The amino acid transporter SLC7A5 drives progression of PI3K-mutant intestinal cancer models and enhances response to MAPK-targeted therapy

doi:10.21203/rs.3.rs-5002395/v1

Download PDF

Article

The amino acid transporter SLC7A5 drives progression of PI3K-mutant intestinal cancer models and enhances response to MAPK-targeted therapy

https://doi.org/10.21203/rs.3.rs-5002395/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted

You are reading this latest preprint version

Colorectal cancer (CRC) is a complex disease with key oncogenic pathways, including Wnt, MAPK, and PI3K, co-operating to drive tumour initiation and progression. Loss-of-function mutations in the Wnt-pathway inhibitor APC are the most prominent genetic alterations and are commonly seen as the tumour-initiating event. Here, we have used genetically engineered mouse models to introduce pathway-activating mutations of PI3K (Pik3ca, Pten) and MAPK (Kras) signalling to the mouse gut. Comprehensive characterization of these models reveals pathway-dependent cooperation, leading to marked allele dose-dependent acceleration of tumour formation, underpinned by MYC-driven transcriptional reprogramming and alterations in downstream signalling pathways. We find the amino acid transporter SLC7A5 to be highly upregulated upon activation of PI3K signalling. In human CRCs, SLC7A5expression correlates with the newly defined PDS1 pathway-derived subtype and highly proliferative tumours. Genetic depletion of Slc7a5 in the newly developed PI3K-hyperactive models drastically extends survival by delaying tumour formation, even in aggressive Kras/Pik3caco-mutant mice. Finally, Slc7a5 gene deletion sensitizes these models to targeted MAPK inhibition. Taken together, SLC7A5 drives progression of PI3K-mutant CRCs and is an attractive (co-)target for mutation-specific inhibitors.

Biological sciences/Cancer/Cancer therapy/Targeted therapies

Biological sciences/Cancer/Gastrointestinal cancer

Loss of the tumour suppressor APC is the most common driver mutation in colorectal cancer (CRC) and considered the tumour-initiating event in many cases¹. Indeed, germline truncating mutation of APC is the cause of the hereditary polyposis syndrome familial adenomatous polyposis and loss of Apc is sufficient to drive intestinal adenoma formation in preclinical mouse models in vivo². Biallelic APC mutation results in dysregulation of Wnt signalling through functional loss of the β-catenin destruction complex, which in turn drives activation of an aberrant Wnt transcriptional programme, uncontrolled growth, and proliferation³.

Human CRCs are genetically complex with aberrant Wnt-pathway activation typically compounded by additional mutations in key oncogenic signalling pathways, including loss of functional TP53, disruption of TGFβ signalling through SMAD4 or TGFBR2 mutations, and alterations in the MAPK (KRAS, BRAF) and PI3K/MTOR (PIK3CA, PTEN) signalling pathways^4,5. More specifically, ca 20% of CRCs harbour activating mutations in the PIK3CA gene, encoding the p110α catalytic subunit of PI3Kα. This in turn elicits downstream AKT and mTOR activation, driving cell proliferation, survival, stemness, and metabolic reprogramming^6,7. Hot-spot mutations in PIK3CA are present in exon 9, which encodes the helical domain (E542 and E545), and in exon 20, which encodes the catalytic domain (H1047); mutations in either of these regions lead to increased PI3K signalling⁶. A subset of PIK3CA mutant tumours harbour multiple mutations, either on the same allele (in cis), or on the second copy of the gene (in trans). This was shown to be functional, affecting oncogenicity and the patient’s response to targeted therapy^8,9. Alternatively, aberrant activation of PI3K signalling is elicited through loss-of-function mutations in the gene encoding PTEN, a dual-specific protein and lipid phosphatase whose primary role is to counteract PI3K signalling by dephosphorylating its product, the lipid second messenger phosphatidylinositol-3,4,5-trisphosphate (PIP₃), to generate phosphatidylinositol-4,5-diphosphate (PIP₂). Consequently, PTEN deficiency leads to the accumulation of PIP_3, which in turn drives cell-membrane recruitment and activation of the pro-survival kinase AKT¹⁰. Indeed, PTEN loss-of-function mutations are found in approximately 8% of CRC patients (Fig. 1A), although this level of PTEN alteration is likely an underestimate as PTEN expression is additionally regulated by post-transcriptional mechanisms^11,12. As such, PTEN functions as a tumour suppressor, and multiple components of the PI3K and downstream AKT/mTOR signalling pathways are frequently mutated across cancer types and are considered key pan-cancer oncogenic drivers, presenting actionable therapeutic vulnerabilities and paving the way for targeted therapies. Nonetheless, it is not clear whether hyperactive PI3K signalling impacts disease outcome in CRC^13–19.

Given these findings, numerous PI3K inhibitors of varying specificity have been developed, have shown promise in preclinical studies, and have been deployed clinically in several disease settings. For example, in the advanced breast cancer setting, patients harbouring PIK3CA-pathway mutated, ER+/HER2- tumours derived benefit from the combination of fulvestrant with targeted inhibition of PI3Kα²⁰ or AKT ²¹. However, in CRC clinical trials, no clear benefit has been observed, e.g. the PI3Kα-specific inhibitor BYL719 (alpelisib) did not significantly improve overall survival when combined with a BRAF/EGFR-targeted therapy regime, while slightly increasing progression free survival at the cost of more frequent adverse effects^22,23. Furthermore, combination strategies, for example with MAPK-pathway inhibitors, have proven difficult to administer due to systemic toxicities, although mutation-specific inhibitors may be able to overcome this issue.

Reactivation of the key downstream PI3K/AKT target, mTORC1, underpins the emergence of resistance in response to targeted upstream inhibition, contributing to therapeutic failure²⁴. mTORC1 integrates multiple inputs from the PI3K, MAPK, and AMPK signalling pathways and senses nutrient availability²⁵ through amino acid sensors that directly activate mTORC1 by detecting lysosomal availability of e.g. leucine and arginine^26–29. Cells can, therefore, utilize amino acid transporters, such as SLC7A5, to balance the amino acid pool in favour of essential amino acids, such as leucine, to fuel mTORC1 signalling³⁰. mTORC1, in turn, regulates multiple anabolic cellular pathways, including lipid synthesis, nucleotide metabolism, autophagy, and protein synthesis²⁵, linking cell growth and proliferation to nutrient availability. While we have previously shown that targeted inhibition of mTORC1 using Rapamycin is effective in simple Wnt-driven intestinal cancer models, compounding Kras mutation drives resistance^31,32. Despite this, we have found that combination strategies, such as targeting translation elongation through MNK1/2 inhibition, or restricting nutrient availability through Slc7a5 deletion, can overcome this resistance to mTORC1 inhibition^33,34.

Mouse models have been invaluable for elucidating disease mechanisms and the pre-clinical testing of promising therapeutic approaches. In this study, we present a set of PI3K-hyperactive preclinical genetically engineered mouse models (GEMMs) of CRC, demonstrating the interplay of mutations within the PI3K-, Wnt- and MAPK-signalling pathways. Further, we show that PI3K hyperactivation drives expression of the amino acid transporter Slc7a5, genetic depletion of which significantly extends survival, even in complex Wnt/MAPK/PI3K-mutant models. Finally, Slc7a5 knockout (KO) can sensitize these models to targeted MAPK inhibition in vivo.

Activated PI3K signalling potentiates intestinal tumorigenesis in vivo and models respond to targeted PI3Kα inhibition

To study oncogenic PI3K signalling in the mouse gut, we crossed a Cre-inducible, kinase-activating PIK3CA mutant allele (Pik3ca^{fl − H1047R−T2A−EYFP−NLS}, henceforth Pik3ca^H1047R) (Fig. 1B)³⁵ to Villin-CreERT2 mice. This allowed for induction of recombination in the intestinal epithelium upon intraperitoneal (IP) administration of tamoxifen. To investigate immediate effects on intestinal homeostasis, we induced Villin-CreERT2 Pik3ca^{H1047R/H1047R} mice and sampled them five days post-induction. Gross intestinal homeostasis was not perturbed, but crypts were enlarged, contained more proliferative cells, and Olfm4 positive area was increased. We found Lysozyme-positive cells throughout crypt-villus axes, which morphologically resembled goblet cells (black arrows) (Supplementary Fig. 1A).

We then crossed the Pik3ca^H1047R allele to a well-characterised model of intestinal tumorigenesis, driven by intestinal epithelial specific (Villin-CreERT2) loss of Apc (Apc^fl/wt). This approach generated cohorts of Villin-CreERT2 Apc^fl/wt Pik3ca^H1047R/wt (Apc^fl/wt Pik3ca^H1047R/wt) and Villin-CreERT2 Apc^fl/wt Pik3ca^{H1047R/H1047R} (Apc^fl/wt Pik3ca^{H1047R/H1047R}) mice. Animals were induced and aged to clinical endpoint. RNA scope in situ hybridisation for EYFP transcripts was conducted to confirm expression of the Pik3ca^H1047R allele in tumours (Supplementary Fig. 1B). Heterozygous activation of PI3K, did not significantly impact overall survival or intestinal tumour number in these mice, while homozygosity for Pik3ca^H1047R significantly accelerated tumorigenesis and decreased intestinal tumour-free survival, demonstrating allele dose-dependent effects of activating Pik3ca mutation in our APC-deficient models of intestinal tumorigenesis (Fig. 1C; Supplementary Fig. 1C).

Production of PIP₃ and hence activation of PI3K signalling is counteracted by the lipid phosphatase and canonical tumour suppressor PTEN. We asked if targeting PTEN would result in the same pro-tumorigenic phenotype and interbred our conditional APC-loss driven model with mice carrying a Pten^fl allele (Fig. 1B), generating Villin-CreERT2 Apc^fl/wt Pten^fl/wt (Apc^fl/wt Pten^fl/wt) and Villin-CreERT2 Apc^fl/wt Pten^fl/fl (Apc^fl/wt Pten^fl/fl) mice. Depletion of Pten resulted in a substantial increase in tumour burden and a consequent reduction in the lifespan of these mice in an allele dose-dependent manner (Fig. 1D; Supplementary Fig. 1D). Activation of PI3K signalling, following deletion of Pten or Pik3ca mutation, resulted in robust phosphorylation of downstream biomarkers, particularly targets of mTORC1/mTORC2. Notably, phosphorylation of the downstream mTOR effector RPS6 (Ser 235/236) was induced in all cases, with enhanced pS6 staining intensity in Apc^fl/wt Pik3ca^{H1047R/H1047R} and Apc^fl/wt Pten^fl/fl lesions (Fig. 1E). Phosphorylation of AKT (Ser 473) was also markedly increased in tumours arising in Apc^fl/wt Pik3ca^{H1047R/H1047R} and Apc^fl/wt Pten^fl/fl mice. Apc^fl/wt Pten^fl/fl lesions exhibited higher staining intensity than Apc^fl/wt Pik3ca^{H1047R/H1047R} counterparts (Fig. 1E), consistent with different degrees of AKT activation downstream of distinct PI3K signalling inputs³⁶.

PIK3CA and PTEN mutations frequently coincide within individual tumours¹¹. We therefore sought to exacerbate PI3K activation through combined mutation of Pik3ca and deletion of Pten in models dependent on Apc-loss in vivo. We generated an allelic series of mice based upon the Villin-CreERT2 Apc^fl/wt model, producing cohorts of Villin-CreERT2 Apc^fl/wt Pten^fl/wt Pik3ca^H1047R/wt (Apc^fl/wt Pten^fl/wt Pik3ca^H1047R/wt), Villin-CreERT2 Apc^fl/wt Pten^fl/wt Pik3ca^{H1047R/H1047R} (Apc^fl/wt Pten^fl/wt Pik3ca^{H1047R/H1047R}), Villin-CreERT2 Apc^fl/wt Pten^fl/fl Pik3ca^H1047R/wt (Apc^fl/wt Pten^fl/fl Pik3ca^H1047R/wt) and Villin-CreERT2 Apc^fl/wt Pten^fl/fl Pik3ca^{H1047R/H1047R} (Apc^fl/wt Pten^fl/fl Pik3ca^{H1047R/H1047R}) mice. In each case, the addition of a compounding PI3K-activating mutation resulted in a marked allele dose-dependent acceleration of tumorigenesis, leading to earlier clinical endpoints (Fig. 1F, Supplementary Fig. 1E, 2A, B).

To corroborate our data on co-operation between Pten loss and Pik3ca mutation, we used the Pik3ca^{H1047R + neo}; Flp-ER^T2 (henceforth Flp^ERPik3ca^H1047R) model³⁷ and crossed it to Villin-CreERT2 Apc^fl/wt Pten^fl/fl mice, generating cohorts of Villin-CreERT2 Apc^fl/wt Pten^fl/wt Flp^ERPik3ca^H1047R/wt (Apc^fl/wt Pten^fl/wt Flp^ERPik3ca^H1047R/wt) and Villin-CreERT2 Apc^fl/wt Pten^fl/fl Flp^ERPik3ca^H1047R/wt (Apc^fl/wt Pten^fl/fl Flp^ERPik3ca^H1047R/wt) mice. The median intestinal tumour-free survival of Apc^fl/wt Pten^fl/wt Flp^ERPik3ca^H1047R/wt mice was drastically reduced compared with Apc^fl/wt Pten^fl/wt counterparts (121 days and 225 days post induction, respectively) (Supplementary Fig. 2C, D). Apc^fl/wt Pten^fl/fl Flp^ERPik3ca^H1047R/wt mice succumbed faster to intestinal tumours than Villin-CreERT2 Apc^fl/wt Pten^fl/fl counterparts (with median survival times of 32 days and 58.5 days post induction, respectively) (Supplementary Fig. 2E, F), suggesting that the presence of a Pik3ca^H1047R mutant allele accelerates tumour progression in a setting of partial or complete loss of PTEN. The longer latencies to endpoint observed in these cohorts are likely due to the outbred genetic background of these mice. These findings further highlight the strong co-operation between loss of PTEN and Pik3ca mutation.

We next asked if Villin-CreERT2 mediated Pten depletion could cooperate with Pik3ca mutation in the absence of predisposition to Apc-loss. Targeted deletion of Pten in the intestine (Villin-CreERT2 Pten^fl/fl) resulted in adenoma formation, with mice developing a small number of early lesions and dysplastic areas. Compounding activating mutation of Pik3ca, significantly accelerated this phenotype. Pten^fl/fl Pik3ca^H1047R/wt had to be sampled at a similar time post induction as their Pik3ca wild type counterparts (median 595 vs 485 days post-induction), while Pten^fl/fl Pik3ca^{H1047R/H1047R} reached clinical endpoint significantly earlier (median 595 vs 344 days post-induction). At endpoint, mice presented with similar intestinal tumour number and burden (Supplementary Figs. 3A, B). We also generated ageing cohorts of Villin-CreERT2 Pten^fl/fl (Pten^fl/fl) and Villin-CreERT2 Pten^fl/fl Flp^ERPik3ca^H1047R (Pten^fl/fl Flp^ERPik3ca^H1047R) mice and found that Pten loss and Pik3ca mutation co-operate to drive intestinal cancer, leading to clinical endpoint and increased tumour number (Supplementary Fig. 3C, D).

Since amplification of PI3K signalling in the gut led to allele dose-dependent phenotypes, we reasoned that this might also translate into a genotype-specific sensitivity to targeted PI3Kα inhibition. To address this, we treated the newly generated Apc/Pten/Pik3ca co-mutant mice with the PI3K α/δ inhibitor AZD8835 from day 5 post induction to clinical endpoint ³⁸. This treatment significantly extended the median intestinal tumour-free survival of Apc^fl/wt Pten^fl/fl mice from 44 to 59 days post induction (Fig. 1G). Treatment benefit scaled with the extent of PI3K hyperactivation: whilst both hetero- and homozygous Pik3ca-mutant mice gained a survival benefit upon drug treatment, homozygous mice derived the greatest relative benefit (with median survival increasing from 28 to 47 days post induction and 18 to 37 days post induction, respectively) (Figs. 1H, I and Supplementary Fig. 4A). Following AZD8835 treatment, the number of small intestinal tumours at clinical endpoint was similar in Apc^fl/wt Pten^fl/fl mice, slightly reduced in Apc^fl/wt Pten^fl/fl Pik3ca^H1047R/wt mice, and significantly reduced in Apc^fl/wt Pten^fl/wt Pik3ca^{H1047R/H0147R} mice, compared with vehicle-treated counterparts (Supplementary Fig. 4B, C, D). Biomarkers of PI3K and mTOR signalling were activated in treated tumours at clinical endpoint, as shown by IHC staining for the phosphorylated forms of key downstream effectors of this pathway, including AKT, S6, and 4E-BP1 (Supplementary Fig. 4E), indicating eventual treatment resistance. PI3Kβ inhibition, using AZD8186 did not delay tumour formation in Apc^fl/wt Pten^fl/wt (Supplementary Fig. 4F, G), further suggesting that PI3Kα signalling is the driving force of tumorigenesis in our models.

These data demonstrate that the acceleration of intestinal tumorigenesis, driven by oncogenic activation of PI3K signalling, can in turn be reduced by pharmacological inhibition of the pathway. While the modest efficacy of this monotherapy is encouraging, targeted therapeutics as single agents have been largely unsuccessful in the clinic, underscoring the need for combination therapy or discovery of novel treatment approaches.

Slc7a5 is upregulated in PI3K-hyperactive mouse tumours and highly proliferative human CRC

Given that PI3K pathway hyperactivation accelerated intestinal tumorigenesis driven by Apc loss, we used a well-characterised acute model of intestinal hyperproliferation, driven by targeted homozygous deletion of Apc, to better understand the direct consequences of these mutations in the intestinal epithelium³⁹. We generated an allelic series of Villin-CreERT2 Apc^fl/fl (Apc^fl/fl), Villin-CreERT2 Apc^fl/fl Pik3ca^H1047R/wt (Apc^fl/fl Pik3ca^H1047R/wt) and Villin-CreERT2 Apc^fl/fl Pik3ca^{H1047R/H1047R} (Apc^fl/fl Pik3ca^{H1047R/H1047R}) mice and used these to assess the acute impact of each mutation in vivo. Bulk transcriptomic analysis of small intestinal tissue from each model identified the amino acid transporter Slc7a5 as the most significantly upregulated transcript in the small intestine of Apc^fl/fl Pik3ca^{H1047R/H0147R} mice (Fig. 2A). Moreover, Slc7a5 transcript expression was elevated in a Pik3ca^H1047R allele dose-dependent manner (Supplementary Fig. 5A). Increased Slc7a5 expression in Pik3ca-mutant tumours was validated with in situ hybridization (RNA scope) (Fig. 2B, C).

Based on bulk and/or single-cell transcriptomics, CRCs have been stratified into four consensus molecular subtypes (CMS)⁴⁰, five CRC cell-intrinsic subtypes (CRIS)⁴¹, and two epithelial cell-intrinsic consensus molecular subtypes (iCMS)⁴². More recently, we reported a refined pathway-derived subtyping (PDS) approach, which identified three distinct subtypes of epithelial-rich colorectal tumours⁴³. PDS1 tumours are highly proliferative, express MYC transcriptional signatures, and exhibit the best prognosis. PDS2 tumours are immune- and stroma-rich with an intermediate prognosis. PDS3 tumours have the worst prognosis and are characterized by activation of the polycomb repressive complex (PRC) and are slow cycling. To investigate the clinical relevance of SLC7A5 expression, we analysed the Marisa clinical cohort comprising 566 mostly stage II–III CRC patients with long-term follow-up⁴⁴. SLC7A5 expression was highly enriched in PDS1 tumours, compared to other subtypes (Fig. 2D). In line with this, rather than associating with specific genetic mutations such as KRAS, SLC7A5 expression correlated with key features of PDS1: High proliferation index⁴⁵ and Ki67 (Fig. 2E, Supplementary Fig. 5B); low stem maturation index⁴⁶ (Fig. 2F); high MYC transcriptional signature⁴⁷ (Fig. 2G); low PRC transcriptional signature⁴⁷ (Fig. 2H).

To validate these transcriptional data from publicly available datasets, we stained a tissue microarray (TMA) containing CRC cores from surgical resections from 787 stage I–III patients for SLC7A5. We classified the epithelial tissue component and generated an analysis algorithm to calculate weighted H-scores for each core. We found a wide range of SLC7A5 staining intensity, ranging from negative to highly positive cores (Fig. 2I). While there was no strong association with cancer-specific survival outcome (Supplementary Fig. 5C, D), SLC7A5 expression correlated with highly proliferative tumours, classified as such based on the clinical parameter of > 30% Ki67-positive cells (Fig. 2J) and elevated cyclin D expression (Fig. 2K).

Slc7a5 KO prolongs survival in PI3K-hyperactive mouse models

Given the upregulation of Slc7a5 in PI3K-driven mouse intestinal tumour models and highly proliferative human CRC, we sought to establish whether genetic deletion of this amino acid transporter would impact tumour initiation and progression of PI3K-hyperactive models in vivo. We have previously shown that Slc7a5 deletion does not extend survival in Lgr5^CreER; Apc^fl/fl mice, indicating that purely Wnt driven intestinal tumours do not rely on SLC7A5⁴⁸. To investigate the role of Slc7a5 in a PI3K-hyperactive setting, we used a previously reported conditional knockout allele for Slc7a5⁴⁹ to generate Villin-CreERT2 Apc^fl/wt Pik3ca^{H1047R/H1047R} Slc7a5^fl/fl (Apc^fl/wt Pik3ca^{H1047R/H1047R} Slc7a5^fl/fl) and Villin-CreERT2 Apc^fl/wt Pten^fl/fl Slc7a5^fl/fl (Apc^fl/wt Pten^fl/fl Slc7a5^fl/fl) mice. We induced genetic recombination by IP administration of tamoxifen and aged mice to clinical endpoint. Slc7a5 deletion led to a substantial survival benefit in Apc^fl/wt Pik3ca^{H1047R/H1047R} mice, extending the median lifespan from 123 days to 195 days post induction (Fig. 3A). Two mice were censored at the end of the study at 412 days post induction without clinical signs of intestinal disease. Similarly, the median survival of Apc^fl/wt Pten^fl/fl Slc7a5^fl/fl mice was extended from 37.5 to 72 days (Fig. 3B). In both models, tumour numbers at endpoint were similar irrespective of Slc7a5 loss, despite being taken at a later timepoint (Supplementary Fig. 6A, B). Taken together, these data indicate that SLC7A5 is required for the effective outgrowth of PI3K-hyperactive tumours in vivo. Furthermore, IHC analysis of PI3K and mTOR signalling markers in endpoint tumours indicated that both pathways are active, suggesting that PI3K signalling can eventually activate mTOR, despite the absence of SLC7A5 (Supplementary Fig. 6C).

To investigate the effects of Slc7a5 loss on the hyperproliferative crypt phenotype, induced upon Apc loss³⁹, we crossed mice carrying the Slc7a5^fl allele to Apc^fl/fl; Pik3ca^{H1047R/H1047R} mice. Crypt proliferation was not altered upon Slc7a5 deletion in Apc^fl/fl; Pik3ca^{H1047R/H1047R} mice (Supplementary Fig. 7A). We performed RNA sequencing of small intestinal tissue from Apc^fl/fl; Pik3ca^{H1047R/H1047R} mice in the presence or absence of Slc7a5. Gene Set Enrichment Analysis (GSEA) using the HALLMARK gene sets did not identify significantly differentially enriched pathways (FDR ≤ 0.05). Despite this, several genes associated with cellular adaptation to stress were significantly upregulated following Slc7a5 deletion. This could be Atf4 dependent, as well-documented Atf4 target genes were upregulated, including Chac1⁵⁰, Gpt2⁵¹, Mthfd2⁵², Slc7a11⁵³, Pycr1⁵⁴, Asns⁵⁵, Pck2⁵⁶, Phgdh⁵⁷, Psat1⁵⁷, Psph⁵⁷, and Shmt2⁵⁷ (Fig. 3C). Furthermore, Slc7a3 was upregulated, which was previously shown to be upregulated in response to low arginine and lysine availability⁵⁸ (Fig. 3C). Taken together, this suggests that cells may be mounting an amino acid deprivation response upon acute genetic depletion of Slc7a5, as many of these genes are functionally involved in de-novo amino acid synthesis and mitochondrial one-carbon metabolism. We performed targeted liquid chromatography - mass spectrometry analysis (LC-MS) on extracts from small intestinal and colonic tissue of Apc^fl/fl, Apc^fl/fl Pik3ca^{H1047R/H1047R}, and Apc^fl/fl Pik3ca^{H1047R/H1047R} Slc7a5^fl/fl mice, sampled four days post induction. There was an overall trend for increased abundance of amino acids in Apc^fl/fl Pik3ca^{H1047R/H1047R} Slc7a5^fl/fl small intestinal tissue compared to compared to Apc^fl/fl Pik3ca^{H1047R/H1047R} controls (Fig. 3D). Methionine was the only altered known SLC7A5 substrate, while multiple other amino acids were more abundant in Slc7a5 deplete tissue (FDR < 5%). Phgdh, Psat1, Psph, and Shmt2 are essential enzymes of Serine/Glycine synthesis. All four were upregulated transcriptionally (Fig. 3E), and both Serine and Glycine abundance was increased upon Slc7a5 deletion (Fig. 3F). In addition, Proline levels were increased, which was in line with upregulation of Pycr1. There were no significant changes in colonic tissue upon Slc7a5 deletion, or between Apc^fl/fl and Apc^fl/fl Pik3ca^{H1047R/H1047R} small intestinal or colonic tissue (Supplementary Fig. 7B, C, D). Taken together, this indicates that genetic deletion of Slc7a5 triggers an alteration in amino acid pools beyond it’s known substrates, which is likely part of a nutrient stress response.

KRAS mutation synergizes with PI3K signalling to further accelerate intestinal cancer development and drives resistance to PI3Kα inhibition

Mutations of the proto-oncogene KRAS occur in approximately 45% of CRC patients (Fig. 1A), and significantly co-occur with activating PIK3CA mutations (Supplementary Fig. 8A)⁵. At the molecular level, GTP-bound KRAS activates PI3K signalling by directly binding to the p110α catalytic subunit^59,60, further suggesting pathway crosstalk. We and others have previously shown that Kras mutation modifies the phenotype of intestinal cancer models in pleiotropic ways, ranging from altering the metabolic landscape, rewiring the translatome through the selective translation of MYC-dependent transcripts, and determining the tumour response to therapy^32,48. For these reasons, we next sought to determine the impact of concurrent mutation of Kras and PI3K hyperactivation on intestinal homeostasis and tumorigenesis.

We crossed the Pik3ca^H1047R allele to Villin-CreERT2 mice carrying an inducible Kras^LSL−G12D (henceforth Kras^G12D) allele⁶¹ to generate Villin-CreERT2 Kras^G12D/wt (Kras^G12D/wt), Villin-CreERT2 Kras^G12D/wt Pik3ca^H1047R/wt (Kras^G12D/wt Pik3ca^H1047R/wt), and Villin-CreERT2 Kras^G12D/wt Pik3ca^{H1047R/H1047R} (Kras^G12D/wt Pik3ca^{H1047R/H1047R}) mice. We induced recombination of the transgenes by administration of tamoxifen and sampled mice 30 days post induction. Co-mutation of Kras and Pik3ca did not perturb gross intestinal homeostasis. Crypt size, BrdU positive cells per crypt, and Olfm4 positive areas were similar in all conditions. Consistent with previously published data ⁶², Kras^G12D expression reduced abundance of Paneth cells in small intestinal crypts. Pik3ca^H1047R co-mutation again led to appearance of Lysozyme expressing cells with goblet cell morphology in villi (Supplementary Fig. 8B).

To investigate the relationship of the Kras^G12D and Pik3ca^H1047R in the cancer setting, we first generated Villin-CreERT2 Apc^fl/wt Kras^G12D/wt (Apc^fl/wt Kras^G12D/wt) mice (Fig. 4A). We induced recombination of the transgenes by administration of tamoxifen and aged mice to clinical endpoint. Apc^fl/wt Kras^G12D/wt mice robustly developed both small intestinal and colonic tumours and had to be sampled at a median of 76 days post induction (Fig. 4A, Supplementary Fig. 8C), which was in line with our previous publications^33,34,63. We then generated Villin-CreERT2 Apc^fl/wt Kras^G12D/wt Pik3ca^H1047R/wt (Apc^fl/wt Kras^G12D/wt Pik3ca^H1047R/wt) and Villin-CreERT2 Apc^fl/wt Kras^G12D/wt Pik3ca^{H1047R/H1047R} (Apc^fl/wt Kras^G12D/wt Pik3ca^{H1047R/H1047R}) mice. Interestingly, even heterozygous Pik3ca mutation was sufficient to vastly accelerate tumorigenesis in both the small intestine and the colon, reducing the time to clinical endpoint from 76 to 39 days post induction (Fig. 4A, Supplementary Fig. 8C). Homozygous Pik3ca mutation further accelerated tumorigenesis and decreased the time to clinical endpoint from 39 to 28 days post induction (Fig. 4A, Supplementary Fig. 8C).

There was no clear histopathological difference between the adenomas, which developed in Kras^G12D/Pik3ca^H1047R double-mutant Apc^fl/wt mice, and those observed in Apc^fl/wt Kras^G12D mice. Analysis of common biomarkers, driven by the activation of PI3K signalling, revealed an increase in the phosphorylation of AKT (Ser473) in Pik3ca^{H1047R/H1047R} small intestinal tumours. RPS6 (Ser 235/236 and Ser240/244) phosphorylation was elevated in most tumours independently of PI3K hyperactivation. Interestingly, PTEN levels were reduced in Kras-mutant SI tumours, compared to the surrounding normal tissue, potentially limiting the extent of PI3K signalling inhibition (Fig. 4B) and underpinning the dramatic impact of a single PI3KCA mutation. IHC phenotyping of colonic tumours did not reveal differences in the levels of PI3K or mTOR signalling effectors upon Pik3ca mutation (Supplementary Fig. 8D). Interestingly, unlike the Kras^wt setting (Supplementary Fig. 1A), detection of the EYFP transcript by RNA scope scaled with Pik3ca^H1047R allele dosage (Supplementary Fig. 8E).

To investigate the interplay between Kras^G12D mutation and Pten loss, we generated Villin-CreERT2 Apc^fl/wt Kras^G12D/wt Pten^fl/wt (Apc^fl/wt Kras^G12D/wt Pten^fl/wt) and Villin-CreERT2 Apc^fl/wt Kras^G12D/wt Pten^fl/fl (Apc^fl/wt Kras^G12D/wt Pten^fl/fl) mice. This resulted in a stepwise acceleration of tumorigenesis, with Apc^fl/wt Kras^G12D/wt Pten^fl/wt mice reaching clinical endpoint at a median survival of 51.5 days, whereas Apc^fl/wt Kras^G12D/wt Pten^fl/fl mice needed to be sampled after only 16 days post induction, showing rampant SI and colonic tumour initiation even in this short timespan (Fig. 4C, Supplementary Fig. 8F).

To assess the impact of PI3K hyperactivation on the transcriptional landscape of Villin-CreERT2 Apc^fl/fl Kras^G12D-mutant (Apc^fl/fl Kras^G12D) guts, we generated Villin-CreERT2 Apc^fl/fl Kras^G12D/wt Pik3ca^H1047R/wt (Apc^fl/fl Kras^G12D/wt Pik3ca^H1047R/wt), Villin-CreERT2 Apc^fl/fl Kras^G12D/wt Pik3ca^{H1047R/H1047R} (Apc^fl/fl Kras^G12D/wt Pik3ca^{H1047R/H1047R}), and Villin-CreERT2 Apc^fl/fl Kras^G12D/wt Pten^fl/fl (Apc^fl/fl Kras^G12D/wt Pten^fl/fl) mice. We induced genetic recombination of the alleles and performed transcriptomic analysis on small intestinal tissue. Given that the recently described PDS1 subtype is characterized by activation of MYC signalling⁴³, we sought to ascertain whether PI3K hyperactivation is associated with an enrichment for MYC-driven transcriptional signatures. Indeed, these signatures were significantly enriched both in the PI3K-mutant and the PTEN-deficient guts (Fig. 4D). Furthermore, Slc7a5 expression was increased upon PI3K hyperactivation (Supplementary Fig. 8G).

In addition to KRAS^G12D, which is the most common KRAS alteration, human CRCs frequently harbour KRAS^G12V mutations⁶⁴. We therefore sought to expand our experiments to a Kras^G12V-mutant model. To this end, we generated Villin-CreERT2 Apc^fl/wt Kras^G12V/wt (Apc^fl/wt Kras^G12V/wt) and Villin-CreERT2 Apc^fl/wt Kras^G12V/wt Flp^ERPik3ca^H1047R (Apc^fl/wt Kras^G12V/wt Flp^ERPik3ca^H1047R) mice, and induced recombination of the transgenes by oral administration of tamoxifen. In this Kras^G12V-mutant background, we observed a drastic reduction in median survival upon Pik3ca mutation, from 182 to 101.5 days post induction (Supplementary Fig. 8H). Small intestinal and colonic tumour numbers at endpoint were similar despite the much shorter time to clinical endpoint (Supplementary Fig. 8I).

Kras mutation drives resistance to PI3K inhibition

Clinically, KRAS mutation is associated with resistance to standard-of-care chemotherapy as well as modern targeted agents such as cetuximab⁶⁵. Moreover, mutant KRAS drives resistance to mTORC1 inhibition in intestinal cancer models³². In line with this, Villin-CreERT2 Apc^fl/wt Kras^G12D/wt mice were resistant to PI3Kα inhibition with AZD8835 (Fig. 4E). Given that AZD8835 effectively suppressed tumorigenesis and extended survival in Kras wild-type Apc^fl/wt Pten^fl/fl and Apc^fl/wt Pten^fl/fl Pik3ca^H1047R mutant models (Fig. 1G, H, I), we next tested whether Kras mutation was able to drive resistance in this setting. Despite the direct impact upon the PI3K pathway downstream of KRAS, AZD8835 could not drive therapeutic benefit in either the Apc^fl/wt Kras^G12D/wt Pik3ca^H1047R/1047R (Fig. 4F) or Apc^fl/wt Kras^G12D/wt Pten^fl/fl (Fig. 4G) models. Of note, the location of the tumours was shifted from the colon to the small intestine in AZD8835-treated Apc^fl/wt Kras^G12D/wt mice, compared with vehicle-treated. This indicates that colonic, but not small intestinal, tumour growth may be impaired/partially susceptible to PI3K-pathway inhibition (Supplementary Fig. 9A). Tumour numbers and anatomical location were not significantly altered, following AZD8835 treatment, in Apc^fl/wt Kras^G12D/wt Pik3ca^H1047R/1047R (Supplementary Fig. 9B) and Apc^fl/wt Kras^G12D/wt Pten^fl/fl mice (Supplementary Fig. 9C).

Taken together, these findings suggest that KRAS mutation drives a complex rewiring of cancer cell signalling that readily adapts to therapeutic pressure.

Slc7a5 deletion delays tumour formation in complex KRAS/PI3K-mutant models and sensitizes to MEK1/2 inhibition

Given the strong cooperation between PI3K-pathway activation and mutant KRAS in the induction of intestinal tumorigenesis in mice, we next examined the impact of Slc7a5 deletion in APC-deficient Kras-mutant mice heterozygous or homozygous for Pik3ca mutation. For this, we generated Villin-CreERT2 Apc^fl/wt Kras^G12D/wt Pik3ca^H1047R/wt (Apc^fl/wt Kras^G12D/wt Pik3ca^H1047R/wt), Villin-CreERT2 Apc^fl/wt Kras^G12D/wt Pik3ca^H1047R/wt Slc7a5^fl/fl (Apc^fl/wt Kras^G12D/wt Pik3ca^H1047R/wt Slc7a5^fl/fl) and Villin-CreERT2 Apc^fl/wt Kras^G12D/wt Pik3ca^{H1047R/H1047R} Slc7a5^fl/fl (Apc^fl/wt Kras^G12D/wt Pik3ca^{H1047R/H1047R} Slc7a5^fl/fl) mice, induced genetic recombination, and aged the mice to clinical endpoint. We observed a slight increase in the median intestinal tumour-free survival of Apc^fl/wt Kras^G12D/wt Pik3ca^H1047R/wt Slc7a5^fl/fl mice, compared with Slc7a5 wild-type counterparts, although this was not statistically significant for this cohort size (Fig. 5A). Small intestinal and colonic tumour numbers were similar upon Slc7a5 deletion (Supplementary Fig. 10A). Importantly, we found that genetic deletion of Slc7a5 prolonged survival in the most aggressive Apc^fl/wt Kras^G12D/wt Pik3ca^{H1047R/H1047R} Slc7a5^fl/fl model, extending median survival from 26 to 37 days post induction (Fig. 5B). SI-tumour numbers were increased in Slc7a5 KO mice, likely due to their extended survival, although colon tumour numbers were similar to those in controls (Supplementary Fig. 10B). We confirmed loss of Slc7a5 in all our models by performing in situ hybridization (RNA scope) for exon 1 of Slc7a5 (Supplementary Fig. 10C). We then performed IHC for the mTORC1 activation markers pS6-235/236 and found persistent mTORC1 activity in tumours that grew out, indicating that PI3K activation may sustain mTORC1 signalling, despite concurrent genetic deletion of Slc7a5 (Fig. 5C).

We proceeded to delete Slc7a5 in Villin-CreERT2 Apc^fl/fl Kras^G12D/wt Pik3ca^{H1047R/H1047R} (Apc^fl/fl Kras^G12D/wt Pik3ca^{H1047R/H1047R}) mice to investigate its role in the acute hyperproliferation model. Slc7a5 KO did not alter crypt proliferation, as determined by scoring BrdU-positive cells (Supplementary Fig. 11A). We performed RNA sequencing of small intestinal tissue. Multiple cell proliferation and metabolism related “Hallmarks” gene sets were negatively enriched upon Slc7a5 depletion, including Myc targets v1, E2F targets, G2M checkpoint, cholesterol homeostasis, and oxidative phosphorylation (Fig. 5D, Supplementary Fig. 11B). We next performed targeted LC-MS analysis Apc^fl/fl Kras^G12D/wt, Apc^fl/fl Kras^G12D/wt Pik3ca^{H1047R/H1047R}, and Apc^fl/fl Kras^G12D/wt Slc7a5^fl/fl small intestinal and colonic tissue. We again focused on amino acid metabolism and could detect significant accumulation of amino acids in Slc7a5 depleted small intestinal and colonic tissue (Supplementary Fig. 11C, D). Leucine, Alanine, Citrulline, Glycine, Proline, Serine, and Threonine were increased both in small intestinal and colonic tissue. Aspartate and Glutamate levels were elevated exclusively in small intestinal tissue, Glutamine and Arginine in colonic tissue. Comparing these data with results presented earlier (Fig. 3H, Supplementary Fig. 6B), there was substantial overlap in the accumulated amino acids, indicating that Slc7a5 depletion drives a similar alteration in amino acid pools in Kras wildtype and mutant tissue. The main difference between the earlier experiment and this is the robust alteration of amino acid pools in colonic tissue, especially the accumulation of Glutamine. This may be the direct result of Kras^G12D mutation, which is capable to drive colonic tumour initiation, unlike Pik3ca^H1047R mutation. PI3K mutation itself only had a minor impact on the abundance of amino acids, as only Tryptophan and Glutamate levels were decreased in Apc^fl/fl Kras^G12D/wt Pik3ca^{H1047R/H1047R} compared to Apc^fl/fl Kras^G12D/wt controls (Supplementary Fig. 11E, F).

Taken together, we show that genetic depletion of Slc7a5 extends survival even in complex Pik3ca/Kras co-mutant mice. Since mutant KRAS induces drug resistance, we sought to pharmacologically inhibit MAPK signalling in combination with Slc7a5 KO^66,67. To this end, we evaluated the response to MEK1/2 inhibition in complex Apc^fl/wt Kras^G12D/wt Pik3ca^{H1047R/H1047R} mice with or without additional Slc7a5 deletion. By treating these mice, from a relatively late timepoint during the course of tumorigenesis, we could show an additive effect of Slc7a5 KO and MEK1/2 inhibition, prolonging survival (Fig. 5E) and reducing tumour number at endpoint (Fig. 5F), indicating the potential for simultaneous therapeutic co-targeting of both MAPK and SLC7A5.

PI3K pathway-activating mutations co-exist in human CRC and are likely selected for during disease progression⁶⁸. Here, we show the cooperation of patient relevant PI3K-pathway activating mutations with co-occurring CRC-driver mutations in genetically engineered mouse models. Our suite of PI3K-hyperactive GEMMs demonstrates an acceleration of tumorigenesis in the gut through marked synergy with Wnt-, and MAPK signalling. Importantly, we show that PI3K hyperactivation drives expression of the amino acid transporter Slc7a5 in intestinal tumours, and that genetic depletion of this gene extends survival in PI3K-driven models, sensitizing them to targeted MAPK inhibition.

The role of PI3K hyperactivation in driving cancer in the mouse gut has been explored before. Transgenic overexpression of a dominant-active PI3K-fusion protein (PIK3ca^*) was targeted to the murine distal small intestine and colon using Fabp1-Cre, leading to the rapid induction of invasive mucinous tumours after 40–60 days without apparent Wnt signalling induction⁶⁹. This allele was then crossed to APC^Min/+ mice, which led to the induction of advanced adenocarcinomas in the small intestine and colon⁷⁰. The same group later induced Pik3ca^H1047R mutation using Fabp1-Cre mice, which led to formation of invasive adenocarcinomas, albeit at a longer latency than the constitutively active fusion protein described earlier⁷¹. By contrast, Hare et al. showed that knock-in of Pik3ca^H1047R into the endogenous locus was not sufficient to induce intestinal tumorigenesis but cooperated with Apc loss to drive the development of highly aggressive and invasive adenocarcinomas in both the small intestine and colon⁷². By using the CAG::FlpeER^T2Pi3kca^H1047R allele, which recombines in the stroma and epithelium, Berenjeno et al. showed that Pik3ca^H1047R mutation induces centrosome amplification and tolerance to genome doubling, and also accelerates the development of intestinal cancer in the context of predisposition to APC-loss³⁷. Here, we induced expression of Pik3ca^H1047R from its endogenous locus throughout the intestinal epithelium, using a Cre recombinase driven by the Villin promoter. In contrast to the earlier models, we saw no increase in the numbers or occurrence of advanced tumours in the colon. There are several potential reasons for the differences between our study findings and those of previous publications: The physiological expression of the Pik3ca^H1047R allele from its endogenous locus, a potential decrease in the signalling strength of the transgenic Pik3ca^H1047R-encoded protein due to its C-terminal fusion with an EYFP reporter via the T2A self-cleaving peptide, the influence of the cell-of-origin on the molecular and phenotypic traits of emergent tumours, and mouse strain differences.

PI3K signalling is not a simple on/off switch, but rather a fine-tuneable cellular mechanism to respond to extracellular stimuli. PIK3CA mutation allele dosage has been shown to be relevant in vitro and in cancer patient cohorts. Homozygous, but not heterozygous, PIK3CA^H1047R mutation was sufficient to drive a self-sustained stemness phenotype in a pluripotent stem cell system, associated with an impaired ability to differentiate and extensive transcriptional remodelling akin to a cancerous state⁷. Double PIK3CA mutations are commonly found in tumours from various tissues, including breast and colorectal cancer. These either occur on the same allele (in cis) or on the other allele (in trans)⁷³. This is, to our knowledge, the first time double PI3K mutations have been studied in mice. Our model system represents double PIK3CA mutations in trans, which colorectal cancers are enriched for⁷³. Homozygous expression of Pik3ca^H1047R in the mouse gut led to enlarged crypts, larger Olfm4 positive area and increased BrdU positive cells. Further, appearance of Lysozyme-expressing, alcian blue positive cells throughout villi indicate mis-differentiation of Paneth cells, resembling a phenotype observed upon genetic LKB1 depletion⁷⁴. Homozygous Pik3ca mutation significantly accelerated time to endpoint in all ageing experiments, and led to increased sensitivity to PI3Kα inhibition, highlighting its functional relevance in driving intestinal cancer.

We have previously shown that mutant KRAS drives expression of Slc7a5 in treatment-refractory intestinal cancer models⁴⁸ and that these models are susceptible to its genetic depletion. Here, we extended this research to investigate the impact of Slc7a5 deletion in PI3K-hyperactive models. We find that it curtails tumour growth, suggesting that strategies to target SLC7A5 activity may be applicable more broadly beyond KRAS mutant cancers. Results from other labs indicate that SLC7A5 expression could be controlled by mTORC1 or MYC, although both factors may be interlinked in cells committed to proliferation^52,75. Both PI3K and MAPK signalling converge on these anabolic factors, and our data suggests that SLC7A5 is an important executor of the programs they regulate. Upon Slc7a5 depletion in the acute model, cellular adaptation was centred around de-novo synthesis and import of amino acids, in line with its function in transmembrane transport of amino acids. This was accompanied by accumulation of multiple amino acids within the tissue, including Serine and Glycine, de-novo synthesis of which is commonly observed as a stress response in amino-acid deprived cancer cells⁷⁶. Slc7a5 deficient tumours at clinical endpoint have seemingly intact mTORC1 signalling, which was not observed in our previous publication in the Kras^G12D mutant setting. This could indicate that PI3K signalling drives resistance to SLC7A5 depletion by activating mTORC1.

Previous publications suggest that loss of functional PTEN leads to dependence on PI3Kβ, rather than PI3Kα^77,78. Here, we show that inhibition of PI3Kα, but not PI3Kβ, is sufficient to extend survival in Pten deficient mice. This could be a tissue specific phenotype, as the previously mentioned studies were investigating breast and prostate cancer. Upon co-mutation of Pten and Pik3ca, relative survival extension in PI3Kα inhibitor-treated mice scaled with Pik3ca^H1047R allele dosage, double Pik3ca^H1047R mutants benefiting most. This mirrored clinical data in breast cancer, although double PIK3CA mutations are mostly found in-cis in this cancer type⁷⁹. While PI3K hyperactive models received at least a mild benefit from targeted PI3Kα mutation, Kras^G12D mutation drives resistance to PI3Kα inhibition, despite concurrent Pik3ca^H1047R mutation, or PTEN depletion. This might contribute to the observed treatment resistance in colorectal cancer trials. Combinatorial inhibition of both MAPK and PI3K signalling has proven difficult due to systemic toxicity. With the advent of both KRAS and PI3K mutation-specific inhibitors, adverse effects could be circumvented. In addition, our findings that high SLC7A5 expression is correlated with PDS1 CRCs, independent of Kras mutation, suggest that newly developed SLC7A5 inhibitors⁸⁰ should be tested as part of combination therapy regimens in this setting. Our newly established autochthonous and immunocompetent mouse models will serve as a useful resource to test these kinds of novel therapeutics.

Mouse experiments

Mice were housed in accordance with UK Home Office Regulations, maintained in a pathogen-free facility under a 12-h light–dark cycle at constant temperature between 19°C–23°C, and 55 ± 10% humidity, and given drinking water and fed a standard chow diet ad libitum. All experiments were conducted in accordance with the UK Home Office guidelines, under project licences 70/8646 and PP3908577, and were reviewed and approved by the University of Glasgow Animal Welfare and Ethical Review Board. Genotyping was performed by Transnetyx according to the previously published protocols provided with references for each allele. Mice were maintained on a C57BL/6J background, back-crossed for at least 9 generations in ageing cohorts of mice carrying the following alleles/transgenes: Villin-CreERT2 transgene⁸¹, Apc^{fl 82}, Kras^{G12D 61}, Pten^{fl 83}, Pik3ca^{fl-H1047R-T2A-EYFP-NLS 35}. For Slc7a5-KO studies, using the Slc7a5^fl allele⁴⁹, mice were on a mixed C57BL/6J background, back-crossed for at least five generations. For studies of mice carrying the CAG::FlpeER^T2Pi3kca^H1047R allele^37,84 or the Kras^LSL-G12V85 allele, mice were on a mixed C57/Bl6 background.

For ageing experiments, Villin-CreERT2 driven genetic recombination was induced by a single intraperitoneal administration of 80 mg/kg tamoxifen, or oral gavage of 80mg/kg of tamoxifen in corn oil for 5 consecutive days when CAG::FlpeER^T2Pi3kca^H1047R mice were induced. For short-term hyperproliferation experiments, Kras wild-type mice were injected with 80 mg/kg tamoxifen on two consecutive days and sampled on day 4 post induction, while Kras-mutant mice were injected with 80 mg/kg tamoxifen and sampled 3 days post induction. Mice were induced at the age of 6–14 weeks of age. Similar numbers of male/female mice were induced in each experiment where possible. For longitudinal in vivo drug treatments, AZD8835⁸⁶ (AstraZeneca), AZD8186⁷⁷ (AstraZeneca), AZD6244⁸⁷ (AstraZeneca) were reconstituted in 0.5% hydroxypropyl methylcellulose (HPMC) + 0.1% Tween-80, and mice were dosed 25 mg/kg for AZD8835 and AZD6244 or 50mg/kg for AZD8186 in 100ul vehicle twice daily by oral gavage. These clinical-grade compounds were supplied by AstraZeneca in a collaborative research agreement. To study tumorigenesis in ageing cohorts, mice were monitored daily and weighed a minimum of three times per week. Humane endpoints were determined based on loss of weight, hunched appearance and paling due to intestinal tumour-induced anaemia. For proliferation analysis, mice were injected 250µl of BrdU (Amersham Biosciences) IP two hours before sampling. 25 half-crypt/villus axes were then manually scored for proliferative cells. For metabolomics experiments, mice were oral gavaged 13C-Glutamine (Cambridge Isotopes) 2x 200mg per kg (40mg/ml in PBS) 15 minutes apart, 30min before sampling.

Power calculations were performed a priori to determine the least amount of animals to show desired effect sizes, in accordance with the 3Rs. Censors in ageing experiments were due to animals being found dead, unless stated otherwise; no histology or tumour scores were collected from these animals. Missing tumour scores in some cohorts were due to blocks being irreversibly lost, or, in studies with Flp^ER mice, not scored upon sampling. For treatment experiments, mice were randomly allocated to treatment and vehicle controls before induction to ensure appropriate sex balance; no formal randomisation method was used. No other confounders were controlled for. The investigators were not blinded to the treatment conditions of individual animals. The ARRIVE guidelines were considered when reporting data related to animal research.

RNA sequencing and gene set enrichment analysis

Tissue for RNA sequencing was frozen in RNAlater (Sigma, #R0901). Samples were homogenized using ceramic bead-filled tubes (Precellys, #03961-1-00) in a Precellys Evolution tissue homogenizer. The RNeasy mini kit (Qiagen, #74104) was used to isolate RNA, according to the manufacturer’s protocol, and β-mercaptoethanol (Sigma, #M6250) was used during this process to preserve RNA integrity. The Agilent 2200 TapeStation in combination with RNA ScreenTape (Agilent, #5067–5576) were used to validate the purity and quality of RNA. cDNA libraries were prepared using a TruSeq RNA LT Kit (Illumina) and quality was evaluated using the Agilent 2200 TapeStation system. DNA library-sequencing was performed on an Illumina Next Seq 500 DNA-sequencer using the Illumina High Output 75-cycle kit (2 × 36 cycles, paired-end reads, single index). Raw sequence quality was assessed using the FastQC algorithm version 0.11.8. The sequences were trimmed to remove adaptor sequences and low-quality base calls, defined by a Phred score of < 20, using the Trim Galore tool version 0.6.4. The trimmed sequences were aligned to the mouse genome build GRCm38.98 using HISAT2 version 2.1.0; then, raw counts per gene were determined using FeatureCounts version 1.6.4. Differential expression analysis was performed using the R package DESeq2 version 1.22.2. Reactome pathway analysis was performed using the R packages ReactomePA version 1.28.0 and org.mM.eg.db version 3.8.2. Gene set expression analysis was conducted using GSEA v4.1.0 [build 27] using standard settings. HALLMARK gene sets were downloaded from the Molecular Signatures Database (MSigDB) (https://www.gsea-msigdb.org/gsea/msigdb/collections.jsp) ⁸⁸.

Immunohistochemistry and RNA in situ hybridisation

Histological samples from mice were fixed between 24 h and 72 h in 10% neutral buffered formalin. For BrdU staining in Apc^fl/fl mice, intestines were fixed in methanol:choloform:acetic acid at a ratio 4:2:1 overnight, then transferred to formalin. Standard histological procedures were used in this study, and the following primary antibodies, raised against the indicated antigens, were used: PTEN (Cell Signaling, #9559, 1:70), pAKT-S473 (Cell Signaling, #4060, 1:45), pS6-S235/236 Cell Signaling, #4858, 1:75), pS6 Ser240/244 (Cell Signaling, #5364, 1:1000), p4E-BP1-T37/46 (Cell Signaling Technology, 2855, 1:500), BrdU (BD Biosciences, 347580, 1:200), and SLC7A5 (Abcam, #EPR17573, 1:50).

RNA in situ hybridisation (ISH) was performed on a Leica Biosystems BOND RX automated IHC/ISH stainer, using the RNA scope 2.5 LS Reagent Kit–BROWN (Advanced Cell Diagnostics, #322100) according to manufacturer’s instructions. The Mm-PPIB probe (Advanced Cell Diagnostics, #313918) was run as a positive control to confirm the integrity of the RNA. The dapB probe (Advanced Cell Diagnostics, #312038) was used as a negative control. Probes used: Slc7a5 (Advanced Cell Diagnostics, #472578) detects the floxed exon 1 of the gene; Lgr5 (Advanced Cell Diagnostics, #312178). Images were acquired using an Olympus BX51 microscope and Zeiss ZEN lite software.

Tissue microarray analysis and association with clinical features

To investigate SLC7A5 protein expression in human tissue, a retrospectively collected CRC-patient cohort (n = 787) was stained via IHC. Ethical approval was in place (MREC/01/0/36) and data are stored within University of Glasgow Safehaven (GSH21ON009). Expression of SLC7A5 was semi-quantitatively measured using the weighted histoscore (WHS) method with HALO digital pathology software (Indica Labs, v3.6.4). SLC7A5 scores were assessed for association with the cell-cycle protein cyclin D1 and the proliferation marker Ki67. Ki67 WHS values were obtained from quantification of manual IHC on tissue microarray (TMA) sections from the patient cohort as part of a previous study ⁸⁹, Nuclear Cyclin D1 as part of this study. Briefly, sections were dewaxed in Histo-Clear (National Diagnostics, #A20101) and rehydrated through a series of graded alcohols. Antigen retrieval was performed by heating TMAs in Tris-EDTA buffer pH 9 and endogenous peroxidases were blocked in 0.3% hydrogen peroxide. Sections were blocked in 5% horse serum (cyclin D1) or 10% casein (Ki67) and then incubated in primary antibody against Ki67 (Agilent Dako, #GA626, 1:50) or cyclin D1 (Abcam, ab273608, 1:750) overnight at 4°C. Slides were washed in Tris-buffered saline solution with 0.1% Tween-20 before incubation with secondary antibody (MP-7800-15, ImmPRESS HRP Universal horse anti-mouse/rabbit IgG PLUS Polymer, Peroxidase Kit; Vector Laboratories). Sections were washed and 3,3′-diaminobenzidine (DAB) chromogen (SK4100, Vector Laboratories) applied before counterstaining in haematoxylin and dehydration through graded alcohols. Stained sections were mounted using Omnimount (HS110, National Diagnostics,) and scanned onto a NZ Connect viewing platform (Hamamatsu Photonics) using a NanoZoomer (Hamamatsu Photonics). Ki67 staining was scored manually as a percentage of positive tumour cells, and cyclin D1 staining intensity was quantified in tumour cell nuclei using the weighted histoscore method. Staining for both proteins was scored by a single observer with 10% of cores independently co-scored to ensure validity. Cyclin D1 was dichotomised into high and low groups based on a weighted histoscore of 111.67 using Survminer (version 0.4.9) in RStudio (version 1.4.0) (Rstudio Inc). Tumours with < 30% of Ki67-positive cells were classified as low Ki67-expressing (i.e., low proliferation). For association of SLC7A5 with survival, data were analysed using using survminer version 0.4.9 and survival version 3.2-7 packages in RStudio version 4.0.4 (RStudio, Boston, MA, US). Kaplan Meier survival analysis and univariate cox regression was performed in SPSS version 28 (IBM SPSS, Chicago, IL, US).

Association of SLC7A5 expression with patient outcome

The publicly available Marisa dataset (GSE39582, Marisa et al., 2013) was accessed from Gene Expression Omnibus on 29/01/2024. The probes-to-genes were collapsed using the collapseRows function within the WGCNA R package (v1.72-1), selecting the probe with the highest mean per gene. Non-tumour samples were removed (n = 19). The remaining n = 566 samples were taken forward for analysis. The PDSpredict function with default settings within the PDSclassifier (v0.1.0) R package was used to assign PDS calls to all samples ⁴³. The calculateSMI function from the PDSclassifier package was used to calculate the stem maturation index (SMI) ⁴³. The calculatePI function within the Proliferative Index (v1.0.1) R package was used to calculate a single sample proliferative index score. For visualisation and statistical analysis ggplot2 (v3.4.2) and ggpubr (v0.6.0) R packages were used. The ssGSEA scores were generated for the Hallmarks collection (n = 50 ontologies), which were accessed via msigdbr R package (v7.5.1), using the GSVA R package (v.1.42.0) with the ‘ssgsea’ method and ssgsea.norm = T.

Mutation frequencies in Human CRC

cBioportal was used to access the MSK metastatic CRC cohort data ⁹⁰. APC, KRAS, PIK3CA, PTEN genes were queried to establish occurrence of mutations and co-occurrences in this cohort (https://bit.ly/3y1AAfw).

Metabolomics experiments

For metabolite extraction, up to 20mg of small intestinal or colonic tissue was homogenized using ceramic bead-filled tubes (Precellys, #03961-1-00) in a Precellys Evolution tissue homogenizer in ice cold solution containing 50% methanol, 30% acetonitrile, 20% water. Samples were centrifuged, supernatants were stored at -80°C, and the CRUK Scotland Institute’s Metabolomics facility analysed the samples. Metabolites from biological extracts were injected (5 µl) and separated using a ZIC-pHILIC column (SeQuant; 150 mm × 2.1 mm, 5 µm; Merck) coupled with a ZIC-pHILIC guard column (SeQuant; 20 mm × 2.1 mm; Merck) using a Vanquish HPLC system (Thermo Fisher Scientific). Chromatographic separation was achieved by a gradient programme that consisted of mobile phases A = 20 mM ammonium carbonate (pH 9.2, 0.1% v/v ammonia, 5µM InfinityLab deactivator (Agilent)) and B = 100% acetonitrile. The elution gradient started with 20% A and held for two minutes, followed by a linear increase to 80% A for 15 minutes and a final re-equilibration step to 20% A. The total run time of the method was 25 min. The column temperature was held at 45°C with a constant flow rate of 200 µl min–1.

A Q Exactive Plus Orbitrap mass spectrometer (Thermo Fisher Scientific) equipped with electrospray ionization was coupled to the HPLC system for both metabolite profiling and metabolite identification. For profiling, the polarity switching mode was used with a resolution (RES) of 70,000 at 200 m/z to enable both positive and negative ions to be detected across a mass range of 75 to 1,000 m/z (automatic gain control (AGC) target of 1 × 10⁶ and maximal injection time (IT) of 250 ms).

Raw data was analysed using Skyline (version 23.1.0.455)⁹¹. Compound identification was assigned by matching the mass and retention time of observed peaks to an in-house library generated using metabolite standards (mass tolerance of 5 ppm and retention time tolerance of 0.5 min). For tracing analysis, the integration range for each isotopologue was manually verified. Results were also corrected for natural isotope abundance using the metabolite Autoplotter tool ⁹². Adapted from ⁹³.

Statistical analyses

All statistical analyses, unless mentioned otherwise, were performed in GraphPad Prism version 10.2.0 (GraphPad Software; www.graphpad.com). For parametric analyses, normality tests were performed (D'Agostino & Pearson; Anderson-Darling; Shapiro-Wilk; Kolmogorov-Smirnov tests) where appropriate. Non-parametric tests were performed if these conditions were not met.

Author Contributions

LBZ: Conceptualisation, Investigation, Methodology, Writing-original draft, writing-review and editing, under the supervision of AC and OJS. CAF, LMM, RAR: Investigation: Supported in vivo experiments. VSM, HBP, TJP: Investigation: Generated in vivo data using the Flp^ERPik3ca^H1047R/wt model. AKN: Investigation: Generated in vivo data on Slc7a5 KO in Apc^fl/fl Pten^fl/fl mice. KG: Formal analysis: Performed RNA-sequencing analysis. NCF, PD: Investigation: Generated data on association of SLC7A5 expression with clinical markers. KP, PH, JE: Investigation: Generated data on association of SLC7A5 with clinical markers in CRC TMA. CN, SRM: Methodology: Supported processing and optimisation of histological techniques. NS: Writing-original draft, writing-review. PHJ, BV: Methodology: Generated PI3K mutant mouse alleles. STB: Resources. AC, OJS: Funding acquisition, Supervision, Writing-original draft, writing-review.

Acknowledgements

We want to thank following CRUK SI Research Services: The Biological Service Unit, Histology, central services and Molecular Technologies services. We also thank Catherine Winchester at the CRUK SI for critical review of the manuscript. We thank Johan Vande Voorde for helpful discussions. We thank members of the Wnt- and colorectal cancer group who supported this work by performing animal welfare-related duties.

Competing interests and funding

This work was funded in part through a research agreement with AstraZeneca. The funder had no role in the conceptualization, study design, data collection, analysis and interpretation, the decision to publish, or the preparation of the manuscript.

L.B.Z. was supported by AstraZeneca Echo awards (Awarded to O.J.S., 10049863/10037842 and 10045935).

L.B.Z, C.A.F., A.K.N., O.J.S. were supported by CRUK Grand Challenge “Rosetta” (A25045).

C.A.F. was supported by AstraZeneca Echo award (10045935) (awarded to O.J.S.).

O.J.S. has received research funding from AstraZeneca, Boehringer Ingelheim, Novartis and Cancer Research Horizons and was supported by the MRC mouse genetics network (MC_PC_20025).

L.B.Z., L.M.M., N.S., R.AR., A.D.C., O.J.S. were supported by CRUK Scotland institute group award (awarded to O.J.S., DRCQQR-May21\100002).

A.H.U., D.S., S.R.M., C.N. were supported by CRUK Scotland institute core funding (A31287).

K.G. was supported by Cancer Grand Challenges “Specificancer” (A29055).

K.A.F.P, P.H., J.E. were supported by grants CSO (EPD/22/13), CSO (TCS/22/02), CTRQQR-(2021\100006).

N.C.F., P.D.D. were supported by a CRUK early detection grant (awarded to P.D.D., A29834).

V.M. was supported by CRUK (C1295/A15937).

H.B.P. was supported by a Cancer Research UK career development fellowship (A27894).

T.P. was supported by funding from the Medical Research Council (MR/R026424/1).

P.H.J. was supported by funding from CRUK (C609/A27326) and the Wellcome Trust (220540/Z/20/A).

B.V. is a consultant for iOnctura (Geneva, Switzerland) and Pharming (Leiden, the Netherlands) and a shareholder of Open Orphan (Dublin, Ireland) and was supported by funding from CRUK (25722).

S.T.B. is an AstraZeneca employee and shareholder. All other authors declare no competing interests.

Data availability statement

RNA-sequencing data that support the findings of this study are available in the GEO database, accession number GSE272639. Human data analyzed in this study are publicly available for the Marisa cohort (GSE39582, Marisa et al., 2013), available on cBioportal for the MSK metastatic CRC cohort (https://bit.ly/3y1AAfw, ⁹⁰), and stored within University of Glasgow Safehaven (GSH21ON009) for the TMA study. Source data files that support the findings of this study are stored at the Cancer Research UK Scotland Institute and are available from the corresponding author upon reasonable request.

Author ORCIDS

Lucas B. Zeiger (0000-0002-8712-3112), Catriona A. Ford (0000-0003-3991-6369), Arafath Kaja Najumudeen (0000-0002-3764-5721), Valerie S. Meniel (0000-0002-9743-2111), Kathryn A.F. Pennel (0000-0003-0186-1331), Natalie C. Fisher (0000-0003-1769-7468), Kathryn Gilroy (0000-0003-4607-194X), Laura M. Millett (0009-0000-7596-4275), Nathalie Sphyris (0000-0003-1860-1958), Alejandro Huerta Uribe (0000-0002-6452-6111), David Sumpton(0000-0002-9004-4079), Phimmada Hatthakarnkul (0000-0002-5274-5480), Sophie R. Mclaughlin (0009-0009-5222-1028), Philip H. Jones (0000-0002-5904-795X), Bart Vanhaesebroeck (0000-0002-7074-3673), Rachel A. Ridgway (0000-0003-0198-8244), Colin Nixon (0000-0002-8085-2160), Helen B. Pearson (0000-0002-3284-0843), Toby J. Phesse (0000-0001-9568-4916), Simon T. Barry (0000-0002-8511-0588), Joanne Edwards (0000-0002-7192-6906), Phillip Dunne (0000-0001-9160-283X), Andrew D. Campbell (0000-0003-3930-1276), Owen J. Sansom (0000-0001-9540-3010)

Fearon, E. R. Molecular Genetics of Colorectal Cancer. Annual Review of Pathology: Mechanisms of Disease 6, 479–507 (2011).
Su, L. K. et al. Multiple intestinal neoplasia caused by a mutation in the murine homolog of the APC gene. Science 256, 668–670 (1992).
Hernández, A. R., Klein, A. M. & Kirschner, M. W. Kinetic responses of β-catenin specify the sites of Wnt control. Science (1979) 338, 1337–1340 (2012).
Muzny, D. M. et al. Comprehensive molecular characterization of human colon and rectal cancer. Nature 2012 487:7407 487, 330–337 (2012).
Yaeger, R. et al. Clinical Sequencing Defines the Genomic Landscape of Metastatic Colorectal Cancer. Cancer Cell 33, 125-136.e3 (2018).
Fruman, D. A. et al. The PI3K Pathway in Human Disease. Cell 170, 605–635 (2017).
Madsen, R. R. et al. Oncogenic PIK3CA promotes cellular stemness in an allele dose-dependent manner. Proc Natl Acad Sci U S A 116, 8380–8389 (2019).
Vasan, N. et al. Double PIK3CA mutations in cis increase oncogenicity and sensitivity to PI3Kα inhibitors. Science 366, 714–723 (2019).
Sivakumar, S. et al. Genetic Heterogeneity and Tissue-specific Patterns of Tumors with Multiple PIK3CA Mutations. Clinical Cancer Research 29, 1125–1136 (2023).
Ebner, M., Lučić, I., Leonard, T. A. & Yudushkin, I. PI(3,4,5)P3 Engagement Restricts Akt Activity to Cellular Membranes. Mol Cell 65, 416-431.e6 (2017).
Serebriiskii, I. G. et al. Comprehensive characterization of PTEN mutational profile in a series of 34,129 colorectal cancers. Nature Communications 2022 13:1 13, 1–17 (2022).
Brito, M. B., Goulielmaki, E. & Papakonstanti, E. A. Focus on PTEN regulation. Front Oncol 5, 166 (2015).
Prenen, H. et al. PIK3CA mutations are not a major determinant of resistance to the epidermal growth factor receptor inhibitor cetuximab in metastatic colorectal cancer. Clinical Cancer Research 15, 3184–3188 (2009).
Roock, W. De, Vriendt, V. De, Normanno, N., Ciardiello, F. & Tejpar, S. KRAS, BRAF, PIK3CA, and PTEN mutations: implications for targeted therapies in metastatic colorectal cancer. Lancet Oncol 12, 594–603 (2011).
Mei, Z., Duan, C. Y., Li, C. B., Cui, L. & Ogino, S. Prognostic role of tumor PIK3CA mutation in colorectal cancer: a systematic review and meta-analysis. Annals of Oncology 27, 1836 (2016).
Sepulveda, A. R. et al. Molecular Biomarkers for the Evaluation of Colorectal Cancer. Am J Clin Pathol 147, 221–260 (2017).
Byun, J. et al. PIK3CA Mutations as a Prognostic Factor in Patients With Residual Rectal Cancer After Neoadjuvant Chemoradiotherapy. Anticancer Res 43, 1513–1520 (2023).
Wu, J. B., Li, X. J., Liu, H., Liu, Y. J. & Liu, X. P. Association of KRAS, NRAS, BRAF and PIK3CA gene mutations with clinicopathological features, prognosis and ring finger protein 215 expression in patients with colorectal cancer. Biomed Rep 19, (2023).
Yang, H. et al. Integrating cfDNA liquid biopsy and organoid-based drug screening reveals PI3K signaling as a promising therapeutic target in colorectal cancer. J Transl Med 22, 1–13 (2024).
André, F. et al. Alpelisib plus fulvestrant for PIK3CA-mutated, hormone receptor-positive, human epidermal growth factor receptor-2–negative advanced breast cancer: final overall survival results from SOLAR-1. Annals of Oncology 32, 208–217 (2021).
Turner, N. C. et al. Capivasertib in Hormone Receptor–Positive Advanced Breast Cancer. New England Journal of Medicine 388, 2058–2070 (2023).
Tabernero, J. et al. Phase 2 results: Encorafenib (ENCO) and cetuximab (CETUX) with or without alpelisib (ALP) in patients with advanced BRAF-mutant colorectal cancer (BRAFm CRC). https://doi.org/10.1200/JCO.2016.34.15_suppl.3544 34, 3544–3544 (2016).
NCT01719380. A Phase Ib/II Multi-center, Open-label, Dose Escalation Study of LGX818 and Cetuximab or LGX818, BYL719, and Cetuximab in Patients With BRAF Mutant Metastatic Colorectal Cancer. (2021).
Dunn, S. et al. AKT-mTORC1 reactivation is the dominant resistance driver for PI3Kβ/AKT inhibitors in PTEN-null breast cancer and can be overcome by combining with Mcl-1 inhibitors. Oncogene 2022 41:46 41, 5046–5060 (2022).
Kim, J. & Guan, K.-L. mTOR as a central hub of nutrient signalling and cell growth. Nat Cell Biol 21, 63–71 (2019).
Rebsamen, M. et al. SLC38A9 is a component of the lysosomal amino acid sensing machinery that controls mTORC1. Nature 2015 519:7544 519, 477–481 (2015).
Wang, S. et al. Lysosomal amino acid transporter SLC38A9 signals arginine sufficiency to mTORC1. Science (1979) 347, 188–194 (2015).
Wyant, G. A. et al. mTORC1 Activator SLC38A9 Is Required to Efflux Essential Amino Acids from Lysosomes and Use Protein as a Nutrient. Cell 171, 642-654.e12 (2017).
Wolfson, R. L. et al. Sestrin2 is a leucine sensor for the mTORC1 pathway. Science (1979) 351, 43–48 (2016).
Nicklin, P. et al. Bidirectional transport of amino acids regulates mTOR and autophagy. Cell 136, 521–34 (2009).
Faller, W. J. et al. MTORC1-mediated translational elongation limits intestinal tumour initiation and growth. Nature 517, 497–500 (2015).
Knight, J. R. P. et al. MNK inhibition sensitizes KRAS-Mutant colorectal cancer to mTORC1 inhibition by reducing eIF4E phosphorylation and C-MYC expression. Cancer Discov 11, 1228–1247 (2021).
Knight, J. R. P. et al. MNK inhibition sensitizes KRAS-Mutant colorectal cancer to mTORC1 inhibition by reducing eIF4E phosphorylation and C-MYC expression. Cancer Discov 11, 1228–1247 (2021).
Najumudeen, A. K. et al. The amino acid transporter SLC7A5 is required for efficient growth of KRAS-mutant colorectal cancer. Nature Genetics 2021 53:1 53, 16–26 (2021).
Herms, A. et al. Levelling out differences in aerobic glycolysis neutralizes the competitive advantage of oncogenic PIK3CA mutant progenitors in the esophagus. bioRxiv 2021.05.28.446104 (2021) doi:10.1101/2021.05.28.446104.
Vasudevan, K. M. et al. AKT-Independent Signaling Downstream of Oncogenic PIK3CA Mutations in Human Cancer. Cancer Cell 16, 21–32 (2009).
Berenjeno, I. M. et al. Oncogenic PIK3CA induces centrosome amplification and tolerance to genome doubling. Nat Commun 8, 1773 (2017).
Barlaam, B. et al. Discovery of 1-(4-(5-(5-amino-6-(5-tert-butyl-1,3,4-oxadiazol-2-yl)pyrazin-2-yl)-1-ethyl-1,2,4-triazol-3-yl)piperidin-1-yl)-3-hydroxypropan-1-one (AZD8835): A potent and selective inhibitor of PI3Kα and PI3Kδ for the treatment of cancers. Bioorg Med Chem Lett 25, 5155–5162 (2015).
Sansom, O. J. et al. Loss of Apc in vivo immediately perturbs Wnt signaling, differentiation, and migration. Genes Dev 18, 1385 (2004).
Guinney, J. et al. The consensus molecular subtypes of colorectal cancer. Nat Med 21, 1350–1356 (2015).
Isella, C. et al. Selective analysis of cancer-cell intrinsic transcriptional traits defines novel clinically relevant subtypes of colorectal cancer. Nat Commun 8, 15107 (2017).
Joanito, I. et al. Single-cell and bulk transcriptome sequencing identifies two epithelial tumor cell states and refines the consensus molecular classification of colorectal cancer. Nature Genetics 2022 54:7 54, 963–975 (2022).
Malla, S. B. et al. Pathway level subtyping identifies a slow-cycling biological phenotype associated with poor clinical outcomes in colorectal cancer. Nature Genetics 2024 1–15 (2024) doi:10.1038/s41588-024-01654-5.
Marisa, L. et al. Gene Expression Classification of Colon Cancer into Molecular Subtypes: Characterization, Validation, and Prognostic Value. PLoS Med 10, e1001453 (2013).
Ramaker, R. C. et al. RNA sequencing-based cell proliferation analysis across 19 cancers identifies a subset of proliferation-informative cancers with a common survival signature. Oncotarget 8, 38668–38681 (2017).
Malla, S. B. et al. Pathway level subtyping identifies a slow-cycling biological phenotype associated with poor clinical outcomes in colorectal cancer. Nature Genetics 2024 1–15 (2024) doi:10.1038/s41588-024-01654-5.
Kim, J. et al. A Myc Network Accounts for Similarities between Embryonic Stem and Cancer Cell Transcription Programs. Cell 143, 313–324 (2010).
Najumudeen, A. K. et al. The amino acid transporter SLC7A5 is required for efficient growth of KRAS-mutant colorectal cancer. Nature Genetics 2021 53:1 53, 16–26 (2021).
Sinclair, L. V. et al. Control of amino-acid transport by antigen receptors coordinates the metabolic reprogramming essential for T cell differentiation. Nature Immunology 2013 14:5 14, 500–508 (2013).
Crawford, R. R. et al. Human CHAC1 Protein Degrades Glutathione, and mRNA Induction Is Regulated by the Transcription Factors ATF4 and ATF3 and a Bipartite ATF/CRE Regulatory Element. Journal of Biological Chemistry 290, 15878–15891 (2015).
Kim, M., Gwak, J., Hwang, S., Yang, S. & Jeong, S. M. Mitochondrial GPT2 plays a pivotal role in metabolic adaptation to the perturbation of mitochondrial glutamine metabolism. Oncogene 2019 38:24 38, 4729–4738 (2019).
Tameire, F. et al. ATF4 couples MYC-dependent translational activity to bioenergetic demands during tumour progression. Nature Cell Biology 2019 21:7 21, 889–899 (2019).
Koppula, P., Zhang, Y., Zhuang, L. & Gan, B. Amino acid transporter SLC7A11/xCT at the crossroads of regulating redox homeostasis and nutrient dependency of cancer. Cancer Commun 38, 1–13 (2018).
Nilsson, R. et al. Metabolic enzyme expression highlights a key role for MTHFD2 and the mitochondrial folate pathway in cancer. Nature Communications 2014 5:1 5, 1–10 (2014).
Krall, A. S. et al. Asparagine couples mitochondrial respiration to ATF4 activity and tumor growth. Cell Metab 33, 1013-1026.e6 (2021).
Méndez-Lucas, A., Hyroššová, P., Novellasdemunt, L., Viñals, F. & Perales, J. C. Mitochondrial Phosphoenolpyruvate Carboxykinase (PEPCK-M) Is a Pro-survival, Endoplasmic Reticulum (ER) Stress Response Gene Involved in Tumor Cell Adaptation to Nutrient Availability. Journal of Biological Chemistry 289, 22090–22102 (2014).
DeNicola, G. M. et al. NRF2 regulates serine biosynthesis in non–small cell lung cancer. Nature Genetics 2015 47:12 47, 1475–1481 (2015).
Rebsamen, M. et al. Gain-of-function genetic screens in human cells identify SLC transporters overcoming environmental nutrient restrictions. Life Sci Alliance 5, (2022).
Sjolander, A., Yamamoto, K., Huber, B. E. & Lapetina, E. G. Association of p21(ras) with phosphatidylinositol 3-kinase. Proc Natl Acad Sci U S A 88, 7908–7912 (1991).
Rodriguez-Viciana, P., Sabatier, C. & McCormick, F. Signaling Specificity by Ras Family GTPases Is Determined by the Full Spectrum of Effectors They Regulate. Mol Cell Biol 24, 4943–4954 (2004).
Jackson, E. L. et al. Analysis of lung tumor initiation and progression using conditional expression of oncogenic K-ras. Genes Dev 15, 3243–3248 (2001).
Feng, Y. et al. Mutant Kras Promotes Hyperplasia and Alters Differentiation in the Colon Epithelium but Does Not Expand the Presumptive Stem Cell Pool. Gastroenterology 141, 1003-1013.e10 (2011).
Cammareri, P. et al. TGFβ pathway limits dedifferentiation following WNT and MAPK pathway activation to suppress intestinal tumourigenesis. Cell Death & Differentiation 2017 24:10 24, 1681–1693 (2017).
Serebriiskii, I. G. et al. Comprehensive characterization of RAS mutations in colon and rectal cancers in old and young patients. Nature Communications 2019 10:1 10, 1–12 (2019).
Ciardiello, F. & Tortora, G. EGFR Antagonists in Cancer Treatment. New England Journal of Medicine 358, 1160–1174 (2008).
Wang, Q. et al. PIK3CA mutations confer resistance to first-line chemotherapy in colorectal cancer. Cell Death & Disease 2018 9:7 9, 1–11 (2018).
Gimple, R. C. & Wang, X. RAS: Striking at the Core of the Oncogenic Circuitry. Front Oncol 9, 965 (2019).
Oda, K. et al. PIK3CA Cooperates with Other Phosphatidylinositol 3′-Kinase Pathway Mutations to Effect Oncogenic Transformation. Cancer Res 68, 8127–8136 (2008).
Leystra, A. A. et al. Mice expressing activated PI3K rapidly develop advanced colon cancer. Cancer Res 72, 2931–6 (2012).
Deming, D. A. et al. PIK3CA and APC mutations are synergistic in the development of intestinal cancers. Oncogene 33, 2245–54 (2014).
Yueh, A. E. et al. Colon Cancer Tumorigenesis Initiated by the H1047R Mutant PI3K. PLoS One 11, e0148730 (2016).
Hare, L. M. et al. Physiological expression of the PI3K-activating mutation Pik3ca H1047R combines with Apc loss to promote development of invasive intestinal adenocarcinomas in mice. Biochemical Journal 458, 251–258 (2014).
Sivakumar, S. et al. Genetic Heterogeneity and Tissue-specific Patterns of Tumors with Multiple PIK3CA Mutations. Clinical Cancer Research 29, 1125–1136 (2023).
Shorning, B. Y. et al. Lkb1 Deficiency Alters Goblet and Paneth Cell Differentiation in the Small Intestine. PLoS One 4, e4264 (2009).
Torrence, M. E. et al. The mtorc1-mediated activation of atf4 promotes protein and glutathione synthesis downstream of growth signals. Elife 10, (2021).
Sun, L. et al. cMyc-mediated activation of serine biosynthesis pathway is critical for cancer progression under nutrient deprivation conditions. Cell Research 2015 25:4 25, 429–444 (2015).
Hancox, U. et al. Inhibition of PI3Kβ signaling with AZD8186 inhibits growth of PTEN-deficient breast and prostate tumors alone and in combination with docetaxel. Mol Cancer Ther 14, 48–58 (2015).
Gao, X. et al. Blocking PI3K p110β Attenuates Development of PTEN-Deficient Castration-Resistant Prostate Cancer. Molecular Cancer Research 20, 673–685 (2022).
Vasan, N. et al. Double PIK3CA mutations in cis increase oncogenicity and sensitivity to PI3Ka inhibitors. Science (1979) 366, 714–723 (2019).
Singh, N. & Ecker, G. F. Insights into the Structure, Function, and Ligand Discovery of the Large Neutral Amino Acid Transporter 1, LAT1. International Journal of Molecular Sciences 2018, Vol. 19, Page 1278 19, 1278 (2018).
El Marjou, F. et al. Tissue-specific and inducible Cre-mediated recombination in the gut epithelium. genesis 39, 186–193 (2004).
Shibata, H. et al. Rapid colorectal adenoma formation initiated by conditional targeting of the APC gene. Science (1979) 278, 120–133 (1997).
Suzuki, A. et al. T Cell-Specific Loss of Pten Leads to Defects in Central and Peripheral Tolerance. Immunity 14, 523–534 (2001).
Hunter, N. L., Awatramani, R. B., Farley, F. W. & Dymecki, S. M. Ligand-activated Flpe for temporally regulated gene modifications. genesis 41, 99–109 (2005).
Guerra, C. et al. Tumor induction by an endogenous K-ras oncogene is highly dependent on cellular context. Cancer Cell 4, 111–120 (2003).
Hudson, K. et al. Intermittent high-dose scheduling of AZD8835, a novel selective inhibitor of PI3Ka and PI3Kd, demonstrates treatment strategies for PIK3CA-dependent breast cancers. Mol Cancer Ther 15, 877–889 (2016).
Davies, B. R. et al. AZD6244 (ARRY-142886), a potent inhibitor of mitogen-activated protein kinase/extracellular signal-regulated kinase kinase 1/2 kinases: mechanism of action in vivo, pharmacokinetic/pharmacodynamic relationship, and potential for combination in preclinical models. Mol Cancer Ther 6, 2209–2219 (2007).
Liberzon, A. et al. The Molecular Signatures Database Hallmark Gene Set Collection. Cell Syst 1, 417–425 (2015).
Roseweir, A. K. et al. Histological phenotypic subtypes predict recurrence risk and response to adjuvant chemotherapy in patients with stage III colorectal cancer. J Pathol Clin Res 6, 283–296 (2020).
Yaeger, R. et al. Clinical Sequencing Defines the Genomic Landscape of Metastatic Colorectal Cancer. Cancer Cell 33, 125-136.e3 (2018).
Adams, K. J. et al. Skyline for Small Molecules: A Unifying Software Package for Quantitative Metabolomics. J Proteome Res 19, 1447–1458 (2020).
Pietzke, M. & Vazquez, A. Metabolite AutoPlotter - an application to process and visualise metabolite data in the web browser. Cancer & Metabolism 2020 8:1 8, 1–11 (2020).
Villar, V. H. et al. Hepatic glutamine synthetase controls N5-methylglutamine in homeostasis and cancer. Nature Chemical Biology 2022 19:3 19, 292–300 (2022).

There is NO Competing Interest.

Download PDF

Version 1

posted

You are reading this latest preprint version

The amino acid transporter SLC7A5 drives progression of PI3K-mutant intestinal cancer models and enhances response to MAPK-targeted therapy

Status:

Version 1

Abstract

Figures

Introduction

Results

Discussion

Materials and Methods

RNA sequencing and gene set enrichment analysis

Declarations

References

Additional Declarations

Supplementary Files

Status:

Version 1