Tumor induces muscle wasting in mice through releasing extracellular Hsp70 and Hsp90

Abstract

Cachexia, characterized by muscle wasting, is a major contributor to cancer-related mortality. However, the key cachexins that mediate cancer-induced muscle wasting remain elusive. Here, we show that tumor-released extracellular Hsp70 and Hsp90 are responsible for tumor's capacity to induce muscle wasting. We detected high-level constitutive release of Hsp70 and Hsp90 associated with extracellular vesicles (EVs) from diverse cachexia-inducing tumor cells, resulting in elevated serum levels in mice. Neutralizing extracellular Hsp70/90 or silencing Hsp70/90 expression in tumor cells abrogates tumor-induced muscle catabolism and wasting in cultured myotubes and in mice. Conversely, administration of recombinant Hsp70 and Hsp90 recapitulates the catabolic effects of tumor. In addition, tumor-released Hsp70/90-expressing EVs are necessary and sufficient for tumor-induced muscle wasting. Further, Hsp70 and Hsp90 induce muscle catabolism by activating TLR4, and are responsible for elevation of circulating cytokines. These findings identify tumor-released circulating Hsp70 and Hsp90 as key cachexins causing muscle wasting in mice.Cachexia affects many cancer patients causing weight loss and increasing mortality. Here, the authors identify extracellular Hsp70 and Hsp90, either in soluble form or secreted as part of exosomes from tumor cells, to be responsible for tumor induction of cachexia.

Fig. 1

Cancer cachexia co-exists with high circulating levels of Hsp70 and Hsp90 associated with tumor cell-released EVs. a Cachexia-inducing tumor cells constitutively release high levels of Hsp70 and Hsp90. Hsp70 and Hsp90 content in conditioned media of cachexia-inducing tumor cells (LLC, C26, H1299, BxPC3 and AGS) are compared with that of non-cachexia-inducing tumor cell EL4 and non-tumorigenic cells (NL20, HPDC and C2C12 myotubes). All cells were incubated for 48 h in regular media (RPMI 160 or DMEM supplemented with 10% fetal bovine serum) and then analyzed for Hsp70 and Hsp90 content by ELISA. Data (n = 3) were analyzed by analysis of variance (ANOVA). b Cachexia co-exists with elevated circulating Hsp70 and Hsp90 in tumor-bearing mice. Hsp70 and Hsp90 levels in the sera of PBS-injected control, EL4 tumor- or LLC tumor-bearing C57BL/6 mice on day 21 of tumor implant, and Apc^min/+ mice that were pre-cachectic (12 weeks of age) or cachectic (20 weeks of age) were determined by ELISA. Data (n = 5) were analyzed by ANOVA (for C57BL/6 mice) or Student t-test (for Apc^min/+ mice). * denotes a difference among bracketed C57BL/6 mice or with pre-cachectic Apc^min/+ mice (P < 0.05). c Particle size of EVs isolated from LCM and the serum of cachectic LLC tumor-bearing mice were analyzed using ZetaView® Nanoparticle Tracking Analyzer. d Tumor-released Hsp70 and Hsp90 are associated with CD9/TSG101/AchE-positive EVs. EVs were isolated from LCM and subjected to immunoprecipitation using CD9 antibody with pre-immune IgG as control. EVs, precipitated pellet and supernatant were analyzed with western blotting. e Cachexia-inducing tumor cells constitutively release high levels of EVs. EVs were isolated from diverse tumor cell-conditioned media and quantified by AchE activity. Data (n = 3) were analyzed by ANOVA. f Cachexia is associated with elevated circulating EVs in tumor-bearing mice. EV levels in serum of indicated mice were determined. Data (n = 5) were analyzed by ANOVA (for C57BL/6 mice) or Student t-test (for Apc^min/+ mice). * denotes a difference among bracketed C57BL/6 mice or pre-cachectic Apc^min/+ mice (P < 0.05)

Fig. 2

Elevated circulating Hsp70 and Hsp90 are critical to LLC tumor-induced catabolic response and muscle wasting. a Neutralizing antibodies of Hsp70 and Hsp90 block tumor cell-conditioned media-induced upregulation of atrogin1 and UBR2 in myotubes. Primary rat myotubes (left) or C2C12 myotubes (right) were treated for 8 h with NCM (NL20-conditioned medium), LCM (LLC-conditioned medium) or CCM (C26-conditioned medium) that were pre-incubated with pre-immune IgG (control) or neutralizing antibodies against Hsp70 and/or Hsp90 (100 ng/ml each) as indicated. Levels of atrogin1 and UBR2 in cell lysate were analyzed by western blotting. Data (n = 3) were analyzed by analysis of variance (ANOVA). * denotes a difference (P < 0.05) from NCM-treated control. ^# denotes a difference between bracketed columns. b Neutralizing antibodies of Hsp70 and Hsp90 block myotube atrophy induced by tumor cell-conditioned media. C2C12 myotubes were treated as described in a and analyzed for p38 MAPK activation (at 1 h) and MHC loss (at 72 h) by western blotting, and myotube diameter of MHC-stained myotubes (at 72 h). Data (n = 3) were analyzed by ANOVA. * denotes a difference (P < 0.05) from NCM-treated control. ^# denotes a difference tween bracketed columns. c Neutralizing antibodies of Hsp70 and Hsp90 prevent muscle catabolism in LLC tumor-bearing mice. Mice implanted with LLC cells were administered neutralizing antibodies against Hsp70 and Hsp90 (or pre-immune IgG as control) subcutaneously through osmotic pumps from day 7 at 3 μg/day each for 14 days and analyzed on day 21 for the activity of catabolic pathways in TA. Data (n = 5) were analyzed by ANOVA. * denotes a difference (P < 0.05). d Neutralizing antibodies of circulating Hsp70 and Hsp90 in LLC tumor-bearing mice prevent muscle wasting. Mice derived above in c were further analyzed for tumor volume and parameters of muscle wasting. Body weight baseline was day 0 of tumor cell implant. Scale bar, 100 μm. Data (n = 5) were analyzed by ANOVA or χ ² analysis (for CSA). * denotes a difference (P < 0.05)

Fig. 3

Elevated circulating Hsp70 and Hsp90 are critical to the development of muscle wasting in Apc^min/+ mice. a Neutralizing antibodies of Hsp70 and Hsp90 block the activation of catabolic pathways in the muscle in Apc^min/+ mice. Pre-cachectic Apc^min/+ mice (16 weeks of age) were administered with neutralizing antibodies against Hsp70 and Hsp90 or pre-immune IgG subcutaneously via osmotic pumps at 3 μg/day each for 4 weeks. Wild-type mice in the same background (C57BL/6) were administered PBS for comparison. At 20 weeks of age the mice were killed and analyzed for activity of the catabolic pathways in TA. b Neutralizing antibodies of Hsp70 and Hsp90 attenuated muscle wasting in Apc^min/+ mice. Mice derived from a were further evaluated for muscle wasting. Body weight baseline was at 12 weeks of age. Scale bar, 100 μm. Data (n = 5) were analyzed by ANOVA or χ ² analysis (for CSA). * denotes a difference (P < 0.05)

Fig. 4

Administration of Hsp70 and Hsp90 causes muscle wasting by activating muscle catabolism. a Exogenous Hsp70 and Hsp90 induce a catabolic response in myotubes. C2C12 myotubes were treated with rHsp70 and/or rHsp90 (100 ng/ml each, controlled with PBS) as indicated, and LCM for comparison. Activation of catabolic pathways was analyzed as described above. Data (n = 3) were analyzed by analysis of variance. * denotes a difference from NCM-treated control (P < 0.05). b Systemically administered Hsp70 and Hsp90 activate muscle catabolism in mice. C57BL/6 mice were injected (i.p.) with rHsp70 and rHsp90 (each at 100 μg/kg/3 days, controlled with PBS) for five times. On day 15, activity of catabolic pathways in TA was analyzed. Data (n = 6) were analyzed by Student t-test. * denotes a difference from PBS-treated control (P < 0.05). c Systemically administered Hsp70 and Hsp90 cause muscle wasting in mice. Mice from b were further analyzed for muscle wasting. Scale bar, 100 μm. Data (n = 6) were analyzed by Student t-test or χ ² analysis (for CSA). * denotes a difference from NCM- or PBS-treated control (P < 0.05)

Fig. 5

Tumor-induced muscle wasting depends on tumor cell-expressed Hsp70 and Hsp90. a Conditioned medium of Hsp70- and Hsp90-deficient LLC cells do not induce myotube catabolism. C2C12 myotubes were treated with LCM from LLC cells transfected with random or Hsp70/Hsp90-specific siRNA as indicated. Catabolic response was evaluated as described in Fig. 2. Data (n = 3) were analyzed by analysis of variance (ANOVA). * denotes a difference (P < 0.05). b Serum Hsp70 and Hsp90 levels in mice bearing Hsp70- and Hsp90-deficient LLC tumor remain unchanged. LLC cells were transduced with lentivirus encoding Hsp70 and Hsp90-specific shRNA, or lentivirus carrying empty vector as control. Mice were implanted with Hsp70&90-deficient or control LLC cells. On day 21 serum levels of Hsp70 and Hsp90 were analyzed after Brij98 treatment. Data (n = 6) were analyzed by ANOVA. * denotes a difference (P < 0.05). c Hsp70&90-deficient LLC tumor does not induce muscle catabolism in mice. TA of mice derived from b was analyzed by western blotting for markers of catabolic pathways. Data (n = 6) were analyzed by ANOVA. * denotes a difference (P < 0.05). d Hsp70&90-deficient LLC tumor do not cause muscle wasting in mice. Mice derived from b were further analyzed for muscle wasting. Scale bar, 100 μm. Data (n = 6) were analyzed by ANOVA or χ ² analysis (for CSA). * denotes a difference (P < 0.05)

Fig. 6

Tumor cell-released Hsp70/90-expressing EVs induce muscle catabolism in myotubes. a CD9-positive EVs induce catabolic response in C2C12 myotubes. EVs isolated from LCM were subjected to immunoprecipitation (IP) against CD9 with pre-immune IgG as control. Resulted pellet and supernatant (Sup) were used to treat C2C12 myotubes and compared with PBS and input EVs for catabolic activity using western blotting. Data (n = 3) were analyzed by analysis of variance (ANOVA) and * denotes a difference from PBS treatment (P < 0.05). b Hsp70/90 release by cachexia-inducing tumor cells is dependent on EV release. Rab27a and Rab27b were knocked down by transfecting LLC and C26 cells with control or specific siRNAs. EVs and Hsp70/90 released into conditioned media were quantified after Brij98 treatment. EVs were quantified by AchE. Data (n = 3) were analyzed by Student t-test. c Tumor cell-induced catabolic response in myotubes is dependent on EV release. C2C12 myotubes were treated with conditioned media of Rab27-deficient LLC or C26 cells and analyzed for p38 MAPK activity (at 1 h), atrogin1 (at 8 h) and MHC levels (72 h). Conditioned medium of non-tumorigenic NL20 cells (NCM) was used as control. Data (n = 3) were analyzed by ANOVA. * denotes a difference (P < 0.05) from NCM-treated control. ^# denotes a difference between bracketed columns

Fig. 7

Tumor cell-released Hsp70/90-expressing EVs are critical to the development of muscle wasting in mice. a Elevation of serum Hsp70/90 in LLC tumor-bearing mice is dependent on EV release. LLC cells transduced with lentivirus encoding Rab27a- and Rab27b-specific shRNA or empty vector were analyzed for knockdown effect by western blotting (top). Mice were then implanted with Rab27-deficient or control LLC cells. In 21 days serum levels of EVs, Hsp70/90 were analyzed. b Rab27-deficient LLC tumor does not induce muscle catabolism in mice. Mice derived from a were analyzed for activity of the catabolic markers in TA. c Rab27-deficient LLC tumor does not induce muscle wasting in mice. Mice derived from a were analyzed for muscle wasting. Scale bar, 100 μm. Data (n = 6) were analyzed by analysis of variance or χ ² analysis (for CSA). * denotes a difference (P < 0.05)

Fig. 8

TLR4 mediates Hsp70/Hsp90-induced catabolic response in myotubes. a Hsp70 and Hsp90-induced catabolic response in myotubes is dependent on TLR4. C2C12 myoblasts were transfected with siRNA as indicated. After differentiation, myotubes were treated with rHsp70 and rHsp90 and analyzed for catabolic response as described above. b LLC-released EVs induce muscle catabolism in C2C12 myotubes through TLR4. C2C12 myotubes with either TLR2 or TLR4 knockdown were treated with EVs isolated from LCM (AchE 6 mU) and analyzed for catabolic response. c CD9-positive EVs activate TLR4 in reporter cells. EVs isolated from LCM were subjected to immunoprecipitation (IP) using an antibody against CD9 with pre-immune IgG as control. Resulting pellet and supernatant were used to treat TLR4 reporter cell line HEK-Blue hTLR4 for 24 h, and compared with original EVs. TLR4 activation was measured as SEAP activity. Data (n = 3) were analyzed by analysis of variance. * denotes a difference (P < 0.05)

Fig. 9

TLR4 is critical to Hsp70/Hsp90-induced muscle wasting in mice. a Hsp70 and Hsp90-induced muscle catabolism in mice is dependent on TLR4. Wild-type and TLR4 null mice were injected with rHsp70 and rHsp90 and analyzed for catabolic response in TA as described in Fig. 4. b Hsp70 and Hsp90-induced muscle wasting is TLR4-dependent. Wild-type and TLR4 null mice injected with rHsp70 and rHsp90 are further analyzed for muscle wasting. Scale bar, 100 μm. Data (n = 5) were analyzed by analysis of variance or χ ² analysis (for CSA). * denotes a difference (P < 0.05)

Fig. 10

Tumor-released Hsp70 and Hsp90-expressing EVs mediate systemic increase of inflammatory cytokines. Sera from mice bearing LLC tumor with or without knockdown of Rab27 on days 0, 7, 14 and 21 of tumor cell implant were analyzed for levels of Hsp70, Hsp90 and EVs as described above, and TNFα, IL6 and IL-1β using Bio-Plex^® Multiplex Immunoassays. For Hsp70/90 measurement sera were treated with Brij98 before ELISA. Data (n = 6) were analyzed by analysis of variance. * denotes a difference (P < 0.05)