Tumor-derived IL-6 and trans-signaling among tumor, fat, and muscle mediate pancreatic cancer cachexia

Abstract

Most patients with pancreatic adenocarcinoma (PDAC) suffer cachexia; some do not. To model heterogeneity, we used patient-derived orthotopic xenografts. These phenocopied donor weight loss. Furthermore, muscle wasting correlated with mortality and murine IL-6, and human IL-6 associated with the greatest murine cachexia. In cell culture and mice, PDAC cells elicited adipocyte IL-6 expression and IL-6 plus IL-6 receptor (IL6R) in myocytes and blood. PDAC induced adipocyte lipolysis and muscle steatosis, dysmetabolism, and wasting. Depletion of IL-6 from malignant cells halved adipose wasting and abolished myosteatosis, dysmetabolism, and atrophy. In culture, adipocyte lipolysis required soluble (s)IL6R, while IL-6, sIL6R, or palmitate induced myotube atrophy. PDAC cells activated adipocytes to induce myotube wasting and activated myotubes to induce adipocyte lipolysis. Thus, PDAC cachexia results from tissue crosstalk via a feed-forward, IL-6 trans-signaling loop. Malignant cells signal via IL-6 to muscle and fat, muscle to fat via sIL6R, and fat to muscle via lipids and IL-6, all targetable mechanisms for treatment of cachexia.

Figure 1.

IL-6 protein expression is associated with increased cachexia severity and mortality in a mouse xenograft model of human PDAC. (A) Experimental outline for the generation of the mouse xenograft model (PDX) of human PDAC. (B) Changes in body weight of human patients with PDAC and their corresponding xenograft mouse avatar. (C–E) Implantation of human tumors into mice shows correlations between muscle loss and mortality (C), mIL-6 and mortality (D), and mIL-6 and muscle loss (E). Mice with the highest levels of mIL-6 also had increased human IL-6 (D;S002, S035, S017). Data shown are mean ± SD, where n = 6–9 mouse avatars per patient tumor. Statistical differences (**, P < 0.005) and r values were determined using Pearson correlation coefficient analysis. (F and G) Human PDAC tumors obtained from US Biomax were reacted for IL-6 using IHC. (F) Tumor sections were classified as having either high or low expression of IL-6 specifically in tumor epithelial cells (arrows). Of the 72 tumors, 40 had low tumor cell IL-6 expression, and 32 had high tumor cell IL-6 expression. (G) Increased magnification to show PDAC tumor cell IL-6 expression, with arrows added to show tumor cells with weak IL-6 expression (top) and strong IL-6 expression (bottom). Scale bar = 40 µm. WL, weight loss.

Figure S1.

Body condition scoring for cachexia avatar studies in Fig. 1, as described. (A) Mice were euthanized when a majority reached a total score of 2. (B) Expression levels of IL6 mRNA across multiple human pancreatic cancer cell lines from Illumina BaseSpace Correlation Engine illustrating heterogeneity in IL6 expression. Blue bar indicates median Il6 expression for each cell line; yellow line indicates median Il6 expression for the entire group; error bars indicate SD where applicable.

Figure 2.

Deletion of IL-6 from KPC cells prevented muscle wasting in vitro and increased survival in mice. (A) Targeted mutagenesis of the Il6 gene was performed, and a transfection control clone (KPC) and an Il6 ablated clone (KPC IL6^KO) were selected for use in downstream experiments based on their Il6 expression relative to the untransfected parental cell line (KPC-P). Clones were cultured individually in triplicate for n = 3 per group, and error bars are mean plus SD. (B) To determine if deletion of IL-6 affected tumor cell growth, a proliferation assay was performed comparing the clones, where clones were cultured in triplicate (n = 3 per group) and measurements were taken every hour for 100 h; growth curves represent the mean proliferation of triplicate wells over time. (C) Myoblasts were differentiated into myotubes and treated in triplicate (n = 3 per group) with 30% CM from tumor clones to measure effects on myotube atrophy and the activation of STAT3. (D) Western blotting (WB) analysis using the pooled myotube protein lysates from n = 3 per group was performed to measure phosphorylation of STAT3 (p-STAT3) as an indication of STAT3 activation. (E and F) Myotubes were visualized using IF with an anti-myosin heavy chain (MHC) antibody (E), and myotube atrophy was measured from 20 random fields per well (n∼150 myotube diameter measurements per well, n = 3 wells per group). Scale bar = 50 µm. Error bars represent mean myotube diameter and SD, and significant differences (**, P < 0.001) were determined using ANOVA and Tukey’s multiple comparisons test (F). (G and H) To verify atrophy was influenced by IL-6, myotubes were treated in triplicate with KPC CM and IL-6 neutralizing antibody as well as KPC IL6^KO CM plus recombinant IL-6 with and without the presence of an anti–IL-6 neutralizing antibody. Myotubes were visualized with MHC IF, and diameter was measured as described in F. Scale bar = 50 µm. Error bars are mean and SD, and significant differences (*, P < 0.05; **, P < 0.001) were determined using ANOVA and Tukey’s multiple comparisons test. (I) KPC tumor cells were orthotopically implanted into mice, and tumors were excised, sectioned, and reacted with anti–IL-6 IHC to verify tumor cell IL-6 deletion. Insets are increased magnification to show staining; scale bar = 20 µm. (J) Survival was measured in mice orthotopically implanted with KPC and KPC IL6^KO tumor cells. The dashed line represents median survival, n = 15 per group, and statistical difference (**, P < 0.001) was determined using a Kaplan-Meier estimate. All panels represent data from single experiments. Diff. media, DM.

Figure S2.

Generation of KPC IL6^KO cells. (A) CRISPR gRNA plasmid map. (B) DNA sequencing was performed at the CRISPR/Cas9 target site (green line) within the mIl6 gene sequence, showing an insertion mutation of 45 nucleotides (red line) into the KPC IL6^KO mutant sequence. Asterisks indicate point mutations.

Figure 3.

Deletion of tumor cell IL-6 attenuates muscle wasting. (A) KPC tumor–bearing mice reached our humane endpoint criteria 17 d after injection, and all groups were euthanized simultaneously. Skeletal muscles and the heart were excised at euthanasia and weighed and normalized to initial body weight (IBW). Data represent the mean plus SD from two individual experiments (n = 13 per group); significant differences (**, P < 0.005; ***, P < 0.0005) were determined using ANOVA and Tukey’s multiple comparisons test. (B and C) Evaluation of muscle fiber CSA was done using sections from excised quadriceps muscles from individual mice reacted for dystrophin expression (B), and mean fiber CSA was determined by comparing myofiber means from muscle cross sections from a subset of samples (n = 4; n > 300 myofibers per animal), quantified from 20 random fields from each cross section (C). Scale bar = 50 µm. Error bars are mean and SD, and significant differences (*, P < 0.05) were determined using ANOVA and Tukey’s multiple comparisons test. (D) Cumulative fiber CSA values from mice in each group were organized based on percent distribution of fiber CSA and plotted to observe shifts in distribution. (E) Tumors were excised, and tumor mass was recorded. Data represent the mean plus SD from two individual experiments (n = 13 per group); significant differences (**, P < 0.005) were determined using Student’s t test. (F and G) Correlation analysis of tumor size and muscle size for the quadriceps and heart showed no correlations.

Figure S3.

Deletion of tumor cell–derived IL-6 is associated with reduced weight loss, carcass and liver wasting, and splenomegaly in mice. (A–D) Tumor-free final body weight was measured at euthanasia (A), and tissue weights for the carcass (B), liver (C), and spleen (D) were also recorded. (E–G) No correlation was observed between tumor mass and gastrocnemius (E), tibialis (F), and epididymal fat pad (G) masses after analysis using the Pearson R coefficient. Data are from two individual experiments (n = 13 per group). (A–D) Data represent the mean and SD, and statistical differences (*, P < 0.05; **, P < 0.01; ***, P < 0.001) were determined using ANOVA and Tukey’s multiple comparisons test. (H) Sections of excised KPC (n = 5) and KPC IL6^KO (n = 5) tumors were stained with H&E and blindly scored by two separate pathologists to determine any differences in tumor grade; N/A indicates no observation. IBW, initial body weight.

Figure 4.

Measurement of protein expression for common molecular pathways associated with muscle wasting. (A–H) Western blotting results evaluating muscle proteins involved in ubiquitination (A and B, left), autophagy (C and D), anabolism (E and F), and mitochondria biogenesis and metabolism (G and H) from protein lysates made from quadriceps harvested at euthanasia. Results are from a single experiment with no tumor (n = 5), KPC (n = 6), and KPC IL6^KO (n = 5) tumor mice, and statistical differences (*, P < 0.05; **, P < 0.01; ***, P < 0.001) were determined using ANOVA and Tukey’s multiple comparisons test. Analysis of qPCR results for mRNA expression of E3 ubiquitin ligases Atrogin-1 and Murf1 in quadriceps was also performed (B, right). Data are from a single experiment with n = 5 per group, and statistical differences (***, P < 0.0001) were determined with ANOVA and Tukey’s multiple comparisons test. Error bars are mean and SD.

Figure 5.

Deletion of IL-6 from KPC cells reduces activation of key cachexia pathways in muscle. (A) Isolated RNA from the quadriceps of no tumor (n = 3), KPC tumor (n = 4), and KPC IL6^KO tumor (n = 4) mice was sequenced, and differentially regulated genes (FC ≥ |1.5| and FDR ≤ 0.05) were compared across groups. (B–F) Ingenuity Pathway Analysis using quadriceps RNA-seq data identified various altered pathways and their associated genes (shown in heat map format) that have roles in muscle wasting, including oxidative stress (B), adipogenesis (C), inflammation (D), FA oxidation (E), and glycolysis (F). The scale bar illustrates increased (red) and decreased (blue) gene expression for the heat maps. (G and H) Measurements of muscle lipids using ORO staining (G, top, and H, left) and SDH activity as a marker for mitochondria oxidative capacity (G, bottom, and H, right) were performed on quadriceps muscle cross sections (n = 5 per group). Scale bar = 50 µm; black arrows in G indicate fibers with increased lipid accumulation and aberrant SDH reactivity at lower magnification; yellow arrows in G indicate fibers with increased lipid accumulation and aberrant SDH reactivity in higher magnification insets. Error bars represent mean and SD, and significant differences (*, P < 0.05; ***, P < 0.0001) were determined using ANOVA and Tukey’s multiple comparisons test.

Figure 6.

Deletion of IL-6 from KPC cells reduces fat wasting and alters affected molecular pathways and upstream regulators in adipose tissue. (A) Epididymal fat pad mass was measured at euthanasia and normalized to initial body weight (IBW). Data represent the mean and SD from two separate experiments (n = 13 per group), and statistical differences (**, P < 0.01; ***, P < 0.0001) were determined using ANOVA and Tukey’s multiple comparisons test. FFA, free fatty acid. (B and C) Characterization of lipolysis using measurements of plasma glycerol (B) and FAs (C) normalized to epididymal fat pad mass from mice. Data represent mean and SD from a single experiment (n = 5 per group), and statistical differences (*, P < 0.05) were determined using ANOVA and Tukey’s multiple comparisons test. (D) Isolated RNA from the epididymal fat pads (n = 4 per group) was sequenced, and differentially regulated genes (compared with no tumor group, FC ≥ |1.5| and FDR ≤ 0.05) were compared across groups. (E) iPathway analysis using epididymal fat pad RNA-seq data identified differentially regulated pathways. (F) iPathway also identified upstream regulators in the adipose tissue of KPC and KPC IL6^KO tumor mice. Red arrows labeled with A indicate an activating interaction, while blue arrows labeled with E indicate an inhibitory interaction. (G) Deconvolution analysis of the RNA-seq data identified changes in cell subtypes present in the adipose tissue; mast, mast cells; B, B cells; a, P < 0.05 versus no tumor; b, P < 0.10 versus no tumor; and c, P < 0.05 versus KPC.

Figure S4.

Characterization of changes in signaling pathways and immune cell populations using RNA-seq from adipose tissue in tumor-bearing mice. (A–D) Analysis of RNA-seq data from adipose tissue usingIngenuity Pathway Analysis, which identified various altered pathways and their associated genes (shown in heat map format) that have roles in inflammation and lipolysis, including granulocyte adhesion/diapedesis (A), liver X receptor/retinoid X receptor (LXR/RXR) activation (B), acute phase signaling (C), and IL-6 signaling (D). (E) A list of immune cell subtypes and their identifying gene sets that are differentially regulated in adipose tissue in the presence of PDAC as determined using deconvolution analysis of adipose RNA-seq data. The scale bar illustrates increased (red) and decreased (blue) gene expression compared with no tumor mice for the heat maps. Data represent single experiments.

Figure 7.

Evidence for an IL-6, IL6R circuit among tumor, adipose tissue, and skeletal muscle in PDAC cachexia. (A and B) Plasma from no tumor, KPC tumor, and KPC IL6^KO tumor mice was harvested immediately before euthanasia and measured for IL-6 (A) and IL6R (B) protein expression using an ELISA. Data represent mean and SD from two separate experiments (n = 13 per group), and statistical differences (*, P < 0.05; **, P < 0.001; ***, P < 0.0001) were determined using ANOVA and Tukey’s multiple comparisons test. (C and D) Isolated RNA from quadriceps, epididymal fat, and liver of mice was used to measure mRNA expression of Il6 (C) and Il6r (D) in tissues and are presented as FC versus no tumor mice. Data represent mean and SD from a single experiment, and statistical differences (*, P < 0.05; **, P < 0.01; ***, P < 0.001) were determined using ANOVA and Tukey’s multiple comparisons test. (E and F) Isolated RNA from KPC (n = 6) and KPC IL6^KO (n = 5) tumors was used to measure mRNA expression of Il6 (E) and Il6r (F) in tumors and are presented as FC versus KPC tumors. Data represent mean and SD from a single experiment, and statistical differences (*, P < 0.05; **, P < 0.01) were determined using ANOVA and Tukey’s multiple comparisons test. (G and H) To determine with increased detail the source of Il6 and Il6r mRNA in muscle and fat, mononuclear (MNC) cell fractions were dissociated from whole tissue, and isolated RNA from the MNC fraction was measured for Il6 (G) and Il6r (H) mRNA expression. Data represent mean and SD from a single experiment with n = 4 per group, and statistical differences (**, P < 0.01) were determined using Student’s t test. (I–K) Protein expression for IL6R and STAT3 phosphorylation (p-STAT3) was quantified in the quadriceps and epididymal fat pads of no tumor (n = 5), KPC (n = 6), and KPC IL6^KO (n = 5) mice using Western blotting. Data represent the mean and SD for a single experiment, and statistical differences (*, P < 0.05; **, P < 0.01; ***, P < 0.001) were determined using ANOVA and Tukey’s multiple comparisons test. (L) IF showing the expression of IL6R protein (red) and nuclei (blue, DAPI) in the quadriceps muscle fibers of mice (L, top; scale bar = 50 µm) and IL6R protein (red), the endothelial protein marker CD31 (green), and nuclei (blue, DAPI) to visualize IL6R expression in and around blood vessels in the quadriceps muscle of mice (L, bottom; scale bar = 100 µm). (M) IHC was used to visualize IL-6 protein localization in the epididymal fat pads. Scale bar = 50 µm.

Figure S5.

Tissue Il6 and Il6r mRNA expression using mean Ct from qPCR and the GTEx Portal database support fat as a significant source of Il6 and muscle as a significant source of Il6r. (A) Comparison of Il6 mean Ct from qPCR analysis between groups for muscle, fat, and liver shows fat from KPC tumor mice with the lowest mean Ct, indicating the highest expression of Il6 among tissues analyzed. (B) Comparison of Il6r mean Ct from qPCR analysis between groups for muscle, fat, and liver shows muscle from KPC tumor mice with the lowest mean Ct, indicating the highest expression of Il6r among tissues analyzed. In A and B, error bars are mean plus SD. (C) Data represent single experiments. Human Il6 mRNA expression across various tissues according to the GTEx Portal Database shows adipose tissues as having the highest Il6 expression. (D) Human Il6r mRNA expression across various tissues according to the GTEx Portal Database shows skeletal muscle as having the highest Il6r expression.

Figure 8.

Modeling of IL-6, IL6R tumor–tissue crosstalk in vitro. (A) C2C12 myotubes were incubated with DM (Control) or CM from KPC tumor cells in triplicate for 48 h, and myotube diameter was measured. (B) Myotubes were incubated in triplicate with Control or KPC CM for 12, 24, and 96 h, and isolated RNA was used to measure mRNA expression of Il6 and Il6r at each time point. (C) 3T3 adipocytes were incubated in triplicate with KPC CM for 1 and 3 h, and media glycerol concentration was measured as a marker of lipolysis. (D) 3T3 adipocyte RNA was harvested at 1 and 3 h after treatment in triplicate with KPC CM and used to measure Il6 and Il6r mRNA expression. (E) The tumor–fat–muscle crosstalk was investigated by incubating adipocytes with KPC CM and then using the adipocyte CM to treat myotubes in triplicate for 48 h and measure myotube diameter. IF was used to measure myotube diameter from 20 random fields per well (n ∼150 diameter measurements per well, n = 3 wells). (A–E) Data represent the mean and SD from single experiments, and statistical differences (*, P < 0.05; **, P < 0.01) were determined using Student’s t test between groups at individual time points. (F) The tumor–muscle–fat crosstalk was investigated by treating myotubes for 12, 24, and 96 h with KPC CM and then using the myotube CM to treat adipocytes in triplicate for 1 and 3 h to measure lipolysis via media glycerol content. (G) To decipher the effects of IL-6 and IL6R in muscle and fat, myotubes and adipocytes were treated in triplicate in vitro with recombinant IL-6 and IL6R. Myotubes were incubated with DM (Diff Media), physiological levels of recombinant mIL-6, physiological levels of recombinant murine IL6R, or the combination of IL-6 and IL6R for 48 h, and myotube diameter was measured. Myotube diameter was measured as described in E. (H) Adipocytes were incubated in triplicate with GM (Control [CTL]), physiological levels of recombinant mIL-6, physiological levels of recombinant murine IL6R, or the combination of IL-6 and IL6R for 3 h, and media glycerol was measured. (I) Myotubes were incubated with IL-6 (20 ng/ml), palmitate (Palm; 0.5 mM), and the combination for 48 h, and myotube diameter was measured. Myotube diameter was measured as described in E. (F–I) Data represent the mean and SD from single experiments, and statistical differences (*, P < 0.05; **, P < 0.01; ***, P < 0.001) were determined using ANOVA and Tukey’s multiple comparisons test.

Figure 9.

Model of tumor–fat–muscle crosstalk in PDAC. Summary of the crosstalk between tumor, muscle, and fat in PDAC-associated cachexia. IL-6 produced by tumor epithelial cells acts to exacerbate both fat and muscle wasting, leading to myosteatosis and systemic inflammation as a result of increased lipolysis and production of IL-6 and FAs by fat. Production of IL6R by muscle is increased in the presence of PDAC and contributes to elevated sIL6R levels in plasma. Ultimately, the increases in both FAs and sIL6R in plasma act in a feed-forward mechanism, where muscle and fat contribute to each other’s wasting in PDAC cachexia.