Cre recombinase expression cooperates with homozygous FLT3 internal tandem duplication knockin mouse model to induce acute myeloid leukemia

Murine models offer a valuable tool to recapitulate genetically defined subtypes of AML, and to assess the potential of compound mutations and clonal evolution during disease progression. This is of particular importance for difficult to treat leukemias such as FLT3 internal tandem duplication (ITD) positive AML. While conditional gene targeting by Cre recombinase is a powerful technology that has revolutionized biomedical research, consequences of Cre expression such as lack of fidelity, toxicity or off-target effects need to be taken into consideration. We report on a transgenic murine model of FLT3-ITD induced disease, where Cre recombinase expression alone, and in the absence of a conditional allele, gives rise to an aggressive leukemia phenotype. Here, expression of various Cre recombinases leads to polyclonal expansion of FLT3ITD/ITD progenitor cells, induction of a differentiation block and activation of Myc-dependent gene expression programs. Our report is intended to alert the scientific community of potential risks associated with using this specific mouse model and of unexpected effects of Cre expression when investigating cooperative oncogenic mutations in murine models of cancer.


INTRODUCTION
Mutations in FMS-like tyrosine kinase 3 (FLT3) are among the most common in myeloid blood cancers. Specifically, FLT3 internal tandem duplications (FLT3 ITD ) are found in 30-40% of all patients with acute myeloid leukemia (AML) and are associated with poor clinical prognosis due to relapse after chemotherapy [1]. Inhibitors of FLT3 combine with chemotherapy to improve overall survival in AML [2], however many patients may become resistant to these targeted inhibitors [3]. Constitutive murine models of FLT3 ITD have been developed that express FLT3 ITD from the endogenous locus [4,5], however these mice only develop AML when crossed with another oncogenic stimulus [6][7][8][9][10][11]. These combinatorial models of leukemogenesis support a hypothesis where an oncogene driving proliferation (such as a constitutively active tyrosine kinase) can cooperate with an oncogene that blocks differentiation (e.g. the loss of function in a hematopoietic transcription factor) to drive overt AML. In light of recent discussions on the significance of FLT3 ITD levels on AML prognosis [12,13], we have crossed mice heterozygous and homozygous for the FLT3 ITD allele with Cre recombinase expressing strains. To our surprise, we found that homozygous FLT3 ITD mice, when crossed with strains expressing Cre alone, had excessive and early mortality. Here we describe an unexpected but important phenomenon where Cre expression alone is able to drive AML in the context of FLT3 ITD/ITD , but not in the Cre:FLT3 WT/WT or Cre:FLT3 ITD/WT state. This process is driven by a block in myeloid differentiation and mediated by aberrant Mycdriven transcription.
Cre activation. To activate Cre in the Scl-CreERT and R26-CreER models, tamoxifen feed purchased from Specialty Feeds (Western Australia) at 400 mg/kg in base Mouse FG 66305 formulation was used. To activate Mx1-Cre in vivo, poly(I):poly(C) (pIpC) (Cytiva, Marlborough, MA) was injected two times every second day intraperitoneally. Injections were halted if mice showed signs of illness prior to completion of treatment. Spontaneous Mx1-Cre activation was noted as previously described consistent with spontaneous activation of Mx1-Cre in an inflammatory milieu [15].
Sample sizes for animal experiments were calculated to detect 25% difference in disease burden parameters between groups at a power of 80% with alpha of 0.05. Treatment groups were pre-specified prior to transplantation and treatment allocation was non-randomized and nonblinded throughout the experiment. All mice were included in analyses.

Blood analysis and bone marrow cytospins
Blood was collected into EDTA-coated tubes and blood counts assessed using a Hemavet 950 analyzer (Drew Scientific) or on a BC-5000Vet (Mindray, China). To analyse cell morphology, 1 × 10 5 BM cells were centrifuged onto glass slides. PB smears and BM cytospins were stained with Wright-Giemsa (BioScientific).

Histological imaging of mouse organs
Spleen, liver and lung were fixed and embedded according to standard protocols. Slides were automatically processed for hematoxylin and eosin staining (Leica AutoStainer XL, Leica Biosystems, Wetzlar, Germany). Images were acquired at 10x magnification on an AxioImager A.2 (Carl Zeiss Microscopy, Jena, Germany). Images were processed and analysed using the ZEN software (blue edition, version 2.3, Carl Zeiss Microscopy GmbH, Jena, Germany).

Molecular protocols and next-generation sequencing
DNA extraction. Isolation of gDNA from Mx1-Cre:FLT3 ITD/ITD AML BM cells or FLT3 WT/WT tail biopsies was performed using the QIAmp DNA Blood Mini Kit or the QIAmp Fast DNA Tissue Kit (Qiagen, Hilden, Germany), respectively, according to the manufacturer's instruction. Genomic DNA was extracted using QIAGEN QIAquick from Cre+FLT3 ITD/ITD tail pre-tamoxifen (control) and BM post-tamoxifen at clinical signs of significant disease (tumor).
Whole Genome sequencing (WGS). Cre+FLT3 ITD/ITD tail pre-tamoxifen and BM post-tamoxifen were sequenced through Macrogen with 150 bp paired end on the Illumina platform at 22x and 38x coverage, respectively. WGS of WT tails and BM of Mx1-Cre:FLT3 ITD/ITD pre-pIpC were performed by Genewiz (Leipzig, Germany) on Illumina NovaSeq 150 bp paired end 30x coverage.
RNA extraction, sequencing and analysis. For Scl-CreERT:FLT3 ITD/ITD and age-matched FLT3 ITD/ITD control BM, LKS + at 4 weeks post tamoxifen and spleen Lineage -Kit + (LK) at significant clinical signs of disease were sorted and RNA isolated using Arcturus PicoPure RNA Isolation Kit (Thermo Fisher Scientific, Waltham, MA). For Mx1-Cre:FLT3 ITD/ITD and control BM, 1 × 10 4 to 2 × 10 5 LKS + cells were sorted 10 days after last pIpC treatment into TRIzol® (Thermo Fisher Scientific, Waltham, MA) and RNA was isolated according to the manufacturer's instruction. RNA libraries were prepared using the NEBnext Ultra RNA Library Prep Kit for Illumina (New England Biolabs), assessed for size and quantified using the 2100 Bioanalyzer (Agilent Technologies) and Kapa Library Quantitation Kit (Illumina) respectively, prior to sequencing on the Illumina NextSeq 500 platform (75 bp single end). Reads were adapter trimmed (Cutadapt [19] v1.11) and aligned (STAR [20] v2.5.2a) to the GRCm38 assembly using the gene, transcript, and exon features of Ensembl (release 70) gene model. Expected gene counts were estimated with RSEM [21] v1.2.30. Genes with zero read counts across all samples were removed prior analysis. Reads were normalisation using edgeR [22] (counts per million, CPM; trimmed mean of M-values; TMM) and used for gene set enrichment analysis (GSEA) analysis [23]. Differential expression analysis comparing Cre+FLT3 ITD/ITD vs. FLT3 ITD/ITD was performed with edgeR using a negative binomial generalised log-linear model paired with genewise likelihood ratio tests.

Statistical Analysis
Kaplan-Meier curves were plotted using GraphPad Prism version 9.0 (GraphPad Software, San Diego, CA) using the log-rank test (Mantel-Cox test). Statistical analyses were performed using ANOVA with FDR p-value correction for comparing more than two groups or t-test for comparing two groups, unless stated otherwise. Significance of p-values in figures are indicated using the following ranges: * p < 0.05; ** p < 0.01; *** p < 0.001; **** p < 0.0001. Each dot represents an individual biological replicate.

RESULTS
Expression of Cre recombinase cooperates with FLT3 ITD to induce acute myeloid leukemia (AML) in mice FLT3 ITD is the most common recurrent genetic mutation found in patients with AML. FLT3 ITD/ITD knockin mice have been a valuable research model for studying the effect of cooperative gene mutations because they only develop a chronic myeloproliferative disease. Recently, other groups have described development of AML in the presence of additional mutations [6]. Scl-CreERT expression is restricted to hematopoietic stem and progenitor cells (HSPC) [14]. Scl-CreERT, Scl-CreERT:FLT3 ITD/WT , FLT3 ITD/ITD and Scl-CreERT:FLT3 ITD/ITD mice were generated and born in expected Mendelian ratios. At 6-8 weeks after birth, mice were treated with tamoxifen-supplemented chow to induce the activity of Cre recombinase. Unexpectedly, Scl-CreERT:FLT3 ITD/ITD mice developed signs of sickness within 4 weeks including marked leukocytosis (Fig. 1A, B). This progressed to a lethal AML characterised by splenomegaly and increased Kit+ HSPC in the spleen, progressive anaemia and marked myeloid skewing in the peripheral blood (PB) (Fig. 1C-F, Fig. S1A). These findings were not present in the Scl-CreERT control groups and significantly greater than in the Scl-CreERT:FLT3 ITD/WT or FLT3 ITD/ITD control groups. Notably, FLT3 ITD/ITD control groups also showed myeloid skewing and splenomegaly, consistent with the previous publications [5]. Phenotypically immature blasts were detectable in PB and BM, consistent with AML (Fig. S1B). To determine whether this effect was present with other Cre recombinases, we crossed FLT3 WT/WT , FLT3 ITD/WT and FLT3 ITD/ITD mice with Mx1-Cre, an interferon inducible Cre recombinase transgene with high expression in hematopoietic cells [24]. Mx1-Cre:FLT3 WT/WT , Mx1-Cre:FLT3 ITD/WT , and Mx1-Cre:FLT3 ITD/ITD mice were initially treated with pIpC at 4 weeks after birth. Consistent with our previous findings, Mx1-Cre:FLT3 ITD/ITD , developed leukocytosis and this progressed to a rapidly fatal AML accompanied by splenomegaly and immature blasts in PB and BM whereas control mice did not develop AML (Fig. 1G-K). Of note, disease development was detectable in a subset of mice without pIpC injection due to low level expression from the Mx1 promoter in the absence of an ectopic stimulus. Together, these unexpected findings show that Cre expression alone is sufficient to cooperate with FLT3 ITD/ITD to induce AML in mouse models.
Cre + FLT3 ITD/ITD AML is transplantable and maintained by transformed HSPC populations To determine whether Cre+FLT3 ITD/ITD AML is fully transformed, we performed transplantation into lethally irradiated (1100 cGy) C57BL/ 6 J recipient mice. Whole BM of Mx1-Cre:FLT3 ITD/ITD or Scl-CreERT:FLT3 ITD/ITD but not FLT3 ITD/ITD was able to transplant a short latency, fatal AML into recipient mice ( Fig. 2A). Next, we performed fractionation by flow cytometry, isolating putative leukemia initiating cell populations (LKS + , GMP) from the BM and spleen (Kit + ) of leukemic mice and transplanted these cells along with 2 × 10 5 wildtype BM cells ("helper BM") into lethally (1100 cGy) irradiated mice (Fig. 2B). AML was able to be transplanted in recipient mice that received 1 × 10 6 Kit+ cells from AML spleens (Fig. 2C, D). Additionally, BM LKS + cells were able to engraft into irradiated recipient, but did not manifest disease, presumably due to the limiting cell number used in transplantation (Fig. 2E). However, these results show that Cre+FLT3 ITD/ITD AML can be propagated in irradiated recipient mice and that the HSPC population contains AML initiating activity in vivo. We therefore sought to characterise the effect of Cre recombinase on HSPC populations.
Cre + FLT3 ITD/ITD AML leads to expansion of committed HSPC populations with leukemia initiating activity Consistent with the aggressive disease phenotype, AML-bearing Scl-CreERT:FLT3 ITD/ITD mice had the fewest immunophenotypically defined long-term HSCs, however this was not significantly different between the Cre+ and Cre-FLT3 ITD/ITD controls (Fig. 3A, B). AML mice from Scl-CreERT:FLT3 ITD/ITD showed marked expansion of BM HSPC populations, particularly in the myeloid progenitor compartment (Lineage -Kit+ Sca-1 -), beyond that observed with FLT3 ITD/ITD alone ( Fig. 3C-E). More specifically, AML-bearing Scl-CreERT:FLT3 ITD/ITD mice had expansion of cells expressing markers consistent with granulocyte macrophage progenitors (GMP) (Fig. 3E). As these data suggest that Cre induction preferentially leads to expansion of a committed myeloid HSPC population, we examined the effect of Cre expression within the GMP and committed myeloid progenitor compartment using LysM-Cre, a constitutive Cre within myeloid cells that has maximal expression from GMP stage but may be active in a small number of HSCs [25]. Consistent with the previous findings, LysM-Cre:FLT3 ITD/ITD mice also developed a rapidly fatal AML (Fig. 3F) characterised by extremely high WBC count, splenomegaly and circulating blast cells (Fig. 3G-I).
Cre + FLT3 ITD/ITD AML is polyclonal and not associated with recurrent genetic mutations We next sought to determine whether Cre recombinase was driving the leukemia phenotype through the generation of additional oncogenic mutations. We first tested whether the leukemic phenotype was associated by a dominant clonal population that expanded and gave rise to AML by generating Scl-CreERT:Confetti LSL :FLT3 ITD/ITD and Scl-CreERT:Confetti LSL : FLT3 WT/WT mice that would randomly express either GFP, YFP, CFP or RFP upon the induction of Cre recombinase, allowing us to trace clonal evolution based on fluorescent marker expression. After tamoxifen induction of Cre recombinase in HSCs, only Scl-CreERT:Confetti LSL :FLT3 ITD/ITD mice developed AML (Fig. 4A). Although there was incomplete expression of each fluorescent marker, we observed that the ratios of each fluorochrome were similar between both conditions, demonstrating that the AML was polyclonal in Scl-CreERT:Confetti LSL :FLT3 ITD/ITD (Fig. 4B, C). Next, we examined whole exome sequencing (WES) of Scl-CreERT:FLT3 ITD/ITD AML cells after tamoxifen administration to determine whether recurrent genetic mutations were found in the AML cells. WES of cells isolated from AML-bearing BM was compared to tail gDNA prior to Cre induction and mutations defined as somatic if present in the BM but not in the tail samples. Somatic mutations were filtered based on high confidence calls with moderate or high predicted impact on protein function. Scl-CreERT:FLT3 ITD/ITD AML had only two somatic mutations, in the genes Muc4 and Arcn1, with low variant allele frequency (VAF 0.075 and 0.155, respectively), whereas FLT3 ITD/ITD controls had a sole mutation in Vmn2r89 (low VAF, 0.039) (Fig. 4D; Supplementary Table 2). As these mutations were found in a minority of cells, we concluded that they were unlikely to be additional driver mutations responsible for the AML phenotype.
FLT3 ITD/ITD cells have a retained Neomycin resistance cassette that is LoxP flanked and excised by Cre recombinase We next performed whole genome sequencing (WGS) of pretamoxifen Cre+FLT3 ITD/ITD tails and post-tamoxifen AML BM cells and FLT3 WT/WT tails and pre-pIpC BM of Mx1-Cre:FLT3 ITD/ITD AML samples. We observed that pre-tamoxifen Cre+FLT3 ITD/ITD samples but not post-tamoxifen AML Cre+FLT3 ITD/ITD cells retained an aberrant neomycin-resistance sequence between exon 15 and 16 of FLT3 flanked by LoxP sites, a residual gene targeting sequence inserted during generation of this mouse model [5] (Fig. 5A). Similar findings have been described from another Flt3 ITD knockin mouse model [4]. Consistent with this, RNA-sequencing from both FLT3 ITD/ITD models contained reads mapping to the neomycinresistance cassette pre but not post Cre induction (Fig. 5B). Altogether, these data suggest that Cre recombinase deletes a retained neomycin-resistance expression cassette, with possible implications for gene regulation and gene expression.
Cre + FLT3 ITD/ITD AML has aberrant gene expression leading to differentiation block and Myc activation To determine whether Cre+FLT3 ITD/ITD AML was driven by broad changes in gene expression that were concordant across different Cre genotypes, we interrogated RNA-sequencing on Kit+ populations from Scl-CreERT:FLT3 ITD/ITD spleen and BM and Mx1-Cre:FLT3 ITD/ITD BM compared to FLT3 ITD/ITD controls. Despite being isolated from immunophenotypically similar cells, there were striking gene expression changes in both Cre-positive groups (Fig. S2A), characterised by a loss of myeloid differentiation and enrichment for stem cell regulatory pathways (Fig. 6A) with differential expression of key myeloid transcription factors (Fig. S2B) compared to FLT3 ITD/ITD controls. There was strong and significant concordance between the gene expression changes seen with Scl-CreERT:FLT3 ITD/ITD AML and Mx1-Cre:FL-T3 ITD/ITD AML (Fig. S2C). FLT3 ITD/ITD mice are characterised by an aberrant inflammatory milieu and disrupted cytokine signalling, and these changes were lost in both FLT3 ITD/ITD Cre-positive groups (Fig. 6B, Supplementary Table 3).
We therefore considered whether Cre-activity may have an effect on remodelling chromatin or binding at specific sites of the genome to regulate these gene expression changes. We therefore performed genome wide ATACSeq on BM LKS + cells from FLT3 ITD/ITD samples and Scl-CreERT:FLT3 ITD/ITD samples to examine chromatin conformation in an unbiased manner. Chromatin accessibility was highly consistent between biological replicates of genotypes (Fig. S3A) with Scl-CreERT:FLT3 ITD/ITD samples overall gaining chromatin accessibility (Fig. S3B, Supplementary Table 4). Interestingly, Scl-CreERT:FLT3 ITD/ITD preferentially lost open chromatin at the sites accessible during myeloid differentiation with enrichment for PU.1 (encoded by Sfpi1) motif (Fig. S3C-F). We reasoned that this was consistent with the block in differentiation phenotype seen in the BM Scl-CreERT:FLT3 ITD/ITD samples overall and did not suggest specific binding of Cre recombinase to directly influence gene expression. We did observe increase in protein FLT3 expression in Mx1-Cre:FLT3 ITD/ITD mice (Fig. S4A-K). Additionally, we see strong and consistent findings using RNA-sequencing analysis of Mx1-Cre, Scl-CreERT in BM and spleen are upregulation of known FLT3 target gene Socs1 (Fig. S4L) with consequent suppression of cytokine and inflammatory signaling (Fig. 6D) particular interferon signaling (Fig. S4L). Socs1 expression is relevant to this as since in a retroviral model it was shown to significantly accelerate the FLT3 ITD myeloproliferative phenotype or leads to leukemia [26].
In addition, RNA-sequencing data from both Scl-CreERT:FLT3 ITD/ITD AML and Mx1-Cre:FLT3 ITD/ITD AML showed marked enrichment for genes associated with Myc activation (Fig. 6B). To confirm that Myc activation was driving the development or maintenance of AML, we assessed response to targeting Myc with the BRD4 inhibitor JQ1 [27]. Cre-negative FLT3 ITD/ITD mice did not develop increased WBC count, splenomegaly or monocyte differentiation block and hence did not show any response in these parameters after 2 weeks of JQ1 treatment (Fig. 6C, D, G). Scl-CreERT:FLT3 ITD/ITD mice showed a marked accumulation of immature monocytes, elevated WBC and splenomegaly compared to Cre-negative controls (Fig. 6C-G). However, Scl-CreERT:FLT3 ITD/ITD mice treated with JQ1 showed a reduction in WBC counts, decreased splenomegaly and partial reversal of abnormal myeloid differentiation (Fig. 6C-G). Gene expression studies on JQ1 treated Scl-CreERT:FLT3 ITD/ITD AML showed reversal of Myc regulated gene expression and restoration of TNFα signalling pathways seen upregulated in Scl-CreERT:FLT3 ITD/ITD mice (Fig. 6B). In aggregate, these findings demonstrate that Cre expression leads to broad gene expression changes, including abnormal Myc pathway activation leading to a block in differentiation that drives AML development in FLT3 ITD/ITD cells.

DISCUSSION
Genetically engineered mouse models faithfully recapitulate many myeloid malignancies and provide important mechanistic insights that may not be evident from studying human samples alone. The transgenic FLT3 ITD/ITD knockin model has been used widely to model the effects of cooperative mutations on leukemogenesis, including NPM1 [9,10] and DNMT3A [6,28]. We present the unexpected findings that homozygous FLT3 ITD/ITD mice are primed for AML development and that co-expression of Cre recombinase is sufficient to give rise to a fully penetrant, yet polyclonal AML. This AML shows many characteristics of human cancer including a differentiation block and activation of a transcriptional Myc signature. Mechanistically, Myc activation appears to be important for the phenotypic manifestations of disease, as treatment with the BRD4 inhibitor JQ1 was able to reverse these gene expression changes and partially restore differentiation in vivo. This phenotype could not be attributed to any spurious damage to the genome by Cre recombinase as we did not find evidence of gene coding mutations. However, we did identify a retained neomycin resistance cassette that was incorporated in the Flt3 locus at the time of gene targeting. This Neo-resistance cassette is intronic in the Flt3 locus, is flanked by LoxP sites and is predicted to reduce expression of FLT3 overall [4]. After Cre-induction, this Neo-resistance cassette is excised, suggesting that retention of this intronic Neo-resistance construct may be having broader impact on gene expression. Importantly, we do not propose this model as a useful preclinical tool that faithfully mimics the progression of human disease. Rather, this is a cautionary report that serves to notify and remind the broader scientific community of the potential caveats of this particular mouse model in modelling AML, but also has impact more generally about the potential for off-target effects of gene targeting as previously emphasized by pioneers of the Cre-loxP system [29]. This finding may limit the applicability of data to clinical scenarios. This finding was only detected by the careful analysis of Cre-positive (and excision positive) controls, as Cre-negative FLT3 ITD/ITD mice never develop AML. Heterozygous expression of FLT3 ITD/WT in the presence of Cre activity did not give rise to off-target AML in any of our experiments, and we believe that this model remains suitable and appropriate for the testing of cooperative event in AML.

DATA AVAILABILITY
ATAC-and RNA-sequencing data are available through NCBI Gene Expression Omnibus under accession numbers GSE212222. Whole Exome-and Whole Genome-Sequencing analysis have been deposited in the NCBI sequencing read archive, BioProject ID: PRJNA909339.