Cooperative Genomic Lesions in HRAS-Mutant Cancers Predict Resistance to Farnesyltransferase Inhibitors

The clinical development of farnesyltransferase inhibitors (FTI) for HRAS-mutant tumors showed mixed responses dependent on cancer type. Co-occurring mutations may affect response. We aimed to uncover cooperative genetic events specific to HRAS-mutant tumors and study their effect on FTI sensitivity. Using targeted sequencing data from MSK-IMPACT and DFCI-GENIE databases we identified co-mutations in HRAS- vs KRAS- and NRAS-mutant cancers. HRAS-mutant cancers had a higher frequency of co-altered mutations (48.8%) in MAPK, PI3K, or RTK pathways genes compared to KRAS- and NRAS-mutant cancers (41.4% and 38.4%, respectively; p < 0.05). Class 3 BRAF, NF1, PTEN, and PIK3CA mutations were more prevalent in HRAS-mutant lineages. To study the effect of comutations on FTI sensitivity, HrasG13R was transfected into ‘RASless’ (Kraslox/lox;Hras−/−;Nras−/−) mouse embryonic fibroblasts (MEFs) which sensitized non-transfected MEFs to tipifarnib. Comutation in the form of Pten or Nf1 deletion or Pik3caH1047R or BrafG466E transduction led to relative resistance to tipifarnib in HrasG13R MEFs in the presence or absence of KrasWT. Combined treatment of tipifarnib with MEK inhibition sensitized cells to tipifarnib, including in MEFs with PI3K pathway comutations. HRAS-mutant tumors demonstrate lineage demonstrate lineage-dependent MAPK/PI3K pathway alterations that confer relative resistance to tipifarnib. Combined FTI and MEK inhibition is a promising combination for HRAS-mutant tumors.


Introduction
The global incidence of HRAS-mutant tumors is approximately 230,000 per year-representing 7% of all RAS-mutant tumors. (1) Activating mutations in the HRAS oncogene have been described with varying prevalence across multiple cancer lineages including bladder, head and neck squamous cell cancer (HNSCC), lung, thyroid and melanoma. (2) Targeting RAS-mutant tumors remains challenging.
(3) Treatment approaches aimed at targeting upstream and downstream effectors have been met with mixed responses (4-7). The recent development of KRAS-G12C inhibitors and their consequent FDA approval for non-small cell lung cancer shows the bene ts and limitations of directly inhibiting RAS drivers (8, 9), and highlights the importance of targeting other members of the RAS family.
Tipifarnib, a farnesyltransferase inhibitor (FTI), is a targeted therapy for HRAS-mutant tumors. (10)(11)(12) FTIs block the addition of the farnesyl lipid moiety on the C-terminus CAAX motif of HRAS and prevents its translocation and subsequent activation at the plasma membrane. (13,14) Although originally developed as therapy for KRAS-mutant tumors, the discovery of rescue prenylation by geranylgeranyl transferase results in an alternative pathway for activation of KRAS and NRAS, but not HRAS. (15) A phase II trial evaluating HNSCC patients with HRAS mutations showed reduction in tumor burden and improved progression free survival, particularly in those with a high variant allelic frequency of the mutant oncoprotein.(16) In addition, pilot trials have demonstrated e cacy of tipifarnib in patients with metastatic salivary, (17) and urothelial cancers(18) with HRAS mutations. However, for unclear reasons, similar responses have not been reported for other cancer types, including thyroid. from the bicistronic vector. Treatment with 600nM tamoxifen (4-OHT) for at least 2 weeks was used to ox-out the KRAS lox/lox allele.

Introduction of Co-Altered Mutations to HRAS G13R -Transfected MEFs
Co-altered mutations were introduced into "Rasless" MEFs transfected with the HRAS G13R allele. To knock out Nf1 a dual-guide RNA targeting RAS-binding domain in exon 21 and for Pten a dual-guide RNA against phosphatase domain in exon 5 were used. Respective dual-guide RNA with CAS9 and mCherry coexpression plasmids were custom designed and purchased from Vector builder (Supplementary Table   2). CRISPR-CAS9 plasmids were transfected using Fugene, transfected cells were selected by RFP ow sorting, single cell clones isolated, and con rmed for respective CRISPR knock outs. Gain-of-function Braf G466E (Class 3 BRAF) and Pik3ca H1047R sequences were cloned into Cs-Mm34053-Lv213 (CMVmutant gene-IRES2-mCherry-IRES-Blasticidin) (Addgene). They were then introduced using lentiviral transduction with 2 ug/mL blasticidin as the selection antibiotic for at least one week. Mass-transduced cells with the lowest 5% expression levels were selected as determined by ow cytometry using an IRES-mCherry signal as a reporter. Treatment with 4-OHT was used to delete KRAS as described previously (23).
Hras G13R /Nf1 LOF MEFs were not viable when sparcely plated and treated with 4-OHT. As such these cells were kept in highly con uent P10 plates, serially split and continuously treated with tamoxifen for a minimum of 3 weeks.

Determination of Cell Viability
Half maximal inhibitory concentration (IC 50 ) assays were done by plating on Day 0 either 70,000 cells in 6-well plates in DMEM media with 15% fetal bovine serum (FBS) and penicillin/streptomycin/L-glutamine (P/S/G; Gemini; #400-110) or 100,000 cells for serum-starved (1% FBS) conditions . Cells were treated with the indicated drug concentrations on day 1 with subsequent drug and media change on day 3. Cells were harvested on day 6 by trypsinization and counted using a Vi-Cell series Cell Viability Analyzer (Beckman Coulter). For combination treatments, we plated 3000 cells/well in a 48-well plate in complete media and assayed responses with a crystal violet assay. Cells were treated on Day 1, with drug and media change on day 3. Cells were subsequently xed and stained with crystal violet on day 6. The following day plates were imaged using Gelcount TM analyzer (Oxford Optronics) and subsequently resuspended in 10% acetic acid and counted using a microplate reader (Spectramax® iD5). IC 50 and crystal violet assays were done in triplicate.
Cell proliferation assays were done by plating 70,000 cells in 6-well plates and treating with either DMSO, monotherapy, or combination therapy for 24h, 48h, 72h, 96h, and 120h. Drug and media change was done at 72h, and cells were harvested at indicated time points. All assays were done in triplicate.

Western Blots
Cells treated in the indicated conditions were rinsed with cold PBS and lysed using RIPA (Millipore) supplemented with protease (Complete Mini, Roche) and phosphatase inhibitors (Phosphatase Inhibitor Cocktail Set I and II; Sigma). Protein lysates were collected by spinning cells in a 4-degree centrifuge at 14,000 RPM for 20min. Protein concentrations were measured using the BCA kit (ThemoFisher Scienti c) on a microplate reader (SpectraMax® iD5). SDS-PAGE and blocking were done as previously described (24), followed by transfer to a nitrocellulose membrane (0.2mm, Amersham) and blocking in 10% BSA for at least one hour. Membranes were incubated overnight with the indicated primary antibody in 10% BSA in a cold room. Membranes were then washed with Tris-buffered saline with 0.1% Tween® 20 detergent (TBST) and blocked with a Goat anti-Rabbit (IRDye® 680RD) or Goat anti-Mouse (IRDye® 800CW) secondary antibody. Membranes were then imaged using a LI-COR Odyssey® DLx Imaging System.

Statistical Analyses
Genomic analysis of gene signature and gene level mutation frequencies were used to compare coaltered lesion prevalence by cancer type and RAS-mutant type. Differences in mutation frequencies between the various cohorts was determined utilizing Fisher's exact test for binomial proportions.
Signi cance was established as p<0.05.
Statistical analysis for the in-vitro studies was performed using GraphPad Prism version 7.0., using twotailed non-parametric t-tests. Signi cance was established as p<0.05. IC 50 curves were generated on GraphPad using non-linear regression.

Prevalence of HRAS-mutant tumors amongst cancer types
Altogether 255 (1.4%) patients with canonical HRAS mutations were identi ed in the MSK-IMPACT and DFCI-Genie cohorts, with the number and relative frequency in the evaluated cancer types shown in Figure  1A, top panel. The prevalence of HRAS codon mutations differed by cancer type with substitutions at Q61 being the most common, which was particularly enriched in thyroid and prostate cancers (p<0.001 for both) ( Figure 1A, bottom).
Overall KRAS and NRAS were more prevalent than HRAS mutations (2516 (14.0%) and 701 (3.9%), respectively; p<0.001 vs HRAS for both), as shown in Figure 1B. A greater frequency of HRAS mutations were found in salivary gland and HNSCC cancers (both p<0.05). KRAS mutations were more prevalent in NSCLC (p<0.01) and NRAS mutations in thyroid and melanoma (both p<0.05).
A signi cantly greater proportion of HRAS-mutant cancers (48.4%) were found to have co-altered mutations along the RTK/MAPK/PI3K pathways compared to KRAS-(41.4%) and NRAS-mutant (38.4%) cancers ( Figure 1C) (p<0.05 for both). Among all cancer types, more than half of all HRAS-mutant cancers had co-altered lesions, with the exception of those of thyroid and prostate ( Figure 1D).

Co-Altered Mutations in RAS-Mutant Cancers
Co-altered mutations prevalent in HRAS-, KRAS-, and NRAS-mutant cancers are outlined in Figure 2. We combined co-altered mutations into signaling pathway categories (MAPK, PI3K/MTOR, and RTK). The prevalence of co-altered mutations in the MAPK pathway in melanomas was the highest among the cancer types we analyzed and was 61.1% for HRAS, 59.4% for KRAS, and 31.3% for NRAS-mutant tumors. In contrast, 27.2% of HRAS-mutant bladder cancers carried a co-altered MAPK mutation compared to 23% and 44% of KRAS and NRAS, respectively ( Figure 2A). Amongst HRAS-mutant melanomas, 42% carried a co-altered NF1 mutation, signi cantly greater than in KRAS-/NRAS-mutant melanoma and other HRAS-mutant cancers (p<0.05). In addition, class 3 BRAF mutations were more prevalent in HRAS-mutant salivary gland (5%), and bladder (9%), and melanoma (11%) cancers, whereas none were present in the respective KRASand NRAS-mutant tumors (p<0.05 for all). Class 2 BRAF mutations were also signi cantly more prevalent in HRAS (8%) compared to KRASand NRAS-mutant melanoma. Clonality analysis of HRAS-mutant/MAPK-co-altered tumors was limited to the MSK-IMPACT samples. With few exceptions HRAS mutations were clonal, whereas the clonality of the co-altered lesions was more variable ( Figure 2C).
Introduction of an Hras WT allele did not confer sensitivity to tipifarnib in the presence of a Kras WT allele (IC50: >3uM) but demonstrated exquisite sensitivity in the presence of tamoxifen (IC 50 : 0.03nM, p<0.001), consistent with the on-target effect of the FTI ( Figure 3C, middle). By contrast expression of Hras G13R ( Figure 3C, right) partially sensitized cells to tipifarnib (IC 50 : 324.7nM) as compared to parental and Hras WT -transfected cells even in the presence of Kras WT (p<0.001), whereas tamoxifen-induced Kras WT loss rendered Hras G13R -expressing cells highly sensitive to treatment (IC 50 : 0.6nM; p<0.001).
Tamoxifen-treated Hras G13R -transfected cells expanded more rapidly after 6 days (1.8 ± 0.1 x 10 6 vs. 1.0 ± 0.2 x 10 6 cells; p<0.01), suggesting that Kras WT expression had a dampening effect on the proliferative drive induced by mutant Hras ( Figure 3D). A time course of the effects of tipifarnib on signaling in Hras G13R -transfected cells treated without or with 4OHT is shown in Figure 1E. HRAS defarnesylation (clear arrow) was detectable 48-72h after exposure to tipifarnib regardless of 4OHT treatment. Baseline Hras and pERK levels were higher in MEFs lacking Kras WT consistent with its inhibitory effects on Hras G13R -induced cell growth, and were inhibited by tipifarnib, whereas pERK levels were not affected by tipifarnib in the absence of 4OHT. Taken together, expression of Kras WT attenuates the MAPK signaling and the proliferative drive induced by Hras G13R , and renders cells less sensitive to tipifarnib.
Introduction of Co-Altered Mutations Leads to Resistance to Tipifarnib: Following transduction of bicistronic lentiviral vectors for Braf WT or Braf G466E cDNA into Hras G13Rexpressing MEFs, cells were sorted for mCherry to select those in the 5 th percentile of expression, which mimicked endogenous levels of Braf ( Figure 4A). Sensitivity to tipifarnib was decreased in Braf G466E compared to Braf WT -expressing cells in the absence of 4OHT (IC 50 Braf G466E vs Braf WT : 800nM vs 161.9nM; p<0.05), whereas in 4OHT-treated cells they were highly sensitive to tipifarnib and not signi cantly different to each other (IC 50 Braf G466E vs Braf WT : 5.9nM vs 6.9nM) (Supplementary Figure 2).
Resistance of Hras G13R /Braf G466E cells to the growth inhibitory effects of tipifarnib was relieved by the combination with trametinib in cells with Kras WT expression, with the combination also being superior to trametinib alone ( Figure 4D). Addition of trametinib signi cantly reduced pERK after 6h, although rebound was observed with trametinib and combination tipifarnib:trametinib therapy ( Figure 4E). 4OHTtreatment partially sensitized cells to tipifarnib. Addition of trametinib inhibited their growth more profoundly, but the effects of the combination were not superior to trametinib monotherapy ( Figure 4F).
Tamoxifen-treated cells ( Figure 4F) cell viability demonstrated sensitivity to tipifarnib therapy by 120hrs however was still signi cantly reduced compared to combination 100nM:10nM tipifarnib:trametininb treatment (tipifarnib:trametininb-0.04 ± 0.01 vs tipifarnib-0.11 ± 0.01; p<0.001). In addition, trametinib demonstrated improved reduction in cell viability at all time points leading up to 120h. The deletion of Kras WT by the addition of 4OHT eliminated the late pErk rebound observed after treatment with trametinib ( Figure 4G).

NF1 Loss-Of-Function (Hras G13R /Nf1 LOF )
To study NF1 loss, sgRNAs targeting exon 21 of Nf1 were designed to create loss-of-function mutations in Hras G13R -transfected cells. Two rounds of CRISPR-Cas9 were required to rst generate a heterozygous clone and subsequently cells with homozygous Nf1 loss (Suppl Fig 3). Signalling was performed amongst various suspected homozygous clones with demonstrated increased pERK signal with NF1 loss-of-function ( Figure 4H). Clone 39, hereafter designated as Hras G13R /Nf1 LOF , was used for analysis.
The introduction of an Nf1 LOF homozygous mutation resulted in resistance to tipifarnib but retained sensitivity to trametinib (non-4OHT IC 50 : 1.73nM vs 4OHT IC 50 : 0.8nM) (Figures 4I and 4J), with consistent ndings in the cell viability and signalling assays ( Figure 4K-N). Interestingly, the rebound of pERK after trametinib treatment was still present in Hras G13R /Nf1 LOF cells treated with 4OHT. This is consistent with a loss of an inhibitory effect of Nf1 on mutant Hras, since the rebound is relieved by the combination of tipifarnib and trametinib.

Pten Loss-of-Function (Hras G13R /Pten LOF )
We generated Pten loss-of-function mutations in Hras G13R -transfected MEFs using sgRNA targeting of exon 5. Western blot analysis identi ed one clone with a homozygous Pten loss-of-function mutation and we observed concomitant increased pAkt-Ser473 signal ( Figure 5A). At baseline Hras G13R /Pten LOF cells were resistant to monotherapy with tipifarnib, which was partially relieved by 4OHT. Cells were insensitive to AZD-8186 (PI3KCB-inhibitor) or Pictilisib (pan-PI3K inhibitor) monotherapy in the absence or presence of 4OHT (Figures 5B and 5C). Crystal violet assays were also used to select an optimal combination drug combination to test cell viability in Hras G13R /Pten LOF cells (Supplementary Figure 4) which demonstrated relative resistance to combination PI3K inhibition.
In the absence of 4OHT, growth assays of Hras G13R /Pten LOF MEFs was insensitive to the tipifarnib:AZD8186 or tipifarnib:pictilisib combinations ( Figure 5D-E). When exposed to 4OHT, there was modest additive growth inhibitory effects of combination tipifarnib:AZD8186 therapy ( Figure 5F-G).
Western blots of Hras G13R /Pten LOF cells treated with AZD8186 or tipifarnib:AZD8186 showed reduction in pAkt-Ser473 and pAkt-Thr308, consistent with on target effects of the PI3KCB-inhibitor. This was also true for pictilisib, particularly in the absence of 4OHT (Supplementary Figure 5).
Next, we tested the combination of tipifarnib/MEK inhibitor in Hras G13R /Pten LOF cells given the sensitivity we observed in Hras G13R , Hras G13R /Braf G466E , and Hras G13R /Nf1 lof Rasless-MEFs. Crystal violet assays were again employed to de ne optimal concentrations of drug (Supplementary Figure 4). Despite the loss of function of Pten, trametinib monotherapy and combination tipifarnib:trametinib had signi cant antiproliferative effects that were augmented by 4OHT ( Figure 5F). Interestingly, trametinib was ineffective in inhibiting ERK phosphorylation in cells expressing wild type Kras. Treatment with 4OHT was associated with transient pERK inhibition by the MEK inhibitor, whereas the combination of tipifarnib and trametinib induced sustained MAPK inhibition, which was more accentuated in Kras-de cient cells. Akt phosphorylation was not inhibited by trametinib or trametinib:tipifarnib combination therapy. Hence, upon loss of Pten, Hras-mutant cells develop resistance to tipifarnib, which is relieved by MAPK pathway inhibition, but not by blocking PI3K signaling.
As in previous experiments, we identi ed sensitive concentrations of drugs using crystal violet assays (Supplementary Figure 6). Resistance to tipifarnib, alpelisib, and picitilisb monotherapy was not signi cantly overcome with combination tipifarnib:alpelisib or tipifarnib:pictilisib therapy ( Figure 6D-E). However, treatment with combination tipifarnib:trametinib signi cantly improved sensitivity, even more so in presence of 4OHT, associated with pERK inhibition (Figure 6F,G). Signaling responses to pictilisib:tipifarnib were similar to those of tipifarnib:alpelisib (Supplementary Figure 7). Baseline Hras G13R MEFs also conferred no added sensitivity with combination tipifarnib:AZD8186 and tipifarnib:alpelisib (Supplementary Figure 8)

Endogenous KRAS expression slows proliferation in HRAS G13R -transfected MEFs
We assessed the effect of endogenous Kras expression on proliferation of Hras G13R -transfected MEFs with or without co-mutations. Cells were grown for 6 days in 15% ( Figure 7A) or 1% FBS ( Figure   7B). Interestingly, Kras WT suppressed cell proliferation in 15% serum across all cell conditions, an effect that was partially relieved in Hras G13R/ Nf1 LOF cells. Kras WT knockdown with 4OHT derepressed cell growth in all the genetic contexts. These effects were serum-dependent, since in 1% serum growth of MEFs was unaffected by the presence or absence of Kras WT .

Discussion
Tipifarnib, a farnesyltransferase inhibitor, impedes the activation of oncogenic HRAS by preventing its translocation to the plasma membrane and subsequent downstream effector signaling. (14) Using targeted exome sequencing data we show that HRAS-mutant cancers are associated more commonly with co-mutations of genes encoding effectors in the MAPK and PI3K pathway than tumors driven by KRAS or NRAS, with speci c lineage differences. We used "RASless" MEF lines expressing mutant HRAS to investigate whether some of the more common co-mutations observed in human tumors affected their response to growth inhibition by tipifarnib, which is currently in clinical development for HRAS-mutant cancers. We found that concurrent mutations of PTEN, PIK3CA, NF1, or a class 3 BRAF mutation reduce the e cacy of tipifarnib. Regardless of whether the co-mutant protein canonically signals in the PI3K or MAPK pathway, combination treatment with a MEK inhibitor, but not with pan-PI3K or isoform selective PI3K inhibitors, enhances sensitivity to the FTI.
The presence of a wild-type KRAS allele attenuated serum-induced growth of the HRAS G13R -expressing MEFs. It is well established that Hras-WT dampens transformation by mutant HRAS in mouse models. (26) This is also true for Kras-WT in KRAS-mutant tumors, and for Nras-WT in NRAS-mutant tumors. (27) Moreover, allelic imbalance, often due to LOH of the corresponding wild-type alleles, is commonly present in oncogenic RAS-driven cancers of different lineages and associated with disease progression, (27) whereby wild-type RAS alleles reduce the activation threshold of mutant RAS alleles. (28, 29) The inhibitory effect of wild-type KRAS on oncogenic KRAS-induced growth has been attributed to homodimerization, since introduction of a mutation that impairs KRAS homodimer formation without affecting its effector function abrogates its growth inhibitory effects. (30) The crystal structure of HRAS identi ed a potential dimer interface similar to that of KRAS,(31) but it is unclear whether HRAS can heterodimerize with other RAS family proteins, or whether HRAS-WT dampens the growth promoting effects of HRAS G13R through alternative mechanisms.
Endogenously expressed wild-type Kras imparted resistance to FTI therapy in Hras G13R -transfected MEFs. Expression of wild-type RAS has been shown to contribute to adaptive resistance to therapeutic targeting of mutant BRAF or MEK. (27,32,33) In our models, wild-type KRAS decreased sensitivity of HRAS G13R -MEFs to tipifarnib even when co-altered mutations in the MAPK and PI3K pathway were introduced. These ndings suggest that allelic balance of wild type RAS proteins likely plays a role in response to tipifarnib therapy. The ratio of wild-type:mutant RAS alleles requires further study as a biomarker for FTI sensitivity of HRAS-mutant cancers.
Wild-type RAS proteins play an important role in determining rebound activation of the MAPK pathway that imparts resistance to targeted therapies. (10,34) In our study, trametinib induced rebound pERK signal by 48-72hrs in cells with any co-altered mutations in the presence of wild type Kras. We also noted rebound signaling of pAkt in the Hras G13R /Pten LOF and Hras G13R /Pik3ca H1047R cells. In the presence of tipifarnib and trametinib, cells that continued to express wild-type KRAS had similar rebound signaling that was abated when cells were treated with 4OHT. This is consistent with data demonstrating combination treatment targeting downstream MAPK signaling and upstream RTKs or SHP2/SOS is an effective anti-tumor strategy in RAS mutant tumors. (34)(35)(36)(37) We identi ed co-mutations in MAPK and PI3K pathway genes that, in the setting of mutant HRAS, imparted resistance to FTI therapy. Class 3 BRAF mutations and NF1 loss were common in various HRASmutant cancers, in particular melanomas. Loss-of function mutations to NF1, a RAS GTPase activating protein, confer resistance to tyrosine kinase inhibitors and targeted therapies across tumor types.(38-40) We previously showed that knock-down of Nf1 in murine Hras-mutant poorly differentiated thyroid cancer cell lines imparted resistance to FTI therapy via feedback activation of wild type RAS proteins, which required MEK inhibition to abate rebound signaling and enhance growth inhibition.(10) CRISPR-Cas9 knockout of NF1 also caused profound resistance to tipifarnib in HRAS G13R MEF, which was only partially abrogated by KRAS deletion. This latter nding is intriguing, as it could imply that loss of NF1 either further enhances mutant HRAS activity, or is acting in a RAS-independent manner in the otherwise RASless cells.
Class 3 BRAF mutations are insensitive to drugs that target class 1 BRAF mutant proteins that signal as monomers, such as dabrafenib or vemurafenib (41,42). They represent a subset of BRAF mutant proteins that heterodimerize with CRAF and are RAS-dependent for activation. (43) As a result, we hypothesized that a class 3 BRAF co-mutation in HRAS G13R -MEFs would render cells sensitive to tipifarnib treatment. This was true when Kras-WT was deleted, but not when KRAS was present, consistent with the dependence of BRAF G466E on RAS signaling.
We found that comutations in the PI3K/mTOR pathway also imparted resistance to FTIs.
Mutations in the PI3K pathway have been reported in patients with KRAS G12C -mutant tumors who developed resistance to the KRAS G12C inhibitor adagarsib in the KRYSTAL-1 trial. (44) Combination tipifarnib and the PI3KA inhibitor alpelisib is currently being used in a Phase II trial to target HRAS-mutant tumors with PIK3CA mutations (KURRENT-HN trial; NCT04809233). We found the combination of tipifarnib and alpelisib to have limited e cacy in the Hras G13R /Pik3ca H1047R MEFs, which was modestly improved by deletion of Kras-WT. Instead the combination of the MEK inhibitor trametinib and tipifarnib was highly effective for Hras G13R MEFs harboring either Pik3ca H1047R or deletion of Pten. Interestingly, Pten loss enhances MAPK activation and MEK inhibitor sensitivity in a mouse model of Her2/neu-driven breast cancer. (45) These ndings indicate that persistent activation of MAPK is a dominant mechanism of resistance to FTIs even in the presence of PI3K pathway activation. An important caveat to our ndingsis is that the experiments were conducted in mouse embryonic broblasts, which may not fully re ect cell lineage-speci c signaling or therapeutic dependencies.
In summary, results from our study highlight the importance of co-altered mutations and of wildtype RAS in driving resistance to targeted therapy of HRAS-mutant cancers. design of the project, data interpretation, and revision/ nalization of the manuscript. WKC was involved in the design of the project, data anaylsis and interpretation, nalization of graphs, and revision/ nalization of the manuscript. KB was involved in acquiring data, data interpretation, and manuscript revision/ nalization. MS assisted in conception and design of the project, data interpretation, and revision/ nalization of the manuscript. HW assisted with data acquisition and analysis, data interpretation, and revision/ nalization of the manuscript. ALH was involved in project conception, data interpretation and analysis, and nalization of the manuscript. NS helped with project conception and design, data anaylsis and interpretation, and revision/ nalization of the manuscript. JAF was involved with conception and design of the manuscript, data anaylsis and interpretation, project direction, and revision and nalization of the manuscript. BRU assisted with conception and design of the project, experiment planning, data analysis and interpretation, and revision/ nalization of the project.       Cell proliferation in HRAS G13R -transfected MEFs is slowed in the preszence of endogenous KRAS WT . Cell proliferation after 6 days of untreated growth were tested in (A) 15% serum and (B) 1% serum conditions. HRAS G13R -transfected cells were tested along with indicated co-mutations witht and without 4-OHT in both high and low serum conditions. (t-test, *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001, ns=not signi cant).