Insights from Genomic Sequencing of Preclinical Breast Cancer Models Establish Human Parallels to Increase Therapeutic Applicability.

Introduction

Background
Breast cancer is classified through a number of systems, ranging from TMN (tumor size, metastasis present, nodes positive) to ER (estrogen receptor), PR (progesterone receptor), and HER2 (Human Epidermal growth factor Receptor 2) status. More recently, breast cancer has been classified by gene expression. Each of these classification systems has revealed substantial heterogeneity across breast cancer patients. With the advent of gene expression profiling of breast cancer, the subsets of genes responsible for variation in growth rate, signaling pathways, and cellular composition became more clear [1, 2]. Ultimately, this resulted in the Pam50 classifier, placing tumors into Luminal A, Luminal B, HER2-enriched, and basal-like subtypes [3]. Further gene expression studies expanded subtype classifications to include claudin-low [4]. The identification of subtypes that harbor unique gene patterns underscored the need to understand the complex gene interactions that impact clinical outcome. In 2012, The Cancer Genome Atlas (TCGA) published the initial findings of integrated genomic information, reinforcing the idea that breast cancer converges into phenotypic classes consistent with the observed genomic and epigenetic alterations. Overall, TCGA observed high incidence (> 10%) of mutations in TP53, PIK3CA, and GATA3 in breast cancer with subtypes enriched in unique patterns. For example, enrichment of GATA3, PIK3CA, and MAP3K1 was noted in Luminal A tumors. TP53 mutations were frequent in basal-like tumors and as expected, HER2 amplification was enriched in the HER2 + subtype [5]. However, there are a variety of additional genetic and epigenetic alterations that are conserved in the four main tumor subtypes of Luminal A, Luminal B, HER2-enriched, and Basal-like. Moreover, invasive lobular cancer (ILC) was comprehensively profiled to reveal enriched alterations in E-cadherin, PTEN, TBX3, and FOXA1 [6]. An examination of luminal A invasive ductal carcinoma (IDC) revealed enrichment of GATA3 mutations and increased expression. These findings clearly demonstrated the importance of sequencing breast cancer to identify new pathways driving the disease with the end goal of developing new therapies [6].
Indeed, sequencing of breast cancer has provided key insights to new possible treatments. For example, a reduction in GATA3 mutations and expression in ILC is indicative of preferential ER occupancy in bound FOXA1 sites. Coupled with reduced ER protein and activity levels in ILC vs IDC, different pharmacological investigations for these two subtypes are warranted: such as using letrozole, an aromatase inhibitor, instead of tamoxifen in ILC [6, 7]. To test these new treatments, the first step is often in a modeling system. However, it is essential that the model itself is relevant for the cancer type being studied.

Models of Breast Cancer
In vitro and in vivo breast cancer models each offer distinct advantages and limitations in their applications. Cell lines are often used to model breast cancer due to their ease of handling, homogeneity, and the ease of propagation and maintenance. Commonly used cell lines represent a range of pathologies: BT-20, MCF-7, MDA-MB-435, T47D as used to model invasive ductal carcinoma while MDA-MB-231 and SKBR3 resemble adenocarcinoma [8–13]. Despite the advantages and utility of cell lines, it is questionable how well they resemble human disease, especially in two dimensional assays [14].
Mouse models are powerful tools used to study the heterogeneity in breast cancer: including carcinogen-based, genetically engineered mouse models (GEMMs), and patient-derived xenografts (PDXs). Carcinogens that have been used to induce mouse mammary carcinogenesis include 7,12-dimethylbenz[a]anthracene (DMBA), N-methyl-N-nitrosourea (MNU), and 2-amino-1-methyl-6-phenylimidazo[4,5-b]pyridine (PhIP). DMBA induced tumors develop a diverse assortment of mammary tumors with resemblance to human subtypes. For example, long latency DMBA tumors have been shown to recapitulate human luminal breast carcinomas with activation of the PI3K-Akt pathway [15]. MNU induces mammary carcinogenesis in conjunction with medroxyprogesterone acetate [16]. Additionally, PhIP has been shown to induce carcinogenesis through DNA damage, mutagenic activity, and ER alpha activation [17]. These models are useful for recapitulating the tumorigenic process but have minimal genes that are mutated. This has limited the utility of these models in studying human breast cancer subtypes.
To better model human breast cancer and the heterogeneity that is present, PDXs have been widely used and are able to overcome this translational gap. With the ever-expanding pool of PDX samples available, this has become an increasingly viable model system. Indeed, a collection of 537 PDX lines from 500 unique patients is representative of a variety of subtypes: 56% TNBC, 36% ER +, and 8% HER2 + [18]. In this system, human breast cancer cells/tissue are xenografted into immunodeficient mice, often for the purpose of drug testing. PDX models are biologically stable, renewable, and allow for molecular interrogation in greater depth than possible in clinical practice. Although these models are derived from human patients, the question remains: how well do these models recapitulate human breast cancer?
Genetically engineered mouse models offer a controlled means to induce tumorigenesis in mice with the manipulation of the initiating event. Commonly used models include MMTV-Myc [19], MMTV-PyMT [20], MMTV-neu [21], and MMTV-v-ha-ras [22]. Each induces tumorigenesis through manipulation of genes linked to human breast cancer. For example, the master transcriptional regulator MYC is amplified in 15% of human breast cancers and overexpressed in up to 35% of patients [23]. It has been clearly demonstrated that overexpression of Myc in mice drives spontaneous mammary tumorigenesis whether through overexpression of the wild type, more stable mutants or through induced overexpression [19, 24, 25]. In the TNBC subtype of human breast cancer, Myc activation and Pten loss are often co-occurring and interestingly a Myc;Ptenfl mouse model demonstrates cooperativity of these two cancer drivers to recapitulate the human phenotype [26]. Likewise, the proto-oncogene HER2/Neu is amplified and/or overexpressed in approximately 15% of all breast cancers [27]. Concordantly, a variety of Neu mouse models demonstrated that overexpression or amplification of this gene drives mammary tumorigenesis [21, 28–30]. Indeed, the conditional knock-in model that develops amplification of the chromosomal locus was also noted to contain extrachromosomal DNA with Neu [31]. While the MMTV-PyMT mouse model initiating event is genetically unlike human breast cancer, it activates similar gene expression pathways. It is widely used as a potent oncogene in the mammary epithelium that closely mimics histological disease progression and often clusters with the human luminal B subtype [20, 32–34]. Although each of these models induce mammary tumorigenesis with a strong initial driving event, like human breast cancer, they accumulate additional genetic alterations over the course of the disease. This is especially evident by the heterogeneity in the MMTV-Myc model, with the characteristic accumulation of Kras activating mutations in the EMT histological subtype [24].
The breast cancer models we employ today have been essential in understanding breast cancer initiation, evolution, and progression. However, it remains to be determined how relevant the models are at a broad scale and thus some of the knowledge gained from the models is limited in relevance to human breast cancer. It is key to understand how well these models recapitulate human breast cancer in terms of genomics, gene expression, pathway activation, and how models manipulating these variables correspond to human disease. Ultimately, a better understanding of the models used to study breast cancer will enhance the ability to understand and treat this disease.

Sequencing of Models Addresses Critical Knowledge Gap
Sequencing human breast cancer through studies such as TCGA, COSMIC, ENCODE, and METABRIC have been pivotal in elucidating conserved genomic alterations. Further integration with gene expression has illustrated linked pathway activation critical to disease progression. This genomic information, and the pathways that are activated, has also been key to understanding how well our models recapitulate this human disease. However, sequencing of tumor models allows us to understand what shared mechanisms drive disease progression. Thus, it is imperative that as new models are developed, or establish models are used, that whole genome sequencing be completed as it informs on copy number alterations (CNAs) and the mutational landscape that determines how relevant each particular model is.

Sequencing of Cell Lines
Cell lines derived from human patients maintained in culture are typically thought to have accumulated additional genomic alterations over time and therefore it is pertinent to know at a genomic level how well they resemble human breast cancer. AURKA amplification, MYC amplification, and CDKN2 deletion in MCF7 cells is observed as well as ERBB2 amplification in SKBR3 cells [35]. SKBR3 cells are an important tool in understanding HER2 + breast cancer, and further sequencing has elucidated additional alterations. Numerous novel structural variants were observed using long-read DNA sequencing that have not been found with short-read DNA technology, such as KLHDC2-SNTB1 [36]. Further analysis of genomic data has revealed cell lines exhibit considerable genomic differences from the metastatic breast cancer they model [14]. This is due to length of time in culture, drift of the line, and absence of a microenvironment in vitro that shapes metastases in vivo among other factors. Thus, while cell lines have utility, they should be used with caution and other in vivo models, that lack widespread genomic alterations, may prove to be superior for studies on the initiation and progression of cancer.

Sequencing Mouse Models of Breast Cancer: Gene Expression Analysis
Interrogating gene expression profiles of mouse models of breast cancer has established the validity of comparing these models to human breast cancer. Whole exome and RNA-seq on long-latency DMBA induced mammary carcinogenesis models showed Pi3kca and/or Pten mutations, with frequent occurrence of a Pi3kca H1047L/R hot-spot mutation and this mutation leads to significant Pi3k-Akt pathway activation. Molecularly, this model recapitulates human luminal breast cancer and as such, may serve as an appropriate preclinical model to test Pi3kca/Akt/mTOR pathway inhibitory therapies [15]. By interrogating a large dataset of mouse tumor models through signaling pathway activation analysis, it was revealed that for the majority of human breast cancer sample, there was a mouse model with a similar profile. Indeed, with analysis of sufficient samples the MMTV-Myc model represents much of the heterogeneity present in human breast cancer [37]. Similar studies have shown GEMMs can recapitulate human subtypes based on commonly expressed pathways [33]. An interesting example of evolution of mouse models to more closely resemble human breast cancer can be seen with HER2 + breast cancer. The initial model drove Neu expression under the control of the MMTV promoter [21]. While this resulted in tumors, the gene expression data from the MMTV-Neu mice does not cluster with HER2 + human samples. In part, this is due to human HER2 + breast cancer having a much larger amplicon including Grb7 and several other genes that may alter gene expression and tumor biology [38]. This model was refined by generation of expression of the activated form [29], and then a switch away from the MMTV promoter to the endogenous promoter. In this strain, a floxed stop cassette was placed between the endogenous promoter and activated NeuNT allele, and this floxed stop was removed in the mammary epithelium by MMTV-Cre. The tumors that developed in this system shared the human HER2 amplification event [30] and using gene expression they were observed to be a much more appropriate representation of human HER2 + breast cancer [39]. By sequencing mouse models, we will understand how alterations such as CNAs, mutations, and differential pathway activation between and within models ultimately recapitulate human breast cancer to refine our use of these preclinical models.

Sequencing Mouse Models of Breast Cancer: Genome Sequencing Reveals Human Parallels
CNAs are known to be key drivers in breast cancer, such as HER2 amplification or PTEN loss. It is therefore important to also understand how CNAs in mouse models compare to human patients. Using methods to predict CNA from gene expression revealed CNAs to be common, but heterogenous, across many mouse models with many of these events being conserved in human patients [40]. Testing these predictions, WGS of MMTV-Neu identified conserved CNAs in extracellular matrix proteins collagen 1 type 1 alpha 1 (Col1a1) and chondroadherin (Chad) and demonstrated their roles in metastasis [41]. In addition to CNAs, mutations in genes can drive tumor initiation and progression. Surprisingly, the rapid tumor formation in the MMTV-PyMT model was revealed through WGS to have a highly conserved mutation of phosphatase Ptprh, resulting in increased phospo-EGFR levels [41]. By interrogating conserved alterations in mouse models and human patients, sequencing of these models has helped us to understand how models recapitulate human disease. This argument that sequencing is essential to reveal conserved alterations and possible vulnerabilities within the heterogeneity of breast cancer was solidified further by the discovery of activating Kras mutations. Gene expression analysis of MMTV-Neu, MMTV-Myc, and MMTV-T58A Myc models revealed segregation both by genotype and histological subtype. Interestingly, EMT tumors display high probability of Ras pathway activation. Sanger sequencing revealed that these tumors were significantly enriched for activating Kras mutations [24, 25, 42]. This Kras activation then predisposes tumors to lose Myc activity, suggesting a dominant role for Ras in tumor progression in the EMT subtype of tumors [24, 43]. Intriguingly, histological subtype is more important in dictating pathway activation than the initiating events in mouse models: Kras activation is present across mouse models with an EMT subtype [44]. This identification of driving mutations underscores the need to sequence tumor models. Indeed, it is interesting to note that in an NRL-PRL mouse model, which mimics ER + breast cancer in postmenopausal women, whole genome and exome sequencing identified activating Ras alterations in 23 out of 29 tumors [45]. This may serve as an appropriate model for luminal B ER + breast cancers in which there is elevated RAS pathways activation [46]. Conserved Kras mutations have also been implicated in metastatic progression revealed by matched whole exome, genome, and RNA sequencing followed by proof-of-concept validation [47].
Given the need for integration of WGS and gene expression to define how models relate to human breast cancer, the heterogenous MMTV-Myc model was subject to multi-omic analysis. Three major subtypes were examined in this model system. Conserved mutations were observed in Kit and Rara in the microacinar subtype while conserved Kras activating mutations and Scrib mutations were noted in the EMT tumors. Implications for how these particular mutations are relevant to human breast cancer are currently a focus of study. It is interesting to note that copy number gains found on chromosomes 11 and 15 in the microacinar subtype correlate to frequently amplified regions in humans on chromosomes 8 and 17, containing Myc and Erbb2 respectively. Investigation into these findings can be useful for delineating mechanisms to target in humans [48]. Indeed, investigating the heterogeneity within mouse models often reveals differential pathway activation, ultimately demonstrating that these different subtypes resemble human subtypes of breast cancer, as discussed above with the MMTV-Myc model. This has also been revealed in other models such as p53 null [49] and MMTV-PyMT [44]. For example, WGS has been performed on the p53-null mammary transplant tumor model to reveal Met amplification [50]. Whole exome sequencing (WES) on Brca2- and Trp53- deficient mice also showed Met amplification and extrachromosomal DNA [51]. TP53- BRCA1- null mice (KPB1) have been shown to resemble human basal like-tumors, with conserved CNAs such as Rb1 loss, Myc amplification, and Kras amplification. This was determined by array comparative genomic hybridization and corroborated by oncogenic pathway signatures [52]. Similar analysis has been completed for the MMTV-Wnt1 model [53] and for a p53- Pten- null model [54]. More recently, WES explored the role of metastasis-associated alterations in MMTV-PyMT, MMTV-Myc, MMTV-Neu, and C3(1)-TAg mouse models [55]. Unfortunately, this study found limited evidence for metastasis-associated point mutations and CNVs.
Genomic analysis of mouse models of breast cancer predicted that E2F transcription factors (E2Fs) are critical in mouse models of breast cancer [37]. Manipulation of E2F transcription factors in mouse models of breast cancer has confirmed key roles of E2Fs controlling genes implicated in angiogenesis, extracellular matrix remodeling, tumor cell survival, and interactions facilitating lung metastasis in MMTV-PyMT [56], as well as important roles in MMTV-Neu [57], and MMTV-Myc models [58]. Subsequent sequencing has demonstrated human parallels, with MMTV-Neu E2F KOs revealing conserved alterations with human HER2 positive patients and high E2F1 activity was associated with worse outcomes [59]. In a MMTV-PyMT mouse model, microarray to analyze gene expression revealed a role of E2F1 as a regulator for many pro-metastatic genes and elucidated target genes, especially Fgf13, that serve a role in metastasis [60]. MMTV-Neu and MMTV-PyMT were then interbred onto an E2F1 null background, resulting in metastatic differences. WGS of MMTV-PyMT and MMTV-Neu in E2F1 null and wild type backgrounds was completed to provide insights into genomic mechanisms that could be driving metastasis. Of interest were alterations in DNA mismatch repair, cell adhesion, and mutational abundance [61].
Given the role for E2Fs in mammary tumor development, it is interesting to note that loss of E2F5 alone was sufficient to drive tumor initiation. In a mammary-specific E2F5 knockout mouse model, RNA-seq suggested altered Cyclin D1 regulation may be driving this highly metastatic tumorigenesis. WGS revealed conserved amplification of Wnt2, Cav1 and Met and SNVs of Fbxo15 and Tshz1 that may be of significance [62] and that are currently being examined. It is evident that WGS resolves the knowledge gap of identifying key events in the progression of these models of breast cancer. A summary of WGS in mouse models is highlighted in Table 1.

Sequencing of Patient Derived Xenografts
In some aspects, PDXs may serve as a more appropriate preclinical model of breast cancer. To understand how well PDXs represent human disease, sequencing is imperative. WGS of 17 matched patient tumors and xenografts shows high conservation of structural variant and copy number events between these paired samples. It is interesting to note that PDXs resistant to endocrine therapy exhibited ESR1 mutations and translocations [73]. PDXs have demonstrated limited genomic instability, with nucleotide changes associated with a minority of xenografts compared to original tumors [73, 74]. CNAs were similar to that of patient samples, evident by the gains in MYC and ERBB2 and deletions of PTEN and PPP2R2A. Interestingly, mutations were well correlated to the subtypes typical from TCGA. For example, ER- PDXs recapitulated the mutational landscape described in TCGA, with both having common alterations in TP53, TTN, and PIK3CA [75]. As expected, PDXs were demonstrated to resemble the patient disease they were derived from, at the molecular level and in terms of growth, metastasis, and pathology [76, 77]. As such, PDXs, and matched organoids, provide a critical tool in the study of human breast cancer for therapy development. There are limitations to consider with this model, especially a lack of the immune system in the tumor microenvironment, since these are human tumors implanted into immunodeficient mice. As there is no human stroma or immune cells, the growth environment is arguably not biologically relevant [78]. Yet, the excellent recapitulation of molecular characteristics compared to patient tumors make PDX models an excellent pre-clinical model for drug discovery.

Sequencing and Breast Cancer Therapeutic Targeting
Current therapies for breast cancer are informed by hormonal and HER2 status. Standard of care for ER + breast cancer includes tamoxifen, a selective ER modulator, or aromatase inhibitors to block estradiol synthesis [79]. HER2 amplified breast cancer is targetable by trastuzumab [80] and pertuzumab [81]. Indeed, these therapies have been pivotal in changing the trajectory of survival for this disease. However, TNBC, by definition, lacks these receptors in which there have been crucial breakthroughs. As such, more precise therapies are needed for treatment of TNBC. Moreover, therapeutic resistance [82] and toxic side effects of current treatments [83, 84] underscores the need for precision therapies. Frequently altered genes are often targeted and therefore PI3K [85], FGFR [86], PARP inhibitors for BRCA2 deficient patients [87], and MET inhibitors [88] have all been developed and show promise in treatment. Sequencing models of breast cancer will be pivotal in identifying which models should be used to develop and test new therapies for breast cancer.
Sequencing of models can also be used to suggest new opportunities for therapeutic intervention. Indeed, WGS revealing conserved Ptprh mutations has suggested a new patient population that may respond to therapy. Ptprh targets EGFR and indeed, with the V438M mutation identified in MMTV-PyMT, pEGFR increases. The mutation also decreased tumor latency. Interestingly, targeting EGFR with erlotinib revealed that those tumors containing Ptprh mutations were more sensitive [41]. While PTPRH is not frequently overexpressed or mutated in breast cancer, it was noted that 5% of all lung cancers contain PTPRH mutations. Indeed, when NSCLC with PTPRH mutation in vitro and in vivo are treated with the tyrosine kinase inhibitor Osimertinib, there was a significant response, indicating PTPRH mutations make these tumors more susceptible to EGFR tyrosine kinase inhibitors. This demonstrates the importance of WGS to reveal novel vulnerabilities and appropriate models for drug development for clinical use [89]. Similarly, in sequencing of the p53-null model, Met was shown to be amplified in the p53null-Luminal subtype. Met was identified to be an ideal candidate drug target, and indeed treatment with crizotinib, a Met inhibitor, shows complete regression in the Met-amplified p53null-Luminal tumors, but not in tumors lacking Met amplification [50]. Sequencing can also be useful for uncovering mechanisms of resistance. For example, Myc was shown to be a driver of mTOR inhibitor resistance by sequencing the K14-Cre;Cdh1F/F;Trp53F/F (KEP) mouse model of ILC [64].
Appropriate molecules are as essential as having an appropriate model to test them in. Mutated Kras has been repeatedly implicated in the progression of mouse models of breast cancer [15, 24, 25, 45, 47, 90], including basal TNBC invasion [91]. Targeting of KRas proved to be a bottleneck in advancing treatment. Many efforts have been made to address this. Sequencing has revealed differential pathway activations that can then be used to predict drug sensitivity. For example, there was significantly increased sensitivity to the RAS pathway inhibitors farnesylthiosalicylic acid (FTS) and a farnesyl transferase inhibitor (L-744,832) in cell lines that showed an increased probability of RAS pathway deregulation [92]. In mouse models of breast cancer, RNA sequencing of a model of basal triple negative breast cancer (b-TNBC) organoids (derived from C3(1)-TAg GEMM) revealed Kras to be required for invasive phenotypes [91]. Given the historic difficulty in targeting KRAS, efforts were made in targeting pathways up and downstream of the dysregulated protein. Upstream inhibition of EGF, Igf-1, or FGF reduced invasion in this b-TNBC model. Interestingly, patients with b-TNBC have been reported to have EGFR amplification and therefore EGFR inhibition is being investigated clinically. Pathways downstream of KRAS include MEK, PI3K, AKT, and RAK1, and inhibition resulted in decreased invasion with ERK inhibition showing strong invasion reduction [91]. Ultimately, sequencing to reveal KRAS pathways as targets was integral in generation of novel pharmacological strategies. Recently, inhibitors specifically targeting KRAS G12C in lung cancers, such as adagrasib [93] and sotorasib [94], have been developed. RMC-7977 as a broad-spectrum RAS inhibitor shows promise [95]. Given the plethora of mouse models of breast cancer exhibiting Kras mutations, testing of novel RAS inhibitors in conjunction with these models would be insightful for clinical outcome and mechanistic insights.
Sequencing of tumors is critical in establishing human relevancy for models in testing mechanisms that can be intervened with regards to developing cancer therapeutics. PDX models are advantageous in that the human relevancy is recapitulated with high fidelity, from the heterogeneity of genomic alterations in patient breast cancers to drug responses. As proof of concept, HER2 + models were sensitive to the ERBB2 inhibitor afatinib and loss of function of BRCA1 in models resulted in sensitivity to PARP inhibition [78]. This is similar to PDX WGS demonstrating BRCA1/2 altered tumors were typically homologous recombination deficient and sensitive to cisplatin [96]. Clearly, it is important to establish how well PDX models recapitulate human disease for the purpose of drug discovery, and WGS provides an avenue for this. In fact, PDX models could be particularly useful for uncovering drivers of treatment resistance [97].

Discussion and Limitations
Sequencing of mouse models of breast cancer is imperative to understand how well they recapitulate the human disease. New mouse models have typically been derived to mimic known oncogenic events [19, 28, 52]. However, as with human breast cancer, heterogeneity within mouse models at both a histological and gene expression level suggested that despite the overexpression of strong oncogenic signals, other events are still required. WGS demonstrated that these genomic events accumulate during tumor formation and progression. Gene expression profiling experiments have characterized differential expression and pathway activations that are present within these models [33, 37]. Indeed, in the various Myc induced models, Kras activation is common in the Squamous/EMT subtype. Beyond Kras, WGS is crucial to understand which CNAs and mutational events are driving histological and gene expression alterations. Indeed, MMTV-Myc EMT tumors that have Kras activating mutations were also noted to contain Fgfr2 amplification, demonstrating a need to precisely refine the contributions of individual and combinatorial genomic events [48]. It is also essential that transcriptomic data be incorporated with genomic data. From Table 1, it is clear transcriptomic analysis has been extensively utilized to characterize mouse models of breast cancer, yet genomic analysis is lacking. Moreover, studies lacking transcriptomic analysis struggle to adequately understand how alterations found by WGS recapitulate human disease [51, 54]. Thus, the combination of sequencing and gene expression information is required to obtain a clear picture of tumor progression (Fig. 1). While sequencing is a powerful tool, it is currently not being utilized to its fullest extent in mouse model systems. While some studies [47, 55] have used exome sequencing, WGS provides a unique perspective in that non-coding regions are sequenced, and such alterations may have significant impacts on events leading to cancer as well. For example, alterations in promoters and long-range enhancers may have a significant impact on the tumor biology [98]. Compared to coding regions of DNA, interrogation of these non-coding regions remains relatively unexplored [99]. Indeed, investigating the role of epigenetics in tumor biology can yield valuable insights. For instance, EZH2 inhibition to reduce H3K27me3 has been shown to enhance HER2/ErbB2 therapy [100]. In addition to differences in sequencing methods (exome to WGS), the literature reveals that mouse models lack a standard method to analyze and call genomic alterations. This can result in discrepancies, such as many more translocations being recognized in the same wild-type Neu and PyMT tumors in two separate studies as methods have evolved even in our own laboratory [41, 61]. Thus, the field could benefit from an expansion of genomic analysis of mouse models and the standardization of best practices in calling these alterations. Sequencing of several key mouse models have provided powerful insights into driving mechanisms of breast cancer. Despite this, there are still many models yet to be sequenced and thus, it is unknown how well they recapitulate human breast cancer. In addition to WGS, as technology advances and it becomes both feasible and cost-effective, spatial techniques (transcriptomics and proteomics) will provide insightful knowledge in tumor heterogeneity and tumor microenvironment. Layering this over well-established RNA sequencing data and whole genome sequencing data [101, 102] will result in a better understanding of how the mutational spectrum influences the immediate tumor microenvironment. This review discusses sequencing of models in the context of breast cancer. It is essential that additional models of breast cancer be sequenced as well, especially those known to be heterogenous in nature. Beyond breast cancer, it is also essential that these techniques be applied to other cancer model systems. While this has been done for Egfr-, Myc-, and Kras-driven GEMMs of lung adenocarcinoma [103], many other models are lacking. As with breast cancer models, this will aid in uncovering key players of cancer initiation and progression that mirror human disease with the goal of yielding mechanistic insights.