells, and in the YUMM1.7-CM-generated tumors. Whole exome sequencing of the mutagenized YUMM1.7-CM clone revealed that nearly 15% of the acquired base substitutions, cause amino acid changes that are likely to contribute to neoantigen formation and improved engagement of immune system. The notion that neoantigen formation enhances antitumor immunogenicity and improves responses to ICB therapy is supported by clinical outcomes demonstrating that tumors with high mutational loads are more likely to respond to ICB therapies (2,37), rendering the mutational burden an important, albeit imperfect predictor of response to ICB (3,4,38,39). Examples from the clinic include subpopulations of colon cancer patients with DNA mismatch repair deficiency characterized by high tumoral mutation burden and elevated neoantigens, which have shown high response rates to anti-PD-1 checkpoint blockade (7,40,41). Similar observations were made in melanomas (8,42,43) and lung cancers (44,45). These encouraging clinical outcomes were recapitulated in animal studies that sought to investigate the links between mutational loads and efficacy of responses to ICB. One approach included the use of CRISPR-Cas9 to inactivate different components of the DNA mismatch repair pathway in several mouse cancer models, which led to significant increases in mutational loads and neoantigen formation in the modified cells, resulting in augmented immune surveillance and restricted growth of subsequently formed tumors (46). Other experimental approaches involved the use of a cyclin-dependent kinase 7 inhibitor to induce replication stress and genomic instability, which resulted in enhanced response to ICB in mouse lung cancer (47), and in mouse melanoma model, the use of UVB-mutagenized YUMM1.7 cells to generate tumors, resulted in favorable response to ICB treatment (20).
We previously developed a mouse melanoma combination treatment protocol that entails ICB regimen given in conjunction with low-dose intratumoral cisplatin delivery designed to block high fidelity replicative DNA synthesis in tumoral cells, while activating the mutagenic DNA damage tolerance pathway (13). The DNA damage tolerance pathway involves a shift to TLS DNA polymerases-catalyzed mutagenic synthesis, which is critical for averting the replication forks collapse following formation of DNA synthesis-blocking cisplatin:DNA crosslinks (6,11,48). The error prone TLS polymerases eta (49)(50)(51) and kappa (52,53) have been specifically implicated in bypass synthesis of cisplatin crosslinks (54)(55)(56)(57), and we detected upregulation of these polymerases in melanoma tumors following intratumoral cisplatin delivery (13). Consistent with the premise that mutagenic DNA synthesis increases mutational loads (58,59) and neoantigen formation, thereby augmenting tumor immunogenicity and antitumor surveillance (7, 60), tumor eradication was achieved with the intratumoral cisplatin delivery/ICB combination protocol that we developed (13).
Here we demonstrate that durable tumor regression is achieved with the in vitro mutagenized YUMM1.7-CM cell-generated tumors in mice subjected to the ICB regimen (Figure 9, left). Involvement of TDLNs in the antitumor immune response is manifested ex-vivo, as augmented potency of the YUMM1.7-CM TDLN derived immune cells at eliminating the co-cultured tumor cells. The corresponding in vivo experiments reveal a markedly greater infiltration of CD8 T lymphocytes and higher granzyme B expression in the TME of YUMM1.7-CM tumors compared to TME of YUMM1.7-generated tumors. Interestingly, while ICB regimen leads to marked increases in the levels of infiltrating CD8 T lymphocytes and granzyme B expression in the YUMM1.7generated tumors, similar increases are not observed in YUMM1.7-CM tumors. Although speculative, it is plausible that as previously suggested (33,61), the ICB treatment may not only reinvigorates extant immune cells in the TME but also potentiates trafficking of T lymphocytes to the TME, which might be the case with the less immunogenic YUMM1.7 tumors. Conversely, in the YUMM1.7-CM tumors, ICB might function chiefly to reinvigorate cytotoxic T lymphocytes, which had been already recruited to the TME due to augmented immunogenicity of the YUMM1.7-CM tumors, facilitating regression of YUMM1.7-CM tumors in response to ICB regimen. Collectively, our findings support the premise that buildup of tumoral mutations driven by an activation of intrinsic tumoral DNA damage tolerance pathway enhances immunogenicity and immune surveillance of solid tumors, charting a path to improve success rates of chemotherapy/ICB combination protocols in the clinic. Since the proposed new protocol relies on low-dose intratumoral delivery of a common chemotherapeutic drug combined with checkpoint blockade, which is currently one of the most successful modes of anticancer therapy, potential translatability of this combination protocol to the clinic warrants thorough investigation.

DATA AVAILABILITY STATEMENT
The raw sequencing data from this study have been submitted to NCBI Sequence Read Archives (SRA) under accession number PRJNA734588.

ETHICS STATEMENT
The animal study was reviewed and approved by the Animal Care and Use Committee of the University of Texas Medical Branch.

AUTHOR CONTRIBUTIONS
EE and DT conceptualized and developed the study. EE developed the experimental design. FM performed the experiments and data analyses. SW designed and conducted the sequencing data analyses and interpretation. EE analyzed and interpreted the data and wrote the manuscript. All authors contributed to the article and approved the submitted version.

ACKNOWLEDGMENTS
We thank Jill Thompson and the University of Texas Medical Branch Next Generation Sequencing Core for performing the sequencing.