Report

LKB1 deficiency in T cells promotes the development of gastrointestinal polyposis

See allHide authors and affiliations

Science  27 Jul 2018:
Vol. 361, Issue 6400, pp. 406-411
DOI: 10.1126/science.aan3975

Inflammation promotes gut polyposis

Peutz–Jeghers Syndrome (PJS) causes benign polyps in the gut and a higher risk of several cancers caused by mutations in the tumor suppressor gene STK11, which encodes liver kinase B1 (LKB1). LKB1's role in this disease is thought to be related to its tumor suppressor function. Now, Poffenberger et al. show that the T cell–specific heterozygous deletion of Stk11 is sufficient to reproduce PJS symptoms in mice (see the Perspective by Hollstein and Shaw). Polyps in mice and humans are characterized by immune cell infiltration, enhanced STAT3 signaling, and increased levels of inflammatory cytokines such as interleukin-6 (IL-6). Targeting STAT3 signaling, IL-6, or T cells ameliorated the polyps, suggesting potential therapies for this disease.

Science, this issue p. 406; see also p. 332

Abstract

Germline mutations in STK11, which encodes the tumor suppressor liver kinase B1 (LKB1), promote Peutz–Jeghers syndrome (PJS), a cancer predisposition syndrome characterized by the development of gastrointestinal (GI) polyps. Here, we report that heterozygous deletion of Stk11 in T cells (LThet mice) is sufficient to promote GI polyposis. Polyps from LThet mice, Stk11+/− mice, and human PJS patients display hallmarks of chronic inflammation, marked by inflammatory immune-cell infiltration, signal transducer and activator of transcription 3 (STAT3) activation, and increased expression of inflammatory factors associated with cancer progression [interleukin 6 (IL-6), IL-11, and CXCL2]. Targeting either T cells, IL-6, or STAT3 signaling reduced polyp growth in Stk11+/ animals. Our results identify LKB1-mediated inflammation as a tissue-extrinsic regulator of intestinal polyposis in PJS, suggesting possible therapeutic approaches by targeting deregulated inflammation in this disease.

Liver kinase B1 (LKB1) is a multifaceted serine/threonine kinase with roles in metabolism, cell polarity, cell size control, and proliferation (1). Somatic mutations in STK11 are associated with several human cancers, including lung cancer (where it is frequently comutated with KRAS) and gynecological cancers (ovary, uterus, and cervix) (2). Heterozygous germline mutations in STK11 predispose both humans and mice to the development of Peutz–Jeghers syndrome (PJS) (37), an autosomal dominant hereditary disease characterized by the development of benign gastrointestinal (GI) hamartomatous polyps (8). PJS patients also have a high cumulative cancer risk, with >90% chance of developing tumors by age 65 (9).

Much of our understanding of PJS development and pathogenesis has been derived from Stk11+/ mice, which develop GI polyps histologically similar to those found in PJS patients, with an average latency of 9 to 12 months (57). Deletion of Stk11 in smooth-muscle cells promotes GI polyp formation with lower penetrance than that in whole-body Stk11+/ mice (~60% versus 85 to 100%) (10). We used a Villin-Cre transgene to delete Stk11 in intestinal epithelial cells (IECs) of the small intestine and colon. We found no evidence of GI polyps in mice heterozygous (Stk11fl/+;Villin-Cre) for Stk11 in IECs, even those >52 weeks of age (table S1), which argues for additional nonepithelial functions for LKB1 that contribute to disease development in Stk11+/ mice and PJS patients.

LKB1 plays an essential role in regulating hematopoietic stem-cell metabolism and survival (1113). Unexpectedly, we found that aged mice (>52 weeks) with heterozygous deletion of Stk11 throughout the hematopoietic system (LHhet: Stk11fl/+;Vav-iCre) developed GI polyps at the junction of the small intestine and stomach reminiscent of PJS polyps (Fig. 1A). Polyps observed in LHhet mice displayed an arborizing smooth-muscle core and were histologically indistinguishable from polyps observed in Stk11+/ mice (Fig. 1A). We next generated mice harboring heterozygous or homozygous Stk11 mutations in either the T cell (LThet: Stk11fl/+;Lck-Cre and LTko: Stk11fl/fl;Lck-Cre) or B cell (LBko: Stk11fl/fl;Cd19-Cre) compartments. LThet mice developed GI polyps at an incidence similar to that of LHhet mice, and homozygous deletion of Stk11 in T cells (LTko) doubled the incidence of polyp formation (Fig. 1, B and C, and table S2). LBko mice did not develop disease (table S1). A subset of male LThet and LTko mice also displayed liver lesions and increased rates of hepatocellular carcinoma (HCC) relative to those of control animals (table S3), a phenotype often observed in Stk11+/ mice (14). Genotyping and immunoblot analysis confirmed reduced LKB1 expression in T cells but not liver tissue of LThet mice (fig. S1). Thus, T cell–specific loss of Stk11 is sufficient to promote intestinal polyposis.

Fig. 1 T cell–specific deletion of LKB1 promotes intestinal polyposis.

(A) Representative hematoxylin and eosin (H&E)–stained sections of stomachs (S), small intestines (D), and polyps (P) from Stk11+/+ (n = 84), Stk11+/ (n = 64), LHwt (n = 28), and LHhet (n = 38) mice aged >52 weeks. Images in the right column are a higher magnification of the boxed region indicated in the left column. Scale bars, 1 mm. (B) Representative H&E-stained sections of stomach (S), small intestine (D), polyps (P), liver, and liver nodules (N) from LTwt (n = 102), LThet (n = 64), and LTko (n = 50) mice aged >52 weeks. Images in the center column are a higher magnification of the boxed region in the left column. Scale bars, 1 mm. (C) Polyp size in Stk11+/ (n = 16), LThet (n = 10), and LTko (n = 9) mouse strains aged >52 weeks. Data are represented as the mean ± SEM. *P < 0.05 compared with control samples (Stk11+/ mice), one-way analysis of variance (ANOVA), Dunnett’s multiple comparisons test.

Many of LKB1’s functions have been attributed to its stimulation of adenosine 5′-monophosphate (AMP)–dependent protein kinase (AMPK)–dependent stress responses (2). However, mice with T cell–specific (ATko: Prkaa1fl/fl;Cd4-Cre) or hematopoietic cell–specific (AHko: Prkaa1fl/fl;Vav-iCre) loss of AMPKα1 did not develop GI polyps (table S1). GI polyps from LThet and LTko mice displayed increased mammalian target of rapamycin complex 1 (mTORC1) signaling (fig. S2A), which is characteristic of polyps from Stk11+/ mice (15). However, mice with conditional deletion of both LKB1 and mTOR in T cells (LTko;T-Frap/: Stk11fl/fl;Frapfl/fl;Lck-Cre+) still developed GI polyps (fig. S2B). Thus, AMPK or mTOR pathway activity in T cells does not appear to contribute to PJS polyp development in these mouse models.

Histological analysis of PJS polyps revealed evidence of prominent immune cell infiltration in polyps from whole-body Stk11+/, LThet, and LTko mice (Fig. 2A and fig. S3A). T cells (CD4+ and CD8+), macrophages, and neutrophils were diffusely present throughout the polyps, with lymphocytes also organizing into concentrated centers in the smooth-muscle core of the polyps (Fig. 2A and fig. S3, A to C). Histological analysis of polyps from two independent, unrelated PJS patients revealed clustering of T cells in the smooth-muscle core and stroma of the polyps (Fig. 2B), similar to our observations in mouse samples. To assess the impact of lymphocytes on disease progression, we crossed Stk11+/ mice onto a Rag2-deficient background (which lack mature lymphocytes because of a block in development). Although Stk11+/;Rag2/ mice still developed polyps, polyp size was significantly decreased in Stk11+/ mice lacking lymphocytes (Fig. 2, C and D).

Fig. 2 Immune-cell infiltration and expression of proinflammatory genes are hallmarks of PJS polyps.

(A and B) Representative H&E-, CD3- (T cell) and Iba-1– (macrophage) stained sections of stomach, small intestine, and polyps from (A) mice aged >52 weeks (n = 3 for each genotype) and (B) PJS patients (n = 2). Scale bars, 100 μm. (C) Representative H&E staining of GI sections from Stk11+/ (n = 18) or Stk11+/;Rag2/ (n = 15) mice aged >52 weeks. Scale bars, 5 mm. (D) Polyp size for animals in (C). Data are represented as the mean ± SEM. *P < 0.05 compared with Stk11+/ mice, unpaired Student’s t test. (E) Normalized relative mRNA levels of inflammatory cytokines (Il1b, Il6, Il11, and Tnfa) in GI tissue from LTwt (WT) (n = 5) or nondiseased (ND) LThet mice (n = 4), or polyps (P) (n = 4) and adjacent nonpolyp tissue (NP) (n = 4) from diseased LThet mice aged >52 weeks. Gene expression relative to Tbp mRNA levels. Data are represented as the mean ± SEM. *P < 0.05 compared with LTwt mice, one-way ANOVA, Dunnett’s multiple comparisons test. (F) Normalized relative cytokine mRNA levels in GI tissue from Stk11+/+ (WT) (n = 16) or polyps (P) (n = 11) and adjacent nonpolyp tissue (NP) (n = 12) from Stk11+/ mice aged >52 weeks. Gene expression relative to Tbp mRNA levels. Data are represented as the mean ± SEM. *P < 0.05 compared with Stk11+/+ mice, one-way ANOVA, Dunnett’s multiple comparisons test. (G and H) Cxcl2 mRNA levels in LThet (n = 4 or 5) and Stk11+/ (n = 6 to 13) mice aged >52 weeks as in (E) and (F). Gene expression relative to Tbp mRNA levels. Data are represented as the mean ± SEM. *P < 0.05 compared with LTwt or Stk11+/+ mice, one-way ANOVA, Dunnett’s multiple comparisons test. (I) Representative flow cytometric plots for frequency of CD11b+ and Gr1+ neutrophils and inflammatory monocytes isolated from GI tissue from control (LTwt) (n = 13) or diseased LThet (n = 13) mice (CD45+ population). (J) Representative H&E staining of stomach (S), small intestine (D), and polyp (P) sections from Stk11+/+ (n = 5) and Stk11+/ (n = 7) mice 16 weeks after irradiation (4.5 Gy). Scale bars, 1 mm. (K) Size of polyps in Stk11+/ mice 16 weeks after irradiation as in (J). Data are represented as the mean ± SEM and are representative of two separate experiments. *P < 0.05 compared with nonirradiated Stk11+/ mice, unpaired Student’s t test.

We next assessed inflammatory cytokine expression in polyps and adjacent nonpolyp tissue from aged control and LThet animals (>52 weeks of age) (Fig. 2E). Polyps from diseased LThet mice consistently expressed increased levels of inflammatory cytokines—including Il1b, Il6, Il11, and Tnfa—compared with GI sections from control animals or LThet mice that did not develop disease (Fig. 2E). Notable in this profile were high levels of Il6 and Il11, cytokines that are associated with chronic gastric inflammation and GI tumor development (1618). Similar changes in inflammatory cytokine gene expression were observed in polyps from Stk11+/ mice (Fig. 2F). The expression of Cxcl2 (MIP-2), a proinflammatory chemokine that directs immune-cell recruitment, was increased in both polyps and adjacent nonpolyp tissue from LThet and Stk11+/ mice (Fig. 2, G and H). Profiling immune-cell subsets from GI tissue of control or polyp-bearing LThet mice by means of flow cytometry revealed the increased presence of inflammatory monocytes (CD11b+Gr1mid) and neutrophils (CD11b+Gr1hi) in polyps of mice heterozygous for LKB1 in T cells (Fig. 2I).

To test whether inflammation could influence PJS polyp formation, we subjected young control and Stk11+/ mice to sublethal irradiation [4.5 grays (Gy)], which stimulates acute mucositis and inflammatory cytokine production in the GI tract (19). All Stk11+/ mice that received radiation developed polyps within 16 weeks of treatment, whereas age-matched nonirradiated Stk11+/ mice did not develop visible polyps during this time (Fig. 2, J and K). Thus, immune infiltration and inflammatory cytokines are hallmarks of PJS polyps, and damage-associated inflammation can initiate polyp formation in genetically susceptible Stk11+/ mice.

We next investigated how LKB1 loss in T cells could account for the proinflammatory nature of PJS polyps. RNA sequencing (RNA-seq) analysis of in vitro activated control and LTko CD8+ T cells revealed enrichment in several pathways central to T cell activation (cytokines/chemokines, ribosome biogenesis, and phosphatidylinositol 3-kinase signaling) in LTko T cells (fig. S4). Among the pathways specifically enriched in activated LTko CD8+ T cells were Janus kinase (JAK)–signal transducers and activators of transcription (STAT) signaling and cytokine–cytokine receptor interactions (figs. S5 and S6), the former scoring highly owing to increased cytokine mRNA expression in activated LTko T cells. The expression of several of these cytokines (Ifng, Il9, and Il24) in activated LThet and LTko CD8+ and CD4+ T cells was confirmed with quantitative polymerase chain reaction (fig. S7, A and B). Stk11-deficient CD4+ and CD8+ T cells activated under nonpolarizing conditions displayed increased secretion of many of the inflammatory factors observed in PJS polyps, including CXCL2, CXCL9, and interleukin 6 (IL-6) (Fig. 3, A and B). The activation of naïve LThet or LTko CD4+ T cells led to the increased generation of interferon-γ (IFN-γ)– and IL-17A–producing CD4+ T cells under T helper 1 (TH1)– and TH17-polarizing conditions, respectively (Fig. 3, C to E). LKB1-deficient TH17 cells also displayed increased IL-6 and tumor necrosis factor–α (TNF-α) production (Fig. 3, E and F).

Fig. 3 Deregulated cytokine production and stimulation of inflammation by LKB1-deficient T cells.

(A and B) Heatmap of cytokine production by activated unpolarized LTwt (n = 6), LThet (n = 7), and LTko (n = 6) (A) CD4+ and (B) CD8+ T cells as measured with MILLIPLEX assay. Shown are cytokines with significant (P < 0.05) up-regulation in LTko T cells. (C) Representative flow cytometry plots for IL-17 and IFN-γ production by LTwt, LThet, and LTko CD4+ T cells cultured under TH1- and TH17-polarizing conditions. Values displayed represent average percent ± SEM for biological replicates (n = 2 or 3). Data are representative of three separate experiments. (D and E) Enzyme-linked immunosorbent assay for IFN-γ production by TH1 cells (D) or IL-17 and IL-6 production by TH17 cells (E) of the indicated genotypes generated as in (C). Data represent the mean ± SEM for technical triplicates. *P < 0.05 compared with LTwt control samples, one-way ANOVA, Dunnett’s multiple comparisons test. (F) Representative flow cytometric plots showing frequency of IL-17 and TNF-α production by LTwt, LThet, and LTko CD4+ T cells cultured under TH17-polarizing conditions as in (C). Values displayed represent average percent ± SEM for biological replicates (n = 2 or 3). (G) Number of IFN-γ–, IL-17–, and IL-6–producing CD4+ and CD8+ T cells in the mesenteric lymph nodes (mLN) of LTwt (n = 19 to 29), LThet (n = 11 to 15), and LTko (n = 5 to 12) mice (>52 weeks of age). Cytokine production by CD4+ and CD8+ T cells was determined by means of intracellular staining after restimulation of T cells directly ex vivo. *P < 0.05 compared with LTwt control samples, one-way ANOVA, Dunnett’s multiple comparisons test. (H) Representative flow cytometric plots for frequency of TNF-α– and IFN-γ–producing CD4+ and CD8+ T cells isolated from healthy GI tissue from control mice (LTwt) (n = 13) or polyp-bearing LThet (n = 13) or LTko (n = 2) mice. (I) Il6 mRNA levels in MEFs after culture (24 hours) with supernatants from activated LTwt, LThet, or LTko CD8+ T cells (n = 3 mice for each genotype). Untreated MEFs cultured with regular medium are shown. Gene expression relative to Tbp mRNA levels. *P < 0.05 compared with control samples (LTwt), one-way ANOVA, Dunnett’s multiple comparisons test. (J) IL-6 production by Stk11+/+ or Stk11/ MEFs after a 6-hour stimulation with LPS from Escherichia coli (LPS-EK), heat-killed Helicobacter pylori (HKHP), Pam3CSK4, FSL-1, or sham treatment. Data are represented as the mean ± SEM for biological duplicates. Data are representative of two separate experiments. *P < 0.05 compared with Stk11+/+ MEF samples, unpaired Student’s t test.

LKB1 deletion in T cells leads to deregulated metabolism and increased IFN-γ production by CD4+ and CD8+ T cells (20). We observed progressive deregulation of T cell homeostasis based on Stk11 gene status, marked by an increased frequency of activated CD4+ and CD8+ T cells (CD44hi) in LThet and LTko mice (fig. S8, A and B). Increased proportions of activated IFN-γ– and IL-17A–producing T cells, as well as IL-6–producing CD4+ T cells, were detected in the GI-draining mesenteric lymph nodes (mLN) of LTko mice (Fig. 3G and fig. S8C). Increased frequency of polyfunctional (IFN-γ+TNF-α+) inflammatory T cells was observed in polyps isolated from LThet and LTko animals (Fig. 3H). The expansion of inflammatory CD4+ TH cells in LTko mice was likely not due to defects in regulatory T cell (Treg cell) populations as control, LThet, and LTko mice displayed similar levels of CD4+Foxp3+ Treg cells (fig. S9, A and B), and we observed no difference in suppressive capacity between control and LKB1-deficient CD4+ Treg cells (fig. S9, C and D).

We next tested whether Stk11-mutant T cells could modulate inflammation in surrounding tissues. Supernatants from Stk11 mutant T cells stimulated a fourfold increase in Il6 expression in nontransformed mouse embryonic fibroblasts (MEFs) compared with supernatants from wild-type T cells (Fig. 3I). Stk11/ MEFs secreted IL-6 in the absence of inflammatory stimuli and produced significantly more CXCL2 and IL-6 when stimulated with lipopolysaccharide (LPS) (Fig. 3J and fig. S10A). Deregulated IL-6 production by Stk1/ MEFs was not limited to LPS treatment because other Toll-like receptor (TLR) ligands strongly induced IL-6 production in LKB1-deficient MEFs (Fig. 3J). IL-6 production by LKB1-deficient MEFs was insensitive to mTORC1 inhibition by use of rapamycin (fig. S10B). Thus, in addition to aberrant immune-cell function, Stk11 mutations in nonimmune tissues (epithelial and/or stromal cells) in PJS patients may confer increased responsiveness to inflammatory stimuli.

In inflammation-driven cancers such as gastric and colon cancer, tumor progression is driven by chronic STAT3 activation downstream of inflammatory cytokines such as IL-6 and IL-11 (21). Consistent with the abundance of STAT3-activating cytokines in the GI microenvironment, polyps from LThet, LTko, and Stk11+/ mice displayed elevated phospho-STAT3 (Y705) staining compared with similar sections from LTwt and Stk11+/+ mice (Fig. 4A). A similar distribution of phospho-STAT3 (Y705) staining was observed in the stromal compartment of polyps from PJS patients (Fig. 4B). Both polyps and adjacent nonpolyp tissue from diseased mice exhibited increased STAT3 phosphorylation compared with that of nondiseased tissue from control mice (Fig. 4, C and D). Expression of several STAT3 target genes, including Ccnd1 and Socs3, was increased in polyps (Fig. 4E and fig. S11A). Additionally, the STAT3 target gene Tlr2, which encodes a pattern-recognition receptor potentiating inflammatory responses to bacterial peptidoglycans, was significantly increased in the polyps of LThet and Stk11+/ mice (Fig. 4F). Increased TLR2 expression and polymorphisms in the TLR2 gene are associated with gastric cancer (22, 23). Conditioned medium from activated LKB1-mutant (LThet and LTko) T cells was sufficient to stimulate STAT3 phosphorylation (Fig. 4G) and promote increased expression of a subset of STAT3 target genes, including Tlr2, in nontransformed MEFs (Fig. 4H and fig. S11B).

Fig. 4 LKB1-deficient T cells promote STAT3 activation and expression of STAT3-dependent growth-promoting and inflammatory genes in PJS polyps.

(A and B) Representative immunohistochemistry (IHC) for P-STAT3 (Y705) staining in healthy tissue from control animals (LTwt and Stk11+/+) (n = 10 and 8, respectively) and polyps from (A) LThet (n = 11), LTko (n = 12), and Stk11+/ (n = 8) mice aged >52 weeks or (B) PJS patients (n = 2). Scale bars, 30 μm. (C and D) Immunoblot for STAT3 activation (pY705 and pS727) in control (LTwt and Stk11+/+), nondiseased LThet (ND), polyp (P), and adjacent healthy tissue (NP) from (C) LThet and (D) Stk11+/ mice aged >52 weeks. (E and F) Relative mRNA expression for STAT3 target genes Ccnd1 (E) and Tlr2 (F) in (left) LThet (n = 4 or 5) and (right) Stk11+/ mice (n = 11 to 16) aged >52 weeks. WT refers to paired control animals (LTwt or Stk11+/+) for each experiment. (G) Immunoblot for STAT3 activation in MEFs treated for 6 hours with IL-6 or supernatants from activated LTwt, LThet, or LTko CD8+ T cells. Cells were treated with the JAK2 inhibitor AG490 for 1 hour before conditioned media treatment where indicated. Data are representative of three separate experiments. (H) Tlr2 gene expression in MEFs after culture (24 hours) with supernatants from activated LTwt, LThet, or LTko T cells (n = 3 mice per genotype). Gene expression is relative to Tbp mRNA levels. Data represent the mean ± SEM. *P < 0.05 compared with control samples (LTwt), one-way ANOVA, Dunnett’s multiple comparisons test. (I) Representative H&E staining of GI sections from Stk11+/ (n = 9) or Stk11+/;Il6/ (n = 20) mice aged >52 weeks. Scale bars, 5 mm. (J) Polyp size for animals in (I). Data represent the mean ± SEM. *P < 0.05 compared with control samples (Stk11+/ mice), unpaired Student’s t test. (K) Representative H&E staining of GI sections from aged Stk11+/ mice treated orally with vehicle control (n = 6) or AZD1480 (n = 3) for 8 weeks. Scale bars, 5 mm. (L) Polyp size of vehicle- and AZD1480-treated Stk11+/ mice as in (K). Data represent the mean ± SEM. *P < 0.05 compared with control samples (vehicle-treated mice), Welch’s t test. (M) Representative IHC for P-STAT3 (Y705), CD3 (T cell), and Iba-1 (macrophage) in GI sections from aged Stk11is mice treated with vehicle control (n = 3) or AZD1480 (n = 3) in (K). Scale bars, 50 μm.

Last, we assessed the contribution of IL-6–STAT3 signaling to PJS polyp growth. GI polyp size was significantly reduced in Stk11+/ mice lacking IL-6 expression (Fig. 4, I and J), similar to that observed in lymphocyte-deficient Stk11+/ mice (Fig. 2, C and D). Polyps from Stk11+/;Il6/ mice still retained phospho-STAT3 (Y705) staining (fig. S11C), suggesting that other STAT3-activating factors such as IL-11 may be active in these tissues. Last, we inhibited STAT3 signaling in Stk11+/ mice in vivo using the JAK2 inhibitor AZD1480 (24, 25). AZD1480-treated Stk11+/ mice displayed significant reductions in polyp size compared with that of controls (Fig. 4, K and L). AZD1480 treatment reduced phospho-STAT3 levels and immune cell infiltration in GI tissues of Stk11+/ mice (Fig. 4M and fig. S12, A and B), whereas mTORC1 activity was unaffected (fig. S12A). Thus, the blockade of STAT3 signaling can affect polyp development in this PJS mouse model.

The development of hamartomatous polyps in PJS patients has been attributed to cell-intrinsic tumor suppressor functions for Stk11 in epithelial and/or stromal tissue cells in the GI tract. Our findings establish immune-mediated inflammation as a hallmark of PJS disease and highlight a critical role for Stk11 mutant T cells in PJS disease progression. Our data argue for a more complex role for Stk11 mutations in PJS disease development, with deregulated inflammatory responses by LKB1 mutant immune cells, in addition to epithelial and stromal tissues (26), reinforcing tumor inflammation and chronic STAT3 activation to drive polyp growth. Consistent with this, elevated IL-6 levels have been observed in LKB1-deficient tumors (27). Our data raise the possibility that inflammatory events in the GI tract (pathogen interactions with Stk11+/ immune cells) may trigger the inflammation we have found associated with PJS polyps in mice and humans to stimulate intestinal polyposis. Targeting chronic GI inflammation may present a novel approach to reducing disease incidence and polyp burden in PJS patients.

Supplementary Materials

www.sciencemag.org/content/361/6400/406/suppl/DC1

Materials and Methods

Figs. S1 to S12

Tables S1 to S3

References (2835)

References and Notes

Acknowledgments: We thank W. Foulkes, C. Duerr, and L. Osborne as well as members of the Jones laboratory for critical reading of the manuscript. We thank M. Ernst, B. Jenkins, A. West, G. Jones, and S. Jones for their experimental advice. We acknowledge technical support from the Histology Facilities of the Goodman Cancer Research Centre (GCRC) and Deeley Cancer Centre (BC Cancer Agency), J.-M. Lapointe of the Comparative Medicine Animal Research Centre (McGill), the Flow Cytometry Facility at McGill University, and the Centre for Applied Genomics (Hospital for Sick Children). We thank AstraZeneca for access to AZD1480 through its Open Innovation program. Funding: This work was supported by grants from the Canadian Cancer Society (CCSRI; 702566 to R.G.J.) and the Canadian Institutes of Health Research (CIHR; MOP-93799 and PJT-156397 to R.G.J., MOP-86582 to N.B., and MOP-142351 to J.J.L.). Fellowship support was provided from the McGill Integrated Cancer Research Training Program (to B.F., J.C., and E.H.M.), the Defi Canderel (to M.C.P.), the Fonds de Recherche du Québec–Santé (FRQS; to M.C.P., E.H.M., G.Z., and J.B.), and the CIHR (to M.C.P. and R.G.J.). Author contributions: Conceptualization was done by M.C.P., J.B., and R.G.J.; investigation was performed by M.C.P., J.B., A.M.-R., E.A., J.C., B.E.H., A.H.W., B.F., B.S., E.H.M., S.-P.G., L.T., L.D., P.K., and E.L.; resources were provided by R.M.J., J.J.L., A.H., S.I., N.B., and M.N.A.; writing and visualization were done by M.C.P. and R.G.J.; supervision and project administration was performed by M.C.P., P.M.S., M.N.A., J.J.L., G.Z., and R.G.J.; and funding acquisition was done by R.G.J. Competing interests: The authors declare no competing interests associated with this work. Data and materials availability: RNA-seq datasets can be found at NCBI (GSE115150). AZD1480 is available from AstraZeneca under a materials transfer agreement with R.G.J. and McGill University.
View Abstract

Stay Connected to Science

Navigate This Article