Sphingosine-1-phosphate lyase downregulation promotes colon carcinogenesis through STAT3-activated microRNAs

Abstract

Growing evidence supports a link between inflammation and cancer; however, mediators of the transition between inflammation and carcinogenesis remain incompletely understood. Sphingosine-1-phosphate (S1P) lyase (SPL) irreversibly degrades the bioactive sphingolipid S1P and is highly expressed in enterocytes but downregulated in colon cancer. Here, we investigated the role of SPL in colitis-associated cancer (CAC). We generated mice with intestinal epithelium-specific Sgpl1 deletion and chemically induced colitis and tumor formation in these animals. Compared with control animals, mice lacking intestinal SPL exhibited greater disease activity, colon shortening, cytokine levels, S1P accumulation, tumors, STAT3 activation, STAT3-activated microRNAs (miRNAs), and suppression of miR-targeted anti-oncogene products. This phenotype was attenuated by STAT3 inhibition. In fibroblasts, silencing SPL promoted tumorigenic transformation through a pathway involving extracellular transport of S1P through S1P transporter spinster homolog 2 (SPNS2), S1P receptor activation, JAK2/STAT3-dependent miR-181b-1 induction, and silencing of miR-181b-1 target cylindromatosis (CYLD). Colon biopsies from patients with inflammatory bowel disease revealed enhanced S1P and STAT3 signaling. In mice with chemical-induced CAC, oral administration of plant-type sphingolipids called sphingadienes increased colonic SPL levels and reduced S1P levels, STAT3 signaling, cytokine levels, and tumorigenesis, indicating that SPL prevents transformation and carcinogenesis. Together, our results suggest that dietary sphingolipids can augment or prevent colon cancer, depending upon whether they are metabolized to S1P or promote S1P metabolism through the actions of SPL.

Figures

Figure 7. Oral SD induces SPL expression, reduces colon S1P levels, prevents STAT3 activation, and reduces cytokines and CAC.

(A) Colon S1P levels in WT mice treated with 25 mg/kg SD (n = 4) or vehicle (n = 8) for 1 week. (B) DAI of SD-treated mice compared with vehicle-treated mice. (C) Percentage of weight loss in SD-treated and vehicle-treated mice. (D) SPL expression in SD-treated DLD1 cells for 24 hours. (E) SPL, phosphorylated STAT3 (STAT3-P), and total STAT3 (STAT3-T) expression in WT mice treated with AOM/DSS plus 25 mg/kg SD or vehicle. Blots shown are from different gels. (F) Colon inflammatory cytokines in vehicle-treated (white bars) and SD-treated (black bars) mice. (G) Average tumor number per mouse in vehicle, SD-treated, or sphingosine-treated (25 mg/kg) mice. (H) IF of activated STAT3 phosphorylated on Tyr705 (STAT3-P, green) in colon tissues of SD-treated and vehicle-treated mice. Nuclei were counterstained with DAPI. Scale bar: 50 μm. (I) PTEN and CYLD protein expression in colons of SD-treated and vehicle-treated mice. *P < 0.05.

Figure 6. SPL downregulation promotes cell transformation in a STAT3-dependent manner.

(A) IB of SPL and STAT3 expression in Stat3+/+ and Stat3–/– MEFs in which SPL was silenced (SPL knockdown) compared with vector-only control (C) cell lines. (B) Intracellular and (C) extracellular S1P levels in MEF lines. (D) Spns2 mRNA expression levels in MEF lines. (E) Extracellular S1P levels in Stat3+/+ SPL knockdown MEFs with or without SPNS2 siRNA and MEF Stat3+/+ control cells. (F) Cell proliferation at day 3 of MEFs grown in 0% FBS in Stat3+/+ SPL knockdown MEFs with and without SPNS2 siRNA and MEF Stat3+/+ control cells. Representative bar graph from 1 individual experiment is shown. Statistical analysis was performed with data from 3 independent experiments. (G) Il6 mRNA expression levels in MEF lines. (H) Protein expression of activated JAK-2 (JAK2-P), total JAK-2 (JAK2-T), activated STAT3 (STAT3-P), and total STAT3 (STAT3-T) in control and SPL knockdown Stat3+/+ MEFs, following treatment with 125 nM S1P. (I) Cell proliferation of MEFs grown in 0% FBS. (J) Time course of tumor growth, following s.c. injection of MEF cell lines into NOD/SCID mice. Blots shown are from different gels. (K) miR-181b-1 expression and (L) Cyld expression in tumors or injection sites of NOD/SCID mice. (M) Cell proliferation at day 3 of MEFs grown in 0% FBS with or without W123 (10 μM), FTY720 (100 nm), rolipram (10 μM), NSC 74859 (25 μM), and WP1066 (1 μM) added at day 0. *P < 0.05 compared with control Stat3+/+ MEFs. #P < 0.05 compared with control SPL knockdown Stat3+/+ MEFs.

Figure 5. The S1P/SPL/STAT3 axis in experimental murine colitis and human IBD.

Control and KO mice were treated for 2 cycles of 3% DSS. (A) DAI. (B) Colon length. (C) Colon S1P levels in WT mice treated with vehicle or DSS and control and KO mice treated with DSS. (D) IHC (original magnification, ×10) of activated STAT3 phosphorylated on Tyr705 in representative colon sections of DSS-treated control and KO mice. (E) Colon miR-181b-1 expression. (F) Colon miR-21 expression. n = 4 per group. (G–K) Colon tissues from patients with IBD and controls evaluated for expression of S1P-related genes. (G) SGPL1. Control, n = 6; IBD, n = 8. (H) SPHK1. Control, n = 6; IBD, n = 8. (I) S1PR1. Control, n = 4; IBD, n = 7. (J) miR-181b-1 expression in control (n = 4) and IBD (n = 5) colons. (K) miR-21 in control (n = 5) compared with IBD (n = 6) colons. *P < 0.05.

Figure 4. Loss of SPL in gut epithelium promotes cytokine production, PTEN and CYLD repression, and colon tumorigenesis in a STAT3-dependent manner.

KO mice were treated with AOM/DSS with or without NSC 74859 (10 mg/kg i.p.) or saline every other day on days 30 to 60. White bars represent KO plus saline, and black bars represent KO plus NSC 74859. (A) Colon inflammatory cytokines. (B) Average colon length. (C) Average tumor number per mouse. (D) miR-181b-1 expression. (E) miR-21 expression. (F) PTEN and CYLD protein expression. n = 6 per group. *P < 0.05.

Figure 3. Loss of SPL in gut epithelium enhances STAT3 activation and induction of STAT3-activated miRNAs.

(A) IF of activated STAT3 phosphorylated on Tyr705 (STAT3-P, green) and counterstained with DAPI (blue) in control and KO mouse colon tissues, following administration of AOM/DSS. Scale bar: 50 μm. (B) Activated STAT3 phosphorylated on Tyr705 IHC in representative colon sections (original magnification, ×10). (C) Protein expression of phosphorylated STAT3 (STAT3-P), total STAT3 (STAT3-T), phosphorylated JAK2 (JAK2-P), total JAK2 (JAK2-T), C-MYC, and MCL-1 in colon tissues of control and KO mice. Note that in the last lane of JAK2-T IB there is a nonspecific band. (D) mRNA expression of murine S1pr1 in nonneoplastic colon tissues of control, KO, and KO mice treated with NSC 74859. (E) mRNA expression of human S1PR1 in SPL knockdown (SPL-KD) and vector control DLD1 cells. (F) miR-181b-1 and (G) miR-21 expression in nonneoplastic colon tissues of control and KO mice. (H) Protein expression of PTEN and CYLD in nonneoplastic colon tissues of control and KO mice. *P < 0.05.

Figure 2. Loss of SPL in gut epithelium promotes susceptibility to AOM/DSS-induced CAC, tumor formation, and crypt cell proliferation.

(A) Kaplan-Meier plot of control (n = 12) and KO (n = 21) survival on AOM/DSS regimen. (B) DAI. Starting on day 16, significance of control (blue line) and KO (red line) mouse DAI was P < 0.05. (C) Plasma and (D) colon inflammatory cytokines. White bars represent controls, and black bars represent KO. (E) Average percentage of Th1 (CD4+IFN-γ+) and Th17 (CD4+IL-17+) cells plotted from 2 independent experiments. No treatment (NT), n = 3; control with AOM/DSS treatment (C), n = 7; SPLGutKO with AOM/DSS treatment (KO), n = 5. (F) Representative dot plots for 1 individual mouse per group from the experiment shown in E. (G) Macrophages (F4/80-positive cells) per ×10 microscopic field. Results represent the average of 3 fields per group. (H) Colon length. (I) Hematocrit. (J) Spleen weight. (K) Average tumor number per mouse. (L) Average tumor size distribution as percentage of total. Green bars represent tumors >16 mm3; red bars represent tumors between 4 and 16 mm3; blue bars represent tumors <4 mm3; no significant difference. (M) Representative colon section Ki-67 IHC (original magnification, ×40). (N) Proliferating cells per crypt (average) determined by counting 50 well-oriented intact crypts with visible lumen per section (n = 5 per group). *P < 0.05.

Figure 1. Global and tissue-specific Sgpl1 KO mouse models.

(A) Conditional KO strategy. A targeting vector introduces a lacZ reporter 5′ to exon 10, resulting in an Sgpl1 truncation. The reporter was deleted by crossing mice with FLPR transgenic mice. The resulting Sgpl1 allele has “floxed” exons 10–12 that can be excised by Cre recombinase. The asterisks indicate that exon 11 contains a critical enzyme cofactor-binding site. (B) mRNA expression of Sgpl1 in tissues of Sgpl1fl/fl (white bars) and Sgpl1fl/fl global Cre+ transgenic (black bars) mice. (C) IB of representative tissues from Sgpl1fl/fl and Sgpl1fl/fl global Cre+ transgenic KO mice. (D) IB of SPLGutKO tissues. Sgpl1fl/fl mice were crossed with mice expressing Cyp1a1-Cre transgene, which was inducible in gut epithelium by oral treatment with NF. After 14 days, WT noninduced (WT), induced (KO), and NF-treated control (C) mice were euthanized. (E) Jejunum SPL activity. (F) Jejunum S1P levels. (G) Plasma S1P levels (no significant difference). (H) SPL IHC (original magnification, ×10) of SPLGutKO mouse colon and thymus tissues. *P < 0.05.