letters

Excessive fatty acid oxidation induces muscle atrophy in cancer cachexia

© 2016 Nature America, Inc. All rights reserved.

Tomoya Fukawa1–3, Benjamin Chua Yan-Jiang4, Jason Chua Min-Wen4, Elwin Tan Jun-Hao4, Dan Huang2, Chao-Nan Qian5, Pauline Ong1,2, Zhimei Li2, Shuwen Chen6, Shi Ya Mak6, Wan Jun Lim7, Hiro-omi Kanayama3, Rosmin Elsa Mohan8, Ruiqi Rachel Wang8, Jiunn Herng Lai9, Clarinda Chua4,7, Hock Soo Ong10, Ker-Kan Tan11, Ying Swan Ho6, Iain Beehuat Tan4,7,12, Bin Tean Teh1,2,7,13–15 & Ng Shyh-Chang4 Cachexia is a devastating muscle-wasting syndrome that occurs in patients who have chronic diseases. It is most commonly observed in individuals with advanced cancer1,2, presenting in 80% of these patients, and it is one of the primary causes of morbidity and mortality associated with cancer3–5. Additionally, although many people with cachexia show hypermetabolism3,6, the causative role of metabolism in muscle atrophy has been unclear. To understand the molecular basis of cachexia-associated muscle atrophy, it is necessary to develop accurate models of the condition. By using transcriptomics and cytokine profiling of human muscle stem cell–based models and human cancerinduced cachexia models in mice, we found that cachectic cancer cells secreted many inflammatory factors that rapidly led to high levels of fatty acid metabolism and to the activation of a p38 stress-response signature in skeletal muscles, before manifestation of cachectic muscle atrophy occurred. Metabolomics profiling revealed that factors secreted by cachectic cancer cells rapidly induce excessive fatty acid oxidation in human myotubes, which leads to oxidative stress, p38 activation and impaired muscle growth. Pharmacological blockade of fatty acid oxidation not only rescued human myotubes, but also improved muscle mass and body weight in cancer cachexia models in vivo. Therefore, fatty acid–induced oxidative stress could be targeted to prevent cancer-induced cachexia. To identify tractable molecular alterations that contribute to muscle atrophy, we generated a human cancer model of cachexia in vivo. On the basis of a mini-screen against five human kidney cancer cell lines, we decided to base our model on the highly cachectic RXF393

(abbreviated RXF) cell line. As compared to other cancer cell lines, RXF accelerated the mortality of mice (Fig. 1a), despite the injection of equal numbers of cells and the relatively less aggressive nature of RXF tumors (Fig. 1b). Indeed, RXF cells induced body-weight loss of ~30%, whereas body weight grew in the presence of all other cancer cell lines (Fig. 1c). SKRC39 (abbreviated SKR) cells were chosen as the noncachectic control, because they resulted in the best prognosis. Histological analysis revealed that RXF-bearing mice sustained severe skeletal muscle atrophy, as compared to the noncachectic SKR-bearing mice (Fig. 1d). Myofiber cross-sectional area in the quadriceps (mixed with type 1 and 2 myofibers) was significantly smaller in RXFbearing mice than in SKR-bearing mice (Fig. 1e and Supplementary Fig. 1a). Expression of the master regulator of myogenesis, Myod1, was significantly lower in the quadriceps of RXF-bearing mice than in those of noncachectic mice (Fig. 1f), which is consistent with previous findings7,8. To recapitulate these effects with a human muscle model in vitro, we derived myotubes from human muscle stem cells isolated from biopsied specimens, and we cultured them in media conditioned with either SKR cells or RXF cells. We observed that early myotubes that were exposed to cachectic RXF media, but not to SKR media, manifested growth defects (Fig. 1g and Supplementary Fig. 1b,c). In addition, Myod1 and pan-myosin heavy chain (MHC) protein expression were lower in early myotubes exposed to the cachectic RXF media than in those exposed to SKR media (Fig. 1h,i and Supplementary Fig. 1d). However, not all differentiation markers were affected; myogenin, α-actinin and fast-twitch MHC protein expression, for example, all remained unchanged in myotubes exposed to RXF media, as compared to SKR media (Supplementary Fig. 1d).

1Laboratory

of Cancer Therapeutics, Program in Cancer and Stem Cell Biology, Duke-National University of Singapore Medical School, Singapore. 2Laboratory of Cancer Epigenome, Division of Medical Science, National Cancer Centre Singapore, Singapore. 3Department of Urology, Institute of Biomedical Sciences, Tokushima University Graduate School, Tokushima, Japan. 4Genome Institute of Singapore, Agency for Science Technology and Research, Singapore. 5Sun Yat-sen University Cancer Center, State Key Laboratory of Oncology in South China, Collaborative Innovation Center for Cancer Medicine, Guangzhou, China. 6Bioprocessing Technology Institute, Agency for Science Technology and Research, Singapore. 7Division of Medical Oncology, National Cancer Centre Singapore, Singapore. 8School of Mechanical and Aerospace Engineering, Nanyang Technological University, Singapore. 9Department of Colorectal Surgery, Singapore General Hospital, Singapore. 10Department of Upper Gastrointestinal and Bariatric Surgery, Singapore General Hospital, Singapore. 11Department of Surgery, Yong Loo Lin School of Medicine, National University of Singapore, Singapore. 12Program in Cancer and Stem Cell Biology, Duke-National University of Singapore Medical School, Singapore. 13Cancer Science Institute of Singapore, National University of Singapore, Singapore. 14Institute of Molecular and Cell Biology, Agency for Science Technology and Research, Singapore. 15SingHealth/Duke-National University of Singapore Precision Medicine Institute, National Heart Centre, Singapore. Correspondence should be addressed to I.B.T. ([email protected]), B.T.T. ([email protected]) or N.S.-C. ([email protected]). Received 8 January; accepted 29 March; published online 2 May 2016; doi:10.1038/nm.4093

nature medicine  advance online publication



letters

RXF (cachectic)

120

d

e

C 39 XF 39 3 R

A4 98

R SK

C 39 XF 39 3

100

R

60 80 Time (d)

12 C

40

R

20

SK

P < 0.0001 0

**

A4 98

0

40 30 20 10 0 –10 –20 –30

Change in body weight (%)

20

SN

c

40

12 C

0

SN

Tumor volume (mm3)

1,000 78 6O

Survival (%)

60

**

2,000

1.4

Myofiber area (fold change)

1.2 1.0 0.8 0.6

**

0.4 0.2 0

**

0.4 0.2 0

he RX ct F ic )

R

0.6

ac (c

(c

(c

(c

0.8

ac

SK

0 SK

R

0

1.0

R

0.2

*

0.5

1.2

he RX ct F ic )

0.4

1.0

RXF (cachectic)

SK

**

0.6

1.5

MHC protein (fold change)

Myod1 mRNA (fold change)

0.8

SKR

i

ac

R

1.0

ac

SK

1.2

he RX ct F ic )

*

h

he RX ct F ic )

1.6 1.4 1.2 1.0 0.8 0.6 0.4 0.2 0

hMyotube growth (fold change)

g

Myod1 mRNA (fold change)

f

stress-response signatures, as compared to SKR media (Fig. 2b and Supplementary Table 2). Glycolysis genes within hypoxia signatures were most dramatically downregulated by RXF media, relative to SKR

b

Relative

Row max

SKR

80

3,000

Row min

c

TX

N

R

D 1

14

3K

PB

AP

M

C EB

M

AF

F

B

JU N

K2

PD

M KN

C EB

FO

S

3

8

d

KM

2

PF

I1

AM

PG

K1

TP

PG

3

M 2

PG

EN O

T4

G

LU

1

O

EN

M

PK

AA C S AC O X2 SP H K1 AC SS H SD 1 11 B 1 C D 36 AC A AC DL AD SB FA B P9

T1

O

2

Fold change

e

AC



AT F

3K

L1

XC

C

M AP

1

R

1

1A

IF

EG

TG

6

KN

3K

C D

M AP

4

IT

C XC

D D

L1 3

Fold change

SKR Figure 2  Cachectic media induce ** 32.0 RXF ** stress-response signatures and fatty ** ** ** ** ** * * 8.0 ** ** acid metabolism as the primary * * ** * ** ** 2.0 transcriptional effects in human 0.5 muscle cells. (a) Clustergram heat map of microarray data from human myotubes after 6 h of exposure to SKR versus RXF (cachectic) conditioned SKR SKR media. (b) GSEA graphs for the top 2.0 32.0 RXF RXF ** upregulated signatures (stress response) ** ** ** 1.5 8.0 * ** ** and top downregulated signatures (hypoxia) * ** * ** * * 1.0 ** ** ** ** ** in human myotubes within 6 h of exposure 2.0 0.5 to RXF (cachectic) conditioned media, 0.5 0 relative to SKR media. (c) Inflammatory stress–related genes upregulated by RXF (cachectic) media, as validated by RT-qPCR. (d) Fatty acid–metabolism genes upregulated by RXF (cachectic) media, as validated by RT-qPCR. (e) Glycolysis genes downregulated by RXF (cachectic) media, as validated by RT-qPCR. For all panels, data are expressed as mean ± s.e.m. *P < 0.05 and **P < 0.01 relative to SKR control, as determined by Student’s t test. Fold change

© 2016 Nature America, Inc. All rights reserved.

100

To clarify these observations, we profiled both our in vitro and in vivo models with microarrays. We first profiled the transcriptomes of mouse myotubes that had been exposed to RXF media for 6 h (Supplementary Table 1). Gene-set enrichment analysis (GSEA) revealed that RXF media, relative to SKR media, most dramatically upregulated NF-κB-related signatures, the FoxO3 signature and the autophagosome signature in mouse myotubes. By contrast, hypoxia, RNA processing, mTOR and TGF-β1 signatures were most dramatically downregulated by RXF media, as compared to SKR media, in mouse myotubes. We also profiled human myotubes after exposure to cachectic RXF media (Fig. 2a). By comparison, we found that the cachectic RXF media most dramatically upregulated the

a

b

RXF393 A498 SN12C 786-O SKRC39

78 6O

a

Figure 1  Modeling cancer-induced cachexia with human cancer cells and myotubes. (a) Survival probability of NOD-SCID mice over time, after subcutaneously implanting five kidney cancer cell lines (RXF393 (n = 7); SKRC39 (n = 10); A498 (n = 3); 786-O (n = 3); and SN12C (n = 4)). (b) Volumes of human tumor xenografts at the point of mouse death. (c) Body-weight changes in xenograft-bearing NOD-SCID mice at the point of death, relative to their initial weight. (d) Representative micrographs of H&E histology (n = 10) of quadriceps muscles in RXF393 (RXF)-bearing mice, relative to SKRC39 (SKR)-bearing mice. Scale bars, 200 µm. (e) Quantification of the myofiber cross-sectional areas in RXF-bearing mice versus SKR-bearing mice. (f) Myod1 mRNA expression in matching quadriceps of RXF-bearing mice versus SKR-bearing mice by RT-qPCR, normalized to Actb (n = 3). (g) Quantification of early myotube growth by measuring total nuclei numbers after 6 d of exposure to media conditioned with cachectic RXF cancer cells, as compared to conditioning with noncachectic SKR cancer cells (n = 3). (h) Myod1 mRNA expression in human myotubes after 6 d of exposure to RXF media versus to SKR media (n = 3). (i) Pan-MHC protein expression in human myotubes by western blot after 6 d of exposure to RXF media versus SKR media (n = 3). For all panels, data are expressed as mean ± s.e.m. *P < 0.05 and **P < 0.01 relative to the SKR control, as determined by Student’s t test.

advance online publication  nature medicine

10.0

b

SKR RXF (cachectic)

SKR RXF (cachectic)

*

1.2 Relative level

a Relative level

1.0

TG Fβ

G F M M P2

LI F

VE

Relative

Row min

Row max

SKR

d

RXF

**

**

**

IL-1b

IL-6

IL-8

1.0

**

**

**

0.8 0.6 0.4 0.2 0

TNF-α

LIF

VEGF

2.0

Metabolite levels (fold change)

c

IL -6 TN Fα

0.1

IL -8 IL -1 b

**

1.5

**

SKR RXF

**

**

**

1.0 0.5

2.5

:3 )

:3 )

(5 2 TG

TG

(5 6

(5 4

:6 )

:2 ) (5 4

TG

e

TG

TG

(5 0

:3 )

0

Metabolite levels (fold change)

SKR RXF

**

2.0

**

1.5

**

**

*

1.0 0.5

e at

O

ct

Ei

ad

co

se

an ec

ac Ar

no

oa

da hi

le no Li

te

e at

at le O 1.5

**

h Mitochondrial ROS (fold change)

**

g Glutathione levels (fold change)

**

Palmitoylcarnitine

**

Oleoylcarnitine

Stearoylcarnitine

Acetylcarnitine

nature medicine  advance online publication

SKR RXF

Hexadecenoylcarnitine

media (Fig. 2b and Supplementary Table 2). f 5.0 Confirmation of the top-ranking genes by ** 4.0 RT-qPCR revealed that many inflamma** 3.0 tory stress–response-related genes (Fig. 2c) 2.0 and fatty acid–metabolism genes (Fig. 2d) were more highly expressed after treatment 1.0 with RXF media than after treatment with 0 SKR media. We also confirmed that a large number of glycolytic genes were suppressed within 6 h after treatment (Fig. 2e). This led us to hypothesize that cachectic RXF media caused inflammation-induced changes in fatty acid metabolism and glycolysis, which represent some of the earliest primary events in muscle atrophy, and that this thereby triggered the rapid onset of stress responses in human myotubes. Several aspects of this proposed primary mechanism, such as the induction of inflammatory stress response signatures and the suppression of glycolysis signatures, were also observed in mouse muscle transcriptomes after RXF xenotransplantation in vivo (Supplementary Table 3). To confirm our hypothesis, we profiled the concentrations of previously described cachectic factors in the RXF and SKR secretomes, both in vivo and in vitro. In the serum of RXF-bearing mice and SKRbearing mice, we found that several inflammatory factors showed higher abundance in RXF serum than in SKR serum, although few of these attained statistical significance (Fig. 3a). By contrast, interleukin (IL)-1b, IL-6, IL-8 and tumor necrosis factor (TNF)-α concentrations were significantly higher in RXF media than in SKR media, which thus reflects the robustness of the in vitro model (Fig. 3b). The diversity of inflammatory factors, however, suggested that putting too much focus on any individual factor risks oversimplifying the cachectic mechanism—as supported by the failure of TNF-α-based and IL-6-antibody-based therapies in previous clinical trials3. Thus, we tried to pinpoint convergent metabolic effects in myotubes. We used liquid chromatography–tandem mass spectrometry

te

0

e

Figure 3  Cachectic conditioned media induces lipolysis and fatty acid oxidation as the primary metabolic effects in human muscle cells. (a) Abundance of secreted factors in RXFbearing (cachectic) versus SKR-bearing mice serum (n = 3). (b) Abundance of secreted factors in media conditioned by RXF cells (cachectic) versus SKR cells (n = 3). (c) Clustergram heat map of intracellular metabolites in human myotubes after exposure to SKR versus RXF (cachectic) conditioned media for 1 h. (d) Top lipid changes induced by RXF (cachectic) conditioned media. TG (x:y), triacylglyceride (number of carbon atoms:number of double bonds). (e) Top fatty acid changes induced by RXF (cachectic) conditioned media. (f) Top fatty acyl-carnitine changes induced by RXF (cachectic) conditioned media. (g) Changes in glutathione abundance in human myotubes after 1 h of exposure to RXF (cachectic) conditioned media. (h) Mitochondrial ROS abundance in human myotubes after 1 h of exposure to RXF (cachectic) conditioned media, as measured by MitoSox Red fluorescence. For all panels, data are expressed as mean ± s.e.m. *P < 0.05 and **P < 0.01 relative to SKR control, as determined by Student’s t test.

Metabolite levels (fold change)

© 2016 Nature America, Inc. All rights reserved.

letters

1.0

0.5

0

3.0

*

2.5 2.0 1.5 1.0 0.5 0

SKR

RXF

SKR

RXF

(LC–MS/MS) to perform metabolomic and lipidomic profiling of human myotubes after just 1 h of exposure to SKR or RXF media (Fig. 3c). We observed that the abundances of many species of tri­ acylglycerides and fatty acids were 50–100% higher after 1 h of treatment with RXF media than with SKR media (Fig. 3d,e). These results suggest that RXF media induced myotube lipid uptake and hydrolysis, a finding that is consistent with our transcriptomic observations of fatty acid metabolism (Fig. 2d). We also observed 50–400% higher abundance of acyl-carnitines after treatment with RXF media than with SKR media, which indicates that RXF media rapidly induced mitochondrial fatty acid oxidation in human myotubes (Fig. 3f). By contrast, we did not observe significant changes in the abundance of glycolytic intermediates after 1 h, which suggests that the decrease observed in glycolysis gene expression after 6 h (Fig. 2e) was a compensatory response to an increase in fatty acid oxidation. We also observed increased levels of glutathione (Fig. 3g), which supports the idea that the human myotubes are responding to an increase in oxidative stress resulting from exposure to RXF media. Direct measurement of mitochondrial superoxides confirmed this conclusion (Fig. 3h and Supplementary Fig. 2a). Thus, excessive fatty acid–induced oxidative stress led to the stress-response signatures induced by the cachectic RXF media (Fig. 2a–d).



a

RXF (cachectic) DMSO

Eto

RXF (cachectic)

SKR

Phospho-p38

p-AKT (T308) p-AKT (S473) AKT p-ERK1/2

GAPDH

c

1.2

Mitochondrial ROS (fold change)

IκBα

DMSO Palmitate Linoleate

p-p38 p38 p-AKT (T308) p-AKT (S473) IκBα

1.0

0.6 0.4 0.2

100

*

6

8

10

12

14

16

18

20

0.8

*

0.6 0.4 0.2

RXF

SKR

g Vehicle

Vehicle Etomoxir 0

1.0

0

RXF (n = 5)

85

DMSO Etomoxir

1.2

Etomoxir

8-oxo-guanine

22

105 100 95 90

G361 (n = 5) Vehicle Etomoxir 0

6

8

*

*

h * **

10 12 14 16 18 20 22 24

105

Etomoxir

SB202190

p38 p-AKT (T308) p-AKT (S473) AKT

100 95

Vehicle p-p38

*

LLC (n = 10)

*

Vehicle Etomoxir

90 0

5

6

IκBα

*

*

MHC GAPDH

7

8 9 Time (d)

10

11

12

i

j

To verify which signaling pathway was mediating the cachectic stress-response signatures, we performed western blots for various kinases. We found that the stressinducible p38 MAPK was more strongly activated after 1 h of treatment with RXF media than after treatment with SKR media (Fig. 4a). Moreover, p38 MAPK activation was blunted by etomoxir, a specific inhibitor of carnitine palmitoyltransferase-1, and thus of fatty acid oxidation. In RXF-bearing mice, quadriceps muscles showed a similar upregulation in phospho-p38 in vivo (Fig. 4b). Indeed, fatty acid supplementation alone upregulated phospho-p38 (Fig. 4c). By contrast, phospho-Akt and the NF-κB inhibitor IκBα did not show consistent trends when exposed to RXF media or fatty acid supplementation, which suggests that excessive fatty acid metabolism exerts more complex effects on the PI3K-Akt and NF-κB pathways than on p38 MAPK. Inhibition of fatty acid oxidation protected against both the mitochondrial oxidative stress (Fig. 4d and Supplementary Fig. 2b) and the myotube growth defects (Fig. 4e

Etomoxir

SB202190

Myofiber area (fold change)

Body weight (%)

* *

1.4

RXF

SKR

*

90

e

0

105

95

**

0.8

Actin

f

DMSO Etomoxir

hMyotube growth (fold change)

d

ERK1/2

Vehicle



b

SKR

DMSO Eto

p-p38 p38

Body weight (%)

Figure 4  Inhibition of fatty acid–induced oxidative stress blocks cachexia. (a) Representative western blot (n = 3) for phospho-p38 MAPK, phospho-AKT, phosphoERK1/2, IκBα and glyceraldehyde 3 phosphate dehydrogenase 1 (GADPH) protein abundance in human myotubes exposed to RXF versus SKR media, with DMSO vehicle or 10 µM of etomoxir (Eto) for 1 h. (b) Representative photomicrograph (n = 10) of immunohistochemical staining for nuclear phospho-p38 in the quadriceps muscles of RXFbearing mice versus SKR-bearing mice. Scale bars, 100 µm. (c) Representative western blot (n = 3) for phospho-p38, phospho-AKT, IκBα and actin levels in human myotubes after 1 h of supplementation with 2 mM of palmitate or linoleate. (d) Quantification of MitoSox Red fluorescence in human myotubes after 1 h of exposure to RXF media, with and without 10 µM etomoxir. (e) Human myotube total nuclei numbers after 6 d growth in SKR versus RXF media, with DMSO vehicle or 10 µM etomoxir. (f) Body weight of mice bearing RXF (n = 5), G361 (n = 5) or Lewis lung carcinoma (LLC; n = 10) tumors, after daily intraperitoneal injections of DMSO vehicle or 20 mg/kg etomoxir. (g) Representative photomicrographs (n = 10) of 8-oxoguanine staining in quadriceps myofibers after DMSO vehicle or etomoxir injections. Scale bars, 100 µm. (h) Western blot for phospho-p38, phospho-AKT, IκBα, MHC and GAPDH levels in quadriceps muscles after daily intraperitoneal injections of DMSO vehicle, 20 mg/kg etomoxir or 5 mg/kg SB202190 into RXF-bearing mice (n = 3 each). (i) Representative photomicrographs (n = 10) of H&E histology of quadriceps muscles after DMSO vehicle, etomoxir or SB202190 injections into RXF-bearing mice. Scale bars, 200 µm. (j) Quantification of the average quadriceps myofiber cross-sectional areas in RXF-bearing mice after injections of DMSO vehicle, etomoxir or SB202190. For all panels, data are expressed as mean ± s.e.m. *P < 0.05 and **P < 0.01 relative to vehicle control, as determined by Student’s t test.

Body weight (%)

© 2016 Nature America, Inc. All rights reserved.

letters

1.4 1.2 1.0 0.8 0.6 0.4 0.2 0

*

Ve

hi

cl

e Et

om

ox

ir 2 SB

*

02

19

0

and Supplementary Fig. 2c) caused by exposure to RXF media. These results support the notion that the cachectic media promoted excessive fatty acid–induced oxidative stress, which activated p38 MAPK and caused a myotube growth defect. To validate our conclusions in vitro as physiologically relevant in vivo, we injected etomoxir into a variety of cachectic mouse models on a daily basis and monitored their body weights until the point of cachexia onset. In the clinic, cachexia is defined as >5% weight loss within 6 months in patients. Etomoxir, when administered alone, rescued the rapid weight loss induced by cachexia in three mouse models, which included human RXF cells, human G361 melanoma cells9 and mouse Lewis lung carcinoma (LLC) cells syngeneically

advance online publication  nature medicine

© 2016 Nature America, Inc. All rights reserved.

letters transplanted into C57BL/6J mice (Fig. 4f). Immunohistochemical staining for 8-oxoguanine, a biomarker for oxidative stress in vivo, demonstrated that etomoxir suppressed the generation of reactive oxygen species (ROS) in cachectic muscles (Fig. 4g). Etomoxir also suppressed aberrant p38 MAPK activation and rescued MHC protein levels in the muscles of all three mouse models, as compared to vehicle controls (Fig. 4h and Supplementary Fig. 3). At the transcriptomic level, etomoxir treatment in mouse muscles suppressed the cachectic activation of proteasome-, NF-κB-, Stat3- and p38-MAPK-related signatures, and also rescued myogenesis and glucosemetabolism-related signatures (Supplementary Tables 3 and 4). The inhibition of p38 MAPK with the inhibitor SB202190 also rescued mouse Akt phosphorylation and MHC expression in muscles in vivo (Fig. 4h). Histopathological analysis revealed that the myofibers were rescued from cachectic atrophy by etomoxir and SB202190 (Fig. 4i,j). Hind-limb muscle-mass measurements confirmed these observations (Supplementary Fig. 4a–d). Etomoxir and SB202190 rescued mice from cachexia without affecting RXF tumor growth (Supplementary Fig. 4e). These observations were also true for the human G361 and mouse LLC models (Supplementary Fig. 4c,d,f,g). To confirm that etomoxir was acting specifically on the muscles, and not systemically on the tumor or other tissues, we formulated etomoxir with hydrogel microspheres for intramuscular controlled-release therapy within the skeletal muscles of one particular hind limb 10,11. We measured hind-limb muscle mass (Supplementary Fig. 5a) and imaged the hind-limb muscles (Supplementary Fig. 5b), which revealed significant (P < 0.01) rescue of the treated hind limb without amelioration of the contralateral, control hind limb. These results also confirm that the etomoxir-containing hydrogel was not leaking into the general systemic circulation. In addition, we found that etomoxir specifically inhibited muscle fatty acid oxidation, as shown by the lower abundance of acyl-carnitines and the higher abundance of fatty acids, as compared to the control, whereas peroxisome-proliferator-activated receptor (PPAR)-α-associated cholesterol and carbohydrate metabolites remained insignificantly changed (Supplementary Fig. 6a–c). Moreover, PPAR-α expression in the muscles was negligible, and PPAR-α targets did not change after etomoxir treatment (Supplementary Fig. 7). These experiments show that etomoxir acts specifically on muscle fatty acid oxidation, and that local treatment of cachectic muscle wasting is possible. Finally, among muscle biopsies from a cohort of subjects with gastrointestinal cancer, we found a significant (P < 0.05) positive correlation between muscle oxidative stress and cachexia (Supplementary Fig. 8a). Cachexia was also positively correlated with phospho-p38 staining (Supplementary Fig. 8b). Furthermore, oxidative stress and phospho-p38 staining showed a strongly positive linear correlation in these samples (r = 0.92; Supplementary Fig. 8c). Taken together, these clinical data suggest that our experimental observations on fatty acid–induced oxidative stress and p38 activation are highly representative of cachexia progression in individuals with cancer. In summary, the transcriptomic- and metabolomic-profiling data from our human cancer-based and muscle stem cell–based models are consistent with the view that excessive fatty acid metabolism has a crucial role in muscle atrophy during cancer-induced cachexia in humans. Previous studies have shown that hypermetabolism is induced by cachexia, but it was long considered to be only one of the multifarious effects—not a driving force—in cachexia1–6. Here we found that complex pro-inflammatory factors converge to trigger excessive fatty acid catabolism in muscles, which causes elevated

nature medicine  advance online publication

oxidative stress and leads to the activation of the aging-associated, p38 signaling pathway12,13; this in turn leads to progressive muscle wasting. Indeed, many previous studies have already shown that cancer and multiple pro-inflammatory factors can induce lipolysis and fatty acid oxidation14–20. Moreover, it is well known that mitochondrial ROS and oxidative stress can potently activate the stress-inducible p38 MAPK pathway21, and that both mitochondrial ROS and p38 MAPK are emerging as crucial regulators of muscle biology12,13,22–26. Remarkably, pharmacological inhibition of fatty acid oxidation rescued p38 MAPK activation and prevented muscle wasting during cancer-induced cachexia both in human myotubes in vitro and in mouse models in vivo. Given that safe pharmacological and dietary methods for inhibiting fatty acid oxidation already exist 27–29, this approach could represent an alternative strategy for preventing the onset of this devastating muscle-wasting condition as a result of cancer or other chronic aging-associated diseases. Methods Methods and any associated references are available in the online version of the paper. Accession codes. Transcriptomics data have been deposited in the Gene Expression Omnibus (GEO) under the accession codes GSE80080 and GSE80081. Note: Any Supplementary Information and Source Data files are available in the online version of the paper. Acknowledgments We thank all our respective lab mates for their helpful discussion, and E. Peh for technical assistance with mass spectrometry. This work was supported by the Agency for Science Technology and Research (ASTAR) Joint Council Office grant 1431AFG128 (N.S.-C.), the Singapore National Medical Research Council grants NMRC/STaR/0024/2014 and NMRC/GMS/CIRG/1332/2012, Duke-NUS Medical School, and National Cancer Centre Singapore (all to B.T.T.), Polaris program grant SPF2012/001 under the ASTAR Strategic Positioning Fund (Y.S.H.) and the Genome Institute of Singapore (I.B.T. and N.S.-C.). AUTHOR CONTRIBUTIONS T.F., I.B.T., B.T.T. and N.S.-C. designed the study. T.F., B.C.Y.-J., J.C.M.-W., D.H., C.-N.Q., P.O., Z.L. and H.K. performed experiments in vivo. T.F., B.C.Y.-J., J.C.M.-W., E.T.J.-H., D.H. and P.O. performed experiments in vitro. W.J.L., J.H.L., C.C., H.S.O., K.-K.T. and I.B.T. provided the clinical samples. S.C., S.Y.M. and Y.S.H. performed mass-spectrometry experiments. R.E.M. and R.R.W. performed quantitative phase imaging. T.F., I.B.T., B.T.T. and N.S.-C. wrote the manuscript. COMPETING FINANCIAL INTERESTS The authors declare no competing financial interests. Reprints and permissions information is available online at http://www.nature.com/ reprints/index.html. 1. Tisdale, M.J. Cachexia in cancer patients. Nat. Rev. Cancer 2, 862–871 (2002). 2. von Haehling, S. & Anker, S.D. Cachexia as a major underestimated and unmet medical need: facts and numbers. J. Cachexia Sarcopenia Muscle 1, 1–5 (2010). 3. Fearon, K.C., Glass, D.J. & Guttridge, D.C. Cancer cachexia: mediators, signaling, and metabolic pathways. Cell Metab. 16, 153–166 (2012). 4. Fearon, K., Arends, J. & Baracos, V. Understanding the mechanisms and treatment options in cancer cachexia. Nat. Rev. Clin. Oncol. 10, 90–99 (2013). 5. Cohen, S., Nathan, J.A. & Goldberg, A.L. Muscle wasting in disease: molecular mechanisms and promising therapies. Nat. Rev. Drug Discov. 14, 58–74 (2015). 6. Zuijdgeest-van Leeuwen, S.D. et al. Lipolysis and lipid oxidation in weight-losing cancer patients and healthy subjects. Metabolism 49, 931–936 (2000). 7. Guttridge, D.C., Mayo, M.W., Madrid, L.V., Wang, C.Y. & Baldwin, A.S. Jr. NF-κBinduced loss of MyoD messenger RNA: possible role in muscle decay and cachexia. Science 289, 2363–2366 (2000). 8. McFarlane, C. et al. Myostatin induces cachexia by activating the ubiquitin proteolytic system through an NF-κB-independent, FoxO1-dependent mechanism. J. Cell. Physiol. 209, 501–514 (2006). 9. Mori, M. et al. Cancer cachexia syndrome developed in nude mice bearing melanoma cells producing leukemia-inhibitory factor. Cancer Res. 51, 6656–6659 (1991).



letters 20. Ebadi, M., Baracos, V.E., Bathe, O.F., Robinson, L.E. & Mazurak, V.C. Loss of visceral adipose tissue precedes subcutaneous adipose tissue and associates with n-6 fatty acid content. Clin. Nutr. S0261-5614(16)00071-6 (2016). 21. Son, Y., Kim, S., Chung, H.T. & Pae, H.O. Reactive oxygen species in the activation of MAP kinases. Methods Enzymol. 528, 27–48 (2013). 22. Mastrocola, R. et al. Muscle wasting in diabetic and in tumor-bearing rats: role of oxidative stress. Free Radic. Biol. Med. 44, 584–593 (2008). 23. Waning, D.L. et al. Excess TGF-β mediates muscle weakness associated with bone metastases in mice. Nat. Med. 21, 1262–1271 (2015). 24. Gao, S. & Carson, J.A. Lewis lung carcinoma regulation of mechanical stretchinduced protein synthesis in cultured myotubes. Am. J. Physiol. Cell Physiol. 310, C66–C79 (2016). 25. Trendelenburg, A.U., Meyer, A., Jacobi, C., Feige, J.N. & Glass, D.J. TAK-1/p38/ nNFκB signaling inhibits myoblast differentiation by increasing levels of Activin A. Skelet. Muscle 2, 3 (2012). 26. Min-Wen, J.C., Jun-Hao, E.T. & Shyh-Chang, N. Stem cell mitochondria during aging. Semin. Cell Dev. Biol. 52, 110–118 (2016). 27. Woods, M.N. et al. Effect of a dietary intervention and n-3 fatty acid supplementation on measures of serum lipid and insulin sensitivity in persons with HIV. Am. J. Clin. Nutr. 90, 1566–1578 (2009). 28. Horowitz, J.D., Chirkov, Y.Y., Kennedy, J.A. & Sverdlov, A.L. Modulation of myocardial metabolism: an emerging therapeutic principle. Curr. Opin. Cardiol. 25, 329–334 (2010). 29. Lopatin, Y.M. et al. Rationale and benefits of trimetazidine by acting on cardiac metabolism in heart failure. Int. J. Cardiol. 203, 909–915 (2016).

© 2016 Nature America, Inc. All rights reserved.

10. Bouhadir, K.H. et al. Degradation of partially oxidized alginate and its potential application for tissue engineering. Biotechnol. Prog. 17, 945–950 (2001). 11. Silva, E.A., Kim, E.S., Kong, H.J. & Mooney, D.J. Material-based deployment enhances efficacy of endothelial progenitor cells. Proc. Natl. Acad. Sci. USA 105, 14347–14352 (2008). 12. Bernet, J.D. et al. p38 MAPK signaling underlies a cell-autonomous loss of stem cell self-renewal in skeletal muscle of aged mice. Nat. Med. 20, 265–271 (2014). 13. Cosgrove, B.D. et al. Rejuvenation of the muscle stem cell population restores strength to injured aged muscles. Nat. Med. 20, 255–264 (2014). 14. Das, S.K. et al. Adipose triglyceride lipase contributes to cancer-associated cachexia. Science 333, 233–238 (2011). 15. Kawakami, M. et al. Human recombinant TNF suppresses lipoprotein lipase activity and stimulates lipolysis in 3T3-L1 cells. J. Biochem. 101, 331–338 (1987). 16. Feingold, K.R., Doerrler, W., Dinarello, C.A., Fiers, W. & Grunfeld, C. Stimulation of lipolysis in cultured fat cells by tumor necrosis factor, interleukin-1, and the interferons is blocked by inhibition of prostaglandin synthesis. Endocrinology 130, 10–16 (1992). 17. Green, A., Dobias, S.B., Walters, D.J. & Brasier, A.R. Tumor necrosis factor increases the rate of lipolysis in primary cultures of adipocytes without altering levels of hormone-sensitive lipase. Endocrinology 134, 2581–2588 (1994). 18. van Hall, G. et al. Interleukin-6 stimulates lipolysis and fat oxidation in humans. J. Clin. Endocrinol. Metab. 88, 3005–3010 (2003). 19. Argilés, J.M., Busquets, S., Toledo, M. & López-Soriano, F.J. The role of cytokines in cancer cachexia. Curr. Opin. Support. Palliat. Care 3, 263–268 (2009).



advance online publication  nature medicine

© 2016 Nature America, Inc. All rights reserved.

ONLINE METHODS

Cell cultures. LLC, 786-O, G361 and A498 cell lines were purchased from the American Type Culture Collection (ATCC). RXF393 and SN12C cell lines were obtained from the National Cancer Institute (Bethesda, MD, USA) and SKRC39 cells were obtained from the Memorial Sloan Kettering Cancer Center. All cells were confirmed for their identity and cultured as recommended by respective sources. No mycoplasma contamination was detected in cultures of any of these cell lines. Conditioned media were obtained by culturing 1.6 × 106 cells for 72–96 h in a 150-mm diameter dish containing 20 ml of RPMI1640 (Gibco) media with 10% FBS (Hyclone), then sterile filtered. Primary muscle stem cells were derived from the rectus abdominus muscles of three patient donors without cachexia (aged 51–72 yrs; weights, 55–59 kg) undergoing tumor-resection surgery, in accordance with ethical legislation and the SingHealth Institutional Review Board guidelines. Muscle stem cells were derived in a 100-mm dish containing growth medium, and then were maintained and differentiated according to previously published protocols30. Early myotubes were obtained after 6 d of growth and differentiation in conditioned media. Nuclei numbers were assessed by DAPI staining. Live cell volume was measured using quantitative 3D microscopy with d’Biomager (d’Optron). MitoSox Red and ethidium (Invitrogen) staining was performed according to the manufacturer’s instructions. Animal experiments. All experiments involving animals were conducted under approved protocols granted by the SingHealth Institutional Animal Care and Use Committee. Eight- to ten-week-old male NOD-SCID mice, NOD.CgPrkdcscid Il2rgtm1Wjl/SzJ (NSG) mice or C57BL/6J mice (Jackson Laboratory) were housed in Academia Singapore. For mini-screening to identify cachectic renal cell carcinoma cell lines, NOD-SCID mice were implanted subcutaneously into their flanks with 5.0 × 106 of RXF393, A498, 786-O, SKRC39 or SN12C cells. When the mice lost 15% of their body weight or when their tumor volumes reached 1,500 mm3, tumors and muscles were rapidly dissected, frozen in liquid nitrogen and stored at −80 °C until ready for further analyses, or fixed in 10% neutral buffered formalin (10% formaldehyde, phosphate buffered) overnight and embedded in paraffin. To check the efficacy of drugs, we administered NSG mice with subcutaneous implantations into their flanks of 5.0 × 106 of RXF393 and LLC, or 1.0 × 107 of G361 cells. Sample sizes were chosen on the basis of previous literature. Mice were excluded if no tumor could be found after cancer cell implantation. 5 d after cell implantation, mice were randomly allocated to each treatment group, and etomoxir (20 mg/kg; Sigma-Aldrich), SB202190 (5 mg/kg; Sigma-Aldrich) or an equal volume of vehicle (PBS containing 50% DMSO) was intraperitoneally injected daily. Investigators were not blinded to the experiments. Mice were weighed daily, and tumor volumes were monitored three times per week. Mice were euthanized as follows: RXF-bearing mice at day 22; G361-bearing mice at day 24; and LLC-bearing mice at day 17. Etomoxir alginate. Alginate gel was generated according to previously published protocols10,11. To incorporate etomoxir, a solution of oxidized alginate was well mixed with etomoxir (final concentration of etomoxir, 100 µg in 50 µl of 2% of oxidized alginate solution) and loaded into a 30G syringe. The solution was transferred drop-wise into 1% solution of calcium chloride to generate etomoxir–alginate microspheres. After washing with 0.9% sodium chloride, vehicle (2% oxidized alginate beads) or etomoxir gel (100 µg in 50 µl) was intramuscularly injected into the hind-limb thighs of RXF-bearing mice every 7 d. Mice were euthanized when the mice lost 15% of their body weight. RNA transcriptomics. Total RNA was extracted from cell lines using Trizol (Invitrogen) and purified with the RNeasy Mini Kit (Qiagen). cRNA libraries

doi:10.1038/nm.4093

were constructed using the TotalPrep RNA Amplification Kit (Illumina) and hybridized to MouseWG-6 v2.0 or HumanHT-12 v4.0 BeadChips (Illumina), according to the manufacturer’s instructions. For validation, RT-qPCR was performed according to previously published protocols31. Secretome cytokine analysis. Serum samples of RXF-bearing mice and SKRbearing mice or secretome-enriched RXF and SKR-conditioned media were collected and stored in aliquots at −80 °C until analysis. IL-1b, IL-6, IL-8, TNF-α, LIF, VEGF, MMP2 and TGF-β levels were quantitatively determined by multiplex cytokine immunoassay platform, Searchlight (Thermo Scientific). Metabolomics analysis. LC–MS/MS metabolomics and lipidomics analyses were performed according to previously published protocols32,33. Immunohistochemical (IHC) staining. Specimens were deparaffinized, and antigen was retrieved using citrate buffer. After quenching endogenous peroxidase activity by incubating the specimen with 3% H2O2, all slides were incubated with the first primary antibody against p-p38 (4511S; Cell Signaling, 1:500), or anti–8-oxoguanine (ab64548; Abcam; 1:400) overnight, followed by incubation with the Dako ChemMate Envision kit/HRP polymer. Staining was completed by incubation with 3,3′-diaminobenzidine (DAB) + substratechromogen, and hematoxylin counterstained. For H scoring of samples, intensity of nucleus staining was evaluated on the basis of the following scoring system: 0, no staining; 1, low; 2, moderate; and 3, high. The percentage of each intensity score for nucleus staining was also recorded. The H score was calculated using the following formula: H score = (% at 0) × 0 + (% at 1) × 1 + (% at 2) × 2 + (% at 3) × 3. Thus, this score produced a continuous variable that ranges from 0 to 300. Western blots. Protein extracts were prepared with RIPA cell lysis buffer (150 mM NaCl, 50 mM Tris-HCl, 0.5% deoxychlorate sodium, 200 mM NaF, 200 mM PMSF, 1.0% NP40, 1 mM EDTA) with the protease-inhibitor cocktail (Roche). Lysates were subjected to SDS–PAGE and transferred to PVDF membrane for immunoblotting analysis. The following antibodies were used: p38 (9212S; Cell Signaling, 1:1,000), p-p38 (4511S; Cell Signaling, 1:1,000), AKT (2920S; Cell Signaling, 1:1,000), p-AKT (S473) (4060S; Cell Signaling, 1:1,000), p-AKT (T308) (2965S; Cell Signaling, 1:1,000), ERK1/2 (9102L; Cell Signaling, 1:1,000), p-ERK1/2 (4370S; Cell Signaling, 1:1,000), IκBα (4814S; Cell Signaling, 1:1,000), Myogenin (sc-576; Santa Cruz Biotech, 1:400), α-Actinin (A7811, Sigma-Aldrich, 1:2,500), fast MHC (A4335; Sigma-Aldrich, 1:50,000), myosin heavy chain (53-6503; eBioscience, 1:500), α-tubulin (sc8035; Santa Cruz Biotech, 1:1,000), β-actin (A1978; Sigma-Aldrich, 1:5,000) and GAPDH (sc-25778; Santa Cruz Biotech, 1:1,000). Statistical analysis. Data are presented as mean ± s.e.m., and Student’s t test (two-tailed distribution, two-sample unequal variance) was used to calculate P values for normally distributed data. All tests were performed using Microsoft Excel. 30. Skuk, D. et al. Intramuscular transplantation of human postnatal myoblasts generates functional donor-derived satellite cells. Mol. Ther. 18, 1689–1697 (2010). 31. Khaw, S.L., Min-Wen, C., Koh, C.G., Lim, B. & Shyh-Chang, N. Oocyte factors suppress mitochondrial polynucleotide phosphorylase to remodel the metabolome and enhance reprogramming. Cell Rep. 12, 1080–1088 (2015). 32. Selvarasu, S. et al. Combined in silico modeling and metabolomics analysis to characterize fed-batch CHO cell culture. Biotechnol. Bioeng. 109, 1415–1429 (2012). 33. Tan, A.H.M. et al. Aberrant presentation of self-lipids by autoimmune B cells depletes peripheral iNKT cells. Cell Rep. 9, 24–31 (2014).

nature medicine

SUPPLEMENTARY INFORMATION Excessive fatty acid oxidation induces muscle atrophy in cancer cachexia Tomoya Fukawa, Benjamin Chua Yan-Jiang, Jason Chua Min-Wen, Elwin Tan Jun-Hao, Dan Huang, Chao-Nan Qian, Pauline Ong, Zhimei Li, Shuwen Chen, Shi Ya Mak, Wan Jun Lim, Hiro-omi Kanayama, Rosmin Elsa Mohan, Ruiqi Rachel Wang, Jiunn Herng Lai, Clarinda Chua, Hock Soo Ong, Ker-Kan Tan, Ying Swan Ho, Iain Beehuat Tan, Bin Tean Teh, Ng Shyh-Chang

Supplementary Figure 1 Human RXF393 cancer cells induce muscle atrophy. (a) Relative frequency distributions of myofiber cross-sectional area in matching quadriceps muscle biopsies from SKR- and RXF-bearing mice. AU, arbitrary units of pixels. Data are expressed as means. P < 0.001 relative to SKR control, as determined by Mann-Whitney test. (b) Representative phase contrast images of early myotubes derived from human muscle stem cells isolated from patient biopsies, after 6d exposure to cachectic RXF media or non-cachectic SKR media. (c) Measurements of total cell volume differences in early myotubes from (b), based on quantitative phase imaging. Data are expressed as means ± s.e.m. *P < 0.05 relative to SKR control, as determined by Student’s t-test. (d) Western blot of human myotubes after 6 days of exposure to cachectic RXF or non-cachectic SKR medias, using antibodies against myogenin, α-actinin, fast myosin heavy chain (MHC), pan-MHC, GAPDH and tubulin.

Supplementary Figure 2 Cachectic RXF media induces mitochondrial oxidative stress in human myotubes. (a) Representative MitoSox Red fluorescence images of live human myotubes after 1h exposure to cachectic RXF versus non-cachectic SKR media. (b) Representative MitoSox Red fluorescence images of live human myotubes after 1h exposure to cachectic RXF with and without the fatty acid oxidation inhibitor etomoxir 10 µM. (c) Quantification of cell death using ethidium dye, after 6d exposure to RXF media. Data are

Nature Medicine: doi:10.1038/nm.4093

expressed as means ± s.e.m. *P < 0.05 relative to SKR control, as determined by Student’s ttest.

Supplementary Figure 3 Human G361 and mouse Lewis lung carcinoma (LLC) cells both cause excessive fatty acid oxidation to induce p38 MAPK signaling in myofibers as well. Western blot for phospho-p38, phospho-AKT, IκBα, MHC, and GAPDH levels in quadriceps myofibers of G361- or LLC-bearing mice after daily intraperitoneal injections of DMSO vehicle or 20 mg/kg etomoxir (n = 3 each).

Supplementary Figure 4 Etomoxir rescues muscle atrophy in mouse models of cachexia. (a) Representative images of RXF-bearing mice’ quadriceps muscle morphology with and without etomoxir treatment. Etomoxir-treated mice’ quadriceps preserved their muscle mass. (b) Forelimb and hindlimb muscle mass (% of body mass) of RXF-bearing mice after daily injections of DMSO vehicle, etomoxir or SB202190 (n = 5 each). (c) Forelimbs’ and hindlimbs’ muscle mass (% body mass) of G361-bearing mice after daily injections of DMSO vehicle, 20 mg/kg etomoxir or 5 mg/kg SB202190 (n = 5 each). Etoxomir and SB202190 rescued limb muscle loss. (d) Representative H&E histology of quadriceps muscles after daily intraperitoneal injections of DMSO vehicle or etomoxir in LLC-bearing C57BL/6J mice. Etomoxir rescued myofiber atrophy. Bar represents 200 µm. (e-g) Tumor growth curves of (e) RXF, (f) G361, and (g) LLC tumors, with daily injections of DMSO vehicle, 20 mg/kg etomoxir or 5 mg/kg of SB202190 (n = 5 each). Data are expressed as means ± s.e.m. *P < 0.05 relative to DMSO vehicle control, as determined by Student’s t-test.

Supplementary Figure 5 Intramuscular controlled-release formulation of etomoxir only rescues treated hindlimbs’ muscle mass. (a) Hindlimb muscle mass (% of body mass) of RXF-bearing

2

Nature Medicine: doi:10.1038/nm.4093

mice after intramuscular injections of vehicle- (2% oxidized alginate beads, n = 6) or etomoxirgel (100 µg / 50 µL, n = 7) into hindlimb thigh muscles every 7 days. Mice were sacrificed when the mice lost 15% weight. Untreated and vehicle-gel-treated hindlimbs atrophied to 3.2% body mass, whereas the etomoxir-gel-treated hindlimbs showed a significant rescue from atrophy (3.8%, P = 0.007), similar to normal non-cachectic hindlimbs (3.6%). (b) Representative images of etomoxir-gel-treated left quadriceps muscle morphology and H&E histology, relative to the untreated right quadriceps muscles. Bar represents 200 µm. Data are expressed as means ± s.e.m. **P < 0.01 relative to vehicle-gel control, as determined by Student’s t-test.

Supplementary Figure 6 Etomoxir inhibited fatty acid oxidation, but not PPARα-associated sterol and carbohydrate metabolism. (a) Acyl-carnitine levels in quadriceps muscles after 20 mg/kg etomoxir treatment (n = 4). (b) Fatty acid and cholesterol levels in quadriceps muscles after 20 mg/kg etomoxir treatment (n = 4). (c) Polar metabolites significantly changed in quadriceps muscles after 20 mg/kg etomoxir treatment (n = 4). Etomoxir altered only 12 of 3743 polar metabolites in quadriceps muscles, none of which were PPARα-associated carbohydrates. Instead only a polar fatty acid, several nucleotide-related metabolites, and redox-related metabolites were affected by etomoxir, supporting a fatty acid oxidation-specific mechanism that regulated the redox state and myocellular growth. CMP, cytidine monophosphate. FAD, flavin adenine dinucleotide. GSSG, glutathione disulfide. Data are expressed as means ± s.e.m. *P < 0.05 and **P < 0.01 relative to DMSO vehicle control, as determined by Student’s t-test.

Supplementary Figure 7 Etomoxir did not affect PPARα target genes in quadriceps muscles. Raw microarray expression values for PPARα target genes and PPARα itself, relative to Pax3 and Pax7, which are at the background level, after 20 mg/kg etomoxir treatment (n = 3). Pdk4, pyruvate dehydrogenase kinase 4. Fabp3, fatty acid binding protein 3. Ldha, lactate

3

Nature Medicine: doi:10.1038/nm.4093

dehydrogenase A. Pcx, pyruvate carboxylase. Pck1, phosphoenolpyruvate carboxykinase 1. Data are expressed as means ± s.e.m. *P < 0.05 relative to DMSO vehicle control, as determined by Student’s t-test.

Supplementary Figure 8 Oxidative damage is correlated with p38 activation in cachexia patients’ muscle biopsies. (a-b) Representative immunohistochemical staining for (a) 8-oxoguanine and (b) nuclear phospho-p38 in non-cachexia and cachexia subject muscle biopsies (n = 11), and their quantitative H-scores. Bar represents 50µm. (c) Correlation plot for the Hscores of 8-oxo-guanine vs. phospho-p38 immunohistochemical staining in non-cachectic (black) and cachectic (red) subjects’ rectus abdominus muscles (n = 11). Data are expressed as means ± s.e.m. *P < 0.05 relative to non-cachectic control, as determined by Student’s t-test.

Supplementary Table 1 Top 4 upregulated and downregulated gene sets in mouse myotubes after 6h exposure to cachectic RXF conditioned media, relative to SKR media (n = 3 each).

Supplementary Table 2 Top 4 upregulated and downregulated gene sets in human myotubes after 6h exposure to cachectic RXF conditioned media, relative to SKR media (n = 3 each).

Supplementary Table 3 Top 4 upregulated and downregulated gene sets in cachectic RXFbearing mouse quadriceps muscles, relative to non-cachectic SKR-bearing mouse quadriceps muscles (n = 3 each).

Supplementary Table 4 Top 10 downregulated and upregulated gene sets in cachectic RXFbearing mouse quadriceps muscles, after daily intraperitoneal injections of 20 mg/kg etoxomir, relative to vehicle control (n = 3 each).

4

Nature Medicine: doi:10.1038/nm.4093

Rlea-ve  frequency  (%)  

a  

50  

RXF   SKR  

40   30  

P  <  0.001  

20   10   0   500  

1,000  

2,000  

8,000  

20,000  

AU   b  

c  

SKR  media  

SKR  

RXF  media  (cachec-c)  

d  

1.2  

RXF  

Total  cell  volume  (fold)  

1  

*  

0.8   0.6   0.4   0.2   0  

SKR  

Nature Medicine: doi:10.1038/nm.4093

RXF  

Supplementary  Figure  1  

.s.  

a   Myosta-n  

RXF  condi-oned  media  

SKR  condi-oned  media  

RXF  media  +DMSO  

RXF  media  +Etomoxir  

b  

ned  media  (Pro-­‐CX)  

ax3  

c   1.8  

Vehicle   Etomoxir   n.s.  

Pax7  

n.s.  

Ppara  

Cell  death     (fold  change  in  ethidium  fluorescence)  

n.s.  

A  

1.6  

Figure  S6  

*   SKR  

RXF  

Pro-­‐CX  +DMSO  

1.4  

n.s.  

1.2   1.0   0.8   0.6   0.4   0.2   0.0   Vehicle  

Nature Medicine: doi:10.1038/nm.4093

Etomoxir  

Supplementary  Figure  2  

G361   Vehicle  

LLC   Etomoxir  

Vehicle  

Etomoxir  

p-­‐p38     p38     p-­‐AKT     (S473)   AKT     IκBα       MHC   GAPDH  

Supplementary  Figure  3  

Nature Medicine: doi:10.1038/nm.4093

a  

Vehicle

b  

9

Limbs’  muscle  mass     (%  body  mass)  

Etomoxir

RXF  (n  =  5)   *  

8

*   *  

7 6 5

Vehicle   Etomoxir   SB202190   c  

Vehicle

Etomoxir

*   *  

8

LLC

Limbs’  muscle  mass     (%  body  mass)  

9

d  

G361  (n  =  5)  

7 6 5

Tumor  Volume  (mm3)  

e  

f   400 350 300 250 200 150 100 50 0

RXF   Vehicle

n.s 1000 .   800

Etomoxir SB202190

600 400

g  

G361   Vehicle

LLC  

600 n.s .   500 400

Etomoxir

Vehicle Etomoxir

300

SB202190

n.s.  

200

200

100

0

0

8 10 12 14 16 18 20

7 9 11 13 15 17 19 21 23 25

Time  (d)  

Time  (d)  

0

5

10

15

Time  (d)  

Supplementary  Figure  4   Nature Medicine: doi:10.1038/nm.4093

a  

b   Eto-­‐gel  

RXF  hindlimb  muscle  (%  body  mass)  

4.00%  

**   3.80%  

Normal  Hindlimb  

3.60%   3.40%   3.20%   3.00%   Eto  

no  treatment  

-­‐gel  

Vehicle  

-­‐gel  

no  treatment  

Supplementary  Figure  5  

*  

**  

b   Fold  change  

200%  

Fold  Change  

150%  

*

*  

*

*   *

*  

50%  

*  

Vehicle  

*  

*  

*  

*  

Etomoxir  

*  

100%   50%   0%  

Cer(d18:1/24:1(1 5Z))  

Glucosylceramide   (d18:1/24:1(15Z))  

Cer(d18:1/22:1(1 3Z))  

200%  

Etomoxir  

*  

*

100%  

*  

250%   Vehicle  

*  

n.s.  

Etomoxir  

Docosapent aenoyl-­‐ carni-ne  

c  

150%  

Vehicle  

0%   Palmitoyl-­‐ carni-ne  

0%  

Stearoyl-­‐ carni-ne  

50%  

Decanoyl-­‐ carni-ne  

Fold  change  

100%  

Nature Medicine: doi:10.1038/nm.4093

Supplementary  Figure  6  

250%   200%   150%   100%   50%   0%  

CMP  

*

2-­‐Octenedioate  

*

Fold  change  

a  

Supplementary  Figure  7   a  

Cachexia Patients

8-­‐Oxo-­‐guanine  

 8-­‐Oxo-­‐guanine  H-­‐score  

Non-cachexia Patients

200   150  

*  

100   50   0   NonCachexia non-­‐CX   CX   cachexia

Phospho-­‐p38  

Phospho-­‐p38  H-­‐score  

b  

8-­‐Oxo-­‐guanine  

c  

200  

*  

150   100   50   0  

NonCachexia non-­‐CX   CX   cachexia

200   150   100  

r  =  0.92  

Supplementary  Figure  8  

50   0   0  

50  

100  

150  

200  

Phospho-­‐p38  

Nature Medicine: doi:10.1038/nm.4093

250  

Supplementary  Table  1   Gene  sets  Up  in  cachec=c  Mouse  myotubes   GALINDO_IMMUNE_RESPONSE_TO_ENTEROTOXIN   ICHIBA_GRAFT_VERSUS_HOST_DISEASE_D7_UP   BAKKER_FOXO3_TARGETS_UP   MIZUSHIMA_AUTOPHAGOSOME_FORMATION   Gene  sets  Down  in  cachec=c  Mouse  myotubes   MANALO_HYPOXIA_DN   REACTOME_MRNA_PROCESSING   PENG_RAPAMYCIN_RESPONSE_DN   KARLSSON_TGFB1_TARGETS_UP  

FDR  q-­‐val   5.23E-­‐03   5.76E-­‐03   7.21E-­‐03   2.03E-­‐02   FDR  q-­‐val   0.00E-­‐00   0.00E-­‐00   0.00E-­‐00   3.00E-­‐03  

Supplementary  Table  2   Gene  sets  Up  in  cachec=c  Human  myotubes   DACOSTA_UV_RESPONSE_VIA_ERCC3_DN   DACOSTA_UV_RESPONSE_VIA_ERCC3_COMMON_DN   GARGALOVIC_RESPONSE_TO_OXIDIZED_PHOSPHOLIPIDS_UP   BUYTAERT_PHOTODYNAMIC_THERAPY_STRESS_UP   Gene  sets  Down  in  cachec=c  Human  myotubes   ELVIDGE_HIF1A_AND_HIF2A_TARGETS_DN   QI_HYPOXIA   WINTER_HYPOXIA_UP   ELVIDGE_HYPOXIA_BY_DMOG_UP  

FDR  q-­‐val   2.72E-­‐02   6.01E-­‐02   6.19E-­‐02   1.64E-­‐01   FDR  q-­‐val   0.00E-­‐00   0.00E-­‐00   5.45E-­‐04   2.57E-­‐03  

Supplementary  Table  3   Top  Gene  sets  Up  in  cachec=c  mouse  muscles   BUYTAERT_PHOTODYNAMIC_THERAPY_STRESS_UP   REACTOME_ACTIVATION_OF_NF_KAPPAB_IN_B_CELLS   DAUER_STAT3_TARGETS_UP   KEGG_PROTEASOME   Top  Gene  sets  Down  in  cachec=c  mouse  muscles   MOOTHA_VOXPHOS   REACTOME_RESPIRATORY_ELECTRON_TRANSPORT   REACTOME_GLYCOLYSIS_GLUCONEOGENESIS   EBAUER_MYOGENIC_TARGETS_OF_PAX3_FOXO1_FUSION  

Nature Medicine: doi:10.1038/nm.4093

FDR  q-­‐val   0.000   0.003   0.003   0.006   FDR  q-­‐val   0.000   0.000   0.003   0.004  

Supplementary  Table  4   Top  Gene  sets  Down  in  cachec=c  mouse  muscles  aVer  Etomoxir   FDR  q-­‐val   KEGG_PROTEASOME   0.000   REACTOME_ACTIVATION_OF_NF_KAPPAB_IN_B_CELLS   0.000   REACTOME_REGULATION_OF_MRNA_STABILITY_BY_PROTEINS_THAT_BIND_AU_RICH_ELEMENTS   0.000   BIOCARTA_PROTEASOME_PATHWAY   0.000   REACTOME_P53_DEPENDENT_G1_DNA_DAMAGE_RESPONSE   0.000   REACTOME_REGULATION_OF_APOPTOSIS   0.000   REACTOME_HOST_INTERACTIONS_OF_HIV_FACTORS   0.000   DAUER_STAT3_TARGETS_UP   0.000   RASHI_RESPONSE_TO_IONIZING_RADIATION_1   0.000   GARGALOVIC_RESPONSE_TO_OXIDIZED_PHOSPHOLIPIDS_BLUE_UP   0.001   Top  Gene  sets  Up  in  cachec=c  mouse  muscles  aVer  Etomoxir   FDR  q-­‐val   CHEMELLO_SOLEUS_VS_EDL_MYOFIBERS_UP   0.000   REACTOME_MUSCLE_CONTRACTION   0.000   EBAUER_MYOGENIC_TARGETS_OF_PAX3_FOXO1_FUSION   0.000   REACTOME_RESPIRATORY_ELECTRON_TRANSPORT_ATP_SYNTHESIS_BY_CHEMIOSMOTIC_COUPLING_A 0.000   ND_HEAT_PRODUCTION_BY_UNCOUPLING_PROTEINS   REACTOME_RESPIRATORY_ELECTRON_TRANSPORT   0.000   REACTOME_TCA_CYCLE_AND_RESPIRATORY_ELECTRON_TRANSPORT   0.000   KUNINGER_IGF1_VS_PDGFB_TARGETS_UP   0.000   KEGG_CARDIAC_MUSCLE_CONTRACTION   0.000   REACTOME_GLUCOSE_METABOLISM   0.001   DELASERNA_MYOD_TARGETS_UP   0.002  

Nature Medicine: doi:10.1038/nm.4093