Marini et al., Sci. Transl. Med. 10, eaat3504 (2018) 25 July 2018

S C I E N C E T R A N S L A T I O N A L M E D I C I N E | R E S E A R C H A R T I C L E

1 of 13

C A N C E R

Inhibition of activin signaling in lung adenocarcinoma increases the therapeutic index of platinum chemotherapyKieren D. Marini1,2, David R. Croucher3,4,5, Rachael A. McCloy3, Vijesh Vaghjiani1, Alvaro Gonzalez-Rajal3, Jordan F. Hastings3, Venessa Chin3,6, Anette Szczepny1, Kaja Kostyrko7, Cesar Marquez7, W. Samantha N. Jayasekara1, Muhammad Alamgeer1,2,8, Vishal Boolell1,8, Jeremy Z. R. Han3, Todd Waugh1, Hong Ching Lee3, Samantha R. Oakes3,4, Beena Kumar9, Craig A. Harrison1,2, Mark P. Hedger1,2, Nirmal Lorensuhewa10, Badia Kita10, Ross Barrow10, Bruce W. Robinson11, David M. de Kretser1,2,10, Jianmin Wu3,4,12,13, Vinod Ganju1,2, E. Alejandro Sweet-Cordero7, Andrew Burgess3,14, Luciano G. Martelotto2,15, Fernando J. Rossello16,17, Jason E. Cain1*, D. Neil Watkins3,4,18*

Resistance to platinum chemotherapy is a long-standing problem in the management of lung adenocarcinoma. Using a whole-genome synthetic lethal RNA interference screen, we identified activin signaling as a critical mediator of innate platinum resistance. The transforming growth factor– (TGF) superfamily ligands activin A and growth differentiation factor 11 (GDF11) mediated resistance via their cognate receptors through TGF-activated kinase 1 (TAK1), rather than through the SMAD family of transcription factors. Inhibition of activin receptor signaling or blockade of activin A and GDF11 by the endogenous protein follistatin overcame this resistance. Consistent with the role of activin signaling in acute renal injury, both therapeutic interventions attenuated acute cisplatin- induced nephrotoxicity, its major dose-limiting side effect. This cancer-specific enhancement of platinum- induced cell death has the potential to dramatically improve the safety and efficacy of chemotherapy in lung cancer patients.

INTRODUCTIONOver one-third of lung cancers are adenocarcinomas, an aggressive malignancy that is frequently inoperable at diagnosis. Despite the introduction of targeted therapy and immunotherapy, most patients still receive platinum chemotherapy. Although platinum agents re-main the cornerstone of treatment, responses remain below 30% (1). Therefore, strategies to improve the efficacy of chemotherapy could improve outcomes for many lung cancer patients. More than 147 potential mechanisms of platinum resistance have been proposed (2), thus presenting a far more complex problem than the expres-

sion of multidrug resistance proteins (2). Here, we report the results of a whole-genome synthetic lethal small interfering RNA (siRNA) screen for innate platinum resistance genes. Although the screen identified several genes that encode therapeutically targetable pro-teins, a network analysis revealed that activin signaling, part of the transforming growth factor– (TGF) superfamily (3), is an impor-tant mediator of resistance. We further show that activin A and the related ligand growth differentiation factor 11 (GDF11) activate TGF- activated kinase 1 (TAK1), an SMAD-independent component of TGF signaling known to be a critical regulator of cell death (4). Finally, we show that inactivation of activin signaling with a small- molecule inhibitor or with the endogenous activin/GDF11-binding protein follistatin (FST) can overcome innate platinum resistance in vivo.

RESULTSOptimization of an in vitro model of platinum-resistant lung adenocarcinomaOn the basis of the work of Hall et al. (5), we concluded that the instability of cisplatin in dimethyl sulfoxide (DMSO) made it un-suitable for a screen of this size, whereas carboplatin is stable in aqueous solution. We therefore assessed a series of human lung cancer cell lines for sensitivity to carboplatin in vitro (Fig. 1A). Tak-ing into account that the peak plasma concentration of carboplatin in humans is reported as 30 to 50 g/ml, and the lack of evidence that platinum compounds are concentrated within tumors (6), we classified A549, NCI-H358, NCI-H2009, and NCI-H727 as platinum- resistant adenocarcinoma cell lines (Fig. 1A).

We further assessed the well-characterized KRAS-mutant lung adenocarcinoma cell line A549, derived from a treatment-naïve patient,

1Hudson Institute of Medical Research, Clayton, Victoria 3168, Australia. 2Faculty of Medicine, Nursing and Health Sciences, Monash University, Clayton, Victoria 3800, Australia. 3The Kinghorn Cancer Centre, Garvan Institute of Medical Research, Darlinghurst, New South Wales 2010, Australia. 4St. Vincent’s Clinical School, Faculty of Medicine, University of New South Wales Sydney, Darlinghurst, New South Wales 2010, Australia. 5School of Medicine, University College Dublin, Belfield, Dublin D04 V1W8, Ireland. 6Department of Medical Oncology, St. Vincent’s Hospital, Darlinghurst, New South Wales 2010, Australia. 7Department of Pediatrics, University of California San Francisco, San Francisco, CA 94158, USA. 8Department of Medi-cal Oncology, Monash Medical Centre, East Bentleigh, Victoria 3165, Australia. 9Department of Pathology, Monash Medical Centre, Clayton, Victoria 3168, Australia. 10Paranta Biosciences Limited, Melbourne, Victoria 3004, Australia. 11School of Med-icine and Pharmacology, Queen Elizabeth II Medical Centre Unit, University of Western Australia, Crawley, Western Australia 6009, Australia. 12Key Laboratory of Carcinogenesis and Translational Research, Ministry of Education, Beijing, China. 13Center for Cancer Bioinformatics, Peking University Cancer Hospital and Institute, Hai-Dian District, Beijing 100142, China. 14ANZAC Research Institute, Concord, New South Wales 2139, Australia. 15Center for Cancer Research, Faculty of Medicine, Dentistry and Health Sciences, University of Melbourne, Melbourne, Victoria 3000, Australia. 16Australian Regenerative Medicine Institute, Clayton, Victoria 3800, Australia. 17Department of Anatomy and Developmental Biology, Monash University, Clayton, Victoria 3800, Australia. 18Department of Thoracic Medicine, St. Vincent’s Hospital, Darlinghurst, New South Wales 2010, Australia.*Corresponding author. Email: n.watkins@garvan.org.au (D.N.W.); jason.cain@hudson.org.au (J.E.C.)

Copyright © 2018 The Authors, some rights reserved; exclusive licensee American Association for the Advancement of Science. No claim to original U.S. Government Works

by guest on October 11, 2020

http://stm.sciencem

ag.org/D

ownloaded from

Marini et al., Sci. Transl. Med. 10, eaat3504 (2018) 25 July 2018

S C I E N C E T R A N S L A T I O N A L M E D I C I N E | R E S E A R C H A R T I C L E

2 of 13

0 2000 4000 6000 8000 10000 12000 14000 16000 180000

5

10

15

20

25

30

Genes in rank order

Sen

siti

zati

on

ind

ex

BA

HEK293

Genes (n = 230)

161Sensitization index

C

F

4 8 12

0 12.5 25 50γH2AX

DAPI

Carboplatin µg/ml

siRNA1siRNA2siRNA3siRNA4

A549

ED

MCL1ACVR1B FKBP3G

ACVR1B

TGFBR1

BCL2L1

FKPB3

PARP1

VEGFB

MCL1

H

Mock NT siACVR1B

ACVR1B

Actin

kDa55

42

Mock NT siFKBP3

FKBP3

Actin

kDa

40

42

Mock NT siMCL1

MCL1

Actin

kDa40

42

LX22CL

NCI-H20

9

NCI-H83

8

NCI-H12

99

NCI-H23

NCI-H19

75A54

9

NCI-H35

8

NCI-H20

09

NCI-H72

70

20

40

60

80

100

IC50

(µg

/ml)

SCLC

Adenocarcinoma resistantAdenocarcinoma sensitive

0.001 0.01 0.1 1 10 1000

20

40

60

80

100

Carboplatin (µg/ml)

Via

bili

ty (

%)

NT siRNAsiACVR1B

IC50

69.4 + 7.819.41 + 4.1

0.001 0.01 0.1 1 10 1000

20

40

60

80

100

Carboplatin (µg/ml)

Via

bili

ty (

%)

NT siRNAsiFKBP3

IC50

69.4 + 7.812.91 + 3.2

0.001 0.01 0.1 1 10 1000

20

40

60

80

100

Carboplatin (µg/ml)

Via

bili

ty (%

)

NT siRNAsiMCL1

IC50

69.4 + 7.89.41 + 1.1

0 10 20 30 40 50 60 70 80 90 1001

1010

100

Carboplatin (µg/ml)

Via

bili

ty (

%)

A549LX22CL

10

Fig. 1. An in vitro siRNA screen for carboplatin synthetic lethality in lung adenocarcinoma cells. (A) Carboplatin median inhibitory concentration (IC50) values in lung cancer cell lines, classified as platinum-resistant or sensitive in vitro. n = 4, mean ± SEM. Dotted line indicates peak plasma concentration for carboplatin in humans. (B) Viability of A549 and LX22CL cells treated with a range of carboplatin concentrations in vitro. n = 4, mean ± SEM. Red dashed line indicates peak plasma concentrations for carboplatin in humans; black dashed line, concentration chosen for screen. (C) H2AX immunofluorescence images in A549 cells treated with carboplatin for 24 hours, counterstained with 4′,6-diamidino-2-phenylindole (DAPI). Scale bar, 5 m. (D) Overview of the synthetic lethal screen methodology. (E) Results of a whole-genome siRNA screen for carboplatin synthetic lethality for each gene in order of sensitization index. The top 5% of hits are shown in red. Potentially targetable genes are listed. (F) Secondary screening of the top 230 hits, depicted as a sensitization index heat map. Deconvolution of the primary screen was performed using individual siRNAs (1 to 4) and pooled siRNAs targeting the same genes in HEK293 cells. (G) Western blot analysis of knockdown efficiency of ACVR1B, FKBP3, and MCL1 siRNAs orthogonal to the siRNAs used in the screen. (H) Viability of A549 cells treated with nontargeting (NT) or specific siRNAs in combination with a range of carboplatin concentrations in vitro. IC50 carboplatin concentrations are shown. n = 4, mean ± SEM.

by guest on October 11, 2020

http://stm.sciencem

ag.org/D

ownloaded from

Marini et al., Sci. Transl. Med. 10, eaat3504 (2018) 25 July 2018

S C I E N C E T R A N S L A T I O N A L M E D I C I N E | R E S E A R C H A R T I C L E

3 of 13

as a model of innate chemoresistance. As shown in Fig. 1B, the plati-num response curve from plotting cell viability as log10 on the y axis according to the method of Stewart (2) indicates that A549 cells exhibit active saturable resistance. By contrast, the small cell lung cancer (SCLC) cell line LX22CL (7) was highly sensitive (Fig. 1B). Detection of DNA damage in A549 cells occurred at carboplatin concentrations well below the IC50 (Fig. 1C), suggesting that carbo-platin export is not the predominant mechanism of innate resist-ance. On the basis of these data, we chose A549 cells as the model for our synthetic lethal screen.

A screen for carboplatin synthetic lethality in A549 cellsThe screen was based on the method of Whitehurst et al. (8), using cell viability as an endpoint. Optimization was performed with siRNA targeting polo-like kinase 1 (siPLK1), a positive control inducing platinum-independent cell death, or mammalian target of rapamycin (siMTOR) as a control known to induce sensitization to platinum (fig. S1A) (2). A sublethal concentration of carboplatin (25 g/ml) was chosen empirically (fig. S1A). These controls revealed that the screen performed within acceptable limits of variability, quality, and reproducibility determined by Z′ factor analysis (fig. S1B) (8). The screen was then performed using the Dharmacon whole-genome siRNA library along with vehicle or carboplatin (Fig. 1D) generat-ing a sensitization index, a measure of the capacity for a given siRNA increase in response to carboplatin, for all 18,178 genes (Fig. 1E). High-confidence hits were defined by the rank order of the sensi-tization index and the false discovery rate (FDR)–corrected P value (table S1).

The top 230 hits underwent deconvolution screening in A549 cells, and as pooled siRNAs in the immortalized human renal epithelial cell line human embryonic kidney (HEK) 293 cells, representing a noncancer control. Of these, 79.1% passed the deconvolution screen-ing, defined as sensitization in response to three of the four individ-ual siRNAs targeting each under the same conditions as the primary screen (Fig. 1F). By contrast, only 24.6% of the pooled siRNAs sen-sitized HEK293 cells to carboplatin. Orthogonal siRNA validation with siRNAs of unrelated sequence was then performed on three targets chosen arbitrarily, activin receptor 1 B (ACVR1B), FK506- binding protein 3 (FKBP3), and myeloid cell leukemia 1 (MCL1). This demonstrated that knockdown of the target protein was asso-ciated with sensitization to carboplatin in vitro (Fig. 1, G and H). These data show that the screening methodology could reliably identify gene targets commonly expressed in lung adenocarcinoma.

Identification of candidate therapeutic targetsThe Dharmacon library contains 4795 genes arbitrarily classified as “druggable” (9). Searching of online databases revealed that 36 of these genes coded for proteins that could be inhibited by large or small mol-ecules, and 15 of these genes were previously implicated in mediating platinum resistance (table S2). Several candidate genes encode pro-teins that can be inhibited by compounds or antibodies in clinical use (table S2). Although these therapies may improve the response to platinum, combinatorial toxicity may represent a barrier to effective translation. We therefore pursued an alternative approach by look-ing for signaling pathways that coordinate innate platinum resistance.

Network analysis of platinum resistance genesThe top 5% of high-confidence hits were submitted for a network analysis using ClueGo/Cytoscape (Fig. 2) (10). Upstream, drivers of

the platinum response were dominated by cytokines and mitogen- activated protein kinase (MAPK) signaling. Downstream, the pre-dominance of DNA repair genes in the analysis is consistent with previous reports that removal of platinum adducts is a major factor in resistance (11). As shown in Fig. 2, signaling through TAK1, a member of the MAP3K family, appeared to be a critical signaling node strongly linked downstream to pathways mediating cell cycle arrest and DNA repair. Because TAK1 has been linked to the re-s ponse to cellular stress, including the generation of reactive oxygen species, necrosis, and apoptosis (4), we explored potential strategies to disable this signaling node as a therapeutic approach.

TGF superfamily genes as mediators of innate platinum resistanceTo further explore TAK1 as a mediator of innate resistance, we re-analyzed the results of the primary RNA interference screen to identify candidate genes related to the TGF signaling pathway. Figure 3A shows that knockdown of the TGF orthologs GDF11 and inhibin A (INHBA), and their receptor heterodimers, en-hanced the efficacy of carboplatin in vitro. GDF11 can activate both TGF and activin receptors (12) and has been implicated in the pathogenesis of hematopoietic failure (13) and sarcopenia (14). In the context of inflammation, INHBA homodimerizes to form activin A, a TGF superfamily ligand implicated in tissue injury, fibrosis, and cachexia (15). Although canonical TGF signaling has been implicated in epithelial cancers as a mediator of drug resistance, our screen did not identify the SMAD family transcription factors, which mediate the related phenotypes of stemness and epithelial- mesenchymal transition (16).

Western blot analysis of A549 cells treated with carboplatin in vitro revealed up-regulation of phospho-TAK1 (p-TAK1), as well as the 55-kDa activin A precursor and its active 25-kDa form, and an increase in the expression of the activin receptor ACVR1B (Fig. 3, B and C). Using an antibody that recognizes both GDF8 and GDF11 (14), we also saw increased expression of the 45-kDa precursor form. Although we observed a modest increase in the expression of the 67-kDa TGF precursor protein, we could not detect expression of the 12-kDa active form in response to carboplatin (Fig. 3B). In keeping with the data shown in Fig. 3A, carboplatin did not induce phosphorylation of SMAD2/3 (Fig. 3B and fig. S2A), thus confirm-ing the results of the primary screen suggesting that the transcrip-tional signaling arm of the TGF superfamily does not play a role in mediating resistance in A549 cells.

To further validate TAK1 signaling as a mediator of platinum treatment resistance, we first showed that the TAK1 inhibitor LLZ-1640-2 (17) sensitized A549 cells to carboplatin in vitro (Fig. 3D). Interrogation of events downstream of TAK1 in this model sup-ported the results of the primary screen by showing that LLZ-1640-2 blocked the phosphorylation of MAPK kinase 4 (MKK4) and P38 (Fig. 3E), as well as phosphorylation of MCL1 threonine 184/187, a P38 target site that mediates MCL1 stabilization (18). These data are supported by orthogonal siRNA validation of MCL1 (Fig. 1, G and H) and are consistent with a model in which TAK1 signaling mediates platinum resistance through MAPK signaling and stabili-zation of MCL1 (Fig. 3A). The importance of TAK1 as a mediator of platinum resistance was further supported in A549 cells by or-thogonal siRNA directed at MAP3K7, the gene encoding TAK1 (Fig. 3F), and a clustered regularly interspaced short palindromic repeats (CRISPR)/CRISPR associated protein 9 (Cas9) knockout

by guest on October 11, 2020

http://stm.sciencem

ag.org/D

ownloaded from

Marini et al., Sci. Transl. Med. 10, eaat3504 (2018) 25 July 2018

S C I E N C E T R A N S L A T I O N A L M E D I C I N E | R E S E A R C H A R T I C L E

4 of 13

Fig. 2. Network analysis of the top 5% of hits from the primary screen in A549 cells. Each node represents an enriched pathway, and the size of each node reflects the statistical significance of the term. The thickness of the interconnecting lines reflects the number of interactions between each term. Colors represent broader categories, where all nodes that share the same color represent interrelated terms. IFN, interferon; TCR, T cell receptor; GM-CSF, granulocyte-macrophage colony-stimulating factor; JNK, c-Jun N-terminal kinase; TRAF1, TNF receptor–associated factor 1; NFB, nuclear factor B; cGMP, guanosine 3′,5′-monophosphate; DCC, dicyclohexylcarbodiimide; LDL, low-density lipoprotein; GPCR, G protein–coupled receptor; RIG-I, retinoic acid–inducible gene-1; MDA5, melanoma differentiation–associated gene 5; ISG15, interferon- stimulated protein, 15 KDa; FGFR, fibroblast growth factor receptor; CaM, calmodulin; IRS, insulin receptor substrate; ATM, ataxia telangiectasia–mutated.

by guest on October 11, 2020

http://stm.sciencem

ag.org/D

ownloaded from

Marini et al., Sci. Transl. Med. 10, eaat3504 (2018) 25 July 2018

S C I E N C E T R A N S L A T I O N A L M E D I C I N E | R E S E A R C H A R T I C L E

5 of 13

model (Fig. 3G and fig. S2, B and C). By contrast, LLZ-1640-2 did not increase the toxicity of carboplatin in the immortalized human embryonic kidney cell line HEK293 (fig. S2D). These results suggest that in the face of platinum-induced stress, TAK1 signaling may con-fer a tumor-specific role in mediating resistance.

Combined inhibition of ACVR1B and TGFBR1, TAK1 signaling, and platinum resistanceTo precisely define the mechanism of TAK1 activation in response to carboplatin in lung adenocarcinoma cells, we used the small- molecule SB-505124, a potent and specific inhibitor of ACVR1B

A B C

E

D

F G

012345

10

15

Rel

ativ

e ex

pre

ssio

n

p-TAK1 Activin A ACVR1B GDF11

Time (hr) 0 12 24 48

**

**

*

0.001 0.01 0.1 1 10 1000

20

40

60

80

100

Carboplatin (µg/ml)

Via

bili

ty (

%)

DMSOLLZ-1640-2

IC50

79.4 + 7.816.55 + 3.2

0.001 0.01 0.1 1 10 1000

20

40

60

80

100

Carboplatin (µg/ml)

Via

bili

ty (

%)

NT siRNAsiTAK1

IC50

69.4 + 7.814.55 + 3.2

0.001 0.01 0.1 1 10 1000

20

40

60

80

100

Carboplatin (µg/ml)

Via

bili

ty (

%)

WT

IC50

53.4 + 7.218.8 + 2.1CRISPR clone 1 12.2 + 3.2CRISPR clone 3

Fig. 3. Activin A and GDF11 signaling and platinum resistance in A549 cells in vitro. (A) Flowchart depicting the sensitization index (change in viability comparing carboplatin alone, siRNA alone, and carboplatin + siRNA) of genes in the activin/TGF pathway from the primary screen. ROS, reactive oxygen species. (B) Western blot analysis for phosphorylated TAK1 (p-TAK1), activin A (precursor, 55 kDa; active fragment, 25 kDa), GDF8/11, TGF (precursor, 67 kDa; active fragment, 12 kDa), ACVR1B, TGFBR1, phospho-SMAD2/3 (p-SMAD2/3), SMAD2/3, and actin in A549 cells treated with carboplatin (25 g/ml) for 0 to 48 hours (hr). Positive controls for p-SMAD2/3 and SMAD2/3 are shown in fig. S4A. (C) Quantitative analysis of the data shown in Fig. 3B. n = 4, mean ± SEM. **P < 0.01, *P < 0.05, one-way analysis of variance (ANOVA)/Bonferroni correction. (D) Viability of A549 cells after treatment with vehicle (DMSO) or the TAK1 inhibitor LLZ-1640-2 (10 M) in combination with a range of carboplatin concentrations. Carboplatin IC50 concentrations are shown. n = 4, mean ± SEM. (E) Western blot analysis of A549 cells treated with vehicle (DMSO) or LLZ-1640-2 (10 M), in combination with carboplatin (Carbo) at 25 mg/ml for 0 to 48 hours. Blots were probed with antibodies specific for phosphorylated TAK1 (p-TAK1, T184/187), TAK1, phosphorylated MKK4 (p-MKK4, S257/T261), MKK4, phosphorylated P38 (p-P38, T180/Y182), P38, phosphorylated MCL1 (p-MCL1, T163), MCL1, and actin. (F) Orthogonal siRNA validation of TAK1. Top: Western blot analysis of TAK1 expression in A549 cells treated with a nontargeting (NT) siRNA, siRNA directed at MAP3K7 (the gene coding for TAK1), orthogonal to the siRNA used in the screen, the gene encoding TAK1. Bottom: Viability of A549 cells after treatment with a NT siRNA or orthogonal siRNA directed at MAP3K7, in combination with a range of carboplatin concentrations in vitro. Carboplatin IC50 concentrations for each experimental group are shown. n = 4, mean ± SEM. (G) CRISPR/Cas9 knockout of TAK1. Top: Western blot analysis of TAK1 expression in control cells and in cells carrying a MAP3K7 mutation induced by CRISPR/Cas9. Bottom: Viability of A549 parental cells and cells carrying a MAP3K7 mutation induced by CRISPR/Cas9. n = 4, mean ± SEM. Carboplatin IC50 concentrations for each group are shown. WT, wild type.

by guest on October 11, 2020

http://stm.sciencem

ag.org/D

ownloaded from

Marini et al., Sci. Transl. Med. 10, eaat3504 (2018) 25 July 2018

S C I E N C E T R A N S L A T I O N A L M E D I C I N E | R E S E A R C H A R T I C L E

6 of 13

and TGFBR1 (19). Treatment with SB-505124 overcame resistance at clinically relevant concentrations of carboplatin without single- agent activity (Fig. 4A). In each case, robust up-regulation of either activin A or GDF8/11 was seen, as well as phosphorylation of TAK1 (p-TAK1), consistent with TAK1 activation in response to carbo-platin (Fig. 4B). Consistent with the data presented in Fig. 3B, we did not observe phosphorylation of SMAD2/3 in response to carbo-platin in these cell lines; however, in each case, the induction of MCL1 expression was inhibited or abolished (Fig. 4B and fig. S3A). By contrast, SB-505124 treatment did not enhance the effect of car-boplatin in platinum-sensitive cell lines (fig. S4A). NCI-H1299 cells did not up-regulate p-TAK1 or MCL1 in response to carboplatin, whereas in NCI-H1975 cells, the modest TAK1 response was not blocked by SB-505124 (fig. S4B). In all three platinum-sensitive lines, we did not detect SMAD2/3 phosphorylation (fig. S3B).

Sensitization to cisplatin over a range of clinically relevant con-centrations was also observed in A549 cells, demonstrating that this effect is seen with the two platinum agents most commonly used in the treatment of lung adenocarcinoma (fig. S4C). Similar results were seen with the alkylating agent busulfan, but not with doxoru-bicin, docetaxel, pemetrexed, or gemcitabine (fig. S4C). These re-sults suggest that adduct-forming cytotoxic drugs such as cisplatin and busulfan specifically activate TAK1 downstream of ACVR1B/TGFBR1.

Expression of TGF superfamily components in lung adenocarcinomaAnalysis of The Cancer Genome Atlas (TCGA) lung adenocarcino-ma data set (20) revealed that genes represented in our primary screen including the ligands INHBA, INHBB, GDF11, the receptors ACVR1B and TGFBR1, and their relevant heterodimerization part-ners were all highly expressed (Fig. 5A). Analysis of the KMPlot lung adenocarcinoma data set (21) showed that expression of GDF11 and its receptors TGFBR1 and ACVR1B was associated with a reduction in overall survival (Fig. 5B). In keeping with published data (22), expression of activin components INHBA and INHBB was also associated with poor outcome. (Fig. 5B). By contrast, expression of TGFB1 was strongly associated with a favorable outcome, once again in keeping with published studies (23). By integrating these data with the results of our primary screen, we postulated that GDF11, acting through both TGF and activin receptors, and activin A, acting through activin receptors, might mediate innate platinum resistance in lung adenocarcinoma.

To further address the relationship of activin A and GDF8/11 ex-pression with platinum resistance, we performed an immunohisto-chemical analysis of diagnostic core biopsy samples from 59 patients with stage IV lung adenocarcinoma without activating mutations in either EGFR or ALK treated with platinum-based chemotherapy (table S3). Expression of activin A, or coexpression of both activin A and GDF8/11, was associated with significant (P < 0.05) reduc-tion in progression-free survival (Fig. 5, C and D, and fig. S5). These results suggest that activin signaling, driven by activin A and/or GDF11, plays an important role in mediating innate platinum resist-ance in lung adenocarcinoma.

Inhibition of ACVR1B/TGFBR1 signaling, platinum resistance, and nephrotoxicity in vivoWe next asked whether the ACVR1B/TGFBR1 inhibitor SB-505124 could increase the efficacy of carboplatin in an athymic nude mouse

model. We chose the A549 cell line based on its engraftment effi-ciency, highly reproducible rate of growth in vivo, and resistance to a single dose of carboplatin (60 mg/kg) (fig. S6A), a dose that reliably induces a complete response in the LX22 SCLC xenograft model (fig. S6A). In keeping with our in vitro data, we also observed robust expression of both ACVR1B and TGFBR1 in A549 xenografts (Fig. 6A).

We administered SB-505124 to mice with A549 flank xenografts in combination with carboplatin according to the regimen depicted in Fig. 6B. As expected from our in vitro data, SB-505124 alone had no effect, but administration of SB-505124 with a single dose of car-boplatin resulted in durable responses without the need for mainte-nance therapy in five animals (Fig. 6, C and D).

Nephrotoxicity is the major dose-limiting toxicity of cisplatin (24), and multiple lines of evidence implicate activation of TAK1 (25) and TGF (26) signaling in the pathogenesis of acute kidney injury. We therefore treated mice with SB-505124 in combination with a single nephrotoxic dose of cisplatin, as described in Fig. 6E. Endpoint measurements of serum creatinine and urea demonstrated that SB-505124 was highly effective at reducing cisplatin-mediated kidney injury (Fig. 6, F and G). Periodic PAS staining of kidney sec-tions from cisplatin-treated animals (Fig.  6G) showed changes characteristic of acute tubular injury, with staining in the medulla of the affected kidneys. In contrast, kidneys from the animals treated with cisplatin in combination with SB-505124 resembled controls (Fig. 6G). High-magnification images (fig. S6B) demonstrated that the typical features of acute tubular necrosis, including loss of the epithelial brush border, tubular necrosis, tubular dilation, and cast formation, were attenuated by the SB-505124 treatment.

Inhibition of activin A/GDF11 signaling, platinum resistance, and nephrotoxicity in vivoOne potential disadvantage of small-molecule inhibitors in combi-nation with chemotherapy is unexpected toxicity due to on-target or off-target effects. We therefore considered whether FST might serve as a more tractable therapeutic approach. FST is an endogenous pro-tein that regulates several developmental, reproductive, and inflam-matory components of the TGF signaling pathway by binding to and inhibiting the activity of activin A, GDF8, and GDF11 (27). FST does not bind to or inhibit TGF1 and TGF2, therefore providing an experimental tool to support our data suggesting that ACVR1B and TGFRB1 are activated by activin A and/or GDF11.

Before testing the efficacy of FST for platinum sensitization in vivo, we determined the expression of its molecular targets, activin A and GDF8/11, in A549, NCI-H358, and NCI-H727 xenografts, as well PDX1.1, a treatment-naïve, chemoresistant patient-derived xeno-graft. In addition, we performed the same assessment in the im-munocompetent murine Lewis lung adenocarcinoma (LLC) model (Fig. 7A). Quantitative analysis showed that expression of both activin A and GDF8/11 was variable, and in the case of NCI-H727 cells, lower than expected based on our in vitro results (Fig. 7B). Tumors generated in the KRas/Tp53 conditional mouse model of lung adenocarcinoma (28) did not express either activin A or GDF8/11 (fig. S7A).

Purified recombinant human FST was administered along with carboplatin according to the regimen outlined in Fig. 7C. Figure 7D shows that FST had no effect on the growth of A549 flank xenografts as a single agent. However, administration of FST enhanced the ef-fect of carboplatin, with 7 of 10 mice achieving a complete response

by guest on October 11, 2020

http://stm.sciencem

ag.org/D

ownloaded from

Marini et al., Sci. Transl. Med. 10, eaat3504 (2018) 25 July 2018

S C I E N C E T R A N S L A T I O N A L M E D I C I N E | R E S E A R C H A R T I C L E

7 of 13

(Fig. 7D). Administration of carboplatin with FST also resulted in a dramatic improvement in survival consistent with durable and bio-logically meaningful therapeutic response (Fig. 7E). Similar results were seen in the NCI-H358 and PDX1.1 models, both of which overexpressed GDF8/11 (Fig. 7, A, B, and F, and fig. S7B). Combi-nation carboplatin/FST treatment also had a significant effect on survival in the fast-growing LLC model; however, no effect was seen in the NCI-H727 xenografts, which did not overexpress activin A or GDF8/11 in vivo (Fig. 7, A, B, and F, and fig. S7B). Using the exper-imental protocol shown in Fig. 7G, administration of FST also pre-vented renal failure (Fig. 7H) and kidney injury (Fig. 7I and fig.

S7C) in response to a single nephrotoxic dose of cisplatin. These data provide strong support for the development of FST as an inhibitor of activin/GDF11 signaling to overcome innate platinum resistance in lung adenocarcinoma.

DISCUSSIONTo increase efficacy, platinum agents are typically given in combi-nation with a taxane, pemetrexed, or gemcitabine (29). These regi-mens are largely based on historical clinical trial data rather than on mechanism and result in additional toxicity when compared to platinum

NCI-H727 NCI-H2009

B

A A549 NCI-H358KRAS G12STP53 WTCDKN2A DelSTK11 Q37*

No prior treatment

KRAS G12CTP53 DelCDKN2A WTSTK11 WT

No prior treatment

KRAS G12VTP53 Q165insYKQCDKN2A WTSTK11 WT

Treatment status unknown

KRAS G12ATP53 R273LCDKN2A WTSTK11 WT

Prior chemo-radiation

0.001 0.01 0.1 1 10 1000

20

40

60

80

100

Carboplatin (µg/ml)

Via

bili

ty (%

)

DMSOSB-505124

IC50

69.4 + 5.817.2 + 4.4

0.001 0.01 0.1 1 10 1000

20

40

60

80

100

Carboplatin (µg/ml)

Via

bili

ty (%

)

DMSOSB-505124

IC50

91.3 + 7.821.8 + 5.2

0.001 0.01 0.1 1 10 1000

20

40

60

80

100

Carboplatin (µg/ml)

Via

bili

ty (%

)

DMSOSB-505124

IC50

69.1 + 3.511.1 + 6.5

0.001 0.01 0.1 1 10 1000

20

40

60

80

100

Carboplatin (µg/ml)

Via

bili

ty (%

)

DMSOSB-505124

IC50

87.3 + 7.89.12 + 2.5

Fig. 4. Activation of TAK1 by ACVR1B and TGFBR1 after treatment with carboplatin in vitro. (A) Viability of A549, NCI-H358, NCI-H727, and NCI-H2009 lung adeno-carcinoma cells after treatment with vehicle (DMSO) or the ACVR1B/TGFBR1 inhibitor SB-505124 (5 M) in combination with a range of carboplatin concentrations. Carboplatin IC50 concentrations for each experimental group are shown. n = 4, mean ± SEM. Genotypes and treatment histories are listed. Del, homozygous deletion; Ins, in-frame insertion. (B) Western blot analysis of phospho-TAK1 (p-TAK1), TAK1, phospho-SMAD2/3 (p-SMAD2/3), SMAD2/3, MCL1, activin A, GDF8/11, and actin expression in A549, NCI-H358, NCI-H727, and NCI-H2009 cells treated with vehicle (DMSO) or SB-505124 (5 M) in combination with carboplatin at 25 mg/ml for 0 to 48 hours. Posi-tive controls for p-SMAD2/3 and SMAD2/3 are shown in fig. S4.

by guest on October 11, 2020

http://stm.sciencem

ag.org/D

ownloaded from

Marini et al., Sci. Transl. Med. 10, eaat3504 (2018) 25 July 2018

S C I E N C E T R A N S L A T I O N A L M E D I C I N E | R E S E A R C H A R T I C L E

8 of 13

given as a single agent (29). Moreover, pharmacologic strategies aimed at improving platinum responses have the potential to increase tox-icity if they target pathways that repair platinum DNA adducts in normal cells. Our results suggest that a mechanism-based approach to reversing platinum resistance could be used to identify more spe-cific and less toxic combination therapies.

A wealth of evidence supports a role for TGF signaling through the SMAD transcription factors as an important mediator of chemo-resistance in many different tumor types (30). Our screen found that the MAP3K TAK1 is an important mediator of platinum resistance in lung adenocarcinoma and that its activation is driven by signal-ing through activin and TGF in response to the ligands activin A

A B

C D Activin A Activin A + GDF8/11Activin A GDF8/11

Fig. 5. Expression of TGF superfamily components in lung adenocarcinoma. (A) Heat map depicting expression of genes belonging to the TGF superfamily in the TCGA lung adenocarcinoma data set. (B) Association between GDF11, TGFB1, INHBA, INHBB, ACVR1B, and TGFBR1 expression and overall survival in the KMPlot lung ade-nocarcinoma data set. (C) Activin A and GDF8/11 expression detected by immunohistochemistry (IHC) in stage IV lung adenocarcinoma. Immunoperoxidase signal is shown in brown, counterstained with hematoxylin. Scale bars, 50 m. (D) Kaplan-Meier analysis of progression-free survival (PFS) in a cohort of patients with stage IV lung adenocarcinoma (n = 59) treated with platinum-based chemotherapy. Survival outcomes are shown with respect to quantitative immunohistochemical analysis of activin A and GDF8/11 staining. Statistical significance was calculated using log-rank analysis. CI, confidence interval; HR, hazard ratio.

by guest on October 11, 2020

http://stm.sciencem

ag.org/D

ownloaded from

Marini et al., Sci. Transl. Med. 10, eaat3504 (2018) 25 July 2018

S C I E N C E T R A N S L A T I O N A L M E D I C I N E | R E S E A R C H A R T I C L E

9 of 13

and GDF11. In contrast to SMAD signaling, the TAK1 component of the TGF pathway is more amenable to small-molecule targeting (30). Notably, our data suggest that ACVR1B/TGFBR1 signaling mediates protection against chemotherapy in cancer cells and pro-motes kidney injury. Up-regulation of activin signaling is also a fea-ture of distal airway epithelial cells in response to inflammation (3). On the basis of the putative cell of origin of lung adenocarcinoma, the up-regulation of activin signaling in the face of chemotherapeu-tic stress may be a consequence of cell lineage rather than genotype.

There are several points that limit the interpretation of our re-sults. First, we have focused on innate platinum resistance by per-forming our screen on the treatment-naïve cell line A549, and thus, we cannot make definitive conclusions regarding acquired resist-ance. Second, we concentrated on KRAS/NRAS mutant models as a clinical priority, and so we cannot generalize our results to lung adenocarcinomas without a defined oncogenic mutation. Third, we

have not identified a mechanism to explain why TAK1 activation rather than SMAD-mediated transcription is favored by lung ade-nocarcinoma cells in response to platinum chemotherapy. Finally, the differences in pharmacodynamics between cisplatin and carbo-platin (11) will need to be considered when considering the poten-tial efficacy of FST in humans.

Recombinant FST as a potential platinum-sensitization agent has several advantages over small molecule–targeting strategies, be-cause it is an endogenous protein that does not require humaniza-tion. Our in vivo data show that FST can dramatically enhance the efficacy of carboplatin in a highly resistant model of lung adenocar-cinoma while protecting against cisplatin nephrotoxicity. Although these animals did not exhibit signs of long-term neurologic deficits, large-animal studies will be needed to exclude an interaction be-tween FST and platinum agents with respect to neurotoxicity. In addition, FST is able to block the activity of GDF8 (myostatin) and

0 20 40 60 80 100 1200

100

200

300

400

500

600

Time (days)

Tum

or

volu

me

(mm

3 )

CarboCarbo + SB-505124

B

A ACVR1B TGFBR1

Urea CreatinineF

C

E

G

D

Vehicle SB-505124 CisPt +Vehicle

CisPt + SB-505124

0 10 20 300

100

200

300

400

500

600

Time (days)

Tum

or

volu

me

(mm

3 )

ControlSB-505124

0 20 40 60 80 100 120 1400

20

40

60

80

100

Time (days)

Su

rviv

al (%

)

Control (8)

Pt (8)SB-505124 (9)

SB-505124 + Pt (10) ***

0255075

100125150

Pla

sma

con

c (m

M)

***

VehicleSB-505124

CisPtCisPt + SB-505124

+–––

+––

–

+–

––

+––

+

0

50

100

150

200

Pla

sma

con

c (µ

M)

**

+–––

+––

–

+–

––

+––

+

Fig. 6. Inhibition of ACVR1B/TGFR1 signaling in vivo. (A) ACVR1B and TGFBR1 expression de-tected by immunohistochemistry in A549 xeno-grafts. Immunoperoxidase signal is shown in brown, counterstained with hematoxylin. Scale bars, 100 m. (B) Schema describing an experiment to determine the effect of combination carboplatin and SB-505124 treatment on the growth of A549 flank xenograft tumors. (C) Growth of A549 xenografts in mice treated with vehicle control, SB-505124, and/or Carbo. Data are presented as the growth of tumors in independent mice. (D) Kaplan-Meier analysis of survival to ethical endpoint of mice from the same experiment. n for each treatment group is shown. ***P < 0.001 compared to mice treated with vehicle + cisplatin (CisPt), log-rank test. (E) Schema describing an experiment to determine the effect of treatment with SB-505124 on acute kidney injury induced by CisPt. (F) Serum urea and creatinine concen-trations in mice treated with vehicle, SB-505124, and/or CisPt. Blood samples were taken on day 5. n = 7, mean ± SEM. **P < 0.01, ***P < 0.001, one-way ANOVA with Bonferroni correction. (G) Top: Macroscopic appearance of kidneys from mice from the experiment in (E) and (F). Scale bar, 2 mm. Bottom: Photomicrographs of periodic acid-Schiff (PAS)–stained sections of whole kidneys. Scale bar, 2 mm.

by guest on October 11, 2020

http://stm.sciencem

ag.org/D

ownloaded from

Marini et al., Sci. Transl. Med. 10, eaat3504 (2018) 25 July 2018

S C I E N C E T R A N S L A T I O N A L M E D I C I N E | R E S E A R C H A R T I C L E

10 of 13

0 40 80 120 160 2000

100

200

300

400

500

600

Time (days)

Tu

mo

r vo

lum

e (m

m3 ) Carbo

Carbo + FST

PBS FSTCisPt + PBS

CisPt + FST

Urea Creatinine

A

B

PDX1.1

Activin A

GDF8/11

D

A549 NCI-H358 NCI-H727 PDX1.1

C

E

F NCI-H727

G H

LLC

LLCNCI-H358

I

0 10 200

100

200

300

400

500

600

Time (days)

Tu

mo

r vo

lum

e (m

m3 )

FSTControl

0 20 40 60 100 200 3000

20

40

60

80

100

Time (days)

Su

rviv

al (

%)

Control (10)Carbo (10)FST (10)Carbo + FST (10)

#

0 20 40 60 800

20

40

60

80

100

Time (days)

Su

rviv

al (%

)

Control (6)FST (6)Carbo (6)Carbo + FST (6) *

0 10 20 30 400

20

40

60

80

100

Time (days)

Su

rviv

al (%

)

Control (6)FST (7)Carbo (7)Carbo + FST (7)

0 50 100 150 2000

20

40

60

80

100

Time (days)

Su

rviv

al (%

)

Control (8)Carbo (8)FST (8)Carbo +FST (9) ***

0 5 10 15 200

20

40

60

80

100

Time (days)

Su

rviv

al (%

)Control (7)

Carbo (7)FST (7)

Carbo + FST (7) *

0255075

100125150

Pla

sma

con

c (m

M)

**

PBSFST

CisPtCisPt + FST

+–––

+––

–

+–

––

+––

+

0

50

100

150

200

Pla

sma

con

c (µ

M)

**

+–––

+––

–

+–

––

+––

+

Fig. 7. Inhibition of activin/GDF11 signaling in vivo. (A) Activin A and GDF8/11 expression detected by immunohistochemistry in A549, NCI-H358, NCI-H727, and PDX1.1 xenografts, and LLC allografts. Immunoperoxidase signal is shown in brown, counterstained with hematoxylin. Scale bars, 100 m. (B) Quantification of the data shown in (A). n = 4, mean ± SEM. (C) Schema describing an experiment to determine the effect of combination carboplatin and FST treatment on the growth of A549 flank xenograft tumors. (D) Growth of A549 flank xenografts in nude mice treated with vehicle control, FST, and/or Carbo. Data are presented as the growth of tumors in inde-pendent mice. (E) Kaplan-Meier analysis of survival to ethical endpoint of mice from the same experiment, n for each experimental group is shown. #P < 0.0001 compared to mice treated with vehicle + cisplatin (Pt), log-rank test. (F) Kaplan-Meier analysis of survival to ethical endpoint of nude mice bearing NCI-H358, NCI-H727, PDX1.1, and LLC flank tumors treated with phosphate-buffered saline (PBS; control), FST, and/or Carbo as shown in (C). n for each group is shown. ***P < 0.001, *P < 0.05 compared to mice treated with vehicle + Pt, log-rank test. (G) Schema describing an experiment to determine the effect of treatment with FST on acute kidney injury induced by CisPt. (H) Serum urea and creatinine concentrations in mice treated with PBS, FST, and/or CisPt. Blood samples were taken on day 5. n = 7, mean ± SEM. **P < 0.01, one-way ANOVA with Bonferroni correction. (I) Top: Macroscopic appearance of kidneys from mice from the experiment in (G) and (H). Scale bar, 2 mm. Bottom: Photomicrographs of PAS-stained sections of whole kidneys from the same experiment. Scale bar, 2 mm.

by guest on October 11, 2020

http://stm.sciencem

ag.org/D

ownloaded from

Marini et al., Sci. Transl. Med. 10, eaat3504 (2018) 25 July 2018

S C I E N C E T R A N S L A T I O N A L M E D I C I N E | R E S E A R C H A R T I C L E

11 of 13

therefore may hold promise as a treatment for cancer cachexia and sarcopenia (15). We conclude that FST could have potential to im-prove the efficacy of chemotherapy, while at the same time attenu-ating a major dose-limiting toxicity with the potential to reverse some of the debilitating systemic effects of cancer.

MATERIALS AND METHODSStudy designIn vitro experiments were performed as quadruplicate technical rep-licates, with the mean value taken as one data point. Each experiment was then repeated as four biological replicates to generate the final data and statistical analysis. In vivo tumor experiments were ran-domized and blinded. Each mouse cohort is reported as the combi-nation of two separate experiments with 3 to 5 mice in each group, resulting in a total of 6 to 10 animals per treatment arm.

ReagentsSources and details of all reagents and antibodies are listed in table S4. Recombinant FST 288 protein (PB01) was supplied by Paranta Biosciences Limited.

Carboplatin response in vitroInitial determination of cell line responses was performed in a 96-well plate format. Cells were seeded at 5000 cells per well in a 96-well plate on day 0, treated with carboplatin on day 2, and viability was deter-mined using AlamarBlue after 72 hours according to the manufacturer’s instructions.

Cell cultureCancer cell lines were grown in advanced RPMI with 1% newborn calf serum and penicillin (100 U/ml) and streptomycin (100 g/ml). HEK293 cells were grown in Dulbecco’s minimum essential medi-um with 10% fetal bovine serum and penicillin-streptomycin at the same concentration. Cell culture experiments were performed with-out antibiotics. Cell line ID and mycoplasma testing was performed every 6 months. All cell lines were purchased from the American Type Culture Collection except for LX22CL (7).

CRISPR/Cas9 genome editingTargeting of MAP3K7 in A549 cells was performed using the GeneScript lentiviral system, with the guide RNA sequence AGAGCCTGAT-GACTCGTTGT. Lentivirus was generated from HEK293T cells transfected with CRISPR/Cas9 plasmids and proprietary packag-ing plasmids using Lipofectamine3000 according to the manufac-turer’s instructions. Cells were infected with lentivirus with polybrene (8 g/ml) and selected with puromycin (1 g/ml). Single-cell clones were isolated and expanded in 96-well plates and then screened for TAK1 protein expression by Western blot. DNA was isolated from candidate clones and Sanger-sequenced using the primers 5′-TT-GAACGTCCGTGAAGTAAAGA-3′ (forward) and 5′-ACAGGCCC-AGAGACAAACAT-3′ (reverse) to confirm knockout of MAP3K7.

DNA damage detectionCells were plated onto 24-well plates for 24 hours and then treated with carboplatin for 24 hours. Cells were fixed in neutral buffered formalin for 20 min at room temperature and permeabilized in 0.1% Triton-X for 5 min, and then washed with PBS before treatment with Odyssey blocking buffer for 1 hour. Cells were then probed with anti-H2AX

for 2 hours at room temperature, then washed with PBS before addi-tion of anti-mouse IgG-568 and DAPI, and then washed and mounted. Images were taken using the Nikon C1 microscope.

Gene expression analysisRNA sequencing data were downloaded from the TCGA lung ade-nocarcinoma study through cBioPortal (31). Heat maps were gen-erated using R/Bioconductor.

ImmunohistochemistryThis study was performed as an anonymized, retrospective cohort analysis of patients with stage IV lung adenocarcinoma in accord ance with the Declaration of Helsinki, Australian law, and the National Health and Medical Research Council guidelines and with the approval of the Monash Health Human Research Ethics Committee. Biopsies were taken from patients at Monash Medical Centre from 2009 to 2015. Formalin-fixed, paraffin-embedded sections were stained as described (32) with the antibodies listed in table S4.

In vivo experimentsAll experiments were conducted in accordance with Australian government guidelines and with the approval of the Monash Uni-versity Animal Ethics Committee. PDX1.1 was generated in-house from an untreated patient with a KRAS mutant lung adenocarcinoma using previously described methods (7). Cells in 50 l of RPMI were added to 50 l of Matrigel and injected into the flanks of Balbc nude mice as follows: 1 × 106 cells per flank for A549, NCI-H727, or PDX1.1; and 1 × 107 per flank for NCI-H358. C57Bl6 mice were injected with 1 × 106 LLC cells per flank. Measurements of tumor volume were calculated using the following formula: V = (xs2 × xl)/2, where xs is the measurement of the shorter axis and xl is the mea-surement of the longer axis.

For treatment with SB-505124, mice with A549 xenografts measur-ing 200 mm3 were randomly assigned to receive vehicle (10% DMSO + 90% saline) only, SB-505124 (5 mg/kg) only, carboplatin (60 mg/kg) + vehicle, or carboplatin (60 mg/kg) + SB-505124 (5 mg/kg) i.p. as de-scribed in Fig. 6B. Mouse weight (fig. S8) and tumor volume were measured daily. Mice were sacrificed when tumors reached 500 mm3. For FST experiments with A549, NCI-H358, NCI-H727, and PDX1.1 models, mice with tumors at 200 mm3 were randomly assigned to ve-hicle (PBS) only, FST (2 g, Paranta Biosciences Limited), carboplatin (60 mg/kg) + vehicle, or FST + carboplatin, intraperitoneally (ip), as described in Fig. 7C. Mouse weight (fig. S8) and tumor volume were measured daily. Mice were sacrificed when tumors reached 500 mm3 or skin ulceration was observed with A549 xenografts, or at 800 mm3 with NCI-H358, NCI-H727, and PDX1.1 xenografts.

Nephrotoxicity experiments were conducted with male C57Bl6 mice randomly assigned to receive cisplatin (15 mg/kg), PBS, vehi-cle (10% DMSO + 90% saline), 2 g of FST, or SB-505124 (5 mg/kg) by intraperitoneal injection as shown in Figs. 6E and 7G. On day 6, mice were sacrificed using CO2 inhalation, and blood was collected. Kidneys were dissected and fixed in neutral buffered formalin for 24 hours, bisected, and then fixed for an additional 24 hours. Serum analysis of urea and creatinine concentrations was performed using a clinical Beckman Coulter analyzer.

Orthogonal siRNA validationThis was performed in a 96-well plate format using Dharmacon On-Target siRNAs. Two days after siRNA treatment, increasing

by guest on October 11, 2020

http://stm.sciencem

ag.org/D

ownloaded from

Marini et al., Sci. Transl. Med. 10, eaat3504 (2018) 25 July 2018

S C I E N C E T R A N S L A T I O N A L M E D I C I N E | R E S E A R C H A R T I C L E

12 of 13

concentrations of carboplatin were added to wells in quadrupli-cate. Cells (5000 per well) were seeded in a 96-well plate and left to attach for 6 to 8 hours. Cells were then transfected with siRNA using Lipofectamine RNAiMax (Thermo Fisher Scientific) in OptiMEM overnight. Medium was then replaced, and after 48 hours, cells were treated with vehicle (PBS) or carboplatin. After 72 hours, viability was determined using AlamarBlue fluorescence assay.

Pharmacological validation of hits generated by siRNA screeningCells (5000 per well) were plated into a 96-well plate and left to attach for 6 to 8 hours. Medium was then replaced with fresh medium containing LLZ-1640-2 (10 M), SB-505125 (5 M), or ve-hicle. After 48 hours, the medium was replaced with medium con-taining increasing concentrations of carboplatin. Plates were then left to incubate for 3 days, and the cell viability was assayed using AlamarBlue.

Primary and secondary siRNA screensA549 cells were used in a parallel 384-well plate format, with or with-out the addition of carboplatin (25 g/ml) using the Dharmacon siGenome Smartpool library. Library plates were obtained lyoph-ilized, and daughter plates were generated, where four 96-well plates were combined to make 384-well plates in batches of four. Cells (1000 per well) were reverse-transfected using Lipofectamine RNAiMax and OptiMEM overnight. Medium was then replaced. Af-ter a further 24 hours, cells were treated with vehicle (PBS) or car-boplatin. After 5 days, viability was determined using AlamarBlue. Using the death and sensitization controls, we generated quality assessment data using the Z′ factor, where Z′ = 1 − [3 x (high control + low control)]/[high control − ow control], = SD, and = mean. Current screening guidelines suggest that Z′ ≥ 0.5 is excellent (±12 SD from mean), Z′ ≥ 0 is acceptable (±6 SD from mean), and Z′ < 0 is un-acceptable and the plate cannot be analyzed (8). The sensitization index is calculated after normalization and averaging and is a func-tion of the viability of cells treated with siRNA and vehicle over the viability of cells treated with siRNA and platinum (8).

Western blot analysisWestern blotting was performed on cell lysates as previously de-s cribed (33). Details of Western blot antibodies are listed in table S4. Examples of full-length blots for each antibody and the image crop-ping used are shown in fig. S9.

Statistical analysisFor the whole-genome siRNA screen, we used a four-part statistical pipeline for data analysis: (i) data triage using visual plate inspec-tion, heat mapping, and linear correlation plotting; (ii) normaliza-tion using the percentage of negative control method, where each value on the plate is divided by the mean of the negative controls on the plate; (iii) quality metrics generated by incorporating positive, negative, death, and sensitization controls in each plate; and (iv) hit identification using a two-step method (8). Genes were ranked by the sensitization index. These data were then analyzed using multi-ple t tests and raw P values subjected to FDR adjustment (34). Each experiment used four plates, where two of these plates received car-boplatin treatment and the other two plates received the vehicle control treatment. Statistical methods for the remainder of the pa-per are described in the legends.

SUPPLEMENTARY MATERIALSwww.sciencetranslationalmedicine.org/cgi/content/full/10/451/eaat3504/DC1Fig. S1. Controls used in the siRNA screen.Fig. S2. Controls for SMAD2/3 expression and carboplatin responses in HEK293 cells in vitro.Fig. S3. Positive controls for SMAD2/3 expression in vitro.Fig. S4. Responses of lung adenocarcinoma cell lines to chemotherapy agents with TGF signaling pathway activation in vitro.Fig. S5. Evaluation of activin A and GDF8/11 expression in lung adenocarcinoma.Fig. S6. Effect of SB-505124 on the platinum response in vivo.Fig. S7. Effect of FST on the platinum response in vivo.Fig. S8. Mouse weights in xenograft/allograft experiments.Fig. S9. Examples of full-length Western blots with image cropping.Table S1. Complete data set from the primary siRNA screen (provided as an Excel file).Table S2. High-confidence hits targetable with existing small or large molecules.Table S3. Characteristics of the patient cohort used to generate immunohistochemical data.Table S4. Details of reagents used (provided as an Excel file).

REFERENCES AND NOTES 1. G. Pilkington, A. Boland, T. Brown, J. Oyee, A. Bagust, R. Dickson, A systematic review of

the clinical effectiveness of first-line chemotherapy for adult patients with locally advanced or metastatic non-small cell lung cancer. Thorax 70, 359–367 (2015).

2. D. J. Stewart, Mechanisms of resistance to cisplatin and carboplatin. Crit. Rev. Oncol. Hematol. 63, 12–31 (2007).

3. D. M. de Kretser, R. E. O’Hehir, C. L. Hardy, M. P. Hedger, The roles of activin A and its binding protein, follistatin, in inflammation and tissue repair. Mol. Cell. Endocrinol. 359, 101–106 (2012).

4. S. R. Mihaly, J. Ninomiya-Tsuji, S. Morioka, TAK1 control of cell death. Cell Death Differ. 21, 1667–1676 (2014).

5. M. D. Hall, K. A. Telma, K.-E. Chang, T. D. Lee, J. P. Madigan, J. R. Lloyd, I. S. Goldlust, J. D. Hoeschele, M. M. Gottesman, Say no to DMSO: Dimethylsulfoxide inactivates cisplatin, carboplatin, and other platinum complexes. Cancer Res. 74, 3913–3922 (2014).

6. S. Oguri, T. Sakakibara, H. Mase, T. Shimizu, K. Ishikawa, K. Kimura, R. D. Smyth, Clinical pharmacokinetics of carboplatin. J. Clin. Pharmacol. 28, 208–215 (1988).

7. V. C. Daniel, L. Marchionni, J. S. Hierman, J. T. Rhodes, W. L. Devereux, C. M. Rudin, R. Yung, G. Parmigiani, M. Dorsch, C. D. Peacock, D. N. Watkins, A primary xenograft model of small-cell lung cancer reveals irreversible changes in gene expression imposed by culture in vitro. Cancer Res. 69, 3364–3373 (2009).

8. A. W. Whitehurst, B. O. Bodemann, J. Cardenas, D. Ferguson, L. Girard, M. Peyton, J. D. Minna, C. Michnoff, W. Hao, M. G. Roth, X.-J. Xie, M. A. White, Synthetic lethal screen identification of chemosensitizer loci in cancer cells. Nature 446, 815–819 (2007).

9. J. Kim, E. McMillan, H. S. Kim, N. Venkateswaran, G. Makkar, J. Rodriguez-Canales, P. Villalobos, J. E. Neggers, S. Mendiratta, S. Wei, Y. Landesman, W. Senapedis, E. Baloglu, C.-W. B. Chow, R. E. Frink, B. Gao, M. Roth, J. D. Minna, D. Daelemans, I. I. Wistuba, B. A. Posner, P. P. Scaglioni, M. A. White, XPO1-dependent nuclear export is a druggable vulnerability in KRAS-mutant lung cancer. Nature 538, 114–117 (2016).

10. G. Bindea, B. Mlecnik, H. Hackl, P. Charoentong, M. Tosolini, A. Kirilovsky, W.-H. Fridman, F. Pagès, Z. Trajanoski, J. Galon, ClueGO: A Cytoscape plug-in to decipher functionally grouped gene ontology and pathway annotation networks. Bioinformatics 25, 1091–1093 (2009).

11. L. Kelland, The resurgence of platinum-based cancer chemotherapy. Nat. Rev. Cancer 7, 573–584 (2007).

12. L. M. Wakefield, C. S. Hill, Beyond TGF: Roles of other TGF superfamily members in cancer. Nat. Rev. Cancer 13, 328–341 (2013).

13. M. Dussiot, T. T. Maciel, A. Fricot, C. Chartier, O. Negre, J. Veiga, D. Grapton, E. Paubelle, E. Payen, Y. Beuzard, P. Leboulch, J.-A. Ribeil, J.-B. Arlet, F. Coté, G. Courtois, Y. Z. Ginzburg, T. O. Daniel, R. Chopra, V. Sung, O. Hermine, I. C. Moura, An activin receptor IIA ligand trap corrects ineffective erythropoiesis in -thalassemia. Nat. Med. 20, 398–407 (2014).

14. M. A. Egerman, S. M. Cadena, J. A. Gilbert, A. Meyer, H. N. Nelson, S. E. Swalley, C. Mallozzi, C. Jacobi, L. L. Jennings, I. Clay, G. Laurent, S. Ma, S. Brachat, E. Lach-Trifilieff, T. Shavlakadze, A.-U. Trendelenburg, A. S. Brack, D. J. Glass, GDF11 increases with age and inhibits skeletal muscle regeneration. Cell Metab. 22, 164–174 (2015).

15. A. Loumaye, M. de Barsy, M. Nachit, P. Lause, L. Frateur, A. van Maanen, P. Trefois, D. Gruson, J.-P. Thissen, Role of Activin A and myostatin in human cancer cachexia. J. Clin. Endocrinol. Metab. 100, 2030–2038 (2015).

16. N. Oshimori, D. Oristian, E. Fuchs, TGF- promotes heterogeneity and drug resistance in squamous cell carcinoma. Cell 160, 963–976 (2015).

17. J. Ninomiya-Tsuji, T. Kajino, K. Ono, T. Ohtomo, M. Matsumoto, M. Shiina, M. Mihara, M. Tsuchiya, K. Matsumoto, A resorcylic acid lactone, 5Z-7-oxozeaenol, prevents inflammation by inhibiting the catalytic activity of TAK1 MAPK kinase kinase. J. Biol. Chem. 278, 18485–18490 (2003).

by guest on October 11, 2020

http://stm.sciencem

ag.org/D

ownloaded from

Marini et al., Sci. Transl. Med. 10, eaat3504 (2018) 25 July 2018

S C I E N C E T R A N S L A T I O N A L M E D I C I N E | R E S E A R C H A R T I C L E

13 of 13

18. A. M. Domina, J. A. Vrana, M. A. Gregory, S. R. Hann, R. W. Craig, MCL1 is phosphorylated in the PEST region and stabilized upon ERK activation in viable cells, and at additional sites with cytotoxic okadaic acid or taxol. Oncogene 23, 5301–5315 (2004).

19. S. DaCosta Byfield, C. Major, N. J. Laping, A. B. Roberts, SB-505124 is a selective inhibitor of transforming growth factor- type I receptors ALK4, ALK5, and ALK7. Mol. Pharmacol. 65, 744–752 (2004).

20. Cancer Genome Atlas Research Network, E. A. Collission, J. D. Campbell, A. N. Brooks, A. H. Berger, W. Lee, J. Chmielecki, D. G. Beer, L. Cope, L. Ding, P. S. Hammerman, J. G. Herman, J. V. Heymach, I. Jurisica, D. Kwiatkowski, M. Ladanyi, G. Robertson, R. Shen, R. Sinha, W. D. Travis, J. N. Weinstein, D. A. Wigle, M. D. Wilkerson, A. D. Cherniack, A. Hadjipanayis, D. J. Weisenberger, L. Averett Byers, S. B. Baylin, R. Govindan, M. Rosenberg, K. Cibulskis, J. Kim, C. Stewart, C. Sougnez, L. Lichtenstein, E. S. Lander, M. S. Lawrence, C. Kandoth, R. Fulton, L. L. Fulton, M. D. McLellan, R. K. Wilson, K. Ye, C. C. Fronick, C. A. Maher, C. A. Miller, M. C. Wendl, C. Cabanski, E. Mardis, D. Wheeler, M. Balasundaram, Y. S. N. Butterfield, R. Carlsen, E. Chuah, N. Dhalla, C. J. Creighton, R. Guin, C. Hirst, D. Lee, H. I. Li, M. Mayo, A. Chu, R. A. Moore, A. J. Mungall, J. E. Schein, P. Sipahimalani, A. Tam, R. Varhol, A. Gordon Robertson, N. Wye, N. Thiessen, R. A. Holt, S. J. M. Jones, M. A. Marra, J. D. Campbell, J. Chmielecki, M. Imielinski, R. C. Onofrio, E. Hodis, T. Zack, E. Helman, C. Sekhar Pedamallu, J. Mesirov, S. E. Schumacher, S. L. Carter, L. Garraway, R. Beroukhim, S. Lee, H. S. Mahadeshwar, A. Pantazi, A. Protopopov, X. Ren, S. Seth, X. Song, J. Tang, L. Yang, B. Hernandez, P.-C. Chen, M. Parfenov, S. B. Gabriel, G. Getz, A. Wei Xu, N. Santoso, M. Meyerson, P. J. Park, J. Todd Auman, S. Meng, Y. Shi, E. Buda, S. Waring, U. Veluvolu, D. Tan, P. A. Mieczkowski, C. D. Jones, J. V. Simons, M. G. Soloway, T. Bodenheimer, S. R. Jefferys, J. Roach, R. Kucherlapati, A. P. Hoyle, J. Wu, S. Balu, D. Singh, J. F. Prins, J. S. Marron, J. S. Parker, C. M. Perou, J. Liu, D. T. Maglinte, P. H. Lai, M. S. Bootwalla, D. J. Van Den Berg, T. Triche Jr., J. Cho, D. DiCara, D. Heiman, P. Lin, W. Mallard, D. Voet, H. Zhang, L. Zou, M. S. Noble, N. Gehlenborg, H. Thorvaldsdottir, M.-D. Nazaire, J. Robinson, L. Danilova, B. Arman Aksoy, G. Ciriello, B. S. Taylor, G. Dresdner, J. Gao, B. Gross, V. E. Seshan, B. Reva, P. W. Laird, S. Onur Sumer, N. Weinhold, C. Sander, L. Chin, S. Ng, J. Zhang, J. Zhu, C. C. Benz, C. Yau, D. Haussler, P. T. Spellman, P. K. Kimes, B. M. Broom, J. Wang, Y. Lu, P. Kwok Shing Ng, L. Diao, G. Saksena, L. Averett Byers, W. Liu, J. V. Heymach, C. I. Amos, R. Akbani, G. B. Mills, E. Curley, J. Paulauskis, K. Lau, S. Morris, T. Shelton, D. Mallery, J. Gardner, R. Penny, C. Saller, K. Tarvin, W. G. Richards, R. Cerfolio, A. Bryant, D. P. Raymond, N. Schultz, N. A. Pennell, C. Farver, C. Czerwinski, L. Huelsenbeck-Dill, S. Ma, M. Iacocca, N. Petrelli, A. Radenbaugh, B. Rabeno, J. M. Stuart, J. Brown, T. Bauer, O. Dolzhanskiy, O. Potapova, D. Rotin, O. Voronina, E. Nemirovich-Danchenko, K. A. Hoadley, K. V. Fedosenko, A. Gal, D. Neil Hayes, M. Behera, S. S. Ramalingam, G. Sica, D. Flieder, J. Boyd, J. Weaver, B. Kohl, D. Huy Quoc Thinh, G. Sandusky, H. Juhl, E. Duhig, E. Gabrielson, J. Shin, B. Lee, K. Rogers, D. Trusty, M. V. Brock, C. Williamson, E. Burks, K. Rieger-Christ, A. Holway, T. Sullivan, M. K. Asiedu, F. Kosari, N. Rekhtman, M. Zakowski, V. W. Rusch, P. Zippile, H. Pass, C. Goparaju, The Cancer Genome Atlas Research Network, J. M. S. Bartlett, S. Kodeeswaran, J. Parfitt, H. Sekhon, M. Albert, J. Eckman, J. B. Myers, R. Cheney, C. Morrison, C. Gaudioso, J. A. Borgia, P. Bonomi, M. Pool, M. J. Liptay, F. Moiseenko, I. Zaytseva, H. Dienemann, M. Meister, P. A. Schnabel, T. R. Muley, M. Peifer, C. Gomez-Fernandez, L. Herbert, S. Egea, M. Huang, L. B. Thorne, L. Boice, A. Hill Salazar, W. K. Funkhouser, W. Kimryn Rathmell, R. Dhir, S. A. Yousem, S. Dacic, F. Schneider, J. M. Siegfried, R. Hajek, M. A. Watson, S. McDonald, B. Meyers, B. Clarke, I. A. Yang, K. M. Fong, L. Hunter, M. Windsor, R. V. Bowman, S. Peters, I. Letovanec, K. Z. Khan, M. A. Jensen, E. E. Snyder, D. Srinivasan, A. B. Kahn, J. Baboud, D. A. Pot, K. R. Mills Shaw, M. Sheth, T. Davidsen, J. A. Demchok, L. Yang, Z. Wang, R. Tarnuzzer, J. Claude Zenklusen, B. A. Ozenberger, H. J. Sofia, B. Clarke, E. Duhig, C. Farver, G. Sica, P. Illei, J. Suh, M.-S. Tsao, Comprehensive molecular profiling of lung adenocarcinoma. Nature 511, 543–550 (2014).

21. B. Gyorffy, P. Surowiak, J. Budczies, A. Lánczky, Online survival analysis software to assess the prognostic value of biomarkers using transcriptomic data in non-small-cell lung cancer. PLOS ONE 8, e82241 (2013).

22. C. W. Seder, W. Hartojo, L. Lin, A. L. Silvers, Z. Wang, D. G. Thomas, T. J. Giordano, G. Chen, A. C. Chang, M. B. Orringer, D. G. Beer, Upregulated INHBA expression may promote cell proliferation and is associated with poor survival in lung adenocarcinoma. Neoplasia 11, 388–396 (2009).

23. M. J. Pajares, J. Agorreta, E. Salvo, C. Behrens, I. I. Wistuba, L. M. Montuenga, R. Pio, A. Rouzaut, TGFBI expression is an independent predictor of survival in adjuvant-treated lung squamous cell carcinoma patients. Br. J. Cancer 110, 1545–1551 (2014).

24. N. Pabla, Z. Dong, Cisplatin nephrotoxicity: Mechanisms and renoprotective strategies. Kidney Int. 73, 994–1007 (2008).

25. F. Y. Ma, G. H. Tesch, E. Ozols, M. Xie, M. D. Schneider, D. J. Nikolic-Paterson, TGF-1-activated kinase-1 regulates inflammation and fibrosis in the obstructed kidney. Am. J. Physiol. Renal Physiol. 300, F1410–F1421 (2011).

26. H. S. Bayomi, N. M. Elsherbiny, A. M. El-Gayar, M. M. H. Al-Gayyar, Evaluation of renal protective effects of inhibiting TGF- type I receptor in a cisplatin-induced nephrotoxicity model. Eur. Cytokine Netw. 24, 139–147 (2013).

27. S.-Y. Lin, J. R. Morrison, D. J. Phillips, D. M. de Kretser, Regulation of ovarian function by the TGF-beta superfamily and follistatin. Reproduction 126, 133–148 (2003).

28. M. DuPage, A. L. Dooley, T. Jacks, Conditional mouse lung cancer models using adenoviral or lentiviral delivery of Cre recombinase. Nat. Protoc. 4, 1064–1072 (2009).

29. D. Morgensztern, M. J. Campo, S. E. Dahlberg, R. C. Doebele, E. Garon, D. E. Gerber, S. B. Goldberg, P. S. Hammerman, R. S. Heist, T. Hensing, L. Horn, S. S. Ramalingam, C. M. Rudin, R. Salgia, L. V. Sequist, A. T. Shaw, G. R. Simon, N. Somaiah, D. R. Spigel, J. Wrangle, D. Johnson, R. S. Herbst, P. Bunn, R. Govindan, Molecularly targeted therapies in non-small-cell lung cancer annual update 2014. J. Thorac. Oncol. 10, S1–S63 (2015).

30. D. Brunen, S. Willems, U. Kellner, R. Midgley, I. Simon, R. Bernards, TGF-: An emerging player in drug resistance. Cell Cycle 12, 2960–2968 (2013).

31. J. Gao, B. A. Aksoy, U. Dogrusoz, G. Dresdner, B. Gross, S. O. Sumer, Y. Sun, A. Jacobsen, R. Sinha, E. Larsson, E. Cerami, C. Sander, N. Schultz, Integrative analysis of complex cancer genomics and clinical profiles using the cBioPortal. Sci. Signal. 6, pl1 (2013).

32. M. Alamgeer, V. Ganju, A. Szczepny, P. A. Russell, Z. Prodanovic, B. Kumar, Z. Wainer, T. Brown, M. Schneider-Kolsky, M. Conron, G. Wright, D. N. Watkins, The prognostic significance of aldehyde dehydrogenase 1A1 (ALDH1A1) and CD133 expression in early stage non-small cell lung cancer. Thorax 68, 1095–1104 (2013).

33. D. R. Croucher, F. Hochgräfe, L. Zhang, L. Liu, R. J. Lyons, D. Rickwood, C. M. Tactacan, B. C. Browne, N. Ali, H. Chan, R. Shearer, D. Gallego-Ortega, D. N. Saunders, A. Swarbrick, R. J. Daly, Involvement of lyn and the atypical kinase SgK269/PEAK1 in a basal breast cancer signaling pathway. Cancer Res. 73, 1969–1980 (2013).

34. Y. Benjamini, D. Drai, G. Elmer, N. Kafkafi, I. Golani, Controlling the false discovery rate in behavior genetics research. Behav. Brain Res. 125, 279–284 (2001).

Funding: This work was supported by the Victorian Cancer Agency (TS10-01), the Cancer Council NSW (RG15-20), the St. Vincent’s Clinic Foundation, the Hudson Institute of Medical Research, Paranta Biosciences Limited, the National Health and Medical Research Council of Australia (GNT100120 and GNT546107), the Petre Foundation, and the Victorian Government’ Operational Infrastructure Support Program. The contents of this manuscript are solely the responsibility of the participating institutions and individual authors and do not reflect the views of these funding agencies. Author contributions: The study was designed by K.D.M., D.R.C., L.G.M., F.J.R., J.E.C., and D.N.W. Experiments were performed by K.D.M., D.R.C., R.A.M., V.C., A.G.-R., J.F.H., W.S.N.J., M.A., V.B., T.W., V.V., J.Z.R.H., H.C.L., B.K., J.W., V.G., F.J.R., and J.E.C. Analysis and interpretation of data were performed by K.D.M., D.R.C., R.A.M., A.G.-R., M.A., H.C.L., S.R.O., B.K., J.W., V.G., A.B., L.G.M., F.J.R., J.E.C., and D.N.W. All authors contributed to the writing and reviewing of the manuscript. Competing interests: K.D.M., D.M.d.K., and D.N.W. are coinventors on a patent application relating to components of this work (U.S. 2018-0125936-A1). B.K., N.L., and R.B. are employees of Paranta Biosciences Limited. C.A.H. and M.P.H. are shareholders and members of the scientific advisory board of Paranta Biosciences Limited. D.M.d.K. and R.B. are directors of Paranta Biosciences Limited. Components of this work were financially supported by Paranta Biosciences Limited. Data and materials availability: All data associated with this study are present in the paper or the Supplementary Materials.

Submitted 20 February 2018Accepted 30 May 2018Published 25 July 201810.1126/scitranslmed.aat3504

Citation: K. D. Marini, D. R. Croucher, R. A. McCloy, V. Vaghjiani, A. Gonzalez-Rajal, J. F. Hastings, V. Chin, A. Szczepny, K. Kostyrko, C. Marquez, W. S. N. Jayasekara, M. Alamgeer, V. Boolell, J. Z. R. Han, T. Waugh, H. C. Lee, S. R. Oakes, B. Kumar, C. A. Harrison, M. P. Hedger, N. Lorensuhewa, B. Kita, R. Barrow, B. W. Robinson, D. M. de Kretser, J. Wu, V. Ganju, E. A. Sweet-Cordero, A. Burgess, L. G. Martelotto, F. J. Rossello, J. E. Cain, D. N. Watkins, Inhibition of activin signaling in lung adenocarcinoma increases the therapeutic index of platinum chemotherapy. Sci. Transl. Med. 10, eaat3504 (2018).

by guest on October 11, 2020

http://stm.sciencem

ag.org/D

ownloaded from

platinum chemotherapyInhibition of activin signaling in lung adenocarcinoma increases the therapeutic index of

Cain and D. Neil WatkinsVinod Ganju, E. Alejandro Sweet-Cordero, Andrew Burgess, Luciano G. Martelotto, Fernando J. Rossello, Jason E.Mark P. Hedger, Nirmal Lorensuhewa, Badia Kita, Ross Barrow, Bruce W. Robinson, David M. de Kretser, Jianmin Wu, Vishal Boolell, Jeremy Z. R. Han, Todd Waugh, Hong Ching Lee, Samantha R. Oakes, Beena Kumar, Craig A. Harrison,Venessa Chin, Anette Szczepny, Kaja Kostyrko, Cesar Marquez, W. Samantha N. Jayasekara, Muhammad Alamgeer, Kieren D. Marini, David R. Croucher, Rachael A. McCloy, Vijesh Vaghjiani, Alvaro Gonzalez-Rajal, Jordan F. Hastings,

DOI: 10.1126/scitranslmed.aat3504, eaat3504.10Sci Transl Med

treatment while also allowing the use of higher doses or longer periods of drug exposure.suggest that activin inhibitors could be a valuable addition to platinum chemotherapy, enhancing the efficacy ofplatinum drugs in mouse models of cancer and also protects the animals from kidney damage. These findings using a small molecule or a protein called follistatin can offer a dual benefit in that it potentiates the effects ofidentified the activin pathway, which can be targeted. The authors demonstrated that inhibition of this pathway

. performed a whole-genome RNA interference screen andet alcontribute to treatment resistance, Marini maycommon problem, as are dose-limiting side effects, particularly kidney toxicity. To search for mechanisms that

Platinum-based chemotherapy is a mainstay of treatment for lung cancer, but resistance to this therapy is aBlocking activin actively treats cancer

ARTICLE TOOLS http://stm.sciencemag.org/content/10/451/eaat3504

MATERIALSSUPPLEMENTARY http://stm.sciencemag.org/content/suppl/2018/07/23/10.451.eaat3504.DC1

CONTENTRELATED

http://stm.sciencemag.org/content/scitransmed/11/482/eaau8680.fullhttp://stke.sciencemag.org/content/sigtrans/12/568/eaau9216.fullhttp://stm.sciencemag.org/content/scitransmed/5/171/171ra18.fullhttp://stm.sciencemag.org/content/scitransmed/7/279/279ra36.fullhttp://stm.sciencemag.org/content/scitransmed/8/331/331ra39.fullhttp://stm.sciencemag.org/content/scitransmed/9/415/eaao4307.full

REFERENCES

http://stm.sciencemag.org/content/10/451/eaat3504#BIBLThis article cites 34 articles, 9 of which you can access for free

PERMISSIONS http://www.sciencemag.org/help/reprints-and-permissions

Terms of ServiceUse of this article is subject to the

registered trademark of AAAS. is aScience Translational MedicineScience, 1200 New York Avenue NW, Washington, DC 20005. The title

(ISSN 1946-6242) is published by the American Association for the Advancement ofScience Translational Medicine

of Science. No claim to original U.S. Government WorksCopyright © 2018 The Authors, some rights reserved; exclusive licensee American Association for the Advancement

by guest on October 11, 2020

http://stm.sciencem

ag.org/D

ownloaded from