brca2-mutant zebrafish exhibit context- and tissue-dependent alterations in cell phenotypes and response to injury

Cancer cells frequently co-opt molecular programs that are normally activated in specific contexts, such as embryonic development and the response to injury. Determining the impact of cancer-associated mutations on cellular phenotypes within these discrete contexts can provide new insight into how such mutations lead to dysregulated cell behaviors and subsequent cancer onset. Here we assess the impact of heritable BRCA2 mutation on embryonic development and the injury response using a zebrafish model (Danio rerio). Unlike most mouse models for BRCA2 mutation, brca2-mutant zebrafish are fully viable and thus provide a unique tool for assessing both embryonic and adult phenotypes. We find that maternally provided brca2 is critical for normal oocyte development and embryonic survival in zebrafish, suggesting that embryonic lethality associated with BRCA2 mutation is likely to reflect defects in both meiotic and embryonic developmental programs. On the other hand, we find that adult brca2-mutant zebrafish exhibit aberrant proliferation of several cell types under basal conditions and in response to injury in tissues at high risk for cancer development. These divergent effects exemplify the often-paradoxical outcomes that occur in embryos (embryonic lethality) versus adult animals (cancer predisposition) with mutations in cancer susceptibility genes such as BRCA2. The altered cell behaviors identified in brca2-mutant embryonic and adult tissues, particularly in adult tissues at high risk for cancer, indicate that the effects of BRCA2 mutation on cellular phenotypes are both context- and tissue-dependent.

www.nature.com/scientificreports/ nuclear margins, suggesting aberrant aggregation and distribution of chromosomal material in brca2 m/m;tp53 m/m zebrafish oocytes (Fig. 1d, arrowheads). Less commonly, brca2 m/m;tp53 m/m oocytes exhibited massive consolidation of nuclear material, often in association with oocyte degeneration (Fig. 1e). Infrequent binucleation was identifiable in mitotic oogonia and all stages of meiotic oocytes from brca2 m/m;tp53 m/m zebrafish (Fig. 1f). The extent of nuclear abnormalities in brca2 m/m;tp53m/m ovaries was most variable in stage III oocytes (compare panels in Fig. 1d-f). www.nature.com/scientificreports/ brca2 mutation is associated with exhibit aberrant adult cell proliferation in a cancer-prone tissue. To determine brca2-associated effects on adult cell phenotypes prior to cancer onset, we focused on the optic nerve pathway (ONP). The natural, well-defined anatomic boundaries of this tissue are ideal for achieving consistent tissue collection, orientation, and histologic sectioning between specimens. Furthermore, the ONP is a predilection site for sarcoma development in both tp53 m/m and brca2 m/m;tp53 m/m zebrafish [25][26][27] , with brca2 m/m;tp53 m/m zebrafish at significantly increased risk for ocular tumors compared to tp53 m/m zebrafish 27 . We performed a time-course analysis of the ONP in wild type, tp53 m/m, and brca2 m/m;tp53 m/m between the ages of three and seven months (the mean age at tumor onset in brca2 m/m;tp53 m/m zebrafish is 8.7 months 25 ). During this period, we noted an abnormality of the choroid rete (CR) in brca2 m/m;tp53 m/m zebrafish. The CR is a vascular plexus located subjacent to the retinal choroid that forms a countercurrent capillary system 45 (Fig. 2a). It is derived from the ophthalmic artery and vein and contributes to maintaining oxygen pressure in the retina 46 . In routine hematoxylin and eosin sections, vascular channels of the normal CR are filled with red blood cells and the cellular meshwork forming this structure is largely obscured (Fig. 2b). However, the CR in brca2 m/m;tp53 m/m zebrafish frequently contained a population of robust spindle cells that were readily apparent between vascular channels (Fig. 2c). The incidence of this lesion progressively increased over time (Fig. 2d). Notably, this lesion was detectable in all brca2 m/m;tp53 m/m specimens after 5.1 months of age. In comparison, incidence was lower and age at onset was higher in tp53 m/m zebrafish (Fig. 2d). No atypical spindle cells were detected in the CR of wild type zebrafish at any time point examined. Similarly, no atypical spindle cells were present in the CR of brca2 m/m zebrafish without concomitant tp53 mutation (Table S4).  www.nature.com/scientificreports/ Optic nerve injury (ONI) induces an enhanced proliferative response in brca2 m/m;tp53 m/m zebrafish. As described above, the ONP is a cancer predilection site in tp53 m/m and brca2 m/m;tp53 m/m zebrafish [25][26][27] and is also notable for unique properties that support complete regeneration of the injured retina and optic nerve 32,33,[36][37][38] . We therefore sought to determine how proliferative and neoplastic phenotypes in the ONP might relate to injury response and regenerative capacity. We first assessed the short-term effects of ONI by performing unilateral ONI in wild type (n = 11), tp53 m/m (n = 11), and brca2 m/m;tp53 m/m (n = 14) zebrafish and assessing the injury response at three days and two weeks post-injury (Fig. 3a, Fig. S1a-c, Tables S2-S3). At both three day and two week time points, the total cellularity of the injured optic nerve was significantly increased compared to the uninjured optic nerve in brca2 m/m;tp53 m/m zebrafish (p = 0.0001 and p < 0.0001, respectively; Fig. 3b,c and Table S2). In comparison, total cellularity was significantly increased in wild type and tp53 m/m zebrafish only at two weeks post-injury (p = 0.0464 and p = 0.0130, respectively; Fig. 3b,c and Table S2). Comparison between genotypes indicated that the injury effect in brca2 m/m;tp53 m/m was significantly greater than in wild type or tp53 m/m cohorts at both three days and two weeks post-injury (Fig. 3b,c and Table S2).
To identify specific cell types that contribute to the increased cellularity observed in injured optic nerves, we performed a series of quantitative analyses by immunohistochemistry and in situ hybridization (details of statistical analyses are in Tables S2 and S3). First, we analyzed injured and uninjured optic nerves from four zebrafish at each time point (three days and two weeks post-injury) for the presence of stem and progenitor cell populations. We assessed the expression of blbp, a marker for radial glial cells 47 (Fig. 3d); sox2, a marker for neural stem cells 48 (Fig S2a); and sox10, a marker for neural crest progenitor cells 49,50 , oligodendrocytes and oligodendrocyte precursors 51 , and Schwann cells and Schwann cell precursors 52 ( Fig S2d). As blbp expression is cytoplasmic, we could not reliably identify and count individual blbp-expressing cells. We therefore quantified the total area of blbp expression in the optic nerve. At three days post-injury, blbp expression was significantly increased in the injured optic nerve compared to the uninjured optic nerve in all cohorts (wild type, p = 0.0039; tp53 m/m, p = 0.0179; brca2 m/m;tp53 m/m, p = 0.0001; Fig. 3e). brca2 m/m;tp53 m/m exhibited a sustained and significant increase in blbp expression at two weeks post-injury that was not observed in wild type or tp53 m/m zebrafish (p = 0.0071; Fig. 3f). On the other hand, neither sox2-expressing cells (Fig. S2b,c) nor sox10-expressing cells (Fig. S2e,f) appeared to contribute to the significant increases in cellularity observed in the injured optic nerve in brca2 m/m;tp53 m/m zebrafish. There were generally no significant differences observed in comparisons between genotypes, indicating that the significant injury effect observed in brca2 m/m;tp53 m/m zebrafish based on total cellularity (Fig. 3b,c) was not attributable to a single cell type.
Next, we analyzed injured and uninjured optic nerves for the presence of inflammatory and reactive cell populations. We assessed the expression of lcp1, a marker for monocytes and macrophages 53 (Fig. 3g), and krt18, a marker for reactive astrocytes in the ONP after injury 54 (Fig. S2g). ln addition to monocyte/macrophage populations, lcp1 is reportedly expressed by microglial cells in zebrafish 55 . However, we were unable to detect lcp1-positive microglial cells in sections of zebrafish brain and therefore considered lcp1 as a marker for monocytes and macrophages. Both lcp1 and krt18 are expressed in the cytoplasm, and therefore the total area of lcp1 or krt18 expression was quantified similar to blbp expression. At both three days and two weeks post-injury, lcp1 expression was significantly increased in the injured optic nerve only in brca2 m/m;tp53 m/m zebrafish (p = 0.0374 and p = 0.0372, respectively; Fig. 3h,i). In comparison, krt18 expression were not significantly different in the injured versus uninjured optic nerves at most time points for any of the three cohorts (Fig. S2h,i).
Finally, we assessed the CR in zebrafish that received ONI for the presence of aberrant spindle cells as observed in uninjured zebrafish. This cell population was present in 7 of 14 brca2 m/m;tp53 m/m zebrafish versus 1 of 11 tp53m/m zebrafish, and was not identified in any wild type zebrafish (Fig. 2e). When present, atypical spindle cells were identified at both three days and two weeks post-injury and there was no clear predilection for the injured or uninjured side. We further found that the atypical spindle cells identified in the CR were uniformly sox10-positive and were distributed both on the periphery and within the body of the CR (Fig. 2f). These expression patterns were similar regardless of whether the spindle cell population had arisen on the injured or uninjured side. In the normal CR, low numbers of sox10-positive cells were present in the CR and were largely confined to the periphery (Fig. 2g).

ONI does not significantly increase ocular tumorigenesis, but affects the incidence and sidedness of ocular lesions in brca2 m/m;tp53 m/m versus tp53 m/m zebrafish. To determine how
injury and regenerative responses affect ocular tumorigenesis in zebrafish, we analyzed tumor development in brca2 m/m;tp53 m/m and tp53 m/m zebrafish that received unilateral ONI at five months of age. The ONI group was compared to brca2 m/m;tp53 m/m and tp53 m/m zebrafish from a previously reported cohort, designated as the control group 27 (see Methods and Table S4 for details). Tumor development in ONI and control groups is summarized in Table 1. First, we compared the proportion of ocular versus non-ocular tumors in the ONI group versus the control group and in ONI and control groups segregated by genotype. In each comparison, ONI was not associated with an increased proportion of ocular tumors (Table 1 and Table S1). Next, we compared the side of ocular tumor development in ONI versus control groups to determine whether there was a side predilection for ocular tumorigenesis in either population. In each comparison, there was no significant difference in the proportions of ocular tumors arising on the right side (ONI side), left side (non-ONI side), or bilaterally in ONI versus control groups (Table 1 and Table S1).
We have previously shown that both ocular and non-ocular tumors from brca2 m/m;tp53 m/m and tp53 m/m zebrafish are predominantly sarcomas that exhibit histologic and immunohistochemical features consistent with malignant peripheral nerve sheath tumor 25,29 . The immunohistochemical expression profile of these tumors is not affected by brca2 genotype 29 . Ocular tumors arising in both the ONI and control cohorts were histologically similar and consistent with our previous identification of these tumors as sarcomas with features of malignant www.nature.com/scientificreports/ peripheral nerve sheath tumor. To further characterize these tumors, we analyzed a subset of ocular tumors from ONI and control zebrafish for expression of blbp, sox2, and sox10 ( Fig. 4a and Fig. S3). Expression of these markers was similar in tumors derived from ONI and control cohorts. Semi-quantitative analysis of marker expression demonstrated that most tumors exhibited little or no expression of either blbp or sox2 ( Fig S3). However, tumors from both ONI and control groups exhibited strong and ubiquitous nuclear sox10 expression (Fig. 4a).
Since we had determined that brca2 m/m;tp53 m/m frequently develop an atypical spindle cell population in the choroid rete of the eye between 3 and 7 months of age (Fig. 2), we investigated the incidence of this lesion in older animals that were followed for tumor development. Because cross-sections of the head were not routinely collected from the control group (derived from a previous study not specifically focused on the ONP; see Methods), our analysis was limited to zebrafish from the ONI group. We first assessed the incidence of atypical spindle cells in the CR in zebrafish from the ONI group that did not develop ocular tumors. Similar to earlier time points, we found that a higher proportion of brca2 m/m;tp53 m/m zebrafish (n = 13 of 25, 52%) exhibited CR atypical spindle cells compared to tp53 m/m zebrafish (n = 12 of 35, 34%), although this increase was not statistically significant (Table S1).
We next compared the side for development of ocular tumors or CR atypical spindle cells in brca2 m/m;tp53 m/m zebrafish versus tp53 m/m from the ONI group (Fig. 4b). There was no significant difference in the proportions of ocular tumors or CR atypical spindle cells arising on the ONI side, non-ONI side, or bilaterally between genotypic groups (Table S1). However, we noted two differences in ocular tumorigenesis versus CR atypical spindle cells in these analyses. First, CR atypical spindle cells in both tp53 m/m and brca2 m/m;tp53 m/m zebrafish were more frequently identified on the ONI side compared to the non-ONI side, in contrast to ocular tumors (Fig. 4b). Second, brca2 m/m;tp53 m/m zebrafish exhibited a relatively greater proportion of bilateral CR atypical spindle cells compared to the proportion of bilateral ocular tumors (Fig. 4b). Lastly, we compared the overall incidence and sidedness of ocular lesions (ocular tumor or CR atypical spindle cells) in brca2 m/m;tp53 m/m zebrafish versus tp53 m/m from the ONI group (Fig. 4c). After ONI, brca2 m/m;tp53 m/m zebrafish were significantly more likely to exhibit an ocular lesion than tp53 m/m zebrafish, and ocular lesions, when present, were more often bilateral in brca2 m/m;tp53 m/m zebrafish (p = 0.0220 and p = 0.1207, respectively; Table S1).

Discussion
The exploration of noncancerous cell phenotypes associated with mutations in cancer susceptibility genes can provide important insights into how such mutations affect cell behaviors and responses to stimuli. In the current study, we analyzed several noncancerous cell phenotypes linked to BRCA2 mutation using a zebrafish model. BRCA2 is required for error-free resolution of double-strand DNA breaks by homologous recombination in both mitotic and meiotic cells, and also participates in processes such as replication fork protection and R loop processing (reviewed in [56][57][58]. Heritable BRCA2 mutations are associated with significantly increased risk for several cancer types in humans, including breast, ovarian, prostate, and pancreatic cancer 56 , and impaired capacity for homologous recombination has been identified more broadly across multiple human cancers ("BRCAness") [59][60][61] .
The processes of embryonic development and the inflammatory/injury response can be partially recapitulated during carcinogenesis, as molecular and cellular programs that are activated during these processes can be coopted by cancer cells [1][2][3][4][5] . Furthermore, these normal processes can be influenced by cancer-associated genetic mutations and thereby directly contribute to cancer initiation and progression. For example, pancreatic cancer can be induced by the cooperating effects of KRAS mutation, inflammation, and tissue injury 7 , which has been attributed to specific epigenetic alterations that are uniquely driven by these combined genetic and microenvironmetal factors 62 . Cancer initiation is also associated with the activation of molecular signaling pathways that normally function during embryogenesis. However, developmental effects caused by mutation or loss of cancerassociated genes are often very different than adult phenotypes, as exemplified by early embryonic lethality versus adult cancer susceptibility [11][12][13][14][15][16][17][18][19][20][21][22] .
Determining how cancer-causing genetic mutations affect adult versus embryonic cell populations is likely to reveal distinct signaling pathways that underlie these phenotypes and are of significant relevance to cancer initiation (i.e., cell proliferation versus cell death). Unfortunately, early embryonic lethality in mouse models for cancer-associated genes such as BRCA2 is a confounding factor. In these cases, zebrafish can provide an excellent complementary model, as they exhibit conserved genetic susceptibility to cancer for many genes [63][64][65][66][67][68] . The brca2-mutant zebrafish model is fully viable in the homozygous condition and captures the collaborative effects of brca2 and tp53 mutations in carcinogenesis that characterize human BRCA2-associated cancers 25,27,28 . We thus used this model system to determine how brca2 mutation affects cell phenotypes during embryogenesis and in  www.nature.com/scientificreports/ the response to tissue injury. As the optic nerve pathway (ONP) is a predilection site for cancer development in brca2 m/m;tp53 m/m zebrafish, evaluations of the injury response in adult animals focused on this tissue. We first assessed the role for maternally provided mRNA for brca2 during early embryonic development in zebrafish. The maternal-to-zygotic transition is characterized by the degradation of maternal mRNA and onset of zygotic gene activation (ZGA) 69 . While mice initiate ZGA at the 1-cell stage and clear most maternal mRNA  www.nature.com/scientificreports/ by the 2-cell stage, zebrafish do not undergo these processes until the mid-blastula transition at cleavage cycle 10 69-71 . These differences in availability of maternally provided transcripts may be a factor in survival of brca2mutant zebrafish embryos. We have previously shown that zebrafish embryos possess abundant maternal mRNA for brca2 25 , and brca2 is both maternally and zygotically expressed in early-stage embryos 70 . In accordance with this, we show here that zebrafish embryos lacking maternally provided mRNA for full-length brca2 exhibit developmental arrest and death at approximately mid-blastula stage. However, detailed analysis of ovaries from female brca2 m/m;tp53 m/m zebrafish reveal abnormalities in developing oocytes that could also contribute to this embryonic phenotype. These include both aberrant localization of nuclear content and evidence for cytokinetic defects; the former observation has been reported in another brca2-mutant zebrafish model, confirming a brca2specific effect on oocyte development 72 . Beyond the canonical role for BRCA2 in dsDNA break repair, in vitro analyses in mitotic cells indicate that BRCA2 participates in cytokinesis 73 and chromosomal alignment/segregation 74,75 . Less is known about the role for BRCA2 in meiosis due to the difficulty in establishing Brca2-knockout mouse models, although meiosis-specific binding partners required for BRCA2 localization to chromosomes were recently characterized [76][77][78] . However, mouse models with oocyte-specific reduction in Brca2 expression 43 or Brca2 deletion 42 displayed nuclear abnormalities suggesting errors in chromosomal localization in oocytes. We identified unfertilized eggs by the absence of cell division. We cannot exclude the possibility that some portion of these eggs may have been fertilized, but failed to initiate cleavage due to severe genetic defects. Given that zebrafish eggs can undergo cell divisions even in the absence of nuclear material 79 , it seems unlikely that genetic perturbations associated with brca2 deficiency would block cell cleavage. Also, eggs derived from zebrafish with genetic mutations that significantly affect genomic integrity (e.g., mlh and mps1) can be fertilized and undergo cell divisions despite demonstrably severe genetic aberrations 80,81 . Loss of functional BRCA2 is likely to disrupt meiotic progression and early embryonic development through multiple mechanisms. Further studies will be required to segregate and clarify the functions of BRCA2 in meiotic oocytes versus early-stage embryos.
We next determined that loss of functional brca2 in adult zebrafish induces aberrant proliferative responses in the ONP, which is a highly cancer-prone tissue in brca2 m/m;tp53 m/m zebrafish. We identified an anomalous spindle cell population that was highly prevalent in the choroid rete of brca2 m/m;tp53 m/m zebrafish in the unperturbed ONP prior to cancer onset. This population was not identified in brca2 m/m zebrafish without tp53 mutation, consistent with low tumor incidence in this genotypic group 25,28 . In humans, BRCA2-associated tumors exhibit frequent TP53 mutation, which suggests that altered or lost P53 function may be critical for BRCA2-associated carcinogenesis in both humans and zebrafish 82,83 . Uniform sox10 positivity suggests that the choroid rete spindle cells are of neural crest, oligodendroglial, or Schwann cell origin. Our current and prior 29 immunohistochemical analyses of ocular tumors in brca2 m/m;tp53 m/m zebrafish are consistent with malignant peripheral nerve sheath tumor (MPNST) and demonstrate widespread sox10 expression in tumors, supportive of Schwann cell origin. The zebrafish choroid, and presumably the choroid rete, contains small myelinated nerve processes that are the likely source for Schwann cells in this location 84 . We therefore hypothesize that tumors in the optic nerve pathway in our model arise from this aberrantly proliferative Schwann cell population. In comparison, conditional deletion of Brca2 in mouse prostatic epithelium induces epithelial hyperplasia and low-grade prostate intraepithelial neoplasia (PIN) that is exacerbated by concurrent Tp53 mutation 85 . On the other hand, Brca2 knockout in mouse T lymphocytes causes a decline in T cell numbers over time 86 . These data from zebrafish and mouse models suggests that BRCA2 mutation or loss affects different cell types differently, and can enhance the growth of certain noncancerous cell populations in specific tissues/contexts. An important next step will be to determine why a particular microenvironment promotes cell proliferation and subsequent cancer initiation in the context of heritable BRCA2 mutation.
We subsequently assessed the injury response in the cancer-prone ONP, since numerous studies have demonstrated key similarities between injury responses and cancer progression at the molecular, cellular, and tissue level [1][2][3]6 . The ONP in zebrafish is uniquely supportive of complete optic nerve regeneration due in part to a permissive microenvironment that supports axonal regrowth. We therefore speculated that cellular responses to optic nerve injury (ONI) might differ in cancer-prone versus non-cancer-prone individuals, and that cancer predisposition might be related to regenerative capacity in this pro-growth environment. In short-term studies, the proliferative response to ONI was significantly enhanced in brca2 m/m;tp53 m/m zebrafish compared to tp53 m/m or wild type cohorts. This included both progenitor cells (radial glia) and inflammatory cells (monocytes/ macrophages); other cell populations that were not investigated here may have added to the overall increase in cellularity. Comparisons of some cell populations did not reach statistical significance, although trends in the data were apparent (Fig. S2). A larger study population will be informative in addressing these potential differences in the injured versus uninjured nerve. Interestingly, cardiomyocyte proliferation during heart regeneration in zebrafish increases in the context of homozygous tp53 mutation 87 . Similar p53-associated effects on cell proliferation are described during early stages of limb regeneration in salamanders 88 . We are not aware of any studies that test the role for BRCA2 in vertebrate regeneration; however, the orthologue for BRCA2 contributes to axonal regeneration in the nematode Caenorhabditis elegans 89 . We also noted some differences in cellular responses to ONI in the current study, e.g., sox10-expressing cells, compared to other reports of optic nerve injury in fish 90,91 . These differences reflect variations in which portion of the injured optic nerve was analyzed, and may also be affected by differences in analytical time points. Together these studies indicate that injury responses are altered by mutations in cancer-associated genes in vertebrate animals.
The potential for cancers to arise from regenerating cell populations in vertebrates is variable. In salamanders and newts, regenerative tissues are highly resistant to chemical carcinogenesis and malignant transformation is suppressed [92][93][94] . Here we found no direct effect (positive or negative) on tumorigenesis following ONI in zebrafish with heritable brca2 and tp53 mutations. Although we did note that bilateral ocular lesions were more common in brca2 m/m;tp53 m/m zebrafish after ONI than might be expected based on the incidence of bilateral ocular tumors in controls, overall our results do not indicate that ONI enhances carcinogenesis in this model. www.nature.com/scientificreports/ In contrast, a zebrafish model for KRAS G12V -driven melanoma subjected to repeated cycles of tail amputation and regeneration developed melanoma at the resection site 95 . The differences in regeneration-associated tumorigenesis in KRAS G12V zebrafish versus brca2-mutant/tp53-mutant zebrafish could reflect the relative impact of chronic repeated injuries versus a single injury event on tumor initiation. Alternatively, differences in the regenerative process might affect tumorigenic potential in zebrafish. While ONI is resolved by regeneration of axonal fibers from surviving retinal ganglion, tail resection is resolved by the more complex process of epimorphic regeneration 31 . Epimorphic regeneration requires repatterning and regrowth of multiple tissue types and is achieved via dedifferentiation and subsequent redifferentiation of mature cell populations 96 . BRCA2 mutations are infrequently reported in human soft tissue sarcomas. However, human sarcomas such as MPNST show evidence of "BRCAness", including karyotypic complexity, frequent alterations of DNA repair genes, and sensitivity to PARP-inhibitors that includes significant genomic instability and deficient DNA damage repair [97][98][99][100] . The proliferative and preneoplastic phenotypes we report in this study may have important parallels in human sarcomagenesis, and further investigation into the relationship between DNA repair deficiency and sarcoma initiation in mammalian species is warranted.
In summary, we demonstrate that phenotypes linked to BRCA2 mutation in mammals, ranging from early embryonic death to cancer predisposition, are captured in the brca2-mutant zebrafish model. We find that brca2associated embryonic lethality is likely to reflect a combination of cellular defects that arise during both mitosis (oogonia, embryos) and meiosis (oocytes). We also identify expansion of several adult cell populations arising under basal conditions and during the post-injury response in a tissue at high risk for cancer onset, including a putative precancerous population. These studies confirm stage-and context-dependent roles for BRCA2 in cell survival and growth that are highly relevant to BRCA2-associated carcinogenesis.

Materials and methods
Zebrafish study cohorts. Experiments were performed with adult wild type (AB) zebrafish and adult zebrafish from the brca2 hg5 and tp53 zdf1 mutant zebrafish lines, corresponding to brca2 Q658X and tp53 M214K mutations 25,26 . Mutant alleles are hereafter referred to as "m". Details of the experimental groups are included in Table S4. Groups including tp53 m/m and brca2 m/m;tp53 m/m were comprised of siblings genotyped for presence or absence of the brca2 Q658X mutation. For analysis of oocyte morphology, groups consisted of age-matched wild type, tp53 m/m, and brca2 m/m;tp53 m/m female zebrafish. For analyses of injury response and tumorigenesis, the ONI group consisted of age-matched cohorts, and zebrafish with brca2 and tp53 mutations were siblings derived from two clutches. For analysis of tumorigenesis, the control group consisted of zebrafish siblings with brca2 and tp53 mutations derived from a single clutch. The control group was previously described in a separate study analyzing tumor ploidy 27 . All animal studies were approved by the Institutional Animal Care and Use Committee, North Carolina State University, Raleigh, NC, performed in accordance with approved protocols, and complied with ARRIVE guidelines.
Zebrafish husbandry and genotyping. Zebrafish used in this study were raised as previously described 27 on a Pentair Z-Hab Duo recirculating aquaculture system and maintained on a 14-h light/10-h dark cycle. The zebrafish colony undergoes routine sentinel testing for infectious organisms and is negative for known zebrafish pathogens. Zebrafish were monitored for gross evidence of tumor development and humanely euthanized with Tricaine methanesulfonate (300 mg/L) in system water buffered with Sodium Bicarbonate to a pH of ~ 7.0 when tumors were visibly apparent. Live adult zebrafish were genotyped for the brca2 Q658X mutation at three months of age by sequencing over the mutation site as previously described 28 . Zebrafish with tp53 zdf1 mutation were maintained as a homozygous mutant line.
Optic nerve injury. Optic nerve injury was performed as previously described 101-103 on randomly selected and anesthetized 5-month-old zebrafish placed in left lateral recumbency on a wet sponge under a stereomicroscope. The right eye was gently displaced, and the optic nerve was crushed with micro-forceps. All zebrafish received equivalent injury and were recovered in system water. Zebrafish were collected and euthanized at 3 days post-injury for short-term analyses and at 2 weeks after injury for long-term analyses. Zebrafish were observed for tumor development for up to 19 months of age.
Tissue collection, immunohistochemistry, and in situ hybridization. Zebrafish from the ONI group were decapitated caudal to the gills after humane euthanasia was performed as described above. Heads were placed in 4% paraformaldehyde for 18-24 h, decalcified in 12% EDTA for two days, and transferred to 70% ethanol. If zebrafish were to be used for tumorigenesis studies, the body was similarly processed. Tissue specimens were embedded in paraffin, and unstained or hematoxylin and eosin stained transverse sections were prepared by the Histology Laboratory, NC State University, College of Veterinary Medicine so that both optic nerves were in the plane of section. Zebrafish tissues from the tumor-bearing control group and from female wild type, tp53 m/m, and brca2 m/m;tp53 m/m zebrafish were collected and processed as previously described 27 .
Imaging and image analysis. For  . The injured nerve and contralateral uninjured nerve were analyzed for each specimen. Quantitative analyses were performed using either a single digital image or multiple aligned digital images captured from hematoxylin and eosin-stained sections. Images were minimally and globally processed with the GNU Image Manipulation Program, version 2.8.6 (http:// www. gimp. org/) and the line tool was used to outline the optic nerve area in each tissue section. Quantitation of total cellularity, total positive cells (sox2, sox10), or percent positive area (blbp, lcp1, krt18) was performed with ImageJ, using the outlined area for each optic nerve to define a region of interest. Total cellularity was determined using the ImageJ Fiji Cell Counter tool (https:// fiji. sc/). Red blood cells were excluded from cell counts based on their appearance as ovoid, nucleated cells with brightly eosinophilic cytoplasm. Expression of blbp, sox2, sox10, lcp1, and krt18 were determined with the IHC Toolbox plugin for ImageJ (https:// imagej. nih. gov/ ij/ plugi ns/). A positive control specimen was used during the training process to generate a model that identified positive pixels in digital images for each marker. Since there was no positive control for krt18, the model generated for lcp1 expression was used because krt18 and lcp1 expression were detected with the same chromogen. sox2 and sox10 expression were determined by quantifying the total cells versus sox2-or sox10-expressing cells, based on nuclear expression of these markers, and calculating the ratio of positive cells to total cells within the outlined nerve. blbp, lcp1, and krt expression were determined by quantifying the percent positive area within the total area of the outlined nerve. Histologic analyses and imaging of tissue specimens were performed by a board-certified veterinary pathologist (HRS) using an Olympus BX43 light microscope with Olympus DP27 digital camera and Olympus cellSens Imaging Software. Images were minimally and globally processed with the GNU Image Manipulation Program, version 2.8.6 (http:// www. gimp. org/). Semi-quantitative analyses of sox2, sox10, and blbp expression were performed in a subset of tumors arising on the right (ONI) side and left (non-ONI) side from optic nerve injury and control cohorts. Tumor expression of these markers was analyzed and scored with an Olympus BX51 light microscope by a single investigator (VAK). Marker expression was scored as a percentage of total tumor tissue in each section (0-25%, 25-50%, 50-75%, or 75-100%) by visual assessment of the entire tumor at 40×, 100×, and 200× magnification.
Embryo phenotyping. Embryos were derived from tp53 m/m and brca2 m/m;tp53 m/m female zebrafish outcrossed to fertile wild type males in two independent experiments. Zebrafish were maintained overnight in breeding chambers without dividers in groups of three to four females per two males, and eggs were collected the following morning upon cessation of breeding behavior. Every egg derived from each clutch was assessed using a Nikon SMZ1000 stereomicroscope and counted as either fertilized (intact egg undergoing cell division), unfertilized (intact egg without cell division), or inviable (degenerate egg). Fertilized eggs were sorted into 100 mm Petri dishes in egg water (60 ug/ml "Instant Ocean" sea salts and 0.0002% methylene blue in distilled water) at a density of up to 55 embryos per dish and incubated at 28 °C degrees in a dedicated incubator. Embryos were periodically observed at zero days post-fertilization to assess developmental progress. At one day post-fertilization, embryos were scored as exhibiting normal phenotype, abnormal phenotype, or inviable using established staging criteria 104 . For one group of embryos derived from tp53 m/m females, 35 fertilized embryos were removed from a total of 522 fertilized embryos on day 0 for an unrelated experiment and are not included in the total on day 1.

Criteria for exclusion.
Individual zebrafish that were (1) found dead; (2) lost the right eye after injury or (3) had histological evidence of unusually severe tissue damage after ONI were excluded from analysis. In addition, specimens for which both optic nerves or both choroid rete could not be identified in tissue sections were excluded from the relevant analyses. See Table S4 for additional details.
Statistical analyses. Statistical analyses were performed using SAS software version 9.4 (SAS Institute Inc., Cary, NC) with statistical significance set at an alpha value of p ≤ 0.05. Comparisons of cellularity and marker expression (sox2, sox10, blbp, lcp1, krt18) were performed using a mixed effect model. Fisher's Exact test was used to compare population proportions for the following assessments: tumor location; sidedness of ocular tumors; presence of atypical spindle cells in the choroid rete; presence of any ocular lesion; sidedness of ocular lesions. Details of statistical analyses and outcomes are shown in Table S1-S3.