Genetically manipulating endogenous Kras levels and oncogenic mutations in vivo influences tissue patterning of murine tumorigenesis

  1. Özgün Le Roux
  2. Nicole LK Pershing
  3. Erin Kaltenbrun
  4. Nicole J Newman
  5. Jeffrey I Everitt
  6. Elisa Baldelli
  7. Mariaelena Pierobon
  8. Emanuel F Petricoin
  9. Christopher M Counter  Is a corresponding author
  1. Department of Pharmacology & Cancer Biology, Duke University Medical Center, United States
  2. Department of Pathology, Duke University Medical Center, United States
  3. Center for Applied Proteomics and Molecular Medicine, School of Systems Biology, George Mason University, United States

Abstract

Despite multiple possible oncogenic mutations in the proto-oncogene KRAS, unique subsets of these mutations are detected in different cancer types. As KRAS mutations occur early, if not being the initiating event, these mutational biases are ostensibly a product of how normal cells respond to the encoded oncoprotein. Oncogenic mutations can impact not only the level of active oncoprotein, but also engagement with proteins. To attempt to separate these two effects, we generated four novel Cre-inducible (LSL) Kras alleles in mice with the biochemically distinct G12D or Q61R mutations and encoded by native (nat) rare or common (com) codons to produce low or high protein levels. While there were similarities, each allele also induced a distinct transcriptional response shortly after activation in vivo. At one end of the spectrum, activating the KrasLSL-natG12D allele induced transcriptional hallmarks suggestive of an expansion of multipotent cells, while at the other end, activating the KrasLSL-comQ61R allele led to hallmarks of hyperproliferation and oncogenic stress. Evidence suggests that these changes may be a product of signaling differences due to increased protein expression as well as the specific mutation. To determine the impact of these distinct responses on RAS mutational patterning in vivo, all four alleles were globally activated, revealing that hematolymphopoietic lesions were permissive to the level of active oncoprotein, squamous tumors were permissive to the G12D mutant, while carcinomas were permissive to both these features. We suggest that different KRAS mutations impart unique signaling properties that are preferentially capable of inducing tumor initiation in a distinct cell-specific manner.

Editor's evaluation

This article addresses the long-standing question of why specific mutations in RAS are associated with tumors arising in distinct tissues; whether this reflects biological expression or functionality. The authors utilize a clever mouse genetic approach to modulate both which Kras mutation is expressed and how much of that Kras mutant is expressed, expressing the same Kras mutation at either a low level (due to rare codon utilization) or at a higher level (through common codon utilization). This study convincingly supports the conclusion that different KRAS mutations impart unique signaling properties that are preferentially capable of inducing tumor initiation in a distinct cell-specific manner.

https://doi.org/10.7554/eLife.75715.sa0

Introduction

Why specific driver mutations track with different cancers is unknown, yet speaks to the very origins of cancer, with implications for early detection and prevention. This is particularly well illustrated with the RAS family of small GTPases, comprised of KRAS, NRAS, and HRAS. Focusing on KRAS, the most commonly mutated of the three (Prior et al., 2020), single-point mutations at one of three hotpot positions (G12, G13, and Q61) result in six possible substitutions that inhibit the intrinsic or extrinsic GTPase activity of the protein. These mutations render KRAS constitutively GTP-bound and active, which is well known to be oncogenic (Simanshu et al., 2017). Although this amounts to 18 possible oncogenic mutations, specific subsets tend to be found in different cancer types. For example, the most common oncogenic KRAS mutation is G12C in non-small cell lung cancer but Q61H in plasma cell myeloma (Prior et al., 2020). Mice similarly exhibit a bias of specific oncogenic Kras mutations towards different cancer types as different mutant Kras alleles have different tumorigenic potential when activated in different tissues (Li et al., 2018; Poulin et al., 2019; Winters et al., 2017; Wong et al., 2020; Zafra et al., 2020). Different oncogenic mutations also affect the ability of Nras to induce tumorigenesis in mice, as do the same mutations in different RAS isoforms (Burd et al., 2014; Haigis et al., 2008; Kong et al., 2016; Murphy et al., 2022). While this tissue ‘tropism’ of cancers towards specific RAS mutations has been appreciated for decades (Bos, 1989), the underlying mechanism is unclear.

Variation in the ability of specific KRAS mutants to be tumorigenic (or not) in different tissues ostensibly results from differences in oncogenic signaling between mutants and the response of normal cells thereof. Generally speaking, oncogenic signaling is a product of the amplitude of the signal (quantitative signaling) and/or the pathways engaged (qualitative signaling). In terms of quantitative signaling, different mutations can exhibit different degrees of activation (GTP-loading) and/or different sensitivities to positive (Ras GTP exchange factors [RasGEFs]) or negative (RAS GTPase activating proteins [RasGAPs]) regulators (Gebregiworgis et al., 2021; Munoz-Maldonado et al., 2019; Simanshu et al., 2017). Various methods to manipulate quantitative Kras signaling, such as through modulating recombination rates (Singh et al., 2021), homozygous expression of the mutant allele (Wang et al., 2011), changing codon usage to increase protein expression (Pershing et al., 2015), additional pharmacological activation of the mitogen-activated protein kinase (MAPK) pathway (Cicchini et al., 2017), and so forth (Li et al., 2018), all affect tumorigenesis. On the other hand, perhaps one of the best examples of qualitative differences in KRAS signaling is the G12R mutant. Unlike the more canonical G12D mutation, the G12R mutant exhibits reduced the RAS effector PI3Kα binding and PI3K/AKT signaling (Hobbs et al., 2020). Activating an inducible KrasLSL-G12R allele in the pancreas led to fewer early premalignant lesions compared to the much more tumorigenic KrasLSL-G12D allele (Zafra et al., 2020). Despite an appreciation that different KRAS mutations can manifest in quantitative or qualitative signaling differences, how each contributes to the mutational patterns of this oncogene in cancer is unclear.

KRAS mutations are often initiating, being sufficient to induce tumorigenesis in mice and truncal in many human cancers. Thus, the bias of specific cancers towards distinct KRAS mutations could arise from potential tissue-specific differences of normal cells in the mutagenic process or repair capacity and/or in their responses to the nature of the signaling imparted by specific KRAS mutations (Li et al., 2018). In regard to the latter, determining the immediate response of normal cells to different KRAS mutations in vivo may help elucidate the mutational patterning of this oncogene. Identifying an oncogenic mutation arising in the KRAS gene from the cell-of-origin prior to becoming a tumor is challenging in humans. Mice, on the other hand, provide an ideal model system to experimentally explore this phenomenon as the point of tumor initiation can be precisely defined using inducible oncogenic Kras alleles. To therefore explore why different cancer types have a bias towards specific KRAS mutations, we created four novel inducible murine Kras alleles with different oncogenic mutations that were expressed at either low or high levels. In this way, we hypothesize that tissues in which tumor initiation is driven by the level of oncogenic signaling would preferentially develop tumors at higher protein levels while those driven by a specific mutant would preferentially develop tumors with just one of the two mutants.

We chose two completely different oncogenic mutations for these alleles, G12D and Q61R. G12D places a negatively-charged headgroup into the catalytic cleft of RAS and blocks extrinsic (RasGAP-mediated) GTPase activity (Parker et al., 2018). On the other hand, Q61R replaces the catalytic amino acid with one that has a positively-charged headgroup, disrupting the position of the active site water molecule necessary for intrinsic GTP hydrolysis (Buhrman et al., 2010). Q61 is also essential for extrinsic GTP hydrolysis, as it stabilizes the transition state via hydrogen bonds to the γ-phosphate and nucleophilic water while providing another hydrogen bond to the RasGAP arginine finger (Grigorenko et al., 2007; Kötting et al., 2008; Rabara et al., 2019; Scheffzek et al., 1997). Comparing these two mutants directly reveals Q61R to have significantly lower GTP exchange and hydrolysis rates than G12D in Nras (Burd et al., 2014) and reduced RasGAP-mediated GTP hydrolysis rates in Kras (Rabara et al., 2019), akin to other substitutions at these two positions (Gebregiworgis et al., 2021; Smith et al., 2013). In those few cases in which the tumorigenic potential of the G12D and Q61R mutants of the same Ras isoform has been directly compared in mice, tissue-specific expression of the G12D mutant was less oncogenic than Q61R in both Nras-induced melanoma (Burd et al., 2014) and Kras-induced myeloproliferative neoplasm (Kong et al., 2016).

To alter the expression of these two mutants, the first three coding exons of Kras were fused and encoded by either their native rare codons, which is known to retard Kras protein expression, or common codons to increase Kras protein expression (Ali et al., 2017; Fu et al., 2018; Lampson et al., 2013; Pershing et al., 2015). We chose the novel approach of altering mammalian codon usage to modulate protein expression in vivo (Li and Counter, 2021; Pershing et al., 2015; Sasine et al., 2018) as no additional elements are required to change protein levels, providing a simple, reproducible, and uniform way of modulating Kras levels in mice and derived cell lines.

These four alleles were activated and immediately thereafter the in vivo transcriptome was determined in the lungs. This revealed that while these four alleles certainly shared similarities, each also induced unique transcriptional responses. Increased expression shifted the transcriptional hallmarks from those indicative of an expansion of multipotent cells to that of hyperproliferation and oncogenic stress. Changing the mutation type shifted the hallmark of estrogen response in the G12D mutant to that of the p53 pathway and DNA repair in the Q61R mutant. All four alleles were then globally activated, revealing that hematolymphopoietic lesions were permissive to the level of active oncoprotein, squamous tumors were preferentially permissive to the G12D mutant, while carcinomas tended to be permissive to both these changes. These findings support tissue-specific responses to the degree and type of signaling imparted by different oncogenic mutations molds the tissue tropism of cancers towards specific oncogenic KRAS mutations.

Results

A panel of inducible oncogenic Kras alleles designed to separate the effects of mutation type from the activity level of the oncoprotein

To explore how specific cancers are driven by distinct KRAS mutations, we reasoned that effects unique to a mutation could be separated from those imparted by activation level by simply comparing two different oncogenic mutants expressed at high or low levels. To this end, we created the four novel Cre-inducible oncogenic Kras alleles KrasLSL-natG12D, KrasLSL-natQ61R, KrasLSL-comG12D, and KrasLSL-comQ61R (Figure 1A). Each began with an LSL transcriptional/translational repressor sequence (STOP) flanked by loxP sites (Jackson et al., 2001) engineered into the intron of Kras following the first non-coding exon to provide temporal and spatial control of gene expression. This was followed by a fusion of the first three coding exons encoded by either their native (nat) rare codons, which are known to retard protein expression, or 93 of these rare codons converted to their common (com) counterparts to increase protein expression (Figure 1—figure supplement 1 and Figure 1—source data 6). Each version contained either a G12D or Q61R mutation. As noted above, these two mutants alter RAS activity in a biochemically different manner (Munoz-Maldonado et al., 2019; Simanshu et al., 2017), with Q61R reported to yield higher levels of active (GTP-bound) Ras (Burd et al., 2014; Kong et al., 2016; Pershing et al., 2015). This was followed by the next intron containing an FRT-NEO-FRT cassette for ES selection, which was excised via Flp-mediated recombination in the resultant animals, after which the Flp transgene was outbred. Finally, the remainder of the gene was left intact so as to generate the two Kras4a and Kras4b isoforms, as both contribute to tumorigenesis (To et al., 2008), potentially through unique protein interactions (Amendola et al., 2019).

Figure 1 with 6 supplements see all
Conditional KrasLSL alleles with different oncogenic mutations and codon usage.

(A) Schematic of generating and activating KrasLSL alleles with the first three coding exons fused and encoded by native (nat) versus common (com) codons with either a G12D or Q61R mutation. (B) PCR genotyping of two independently derived mouse embryonic fibroblast (MEF) cultures (two biological replicates) with the indicated Kras alleles in the absence and presence of Cre recombinase (CRE) to detect the unaltered wild-type Kras allele product (WT, 488 bp) and the unrecombined (KrasLSL*, 389 bp) and recombined (LoxP recombined, 616 bp) KrasLSL allelic products. Gel images were cropped and color inverted for optimal visualization. Full-length gel images are provided in Figure 1—source data 2. (C) Expression levels, determined by immunoblot with an anti-Kras antibody, RAS activity levels, determined by RBD pull-down (RBD-PD) of lysates collected from MEF cells derived from mice with the indicated Kras alleles in the presence of Cre recombinase (CRE). MEF cultures derived from KrasLSL-natG12D/+ mice in the absence of Cre recombinase were used as negative control. MEF cultures were either serum starved overnight (starved) or serum starved overnight followed by serum stimulation for 5 min (stimulated). 20% of the elute from RBD-PD and 30 μg total protein from the total cell lysates were loaded. Tubulin serves as loading control. One of two biological replicates; see Figure 1—figure supplement 3 for the second biological replicate. Full-length gel images are provided in Figure 1—source data 3.

Figure 1—source data 1

Full-length gel images of RBD pull downs.

Full-length gel images from RAF1-RBD pull downs (top) and whole-cell lysates (bottom) from HEK-HTs ectopically expressing engineered Kras constructs shown in Figure 1—figure supplement 2.

https://cdn.elifesciences.org/articles/75715/elife-75715-fig1-data1-v1.png
Figure 1—source data 2

Full-length gel image of genotyping of mouse embryonic fibroblast (MEF) cultures derived from the KrasLSL alleles in Figure 1B.

PCR genotyping of two independently derived MEF cultures with the indicated KrasLSL alleles in the absence and presence of Cre recombinase (CRE) to detect the unaltered wild-type Kras allele product (WT, 488 bp) as well as the unrecombined (KrasLSL, 389 bp) and recombined (LoxP recombined, 616 bp) Kras allelic products. Gel images were color inverted for better visualization. Red box depicts region shown in Figure 1B.

https://cdn.elifesciences.org/articles/75715/elife-75715-fig1-data2-v1.zip
Figure 1—source data 3

Full-length gel images of RBD pull-downs in MEFs.

Full-length gel images from RBD pull-downs (left) and whole-cell lysates (right) from MEFs derived from KrasLSL alleles with serum starvation, or serum starvation followed by serum stimulation shown in Figure 1C, and same conditions with a second clone of MEF cultures with serial dilutions of 500 μg lysate (left) and 200 μg lysate (right) used for RBD-PD as shown in Figure 1—figure supplement 3. Red box depicts regions shown in Figure 1C and Figure 1—figure supplement 3.

https://cdn.elifesciences.org/articles/75715/elife-75715-fig1-data3-v1.png
Figure 1—source data 4

Full-length gel images of Kras expression and activity using RBD pull-downs from lung tissue.

Full-length gel images from RAF1-RBD pull-downs (RBD-PD, top) and whole-cell lysates (bottom) of lungs from mice with KrasLSL alleles seven days after tamoxifen injection as shown in Figure 1—figure supplement 4A. Immunoblots of two separate pull downs from two biological replicates are shown as in Figure 1—figure supplement 4B. Red box depicts the regions shown in Figure 1—figure supplement 4B.

https://cdn.elifesciences.org/articles/75715/elife-75715-fig1-data4-v1.png
Figure 1—source data 5

Ct values from the qRT-PCR analysis in Figure 1—figure supplement 5.

https://cdn.elifesciences.org/articles/75715/elife-75715-fig1-data5-v1.xlsx
Figure 1—source data 6

Sequence of coding exons 1 to 3 of the four KrasLSL alleles.

Sequence alignment of the codons in the coding exons 1 to 3 of the indicated KrasLSL alleles in comparison to the murine wild-type sequence (Kras). Top: amino acid sequence. Red nucleotides: optimized codons. Green highlight: G12D mutation. Blue highlight: Q61R mutation.

https://cdn.elifesciences.org/articles/75715/elife-75715-fig1-data6-v1.xlsx

To confirm that the expression and activity of these four versions of oncogenic Kras was reflected in their design, we ectopically expressed the corresponding FLAG-tagged murine Kras cDNAs in HEK-HT cells that critically depend upon oncogenic RAS for tumorigenesis (Hahn et al., 1999). The total amount of ectopic Kras protein was then determined by immunoblot with an anti-FLAG antibody, and the level of GTP-bound and active Kras was determined by pull-down with a Ras binding domain (RBD) peptide followed by immunoblot or ELISA. Based on common codons increasing protein expression and Q61R leading to higher levels of GTP-bound RAS, the four constructs display the expected stepwise increase in RBD-bound Kras in the ascending order of KrasnatG12D<KrasnatQ61R <KrascomG12D<KrascomQ61R in both assays (Figure 1—figure supplement 2, Figure 1—source data 1).

The effect of codon usage and mutation type on Kras activity in derived murine cells

To evaluate the endogenous functionality of this allelic set, we derived and characterized two separate mouse embryonic fibroblast (MEF) cultures from each of the genotypes KrasLSL-natG12D/+, KrasLSL-natQ61R/+, KrasLSL-comG12D/+, and KrasLSL-comQ61R/+, and validated that all four alleles could be recombined upon the expression of Cre recombinase (Figure 1B, Figure 1—source data 2). We then confirmed that this allelic set yields an increase in endogenous Kras activity consistent with codon usage and mutation type. These pairs of MEF cultures expressing Cre were either serum starved or stimulated to assess basal or stimulated Kras activity, respectively. RBD pull-down followed by immunoblot demonstrated the stepwise increase in Kras expression and activity levels in the ascending order of KrasnatG12D<KrasnatQ61R <KrascomG12D<KrascomQ61R in both conditions (Figure 1C, Figure 1—figure supplement 3, Figure 1—source data 3). Thus, endogenous expression of these four alleles revealed that the Q61R mutant is more active than the G12D mutant, and changing rare codons to common increases the expression and activity of endogenous Kras oncoprotein.

The effect of codon usage and mutation type on Kras activity in vivo

To validate these findings in vivo, we crossed the KrasLSL-natG12D/+, KrasLSL-natQ61R/+, KrasLSL-comG12D/+, and KrasLSL-comQ61R/+ genotypes into a Rosa26CreERT2/+ background, which expresses a tamoxifen-inducible Cre from the endogenous Rosa26 promoter that is active in a broad spectrum of tissues (Ventura et al., 2007). Two adult mice from each of the four derived cohorts, as well as the control strain (Rosa26CreERT2/+), were injected with tamoxifen, and seven days later humanely euthanized, their lungs removed, and duplicate samples of derived protein lysates subjected to RBD pull-down followed by immunoblot (Figure 1—figure supplement 4A). As in the case of the MEF cultures, we observed a stepwise increase in Kras expression and activity consistent with the codon usage and oncogenic mutation in this allelic set (Figure 1—figure supplement 4B C, Figure 1—source data 4). To assess whether this was reflected in the expression of downstream target genes, we generated the same four cohorts of mice, but in this case isolated RNA from the lungs of euthanized mice seven days after tamoxifen injection and determined the level of mRNA encoded by five known Ras target genes by qRT-PCR (Figure 1—figure supplement 5A). This revealed the expected increase in three such transcripts when Kras was encoded with common codons, which was further increased in the Q61R-mutant background (Figure 1—figure supplement 5B C, Figure 1—source data 5). To further validate this effect at the protein level, we performed reverse-phase protein array (RPPA) analysis of lung tissue upon activation of the two extreme alleles, KrasLSL-natG12D versus KrasLSL-comQ61R in the Rosa26CreERT2/+ background, with Rosa26CreERT2/+ serving as a control. Three mice from each genotypes were injected with tamoxifen and seven days later euthanized, their lungs removed, embedded, and derived protein lysates subjected to RPPA analysis. When normalized to the control tissue, there was a clear increase in the level of phosphorylated proteins of the MAPK pathway and downstream transcription factors upon activating the KrasLSL-comQ61R allele compared to activating the KrasLSL-natG12D allele (Figure 1—figure supplement 6, Supplementary file 1). Thus, the increase in Kras activity imparted by the Q61R mutant or by common codons observed in cultured cells is recapitulated in lung tissue.

The effect of codon usage and mutation type on Kras biological activity in vivo

We next assessed whether these alleles were proportionally tumorigenic in a side-by-side comparison in the same organ. Each of these four alleles were crossed into a CC10CreER/+ (and Rosa26CAG-fGFP/+) background, and the resultant mice were injected with tamoxifen to specifically induce Cre-mediated recombination in the lung, after which every month thereafter for 6 months, five mice from each of the four cohorts were euthanized (Figure 2—figure supplement 1A). The lungs from all mice were visually analyzed for the presence of surface pulmonary tumors (Figure 2—figure supplement 1B), and in addition, two H&E-stained sections from pairs of mice were assayed for the presence and type of pulmonary tumors by a veterinarian pathologist blinded to the genotype (Figure 2A). This revealed a stepwise increase in early-onset and tumor burden of pulmonary lesions in lock step with the increased biochemical activity of the oncoproteins in the ascending order of KrasnatG12D<KrasnatQ61R <KrascomG12D<KrascomQ61R (Figure 2A, B). These tumorigenic phenotypes were the result of activating the inducible Kras alleles in the lungs as five control wild type (Kras+/+) mice in the same background that were injected with tamoxifen failed to develop tumors after 13 months, twice the length of the study (Figure 2—figure supplement 1C, D). Finally, we find preliminary evidence in a few animals that the G12D-mutant alleles preferentially developed atypical alveolar hyperplasia (AAH), whereas bronchiolar hyperplasia/dysplasia (BH) lesions were more prevalent with the Q61R-mutant alleles (Figure 2—figure supplement 1E F). We conclude that this allelic set exhibits an increase in tumorigenic potential consistent with the activity of the encoded oncoproteins, but potentially may also display evidence of mutation-specific effects on tumorigenesis.

Figure 2 with 1 supplement see all
Biological effect upon activating each oncogenic Kras allele.

(A) Examples of H&E-stained lung sections and (B) the mean ± SD % tumor burden from microscopic analysis of two lung sections from five mice with the indicated KrasLSL alleles in a CC10CreER/+;Rosa26CAG-fGFP/+ background at each of the indicated times post-tamoxifen injection.

The response of lung tissue to Kras alleles with different codon usage or mutations in vivo

As the different tumor types and grades that arose upon activation of each of these four inducible oncogenic Kras alleles are presumably a product of tumor initiation, understanding these effects presumably lies in how cells immediately respond to these different oncoproteins. We thus compared the immediate transcriptional response, as a measure of the cellular changes upon activation of each allele, in the lung. Cohorts of three adult mice from each of the four inducible oncogenic Kras alleles in Rosa26CreERT2/+ background were injected with tamoxifen, and seven days later the animals were euthanized, their lungs removed, and RNA was isolated for bulk transcriptome sequencing (Figure 3—figure supplement 1A).

The resultant transcriptomes (GSE181628) exhibited a gradual increase in the number of genes differentially expressed with the level of active Kras (as assessed above by the RBD pull-down assay in lung tissue). Namely, we identified 7, 15, 73, and 701 genes were differentially expressed (with absolute log2 fold change larger than 2, and adjusted p-value less than 5%) upon activation of the KrasLSL-natG12D, KrasLSL-natQ61R, KrasLSL-comG12D, and KrasLSL-comQ61R alleles, respectively, compared to the wild-type Kras allele (Figure 3A, Figure 3—figure supplements 26, Figure 3—source data 1). Principal component analysis of these four transcriptomes revealed that the transcriptome upon activating both alleles encoded by native rare codons differed from both alleles encoded with common codons,while the G12D mutant was most distinct from the Q61R mutant in the KrasLSL-com background (Figure 3—figure supplement 1B). Similarly, when two mutants were compared, the ratio of uniquely differentially expressed genes specific to each allele increased from the G12D to the Q61R mutant when Kras was encoded with common codons (34–93%) more than it did when encoded with native rare codons (0–6%) (Figure 3—figure supplement 6, Figure 3—source data 1). Comparing the transcriptomes of the two Kras alleles encoded with native rare versus common codons revealed that the two native-encoded alleles decreased KRAS Signaling UP hallmarks while the two common-encoded alleles induced signaling events related to oncogenic stress, such as an increase in Oxidative Phosphorylation, Glycolysis, Reactive Oxygen Species, MTORC1 signaling, Peroxisome, Xenobiotic Metabolism, and UV Response UP hallmarks and a decrease in UV Response DN, Hedgehog Signaling, and Apical Junction hallmarks (Figure 3B, Figure 3—figure supplement 1C, Figure 3—source data 2). We interpret this as expression level largely dictating the degree of oncogenic signaling in this allelic set. Comparing the G12D versus Q61R transcriptomes revealed both G12D-mutant alleles induced the Estrogen Response Late hallmark while the Q61R-mutant alleles induced DNA Repair and P53 Pathway and decreased Interferon Alpha Response hallmarks, suggesting that there is a strong tumor-suppressive response uniquely activated by the Q61R-mutant alleles (Figure 3C, Figure 3—figure supplement 1C, Figure 3—source data 2).

Figure 3 with 9 supplements see all
Lung transcriptome induced by each oncogenic Kras allele.

(A) Volcano plot of the log2 fold change versus p-value of the genes showing differential expression in each allele compared to the wild-type (+/+) Kras allele in the lung. Full-sized plots are provided in Figure 3—figure supplements 25. (B, C) Normalized enrichment score and patterns of the indicated Gene Set Enrichment Analysis (GSEA) hallmarks differentially enriched by Kras codon usage (expression) (B) and mutation type (C). Only hallmarks with a false discovery rate (FDR) < 5% are shown. Dot size is adjusted to RAS activity for better visualization. All GSEA hallmarks differentially enriched upon activating KrasLSL alleles with an FDR < 5% are provided in Figure 3—figure supplement 7. (D) STRING analysis of the top ten genes in the differentially enriched GSEA hallmarks identified in RNA-seq analysis of the lungs of Rosa26CreERT2/+;KrasLSL-natG12D/+ versus Rosa26CreERT2/+;KrasLSL-comQ61R/+ mice seven days after tamoxifen injection.

To validate the transcriptional responses detected by bulk RNA-seq analysis, we quantified four to six marker genes in a subset of selected hallmarks. Two adult mice from each of the four cohorts and control mice were injected with tamoxifen as above, and seven days later the animals were euthanized, their lungs removed, RNA isolated, and the level of select transcripts determined by qRT-PCR. We identified similar expression patterns to the transcriptome signature in the hallmarks of TNFα Signaling via NFκβ and Interferon-γ, which increase with Kras activity (Figure 3—figure supplement 7A B, Figure 3—source data 3), and EMT and Myogenesis, as these hallmarks were enriched in G12D mutants but depleted in Q61R mutants (Figure 3—figure supplement 7A C, Figure 3—source data 3).

To probe these transcriptomes for evidence of an orchestrated response of normal cells to different Kras mutants, we repeated transcriptome analysis of lung tissue on the two extreme cases, namely, KrasLSL-natG12D versus KrasLSL-comQ61R (GSE181627). Activation of the KrasLSL-natG12D allele resulted in transcriptional signatures suggestive of an expansion of multipotent cells (Figure 3—figure supplement 8A B). Namely, Gene Set Enrichment Analysis (GSEA) hallmarks indicative of Epithelial Mesenchymal Transition (EMT, TGFβ Signaling, Wnt/βcatenin Signaling, Notch Signaling, Apical Junction, Apical Surface, and Estrogen Response Early) and multiple cell lineages (Adipogenesis, Myogenesis, Hedgehog Signaling, and Pancreas β Cells). Conversely, activation of the KrasLSL-comQ61R allele had all features of a potent oncogenic signaling leading to hyperproliferation and oncogenic-induced stress. Namely, GSEA hallmarks indicative of high oncogenic signaling (KRAS Signaling UP) and unrestrained proliferation (MYC Targets V1, and E2F Targets), leading to reactive oxygen species (Oxidative Phosphorylation, and Reactive Oxygen Species Pathway) and a DNA damage response (DNA Repair, P53 Pathway, and G2M Checkpoint) followed by apoptosis (Apoptosis) and inflammation (Inflammatory Response, IL6 JAK STAT3 Signaling, TNFα Signaling via NF-κβ, and Allograft Rejection).

To assess the response of cells at the protein level, we performed RPPA analysis of lung tissue from these same two genotypes. This revealed that the level of protein/phosphoproteins of RAS/MAPK, PI3K/AKT, Growth, DNA repair, senescence/autophagy/apoptosis, and IL/Jak/Stat pathways was higher upon activation of the KrasLSL-comQ61R compared to the KrasLSL-natG12D allele (Figure 3—figure supplement 9 and Supplementary file 1), consistent with the transcriptome analysis. Finally, STRING analysis of the top ten genes from each of these GSEA hallmarks revealed potential crosstalk between the various signaling pathways, suggesting a consolidated signaling program by the two different oncoproteins manifests in very different responses by normal cells (Figure 3D). We suggest that this allelic set moves the response of normal lung cells from an expansion of multipotent cells to one of extreme oncogenic signaling and stress, with the type of oncogenic mutation potentially further modulating these responses.

Tissue sensitivities to the different oncogenic Kras alleles

We next addressed the ‘tropism’ of tissues towards specific Kras mutants by determining the tumor landscape upon globally activating each allele. As each allele is expected to have a different oncogenic potential, we opted for a moribundity endpoint, as opposed to a fixed endpoint, to identify the tissues most permissive to tumorigenic conversion in a competition-based approach. To this end, each allele was again activated by tamoxifen using the aforementioned ubiquitous Cre driver Rosa26CreERT2. Mice were regularly monitored for moribundity endpoints, indicative of ensuing mortality due to cancer, at which time the animals were euthanized (Figure 4—figure supplement 1). As a control to rule out a tumor phenotype being a product of variations in gene activation, we validated Cre-mediated recombination between all four alleles across 14 diverse tissues seven days after tamoxifen injection. Only the ovary displayed reduced recombination, and hence was not included in the study (Figure 4—figure supplement 2, Figure 4—source data 1, and Supplementary file 2).

Plotting the percent survival for each of the four genotypes by the Kaplan–Meier approach revealed that the median life span of mice progressively increased from 14 days upon activating the KrasLSL-comQ61R allele to 150 days upon activating the KrasLSL-natG12D allele (Figure 4A). Not surprisingly, the number of tumors per animal mirrored these survival differences (Figure 4—figure supplement 3). Pairwise comparisons revealed that median survivals were statistically different, except between the KrasLSL-natG12D/+ versus KrasLSL-natQ61R/+ cohorts (Figure 4—figure supplement 4 and Supplementary file 3). Of note, median survival was significantly decreased in mice with when Kras was encoded with common compared to native rare codons. Survival was also decreased in the mice with the Q61R compared to the G12D mutation when the allele contained the same codon usage (KrasLSL-natQ61R versus KrasLSL-natG12D and KrasLSL-comQ61R versus Kras LSL-comG12D). As both common codons and the Q61R mutation increase Kras activity, we suggest that the level of active oncoprotein is the dominant determinant of cancer survival in this model.

Figure 4 with 7 supplements see all
Tissue atlas of sensitivities to each oncogenic Kras allele.

(A) Kaplan–Meier survival curve of the mice with the indicated KrasLSL alleles after activation by tamoxifen. Dotted lines: 50% survival. (B) Number of mice with the indicated number of different tumor types. Examples of H&E-stained slides of the indicated tissues are provided in Figure 4—figure supplement 5. (C–F) Percentage of the indicated grades of hematolymphopoietic (C), forestomach (D), oral (E), and lung (F) lesions at moribundity endpoint in Rosa26CreERT2/+ mice (n = 8–10) with one of the four indicated KrasLSL alleles after activation by tamoxifen.

To identify the tissues permissive to tumorigenesis by each of these different alleles, eight different organs were removed from the mice during necropsy and analyzed for pathological changes as above. While there was much overlap, the prevalence and severity of specific cancer types varied between alleles, arguing that the different alleles can lead to differences in the tumor landscape (Figure 4B, Figure 4—figure supplement 5). This manifested in four general patterns of tissue permissivity: hematolymphopoietic neoplasias increased in severity with the level of active oncoprotein, squamous tumors were preferentially induced by the G12D-mutant alleles, carcinomas were induced by all four alleles, while many organs were resistant to oncogenic RAS-driven tumorigenesis within the time frame of the study (Figure 4B, Figure 4—figure supplement 5, and Supplementary file 4).

A tissue sensitive to Kras activity

The incidence of hematolymphopoietic neoplasias increased with the biochemical activity of the Kras oncoprotein, as determined from analysis of MEF cultures and lung tissue. In more detail, the spleen and thymus from eight to ten mice from each of the four cohorts were removed at necropsy, after which H&E-stained sections were assayed for the presence and grade of hematolymphopoietic neoplasias as above. Beginning with the least active oncoprotein, most of the KrasLSL-natG12D mice after tamoxifen injection had no evidence of hematolymphopoietic neoplasms, although some mice had pathological features consistent with malignant lymphoma. The incidence of malignant lymphoma increased with KrasLSL-natQ61R allele, and then again upon activating the KrasLSL-comG12D allele. Pathological analysis also revealed medullary hyperplasia in the thymus, as well as leukemic infiltrates in the kidneys and pancreas of these latter mice, suggesting further progression of the lymphomas with this allele. Lastly, activating the KrasLSL-comQ61R allele induced severe myeloproliferative disease (MPD) with 100% penetrance, with extensive myeloproliferative infiltrates throughout many tissues (Figure 4B, C, Figure 4—figure supplement 5, Supplementary file 4 and 5, and not shown). While we cannot discount high oncogenic activity leading to a different hematopoietic disease in these later mice, we suggest that the incredibly short latency of the onset of severe systemic myeloid neoplasia may instead preclude development of longer latency tumors, such as lymphopoietic neoplasms. Thus, with this proviso, we suggest that hematolymphopoietic neoplasia are sensitive to the level of oncogenic activity, being induced at the lowest level of active Kras and progressively becoming more aggressive with increased activity.

Tissues sensitive to Kras mutation type

Proliferative lesions of forestomach and oral squamous epithelium were preferentially induced by the oncogenic G12D mutant of Kras encoded by either native or common codons. In the case of the forestomach tumors, pathological analysis performed as above revealed that activation of the KrasLSL-natG12D allele induced squamous hyperplasia as well as mild and moderate grades of ‘atypical’ or dysplastic squamous lesions in the forestomach mucosa. Conversely, we did not detect any squamous proliferative changes upon activating the KrasLSL-natQ61R allele (Figure 4B and D, Figure 4—figure supplement 5, and Supplementary files 4 and 5), despite the previously detected higher activity of the KrasnatQ61R oncoprotein in MEFs and lung tissue. The same trend was observed when Kras was encoded with common codons, except the difference was less extreme between the two mutants, and shifted towards more aggressive disease. Specifically, activating the KrasLSL-comG12D allele induced more severe grades of forestomach squamous lesions while activation of the KrasLSL-comQ61R allele now induced lesions that were of moderate grades (Figure 4B and D, Figure 4—figure supplement 5, and Supplementary file 5). Similar analysis of oral tumors revealed that activating the KrasLSL-natG12D allele induced minimal to mild grade squamous lesions, while activating the KrasLSL-natQ61R allele was not tumorigenic. Again, when Kras was encoded with common codons there was a shift to a more aggressive disease. Namely, activating the KrasLSL-comG12D allele induced severe squamous papilloma in all mice, while activating the KrasLSL-comQ61R allele induced a mixture of moderate and severe grade squamous papillomas (Figure 4B and E, Figure 4—figure supplement 5, and Supplementary file 5). As such, in these two organs, the G12D mutant is associated with more severe phenotypes than the Q61R mutant, and changing codon usage to increase expression shifts this difference to a more advanced stage.

Tissues sensitive to both Kras oncogenic activity and mutation type

Pathological analysis performed as above in the lung revealed that the G12D mutation consistently induced more AAH and/or adenomas than the Q61R mutation when Kras was encoded with either common or native rare codons, both in terms of the number of animals with these lesions and the total number of these lesions per animal (Figure 4B and F, Figure 4—figure supplements 5; 6A, and Supplementary file 5). As was the case with forestomach and oral lesions, converting rare codons to common amplified the severity of lesions detected, which was particularly evident in the confluence of large peripheral AAH lesions induced preferentially by theKrasLSL-comG12D allele (Figure 4B and F, Figure 4—figure supplements 5; 6B). However, no BH lesions were induced by either mutant when Kras was encoded with native rare codons and instead were only prevalent when Kras was encoded with common codons. Further, the number of animals with BH lesions was higher upon activating the KrasLSL-comQ61R compared to the KrasLSL-comG12D allele (Figure 4—figure supplements 5; 6C). Assuming that AAH and BH lesions represent different types of tumors (and not different stages of the same tumor type), we suggest that tumorigenesis in the lung is influenced by both the degree of Kras activation and the mutation type, although temporal analysis argues (Figure 2) that the level of Kras activation nevertheless dominates tumorigenesis in this tissue.

Tissues resistant to oncogenic Kras

Despite widespread tumorigenesis, we note that no overt proliferative lesions were detected at necropsy or by histopathological analysis in the pancreas, kidney, or liver (Figure 4—figure supplement 5 and Supplementary file 4). A gross survey of other organs such as the colon, intestine, heart, skin, and mammary glands similarly failed to reveal macroscopically detectable tumors (not shown). In agreement, many of these same tissues, including pancreatic, were reported to be refractory to tumorigenesis upon activating a KrasLSL-G12D allele in the adult mice by CreER expressed from the Rosa26 (Parikh et al., 2012; van der Weyden et al., 2011), CK19 (Ray et al., 2011), or Ubc9 (Matkar et al., 2011) loci. Thus, many organs appear to be intrinsically resistant to the tumorigenic effects of oncogenic Kras, regardless of the mutation type or expression levels tested, at least within the time frame of this study and in the tested Rosa26CreERT2/+ background.

Discussion

Here, we describe the effect of activating four inducible oncogenic Kras alleles encoding two very different mutants in a native rare versus common codon background to explore the mechanisms underlying the bias of specific oncogenic KRAS mutations towards distinct cancer types. We suggest that the unique tumor patterns arising from each of these mutant alleles implies that tissues differ in their sensitivities to quantitative and/or qualitative RAS signaling, both as a product of oncoprotein signaling and the cellular response thereof. We acknowledge four caveats to this approach. First, these four alleles were generated by fusing the first three coding exons, an artificial gene architecture, and hence can only be compared to themselves and not to other types of Kras alleles. Second, these alleles were induced by an injection of tamoxifen to activate CreER expressed from the Rosa26 locus. Admittedly, this is an unnatural situation whereby oncogenic Kras is expressed all at once in a multitude of tissues, potentially perturbing homeostasis in the whole animal. Nevertheless, as the identical design was applied to all four alleles, comparisons can be made within this allelic set. Third, Kras activity of this allelic set was defined in MEFs and lung tissue. We therefore acknowledge that cell-type differences in regulatory feedback pathways (Lake et al., 2016; Liu et al., 2018) or codon-dependent expression (Peterson et al., 2020) could result in different levels of Kras expression, activation, or signaling compared to that observed in MEF cultures and lung tissue, which were used to define the activity of each Kras oncoprotein. Fourth, by the nature of the experimental design, tissue types with slower tumor progression are underrepresented, and hence may respond differently to activation of the different Kras alleles when assessed more specifically.

With these limitations in mind, transcriptome analysis shortly after activating these alleles revealed similarities between the four, but also distinct transcriptional changes. This began with activation of the KrasLSL-natG12D allele, which induced EMT and differentiation hallmarks. EMT is known to promote stem cell-like fates (Floor et al., 2011; Mani et al., 2008) and activation of oncogenic Kras in the murine lung generates tumors with many tissue lineages (Tata et al., 2018). Taken together, we suggest that the transcriptional signature induced by the KrasnatG12D oncoprotein reflects either reprogramming towards or an expansion of cells with multipotent characteristics. In support, single-cell transcriptome profiling of lung tumors induced by targeted delivery of AAV-Cre in an KrasLSL-G12D/+;Trp53fl/fl background identified a distinct population with a mixed cellular identity (Marjanovic et al., 2020). At the other end of the spectrum, activating the KrasLSL-comQ61R allele induced transcriptional hallmarks consistent with overt oncogenic signaling. Thus, as the level of Kras biochemical activity increased, so did the transcriptional signatures of RAS signaling, which at its crescendo resulted in transcriptional signatures indicative of hyperproliferation and oncogenic stress. Nevertheless, within this overarching pattern of increased signaling we also find evidence for both mutation- and codon (expression)-specific transcriptional responses.

Globally activating these four alleles revealed four tumorigenic patterns. First, hematolymphopoietic neoplasias appeared to be largely driven by the amount of active Kras, as defined by RBD pull-down in both MEFs and lung tissue, as well as qRT-PCR of Kras target genes, transcriptome analysis, and RPPA analysis in lung tissue. Namely, these neoplasias were induced at the lowest level of oncogenic activity and increased in aggressiveness with increased Kras activity. To independently validate this result, we found that globally activating a ‘super-rare’ version of KrasLSL-rareG12D allele encoded by the rarest codons induced hematolymphopoietic neoplasias, although in a much-protracted time frame (Figure 4—figure supplement 7 and Supplementary file 4). This further supports hematolymphopoietic tissues as being more permissive to oncogenic Ras activity, suggestive of a dependency on quantitative signaling. Second, proliferative lesions of forestomach and oral squamous epithelium were preferentially induced by G12D-mutant alleles. Such a finding points towards qualitative signaling differences potentially driving these tumors, with higher expression shifting the effects to more aggressive grades. Equally plausible, however, perhaps either a RasGEF or RasGAP is uniquely expressed in these tissues that preferentially interacts with or affects signaling of only one of these mutants, implying quantitative signaling unique to the G12D mutant in these specific tissues. In support, GEF-mediated GTP exchange is more rapid with a Q61L versus G12V mutant of Kras (Smith et al., 2013), the RasGEF SOS1 is inactive toward KRASG12R (Hobbs et al., 2020), and KRASQ61H has diminished sensitivity to SHP2 inhibitors when compared to G12 or G13 mutants (Gebregiworgis et al., 2021). Third, the lung was sensitive to both the activation level and mutation type, with the G12D mutants favoring AAH and adenomas while high Ras activity favored BH lesions, perhaps reflecting a different cell-of-origin for these different lesions (Sutherland et al., 2014; Xu et al., 2014). AAH lesions may also speak to mutation-specific signaling, as here is a case in which we document Q61R mutants are more active, yet G12D mutants are more tumorigenic. With the caveat that Kras-GTP levels were determined in the entire lung, which may not reflect the actual levels in the cell-of-origin for AAH lesion, this implies qualitative differences may underlie these specific lesions. Fourth, many organs failed to develop lesions. We acknowledge that the short life span of some of the tested mice may very well prevent longer latency tumors from developing, and even though Rosa26-restricted Cre expression activated the four inducible oncogenic Kras alleles in tissues that did not form tumors, Cre may not be expressed in the tumor cell-of-origin within these tissues. With these caveats mind, these data suggest that many tissue are intrinsically resistant to the tumorigenic effects of oncogenic Kras. Instead, the hematopoietic system, lungs, forestomach, and oral mucosa are unique in being permissive to the tumorigenic potential of Kras oncoproteins, again, however, within the confines of the experimental design. Interestingly, these tissue sensitivities share some similarity to that of humans, namely, both species have RAS-associated cancers in the lung, mouth, and hematopoietic system but not in the mammary gland, skin, central nervous system, and so forth, but there is also some discordance (Prior et al., 2020).

How these different alleles drive the observed tumor patterns remains to be determined. However, comparing the two extreme cases – activation of the KrasLSL-natG12D versus the KrasLSL-comQ61R alleles in the lung – reveals transcription factors linked to the unique transcriptional responses of these two alleles. Specifically, we cross-referenced the transcriptome with a curated eukaryotic transcription factor database (Matys et al., 2003; Matys et al., 2006), identifying a set of upregulated transcription factors that tracked with the GSEA hallmarks of each oncoprotein (Figure 5—figure supplement 1A, B and Supplementary file 6). Namely, transcription factors tracking with GSEA hallmarks EMT, Estrogen Response Early, KRAS Signaling DN, and Apical Junction in the case of activating the KrasLSL-natG12D allele in the lung, and GSEA hallmarks TNFα Signaling via NFκβ, Complement, P53 Pathway, KRAS Signaling UP, and Inflammatory Response in the case of activating the KrasLSL-comQ61R allele, again in the lung (Figure 5—figure supplement 1A, B). Further, our finding that oncogenic RAS mutations are not identical in terms of activity (GTP-loading), oncogenic potential (induction of lung tumors), and cellular response (transcriptome and RPPA analysis) suggests the intriguing possibility that different initiating RAS mutations may have different therapeutic sensitivities. Indeed, again censoring the GSEA hallmarks in these two extreme cases for pharmacological targets already drugged in the clinic identified FLT4/PDGFRB/KIT, EGFR, and ABL1/SRC as specific to an activated KrasLSL-natG12D allele, while AKT1/2, PI3KCA/CD/CG, SYK, RAF1, JAK2, TGFBR1, and CDK1/2/9 were specific to an activated KrasLSL-comQ61R allele (Figure 5).

Figure 5 with 1 supplement see all
Transcriptome analysis predicts unique pharmacological vulnerabilities.

Druggable kinases positively enriched in GSEA hallmarks identified in Figure 3D (blue KrasLSL-natG12D/+, red KrasLSL-comQ61R/+). * Adjusted p-value<5%.

In summary, we suggest that tissues differ in their sensitivities to quantitative and/or qualitative RAS signaling both as a product of oncoprotein signaling and the cellular response thereof. We find that the level of oncogenic activity favors hematolymphopoietic neoplasias, the G12D mutant uniquely gives rise to oral and forestomach squamous tumors, while lung adenocarcinomas are sensitive to both mutation type and expression levels. The unique signaling dependencies of these tissues may, in turn, be capitalized upon to identify new therapeutic opportunities to target early tumorigenesis, when the tumors are particularly vulnerable, perhaps either as an early intervention or as a preventative measure in high-risk populations.

Materials and methods

Key resources table
Reagent type (species) or resourceDesignationSource or referenceIdentifiersAdditional information
AntibodyAnti-FLAG (mouse monoclonal)SigmaF1804WB (1:1000)
AntibodyAnti-KRAS (mouse monoclonal)Santa Cruz Biotechnologysc-30WB (1:500)
AntibodyAnti-βTubulin (mouse monoclonal)SigmaT5201WB (1:10,000)
AntibodyAnti-SV40 large T antigen (rabbit monoclonal)Cell Signaling15729WB (1:1000)
Strain, strain background (Escherichia coli)STBL3Thermo Fisher ScientificC737303Chemically competent cells
Commercial assay or kitFuGENE 6 Transfection ReagentPromegaE2691
Commercial assay or kitDC Protein AssayBio-Rad5000112
Commercial assay or kitDNeasy Blood and Tissue DNA extraction KitQIAGEN69504
Chemical compound, drugTamoxifenSigma-AldrichT5648-5G
Commercial assay or kitRNAeasy KitQIAGEN74104
Commercial assay or kitRNase-Free DNase SetQIAGEN79254
Commercial assay or kitGenPoint kitAgilentK062011-2
Commercial assay or kitMycoAlert PLUS Mycoplasma detection kitLonzaLT07-703
AntibodyAnti-rabbit IgG antibodyVector LaboratoriesBA-1000-1.5RPPA
AntibodyAnti-mouse IgG antibodyVector LaboratoriesBA-9200-1.5RPPA
Commercial assay or kitActive Ras Detection KitCell Signaling8821
Commercial assay or kitRas GTPase ELISA KitAbcamab134640
Cell line (Homo sapiens)HEK-HTCounter et al., 1992Cell line created and maintained in C. Counter lab, validated by immunoblot for SV40 Large T antigenCultured in DMEM and 10% FBS

Tested negative for mycoplasma
Cell line (Mus musculus)MEF derived from KrasLSL-natG12D/+ miceThis paperValidated by genotyping PCRCultured in DMEM and 10% FBS

Tested negative for mycoplasma
Cell line (M. musculus)MEF derived from KrasLSL-natQ61R/+ miceThis paperValidated by genotyping PCRCultured in DMEM and 10% FBS

Tested negative for mycoplasma
Cell line (M. musculus)MEF derived from KrasLSL-comG12D/+ miceThis paperValidated by genotyping PCRCultured in DMEM and 10% FBS

Tested negative for mycoplasma
Cell line (M. musculus)MEF derived from KrasLSL-comQ61R/+ miceThis paperValidated by genotyping PCRCultured in DMEM and 10% FBS

Tested negative for mycoplasma
Strain, strain background (M. musculus)KrasLSL-natG12D/+This paperGenerated in Counter Lab
Strain, strain background (M. musculus)KrasLSL-natQ61R/+This paperGenerated in Counter
Strain, strain background (M. musculus)KrasLSL-comG12D/+This paperGenerated in Counter Lab
Strain, strain background (M. musculus)KrasLSL-comQ61R/+This paperGenerated in Counter Lab
Strain, strain background (M. musculus)KrasLSL-rareG12D/+This paperGenerated in Counter Lab
Strain, strain background (M. musculus)ACTBFLPe/FLPeJackson Laboratory003800
Strain, strain background (M. musculus)CC10CreER/CreER; Rosa26CAG-fGFP/CAG-fGFPXu et al., 2012
Strain, strain background (M. musculus)Rosa26CreERT2/CreERT2Jackson Laboratory008463
Recombinant DNA reagentpcDNA3.1Thermo Fisher ScientificV79020Mammalian expression vector backbone
Recombinant DNA reagentpcDNA3.1+FLAG-KrasnatG12DThis studyGenerated in Counter Lab
Recombinant DNA reagentpcDNA3.1+FLAG-KrasnatQ61RThis studyGenerated in Counter Lab
Recombinant DNA reagentpcDNA3.1+FLAG-KrascomG12DThis studyGenerated in Counter Lab
Recombinant DNA reagentpcDNA3.1+FLAG-KrascomQ61RThis studyGenerated in Counter Lab
Recombinant DNA reagentpBABE-neo largeTcDNAHahn et al., 2002Addgene #1780
Recombinant DNA reagentMSCV-Cre-HygroWang et al., 2010Addgene #34565
Software, algorithmImageJ version 1.52k with Java 1.8.0_172Schneider et al., 2012https://imagej.nih.gov/ij/
Software, algorithmImage LabBio-Radhttps://www.bio-rad.com/en-us/product/image-lab-software
Software, algorithmGraphPad Prism v8GraphPadhttps://www.graphpad.com/
Software, algorithmbcl2fastqIlluminahttps://support.illumina.com/sequencing/sequencing_software/bcl2fastq-conversion-software.html
Software, algorithmSTAR RNA-Seq alignment tool v2.7.8aDobin et al., 2013https://github.com/alexdobin/STARSTAR (RRID:SCR_004463)
Software, algorithmDESeq2Love et al., 2014http://www.bioconductor.org/packages/release/bioc/html/DESeq2.html
Software, algorithmGSEAMootha et al., 2003https://www.gsea-msigdb.org/gsea/
Software, algorithmR StudioR Development Core Team, 2020https://www.R-project.org
Software, algorithmMicroVigene Software Version 5.1.0.0VigeneTechhttp://www.vigenetech.com/Protein.htm
Software, algorithmBioRenderhttp://www.biorender.com

Generation of KrasLSL-natG12D, KrasLSL-natQ61R, KrasLSL-comG12D, and KrasLSL-comQ61R alleles

Request a detailed protocol

A bacteria artificial chromosome was engineered with 7.5 kbp of 5′ flanking sequence Kras intron 1 DNA, a Lox-STOP-Lox cassette (LSL) (Feil et al., 1996), the first three coding exons fused together and encoded by either native (nat) codons or with 93 rare codons converted to the most commonly used codons in the mouse genome (com) and either a G12D or Q61 oncogenic mutation, followed by the N-terminal 564 bp of intron 4, an FRT-Neomycin-FRT cassette, and a further 1.5 kbp of 3′ flanking sequence (Figure 1A, Figure 1—figure supplement 1). The targeting cDNAs were each cloned into the targeting PL253 vector (Liu et al., 2003) and electroporated into 129S6/C57BL/6N (G4) ES cells. After selection, clones were screened using PCR and positive clones were selected for each engineered Kras allele, expanded, and frozen. At least two targeted clones per allele confirmed with a Southern hybridization were then microinjected into blastocysts to produce chimeras using standard procedures (Behringer, 2014). KrasLSL-natG12D(+neo)/+, KrasLSL-natQ61R(+neo)/+, KrasLSL-comG12D(+neo)/+, and KrasLSL-comQ61R(+neo)/+, chimeras (still retaining the neo cassette in the engineered Kras alleles) were crossed back to 129S6 mice. A genotyping PCR specific to the engineered Kras alleles with neo cassette was used to screen for germline transmission in clones. Each 129S6-KrasLSL-nat(+neo)/+ and 129S6-KrasLSL-com(+neo)/+ cohort was crossed with ACTBFLPe/FLPe (Jackson Laboratory, strain 003800) mice to remove the selection marker via FLP-mediated excision of the neo cassette (Dymecki, 1996). Removal of the neo cassette was confirmed with genotyping PCR. Resultant strains were backcrossed with 129S6 mice for five generations, generating the KrasLSL-natG12D/+, KrasLSL-natQ61R/+, KrasLSL-comG12D/+, and KrasLSL-comQ61R/+ strains used in this study. All mouse care and experiments were performed in accordance with a protocol approved by the Institutional Animal Care and Use Committee (IACUC) of Duke University (protocol no. A195-19-09).

Codon usage plots

Request a detailed protocol

The codon usage index (Sharp and Li, 1987) was calculated using the relative codon frequency derived from codon usage in the mouse exome (Nakamura et al., 2000) with a sliding windows of 25 codons across the open reading frame (ORF) of each transcript. A theoretical murine Kras ORF encoded by the rarest codons at each position (gray dotted line) was plotted for reference (Figure 1—figure supplement 1).

Genotyping

Request a detailed protocol

Genomic DNA was isolated from 1 to 2 mm piece of toes by boiling for 30 min in 100 μl Toe Lysis Buffer (25 mM NaOH and 0.2 mM EDTA), followed by neutralization with 1.5 volume of Neutralization Buffer (40 mM Tris–HCl pH 6.0). 1.2 μl of the isolated genomic DNA was subjected to PCR. Genotyping cell lines and mouse tissue was performed as above with 20 ng of genomic DNA isolated with a QIAGEN DNeasy Blood and Tissue DNA Extraction Kit (QIAGEN, #69504). All genotyping PCR reactions were performed using 0.4 U (0.08 μl) Platinum Taq polymerase (Invitrogen, #10342046) in 12.5 μl reaction volume with final concentration of 1.5–2 mM MgCl2, 0.2 mM of each dNTP, and 0.5 μM of each primer. Full-length gels and replicates are provided (Figure 1—source data 2, Figure 4—source data 1). Mice were genotyped using the following primers (also provided in Appendix 1):

KrasLSL(+neo) alleles:

  • Kras.in3.F: 5'-TTGGTGTACATCACTAGGCTTCA-3'

  • Kras.in3.R: 5'-TGGAAAGAGTAAAGTGTGGTGGT-3'

  • Kras.neo.F: 5'-GTGGGCTCTATGGCTTCTGA-3'

  • Products: 590 bp (Targeted Allele[+neo]) or 240 bp (WT allele)

KrasLSL-com alleles:

  • KrasCOM5.F: 5'-CTTCCATTTGTCACGTCCTGC-3'

  • KrasCOM5.R: 5'-TCTTCGGTGGAAACAACGGT-3'

  • Product: 448 bp (KrasLSL-com)

KrasLSL-nat alleles:

  • F-LSL: 5'-TAGTCTGTGGGACCCCTTTG-3'

  • R-LSL: 5'-GCCTGAAGAACGAGATCAGC-3'

  • Product: 448 bp (KrasLSL-nat)

Recombination PCR:

  • KRASOP.A2: 5'-CTAGCCACCATGGCTTGAGT-3'

  • KRASOP.B: 5'-GTAATCACAACAAAGAGAATGCAG-3'

  • LSL-F: 5'-GGGGAACCTTTCAGGCTTA-3'

  • Products: 616 bp (LoxP Recombined), 488 bp (WT allele), or 389 bp (KrasLSL)

CC10CreER alleles:

  • F-CC10 WT: 5'-ACTCACTATTGGGGGTGTGG-3'

  • R-CC10 WT: 5'-GGAGGACTTGTGGATCTTG-3'

  • F-Cre: 5'-TCGATGCAACGAGTGATGAG-3'

  • R-Cre: 5'-TTCGGCTATAGGTAACAGGG-3'

  • Products: 450 bp (CC10CreER allele) or 350 bp (WT allele)

Rosa26CAG-fGFP alleles:

  • F-Rosa-01: 5'-CACTTGCTCTTCCAAAGTCG-3'

  • R-Rosa-02B: 5'-TAGTCTAACTCGCGACACTG-3'

  • F-CAG-02B: 5'-GTTATGTAACGCGGAACTCC-3'

  • Products: 500 bp (WT allele allele) or 350 bp (Rosa26fGFP allele)

Rosa26CreERT2 alleles:

  • R26R-univF: 5'-AAAGTCGCTCTGAGTTGTTAT-3'

  • R26R-wtR: 5'-GGAGCGGGAGAAATGGATATG-3'

  • CreER-R1: 5'-CCTGATCCTGGCAATTTCG-3'

  • Products: 800 bp (Rosa26CreERT2 allele) or 600 bp (WT allele)

Kras LSL element:

  • KRASOP.A2: 5'-CTAGCCACCATGGCTTGAGT-3'

  • KRASOP.B: 5'-GTAATCACAACAAAGAGAATGCAG-3'

  • Product: 389 bp (KrasLSL alleles)

CreER validation:

  • F-Cre: 5'-GGAGGACTTGTGGATCTTG-3'

  • CREER-R1: 5'-CCTGATCCTGGCAATTTCG-3'

  • Product: 500 bp (CreER)

KrasLSL-rare alleles:

  • KrasRAR.F: 5'-TATGCGTACGGGTGAAGGTT-3'

  • KrasRAR.R: 5'-GCAGAGCACAGACTCACGTC-3'

  • Product: 275 bp (KrasLSL-rare)

Plasmids

N-terminal FLAG-tagged murine Kras cDNAs encoding KrasnatG12D, KrasnatQ61R, KrascomG12D, and KrascomQ61R were designed using the same sequence of the first three coding exons as described in the engineered alleles with the addition of the native Kras4B to the C-terminus and cloned into pcDNA3.1+ (Thermo Fisher). Plasmid sequences were verified by sequencing.

Ectopic expression, immunoblots, and Ras activity assay

Request a detailed protocol

To validate the expression levels of the FLAG-tagged murine Kras cDNA constructs, 2 × 106 HEK-HT cells (Counter et al., 1992) were seeded in 10 cm tissue culture plates in DMEM with high glucose (Sigma-Aldrich, D5796) supplemented with 10% FBS (VWR, 97068-085), and transiently transfected the next day with the pcDNA3.1+ (empty vector) or the same plasmid encoding FLAG-KrasnatG12D, FLAG-KrasnatQ61R, FLAG-KrascomG12D, or FLAG-KrascomQ61R using FuGene 6 reagent (Promega, E2691) according to the manufacturer’s protocol. 48 hrs later, transfected cells were washed with cold PBS and pelleted. Cell pellets were lysed in 5 volumes of 1× lysis buffer (50 mM Tris–HCl pH 8.0, 150 mM NaCl, 1% NP40, 0.5% sodium deoxycholate, 0.1% SDS, and 5 mM EDTA) containing Protease Inhibitor Cocktail (Roche, 11836170001). The DC Protein Assay (Bio-Rad, 5000112) was used to measure protein concentration and 30 μg protein of each sample was resolved by SDS-PAGE, transferred to a PVDF membrane (Bio-Rad, 1704273), blocked in 5% milk, and immunoblotted by the following antibodies in 5% BSA (Sigma, A7906-500G): FLAG (Sigma, F1804; diluted 1:1000), KRAS (Santa Cruz, SC-30; diluted 1:500), and ß-Tubulin (Sigma, T5201; diluted 1:10,000). Primary antibody incubation was performed at room temperature for 1 hr followed by the secondary antibody incubation at 4°C overnight. To measure biochemical activity of the FLAG-tagged Kras constructs, cells were seeded, transfected, washed with PBS as above, after which cell pellets were lysed and immediately assayed with the Active Ras Detection Kit (Cell Signaling, #8821) or Ras GTPase ELISA Kit (Abcam, ab134640) according to the manufacturer’s protocol. Measurements were done with two technical replicates of four serial dilutions and data shown are representative of two independent biological replicates. HEK-HT cells were confirmed to be mycoplasma-negative. Full-length images of immunoblots and replicates are provided (Figure 1—source data 1).

Mouse embryonic fibroblasts (MEFs)

Request a detailed protocol

KrasLSL-natG12D/+, KrasLSL-natQ61R/+, KrasLSL-comG12D/+, and KrasLSL-comQ61R/+ mice were crossed with 129S6 mice for timed pregnancies to isolate embryos 12.5 to 13.5 days postcoitum, as previously described (Serrano et al., 1997). Pairs of primary MEF cultures derived from two separate embryos from each cohort were cultured in DMEM with high glucose (Sigma-Aldrich, D5796) supplemented with 10% FBS (VWR, 97068-085) and then immortalized by infection with an amphotropic retrovirus derived from plasmid pBABE-neo largeTcDNA (a gift from Robert Weinberg, Addgene plasmid #1780) (Hahn et al., 2002) using standard methodologies (O’Hayer and Counter, 2006). Paired parental MEF lines that were not infected with the pBABE-neo largeTcDNA were used as a negative control for antibiotic selection. Stable immortalized MEF cultures were selected in 400 μg/ml Geneticin (Thermo Fisher, 10131035) and again infected with an amphotrophc retrovirus derived from plasmid MSCV-Cre-Hygro (a gift from Kai Ge, Addgene plasmid #34565) to activate the mutant Kras allele (Wang et al., 2010). The stable Cre-expressing MEF cultures were selected in 500 μg/ml Hygromycin (Thermo Fisher, 10687010). Paired immortalized KrasLSL-natG12D/+ MEF lines that were not infected with MSCV-Cre-Hygro were used as negative control for Hygromycin selection. Genomic DNA was isolated from the resultant cells using QIAGEN DNeasy Blood and Tissue DNA Extraction Kit (QIAGEN, #69504) and subjected to PCR to detect recombination of the KrasLSL alleles as described above. Paired immortalized MEF lines for each clone that were not infected with the MSCV-Cre-Hygro were used as negative control for recombination PCR. MEF cultures were confirmed to be mycoplasma-negative. Full-length gel image is provided (Figure 1—source data 2).

Immunoblot and Ras activity assays in MEF cultures

Request a detailed protocol

The aforementioned pairs of independently derived KrasLSL-natG12D/+, KrasLSL-natQ61R/+, KrasLSL-comG12D/+, and KrasLSL-comQ61R/+ MEF cultures stably expressing Cre were grown in DMEM with high glucose (Sigma-Aldrich, D5796) supplemented with 10% FBS (VWR, 97068-085) in 15 cm dishes to prepare lysates for immunoblots and Ras activity assays. We note here that immortalized MEF cultures derived from KrasLSL-comQ61R/+ mice exhibited decreased recombination rate compared to MEFs derived from the other three genotypes, potentially due to resistance to hygromycin. Given this, we generated single colonies from this line after hygromycin selection for all experiments investigating Kras expression and activity. Two independent cell lines from immortalized KrasLSL-natG12D/+ MEF cultures without Cre were used as negative control to Cre-induced Kras expression. Once cells reached 85 to 90% confluency, FBS supplemented media was removed and replaced with plain DMEM with high glucose (Sigma-Aldrich, D5796) without FBS overnight. For serum stimulated lysates, cells were serum stimulated for 5 min by adding FBS to a final concentration of 10%. After serum starvation and stimulation, MEF cells were washed with cold PBS and pelleted. Cell pellets were lysed in 1 ml of lysis buffer provided in the Active Ras Detection Kit (Cell Signaling, #8821) supplemented with Halt Protease Inhibitor Cocktail (Thermo Fisher, 78430). The DC Protein Assay (Bio-Rad, 5000112) was used to measure protein concentration and 20 μg protein of each sample was used to immunoblot for total cell lysates. To measure biochemical activity of the endogenous Kras, 500 μg or 200 μg of each cell lysate was immediately assayed with the Active Ras Detection Kit (Cell Signaling, #8821) according to the manufacturer’s protocol, except that incubation of lysates with RBD peptide was performed at 4°C overnight. Total cell lysates and lysates after RBD pull-down were resolved by 18% SDS-PAGE, transferred to a PVDF membrane (Bio-Rad, 1704273), blocked in 5% milk, and immunoblotted by the following antibodies in 5% BSA (Sigma, A7906-500G): KRAS (Santa Cruz, SC-30; diluted 1:500) and ß-Tubulin (Sigma, T5201; diluted 1:10,000). Primary antibody incubation was performed at 4°C overnight followed by the secondary antibody incubation for 3 hrs at room temperature. Immunoblots shown are representative of two independent MEF lines. Full-length images of immunoblots and replicates are provided (Figure 1—source data 3).

Tumorigenesis studies

Request a detailed protocol

KrasLSL-natG12D/+, KrasLSL-natQ61R/+, KrasLSL-comG12D/+, and KrasLSL-comQ61R/+ mice were crossed with either CC10CreER/CreER;Rosa26CAG-fGFP/CAG-fGFP (a gift from Mark Onaitis) (Xu et al., 2012) or Rosa26CreERT2/CreERT2 (Jackson Laboratory, strain 008463) mice and resultant offspring with the desired alleles were selected by genotyping. At six to eight weeks of age, selected littermates with random distribution of males and females received four intraperitoneal injections of tamoxifen (Sigma-Aldrich, T5648-5G, CAS# 10540-29-1) dissolved in corn oil (Sigma-Aldrich, C8267) and filter sterilized for a dose of 250 μg/g body weight to induce Cre-mediated recombination of the LSL cassette, activating the engineered oncogenic mutant Kras alleles. Mice from the first cross were humanely euthanized 1, 2, 3, 4, 5, or 6 months later while mice from the second cross were euthanized upon moribundity. To prevent age-related outcomes, four mice that did not reach moribundity by 300 days were euthanized at one year of age, which we reasoned to suffice to show tumorigenesis as it is about eight times longer than the previously identified average moribundity of five to six weeks with the tamoxifen-treated Rosa26CreERT2/+;KrasLSL-G12D/+ mice (Parikh et al., 2012). Selected tissues were removed at necropsy and fixed in 10% formalin (VWR, 89370-094) for 24 to 48 hrs, then post-fixed in 70% ethanol (VWR, 89125-166) until analysis.

Tissue analysis

Request a detailed protocol

Animals were euthanized with inhaled carbon dioxide and subjected to a complete necropsy. Selected organs were sampled for microscopic examination, including lung, liver, kidney, spleen, thymus, stomach, pancreas, and macroscopic lesions. All tissues were fixed for 48 hrs in 10% neutral buffered formalin (VWR, 89370-094) and then post-fixed in 70% ethanol (VWR, 89125-166), processed routinely, embedded in paraffin with the flat sides down, sectioned at a depth of 5 μm, and stained by the H&E method. Routine processing of the lungs from CC10CreER/+;Rosa26CAG-fGFP/+ mice was performed by the Duke Research Immunohistology Lab, while all tissues from Rosa26CreERT2/+ mice were processed by IDEXX Laboratories. Tissues and H&E slides were evaluated by a board-certified veterinary anatomic pathologist with experience in murine pathology (Supplementary file 4 and 5).

Tissue recombination analysis

Request a detailed protocol

At six to eight weeks of age, two female mice from each of the genotypes KrasLSL-natG12D/+, KrasLSL-natQ61R/+, KrasLSL-comG12D/+, and KrasLSL-comQ61R/+ in a Rosa26CreERT2/+ background were injected with tamoxifen as above, and seven days later euthanized. One age-matched female Rosa26CreERT2/+;KrasLSL-comG12D/+ mouse was alsoeuthanized as a no-tamoxifen control. At necropsy, the colon, duodenum, ileum, cecum, jejunum, pancreas, spleen, glandular stomach, forestomach, kidney, liver, lung, heart, and ovaries were removed, genomic DNA extracted, and the status of recombination of each of the four oncogenic mutant KrasLSL alleles determined by recombination PCR in duplicate, as described above. The intensities of bands corresponding to the unaltered wild-type Kras allele (WT) as well as the unrecombined and recombined oncogenic mutant KrasLSL alleles were quantified with ImageJ software and the recombination rates for each tissue type from each mouse calculated by dividing the densitometry of recombined allele to that of the wild-type allele. Full-length gels and replicates are provided (Figure 4—source data 1).

Statistical analysis

Request a detailed protocol

Statistical analyses were performed using GraphPad Prism software version 8 (GraphPad Software). One-way ANOVA with Bonferroni’s multiple-comparisons test with a single pooled variance and a 95% CI were used for experiments with more than two groups. Reported p-values are adjusted to account for multiple comparisons. A p-value of <0.05 was considered statistically significant.

Ras activity assay on lung tissue

Request a detailed protocol

At six to eight weeks of age, two mice with random distribution of males and females from each of the genotypes Rosa26CreERT2/+;KrasLSL-natG12D/+, Rosa26CreERT2/+;KrasLSL-natQ61R/+, Rosa26CreERT2/+;KrasLSL-comG12D/+, Rosa26CreERT2/+;KrasLSL-comQ61R/+, and Rosa26CreERT2/+ (control) were injected with tamoxifen as above, seven days latereuthanized, and their lungs were removed. Lungs were lysed in 5 ml of lysis buffer provided in the Active Ras Detection Kit (Cell Signaling, #8821) supplemented with Halt Protease Inhibitor Cocktail (Thermo Fisher, 78430). The DC Protein Assay (Bio-Rad, 5000112) was used to measure protein concentration, and 30 μg protein of each sample was used to immunoblot for total cell lysates. To measure biochemical activity of the endogenous Kras, 1 mg of each cell lysate was immediately assayed with the Active Ras Detection Kit (Cell Signaling, #8821) according to the manufacturer’s protocol, except for overnight incubation of the RBD pull-down reaction at 4°C. 30 μg of total cell lysates and 20% of the RBD pull-down mixture were resolved by 8 to 18% gradient SDS-PAGE, transferred to a PVDF membrane (Bio-Rad, 1704273), blocked in 5% milk, and immunoblotted by the following antibodies in 5% BSA (Sigma, A7906-500G): KRAS (Santa Cruz, SC-30; diluted 1:500), and ß-Tubulin (Sigma, T5201; diluted 1:10,000). Primary antibody incubation was performed at 4°C overnight followed by the secondary antibody incubation for 2 hrs at room temperature. Full-length images of immunoblots and replicates are provided (Figure 1—source data 4).

RNA-seq

Request a detailed protocol

For the first experiment, at six to eight weeks of age, three mice with random distribution of males and females from each of the genotypes Rosa26CreERT2/+;KrasLSL-natG12D/+, Rosa26CreERT2/+;KrasLSL-natQ61R/+, Rosa26CreERT2/+;KrasLSL-comG12D/+, Rosa26CreERT2/+;KrasLSL-comQ61R/+, and Rosa26CreERT2/+ (control) were injected with tamoxifen as above, seven days later euthanized, and their lungs were removed. For the second RNA-seq experiment, the experiment was repeated exactly as before except with Rosa26CreERT2/+;KrasLSL-natG12D/+ versus Rosa26CreERT2/+;KrasLSL-comQ61R/+ mice. Tissue lysis and RNA extraction steps were performed in a chemical hood, and all instruments and tools were sprayed with RNaseZAP (Sigma, R2020) to prevent RNA degradation. Isolated lungs were immediately stored in RNA stabilizing solution RNAlater (Sigma, R0901) at 4°C overnight and then transferred to –80°C for long-term storage. Lung tissues were thawed, weighed, and pulverized with mortar and pestle in the presence of liquid nitrogen. RNA extraction was performed immediately thereafter using RNAeasy Kit (QIAGEN, 74104) according to the manufacturer’s instructions. Briefly, tissue lysates were prepared according to the kit instructions and tissue clumps were removed using QIAshredder columns (QIAGEN, 79654). Cleared lysates were applied to RNeasy silica columns and on-column DNase digestion was performed using RNase-Free DNase Set (QIAGEN, 79254) to remove DNA in silica membrane. Following wash steps, RNA was eluted in RNase-free water that was treated with RNAsecure RNase Inactivation Reagent (Thermo Fisher, AM7005) according to the manufacturer’s instructions. Extracted total RNA quality and concentration were assessed on Fragment Analyzer (Agilent Technologies) and Qubit 2.0 (Thermo Fisher Scientific), respectively. Samples with RIN less than 7 were not sequenced. RNA-seq libraries were prepared using the commercially available KAPA Stranded mRNA-Seq Kit (Roche). In brief, mRNA transcripts were captured from 500 ng of total RNA using magnetic oligo-dT bead. The mRNA was then fragmented using heat and magnesium, and reverse transcribed using random priming. During second-strand synthesis, the cDNA:RNA hybrid was converted into to double-stranded cDNA (dscDNA) and dUTP was incorporated into the second cDNA strand, effectively marking this strand. Illumina sequencing adapters were ligated to the dscDNA fragments and amplified to produce the final RNA-seq library. The strand marked with dUTP was not amplified, allowing for strand-specific sequencing. Libraries were indexed using a dual-indexing approach allowing for all the libraries to be pooled and sequenced on the same sequencing run. Before pooling and sequencing, fragment length distribution for each library was first assessed on a Fragment Analyzer (Agilent Technologies). Libraries were also quantified using Qubit. Molarity of each library was calculated based on Qubit concentration and average library size. All libraries were then pooled in equimolar ratio and sequenced. Sequencing was done on an Illumina NovaSeq 6000 sequencer. The pooled libraries were sequenced on one lane of an S-Prime flow cell at 50 bp paired-end. Once generated, sequence data was demultiplexed and Fastq files generated using bcl2fastq v2.20.0.422 file converter from Illumina. The RNA-seq data has been deposited in NCBI’s Gene Expression Omnibus (Edgar et al., 2002). Initial and secondary RNA-seq data are available through GEO Series accession numbers GSE181628 and GSE181627, respectively, and are accessible under the project GSE181629 (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE181629).

Transcriptome analysis

Request a detailed protocol

RNA-seq data was processed using the TrimGalore toolkit (http://www.bioinformatics.babraham.ac.uk/projects/trim_galore), which employs Cutadapt (Martin, 2011) to trim low-quality bases and Illumina sequencing adapters from the 3′ end of the reads. Only reads that were 20 nucleotides or longer after trimming were kept for further analysis. Reads were mapped to the GRCm38v73 version of the mouse genome and transcriptome (Kersey et al., 2012) using the STAR RNA-seq alignment tool (Dobin et al., 2013). Reads were kept for subsequent analysis if they mapped to a single genomic location. Gene counts were compiled using the HTSeq tool (http://www-huber.embl.de/users/anders/HTSeq/). Only genes that had at least ten reads in any given library were used in subsequent analysis. Normalization and differential expression were carried out using the DESeq2 (Love et al., 2014) Bioconductor (Huber et al., 2015) package with the R statistical programming environment (http://www.R-project.org). The false-discovery rate was calculated to control for multiple hypothesis testing (GSE181628). GSEA (Mootha et al., 2003) was performed to identify hallmarks and pathways associated with altered gene expression for each of the comparisons performed (Figure 3B and C, Figure 3—figure supplement 8, Figure 3—source data 2). To identify transcription factors significantly enriched, the differential expression data from KrasLSL-natG12D/+ and KrasLSL-comQ61R/+ transcriptomes was trimmed to genes whose adjusted p-value is less than 5% and whose log2 fold-change either larger than +1 or lower than –1. The trimmed dataset was cross-referenced with TRANSFAC, a curated eukaryotic transcription factor dataset, retrieved from Harmonizome web portal (Matys et al., 2003; Matys et al., 2006; Rouillard et al., 2016). Retrieved gene transcription factor pairs from TRANSFAC were then censored to the transcription factors enriched in KrasLSL-natG12D/+ and KrasLSL-comQ61R/+ specific GSEA hallmarks including more than four such gene transcription factor pairs both with p-value less than 5% (Figure 5—figure supplement 1). To identify low and high Kras-specific druggable kinases, the transcriptome enriched in KrasLSL-natG12D/+ and KrasLSL-comQ61R/+ specific GSEA hallmarks were cross-referenced with kinases whose clinical inhibitors were previously surveyed (Klaeger et al., 2017). The kinases enriched in KrasLSL-natG12D/+ and KrasLSL-comQ61R/+ are shown in blue and red, respectively, with adjusted p-value less than 5% highlighted (Figure 5).

Reverse transcription quantitative real-time PCR (qRT-PCR)

Request a detailed protocol

At six to eight weeks of age, two to three mice with random distribution of males and females from each of the genotypes Rosa26CreERT2/+;KrasLSL-natG12D/+, Rosa26CreERT2/+;KrasLSL-natQ61R/+, Rosa26CreERT2/+;KrasLSL-comG12D/+, Rosa26CreERT2/+;KrasLSL-comQ61R/+, and Rosa26CreERT2/+ (control) were injected with tamoxifen as above, and seven days later euthanized to harvest lungs. RNA extraction was performed as mentioned above, followed by first-strand cDNA synthesis from 2 μg RNA, and real-time quantitative PCR using GoTaq 2-Step RT-qPCR kit (Promega, A6110). All measurements were normalized against Actin as the internal control using the 2-ΔΔCt method (Figure 1—source data 5, Figure 3—source data 3). For qRT-PCR analysis for the second biological replicate, one of two negative controls did not have numerical data for two transcripts and hence was not used for plotting. Data shown are representative of two independent biological replicates, and the primer sequences are provided in Appendix 1.

Reverse-phase protein microarray (RPPA)

Request a detailed protocol

At six to eight weeks of age, three to four mice with random distribution of males and females from each of the genotypes Rosa26CreERT2/+;KrasLSL-natG12D/+, Rosa26CreERT2/+;KrasLSL-comQ61R/+, and Rosa26CreERT2/+ (control) were injected with tamoxifen as above, seven days later euthanized, and their lungs were removed. Lungs were snap froze, and embedded in Optimal Cutting Temperature (O.C.T.) compound (Tissue-Tek, 4583). Five sections of 8 μm were cut, mounted on uncoated glass slides, and stored at −80°C until lysed. Whole-tissue lysates were obtained directly from the glass slides after sections were fixed in 70% ethanol, washed with deionized water, stained with hematoxylin (Sigma-Aldrich, H9627, CAS# 517-28-2) and Scott’s Tap Water (Electron Microscopy Sciences, 2607006), and dehydrated in an ascending series of ethanol (70, 95, and 100%). Solutions were supplemented with cOmplete Mini Protease Inhibitors (Roche Applied Science, 11836153001) to avoid proteins and phosphorylated residues degradation. Stained tissues were lysed in a 1:1 solution of Tissue Protein Extraction Reagent (Pierce, 78510) and 2× Tris-Glycine SDS Sample buffer (Invitrogen Life Technologies, LC2676) supplemented with 2.5% β-mercaptoethanol (Sigma-Aldrich, M3148, CAS# 60-24-2). Tissue lysates were immobilized onto nitrocellulose-coated slides (Grace Bio-labs, 505278) in three technical replicates using an Aushon 2470 arrayer (Quanterix) equipped with 185 μm pins as previously described (Baldelli et al., 2017). Reference standard curves for internal quality control were arrayed along with the tissue lysates along with a bovine serum albumin serial dilution curve to quantify total protein amount on each sample. To assess the amount of protein in each sample and for normalization purposes, selected arrays were stained with Sypro Ruby Protein Blot Stain (Molecular Probes, S12000) following the manufacturer’s instructions as previously described (Baldelli et al., 2021). Reblot Antibody Stripping solution (Chemicon, 2500) was used to treat the remaining arrays for 15 min at room temperature before antibody staining. Arrays were subsequently washed with PBS and incubated for at least 4 hrs in I-block reagent (Tropix, T2015). Using an automated system (Dako Cytomation), arrays were probed with selected antibodies targeting total proteins and post-translationally modified epitopes of kinases and their downstream substrates (Supplementary file 1). For their use on the array, antibodies were validated as previously described (Signore et al., 2017). After incubation with a primary antibody, each array was probed with a biotinylated anti-rabbit (Vector Laboratories, Inc BA-1000-1.5) or anti-mouse (Vector Laboratories, BA-9200-1.5) secondary antibodies matching the species of the primary antibody. Signal detection was performed using the GenPoint kit (Agilent, K062011-2), a commercially available tyramide-based avidin/biotin amplification system, coupled with a fluorescent streptavidin-conjugated IRDye680 dye (LI-COR Biosciences, 680RD) according to the manufacturer’s recommendation. As negative controls for the background signal, selected arrays were probed with the secondary antibody alone. Antibody, negative control, and Sypro Ruby-stained slides were scanned at the appropriate wavelength channel using a laser scanner (TECAN). Image analysis was performed by spot finding, subtraction of local background and unspecific binding generated by the secondary antibody, and normalization to the amount of total protein obtained from the Sypro Ruby-stained slide by MicroVigene Software Version 5.1.0.0 (VigeneTech) as previously described (Pin et al., 2014). Replicate values are averaged to generate a single-intensity RPPA value per sample. Raw data with these average values are shown in Supplementary file 1.

RPPA analysis

Request a detailed protocol

RPPA data for each tissue was first processed by replacing intensity values of 0 with 1 to ease computation. Each value was then converted to log2 scale. To calculate log2 fold-change, we calculated the average log2 values of each target in each genotype and subtracted the average log2 values of the wild-type Kras allele of that target. Heatmaps for the log2 fold-change were plotted using R statistical programming environment (http://www.R-project.org). Note that RPPA analysis of one of the Rosa26CreERT2/+;KrasLSL-natG12D/+ mouse lungs was deemed an outlier as it shared little commonality with the other three biological replicates. As such, the RPPA analysis of this mouse was not included in Figure 1—figure supplement 6 and Figure 3—figure supplement 9.

Generation of KrasLSL-rareG12D/+ mice and tumorigenesis study

Request a detailed protocol

An additional inducible Kras allele with the rarest codons was generated in the same manner as mentioned above from BAC design to chimera production (Figure 1—figure supplement 1). KrasLSL-rareG12D(+neo) chimeras were crossed back to 129S6 mice followed by ACTBFLPe/FLPe (Jackson Laboratory, strain 003800) mice to remove the neo selection marker via FLP-mediated excision (Dymecki, 1996). Both germline transmission and the removal of the neo cassette were confirmed with genotyping PCR as mentioned above. Resultant strain was backcrossed with 129S6 mice for five generations, generating the KrasLSL-rareG12D/+ strain. KrasLSL-rareG12D/+ mice were crossed with Rosa26CreERT2/CreERT2 (Jackson Laboratory, strain 008463) mice and animals with the desired alleles selected by genotyping. At six to eight weeks of age, seven mice with random distribution of males and females were injected with tamoxifen as above, and 22 months later euthanized. Lung, liver, kidney, spleen, thymus, stomach, pancreas, and femurs were removed at necropsy and fixed in 10% formalin (VWR, 89370-094) for 24 to 48 hrs, then post-fixed in 70% ethanol (VWR, 89125–166), processed routinely, embedded in paraffin with the flat sides down, sectioned at 5 μm, and stained with H&E. Routine processing of the tissues was performed by IDEXX Laboratories. Tissues and H&E slides were evaluated by a board-certified veterinary anatomic pathologist with experience in murine pathology.

Appendix 1

Primer sequences for genotyping and qRT-PCR analysis.

Appendix 1—table 1
Primer sequences for genotyping and qRT-PCR analysis.
Reagent type (species) or resourceDesignationSource or referenceIdentifiersAdditional information
Sequence-based reagentKras.in3.FThis paperPCR primersTTGGTGTACATCACTAGGCTTCA
Sequence-based reagentKras.in3.RThis paperPCR primersTGGAAAGAGTAAAGTGTGGTGGT
Sequence-based reagentKras.neo.FThis paperPCR primersGTGGGCTCTATGGCTTCTGA
Sequence-based reagentKrasCOM5.FThis paperPCR primersCTTCCATTTGTCACGTCCTGC
Sequence-based reagentKrasCOM5.RThis paperPCR primersTCTTCGGTGGAAACAACGGT
Sequence-based reagentF-LSLThis paperPCR primersTAGTCTGTGGGACCCCTTTG
Sequence-based reagentR-LSLThis paperPCR primersGCCTGAAGAACGAGATCAGC
Sequence-based reagentKRASOP.A2This paperPCR primersCTAGCCACCATGGCTTGAGT
Sequence-based reagentKRASOP.BThis paperPCR primersGTAATCACAACAAAGAGAATGCAG
Sequence-based reagentLSL-FThis paperPCR primersGGGGAACCTTTCAGGCTTA
Sequence-based reagentKrasRAR.FThis paperPCR primersTATGCGTACGGGTGAAGGTT
Sequence-based reagentKrasRAR.RThis paperPCR primersGCAGAGCACAGACTCACGTC
Sequence-based reagentF-CC10 WTXu et al., 2012PCR primersACTCACTATTGGGGGTGTGG
Sequence-based reagentR-CC10 WTXu et al., 2012PCR primersGGAGGACTTGTGGATCTTG
Sequence-based reagentF-CreXu et al., 2012PCR primersTCGATGCAACGAGTGATGAG
Sequence-based reagentR-CreXu et al., 2012PCR primersTTCGGCTATAGGTAACAGGG
Sequence-based reagentF-Rosa-01Xu et al., 2012PCR primersCACTTGCTCTTCCAAAGTCG
Sequence-based reagentR-Rosa-02BXu et al., 2012PCR primersTAGTCTAACTCGCGACACTG
Sequence-based reagentF-CAG-02BXu et al., 2012PCR primersGTTATGTAACGCGGAACTCC
Sequence-based reagentR26R-univFJackson LaboratoryPCR primersAAAGTCGCTCTGAGTTGTTAT
Sequence-based reagentR26R-wtRJackson LaboratoryPCR primersGGAGCGGGAGAAATGGATATG
Sequence-based reagentCreER-R1Jackson LaboratoryPCR primersCCTGATCCTGGCAATTTCG
Sequence-based reagentGLI1-FThis paperPCR primersCCCATAGGGTCTCGGGGTCTCAAAC
Sequence-based reagentGLI1-RThis paperPCR primersGGAGGACCTGCGGCTGACTGTGTAA
Sequence-based reagentCDH1-FThis paperPCR primersGTCTCCTCATGGCTTTGC
Sequence-based reagentCDH1-RThis paperPCR primersCTTTAGATGCCGCTTCAC
Sequence-based reagentTWIST1-FThis paperPCR primersAGCGGGTCATGGCTAACG
Sequence-based reagentTWIST1-RThis paperPCR primersGGACCTGGTACAGGAAGTCGA
Sequence-based reagentZEB2-FThis paperPCR primersGAGCTTGACCACCGACTC
Sequence-based reagentZEB2-RThis paperPCR primersTTGCAGGACTGCCTTGAT
Sequence-based reagentSOX5-FThis paperPCR primersATTGTGCAGTCCCACAGGTTG
Sequence-based reagentSOX5-RThis paperPCR primersCTGCCTTTAGTGGGCCAGTG
Sequence-based reagentIL6-FThis paperPCR primersCCGGAGAGGAGACTTCACAG
Sequence-based reagentIL6-RThis paperPCR primersCAGAATTGCCATTGCACAAC
Sequence-based reagentIL10-FThis paperPCR primersGGTTGCCAAGCCTTATCGGA
Sequence-based reagentIL10-RThis paperPCR primersACCTGCTCCACTGCCTTGCT
Sequence-based reagentTNFa-FThis paperPCR primersCCCCAAAGGGATGAGAAGTT
Sequence-based reagentTNFa-RThis paperPCR primersGTGGGTGAGGAGCACGTAGT
Sequence-based reagentCCL2-FThis paperPCR primersAGGTCCCTGTCATGCTTCTG
Sequence-based reagentCCL2-RThis paperPCR primersTCTGGACCCATTCCTTCTTG
Sequence-based reagentLIF-FThis paperPCR primersAATGCCACCTGTGCCATACG
Sequence-based reagentLIF-RThis paperPCR primersCAACTTGGTCTTCTCTGTCCCG
Sequence-based reagentTNFSF9-FThis paperPCR primersGCAAGCAAAGCCTCAGGTAG
Sequence-based reagentTNFSF9-RThis paperPCR primersTCCAGGAACGGTCCACTAAC
Sequence-based reagentIFNg-FThis paperPCR primersCGGGAGGTGCTGCTGATGG
Sequence-based reagentIFNg-RThis paperPCR primersAGGGACAGCCTGTTACTACC
Sequence-based reagentFOS-FThis paperPCR primersCCTGCCCCTTCTCAACGAC
Sequence-based reagentFOS-RThis paperPCR primersGCTCCACGTTGCTGATGCT
Sequence-based reagentFOSL2-FThis paperPCR primersGAGTCCTACTCCAGCGGTG
Sequence-based reagentFOSL2-RThis paperPCR primersGACTGTAGGGATGTGAGCGT
Sequence-based reagentDUSP6-FLi and Counter, 2021PCR primersACTTGGACGTGTTGGAAGAGT
Sequence-based reagentDUSP6-RLi and Counter, 2021PCR primersGCCTCGGGCTTCATCTATGAA
Sequence-based reagentEGR1-FLi and Counter, 2021PCR primersCCTGACCACAGAGTCCTTTTCT
Sequence-based reagentEGR1-RLi and Counter, 2021PCR primersAGGCCACTGACTAGGCTGA
Sequence-based reagentFOSL1-FLi and Counter, 2021PCR primersCAGGAGTCATACGAGCCCTAG
Sequence-based reagentFOSL1-RLi and Counter, 2021PCR primersGCCTGCAGGAAGTCTGTCAG
Sequence-based reagentACTIN-FLi and Counter, 2021PCR primersCGTGAAAAGATGACCCAGATCATGT
Sequence-based reagentACTIN-RLi and Counter, 2021PCR primersCGTGAGGGAGAGCATAGCC

Data availability

Initial and secondary RNAseq data are available through GEO Series accession numbers GSE181628 and GSE181627, respectively, and are accessible under the project GSE181629 (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE181629).

The following data sets were generated
    1. Erdogan O
    2. Counter CM
    (2021) NCBI Gene Expression Omnibus
    ID GSE181628. Signaling response to early oncogenesis by a novel panel of murine mutant Kras alleles in the mouse lung transcriptome.
    1. Erdogan O
    2. Counter CM
    (2021) NCBI Gene Expression Omnibus
    ID GSE181627. Signaling response to early oncogenesis by a novel panel of murine mutant Kras alleles in the mouse lung transcriptome.

References

  1. Book
    1. Behringer R
    (2014)
    Manipulating the Mouse Embryo: A Laboratory Manual
    Cold Spring Harbor Laboratory Press.
    1. Bos JL
    (1989)
    Ras oncogenes in human cancer: a review
    Cancer Research 49:4682–4689.
    1. Matkar SS
    2. Durham A
    3. Brice A
    4. Wang TC
    5. Rustgi AK
    6. Hua X
    (2011)
    Systemic activation of K-ras rapidly induces gastric hyperplasia and metaplasia in mice
    American Journal of Cancer Research 1:432–445.
  2. Software
    1. R Development Core Team
    (2020) R: A language and environment for statistical computing
    R Foundation for Statistical Computing, Vienna, Austria.

Decision letter

  1. Erica A Golemis
    Senior and Reviewing Editor; Fox Chase Cancer Center, United States
  2. Jonathan Chernoff
    Reviewer; Fox Chase Cancer Center, United States
  3. Aaron Hobbs
    Reviewer

In the interests of transparency, eLife publishes the most substantive revision requests and the accompanying author responses.

Decision letter after peer review:

Thank you for submitting your article "RAS mutation patterns arise from tissue-specific responses to distinct oncogenic signaling" for consideration by eLife. Your article has been reviewed by 3 peer reviewers, and the evaluation has been overseen by a Reviewing Editor and Erica Golemis as the Senior Editor. The following individuals involved in review of your submission have agreed to reveal their identity: Jonathan Chernoff (Reviewer #1); Aaron Hobbs (Reviewer #3).

The reviewers have discussed their reviews with one another, and the Reviewing Editor has drafted this to help you prepare a revised submission.

Essential revisions:

1) Based on their discussion, the reviewers are in agreement that use of some method is required to demonstrate that the RNA changes documented in the study result in real changes in protein-level expression of key signaling molecules. The reviewers indicate this could be addressed using various approaches, including 1) total proteome analysis; 2) immunoblots for proteins that are important to the analysis; or 3) RPPA. However, some method of evaluating protein changes must be applied, and the results integrated into the analysis.

2) In addition, each of the three reviewers made some suggestions regarding phrasing and interpretation. The authors are strongly encouraged to address these.

Reviewer #1 (Recommendations for the authors):

The authors try to establish biological differences between commonly encountered KRAS mutants. Distinct mutations exhibit tissue tropism (e.g., KRAS G12C in lung, G12D and G12R in pancreas, etc), but the reasons for these distributions have never been fully clarified. By establishing mouse models in which the type of KRAS mutant and its level of expression can be readily controlled, the authors hope to establish whether the type of amount of KRAS mutant expressed will result in distinct outcomes.

Major strengths include the extensive body of work from the Counter lab regarding the key role of codon usage in KRAS oncogenicity, the generation of several sets of matched mouse models (two of which are studied in detail, but the others are also of value), and the examination of KRAS effects in many tissues.

Weaknesses include the fact (acknowledged by the authors) that the system is highly artificial, in that KRAS is being induced globally at one point in time, that its expression levels are artificially boosted by codon changes that would not occur in vivo (though perhaps might happen if the KRAS gene were amplified, as sometimes happens), and a lack of any protein level data to corroborate the RNA profiling performed in the study.

It would strengthen the work if the authors would back up some of the key RNA signatures with western blots of RPPA-type data.

It could be instructive if the authors performed exon sequencing on some of their tumors. They saw no pancreatic tumors, which is a bit surprising, though it was perhaps because tp53 was not also mutated. One wonder what mutations might have occurred in those tissues where tumors were seen, such as lung.

Reviewer #2 (Recommendations for the authors):

Within this manuscript, the authors utilize mouse genetic approaches to modulate both which Kras mutation is expressed and how much of that Kras mutant is expressed. By utilizing different codons, they can have the same Kras mutation expressed at either a low level (due to rare codon utilization) or at a higher level (through common codon utilization). This is a clever strategy that this group pioneered for studying RAS signaling, and by now comparing two different Kras mutants (G12D and Q61R) they can also introduce additional quantitative and qualitative variation.

The authors define quantitative variation as variation in the intensity of active Kras-GTP, and qualitative variation as how the Kras-GTP may activate different downstream effector proteins. This is a well-defined concept that helps provide a structure to their analysis.

The genetic approach, and the setting of definitions, are major strengths as the authors dissect this complicated problem in a logical, systematic manner. Additional strengths include studying how the different Kras strains differ for different tissue phenotypes.

However, there are a few weaknesses to the manuscript in its current form. Although the authors' hypotheses about quantitative and qualitative variation are likely true, at times the authors appear to interpret their data more strongly than the data seem to permit. In many of these cases, a rephrasing that describes their findings objectively would be valuable because the data – at minimum – highlight the need for these quantitative and qualitative factors to be better considered within the field.

One seemingly striking weakness is that the allelic series does not seem to be sufficiently well demonstrated to have the progressive increase in mutant strengths. Another, and partially overlapping weakness, is that the authors seem to invoke tissue-specific differences when convenient and to ignore them when convenient.

The field of RAS biology has made great progress ignoring the potential effects of signal intensity and mutant-to-mutant variation, but the reinvigoration of the field in recent years has revealed important mutant-to-mutant specific differences and instances where intensity matters. A deeper understanding of these factors, and improved tools to study these differences, are both important to the field. The authors are here advancing tools and approaches that could be used in future studies and are providing a foundation for future studies. Thus, the work is likely to be impactful and influential.

One concern is the demonstration of the progressive quantitative strength of the allelic series. Even though a mouse genetic approach is utilized for the manuscript, the comparisons of Kras mutant strength is made with transiently transfected HEK-HT cells. Could the authors not do this in the MEFs they utilize, which would be endogenous expression from these cells derived from the genetically engineered mice?

The authors later claim, and acknowledge, tissue differences in factors that can contribute to the overall activation of Kras, and the authors also acknowledge qualitative differences. It seems reasonable to consider that there could be tissue-specific differences in which GEFs are activated, and that active GEFs might act upon the G12D and Q61R mutants differently – just as effector proteins may bind the mutants differently. The authors seem to be assuming the relative strengths observed are invariant between cell/tissue types and that their observations are either reflective of these quantitative differences, or that if they differ, that they must be qualitative. But why can't there be qualitative upstream differences that impact the quantity of Kras-GTP? In other words – quantitative and qualitative influences still seem to be intertwined in this system. The authors may want to state that these are strong (and unproven) assumptions, and that their interpretations are based upon these assumptions.

The authors refer to "normal cells" in the introduction; they appear to be referring to HEK-HT cells. It seems misleading to refer to them as "normal". The authors should be more precise.

The title seems to be too strongly stated. Mutation patterns also arise from the mutagens (cigarette smoking in lung causing G12C, etc.). Additionally, the authors do not show this or prove this – this is a speculation down the road from their findings, by my interpretation of their study. A more objective title would be more appropriate.

The authors state "determining the immediate response of normal cells to different Kras mutations in vivo thus holds the key to understanding the mutational patterning of this oncogene". The logic to make such a strong conclusion is not apparent. This rather seems to be a hypothesis, that studying the immediate response may offer some insight into mutational patterning. Several specific details in the sentence appear overly strong. Can this be rephrased?

The sentence "In this way, tissues sensitive to quantitative signaling…". This is important, as it highlights the authors interpretation of their data (and possible flaws with this assumption are mentioned above). Could the authors rephrase this a bit more mildly to describe that the authors interpret their data to indicate quantitative signaling when… and qualitative signaling when…

Mention of NRAS G12D and KRAS G12D does not mention the study by Haigis et al., Nat Genetics, 2008. Is that study not relevant?

When discussing Figure 3A and Figure 3 —figure supplement 2-5, the authors describe a progressive change. Do they simply mean more genes that are significantly different? Or do the identities of genes progressively change? IT is difficult to dissect finer details on the genes from the plot.

In general, increased clarity would be helpful, and more objective descriptions of findings without stronger conclusions than are justified would be helpful.

The main experiment this reviewer would like to see is measurements of RAS-GTP in the MEFs, and – if possible (very likely not) – in specific tissues. (Are mouse macrophages or immune cells not isolatable for measurement?) Would western blots of RAF, MEK, or ERK phosphorylation not be useful to assess signaling? Qualitative differences are of course possible between G12D and Q61R, but it would seem that one could at least evaluate if natural and common codons alter expression of Kras enough to result in changes to signaling downstream from the same mutant (G12D or Q61R).

Reviewer #3 (Recommendations for the authors):

The manuscript by Erdogan et al., asks a straightforward question and tightly focusses on answering that question. Further, this manuscript benefits from being very well written and clearly stating the limitations of this study. The authors ask whether the mutational prevalence of KRAS mutations observed in human tumors is due to qualitative (which pathways are engaged) or quantitative (the amplitude of signal) signaling. The authors chose to use a variation on a mouse model that the Counter Lab has previously utilized to approach this question. By using genetically engineered mouse models that have had the first three coding exons of the KRAS locus exchanged to use only rare or common codons to regulate KRAS protein expression levels, and by using two distinct mutations, the common G12D and the rare Q61R, under control of the Rosa26 promoter, this system allowed for mutant protein expression in a broad range of tissues. By measuring the overall GTP-bound levels, a measure of RAS activity, the authors achieved their goal of driving a step-wise increase of total GTP-bound KRAS in the four different models. They reason that quantitative signaling can be observed by changing the expression levels of the same mutant while quantitative signaling is observed by expression of two different mutants that have similar activity levels (KrasnatQ61R vs KrascomG12D).

One limitation to this study is the use of mouse models where the Kras locus has been heavily modified. Further, the use of rare and common codons could have tissue specific effects. However, the authors reason that each mouse has been subjected to the same types of manipulations and RAS function was verified in lung tissue. Because of these limitations, the authors do caution that the degree of signaling cannot be readily compared to previous studies but that these mice can be compared to each other.

It is accepted that the KRAS Q61R mutation has greater overall GTP activity compared to G12D. This was shown in this manuscript using GTP pulldown assays and an ELISA based luciferase assay, showing increased GTP loading in the Q61R mutant relative to G12D and increased expression resulted in increased GTP-bound KRAS. A CRAF RBD pulldown confirmed increased GTP loading. While RBD-pulldowns can be affected by the binding affinity of the RAS mutants to the RBD domain, the use of two distinct assays helps alleviate that concern.

After confirming the model system was behaving as expected, the authors looked at lung tumors and found that increasing activity directly correlated with increasing tumorigenesis, in support of quantitative signaling playing a significant role in tumor formation. However, they also observed that the type of lesions that developed were dependent on the specific KRAS mutation, indicating that qualitative signaling also dictates tumorigenic potential.

The authors performed bulk RNAseq on the total lung tissue to determine how each KRAS allele altered gene transcription levels. Here, the authors found that a number of transcripts involved in numerous connected pathways, analyzed by GSEA analysis, were modulated by the RAS mutants. Importantly, the total number of upregulated genes correlated with total RAS activity levels, indicating that quantitative signaling drives different transcription programs. It is surprising that there is not more overlap between the different mutant RAS constructs in this system. It would be reasonable to expect active RAS to have a core set of effectors that are independent of quantitative and qualitative signaling. However, the extent to which there is any overlap between RAS mutants or by degree of expression is not broadly discussed.

While the active RAS GTP levels appeared to be dependent on both the KRAS mutation and expression levels, the overall survival of the mice expressing the natural codons of both KRASG12D and Q61R were similar. While survival is a global readout of total oncogenic activity, this result does imply that a certain degree of quantitative signaling is necessary to overcome differences in qualitative signaling, a conclusion that the author's also state. Further, looking at tumorigenesis in 8 different tissue types, the authors observed that in some tissue types, quantitative signaling appears to drive the overall malignancy (hematopoietic), while in others (forestomach), qualitative signaling is a larger driver of malignancy.

In conclusion, this study supports the possibility that qualitative signaling preferences of KRAS mutants may drive tumor initiation, in agreement with several recent studies showing mutation-specific signaling, such as KRAS A146T and KRAS G12R. While this study is limited to only two KRAS mutations, the authors suggest that these differences could drive differential inhibitor sensitivities. However, one hallmark of many RAS mutant cancers is the development of increased RAS activity during tumor development, which implies that quantitative signaling may require greater consideration for therapeutic intervention of established tumors. Further, this result leads one to question how successful mutation-specific therapeutics may be in the clinic in qualitative signaling can be masked by quantitative signaling. However, these results also shed light on tumor initiation, which may be useful in determining early detection markers that may be mutation dependent.

Overall, the manuscript has a central theme and the authors do not deviate from this theme. This results in a highly focused paper that clearly draws its main points while being very aware of the limitations of the approach. With these limitations in mind, this manuscript clearly demonstrates mutation specific signaling in an in vivo model, demonstrating yet again that all KRAS mutations are not created equal.

1. Overall, I note very few problems with this text. However, Figure 1 C needs additional clarification. The methods state that the abcam RAS GTPase ELISA kit was used. The product manual states that this kit detects activated H and KRAS human and HRAS rodent samples, if employed per the manufacturer's kit. Was a different primary used? According to the methods, the FLAG-tagged constructs were Kras, which implies a murine system. Given that the empty vector increases with increasing protein extract, it is likely that this system is detecting the endogenous human GTPases and not the ectopically expressed murine construct. While the supplemental data do agree with the expected results based on the published literature, this figure could possibly be improved.

2. In figure 1, supplement 4 panel B, the KrasnatG12D/+ bar does not match the legend.

3. In Figure 5, the gene name for the PI3Ks are incorrect. The correct gene names for p110 isoforms are PIK3CA, PIK3CD and PIK3CG.

4. While not necessary but could improve Figure 5, could testing which inhibitor class on overall survival in the common codon mutants add improve survival? If done in the Q61 mutant, it could be quick to validate the specificity between the classes of inhibitors.

5. Is it possible that expression levels may alter mutant-specific (qualitative) signaling as well? Would this appear as quantitative? More specifically, what was the degree of overlap within the RAS mutants when changing from natural to rare codons? Were all of the hits observed in the natural codon mice detected in the common codon mice? This might be a more useful figure than the STRING analysis.

6. On page 4, it is mentioned that Q61R has slower GTP exchange and hydrolysis rates than G12D. It is worth pointing out that the Burd et al., reference is NRAS while the Rabara et al., reference compared GAP-mediated hydrolysis. However, G12D is interesting because its GTP exchange rate has been reported to be elevated (PMID 8955068 and PMID 24224811), being faster than G13D in biochemical studies, yet this is never discussed. Thus, the author is not wrong in their statement, but additional references and clarification could be useful.

7. On page 6, the authors wrote “special control of gene expression” although I believe they meant “special control”.

8. On page 9, first full paragraph. Discussing RNAseq data as a direct proxy for protein signaling is not entirely valid. RNA levels, which do indicate transcript levels, do not necessarily mean that signaling in those pathways is elevated.

9. On page 11, first paragraph. “Of note, median survival was significantly decreased in mice with either of the Krascom alleles or with either of the Q61R mutants.” This sentence is not entirely clear. Median survival was decreased compared to what? And didn’t the Q61Rnat and G12Dnat have similar survival rates? Please clarify.

https://doi.org/10.7554/eLife.75715.sa1

Author response

Essential revisions:

1) Based on their discussion, the reviewers are in agreement that use of some method is required to demonstrate that the RNA changes documented in the study result in real changes in protein-level expression of key signaling molecules. The reviewers indicate this could be addressed using various approaches, including 1) total proteome analysis; 2) immunoblots for proteins that are important to the analysis; or 3) RPPA. However, some method of evaluating protein changes must be applied, and the results integrated into the analysis.

As requested, we validate the major finds that the four LSL-Kras alleles generate different levels of active Kras (as assessed by RBD affinity capture) that imparts distinct cellular responses at the protein level (as assessed by RPPA) at the endogenous level and in MEF cultures in vitro and/or actual lung tissue a mere days after oncogene activation in vivo, as summarized below:

1. Confirmed a stepwise increase in activation of the Kras oncoprotein in the allelic LSL-Kras set at the endogenous level by Ras Binding Domain (RBD) affinity capture in MEF cultures in vitro. We generated immortalized mouse embryonic fibroblast (MEF) from two independent embryos (i.e. two biological replicates) from each of the genotypes LSL-KrasnatG12D/+, LSL-KrasnatQ61R/+, LSL-KrascomG12D/+, and LSL-KrascomQ61R/+. All eight cell lines were then stably infected with a Cre-expressing retrovirus and successful recombination confirmed by genomic PCR. Two independent immortalized MEF cell lines derived from LSL-KrasnatG12D/+ mice but not expressing Cre recombinase were cultured in parallel as a control. Duplicate cultures of each of the ten MEF cell lines were serum-starved overnight and the next day one plate was collected as the serum-starved condition to assay basal Kras activity and another plate was treated with serum for 5 minutes before lysates were collected to assess the effect of serum stimulation. Protein lysates were prepared from these 20 samples, protein concentrations were determined, and equal amount of total protein was subjected to affinity capture with a Ras Binding Domain (RBD) polypeptide. Total and RBD pull-down protein samples were resolved by SDS-PAGE and immunoblotted with an anti-KRAS antibody. This analysis confirmed our previous results from HEK-HT cells (now Figure 1—figure supplement 3 and Figure 1—source data 1), namely that changing rare codons to common increases the total amount of Kras protein detected, and that the Q61R mutant leads to higher levels of Kras-GTP than the G12D mutant. This resulted in a stepwise increase in endogenous Kras-GTP levels, beginning with the control cells lacking detectable GTP-Kras, followed in order of increasing activity, the Cre-activated LSL-KrasnatG12D/+, LSL-KrasnatQ61R/+, LSL-KrascomG12D/+, and finally LSL-KrascomQ61R/+ MEF cultures, which was observed in both the serum-starved and serum-stimulated conditions. These new data are provided in the new panel Figure 1C and Figure 1—source data 3. To ensure reproducibility and rigor, we repeated this experiment (i.e. two technical replicates). These new data are provided as the new Figure 1—figure supplement 4 and Figure 1—source data 3.

2. Confirmed a stepwise increase in activation of the Kras oncoprotein in the allelic LSL-Kras set at the endogenous level by Ras Binding Domain (RBD) affinity capture in lung tissue in vivo. We also performed the above analysis from lung tissue isolated from the same allelic set. Namely, we crossed the LSL-KrasnatG12D/+, LSL-KrasnatQ61R/+, LSL-KrascomG12D/+, and LSL-KrascomQ61R/+ genotypes into a Rosa26-CreERT2 background. Two adult mice from each of the four derived cohorts, as well as the control strain (Rosa26-CreERT2/+), were injected with tamoxifen and seven days later humanely euthanized, their lungs removed, and duplicate samples of derived protein lysates were subjected to RBD affinity capture followed by immunoblot (Figure 1—figure supplement 5A). As in the case of MEF cultures described above, we again observed a stepwise increase in Kras activity consistent with the codon usage and oncogenic mutation in vivo. These new data are provided in the new Figure 1—figure supplement 5B,C and Figure 1—source data 4.

3. Confirmed increased activation of each protein in the MAPK pathway in the two most phenotypically different alleles, LSL-KrasnatG12 versus LSL-KrascomQ61R, at the endogenous level by RPPA in lung tissue in vivo. Cohorts of three Rosa26::CreERT2/+;LSL-KrasnatG12D/+ versus three Rosa26::CreERT2/+;LSL-KrascomQ61R/+ (and as a control, Rosa26::CreERT2/+;Kras+/+) mice comprised of an equal distribution of males and females were injected with tamoxifen at 6 to 8 weeks of age to induce recombination and activation of these inducible Kras alleles. We chose these two genotypes as they exhibit the greatest difference in transcriptional profiles (see Figure 3A,B) and tumorigenesis (see Figures 2,4) and encompass both differences in codon usage and mutation type. Seven days later, all mice were humanely euthanized and the lungs removed at the time of necropsy. We chose the lung both because it was the same tissue for which we previously performed RNAseq analysis (see Figure 3) and because this organ exhibits evidence of both the individual mutations as well as codon bias influencing tumor initiation (see Figure 4F). We chose seven days post-injection for this analysis, both because it is earliest point that we validate robust recombination of the LSL-Kras alleles in order to capture the response of normal cells to the oncoprotein (see Figure 4—figure supplement 2) and, again, because this is the same timepoint we previously performed RNAseq analysis (see Figure 3). Protein lysates were then isolated from lung sections and subjected to RPPA analysis. To ensure reproducibility and rigor, lungs from all mice (three biological replicates) were tested in triplicate (three technical replicates). We note that originally we had four mice in the Rosa26::CreERT2/+;LSL-KrasnatG12D/+ cohort, but one sample was censored as it had a completely different RPPA profile compared to the other three mice in this cohort. RPPA was performed on 140 proteins/phosphoproteins representative of pathways altered the most in the two extreme transcriptional profiles, namely derived from the lungs of LSL-KrasG12D(nat) versus LSL-KrasQ61R(com) mice shortly after recombination was induced. Focusing on the MAPK pathway, we report increased phosphorylation of the proteins of this pathway in the Rosa26::CreERT2/+;LSL-KrascomQ61R/+ murine lung tissue. Again, in complete concordance with the immunoblot analysis of Kras-GTP levels in MEFs and/or lung tissue, as well as the transcriptome analysis of lung tissue. These new data are provided in the new Figure 1—figure supplement 1 and Supplementary file 1.

4. Confirmed that the transcriptomes manifest in differences at the protein level in the two most different alleles, LSL-KrasnatG12 versus LSL-KrascomQ61R, at the endogenous level by RPPA in lung tissue in vivo. As noted, our RPPA analysis focused on proteins representative of pathways altered in the two extreme transcriptional profiles. Specifically, proteins within the RAS/MAPK, PI3K/AKT/mTOR, Growth, IL/Jak/Stat, DNA repair, Apoptosis, Autophagy, and Senescence pathways (see Figure 3B,C and Figure 3—figure supplement 1C). We report concordance between the transcriptional and proteomic profiles, namely an upregulation in proteins in the MAPK pathway, those involved in stress response and DNA damage, and cytokine activation. These new data are provided in the new Figure 3—figure supplement 9 and Supplementary file 1.

2) In addition, each of the three reviewers made some suggestions regarding phrasing and interpretation. The authors are strongly encouraged to address these.

As requested, all other concerns were positively addressed, either by performing the suggested experiments or significantly revising the text accordingly.

Reviewer #1 (Recommendations for the authors):

The authors try to establish biological differences between commonly encountered KRAS mutants. Distinct mutations exhibit tissue tropism (e.g., KRAS G12C in lung, G12D and G12R in pancreas, etc), but the reasons for these distributions have never been fully clarified. By establishing mouse models in which the type of KRAS mutant and its level of expression can be readily controlled, the authors hope to establish whether the type of amount of KRAS mutant expressed will result in distinct outcomes.

Major strengths include the extensive body of work from the Counter lab regarding the key role of codon usage in KRAS oncogenicity, the generation of several sets of matched mouse models (two of which are studied in detail, but the others are also of value), and the examination of KRAS effects in many tissues.

Weaknesses include the fact (acknowledged by the authors) that the system is highly artificial, in that KRAS is being induced globally at one point in time, that its expression levels are artificially boosted by codon changes that would not occur in vivo (though perhaps might happen if the KRAS gene were amplified, as sometimes happens), and a lack of any protein level data to corroborate the RNA profiling performed in the study.

We thank the reviewer for both their encouraging review and thoughtful suggestions. We note here that while admittedly artificial, whole-body activation of each of the four novel LSL-Kras alleles was key to identify the sensitivities of different tissues to each mutant. We also note here that changing rare codons to common is essentially a hypermorphic allele, which is a staple of mouse genetics. Nevertheless, we see your point, so if there are specific recommendations to the text beyond what we have already included to better describe the limitations of these two genetic approaches, we would be glad to include them- we want to get this right! As requested, please see comment 1 below describing RPPA analysis.

It would strengthen the work if the authors would back up some of the key RNA signatures with western blots of RPPA-type data.

As requested, cohorts of three Rosa26::CreERT2/+;LSL-KrasnatG12D/+ versus three Rosa26::CreERT2/+;LSL-KrascomQ61R/+ (and as a control, Rosa26::CreERT2/+;Kras+/+) mice comprised of an equal distribution of males and females were injected with tamoxifen at 6 to 8 weeks of age to induce recombination and activation of these inducible Kras alleles. We chose these two genotypes as they exhibit the greatest difference in transcriptional profiles (see Figure 3A,B) and tumorigenesis (see Figures 2,4). Seven days later, all mice were humanely euthanized and the lungs removed at the time of necropsy. We chose the lung both because it was the same tissue for which we previously performed RNAseq analysis (see Figure 3) and because this organ exhibits evidence of both the individual mutations as well as codon bias influencing tumor initiation (see Figure 4F). We chose seven days post-injection for this analysis, both because it is earliest point that we validate robust recombination of the LSL-Kras alleles in order to capture the response of normal cells to the oncoprotein (see Figure 4—figure supplement 2) and, again, because this is the same timepoint we previously performed RNAseq analysis (see Figure 3). Protein lysates were then isolated from lung sections and subjected to RPPA analysis. To ensure reproducibility and rigor, lungs from all mice (three biological replicates) were tested in triplicate (three technical replicates). We note that originally we had four mice in the Rosa26::CreERT2/+;LSL-KrasnatG12D/+ cohort, but one sample was censored as it had a completely different RPPA profile compared to the other three mice in this cohort. RPPA was performed on 140 proteins/phosphoproteins representative of pathways altered the most in the two extreme transcriptional profiles, namely derived from the lungs of LSL-KrasG12D(nat) versus LSL-KrasQ61R(com) mice shortly after recombination was induced. Focusing on the MAPK pathway, we report increased phosphorylation of the proteins of this pathway in the Rosa26::CreERT2/+;LSL-KrascomQ61R/+ compared to the Rosa26::CreERT2/+;LSL-KrasnatG12D/+ murine lung tissue. These new data are provided in the new Figure 1—figure supplement 7 and Supplementary file 1. Additionally, we also included analysis of proteins/phosphoproteins from pathways that transcriptome analysis identified as being upregulated in the lung upon activation of the KrascomQ61 compared to the KrasnatG12D allele. Bulk transcriptome analysis revealed that activation of the KrascomQ61R allele had all features of a potent oncogenic signaling leading to hyperproliferation and oncogenic-induced stress. Namely, GSEA hallmarks indicative of high oncogenic signaling (KRAS Signaling UP) and unrestrained proliferation (MYC Targets V1, E2F Targets), leading to reactive oxygen species (Oxidative Phosphorylation, Reactive Oxygen Species Pathway) and a DNA damage response (DNA Repair, P53 Pathway, G2M Checkpoint) followed by apoptosis (Apoptosis) and inflammation (Inflammatory Response, IL6 JAK STAT3 Signaling, TNFa Signaling via NF-kb, Allograft Rejection) (see Figure 3). Comparing the level of proteins/phosphoprotein by RPPA of KrasnatG12D to KrascomQ61 (normalized to the control strain) revealed a clear increase in the latter in RAS/MAPK (as already noted), PI3K/AKT, Growth, DNA repair, senescence/autophagy/apoptosis, and IL/Jak/Stat, thereby validating the transcriptome analysis. These new data are included as Figure 3—figure supplement 9 and Supplementary file 1.

It could be instructive if the authors performed exon sequencing on some of their tumors. They saw no pancreatic tumors, which is a bit surprising, though it was perhaps because tp53 was not also mutated. One wonder what mutations might have occurred in those tissues where tumors were seen, such as lung.

This is a really great suggestion, and one that we will most certainly perform in the future as it would be ever-so-interesting to see if the type of foundational Kras mutation influences the nature of mutations selected to drive tumor progression. However, since the current study focuses on tumor initiation and exploring this new idea will be a major undertaking to do correctly, we propose to address this new concept in the future. Thank you for this idea! With regards to why pancreatic tumors fail to develop in the genetic backgrounds we employed in this study, we are not the first to observe this. Indeed, we now clarify in the revised text that Parikh, et al., (PMID: 22532587) and Van der Weyden, et al., (PMID:21381032) both published that inducing recombination of an LSL-KrasG12D allele in the identical Rosa26-CreERT2 background in adult mice did not generate pancreatic tumors. Instead, we note here that an oncogenic Kras allele induces pancreatic tumors if activated either during development or in adult when accompanied by either pancreatitis or loss of tumor suppressors (Guerra and Barbacid, Mol Oncol, 2013, PMID: 23506980), but not in adults without further perturbations.

Reviewer #2 (Recommendations for the authors):

Within this manuscript, the authors utilize mouse genetic approaches to modulate both which Kras mutation is expressed and how much of that Kras mutant is expressed. By utilizing different codons, they can have the same Kras mutation expressed at either a low level (due to rare codon utilization) or at a higher level (through common codon utilization). This is a clever strategy that this group pioneered for studying RAS signaling, and by now comparing two different Kras mutants (G12D and Q61R) they can also introduce additional quantitative and qualitative variation.

The authors define quantitative variation as variation in the intensity of active Kras-GTP, and qualitative variation as how the Kras-GTP may activate different downstream effector proteins. This is a well-defined concept that helps provide a structure to their analysis.

The genetic approach, and the setting of definitions, are major strengths as the authors dissect this complicated problem in a logical, systematic manner. Additional strengths include studying how the different Kras strains differ for different tissue phenotypes.

However, there are a few weaknesses to the manuscript in its current form. Although the authors' hypotheses about quantitative and qualitative variation are likely true, at times the authors appear to interpret their data more strongly than the data seem to permit. In many of these cases, a rephrasing that describes their findings objectively would be valuable because the data – at minimum – highlight the need for these quantitative and qualitative factors to be better considered within the field.

One seemingly striking weakness is that the allelic series does not seem to be sufficiently well demonstrated to have the progressive increase in mutant strengths. Another, and partially overlapping weakness, is that the authors seem to invoke tissue-specific differences when convenient and to ignore them when convenient.

The field of RAS biology has made great progress ignoring the potential effects of signal intensity and mutant-to-mutant variation, but the reinvigoration of the field in recent years has revealed important mutant-to-mutant specific differences and instances where intensity matters. A deeper understanding of these factors, and improved tools to study these differences, are both important to the field. The authors are here advancing tools and approaches that could be used in future studies and are providing a foundation for future studies. Thus, the work is likely to be impactful and influential.

We thank the reviewer for both their encouraging review and thoughtful suggestions. As requested, we extensively revised the text to better describe the data and our interpretations thereof, especially with regards to different interpretations of qualitative signaling, although if there are specific sentences or points the reviewer would like us to change or add, we would be glad to abide by their suggestions. As requested, we also positively addressed the issue of allelic strength in reply to your comment 1 below.

One concern is the demonstration of the progressive quantitative strength of the allelic series. Even though a mouse genetic approach is utilized for the manuscript, the comparisons of Kras mutant strength is made with transiently transfected HEK-HT cells. Could the authors not do this in the MEFs they utilize, which would be endogenous expression from these cells derived from the genetically engineered mice?

As requested, we generated immortalized mouse embryonic fibroblast (MEF) from two independent embryos (i.e. two biological replicates) from each of the genotypes LSL-KrasnatG12D/+, LSL-KrasnatQ61R/+, LSL-KrascomG12D/+, and LSL-KrascomQ61R/+. All eight cell lines were then stably infected with a Cre-expressing retrovirus and successful recombination confirmed by genomic PCR. Two independent immortalized MEF cell lines derived from LSL-KrasnatG12D/+ mice but not expressing Cre recombinase were cultured in parallel as a control. Duplicate cultures of each of the ten MEF cell lines were serum-starved overnight and the next day one plate was collected as the serum-starved condition to assay basal Kras activity and another plate was treated with serum for 5 minutes before lysates were collected to assess the effect of serum stimulation. Protein lysates were prepared from these 20 samples, protein concentrations were determined, and equal amount of total protein was subjected to affinity capture with a Ras Binding Domain (RBD) polypeptide. Total and RBD pull-down protein samples were resolved by SDS-PAGE and immunoblotted with an anti-KRAS antibody. This analysis confirmed our previous results from HEK-HT cells (now Figure 1—figure supplement 3 and Figure 1—source data 1), namely that changing rare codons to common increases the total amount of Kras protein detected, and that the Q61R mutant leads to higher levels of Kras-GTP than the G12D mutant. This resulted in a stepwise increase in endogenous Kras-GTP levels, beginning with the control cells lacking detectable GTP-Kras, followed in order of increasing activity, the Cre-activated LSL-KrasnatG12D/+, LSL-KrasnatQ61R/+, LSL-KrascomG12D/+, and finally LSL-KrascomQ61R/+ MEF cultures, which was observed in either the serum-starved or serum-stimulated conditions. These new data are provided in the new panel Figure 1C and Figure 1—source data 3. To ensure reproducibility and rigor, we repeated this experiment (i.e. two technical replicates). These new data are provided as the new Figure 1—figure supplement 4 and Figure 1—source data 3. Additionally, please see our response to your comment 10b, as we also repeated this analysis on the lungs of mice from these four backgrounds, the results of which mirrored those from MEF cultures.

The authors later claim, and acknowledge, tissue differences in factors that can contribute to the overall activation of Kras, and the authors also acknowledge qualitative differences. It seems reasonable to consider that there could be tissue-specific differences in which GEFs are activated, and that active GEFs might act upon the G12D and Q61R mutants differently – just as effector proteins may bind the mutants differently. The authors seem to be assuming the relative strengths observed are invariant between cell/tissue types and that their observations are either reflective of these quantitative differences, or that if they differ, that they must be qualitative. But why can't there be qualitative upstream differences that impact the quantity of Kras-GTP? In other words – quantitative and qualitative influences still seem to be intertwined in this system. The authors may want to state that these are strong (and unproven) assumptions, and that their interpretations are based upon these assumptions.

As requested, we revised the text to highlight this point.

The authors refer to "normal cells" in the introduction; they appear to be referring to HEK-HT cells. It seems misleading to refer to them as "normal". The authors should be more precise.

We revised the text so that “normal cells” does not refer to HEK-HT cells.

The title seems to be too strongly stated. Mutation patterns also arise from the mutagens (cigarette smoking in lung causing G12C, etc.). Additionally, the authors do not show this or prove this – this is a speculation down the road from their findings, by my interpretation of their study. A more objective title would be more appropriate.

As requested, we changed the title to “Genetically manipulating endogenous Kras levels and oncogenic mutations in vivo influences tissue patterning of murine tumorigenesis”.

The authors state "determining the immediate response of normal cells to different Kras mutations in vivo thus holds the key to understanding the mutational patterning of this oncogene". The logic to make such a strong conclusion is not apparent. This rather seems to be a hypothesis, that studying the immediate response may offer some insight into mutational patterning. Several specific details in the sentence appear overly strong. Can this be rephrased?

As requested, we rephrased this sentence.

The sentence "In this way, tissues sensitive to quantitative signaling…". This is important, as it highlights the authors interpretation of their data (and possible flaws with this assumption are mentioned above). Could the authors rephrase this a bit more mildly to describe that the authors interpret their data to indicate quantitative signaling when… and qualitative signaling when…

As requested, we rephrased this and other similar sentences to address this point.

Mention of NRAS G12D and KRAS G12D does not mention the study by Haigis et al., Nat Genetics, 2008. Is that study not relevant?

As requested, while the focus of this study was to compare the same isoform with two different mutations and expression levels to parse out the potential contribution of quantitative versus qualitative signaling (with the caveats of this approach noted), we agree that there is great value in expanding our interpretations to isoforms, and now cite this informative study in the introduction.

When discussing Figure 3A and Figure 3 —figure supplement 2-5, the authors describe a progressive change. Do they simply mean more genes that are significantly different? Or do the identities of genes progressively change? IT is difficult to dissect finer details on the genes from the plot.

As requested, we rephrased the explanation emphasizing that the total number of different genes differentially expressed in the mutant alleles in comparison to the WT allele gradually increase as the RAS activity (as defined by RBD pull-downs) of the oncoprotein encoded by these alleles increase (i.e. the identity of the genes changes). As requested, to make this easier to see we include full-page volcano plots in Figure 3—figure supplements 2-5. We also clarify that genes differentially expressed in lung tissue with the lowest level of Kras activity remain activated in when Kras is more highly activated (i.e. expression of the same genes throughout the genotypes). As requested, to clarify this point we also add a Venn diagram breaking down the number of genes shared between all possible comparisons in the new Figure 3—figure supplement 6.

In general, increased clarity would be helpful, and more objective descriptions of findings without stronger conclusions than are justified would be helpful.

As requested, we extensively revised the text accordingly.

The main experiment this reviewer would like to see is measurements of RAS-GTP in the MEFs, and – if possible (very likely not) – in specific tissues. (Are mouse macrophages or immune cells not isolatable for measurement?)

As requested, we performed the suggested experiment in MEF cultures which a stepwise increase in endogenous Kras-GTP levels, beginning with the control cells lacking detectable GTP-Kras, followed in order of increasing activity, the Cre-activated LSL-KrasnatG12D/+, LSL-KrasnatQ61R/+, LSL-KrascomG12D/+, and finally LSL-KrascomQ61R/+ MEF cultures. Please see our reply to your comment 1 for details and Figure 1C, Figure 1—figure supplement 4, and Figure 1—source data 3. As requested, we additionally performed RBD affinity capture on lysates derived from lung tissue isolated from the same allelic set. Namely, we crossed the LSL-KrasnatG12D/+, LSL-KrasnatQ61R/+, LSL-KrascomG12D/+, and LSL-KrascomQ61R/+ genotypes into a Rosa26-CreERT2 background. Two adult mice from each of the four derived cohorts, as well as the control strain (Rosa26-CreERT2/+), were injected with tamoxifen and seven days later humanely euthanized, their lungs removed, and duplicate samples of derived protein lysates were subjected to RBD affinity capture followed by immunoblot (Figure 1—figure supplement 5A). As in the case of MEF cultures, we again observed a stepwise increase in Kras activity consistent with the codon usage and oncogenic mutation in vivo. These new data are provided in the new Figure 1—figure supplement 5 and Figure 1—source data 4.

Would western blots of RAF, MEK, or ERK phosphorylation not be useful to assess signaling? Qualitative differences are of course possible between G12D and Q61R, but it would seem that one could at least evaluate if natural and common codons alter expression of Kras enough to result in changes to signaling downstream from the same mutant (G12D or Q61R).

As requested, cohorts of three Rosa26::CreERT2/+;LSL-KrasnatG12D/+ versus three Rosa26::CreERT2/+;LSL-KrascomQ61R/+ (and as a control, Rosa26::CreERT2/+;Kras+/+) mice comprised of an equal distribution of males and females were injected with tamoxifen at 6 to 8 weeks of age to induce recombination and activation of these inducible Kras alleles. We chose these two genotypes as they exhibit the greatest difference in transcriptional profiles (see Figure 3A,B) and tumorigenesis (see Figures 2,4) as well as encompass differences in both codon usage and mutation type. Seven days later, all mice were humanely euthanized and the lungs removed at the time of necropsy. We chose the lung both because it was the same tissue for which we previously performed RNAseq analysis (see Figure 3) and because this organ exhibits evidence of both the individual mutations as well as codon bias influencing tumor initiation (see Figure 4F). We chose seven days post-injection for this analysis, both because it is earliest point that we validate robust recombination of the LSL-Kras alleles in order to capture the response of normal cells to the oncoprotein (see Figure 4—figure supplement 2) and, again, because this is the same timepoint we previously performed RNAseq analysis (see Figure 3). Protein lysates were then isolated from lung sections and subjected to RPPA analysis. To ensure reproducibility and rigor, lungs from all mice (three biological replicates) were tested in triplicate (three technical replicates). We note that originally we had four mice in the Rosa26::CreERT2/+;LSL-KrasnatG12D/+ cohort, but one sample was censored as it had a completely different RPPA profile compared to the other three mice in this cohort. RPPA was performed on 140 proteins/phosphoproteins representative of pathways altered the most in the two extreme transcriptional profiles, namely derived from the lungs of LSL-KrasG12D(nat) versus LSL-KrasQ61R(com) mice shortly after recombination was induced. Focusing on the MAPK pathway proteins in this analysis, we report increased phosphorylation of the proteins of this pathway in the Rosa26::CreERT2/+;LSL-KrascomQ61R/+ compared to the Rosa26::CreERT2/+;LSL-KrasnatG12D/+ mice lung tissue. Again, in complete concordance with the immunoblot analysis of Kras-GTP levels in MEFs and lung tissue, as well as the transcriptome of these cells. These new data are provided in the new Figure 1—figure supplement 7 and Supplementary file 1.

Reviewer #3 (Recommendations for the authors):

The manuscript by Erdogan et al., asks a straightforward question and tightly focusses on answering that question. Further, this manuscript benefits from being very well written and clearly stating the limitations of this study. The authors ask whether the mutational prevalence of KRAS mutations observed in human tumors is due to qualitative (which pathways are engaged) or quantitative (the amplitude of signal) signaling. The authors chose to use a variation on a mouse model that the Counter Lab has previously utilized to approach this question. By using genetically engineered mouse models that have had the first three coding exons of the KRAS locus exchanged to use only rare or common codons to regulate KRAS protein expression levels, and by using two distinct mutations, the common G12D and the rare Q61R, under control of the Rosa26 promoter, this system allowed for mutant protein expression in a broad range of tissues. By measuring the overall GTP-bound levels, a measure of RAS activity, the authors achieved their goal of driving a step-wise increase of total GTP-bound KRAS in the four different models. They reason that quantitative signaling can be observed by changing the expression levels of the same mutant while quantitative signaling is observed by expression of two different mutants that have similar activity levels (KrasnatQ61R vs KrascomG12D).

One limitation to this study is the use of mouse models where the Kras locus has been heavily modified. Further, the use of rare and common codons could have tissue specific effects. However, the authors reason that each mouse has been subjected to the same types of manipulations and RAS function was verified in lung tissue. Because of these limitations, the authors do caution that the degree of signaling cannot be readily compared to previous studies but that these mice can be compared to each other.

It is accepted that the KRAS Q61R mutation has greater overall GTP activity compared to G12D. This was shown in this manuscript using GTP pulldown assays and an ELISA based luciferase assay, showing increased GTP loading in the Q61R mutant relative to G12D and increased expression resulted in increased GTP-bound KRAS. A CRAF RBD pulldown confirmed increased GTP loading. While RBD-pulldowns can be affected by the binding affinity of the RAS mutants to the RBD domain, the use of two distinct assays helps alleviate that concern.

After confirming the model system was behaving as expected, the authors looked at lung tumors and found that increasing activity directly correlated with increasing tumorigenesis, in support of quantitative signaling playing a significant role in tumor formation. However, they also observed that the type of lesions that developed were dependent on the specific KRAS mutation, indicating that qualitative signaling also dictates tumorigenic potential.

The authors performed bulk RNAseq on the total lung tissue to determine how each KRAS allele altered gene transcription levels. Here, the authors found that a number of transcripts involved in numerous connected pathways, analyzed by GSEA analysis, were modulated by the RAS mutants. Importantly, the total number of upregulated genes correlated with total RAS activity levels, indicating that quantitative signaling drives different transcription programs. It is surprising that there is not more overlap between the different mutant RAS constructs in this system. It would be reasonable to expect active RAS to have a core set of effectors that are independent of quantitative and qualitative signaling. However, the extent to which there is any overlap between RAS mutants or by degree of expression is not broadly discussed.

While the active RAS GTP levels appeared to be dependent on both the KRAS mutation and expression levels, the overall survival of the mice expressing the natural codons of both KRASG12D and Q61R were similar. While survival is a global readout of total oncogenic activity, this result does imply that a certain degree of quantitative signaling is necessary to overcome differences in qualitative signaling, a conclusion that the author's also state. Further, looking at tumorigenesis in 8 different tissue types, the authors observed that in some tissue types, quantitative signaling appears to drive the overall malignancy (hematopoietic), while in others (forestomach), qualitative signaling is a larger driver of malignancy.

In conclusion, this study supports the possibility that qualitative signaling preferences of KRAS mutants may drive tumor initiation, in agreement with several recent studies showing mutation-specific signaling, such as KRAS A146T and KRAS G12R. While this study is limited to only two KRAS mutations, the authors suggest that these differences could drive differential inhibitor sensitivities. However, one hallmark of many RAS mutant cancers is the development of increased RAS activity during tumor development, which implies that quantitative signaling may require greater consideration for therapeutic intervention of established tumors. Further, this result leads one to question how successful mutation-specific therapeutics may be in the clinic in qualitative signaling can be masked by quantitative signaling. However, these results also shed light on tumor initiation, which may be useful in determining early detection markers that may be mutation dependent.

Overall, the manuscript has a central theme and the authors do not deviate from this theme. This results in a highly focused paper that clearly draws its main points while being very aware of the limitations of the approach. With these limitations in mind, this manuscript clearly demonstrates mutation specific signaling in an in vivo model, demonstrating yet again that all KRAS mutations are not created equal.

We thank the reviewer for both their encouraging review and thoughtful suggestions.

1. Overall, I note very few problems with this text. However, Figure 1 C needs additional clarification. The methods state that the abcam RAS GTPase ELISA kit was used. The product manual states that this kit detects activated H and KRAS human and HRAS rodent samples, if employed per the manufacturer's kit. Was a different primary used? According to the methods, the FLAG-tagged constructs were Kras, which implies a murine system. Given that the empty vector increases with increasing protein extract, it is likely that this system is detecting the endogenous human GTPases and not the ectopically expressed murine construct. While the supplemental data do agree with the expected results based on the published literature, this figure could possibly be improved.

As requested, we clarify in the Materials and methods the specific antibody used (which was not a different antibody than that provided by this kit). As requested, we clarify in text that the ectopic Kras expressed in HEK-HT cells was derived from a FLAG-tagged murine Kras cDNA. As requested, we note here that the manufacturer indicated that this kit was validated with full-length human Kras/Hras and murine Hras, but not murine Kras. Given this, we tested and found that this kit captured FLAG-tagged murine Kras ectopically expressed in HEK-HT cells, while no signal was present in the vector control cells. This finding is consistent with the kit detecting the over-expressed murine Kras protein in the GTP-bound state.

2. In figure 1, supplement 4 panel B, the KrasnatG12D/+ bar does not match the legend.

As requested, we first thank the reviewer for catching this oversight and have corrected the figure legend.

3. In Figure 5, the gene name for the PI3Ks are incorrect. The correct gene names for p110 isoforms are PIK3CA, PIK3CD and PIK3CG.

As requested, we first thank the reviewer for catching this and have corrected the nomenclature in this figure.

4. While not necessary but could improve Figure 5, could testing which inhibitor class on overall survival in the common codon mutants add improve survival? If done in the Q61 mutant, it could be quick to validate the specificity between the classes of inhibitors.

This would be a natural progression of this study. However, since the current study focuses on tumor initiation, we propose to address the concept of targeting specific mutants in mice would be best suited for future exploration.

5. Is it possible that expression levels may alter mutant-specific (qualitative) signaling as well? Would this appear as quantitative?

As requested, we include this possibility in the text.

More specifically, what was the degree of overlap within the RAS mutants when changing from natural to rare codons? Were all of the hits observed in the natural codon mice detected in the common codon mice? This might be a more useful figure than the STRING analysis.

As requested, we compared NatG12D to ComG12D, and found all seven differentially expressed genes in the NatG12D with respect to the WT allele were also differentially expressed in the ComG12D transcriptome with very similar log2FC and adjusted p-values. Five of these genes were differentially expressed in all alleles. Comparing NatQ61R to ComQ61R, we found 12 out of 15 differentially expressed genes in the NatQ61R allele were also differentially expressed in the ComQ61R transcriptome. These new data are now provided in the form of a Venn diagram in Figure 3—figure supplement 6.

6. On page 4, it is mentioned that Q61R has slower GTP exchange and hydrolysis rates than G12D. It is worth pointing out that the Burd et al., reference is NRAS while the Rabara et al., reference compared GAP-mediated hydrolysis. However, G12D is interesting because its GTP exchange rate has been reported to be elevated (PMID 8955068 and PMID 24224811), being faster than G13D in biochemical studies, yet this is never discussed. Thus, the author is not wrong in their statement, but additional references and clarification could be useful.

As requested, we indicate that the Burd et al., reference refers to NRAS and that the Rabara et al., reference compared GAP-mediated hydrolysis. Since the focus of this study is between G12D and Q61R mutations we elected not to bring in comparisons with G13 mutations. However, if we have missed the point of the reviewer, we are glad to make further changes to this section.

7. On page 6, the authors wrote "special control of gene expression" although I believe they meant "special control".

As requested, we first thank the reviewer for catching this and change this to spatial.

8. On page 9, first full paragraph. Discussing RNAseq data as a direct proxy for protein signaling is not entirely valid. RNA levels, which do indicate transcript levels, do not necessarily mean that signaling in those pathways is elevated.

As requested, we rephrased the rationale to use RNAseq as a way to capture changes to cells at the level of the transcriptome.

9. On page 11, first paragraph. "Of note, median survival was significantly decreased in mice with either of the Krascom alleles or with either of the Q61R mutants." This sentence is not entirely clear. Median survival was decreased compared to what? And didn't the Q61Rnat and G12Dnat have similar survival rates? Please clarify.

As requested, we revised the text to indicate that the median survival of mice in which either of the LSL-Krascom allele were activated was lower than when either of the LSL-Krasnat alleles were activated. As requested, we revised the text to indicate that the median survival of mice in which the LSL-KrascomQ61R allele was activated was lower than when the LSL-KrascomG12D allele was activated. Finally, and as requested, we revised the text to indicate that the median survival of mice in which the LSL-KrasnatQ61R allele was activated trended, but did not reach significance as being lower than when the LSL-KrasnatG12D allele was activated.

https://doi.org/10.7554/eLife.75715.sa2

Article and author information

Author details

  1. Özgün Le Roux

    Department of Pharmacology & Cancer Biology, Duke University Medical Center, Durham, United States
    Contribution
    Conceptualization, Resources, Data curation, Formal analysis, Validation, Investigation, Visualization, Methodology, Writing – original draft, Writing – review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-9885-6872
  2. Nicole LK Pershing

    Department of Pharmacology & Cancer Biology, Duke University Medical Center, Durham, United States
    Present address
    Departments of Pathology and Pediatric Infectious Disease, University of Utah, Salt Lake City, United States
    Contribution
    Conceptualization, Writing – review and editing
    Competing interests
    No competing interests declared
  3. Erin Kaltenbrun

    Department of Pharmacology & Cancer Biology, Duke University Medical Center, Durham, United States
    Present address
    Nighthawk Biosciences Inc, Morrisville, United States
    Contribution
    Conceptualization, Resources, Writing – review and editing
    Competing interests
    No competing interests declared
  4. Nicole J Newman

    Department of Pharmacology & Cancer Biology, Duke University Medical Center, Durham, United States
    Present address
    West Virginia School of Osteopathic Medicine, Lewisburg, United States
    Contribution
    Investigation, Writing – review and editing
    Competing interests
    No competing interests declared
  5. Jeffrey I Everitt

    Department of Pathology, Duke University Medical Center, Durham, United States
    Contribution
    Data curation, Methodology, Writing – review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-0273-6284
  6. Elisa Baldelli

    Center for Applied Proteomics and Molecular Medicine, School of Systems Biology, George Mason University, Manassas, United States
    Contribution
    Data curation, Formal analysis, Methodology
    Competing interests
    EB, MP, and EP are inventors on US Government and University assigned patents and patent applications that cover aspects of the technologies discussed such as the Reverse Phase Protein Microarrays. As inventors, they are entitled to receive royalties as provided by US Law and George Mason University policy. MP and EP receive royalties from and are consultants of TheraLink Technologies, Inc EP is shareholders and consultants of TheraLink Technologies, Inc EP is shareholder and consultant of Perthera, Inc
  7. Mariaelena Pierobon

    Center for Applied Proteomics and Molecular Medicine, School of Systems Biology, George Mason University, Manassas, United States
    Contribution
    Data curation, Formal analysis, Methodology
    Competing interests
    EB, MP, and EP are inventors on US Government and University assigned patents and patent applications that cover aspects of the technologies discussed such as the Reverse Phase Protein Microarrays. As inventors, they are entitled to receive royalties as provided by US Law and George Mason University policy. MP and EP receive royalties from and are consultants of TheraLink Technologies, Inc EP is shareholders and consultants of TheraLink Technologies, Inc EP is shareholder and consultant of Perthera, Inc
  8. Emanuel F Petricoin

    Center for Applied Proteomics and Molecular Medicine, School of Systems Biology, George Mason University, Manassas, United States
    Contribution
    Resources, Supervision
    Competing interests
    EB, MP, and EP are inventors on US Government and University assigned patents and patent applications that cover aspects of the technologies discussed such as the Reverse Phase Protein Microarrays. As inventors, they are entitled to receive royalties as provided by US Law and George Mason University policy. MP and EP receive royalties from and are consultants of TheraLink Technologies, Inc EP is shareholders and consultants of TheraLink Technologies, Inc EP is shareholder and consultant of Perthera, Inc
  9. Christopher M Counter

    Department of Pharmacology & Cancer Biology, Duke University Medical Center, Durham, United States
    Contribution
    Conceptualization, Resources, Supervision, Funding acquisition, Visualization, Writing – original draft, Project administration, Writing – review and editing
    For correspondence
    chris.counter@duke.edu
    Competing interests
    co-founder of the company Merlon Inc
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-0748-3079

Funding

National Cancer Institute (R01CA94184)

  • Christopher M Counter

National Cancer Institute (R01CA269272)

  • Christopher M Counter

National Cancer Institute (P01CA203657)

  • Christopher M Counter

Duke Cancer Institute

  • Christopher M Counter

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

We thank members of the Counter Laboratory for technical assistance, David Kirsch, James Alvarez, Mark Onaitis, Seth Zimmerman, Channing Der, and Aaron Hobbs for advice, Cheryl Block, Gary Kucera, and the Duke Cancer Institute Transgenic Mouse Facility for the generation of mice, and Nicolas Devos, David Corcoran, Wei Chen, and Duke Center for Genomic and Computational Biology for their assistance in RNA-Seq. All mouse and organ pictures were created with Biorender.com. This work was supported by the National Cancer Institute (R01CA94184, R01CA269272, and P01CA203657 to CMC) and the Shared Resources of the Duke Cancer Institute (P30CA014236). The authors declare no competing interests.

Ethics

All mouse care and experiments were performed in accordance with a protocol approved by the Institutional Animal Care and Use Committee (IACUC) of Duke University (Protocol no. A195-19-09).

Senior and Reviewing Editor

  1. Erica A Golemis, Fox Chase Cancer Center, United States

Reviewers

  1. Jonathan Chernoff, Fox Chase Cancer Center, United States
  2. Aaron Hobbs

Publication history

  1. Received: November 20, 2021
  2. Preprint posted: December 10, 2021 (view preprint)
  3. Accepted: August 2, 2022
  4. Version of Record published: September 7, 2022 (version 1)

Copyright

© 2022, Le Roux et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 585
    Page views
  • 127
    Downloads
  • 0
    Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Open citations (links to open the citations from this article in various online reference manager services)

Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)

  1. Özgün Le Roux
  2. Nicole LK Pershing
  3. Erin Kaltenbrun
  4. Nicole J Newman
  5. Jeffrey I Everitt
  6. Elisa Baldelli
  7. Mariaelena Pierobon
  8. Emanuel F Petricoin
  9. Christopher M Counter
(2022)
Genetically manipulating endogenous Kras levels and oncogenic mutations in vivo influences tissue patterning of murine tumorigenesis
eLife 11:e75715.
https://doi.org/10.7554/eLife.75715
  1. Further reading

Further reading

    1. Biochemistry and Chemical Biology
    2. Cancer Biology
    Stefania Monterisi, Johanna Michl ... Pawel Swietach
    Research Article Updated

    Growth of cancer cells in vitro can be attenuated by genetically inactivating selected metabolic pathways. However, loss-of-function mutations in metabolic pathways are not negatively selected in human cancers, indicating that these genes are not essential in vivo. We hypothesize that spontaneous mutations in ‘metabolic genes’ will not necessarily produce functional defects because mutation-bearing cells may be rescued by metabolite exchange with neighboring wild-type cells via gap junctions. Using fluorescent substances to probe intercellular diffusion, we show that colorectal cancer (CRC) cells are coupled by gap junctions assembled from connexins, particularly Cx26. Cells with genetically inactivated components of pH regulation (SLC9A1), glycolysis (ALDOA), or mitochondrial respiration (NDUFS1) could be rescued through access to functional proteins in co-cultured wild-type cells. The effect of diffusive coupling was also observed in co-culture xenografts. Rescue was largely dependent on solute exchange via Cx26 channels, a uniformly and constitutively expressed isoform in CRCs. Due to diffusive coupling, the emergent phenotype is less heterogenous than its genotype, and thus an individual cell should not be considered as the unit under selection, at least for metabolite-handling processes. Our findings can explain why certain loss-of-function mutations in genes ascribed as ‘essential’ do not influence the growth of human cancers.

    1. Cancer Biology
    2. Computational and Systems Biology
    Erika K Ramos, Chia-Feng Tsai ... Huiping Liu
    Research Article

    Tumor-initiating cells with reprogramming plasticity or stem-progenitor cell properties (stemness) are thought to be essential for cancer development and metastatic regeneration in many cancers; however, elucidation of the underlying molecular network and pathways remains demanding. Combining machine learning and experimental investigation, here we report CD81, a tetraspanin transmembrane protein known to be enriched in extracellular vesicles (EVs), as a newly identified driver of breast cancer stemness and metastasis. Using protein structure modeling and interface prediction-guided mutagenesis, we demonstrate that membrane CD81 interacts with CD44 through their extracellular regions in promoting tumor cell cluster formation and lung metastasis of triple negative breast cancer (TNBC) in human and mouse models. In-depth global and phosphoproteomic analyses of tumor cells deficient with CD81 or CD44 unveils endocytosis-related pathway alterations, leading to further identification of a quality-keeping role of CD44 and CD81 in EV secretion as well as in EV-associated stemness-promoting function. CD81 is co-expressed along with CD44 in human circulating tumor cells (CTCs) and enriched in clustered CTCs that promote cancer stemness and metastasis, supporting the clinical significance of CD81 in association with patient outcomes. Our study highlights machine learning as a powerful tool in facilitating the molecular understanding of new molecular targets in regulating stemness and metastasis of TNBC.