Mitotic Errors Promote Genomic Instability and Leukemia in a Novel Mouse Model of Fanconi Anemia

Fanconi anemia (FA) is a disease of genomic instability and cancer. In addition to DNA damage repair, FA pathway proteins are now known to be critical for maintaining faithful chromosome segregation during mitosis. While impaired DNA damage repair has been studied extensively in FA-associated carcinogenesis in vivo, the oncogenic contribution of mitotic abnormalities secondary to FA pathway deficiency remains incompletely understood. To examine the role of mitotic dysregulation in FA pathway deficient malignancies, we genetically exacerbated the baseline mitotic defect in Fancc-/- mice by introducing heterozygosity of the key spindle assembly checkpoint regulator Mad2. Fancc-/-;Mad2+/- mice were viable, but died from acute myeloid leukemia (AML), thus recapitulating the high risk of myeloid malignancies in FA patients better than Fancc-/-mice. We utilized hematopoietic stem cell transplantation to propagate Fancc-/-; Mad2+/- AML in irradiated healthy mice to model FANCC-deficient AMLs arising in the non-FA population. Compared to cells from Fancc-/- mice, those from Fancc-/-;Mad2+/- mice demonstrated an increase in mitotic errors but equivalent DNA cross-linker hypersensitivity, indicating that the cancer phenotype of Fancc-/-;Mad2+/- mice results from error-prone cell division and not exacerbation of the DNA damage repair defect. We found that FANCC enhances targeting of endogenous MAD2 to prometaphase kinetochores, suggesting a mechanism for how FANCC-dependent regulation of the spindle assembly checkpoint prevents chromosome mis-segregation. Whole-exome sequencing revealed similarities between human FA-associated myelodysplastic syndrome (MDS)/AML and the AML that developed in Fancc-/-; Mad2+/- mice. Together, these data illuminate the role of mitotic dysregulation in FA-pathway deficient malignancies in vivo, show how FANCC adjusts the spindle assembly checkpoint rheostat by regulating MAD2 kinetochore targeting in cell cycle-dependent manner, and establish two new mouse models for preclinical studies of AML.

Fanconi anemia (FA) is a disease of genomic instability and cancer. In addition to DNA damage repair, FA pathway proteins are now known to be critical for maintaining faithful chromosome segregation during mitosis. While impaired DNA damage repair has been studied extensively in FA-associated carcinogenesis in vivo, the oncogenic contribution of mitotic abnormalities secondary to FA pathway deficiency remains incompletely understood. To examine the role of mitotic dysregulation in FA pathway deficient malignancies, we genetically exacerbated the baseline mitotic defect in Fancc-/-mice by introducing heterozygosity of the key spindle assembly checkpoint regulator Mad2. Fancc-/-;Mad2+/-mice were viable, but died from acute myeloid leukemia (AML), thus recapitulating the high risk of myeloid malignancies in FA patients better than Fancc-/mice. We utilized hematopoietic stem cell transplantation to propagate Fancc-/-; Mad2+/-AML in irradiated healthy mice to model FANCC-deficient AMLs arising in the non-FA population. Compared to cells from Fancc-/-mice, those from Fancc-/-;Mad2+/-mice demonstrated an increase in mitotic errors but equivalent DNA cross-linker hypersensitivity, indicating that the cancer phenotype of Fancc-/-;Mad2+/-mice results from error-prone cell division and not exacerbation of the DNA damage repair defect. We found that FANCC enhances targeting of endogenous MAD2 to prometaphase kinetochores, suggesting a mechanism for how FANCC-dependent regulation of the spindle assembly checkpoint prevents chromosome mis-segregation. Whole-exome sequencing revealed similarities between human FA-associated myelodysplastic

INTRODUCTION
Fanconi anemia (FA) is a heterogeneous genetic disorder characterized by bone marrow failure, developmental abnormalities, and cancer predisposition (1). Germline mutations in any one of the 22 genes that encode FA pathway proteins causes Fanconi Anemia, whereas monoallelic inactivation of specific FA genes, such as BRCA1 and BRCA2, results in an adult-onset predisposition for breast, ovarian, and pancreatic cancer (1). Furthermore, inactivation of FA pathway proteins via somatic mutation or epigenetic silencing is now being recognized as a key event contributing to the development of several types of sporadic cancers, including more than 40% of acute myelogenous leukemias (AML) (1)(2)(3).
The increased risk of malignant transformation in the setting of FA-deficiency is ultimately a consequence of increased genomic instability. However, the specific mechanisms underlying the origins of this genomic instability are incompletely understood (4). The best characterized role of the FA pathway is in the detection and repair of DNA damage (1). Although several mouse models have been developed to study the molecular pathogenesis of FA, most have been unable to precisely recapitulate the hematologic malignancies of human FA patients [reviewed in (5)]. Fancc-/-mice demonstrate many of the phenotypic characteristics consistent with FA pathway deficiency, including hypersensitivity to DNA cross-linking agents and impaired hematopoietic stem cell repopulating capacity (6,7). However, our group previously demonstrated that while aged Fancc-/-mice eventually develop hematologic malignancies consistent with human AML, they do not develop early onset disease like their human FA counterparts. Importantly, we found that leukemogenesis in Fancc-/-mice was preceded by an increase in abnormal mitotic figures and aneuploidy, which was noted to be exacerbated upon malignant transformation (8). Consistent with these observations, we have previously shown that loss of FA pathway proteins results in a weakening of the spindle assembly checkpoint (SAC), a complex signaling cascade that delays anaphase onset until each sister chromatid is properly attached to the mitotic spindle (4,9). These findings suggest that defects in both DNA damage repair and mitotic fidelity may contribute to progressive genomic instability in the setting of FA pathway compromise.
Accordingly, emerging evidence suggests that impaired DNA damage repair and compromised mitotic fidelity may cooperate to facilitate tumorigenesis (10). Specifically, previous studies have suggested that FA pathway disruption and SAC inactivation cooperate to promote malignant transformation.
In their elegant study, Lee et al., demonstrated that inactivation of the SAC through loss of BUB1 function promoted transformation in cells lacking BRCA2, a prevalent FA pathway protein (1,(11)(12)(13). Additionally, patients with FA pathway mutations are thought to be at an increased risk of developing HPV-associated cancers (14). Interestingly, a recent study demonstrated that the HPV E6 protein disrupts the SAC by causing a decrease in kinetochore-associated MAD2 (15). Given these results and our previous findings, we hypothesized that FA-deficient cells are hyper-dependent on the SAC and that further SAC compromise would be sufficient to drive leukemogenesis in the setting of FA pathway dysfunction.
To test this hypothesis, we generated a novel mouse model in which the weakened SAC of Fancc-/-mice was further impaired by direct genetic disruption. For this model, it was necessary to select a component of the SAC that could be manipulated without altering the gross structure of the kinetochore or other important mitotic pathways. Previous studies have suggested a role for the SAC protein BUB1B in promoting leukemogenesis (16,17). However, in addition to its role in the SAC, BUB1B is also known to be critical for chromosome alignment, end-on capture of microtubules, and chromosome congression (18)(19)(20). Likewise, the SAC protein BUB1 is required for kinetochore localization of BUB1B, centromere proteins CENP-E and CENP-F, as well as for proper chromosome congression and sister centromere cohesion (21,22). Lastly, the recruitment of BUB1 and BUB1B to the kinetochore is dependent upon BUB3 (23). Therefore, given their important SAC-independent mitotic functions, we excluded Bub1, Bub1b, and Bub3 from consideration for our model (24,25). Complete loss of the SAC component Mad2 results in embryonic lethality, but Mad2 heterozygous mice develop normally and are not known to develop cancer at a young age (26). Additionally, Mad2+/mice do not develop hematologic malignancies and develop solid tumors only after a long latency period, suggesting that additional events are required to drive tumorigenesis (26). Thus, we reasoned that introducing Mad2 heterozygosity into Fancc-/-mice would allow for the specific evaluation of the contribution of SAC insufficiency to leukemogenesis in the setting of Fancc-deficiency (6,27).
In this study, we strategically crossed Fancc-/-and Mad+/mice to further genetically compromise the SAC. As we hypothesized, heterozygous deletion of Mad2 accelerated the onset of hematologic malignancies in Fancc-/-mice, resulting predominantly in AML. Compared to other FA murine models, this phenotype more closely recapitulates the childhood AML predisposition of human FA patients (1). Importantly, the increased incidence of hematologic malignant transformation was associated with an exacerbation of the underlying SAC impairment, but not of the underlying DNA damage repair defect. This novel AML-prone mouse model provides a valuable tool for studying leukemogenesis in FA and provides in vivo evidence implicating SAC insufficiency as a driver of tumorigenesis in the setting of FA pathway deficiency.
Mice were genotyped as described (28). B6.SJL-Ptprc a Pepc b / BoyJ mice were purchased from IU In Vivo Therapeutics Core. All studies were performed in accordance with the policies of the IU IACUC.

Histology
Mouse tissues were fixed in 10% formalin, paraffin-embedded, sectioned (5mM sections), and stained with hematoxylin and eosin (H&E). The following antibodies were used: anti-c-kit (C19, Santa Cruz, 1:50), anti-cd11b (Novus, 1:50), CD3 (Dako, IR503), B220 (Clone RA3-6B2, BD Pharmingen). Peripheral blood smears and marrow cytospins were stained with Giemsa using the Siemens Hematek 3000 system. Diagnoses of leukemia and lymphoma in the Fancc-/-; Mad2+/-mice were made using Bethesda criteria for the non-lymphoid and lymphoid neoplasms in mice (29) and confirmed by a mouse veterinary pathologist at IU School of Medicine. Confirmation of leukemia posttransplant was determined by primarily by histology upon presence of leukemic infiltrates in non-hematopoietic organs (such as liver, kidneys, lungs) at time of necropsy, as well as presence of blasts in the peripheral blood, marrow and/or spleen.

Metaphase Spreads
Marrow cells were cultured in IMDM with 20% FBS, murine SCF (100ng/ml), and IL-6 (200ng/ml) for two days. Cells were then exposed to 0.2 mg/ml colcemid (Life Tech) for 4 hours and pelleted (800 rpm) for 5 minutes. Cells were resuspended dropwise in pre-warmed (37°C) 75mM KCl while vortexing gently and incubated at 37°C for 15 minutes. After pelleting, cells were resuspended in a 3:1 methanol: glacial acetic acid. Cells were pelleted and resuspended in fixative solution two additional times before being dropped onto slides and dried overnight. Spreads were then mounted in Vectashield medium with DAPI (Vector Laboratories). For the chromosome breakage test, cells were cultured for 48 hours in 50nM MMC before colcemid addition. For spectral karyotyping (SKY), samples were imaged and analyzed by the MD Anderson Cancer Center Molecular Cytogenetics Facility (Houston, TX).

Microscopy
Histology images were obtained using a Zeiss Axiolab system. Metaphase images were acquired on a Deltavision personalDx deconvolution microscope (Applied Precision). Image stacks (distance between z-sections: 200 nM) were deconvolved using Softworx suite (10 iterations; ratio: conservative) and analyzed using Imaris (Bitplane). For live mitosis studies, a Biostation liveimaging microscope (Nikon) at 37°C/5% CO 2 was used, with phase-contrast image stacks (z-sections distance: 2mm) captured in two-minute intervals for 24 hours. Videos were analyzed using NIS-Elements Viewer (Nikon).

Immunofluorescence for MAD2-KT Quantification
MEFs were grown under the above indicated conditions and fixed on glass coverslips (Fisher Scientific) in 4% paraformaldehyde/ 0.5% Triton X-100 in PBS for 10 minutes at room temperature. Coverslips were washed with PBS-T (0.03% Triton X-100), and then permeabilized with 0.5% Triton X-100/0.05% SDS in PBS for 30 minutes at room temperature. The coverslips were then blocked in 10% FBS/0.03% Triton X-100 in PBS for 1 hour at room temperature. Coverslips were stained with rabbit polyclonal antihuman Mad2L1 (1:100 in PBS, Life Span Biosciences; LS-B13367) and mouse monoclonal anti-Hec1 (1:100 in PBS, Santa Cruz Biotechnology; SC-515550) for 2 hours at room temperature, then washed three times with PBST-T, 5 minutes per wash. Coverslips were then incubated with fluorophore-conjugated secondary antibodies (1:2000 in PBS, Life Technologies) at room temperature. Coverslips were washed three times in PBS-T, 5 minutes each, rinsed once in distilled water, then mounted in ProLong Diamond antifade mountant with DAPI (Molecular Probes) and cured at room temperature at least 24 hours. The fluorescence signals were captured using a Deltavision personalDx deconvolution microscope (Applied Precision). Image stacks (distance between z-sections: 200 nM) were deconvolved using Softworx suite (10 iterations; ratio: conservative). Quantification of fluorescence intensity at kinetochores and subsequent statistical analysis were performed using Imaris (Bitplane) and GraphPad Prism, respectively.

Quantitative Western Blotting
Whole-cell lysates were prepared for Western blotting by incubating cells in M-PER mammalian protein extraction reagent (Life Technologies) with protease (Complete Mini, EDTA-free; Roche) and phosphatase inhibitors (Pierce Phosphatase Inhibitor Mini tablets; Thermo Scientific) on ice (10 min). Lysates were centrifuged at top speed in a microcentrifuge for 10 min. Before loading onto gels, samples were diluted with NuPAGE sample-reducing agent and NuPAGE lithium dodecyl sulfate (LDS) sample buffer (Life Technologies) and boiled (95°C, 5 min). Following protein separation on NuPAGE 4 to 12% polyacrylamide-bis-Tris gels (Life Technologies) and transfer to nitrocellulose, membranes were probed with primary antibodies. Primary antibodies used include monoclonal mouse anti-Mad2 (Santa Cruz Biotechnology; sc-374131), monoclonal mouse anti-actin (Sigma; A5441), monoclonal rabbit anti-BUB1 (Abcam; ab195268), monoclonal mouse anti-BUBR1 (Abcam; ab54894), monoclonal rabbit anti-BUB3 (Abcam; ab133699) and mouse anti-beta Tubulin (Invitrogen; A11126). Fluorescent dyeconjugated secondary antibodies (Li-Cor Biosciences) were used for infrared fluorescence-based detection (Odyssey CLX). Protein levels were quantified by measuring the relative fluorescence intensities of bands (normalized against actin) using Image Studio 2.1 software.

Whole-Exome Sequencing and CNV Evaluation
The sequencing and bioinformatics analysis was performed at Indiana University Center for Computational Biology and Bioinformatics Core. The Agilent SureSelect XT2 Mouse All Exon kit was used to prepare the whole exome sequencing libraries for the paired normal and tumor samples. Four Fancc-/-; Mad2+/-mouse malignancies were sequenced (mouse IDs 501, 504, 520 and 576). The capture/target regions of the Agilent SureSelect design were downloaded from the Agilent website (https://earray.chem.agilent.com/suredesign/). These target regions were lift-overed (http://hgdownload.soe.ucsc.edu/ admin/exe/linux.x86_64/liftOver) to mm10 coordinates since they were designed based on the mouse reference mm9. The resulting WES libraries were sequenced on Illumina HiSeq4000. 100 million and~200 million paired-end 75 bp reads were generated for the germline and tumor samples, respectively. The sequencing reads were first assessed for quality using FastQC (v.0.11.5, Babraham Bioinformatics, Cambridge, UK). Illumina adapter sequence and low-quality base calls were trimmed off the paired-end reads with Trim Galore (v0.4.3, http://www. bioinformatics.babraham.ac.uk/projects/trim_galore/). The remaining high-quality reads were next aligned to the mouse reference genome mm10 using BWA v7.15 (30). Duplicate reads were marked with Picard (https://broadinstitute.github.io/ picard/) using the specific duplicate pixel distance parameter for HiSeq4000. Genome Analysis Toolkit (GATK v 3.7) (31) was subsequently applied for simultaneous indel realignment for each pair of the samples, to use MuTect 1 for somatic variant identification. GATK was used further to perform a base quality recalibration. MuTect 1, MuTect 2 (32) and Strelka (33) were used to identify somatic variants in the SureSelect capture regions. The variants identified from different tools were combined with GATK CombineVariants. To annotate the variants with ANNOVAR (34), we built a gene annotation database for mm10 according to the instructions from ANNOVAR. dbsnp142 was used for the annotation. The variants were also annotated with SnpEff (35). For detection of copy number variation (CNV), ExomeCNV (36) and CNVkit (37) were used. Depth coverage files were generated with GATK DepthOfCoverage for analysis with ExomeCNV. For the analysis with CNVkit, the blacklist regions for mm10 (downloaded from https://www.encodeproject.org/annotations/ENCSR636HFF/ by choosing mm10) were excluded in the access file.

Evaluation of Predicted Variant Impact on Protein Function and Pathway Analysis
The Ensembl Variant Effect Predictor (VEP) and SIFT (Sorting Intolerant from Tolerant) bioinformatic algorithms (38,39) were employed to evaluate genetic variants identified via WES. Variants (i) predicted by VEP to have a moderate or high impact, (ii) predicted by SIFT to be harmful to protein function and (iii) not present in reference mouse genomes as SNPs were selected for further analysis. The PANTHER bioinformatics strategy (40) was used for overrepresentation tests (GO biological process complete) using GO Ontology database (released 2018-08-09) with all Mus musculus genes serving as a reference. Fisher's Exact with FDR (false discovery ratio) multiple test correction was used to determine the statistical significance of candidate pathway enrichment.

Statistical Analysis
All statistical analyses were performed using GraphPad Prism. The specific statistical test(s) performed for each experiment are detailed in the respective figure legends.
We next evaluated the malignant potential of Fancc-/-;Mad2 +/-hematopoietic cells through competitive hematopoietic stem cell transplant (HSCT) experiments, in which marrows from five moribund Fancc-/-;Mad2+/-mice were transplanted into lethallyirradiated wt recipients (Figure 2A). Engraftment was confirmed based on the presence of CD45.2+ LDMNCs in peripheral blood. Despite developing hematologic malignancies consistent with leukemia, cells from post-transplant peripheral blood samples obtained from the recipient mice of one moribund Fancc-/-; Mad2+/-donor mouse failed to demonstrate CD45.2 positivity. As such, data from these recipient mice were not included in our analyses. All mice transplanted with leukemic Fancc-/-;Mad2+/marrows for which engraftment was confirmed by CD45.2+ LDMNCs in peripheral blood (recipients of 4/5 donor mice) had leukemia at necropsy one-year after transplant ( Figure 2B), and 5/13 recipients of Fancc-/-;Mad2+/-leukemic marrows died within one-year post-HSCT ( Figure 2C). Immunophenotype and histopathologic findings were the same in leukemic donor and recipients, as determined by myeloid marker flow cytometry ( Figures 2D, E) and evaluation of bone marrow architecture ( Figure 2F) as well as liver, spleen and lymph node infiltrates ( Figures 2G, H). Disease latency was similar in recipients of the same donor, but time to death in recipients differed depending on the donor ( Figure 2I). Among those with confirmed engraftment, initial chimerism of Fancc-/-; Mad2+/-transplanted cells at 3 months posttransplant was varied in the peripheral blood of recipient mice disorder, one large Ki67+, cytokeratin-negative retroperitoneal solid tumor with gland-like structures, and one lymphoma (Supplemental Table 2). We found one hepatocellular carcinoma and two solid tumors involving lungs in 3/14 aged Mad2+/-mice, as well as two lymphomas and two solid tumors in 4/12 aged Fancc-/-mice (Supplemental Figure 5). The incidence of malignancies was higher in surviving 2-year-old Fancc-/-;Mad2+/-mice compared to age-matched wt controls (1/ 17 mice) (p=0.0145). These findings expand the known patterns of malignancies for mice of these genotypes and, importantly, are consistent with the phenotype of human FA patients (8,26).

Combined Loss of Fancc and Mad2 Compromises Mitotic Fidelity Through Exacerbation of Underlying SAC Defect
FA-deficient cells consistently demonstrate features suggestive of abnormal cell division, and we and others have previously shown that the loss of FA-pathway proteins produces a phenotype consistent with SAC impairment (4,(44)(45)(46)(47)(48)(49). Above, we have confirmed the presence of an underlying SAC defect in Fancc-/cells and demonstrated that this defect is exacerbated in Fancc-/-; Mad2+/-cells. Previous data suggests that the SAC does not function as an "all or nothing" response, but rather as a rheostat that can be engaged to various degrees (9). SAC strength is known to be dependent on the amount of MAD2 recruited to the kinetochore such that depletion of kinetochore-bound MAD2 below a certain threshold causes functional compromise of the SAC, allowing cells to exit mitosis prematurely (9). To assess the role of FANCC in MAD2 localization, we employed HEC1 as a kinetochore marker and quantified the fluorescence signal intensity of MAD2 at prometaphase kinetochores via immunofluorescence microscopy in wt, Fancc-/-, Mad2+/-, and Fancc-/-;Mad2+/-mouse embryonic fibroblasts (MEFs). We validated our approach by demonstrating the expected decrease in MAD2 at prometaphase kinetochores in Mad2+/-cells. We found that MAD2 intensity was significantly decreased at Fancc-/kinetochores relative to wt kinetochores, and the amount of MAD2 on Fancc-/-;Mad2+/-prometaphase kinetochores was more significantly decreased than both Mad2+/-and Fancc-/-MEFs ( Figures 5A, B). Western blot analysis revealed comparable levels of total MAD2 protein in Fancc-/-and wt cells (p=0.4789), indicating that reduced accumulation of MAD2 at Fancc-/kinetochores was not due to reduced protein expression ( Figures 5C, D). Similarly, MAD2 protein expression was not significantly different between Mad2+/-and Fancc-/-;Mad2+/-(p=0.9896). Western blot analysis of SAC proteins BUB1, BUB1B and BUB3 also demonstrated comparable expression across all four genotypes (Supplemental Figure 9). These results support a model in which loss of Fancc impairs kinetochore recruitment of MAD2, such that further depletion of the kinetochore-bound pool of MAD2 through heterozygous deletion of Mad2 results in a compromised SAC that is insufficient to ensure mitotic fidelity. (Figure 5E).

Fancc-/-; Mad2+/-Malignancies Are Characterized by Mitotic Infidelity and Chromosomal Abnormalities Implicated in Human FA-Associated Malignancies
Given the above results, we suspected that the progressive chromosome instability resulting from increased error-prone mitoses played a critical role in the malignant transformation of Fancc-/-;Mad2+/-cells. Indeed, we found abundant abnormal mitoses in all Fancc-/-;Mad2+/-AMLs in situ and marrow cytospins ( Figures 6A, B). Further, SKY chromosome-painting confirmed karyotype instability in a moribund Fancc-/-;Mad2+/-AML mouse ( Figure 6C). To explore the similarities between FA MDS/AML in humans and mice, we first mapped the chromosomes frequently mutated in human FA MDS/AML (chromosomes 1, 3, 7 and 21) (50) to the mouse genome. Our comparative genomics analysis (51,52) mapped the relevant human chromosome regions to mouse chromosomes 3, 6 and 16; furthermore, portions of human chromosome 1 map to mouse chromosomes 1 and 4, and portions of human chromosome 3 are represented by mouse chromosome 9 ( Figure 6D). We next examined genomic instability in four Fancc-/-;Mad2+/-malignancies ( Figures 6E-H) through whole-exome sequencing (WES) coupled with copy number variation (CNV) analysis. We noted copy number abnormalities of chromosome 16 in 4/4 analyzed malignancies; chromosome 1 was affected in 2/4 and chromosome 6 in 1/4 malignancies ( Figure 6I). PANTHER pathway analysis of WES results in four Fancc-/-;Mad2+/-malignancies detected enrichment of variants within genes involved in hematopoiesis, mitosis, apoptosis, growth factor signaling, and stress response (Supplemental Table 3) (40). We also noted acquired splicing variants within Kmt2e and Kmt2c MLL (mixed lineage leukemia) lysine methyltransferase genes, and loss of Tp53 locus ( Figure 6J and data not shown). These studies provide initial insights into the somatic evolution of hematopoietic genome instability in Fancc-/-; Mad2+/-mice as they progress towards MDS/AML and suggest features that are similar to human disease.

DISCUSSION
Error-prone mitosis occurring secondary to disruption of the SAC can facilitate the acquisition of genetic alterations necessary to promote tumorigenesis (49,53). Under physiologic conditions, the , with quantification of MAD2 signal normalized to actin shown in (D) ("ns" denotes "not significant"). (E) Summary schematic illustrating that Fancc-/-; Mad2+/-mice have decreased targeting of MAD2 to KTs, resulting in impaired SAC function and heightened genomic instability. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. SAC protects cells from chromosomal instability and aneuploidy by ensuring accurate chromosome segregation (54, 55). While somatic mutations of genes encoding SAC proteins are reported infrequently, they occur at a nearly 4-fold increased frequency in FA pathway deficient cancers compared to FA pathway proficient cancers ( Figure 7). Additionally, disruption of SAC protein function is widely reported in a variety of cancers and nonmutational alterations in SAC proteins, including MAD2, have specifically been reported in AML/MDS (11,16,17,(56)(57)(58)(59).
The role of SAC dysfunction in driving tumorigenesis has previously been investigated utilizing mouse strains deficient in BUB1B, BUB3, and MAD2. These studies demonstrated that reduced expression of Bub3 (60, 61) and Bub1b (62) did not significantly increase the tumor incidence, and haploinsufficiency of Mad2 (26) caused only a mild increase in spontaneous tumors. Notably, a shared feature of the of these models is that the tumors developed only in old age, suggesting that loss of SAC protein function alone is not sufficient to drive tumorigenesis and that cells must acquire additional "hits" to undergo transformation.
Likewise, mice deficient in key FA pathway proteins, including Fancc, Fancg, Fanca, and Fancd2, also do not develop cancers until later in life (14 months of age), and only Fancc-/-mice are known to develop AML/MDS in old age (5,8). Fanconi anemia confers a several hundred-fold increased risk of developing AML/MDS compared to the general population, and treatment of these patients is challenging due to cytotoxic chemotherapeutic hypersensitivity (1). As such, the development of mouse models of spontaneous AML associated with loss of core FA genes is crucial for preclinical studies and therapeutics development (63). Several genetic strategies to accelerate the development of AML in FA core complex-deficient mice have previously been explored; however, while these mice are known to develop solid tumors and lymphoid malignancies, they do not develop early-onset AML/MDS (64)(65)(66)(67)(68)(69)(70).
Previous studies suggest that FA protein loss and SAC inactivation cooperate to promote malignant transformation (1,(10)(11)(12)(13)(14)(15). Thus, we hypothesized that SAC compromise would be sufficient to drive leukemogenesis in the setting of FA pathway dysfunction. The strength of the SAC is directly correlated with the amount of MAD2 associated with prometaphase kinetochores, which allowed us to fine-tune the frequency of mitotic errors in Fancc-/-mice by introducing Mad2 heterozygosity (9). The resultant Fancc-/-; Mad2+/-mice developed early-onset AML and exhibited abnormal hematopoiesis characterized by error-prone mitoses and progressive chromosomal instability preceding malignant transformation. We found that overt leukemogenesis in Fancc-/-; Mad2+/-mice is associated with increased mitotic infidelity secondary to compromise of the spindle assembly checkpoint (SAC) and that loss of Fancc impairs recruitment of MAD2 to prometaphase kinetochores. Accordingly, we show that leukemogenesis in Fancc-/-; Mad2+/-mice does not arise from exacerbation of the underlying DNA damage repair defect, but rather as a result of SAC insufficiency due to decreased MAD2 localization to prometaphase kinetochores. Furthermore, analysis of Fancc-/-; Mad2+/-murine leukemias through WES, CNV and SKY suggests the presence of genetic lesions that overlap with acquired chromosomal changes reported in human FA AML/MDS (50,71,72). Importantly, we noted copy number alterations in chromosomes that map to human chromosomes 1, 3, and 7, which are frequently affected in FA patients with MDS and AML (73). These findings suggest that leukemogenesis in this murine model reflects the molecular pathogenesis that underlies malignant transformation in FA patients.
Here, we have presented a novel murine model that reflects the AML-prone phenotype of human FA patients and provides insight into the role of mitotic abnormalities as drivers of leukemogenesis in the setting of FA pathway deficiency. The propensity of our Fancc-/-; Mad2+/-mice to develop early-onset AML provides a valuable tool for the development of evidencedriven therapies to optimize outcomes and decrease therapyrelated toxicities in patients with FA who suffer from AML/MDS.

DATA AVAILABILITY STATEMENT
The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found below: NCBI Sequence Read Archive (SRA) under accessions SAMN21889475, SAMN21889476, SAMN21889477, SAMN21889478, SAMN21889479, SAMN21889480, SAMN21889481, SAMN21889482.

ETHICS STATEMENT
The animal study was reviewed and approved by Institutional Animal Care and Use Committee at Indiana University.

AUTHOR CONTRIBUTIONS
DE and ES performed experiments, analyzed data, and prepared the manuscript. DM analyzed data and prepared manuscript. ZA, AS, and YH performed experiments and contributed to data analysis. K-KC performed experiments and analyzed mouse CNV/WES data. P-JC and YL performed and analyzed telomere quantification and SCE experiments. DWC and HB provided intellectual input and assisted in preparation and editing of the manuscript. RS and GC provided intellectual input. LJ and JY helped with mouse colony maintenance. GN conceived the study and assisted in manuscript preparation. All authors contributed to the article and approved the submitted version.