GDNF family receptor alpha-like antagonist antibody alleviates chemotherapy-induced cachexia in melanoma-bearing mice

Abstract

Background

Patients with cancer undergoing chemotherapy experience cachexia with anorexia, body weight loss, and the depletion of skeletal muscles and adipose tissues. Effective treatment strategies for chemotherapy-induced cachexia are scarce. The growth differentiation factor 15 (GDF15)/GDNF family receptor alpha-like (GFRAL)/rearranged during transfection (RET) axis is a critical signalling pathway in chemotherapy-induced cachexia. In this study, we developed a fully human GFRAL antagonist antibody and investigated whether it inhibits the GDF15/GFRAL/RET axis, thereby alleviating chemotherapy-induced cachexia in tumour-bearing mice.

Methods

Anti-GFRAL antibodies were selected via biopanning, using a human combinatorial antibody phage library. The potent GFRAL antagonist antibody A11 was selected via a reporter cell assay and its inhibitory activity of GDF15-induced signalling was evaluated using western blotting. To investigate the in vivo function of A11, a tumour-bearing mouse model was established by inoculating 8-week-old male C57BL/6 mice with B16F10 cells (n = 10–16 mice per group). A11 was administered subcutaneously (10 mg/kg) 1 day before intraperitoneal treatment with cisplatin (10 mg/kg). Animals were assessed for changes in food intake, body weight, and tumour volume. Plasma and key metabolic tissues such as skeletal muscles and adipose tissues were collected for protein and mRNA expression analysis.

Results

A11 reduced serum response element-luciferase reporter activity up to 74% (P < 0.005) in a dose-dependent manner and blocked RET phosphorylation up to 87% (P = 0.0593), AKT phosphorylation up to 28% (P = 0.0593) and extracellular signal regulatory kinase phosphorylation up to 75% (P = 0.0636). A11 inhibited the action of cisplatin-induced GDF15 on the brainstem and decreased GFRAL-positive neuron population expressing c-Fos in the area postrema and nucleus of the solitary tract by 62% in vivo (P < 0.05). In a melanoma mouse model treated with cisplatin, A11 recovered anorexia by 21% (P < 0.05) and tumour-free body weight loss by 13% (P < 0.05). A11 significantly improved the cisplatin-induced loss of skeletal muscles (quadriceps: 21%, gastrocnemius: 9%, soleus: 13%, P < 0.05) and adipose tissues (epididymal white adipose tissue: 37%, inguinal white adipose tissue: 51%, P < 0.05).

Conclusions

Our study suggests that GFRAL antagonist antibody may alleviate chemotherapy-induced cachexia, providing a novel therapeutic approach for patients with cancer experiencing chemotherapy-induced cachexia.

Introduction

Cancer cachexia is a complex syndrome that causes involuntary weight loss in cancer patients and is characterized by increased energy expenditure and disrupted protein homeostasis.^1,2 Cancer cachexia patients suffer from poor quality of life due to anorexia and wasting of skeletal muscles and adipose tissues. Cancer cachexia affects 50–80% of patients with advanced cancer³ and is responsible for 20–40% of cancer mortality.^1,2 Many factors contribute to cancer cachexia including metabolic and psychological changes, inflammation, and radiation therapy.⁴ Ironically, chemotherapy, the first line of treatment for cancer, is also a potent trigger of cachexia.^4–6 For example, cisplatin caused a reduction in cumulative food intake from 45% in healthy rats to 77% in tumour-bearing rats,⁵ and cisplatin-treated mice suffered from loss of body weight by 29.8%, lean body mass by 20.6%, muscle strength by 22.5%, and anorexia.⁶ Furthermore, cancer patients undergoing chemotherapy usually experience chemotherapy-induced nausea and vomiting (CINV), regardless of the emetic risk of chemotherapeutic agents.^7,8 Unfortunately, the underlying mechanism of cancer cachexia is unclear, making it difficult to develop a cancer cachexia treatment. However, in the case of chemotherapy-induced cachexia, the fundamental mechanism inducing cachectic symptoms has been clearly identified to be growth differentiation factor 15 (GDF15)/GDNF family receptor alpha-like (GFRAL)/rearranged during transfection (RET) axis dependent.^9–11 Therefore, prevention of the GDF15/GFRAL/RET axis has been suggested as a promising approach to treat chemotherapy-induced cachexia.^12–14

GDF15 is a stress-related cytokine that is released from various tissues, such as the placenta, liver, kidneys, and prostate.¹⁵ Under physiological conditions, the blood levels of GDF15 are low.¹⁵ However, in stress conditions, the blood level increases by up to 100-fold.^16,17 Elevated GDF15 levels are associated with diseases such as heart failure, liver fibrosis, and particularly cancer cachexia.^12,16 Interestingly, recent studies proved that GDF15 is a major causative factor in chemotherapy-induced cachexia.^9,10 Treatment with chemotherapeutic agents such as cisplatin and dacarbazine increased blood levels of GDF15 in mice, which was positively correlated with weight loss.⁹ Cisplatin-induced GDF15 triggered emesis in both rats and musk shrews, but this symptom was not observed in GDF15 knockout mice.¹⁰ Therefore, inhibiting the action of GDF15 could effectively alleviate chemotherapy-induced cachexia.

The cachectic effect of GDF15 is mediated by the formation of a tertiary complex with GFRAL and its co-receptor RET.^11,18–21 GFRAL is specifically expressed in the area postrema (AP) and the nucleus of the solitary tract (NTS) in the brainstem.^18–21 As GDF15 and RET can regulate GFRAL-independent bioactivity in various peripheral tissues, they are not appropriate targets to specifically block GDF15's cachectic activity.^16,22–25 Therefore, a strategy targeting GFRAL is required for the treatment of cancer cachexia by chemotherapy. A humanized GFRAL antagonist antibody can interfere with tumour-derived GDF15 to induce weight gain in a mouse model of cancer cachexia, without affecting food intake.¹⁴ In addition, there are no studies on whether GFRAL inhibition through antibodies can improve chemotherapy-induced cancer cachexia.

Here, we developed a fully human GFRAL antagonist antibody A11 and demonstrated that blocking the GDF15/GFRAL/RET axis attenuates anorexia and weight loss in a melanoma mouse model treated with cisplatin. Our study provides a novel approach to the treatment of chemotherapy-induced cachexia. We expect that GFRAL antagonist antibody may help prevent CINV and discontinuation of chemotherapy in cancer patients.

Methods Biopanning

Solution-phase panning was performed using a recombinant human GFRAL extracellular domain (9647-GR; R&D Systems, Minneapolis, MN, USA), a recombinant mouse GFRAL extracellular domain (9844-GR; R&D Systems), and Dynabeads M-270 Epoxy beads (14301; Invitrogen; Thermo Fisher Scientific, Waltham, MA, USA). The beads were coupled with either the human or mouse GFRAL extracellular domain by incubation at 37°C overnight and alternatively used for each round of panning. GFRAL-coupled beads were incubated with human combinatorial antibody phage library²⁶ for 2 h at 25°C. The beads were washed with PBST (0.05% Tween 20 in phosphate-buffered saline [PBS]), and the remaining phages were eluted in 0.2 M glycine-HCl with 0.1% bovine serum albumin (BSA) (pH 2.2). The eluted phages were amplified by infecting Escherichia coli XL1-Blue competent cells with 100 mg/mL of carbenicillin. After 2 h at 37°C, the culture was diluted in super broth (SB) medium, and M13KO7 helper phages (N0315S; New England Biolabs, Ipswich, MA, USA) were added and cultured for an additional 2 h at 37°C. Next, 70 μg/mL kanamycin sulfate was added, and the cells were cultured at 37°C overnight. The following day, the phage particles were concentrated using polyethylene glycol/NaCl precipitation, purified, and resuspended in Tris-buffered saline containing 1% BSA.

Enzyme-linked immunosorbent assay

A 96-well half-area plate (3690; Corning, Corning, NY, USA) was coated with recombinant human GFRAL extracellular domain, recombinant mouse GFRAL extracellular domain, BSA, or recombinant human GDNF family receptor alpha-1 (GFRα1) (10330-H08H; Sino Biological, Beijing, China) and incubated overnight at 4°C with carbonate coating buffer. The plate was washed once with PBS and blocked for 1 h at 37°C with PBS containing 5% skim milk. Without washing, the anti-GFRAL antibodies were added and left to react for 2 h at 37°C. Each well was washed five times with PBS, and horseradish peroxidase (HRP)-conjugated goat anti-human IgG FC (1:1000, ab97225; Abcam, Cambridge, UK) was added and incubated for 1 h at 37°C.

For phage enzyme-linked immunosorbent assay (ELISA), phage supernatants were added and left to react for 2 h at 37°C after blocking the plate. Each well was washed five times with PBS, then HRP-conjugated anti-M13 bacteriophage antibody (1:2000, 11973-MM05T-H; Sino Biological) was added and incubated for 1 h at 37°C.

For ELISA to analyse the effect of A11 on the interaction between GDF15 and GFRAL, a 96-well half-area plate was coated with goat anti-human IgG Fc antibody (G-102-C; R&D Systems) and incubated overnight at 4 °C, then washed once with PBS and blocked with 5% milk in PBS for 1 h at 37°C. Next, recombinant human GFRAL-Fc chimera (9697-GR; R&D Systems) was added to the plates for 1 h at 37°C and then washed five times with PBS. Following that, 5-fold serial dilutions of A11 starting at 20 μg/mL or unbiotinylated recombinant human GDF15 (8146-GD; R&D Systems) controls starting at 180 μM were added to plates for 1 h at 37°C. Biotinylated recombinant human GDF15 (using a biotinylation kit [21935; Thermo Fisher Scientific]) was added to the plates for 1 h at 37°C. After washing each well five times with PBS, streptavidin–HRP (N100; Thermo Fisher Scientific) was added to detect GFRAL-bound biotinylated GDF15.

Finally, tetramethylbenzidine (TMB) was added to the plate for 10 min, and the reaction was terminated with 2 N sulfuric acid. Absorbance was measured at 450 nm using a microplate reader.

Antibody purification

The gene encoding the scFv antibody was cloned into the pFUSE-Fc expression vector (pfuse-hg1fc2; InvivoGen, San Diego, CA, USA) and transfected to Expi293F cells (A14527; Gibco; Thermo Fisher Scientific) using an ExpiFectamine 293 Transfection Kit (A14524; Gibco; Thermo Fisher Scientific). After 3 days, the supernatant was filtered with a 0.2 μm filter and purified using an AKTA Prime Plus protein purification system (Cytiva, Marlborough, MA, USA) with a HiTrap Protein G HP column (17–0404-01; Cytiva).

Surface plasmon resonance spectrometry

Binding kinetics of A11 to the GFRAL extracellular domain were determined by surface plasmon resonance (SPR) using a Biacore T200 SPR system (Cytiva). Recombinant human GFRAL extracellular domain was immobilized on a CM5 sensor chip (BR-1005-30; Cytiva), and A11 was dissolved in HBS-EP Buffer (BR100188; Cytiva) prepared at final concentrations of 1.89–31.25 nM by 2-fold serial dilutions. Data were processed using Biacore T200 Evaluation Software v3.2 (Cytiva) and 1:1 binding model analysis. The equilibrium dissociation constant (K_D) was calculated according to the following formula: K_D = k_off (dissociation rate, s⁻¹)/k_on (association rate, M⁻¹ s⁻¹).

Cells and cell culture

HEK293 cells (CRL-1573) were obtained from the American Type Culture Collection (Manassas, VA, USA), Expi293F cells were obtained from Thermo Fisher Scientific, and B16F10-luc-G5 cells were obtained from Caliper Life Sciences (Hopkinton, MA, USA). The cells were cultured at 37°C with 5% CO₂. HEK293 and B16F10-luc-G5 cells were cultured in Dulbecco's modified Eagle's medium (Hyclone, Logan, UT, USA) with 10% fetal bovine serum and 1X Antibiotic-Antimycotic solution (15240062, Gibco; Thermo Fisher Scientific). Expi293F cells were cultured in Expi293 Expression Medium (A1435101, Gibco; Thermo Fisher Scientific). For in vitro analyses, HEK293 cells were transfected with vectors expressing human GFRAL (OHu31183D; GenScript, Piscataway, NJ, USA), human RET (HG11997-CF; Sino Biological), and luciferase under a serum response element (SRE) promoter (SRE-luciferase) (E1340; Promega, Madison, WI, USA) using Lipofectamine 3000 Transfection Reagent (L3000001; Thermo Fisher Scientific).

Luciferase reporter assay

HEK293 cells transfected with human GFRAL, human RET, and SRE-luciferase genes were seeded in 96-well white plates (3610; Corning) and tested under serum starvation conditions. After 2 h of starvation, anti-GFRAL antibodies were added at concentrations of 0, 1, 2, 5, 10, 20, 50, and 100 μg/mL and incubated for 2 h at 37°C. Then, recombinant human GDF15 (10 nM) was added and incubated for 4 h at 37°C. Bright-Glo Luciferase Assay System (E2620; Promega) was used to activate the luciferase, and the luminescent signal was measured using an EnVision Multimode Plate Reader (PerkinElmer, Waltham, MA, USA).

Flow cytometry analysis

Human GFRAL gene-transfected HEK293 cells were fixed with 4% paraformaldehyde (PFA) solution in PBS for 10 min at 25°C. After washing with PBS, the cells were blocked using PBST (0.1% Tween in PBS) with 5% normal goat serum and 1% BSA at 4°C for 1 h. Then, the samples were incubated with a control antibody or A11 (0.5, 5, and 50 nM) at 4°C for 1 h. Next, the cells were washed with PBS and incubated with FITC-conjugated goat anti-human IgG antibody (1:1000, ab97224; Abcam) at 4°C for 1 h. After washing with PBS, the cells were analysed by flow cytometry using a BD FACSAria III Cell Counter (BD Life Sciences, Franklin Lakes, NJ, USA). Data were analysed with FlowJo v10.7 software (BD Life Sciences).

Immunofluorescence analysis

For immunocytochemistry, the cells were seeded onto 4-well chamber slides (Nalge Nunc International, Rochester, NY, USA) and fixed with 4% PFA at 25°C for 10 min. As a primary antibody, commercial anti-GFRAL antibody (1:200, ab235111; Abcam) or A11 (35 μg/mL) was added and incubated at 25°C for 1 h. As a secondary antibody, Alexa Flour 488-conjugated goat anti-rabbit IgG antibody (1:1000, A11008; Invitrogen; Thermo Fisher Scientific) or FITC-conjugated goat anti-human IgG antibody was added and incubated at 25°C for 1 h. The cells were stained with DAPI solution and fluorescence images were obtained using an FV1200 laser scanning fluorescence confocal microscope (Olympus Corp., Tokyo, Japan).

For immunohistochemistry, 30-μm-thick coronal sections of mice brains were obtained using a cryostat microtome (CM3050S; Leica, Wetzlar, Germany) after perfusion fixation. The sections were blocked using PBST (0.3% Triton X-100 in PBS) with 5% donkey serum at 25°C for 1 h before incubation at 4°C overnight in a humidified chamber with primary antibodies; anti-GFRAL antibody (1:100, AF5728; R&D systems) and anti-Fos/c-Fos antibody (E-8) (1:100, sc-166940; Santa Cruz Biotechnology, Dallas, TX, USA). The following day, the sections were incubated with secondary antibodies; Alexa Fluor 594-conjugated donkey anti-rabbit IgG antibody (1:1000, A21203; Invitrogen; Thermo Fisher Scientific) and Alexa Flour 488-conjugated donkey anti-mouse IgG antibody (1:1000, A11015; Invitrogen; Thermo Fisher Scientific) at 25°C for 2 h. The sections were stained with DAPI solution, and fluorescence images were obtained using an FV1200 laser scanning fluorescence confocal microscope.

Western blot analysis

Human GFRAL and human RET gene-transfected HEK293 cells were seeded in a 24-well plate and tested under serum starvation conditions. After 2 h, A11 was added into the wells at concentrations of 0, 1, 5, 10, 20, and 50 μg/mL and incubated for 2 h at 37°C. Then, recombinant human GDF15 (50 nM) was added and incubated for 5 min at 37°C. Next, the cells were harvested, and lysates were denatured in sodium dodecyl sulfate (SDS) sample buffer containing 2% 2-mercaptoethanol. The samples were boiled for 10 min, separated by SDS-polyacrylamide gel electrophoresis, and then transferred onto nitrocellulose membranes (IB301002; Invitrogen; Thermo Fisher Scientific). Primary antibodies (1:1000): anti-RET (C31B4) rabbit antibody, anti-AKT antibody, anti-p-AKT (Ser473) antibody, anti-p44/42 MAPK (Erk 1/2) antibody, anti-p-p44/42 MAPK (Erk 1/2) (Thr202/Tyr204) antibody, anti-p-RET (Tyr905) antibody (3223S, 9272S, 9271S, 9102S, 9101S, and 3221S respectively; Cell Signaling Technologies), and a secondary antibody (1:1000): anti-rabbit-IgG, HRP-linked antibody (7076S; Cell Signaling Technologies) were used and the protein bands were visualized using the chemiluminescent substrates. ChemiDoc XRS+ system (Bio-rad, Hercules, CA, USA) was used for quantification. The relative band intensity was measured using ImageJ software (NIH, Bethesda, MD, USA). The data were expressed as fold-change compared with the human GDF15 only-treated group across three independent experiments.

Animals and in vivo experiments

All animal experimental protocols were approved by the DGIST Institutional Animal Care and Use Committee (DGIST-IACUC-20112405-0000). For drug administration, cisplatin (1134357; Sigma-Aldrich, St. Louis, MO, USA) was dissolved in normal saline and administered to the mice via intraperitoneal injection (10 mg/kg). Control antibody or A11 were dissolved in PBS and administered to the mice via subcutaneous injection (10 mg/kg) 1 day prior to the cisplatin treatment. Drugs were administered every 3 days. For the establishment of a cisplatin-induced cachexia model, 8-week-old male C57BL/6 mice weighing ~20 g were housed in individual cages (n = 7 mice per group) and maintained under standard conditions (20–24°C, 45–65% relative humidity) on a 12 h light–dark cycle. The mice were provided a normal chow diet and water ad libitum. For the syngeneic tumour model, B16F10-luc-G5 cells (5 × 10⁵) were suspended in PBS and subcutaneously injected into the right flank of 8-week-old male C57BL/6 mice (n = 64). After tumour cell inoculation, tumour-bearing mice were randomly divided into four groups (n = 16 mice per group), and control mice without tumours (n = 10) were housed in individual cages and provided a normal chow diet and water ad libitum. After the mice were euthanized, blood and various tissues, including skeletal muscles (quadriceps, gastrocnemius, and soleus) and adipose tissues (epididymal white adipose tissue [eWAT] and inguinal white adipose tissue [iWAT]) were collected and analysed. In the case of death or tissue preparation limitations due to tumour growth, these mice were excluded from the statistical analysis.

Quantitative real-time polymerase chain reaction (qPCR)

Total RNA was extracted from tissues using the MiniBEST Universal RNA Extraction Kit (9767; TaKaRa, San Jose, CA, USA) and complementary strands of DNA were generated from total RNA by PrimeScipt 1st strand cDNA Synthesis Kit (6110A; TaKaRa). PCR products were detected using TB Green Premix Ex Taq Kit (RR420A; TaKaRa), and all amplifications were performed in triplicate in a 20 μL reaction volume with the indicated primer pairs (GAPDH; FW: CATCACTGCCACCCAGAAGACTG, RV: ATGCCAGTGAGCTTCCCGTTCAG, MuRF1 [Trim63]; FW: TACCAAGCCTGTGGTCATCCTG, RV: ACGGAAACGACCTCCAGACATG, Atrogin-1 [Fbxo32]; FW: CTTCTCGACTGCCATCCTGGAT, RV: TCTTTTGGGCGATGCCACTCAG, ATGL [Pnpla2]; FW: GGAACCAAAGGACCTGATGACC, RV: ACATCAGGCAGCCACTCCAACA, HSL [Lipe]; FW: GCTCATCTCCTATGACCTACGG, RV: TCCGTGGATGTGAACAACCAGG). Analysis was performed using StepOnePlus qRT-PCR System (Applied Biosystems; Thermo Fisher Scientific). The relative quantity of mRNA samples was calculated by comparative threshold cycle (Ct) analysis after normalization to GAPDH.

Statistical analysis

Data are presented as mean ± standard error of the mean (SEM). Statistical analyses were performed using GraphPad Prism 9 (GraphPad Software, San Diego, CA, USA). The Kruskal–Wallis test followed by the uncorrected Dunn's test were used for multiple comparisons and post hoc analysis. The Mann–Whitney U test was used for comparison of two individual groups.

Results GFRAL antagonist antibody was selected by biopanning and cell-based reporter assay

To select the fully human anti-GFRAL antibody, we conducted solution-phase biopanning using a human combinatorial antibody phage library containing 10¹¹ diversity derived from peripheral blood mononuclear cells of 50 donors²⁶ and GFRAL-coupled magnetic beads (Figure 1A). After four rounds of biopanning, we performed polyclonal phage ELISA to investigate the ratios of GFRAL-binding phages among the phage pools from each round. We confirmed that GFRAL-binding phages were successfully amplified from the second round (Figure 1B). To find a monoclonal anti-GFRAL antibody, we randomly picked 76 single colonies from phage-infected E. coli eluted in the fourth round of panning (Figure 1A). Monoclonal phages expressing GFRAL-binding antibodies were produced in deep well plates and tested for human and mouse GFRAL binding by monoclonal phage ELISA (Figure S1). Amino acid sequences of the 30 clones displaying the strongest binding signals were analysed based on ImMunoGeneTics unique numbering. Four distinct clones were identified based on the third complementarity determining region of the heavy chain (Figure 1C). Next, we expressed four anti-GFRAL antibody clones using a mammalian transient expression system (Figure 1A) and confirmed that all anti-GFRAL antibodies bound to both human and mouse GFRAL extracellular domains (Figure 1D). To evaluate the antagonist activity of the anti-GFRAL antibodies, a cell-based luciferase reporter assay was performed using HEK293 cells transfected with human GFRAL, human RET, and SRE-luciferase genes. After the cells were sequentially treated with anti-GFRAL antibodies and GDF15, the GDF15-dependent luminescent signal was measured. Among four clones, only antibody clone A11 showed inhibitory effect on GDF15-induced luminescent signals, with approximately 74% reduction (P < 0.005) (Figure 1E).

Figure 1. Anti-GFRAL antibodies were selected by solution-phase biopanning. (A) Strategy for anti-GFRAL antibody selection using a human combinatorial antibody phage library. Black dotted lines represent GFRAL-binding phage pool. (B) Polyclonal phage ELISA for phage pools from each round of biopanning. (C) Heavy-chain complementarity-determining region 3 (H-CDR3) amino acid sequences of the anti-GFRAL antibody clones. (D) Binding of the anti-GFRAL antibodies to both human and mouse GFRAL extracellular domains verified by ELISA. (E) Luciferase reporter assay of HEK293 cells transfected with human GFRAL, human RET, and SRE-luciferase genes for the selection of the most potent GFRAL antagonist antibody with inhibitory activity against GDF15-induced luminescent signal (n = 5). Data are presented as the mean ± standard error of the mean (SEM), analysed with the Kruskal–Wallis test, followed by the uncorrected Dunn's test. Statistical differences in post hoc testing are indicated as ns = non-significant, ***P [less than] 0.005.

The antibody A11 binds to GFRAL and potently suppresses the GDF15/GFRAL/RET axis

First, the specificity of A11 binding to human GFRAL was determined using ELISA. A11 was found to selectively bind to the human GFRAL extracellular domain in a dose-dependent manner but not to the human GFRα1 extracellular domain, a GDNF receptor family member with a structure similar to GFRAL (Figure S2). Next, kinetic parameters between A11 and human GFRAL extracellular domain were analysed using SPR spectrometry. The binding affinity of A11 to the human GFRAL extracellular domain had a K_D value in the nanomolar range (1.95 nM) calculated using the formula: K_D = k_off (dissociation rate between 180 s and 540 s, s⁻¹)/k_on (association rate between 0 s and 180 s, M⁻¹ s⁻¹) (Figure 2A). Flow cytometry and immunocytochemical analyses were performed to test whether A11 binds to the native structure of GFRAL expressed on the cell surface. To achieve this, a control antibody or A11 was labelled in HEK293 cells transfected with an empty vector or human GFRAL gene. Flow cytometry analysis showed that the control antibody did not bind to the HEK293 cells transfected with the empty vector or human GFRAL gene; however, A11 specifically and concentration-dependently bound to HEK293 cells transfected with the human GFRAL gene (Figure 2B). Immunocytochemical analysis revealed that A11 selectively bound to HEK293 cells transfected with the human GFRAL gene at a level similar to that of a commercial anti-GFRAL antibody (Figure 2C). To investigate the effect of A11 on intracellular signalling of the GDF15/GFRAL/RET axis, we pretreated HEK293 cells overexpressing both human GFRAL and human RET genes with A11, added GDF15 and then performed western blotting. In the GDF15/GFRAL/RET axis-mediated signalling pathway, A11 effectively inhibited GDF15-induced phosphorylation up to 87% in RET (P = 0.0593), 28% in AKT (P = 0.0593), and 75% in extracellular signal-regulated kinase (ERK) (P = 0.0636) in a dose-dependent manner (Figures 2D and S3A and S3B). To examine whether A11 interferes with the binding of GDF15 to GFRAL, we conducted ELISA for biotinylated GDF15 binding to GFRAL in the presence of A11 or unbiotinylated GDF15. In contrast to unbiotinylated GDF15 (the positive control), A11 did not prevent binding of biotinylated GDF15 to GFRAL (Figure S4). Therefore, we suggest the possibility that A11 may inhibit GDF15-induced signalling by blocking the interaction between GFRAL and RET but not that between GFRAL and GDF15. Collectively, our data indicate that A11 is a potent antagonist that inhibits the GDF15/GFRAL/RET-dependent signalling pathway through physical binding to GFRAL.

Figure 2. The antibody A11 binds to GFRAL and inhibits the signal transduction of the GDF15/GFRAL/RET axis. (A) SPR sensorgram showing the binding kinetics of the antibody A11 to the recombinant human GFRAL extracellular domain. Black box indicates concentration of antibody A11 (1.89–31.25 nM). (B) Flow cytometry analysis of the control antibody and antibody A11 binding in wild-type (WT) and human GFRAL gene-transfected HEK293 cells. Black box indicates concentration of antibody A11 (0–50 nM). (C) Representative immunostaining images of commercial anti-GFRAL antibody (left) and antibody A11 (right) binding in WT and human GFRAL gene-transfected HEK293 cells. The white scale bar represents 30 μm. (D) Western blot analysis showing the inhibitory effect of the antibody A11 on the phosphorylation of RET, AKT, and ERK in human GFRAL/RET gene-transfected HEK293 cells. The gel is representative of three independent experiments.

The antibody A11 attenuates cisplatin-induced cachexia through inhibition of the GDF15/GFRAL/RET axis in vivo

To determine the in vivo efficacy of A11 on cisplatin-induced cachexia, we established a mouse model in 8-week-old male C57BL/6 mice. A11 was administered subcutaneously 1 day before intraperitoneal treatment of cisplatin (Figure S5). To induce severe cachectic symptoms by rapid accumulation of cisplatin in a short period of time, we treated 10 mg/kg of cisplatin every 3 days.¹⁰ We administered 10 mg/kg of A11, as in previous studies investigating the anti-cachectic effects of antibodies targeting GDF15/GFRAL/RET axis.^9,11 Cisplatin treatment reduced food intake, whereas the administration of A11 recovered the cisplatin-induced anorexia by 29% (P < 0.05) (Figure 3A). Reduction of food intake in cisplatin-treated mice led to body weight loss, but the administration of A11 also improved cisplatin-induced weight loss by 7% (P = 0.053) on day 5 (Figure 3B). In addition to anorexia and body weight loss, cisplatin has been reported to cause loss of skeletal muscles and adipose tissues.^5,6,10 Mice were euthanized on day 5, and various tissues were collected and weighed. A11 recovered cisplatin-induced muscle atrophy by 20% in quadriceps (P < 0.05), 5% in gastrocnemius (P = 0.17), and 13% in soleus (P < 0.05) (Figure 3C). In the adipose tissues, administration of A11 improved cisplatin-induced loss of adipose tissues by 13% in eWAT (P = 0.3) and 23% in iWAT (P < 0.05) (Figure 3D). Next, we investigated whether the improvement in cisplatin-induced cachexia by A11 was mediated through the regulation of blood GDF15 level. Plasma GDF15 levels in cisplatin-treated mice increased up to 20-fold compared with control mice, but there was no significant change in plasma GDF15 levels after A11 administration (Figure 3E). These data suggest that A11-induced amelioration of cancer cachexia is due to blocking of cisplatin-dependent GDF15 action rather than reduction of GDF15 plasma levels. Cisplatin-induced GDF15 is known to interact with GFRAL-positive neurons in the AP and NTS of the hindbrain, and these activated neurons induce anorexia.^10,11 Immunohistochemical analysis was performed to test whether A11 inhibits the GDF15/GFRAL/RET-dependent signalling pathway in AP and NTS. Because a single injection of 10 mg/kg cisplatin induced elevation of plasma GDF15 levels within 2 h, which was maintained for 3 days,¹⁰ we harvested brain tissue at 4 h post-injection of cisplatin. Cisplatin treatment promoted expression of the neuronal activation marker c-Fos on brainstem, whereas A11 administration prevented c-Fos accumulation (Figure 4A,B). Specifically, A11 administration decreased GFRAL-positive neuron population expressing c-Fos in AP and NTS by 62% (P < 0.05), compared with the cisplatin-only treatment group (Figure 4C). These data indicate that A11 effectively suppresses the activity of GFRAL-positive neurons in the hindbrain by blocking the action of cisplatin-induced GDF15. Conclusively, A11 successfully attenuates cisplatin-induced cachectic symptoms by inhibiting GDF15/GFRAL/RET axis on GFRAL-expressing neurons in vivo.

Figure 3. The antibody A11 attenuates cisplatin-induced cachexia in vivo Normal: control mice with PBS injections, Cis/Con: mice with cisplatin and control antibody injections, Cis/A11: mice with cisplatin and the antibody A11 injections (n = 7 mice per group). (A) Cumulative food intake per mouse on days 2–5. (B) Body weight change compared with the weight before drug administration followed up to day 5. (C) Weights of isolated skeletal muscles (quadriceps, gastrocnemius, and soleus). (D) Weights of isolated adipose tissues (eWAT and iWAT). (E) Plasma levels of GDF15 on day 5, determined by ELISA. Data are presented as the mean ± SEM, analysed with the Kruskal–Wallis test, followed by the uncorrected Dunn's test. Statistical differences in post hoc testing are indicated as ns = non-significant, *P [less than] 0.05, **P [less than] 0.01, ***P [less than] 0.005, and ****P [less than] 0.001.

Figure 4. The antibody A11 suppresses c-Fos accumulation on the hindbrain of cisplatin-treated mice Normal: control mice with PBS injections, Cis/Con: mice with cisplatin and control antibody injections, Cis/A11: mice with cisplatin and the antibody A11 injections (n = 5 mice per group). (A) Representative sections of the area postrema (AP) and the nucleus of the solitary tract (NTS) showing c-Fos expression in GFRAL-expressing neurons 4 h after cisplatin treatment on day 4. White arrows represent neurons stained with both GFRAL and c-Fos, indicating activation by cisplatin-induced GDF15. The white scale bars represent 150 μm and the yellow scale bar represents 20 μm. The CC represents central canal. (B) Quantification of cisplatin-induced c-Fos expression in the AP/NTS 4 h after cisplatin injection. Each point represents one mouse. Data from each mouse represent quantification from 6 alternative sections. (C) Number of GFRAL-positive neurons in the AP/NTS that co-express c-Fos 4 h after cisplatin injection. Each point represents one mouse. Data from each mouse represent quantification from 6 alternative sections. Data are presented as the mean ± SEM, analysed with the Kruskal–Wallis test, followed by the uncorrected Dunn's test. Statistical differences in post hoc testing are indicated as ns = non-significant, *P [less than] 0.05, **P [less than] 0.01.

The antibody A11 alleviates cisplatin-induced cachexia in melanoma-bearing mice

To investigate whether A11 ameliorates cisplatin-induced cachexia in tumour-bearing mice, we first established a melanoma mouse model by inoculating B16F10-luc-G5 cells into 8-week-old male C57BL/6 mice. In this experiment, we chose B16F10, a GDF15-non-expressing tumour, to examine the effect of the GFRAL antagonist antibody on alleviating cachectic symptoms specifically caused by cisplatin-induced GDF15 but not by tumour-derived GDF15. When the tumour volume reached 50 mm³, cisplatin was administered by intraperitoneal injection (10 mg/kg), and A11 was administered by subcutaneous injection (10 mg/kg) one day before cisplatin treatment (Figure S6A). Cisplatin treatment significantly suppressed tumour growth and splenomegaly, but A11 administration did not affect tumour volume and weight of the spleen, indicating that it had no impact on the anti-cancer effects of cisplatin (Figure S6B–D). In this model, it was observed that treatment with cisplatin significantly reduced food intake while the tumour per se had no effect on appetite compared with the normal group (Figure 5A). Interestingly, A11 administration potently recovered cisplatin-induced anorexia. Food intake increased by 21% (P < 0.05) on days 13–18 (Figure 5A), leading to improving cisplatin-induced loss of body weight by 13% (P < 0.05) in tumour-free body weight on day 17 (Figure 5B). These data indicate that cisplatin treatment increased the severity of anorexia and body weight loss independent of tumour growth, and A11 administration effectively ameliorated cisplatin-induced cachexia. The mice were sacrificed on day 18, and various tissues were collected. Tumour growth slightly reduced the weight of skeletal muscles and adipose tissues, whereas cisplatin treatment caused additional weight loss. A11 administration significantly recovered the cisplatin-induced loss of skeletal muscles by 21% in quadriceps (P < 0.05), 9% in gastrocnemius (P < 0.05), and 13% in soleus (P < 0.05) (Figures 5C and S7A). In the adipose tissues, administration of A11 improved cisplatin-induced depletion by 37% in eWAT (P < 0.05) and 51% in iWAT (P < 0.05) (Figures 5D and S7A). To elucidate the molecular mechanism underlying chemotherapy-induced cachexia, we analysed mRNA expression for key markers of muscle atrophy and lipid oxidation from collected tissues on day 12. In gastrocnemius muscle, cisplatin did not affect the expression of MuRF1 and Atrogin-1, so cisplatin-induced muscle wasting could be solely a result of anorexia (Figure S7B). In contrast, in iWAT, cisplatin caused elevated expression of ATGL and HSL, thus cisplatin-induced wasting of adipose tissue could be a combined result of anorexia and cisplatin-induced lipid oxidation (Figure S7C). In both cases, A11 administration did not affect the mRNA expression levels of key markers. To determine whether tumour growth or cisplatin treatment affects blood levels of GDF15, we measured GDF15 levels in the plasma of melanoma-bearing mice. Tumour growth slightly increased plasma levels of GDF15; however, cisplatin treatment significantly increased plasma levels of GDF15 up to 7-fold compared with controls, implying that cisplatin treatment is a major cause of loss in skeletal muscles and adipose tissues, irrespective of wasting due to tumour growth (Figure 5E). Finally, A11 alleviates cisplatin-induced anorexia and depletion of skeletal muscles and adipose tissues in a mouse model of melanoma.

Figure 5. The antibody A11 ameliorates cisplatin-induced cachexia in a mouse model of melanoma Normal: control mice without tumour, TB/Con: tumour-bearing mice with control antibody injections, TB/A11: tumour-bearing mice with the antibody A11 injections, TB/Cis/Con: tumour-bearing mice with cisplatin and control antibody injections, TB/Cis/A11: tumour-bearing mice with cisplatin and the antibody A11 injections (n = 10–16 mice per group). (A) Cumulative food intake per mouse on days 13–18. (B) Body weight change compared with the weight before drug administration followed up to day 17. (C) Weights of isolated skeletal muscles (quadriceps, gastrocnemius, and soleus). (D) Weights of isolated adipose tissues (eWAT and iWAT). (E) Plasma GDF15 levels on day 18, determined by ELISA. Data are presented as the mean ± SEM, analysed with the Kruskal–Wallis test, followed by the uncorrected Dunn's test. Statistical differences in post hoc testing are indicated as ns = non-significant, *P [less than] 0.05, **P [less than] 0.01, ***P [less than] 0.005, and ****P [less than] 0.001.

Discussion

Chemotherapy induces an increase in GDF15 blood levels,^9,27,28 and GDF15-induced activation of the GDF15/GFRAL/RET axis primarily contributes to chemotherapy-induced cachexia symptoms.^9–11 Therefore, inhibiting the GDF15/GFRAL/RET axis is essential in alleviating chemotherapy-induced cachexia.⁹ In this study, we developed the first fully human GFRAL antagonist antibody, which potently blocks the GDF15/GFRAL/RET axis both in vitro and in vivo. We confirmed that the GFRAL antagonist antibody prevents the cisplatin-induced activation of GFRAL-expressing neurons in the brainstem of cisplatin-treated mice. Furthermore, we demonstrated that GFRAL antagonist antibody ameliorates cisplatin-induced cachectic symptoms in a melanoma-bearing mouse model.

The main cause of chemotherapy discontinuation is the deterioration of general health that occurs regardless of the anti-cancer effect.²⁹ Chemotherapy causes involuntary weight loss, decreased immunity, and organ function, which in turn reduces responsiveness and tolerability in cancer patients.⁸ Moreover, restoring anorexia and wasted skeletal muscles and adipose tissues in cancer patients undergoing chemotherapy is essential for the continuation of chemotherapy and prolonging life expectancy, as well as improving nutritional status and recovering physical strength. Therefore, alleviating chemotherapy-induced cachexia through administration of GFRAL antagonist antibody could be a novel anti-cancer therapy that improves the chance of survival of cancer patients beyond improving their quality of life. Previously, Suriben et al. showed that tumour-derived GDF15 induced depletion of skeletal muscles and adipose tissues, but not anorexia, in tumour-bearing mice and that their humanized anti-GFRAL antibody restored weight loss by inhibiting lipid oxidation.¹⁴ This study confirmed the efficacy of anti-GFRAL antibody in tumour-bearing mice with mild cachexia, but not in chemotherapy-treated mice with severe cachexia. Breen et al. demonstrated that the anti-GDF15 antibody ameliorates severe cachectic symptoms and improves survival in chemotherapy-treated tumour-bearing mice.⁹ However, because they used GDF15-expressing tumours, both tumour-derived GDF15 and cisplatin-induced GDF15 may collectively cause cachectic symptoms. Therefore, it is necessary to analyse the inhibitory effect on GDF15/GFRAL/RET axis in chemotherapy-treated tumour-bearing mice with severe cachectic symptoms specifically caused by cisplatin-induced GDF15 rather than tumour-derived GDF15. In this study, we demonstrated that fully human anti-GFRAL antibody restored body weight by simultaneously alleviating cachectic symptoms caused only by cisplatin-induced GDF15, such as anorexia and wasting of skeletal muscles and adipose tissues, in cisplatin-treated melanoma-bearing mice. Therefore, our study is a translational study to address the unmet clinical need for treatment of chemotherapy-induced cancer cachexia patients and enhance the clinical applicability of the GFRAL antagonist antibody.

While chemotherapeutic drugs have been successfully used to treat cancer patients, approximately 70–80% of cancer patients undergoing chemotherapy suffer from CINV.³⁰ CINV causes metabolic imbalances, leading to malnutrition and general weakness due to eating disorders, ultimately reducing the overall quality of life.^7,30 Chemotherapeutic agents that show high emetic scores elevate blood GDF15 levels.⁹ GDF15 plays a crucial role in triggering nausea and emesis through the GDF15/GFRAL/RET axis in cisplatin-treated animal models.^9,10 In this study, we verified that our GFRAL antagonist antibody potently suppresses GDF15/GFRAL/RET axis in vivo. Therefore, we expect that GFRAL antagonist antibody will improve the health-related quality of life of chemotherapy-induced cachexia patients through antiemetic and anti-cachectic actions.

GDF15 has been suggested as a biomarker for the diagnosis and prognosis of tumour progression as well as cancer cachexia.¹⁶ High levels of GDF15 are significantly correlated with cancer cell invasion, proliferation, and metastasis in advanced tumours.^31–33 For example, Zhao et al. recently showed that pancreatic ductal adenocarcinoma cells express GFRAL, and GDF15 promotes cancer cell growth and metastasis through GDF15/GFRAL/RET axis.³⁴ Considering the functional role of GDF15 and GFRAL in the tumour microenvironment, the GFRAL antagonist antibody could be a potential anti-cancer agent for GDF15/GFRAL/RET axis-dependent tumorigenesis.

This study has some limitations as well. First, we observed that A11 inhibited GDF15-induced phosphorylation of RET and ERK in vitro and prevented c-Fos accumulation in cisplatin-treated mice. However, to more clearly determine whether A11 directly blocks the GDF15/GFRAL/RET axis, it would be helpful to examine A11-mediated inhibition of kinases potentially downstream of RET such as AKT, ERK, and ribosomal protein S6 in c-Fos/GFRAL double-positive neurons of the mouse brainstem. Second, this study only showed the effect of A11 on alleviation of anorexia and depletion of skeletal muscles and adipose tissue in a singular chemotherapy-treated tumour-bearing mouse model. To further expand the clinical applicability of A11 in cachexia as well as in tumour growth and survival, it is necessary to evaluate the efficacy of our antibody in various chemotherapy-induced cachexia conditions in multiple tumour-bearing mice. Moreover, it needs to be evaluated across a wide range of ages in both male and female subjects. Third, the molecular mechanism underlying A11 action in cisplatin-induced skeletal muscle and adipose tissue depletion are still insufficient. While we show mRNA expression levels of key genes associated with muscle atrophy and lipid oxidation, unbiased approaches such as RNA seq analysis would be helpful for more in-depth mechanistic analysis. Finally, tissue histology, such as cross-sectional area measurement or grip strength analysis, may be more advantageous for a comprehensive evaluation of the effect of the A11 on muscle atrophy.

In conclusion, we found that the GDF15/GFRAL/RET axis is a promising target for the treatment of chemotherapy-induced cachexia and confirmed that our GFRAL antagonist antibody alleviates chemotherapy-induced cachectic symptoms. This is the first study to demonstrate that GFRAL antagonist antibody can ameliorate chemotherapy-induced cachexia in tumour-bearing mice. The GFRAL antagonist antibody could be a novel therapeutic approach to improve the health-related quality of life of cancer patients receiving chemotherapy.

Acknowledgements

The authors certify that they comply with the ethical guidelines for authorship and publishing in the Journal of Cachexia, Sarcopenia and Muscle.³⁵ All animal experiments were performed according to the protocols approved by the DGIST Institutional Animal Care and Use Committee (DGIST-IACUC-20112405-0000).

Funding

This study was supported by the Daegu Gyeongbuk Institute of Science and Technology Start-up Fund Program of the Ministry of Science and ICT (MSIT) (2018010109) and the Bio & Medical Technology Development Program of the National Research Foundation of Korea funded by the Korean government (MSIT) (2019M3A9H1103607 and 2020M3A9I4039539).

Conflict of interest

There are no conflicts of interest to declare.