**3. Results**

#### *3.1. Host Strain Determines Tumor Onset for Genetically-Identical MPNSTs*

To determine the impact of murine background strain on MPNST development, we generated somatic CRISPR/Cas9-induced tumors in four commonly-used laboratory strains: 129/SvJae, C57BL/6, 129X1, and BALB/c. Importantly, the 129/SvJae mice serve as reference controls, as this strain was used in our prior study [27]. We injected the sciatic nerve of 10–13 mice per background with adenovirus containing Cas9 and guide RNAs for *Nf1* and *p53* (Ad-Cas9 + gNF1 + gp53). This approach was previously shown to generate high-fidelity, *Nf1*/*p53*-null MPNSTs at the site of injection within 3–4 months. Similar to our prior data, 129/SvJae mice in the current study develop tumors at ~80 days post-injection (Figure 1A). Tumor onset is similar in C57BL/6 and 129X1 mice, arising at an average of 82 and 93 days, respectively. In contrast, BALB/c mice develop MPNSTs earlier than other strains, with tumors developing with an average onset of 61 days. After tumor detection, MPNSTs were measured 3x/weekly to obtain proliferative rates, which are calculated from a uniform initiating size of 150 mm3. The average time for tumors to double in volume is 7–8 days, which is similar across all backgrounds (Figure 1B). Tumor proliferation was also examined by immunohistochemistry for Ki67 in terminally-harvested MPNSTs. Ki67 indices are similar in tumors from all strains, supporting the observation that host strain does not influence MPNST proliferation (Figure 1C). Histological analysis confirms MPNST morphology in all tumors, with S100 positivity noted in tumors from each background (Figure 1D). Taken together, these data show that somatic CRISPR/Cas9 tumorigenesis approaches can generate MPNSTs in a broad spectrum of wild-type mice, and that background strain can influence tumor initiation in genetically-matched tumors.

**Figure 1.** Host strain determines tumor onset but does not alter tumor growth kinetics. (**A**) Kaplan– Meyer curve of tumor-free survival. Formation of *Nf1*/*p53*-deleted malignant peripheral nerve sheath tumors (MPNSTs) is accelerated in BALB/c mice. Tumor initiation occurs within a similar timeframe in mice from 129/SvJae, C57BL/6, and 129X1 backgrounds. (**B**) Growth kinetics are similar across all background strains for genetically-identical MPNSTs (*<sup>n</sup>*= 6–8 tumors per strain). Growth rates are calculated as the number of days required for tumors to double from an initial volume of 150 mm3. 129X1 (red circles), C57BL/6 (black triangles), BABL/c (white squares), and 129/SvJae (blue triangles). (**C**) Representative images of MPNSTs from different host strains stained for H&E (20×), S100 (20×), and Ki67 (20×). (**D**) Quantification of Ki67 confirms that background strain does not alter the rate of tumor proliferation (*<sup>n</sup>*= 5 tumors per strain). (**B**,**D**) analyzed by one-way ANOVA with Tukey's multiple comparison test.

#### *3.2. Indel Analysis Reveals Unique Patterns of Gene Disruption*

Indel signatures can determine the spectrum and frequency of CRISPR/Cas9-induced events in individual tumors. We generated tumor-derived cell lines to evaluate the unique indel patterns within each MPNST (Figure 2). Our analysis confirms the presence of *Nf1* and *p53* indels in all tumors. Additionally, no wild-type sequence is detectable in any cell line, suggesting complete disruption of the targeted regions. As CRISPR/Cas9 generates indels by random reassembly of DNA, we investigated the types of indels generated with each guide RNA. To focus this analysis, we evaluated indels that occur at > 5% frequency. Across 14 tumor-derived cell lines, we observe 24 indels in *Nf1* and 33 indels in *p53*. Several cell lines have a simple signature, containing predominantly one indel, while others have complex signatures comprised of up to five distinct variants per gene. The majority of cell lines contain multiple *p53* indels, as a single dominant indel of *p53* is detected in only 4/14 (29%) of cells. Single indels in *Nf1* are more frequent, with 7/14 (50%) of cell lines containing a solitary *Nf1* indel event. Insertions are less common than deletions, with only 1/14 (7%) of cell lines harboring *Nf1* insertions and 6/14 (43%) of cell lines harboring *p53* insertions. Indeed, only one cell line does not have a deletion event in *p53*, with a single predominant insertion being the only indel event detected within the sample. In our analysis, CRISPR-generated insertions are genetically small (1–2 bp), while deletions occur within a larger range (1 bp to > 20 bp). In *p53* indels, we observe a trend towards smaller deletions (<10 bp), which occur in 23/27 (85%) of deletion events. All of the indels detected in *Nf1* were either frameshift (FS) mutations (20/24) or indels ≥ 20 bp (4/24) that are the most likely to disrupt protein function by shifting the reading frame and inducing premature termination, nonsense mediated decay (NMD), or alterations in protein structure [33,34]. For indels detected in *p53*, 24/33 were FS mutations and 3/33 were deletions ≥ 20 bp. We did not identify any strain-specific trends in indel type, size, or frequency in this analysis, suggesting that in vivo CRISPR/Cas9 genomic editing occurs similarly across different murine backgrounds.

**Figure 2.** CRISPR/Cas9-induced insertions and deletions detected in *Nf1* and *p53* in MPNST-derived cell lines from different genetic backgrounds. Indel pattern analysis of the sgRNA-targeted regions of *Nf1* (**A**) and *p53* (**B**) demonstrates disruption of genomic targets in all tumors. The majority of indels detected in both *Nf1* and *p53* are frameshift mutations that result in inactivation of targeted proteins.

#### *3.3. Immunological Diversity of MPNSTs Is a Hallmark of Genetic Background*

Data from genetically-engineered mouse models strongly support a role for host strain in distinct patterns of immune cell activation [18,24]. Therefore, we hypothesized that there are strain-dependent differences in the composition of the immune landscape in our CRISPR/Cas9 generated MPNSTs. To examine the tumor microenvironment in genetically-identical tumors from different mouse strains, we performed histological analysis for populations of innate and adaptive immune cells that play key roles in MPNST biology, including CD4+ T cells, CD8+ T cells, regulatory T lymphocytes (Tregs), macrophages and mast cells in five tumors per genetic background (Supplementary Figure S1).

Levels of tumor-infiltrating cytotoxic CD8+ T lymphocytes are similar across all host strains (Figure 3A). In contrast, amounts of CD4+ T lymphocytes are highly dependent on background strain, with MPNSTs from C57BL/6 mice having lower CD4+ T infiltration than tumors on 129Sv/Jae, BALB/c, and 129X1 backgrounds (Figure 3B). MPNSTs from 129Sv/Jae mice display a heterogenous distribution of CD4+ T lymphocytes, with a wide variability of cell number across individual tumors. Regulatory T cells levels are highly variable across individual tumors, most likely due to the rare nature of these cells. In several tumors, we were unable to detect a single Treg in the sample. Analysis of multiple tumors determined that MPNSTs from 129X1 mice have higher levels of Tregs than MPNSTs from C57BL/6 or BALB/c mice (Figure 3C). Analysis of macrophage levels by F4/80 staining shows increased macrophage infiltration in MPNSTs from BALB/c mice (Figure 3D). Mast cells, histamine-rich myeloid cells with a strong role in MPNST biology [30,35], are enriched in MPNSTs from BALB/c mice (Figure 3E). The lowest levels of mast cells are observed in tumors from C57BL/6 mice. Taken together, these observations demonstrate the broad diversity of immune landscapes in MPNSTs from different background strains.

**Figure 3.** The MPNST immune landscape is determined by genetic background. (**A**) Levels of CD8+ T cells in terminally-harvested MPNSTs are similar across all host strains. (**B**) Infiltration of CD4+ T cells are significantly lower in tumors from C57BL/6 mice compared to MPNSTs in mice from 129X1, BALB/c, and 129/SvJae backgrounds. (**C**) Foxp3+ Tregs are detected at higher levels in tumors from 129X1 mice compared to C57BL/6 and BALB/c mice. (**D**) MPNSTs from BALB/c mice have significantly higher levels of infiltrating F4/80+ macrophages compared to C57BL/6 mice. (**E**) Mast cell infiltration is higher in tumors from BALB/c mice compared to 129/SvJae, C57BL/6, and 129X1 mice. Mast cell levels are lowest in MPNSTs from C57BL/6 mice. 129X1 (red circles), C57BL/6 (black triangles), BABL/c (white squares), and 129/SvJae (blue triangles). Analyzed by one-way ANOVA with Tukey's multiple comparison test. A *p*-value of less than 0.05 is considered statistically significant and is denoted by "\*" (*n* = 5 tumors per strain).

#### *3.4. Gene Expression of the MPNST Microenvironment*

Given the broad variability of strain-dependent immune infiltration observed in our IHC data, we chose to perform extensive gene expression analysis of key tumor microenvironmental markers [24]. Using real-time qPCR analysis of whole tumor lysates from five tumors per background, we evaluated expression levels of pathways involved in innate immunity, adaptive immunity, angiogenesis, and cytokine signaling (Figure 4A and Supplementary Figure S2). These data provide insight into key tumor–stroma interactions and reveal extensive heterogeneity across host strains and individual tumors.

We first examined expression of tumor-associated macrophage (TAM) genes, since they are one of the most differentially-regulated immune cell populations between host strains. Expression of *Arg1* mRNA, a marker of immunosuppressive M2 macrophages, is elevated in MPNSTs from BALB/c mice (Figure 4B). Of note, *Arg1* is the only gene in our analysis that is statistically different between host backgrounds (*p* = 0.0156, one-way ANOVA). There were no differences in levels of the M1 macrophage marker *iNos1*/*Nos2* in tumors from different host strains (Figure 4C), suggesting that the influx of macrophages in MPNSTs from BALB/c mice consists of *Arg1*-expressing TAMs of the M2 subtype. This finding is consistent with data demonstrating that expression of the pro-immunogenic, M1 macrophage transcription factor *Stat3* is similar across backgrounds.

**Figure 4.** Expression of key genes in the MPNST microenvironment. (**A**) RT-qPCR analysis of markers for innate immunity, adaptive immunity, angiogenesis, and cytokine signaling in terminally-harvested tumors shows a large degree of heterogeneity between host strains and individual tumors. Samples are normalized to a single C57BL/6 tumor, shown as reference (*n* = 5 tumors per strain). (**B**) Expression analysis determines that MPNSTs from BALB/c mice express significantly higher levels of *Arg-1* mRNA, a marker of immunosuppressive M2 macrophages, when compared to tumors from 129/SvJae, C57BL/6, and 129X1 mice. (**C**) In contrast, levels of *Nos2* mRNA, a marker of M1 macrophages, is similar in tumors from all background strains. 129X1 (red circles), C57BL/6 (black triangles), BABL/c (white squares), and 129/SvJae (blue triangles). Analyzed by one-way ANOVA with Tukey's multiple comparison test. A *p*-value of less than 0.05 is considered statistically significant and is denoted by "\*".

To further explore T lymphocyte populations, we examined genes involved in T cell activation and signaling. Expression of APC-resident co-stimulatory molecules—including *CD80*, *CD86*, *OX40L*, and *PDL1*—are similar across host strains. Similarly, expression of *CTLA-4*, an inhibitory receptor that negatively regulates T cell responses, and *CD83*, a marker of activated CD4+ T lymphocytes and dendritic cells, is not strain dependent. Levels of the regulatory T cell marker *FoxP3* are not statistically different across strains due to extensive heterogeneity between tumors, although trends are similar to IHC findings in Figure 3.

We next examined expression of angiogenesis genes, including *Vegf*, *Vegfr1*, and *Vegfr2*, in addition to the lymphangiogenic growth factor *Vegfc*. While several individual tumors display high expression of these growth factors, there are no statistically significant differences between host strains. Finally, we examined expression of key cytokines involved in immune activation, including proinflammatory molecules (*Tnfa*, *Ifng*, *IL4*, *IL1b*, and *Ccl21*) and immune-suppressive cytokines (*IL10* and *Tgfb*). Several cytokines have similar expression across all tumors, including *Tnfa*, *Ifng*, and *Ccl21*. Other cytokines (including *Tgfb*, *IL4*, *IL10*, and *IL1b*) display more variability across individual tumors, although this was not associated with specific background strains. Taken together, this gene expression analysis highlights key strain-dependent differences in the composition of the tumor microenvironment—most notably, the elevation of M2 macrophages in MPNSTs from BALB/c mice.
