Identification and Therapeutic Targeting of GPR20, Selectively Expressed in Gastrointestinal Stromal Tumors, with DS-6157a, a First-in-Class Antibody–Drug Conjugate

Gastrointestinal stromal tumors (GIST) are the most common mesenchymal tumors of the gastrointestinal tract, derived from the interstitial cells of Cajal (ICC; ref. 1). Most adult GISTs are driven by activating mutations in KIT proto-oncogene receptor tyrosine kinase (KIT) or platelet-derived growth factor receptor alpha (PDGFRA; refs. 2, 3). However, approximately 10% to 15% of adult and 85% of pediatric GISTs lack KIT/PDGFRA mutations, which is known as KIT/PDGFRA wild-type (WT) GIST. The KIT/PDGFRA WT GISTs are a heterogeneous group, in which various genetic alterations in BRAF, RAS, NF1, or the genes of the succinate dehydrogenase (SDH) complex have been reported (4–6). SDH-deficient GIST is the most common form of KIT/PDGFRA WT GIST. Currently, the only approved treatments for GIST are KIT/PDGFRA-directed tyrosine kinase inhibitors (TKI). TKIs such as imatinib, sunitinib, and regorafenib have improved the survival of patients with KIT-mutant GIST (7–9). However, patients ultimately experience disease progression most often due to the development of heterogeneous secondary resistance mutations in KIT (10–12). Recently, two novel TKIs were approved by the FDA: avapritinib for the treatment of GIST with exon 18 mutations of PDGFRA and ripretinib for patients with advanced GIST who have received at least three prior kinase inhibitors, including imatinib, respectively (13). Currently approved TKIs have limited activity against KIT/PDGFRA WT GIST. Therefore, it is essential to develop novel therapeutic strategies with unique modes of action for advanced GIST.

Multiple previous transcriptional studies have indicated that GIST cells may be enriched for a novel target known as G protein–coupled receptor 20 (GPR20), which is a 358-amino acid orphan GPCR (14, 15). GPR20 constitutively activates Gi proteins without ligand stimulation when exogenously expressed in HEK293 cells (16). GPR20 is abundantly expressed in GIST (14). The expression of GPR20 is regulated by Forkhead box F1 (FOXF1) and ETS variant transcription factor 1 (ETV1) transcription factors that play critical roles in GIST initiation, proliferation, and survival (17, 18). In mice, GPR20 is expressed in subsets of ICC (18). The function of GPR20 in GIST or ICC has not yet been elucidated.

Antibody–drug conjugates (ADC) are a promising treatment modality that take advantage of the specificity of monoclonal antibodies (mAb) to deliver potent cytotoxic drugs selectively to antigen-expressing tumor cells. The mechanisms of action of ADCs are generally thought to be as follows: after binding of the ADC to the cell-surface antigen, the ADC-antigen complex undergoes internalization and translocation to the lysosomal compartment where the ADC is cleaved to release a payload drug that causes cell death (19, 20). DXd-ADC is a DNA topoisomerase I (TOP1) inhibitor–based ADC technology that utilizes an exatecan derivative (DXd) as the cytotoxic drug and a protease cleavable maleimide Gly–Gly–Phe–Gly (GGFG) tetrapeptide-based linker (21–23). A key function of TOP1 is the relaxation of DNA supercoiling by inducing transient single-strand DNA breaks for DNA replication and transcription (24). TOP1 inhibitors bind to TOP1–DNA cleavage complexes and stabilize them, resulting in the induction of double-strand DNA breaks and cell apoptosis. DXd exhibits stronger TOP1 inhibitory activity than SN-38 (25). The maleimide GGFG tetrapeptide-based linker is designed to be preferentially cleaved by lysosomal enzymes such as cathepsins, which are highly expressed in tumor cells (26, 27), thereby limiting free payload in plasma (28). This linker-payload technology enables a stable and high (8 DXds per antibody) drug-to-antibody ratio (DAR) without compromising the physicochemical properties. Its utility has been demonstrated in [fam]-trastuzumab deruxtecan (DS-8201a), a HER2-targeting ADC, and U3-1402, a HER3-targeting ADC, both of which have been demonstrated potent antitumor activity in preclinical studies (25, 29, 30) and promising efficacy and safety in patients with cancer (31). Recently, the FDA has approved DS-8201a for the treatment of adult patients with unresectable or metastatic HER2-positive breast cancer who have received two or more prior anti-HER2–based regimens in the metastatic setting.

We hypothesized that GPR20 is a potential therapeutic target for ADC development for the treatment of GIST. We now describe expression of GPR20 protein assessed by IHC staining on GIST samples from Dana-Farber Cancer Institute (DFCI; n = 139), and National Cancer Center Hospital East, Japan (NCCHE; n = 100), as well as on normal and malignant tissue microarrays (TMA) obtained commercially. In addition, an anti-GPR20 DXd-ADC (DS-6157a) was generated to assess antitumor activity in GIST models as well as pharmacokinetics and toxicologic profiles in rats and monkeys.

We generated rat mAbs raised against human GPR20 by a DNA immunization method and obtained three rat anti-GPR20 mAbs, designated 04-046, 04-093, and 04-021 (Fig. 1A), that showed different binding properties (Fig. 1B). The 04-046 and 04-093 mAbs bound to both human GPR20 and N-terminal FLAG epitope-tagged human GPR20 but not to mouse GPR20. Utilizing this information, we conducted binding tests against four recombinant human GPR20 proteins with each of four extracellular domains (ECD1 to ECD4) replaced with the corresponding mouse GPR20 sequence. Binding of 04-046 to human GPR20 was significantly reduced when ECD1 or ECD2 were replaced with those of mouse GPR20, indicating that 04-046 mAb binds to the conformational structure composed of ECD1 and ECD2. On the other hand, binding of 04-093 mAb to human GPR20 was reduced by replacing ECD1 with that of mouse GPR20, indicating 04-093 mAb bound to a linear epitope located in the 48 amino acid sequence of ECD1 (Fig. 1C). Binding of 04-021 mAb to human GPR20 was abolished by addition of a FLAG-tag at N-terminus, indicating that the N-terminal portion of GPR20 was associated with its binding. The 04-046 mAb showed high internalization activity (Fig. 1D). Therefore, we used 04-046 mAb to generate an anti-GPR20 ADC, DS-6157a, after humanization of the antibody to reduce immunogenicity. Another 04-093 mAb was found to be suitable for IHC of GPR20 protein on formalin-fixed paraffin-embedded (FFPE) cells as well as tissues. GPR20-specific signal was observed from 293T cells transfected with a human GPR20 expression vector, whereas no signal was observed on 293T cells transfected with the empty vector (Fig. 1E). The rat 04-093 mAb was converted to a rabbit anti-GPR20 mAb, designated 04-093OcH1L1 for diagnostic use.

Expression of GPR20 and KIT proteins was assessed by IHC on GIST tissue samples from DFCI (n = 139), using the 04-093OcH1L1 and rabbit anti-CD117, c-Kit (DAKO), respectively. Overall sample characteristics and statistical analyses are summarized in Supplementary Tables S1 and S2, respectively. GPR20 was detected in 126 (91%) samples. GIST specimens generally expressed both GPR20 and KIT homogenously (Fig. 2A). We found a significant positive correlation between GPR20 and KIT expression (Fig. 2B). In 30 samples with strong positive KIT expression (IHC score 3+), 29 samples (97%) expressed GPR20, including 21 samples (70%) that were strong GPR20-positive. More than 90% of KIT-positive GIST samples expressed GPR20. In 23 samples with a KIT expression score of zero, 15 samples expressed GPR20 (65%), including five samples (22%) that were strongly GPR20-positive. GPR20 was expressed in 85% or more of GIST, regardless of tumor location. The proportion of GPR20 strong positive (IHC score 3+) in GIST from small intestine and abdominal cavity (metastatic lesions) was 84% and 71%, respectively (Fig. 2C). High-level expression of GPR20 (IHC score 3+) was more common in small intestinal GIST than in gastric GIST. Genotype data were available for 124 samples (Fig. 2D). We did not find a difference in GPR20 expression levels between KIT-mutant GIST and KIT/PDGRFA WT GIST, whereas GPR20 expression was lower in PDGFRA-mutant GIST. Although all GIST genotypes exhibited a high fraction of GPR20 positivity (≥89%), the ratio of GPR20 strong positive (IHC score 3+) was higher in the KIT-mutant GIST (55%) and KIT/PDGFRA WT GIST (69%) than in the PDGFRA-mutant GIST (0%). Seven of eight GPR20-positive PDGFRA-mutant GIST harbored D842V point mutation.

Among 99 KIT-mutated samples, GPR20 was expressed in both GIST harboring a primary mutation and GIST with primary and secondary mutations in KIT (Fig. 2E). The GPR20 expression level was higher in KIT exon 9 mutant samples in which all 23 samples (100%) expressed GPR20, with 22 (96%) of them having an IHC score of 3+, compared with KIT exon 11 mutant samples in which 27 samples (77%) expressed GPR20, including seven (20%) having an IHC score of 3+. GPR20 was also expressed in single KIT exon 13 or 17 mutant samples as well as in double KIT exons 9+11, 9+17, 11+ 13/14, and 11+17 mutant samples. The mutations in KIT exon 13/14 and 17 are associated with TKI resistance. One sample harbored a mutation in 4 KIT exons: 11+13+17+18. This sample did not show expression of GPR20 by IHC (Fig. 2E). In comparison between single and two KIT mutations, a significantly higher expression was observed in GIST with KIT exon 11+17 mutations than in KIT exon 11 mutant GISTs.

GPR20 was expressed in all 16 KIT/PDGFRA WT GIST samples, of which 11 samples (69%) had strong IHC staining (Supplementary Table S3). We tried to understand the underlying genetic causes and identified that these WT GISTs harbored germline mutations of SDH, NF1 gene mutation, were Carney triad, Carney Stratakis syndrome, or quad-negative (KIT^WT/PDGFRA^WT/SDH^WT/BRAF^WT) GISTs.

The 139 GIST samples were classified by treatment history into 46 treatment-naïve samples and 93 samples that had received prior TKI treatment (Fig. 2F). GPR20 expression was detected in more than approximately 80% of GIST specimens across all treatment lines (IHC score ≥ 1+). Strong GPR20-positive (IHC score 3+) populations were observed in 26% of treatment-naïve samples and 48%, 50%, 71%, and 58% of samples progressed or refractory to the first, second, third treatment and samples after neoadjuvant, respectively. The strong GPR20 positive population was higher in GISTs progressed on third treatment, compared with treatment-naïve GIST samples (Fig. 2F).

GPR20 expression was also assessed on treatment-naïve GIST samples from NCCHE (n = 100, Supplementary Tables S4 and S5), by a prototype GPR20 IHC using rat anti-GPR20 antibody 04-093. The sensitivity of the detector rat 04-093 Ab is lower than the rabbit 04-093OcH1L1 Ab used for testing GIST samples from DFCI. Most of GIST samples in a commercially available GIST TMA stained with rat 04-093 to GPR20 IHC score 3+ and 2+ corresponded to GPR20 strong positive (IHC score 3+) when stained with rabbit 04-093OcH1L1 (Supplementary Fig. S1A). Similar GPR20 expression results were obtained in the GIST samples derived from Japanese patients. Eighty-eight samples (88%) from NCCHE were GPR20-positive (IHC score ≥1+). Similar to the results of the DFCI samples, we found a significant positive correlation between GPR20 and KIT IHC scores (Supplementary Fig. S1B). Higher expression of GPR20 was observed in small intestinal GIST where 17 samples (94%) were strong GPR20 positive (IHC score ≥2+) compared with gastric GIST (34%; Supplementary Fig. S1C). Genotype data were available for 86 samples. The proportion of high GPR20 IHC score ≥2+ was higher in KIT-mutant GISTs compared with PDGFRA-mutant GISTs (Supplementary Fig. S1D). It should be noted that high expression of GPR20 in GIST with KIT exon 9 mutations, which was observed in DFCI samples, was not observed in NCCHE samples (Supplementary Fig. S1E).

In the IHC evaluation of large sections of GIST samples from DFCI, we confirmed KIT expression on both mast cells and the ICC; however, GPR20 was expressed only on ICC and not expressed on mast cells (Fig. 2G). Expression of GPR20 was tested on FDA human multiple organ normal TMA using rabbit 04-093OcH1L1 (Supplementary Table S6). No specific staining was noted in any human normal tissue including gastrointestinal sections that did not include the myenteric plexus of gut.

To evaluate the expression of GPR20 in tumor tissues, a commercially available multiple organ TMA (Supplementary Table S7) and a sarcoma TMA provided by the University of Basel containing various sarcomas of the bone and soft tissue including 17 GIST cases (Supplementary Table S8; Supplementary Fig. S2A) were stained. GPR20 was expressed on all of the GIST samples including 13 samples (76%) with strong GPR20 positive (IHC score 3+), but no GPR20 expression was detected in any other tumors, except for a single leiomyosarcoma case that was negative for KIT or DOG1 (Supplementary Fig. S2B). The other 15 leiomyosarcoma cases were GPR20 negative. As gain-of-function mutations in KIT are detectable in more than 90% of systemic mastocytosis (32), we evaluated GPR20 expression in 20 cases of systemic mastocytosis. All cases were GPR20 negative (Supplementary Fig. S3A and S3B), indicating that activated KIT signaling is not sufficient for GPR20 expression.

In summary, among all the tested samples, GIST and ICC, the cell of origin of GIST and one leiomyosarcoma case were the tissues found to express GPR20 by IHC.

To evaluate the potential activity of GPR20-targeting ADC for the treatment of GIST, we generated DS-6157a with humanized anti-GPR20 antibody derived from the rat 04-046 mAb utilizing the DXd-ADC technology (Fig. 3A). Hydrophobic interaction chromatography showed a homogeneous drug distribution, and the DAR of synthesized DS-6157a was designed to be approximately eight (Fig. 3B). DS-6157a bound to human and cynomolgus monkey GPR20, but not to mouse or rat GPR20 (Fig. 3C). Expression of all GPR20 orthologs on the CHO-KI cells was detected by anti-FLAG antibody (Supplementary Fig. S4). Similar binding affinity of DS-6157a to human and cynomolgus monkey GPR20 supports the appropriateness of using cynomolgus monkeys for nonclinical pharmacokinetic (PK) and toxicologic evaluation of DS-6157a. The cell growth–inhibitory activity of DS-6157a was evaluated in GIST-T1/GPR20 cells that were derived from the human GIST-T1 cell line by stable integration of a human GPR20 expression vector, and in the human gastric carcinoma cell line NCI-N87. Flow-cytometric analysis showed that GPR20 was markedly expressed on the cell surface of GIST-T1/GPR20 cells, whereas no significant expression of GPR20 was detected on the NCI-N87 cells (Fig. 3D). Both GIST-T1/GPR20 and NCI-N87 cells were susceptible to DXd with IC₅₀ of 1.37 nmol/L and 3.57 nmol/L, respectively (Fig. 3E). DS-6157a showed cell growth–inhibitory activity in GPR20-positive GIST-T1/GPR20 cells with IC₅₀ of 156 ng/mL but not in GPR20-negative NCI-N87 cells, whereas neither the parent anti-GPR20 antibody nor a nonspecific IgG-ADC showed cell growth–inhibitory activity in both cells (Fig. 3F), indicating that DS-6157a exerts GPR20-dependent as well as payload-dependent cytotoxic activity.

Intracellular trafficking of DS-6157a was evaluated by time-lapse imaging analysis in GIST-T1/GPR20 cells, using DS-6157a and IgG-ADC labeled with a pH-sensitive dye, pHrodo. After binding to cell-surface GPR20, DS-6157a was internalized by the cells and translocated to the lysosome, which was traced by monitoring pHrodo probe label-derived orange dots, compared with IgG-ADC (Fig. 4A and B). To verify the contribution of payload drug to the cytotoxic activity of DS-6157a, we analyzed the induction of phosphorylated checkpoint kinase 1 (pChk1) as a marker of DNA damage (33, 34) and cleaved poly(adenosine diphosphate-ribose) polymerase (PARP) as a marker of apoptosis (35), which are known to be induced by DNA topoisomerase I inhibitors. Both pChk1 and cleaved PARP were induced in the cells treated with either DS-6157a or DXd (Fig. 4C). In contrast, neither pChk1 nor cleaved PARP were induced in the cells treated with the parent anti-GPR20 antibody or IgG-ADC. These results indicate that DS-6157a induced DNA damage and apoptosis in the same manner as DXd, and suggests that these changes are caused by the DNA topoisomerase I inhibition activity of the DXd released from DS-6157a.

Antibody-dependent cellular cytotoxicity (ADCC) activity of DS-6157a was evaluated by detecting specific cell lysis of GIST-T1/GPR20 cells mediated by human peripheral blood mononuclear cells (PBMC). DS-6157a showed ADCC activity (Fig. 4D). On the other hand, there was no complement-dependent cytotoxicity (CDC) activity associated with DS-6157a or anti-GPR20 antibody against GIST-T1/GPR20 cells (Fig. 4E), whereas rituximab showed CDC activity against CD20-positive Ramos cells in the presence of the same human complement serum.

The in vivo antitumor activity of DS-6157a was evaluated in GIST cell line–derived xenograft (CDX) models and in more clinically relevant GIST patient-derived xenograft (PDX) models. In the GIST-T1/GPR20 and GIST-T1 (the parent cell line of GIST-T1/GPR20) models administered intravenously with an escalating dose of DS-6157a, dose-dependent tumor growth inhibition (TGI) and tumor regression at doses of 3 to 10 mg/kg were observed (Fig. 5A and B). Comparing tumor volumes of animals receiving DS-6157a at 3 mg/kg, GIST-T1/GPR20 CDX was more sensitive to DS-6157a than GIST-T1 CDX due to higher expression of GPR20 (IHC score 3+ vs. 2+) in the same genetic background of the cells. Furthermore, treatment with 10 mg/kg DS-6157a (every 3 weeks, 2 times) showed potent antitumor activity against GIST-T1 where the tumor regression continued 8 weeks after last dosing. Induction of DNA damage (γH2AX, pKAP1, and pChk1) and apoptosis (cleaved PARP) markers after administration of DS-6157a were observed in both GIST-T1/GPR20 and GIST-T1 xenograft tumors (Fig. 5C). In mice with GIST1 PDX (Fig. 5D) and GIST6 PDX (Fig. 5E) treated with 10 mg/kg DS-6157a every 3 weeks, significant TGI was observed. Importantly, treatment with 10 mg/kg of the parent anti-GPR20 antibody or IgG-ADC had no impact on tumor growth in the GIST6 PDX model study (Fig. 5E), indicating GPR20- and payload-dependent in vivo antitumor activity of DS-6157a. Supporting this, DS-6157a was ineffective in the GPR20-negative KPL-4 model (Supplementary Fig. S5).

Next, DS-6157a was tested in GIST#1001 PDX, GIST430/654 CDX, and GIST#1338 PDX models, comparing its antitumor activity with approved TKIs such as imatinib, sunitinib, and regorafenib (Fig. 6). Treatment with DS-6157a at 5 and 10 mg/kg blocked the tumor growth against GIST#1001, whereas IgG-ADC had no antitumor activity (Fig. 6A). TGI by two TKIs were almost equivalent to that of DS-6157a in GIST#1001 models. GIST430/654 was resistant to imatinib due to V654A mutation in KIT exon 13, which is a known imatinib-resistant secondary mutation (Fig. 6B). DS-6157a showed TGI against GIST430/654; its antitumor activity was equivalent to or stronger than sunitinib or regorafenib. Finally, we tested DS-6157a against GIST#1338 PDX derived from a patient with GIST who progressed on regorafenib treatment. In the GIST#1338 PDX model treated with DS-6157a, the greatest TGI was observed, compared with imatinib, sunitinib, and regorafenib (Fig. 6C). Taken together, these studies demonstrate that DS-6157a has potential antitumor activity against GPR20-positive GIST, regardless of TKI resistance mutations.

Pharmacokinetic profiles of DS-6157a, total antibody, and DXd in cynomolgus monkeys were evaluated after single intravenous administration of DS-6157a at doses of 0.1 to 30 mg/kg (Fig. 6D). DS-6157a increased the plasma exposure in a dose-dependent manner. The volume of distribution at steady state of DS-6157a (mean range, 53.2–59.9 mL/kg) was similar to the plasma volume seen in cynomolgus monkeys (36). No clear difference was observed in the plasma concentration–time profile between DS-6157a and the total antibody, indicating that the peptide linker of DS-6157a is stable in plasma. Plasma DXd exposure was also increased in a dose-dependent manner at more than 1 mg/kg, although DXd was not detected due to the concentration below the limit of quantitation of 0.100 ng/mL when animals were dosed at from 0.1 to 1 mg/kg.

Safety profiles of DS-6157a were characterized in intermittent intravenous dosing studies in rats and cynomolgus monkeys. DS-6157a was administered once every three weeks for six weeks (three times in total), followed by a two-month recovery period. Target organs of toxicity in rats and monkeys are summarized in Table 1. Major toxicity findings were found in the intestines, kidneys, liver, lungs, lymphatic/hematopoietic organs, reproductive organs, and skin in rats and/or monkeys. All these findings were minimal or mild. All toxicologic changes except for the testicular changes in rats showed recovery. More importantly, no histopathologic changes in the GPR20-positive cells (e.g., Cajal cells) were observed in cynomolgus monkeys, a cross-reactive species to DS-6157a. On the basis of the toxicology findings, the severely toxic dose of 10% (STD₁₀) was considered to be more than 199 mg/kg in rats [the human equivalent dose (HED) = 32 mg/kg], and the highest nonseverely toxic dose (HNSTD) for monkeys was considered to be 30 mg/kg (HED = 9.6 mg/kg; ref. 37). These nonclinical toxicologic findings suggest acceptable safety profiles of DS-6157a, which support its entry into human trials with a starting dose of 1.6 mg/kg (1/6 of the HED of HNSTD in monkey).

GPR20 is selectively and abundantly expressed in GIST. We aimed to test the hypothesis that GPR20 is a potential therapeutic target for ADC for the treatment of GIST by assessing GPR20 expression on GIST samples by IHC as well as on normal and malignant tissues, and evaluating antitumor activity of DS-6157a, a GPR20-targeting DXd-ADC, in GIST models, and its pharmacokinetic and safety profiles.

GPR20 was expressed in more than 80% of GIST specimens across all treatment lines. Higher GPR20 expression levels were observed in GIST samples that (i) received multiple treatment lines compared with TKI-naïve samples, (ii) expressed higher KIT levels, (iii) had small intestinal primary origin, and (iv) were WT GIST (no KIT/PDGFRA mutations). The ICCs were the only normal cells positive for GPR20 by IHC. The positive correlation between GPR20 and KIT IHC scores might be consistent with a previous report that both KIT and GPR20 were ICC/GIST lineage-specific genes regulated by FOXF1 and ETV1 (18). Similar gene expression changes by siRNA-mediated downregulation of FOXF1 or ETV1 in GIST cell lines were observed between KIT and GPR20. It is not clear whether the correlation between GPR20 and KIT expression contributes to the functional relationship between both. Comparing growth rates of GIST-T1 and GIST-T1/GPR20 in xenograft models, no significant difference was observed (Supplementary Fig. S6). Different from KIT or PDGFRA, GPR20 may not have an impact on the proliferation of GIST cells. We found that 15/23 (65%) samples with a KIT expression score of zero by IHC expressed GPR20, including five samples (22%) that were strong GPR20-positive, suggesting that there may be some factors either upregulating only GPR20 expression or downregulating only KIT in addition to the transcriptional regulation by the FOXF1/ETV1 axis. GPR20 was highly expressed in small intestinal GIST compared with gastric GIST. In comparison with other types of sarcomas, GISTs have been found to have distinctly homogeneous gene-expression profiles. However, differences in the gene-expression program between gastric and small intestinal GISTs have been described (15). In relation to KIT/PDGFRA mutation status, GPR20 expression level was lower in PDGFRA-mutant GIST compared with KIT-mutant (DFCI and NCCHE) and KIT/PDGFRA WT (DFCI only) GIST. None of the evaluated PDGFRA-mutated samples from DFCI strongly expressed GPR20. The nine samples of PDGFRA-mutant GIST were all gastric GIST, in which KIT expression was low (seven samples were KIT IHC score 0, two were 1+). These genetic and cellular backgrounds may affect low GPR20 expression level in PDGFRA-mutant GIST. In KIT-mutant GIST, TKI-resistant secondary mutations for imatinib, sunitinib, and regorafenib are accumulated in the ATP-binding domain (exon 13/14) and/or activation loop (exon 17/18) in the KIT gene. As intra- and interlesional resistance mutations accumulate in the body of patients, these inhibitors cannot substantially address the molecular basis of the disease and lose efficacy (11). Our IHC study indicated that GPR20 was highly expressed in most GIST harboring these TKI-resistant secondary mutations in KIT, and in KIT/PDGFRA WT GIST samples, suggesting that GPR20 may be a unique and novel target in patients with a broad spectrum of molecular subtypes of GIST. There were higher GPR20 expression levels in GIST with KIT exon 9 mutation in DFCI compared with NCCHE samples. To clarify this, we need to evaluate more samples.

The current work provides a better understanding of GPR20 expression across different molecular subtypes of GIST, which may present hope for patients with either SDH-deficient GISTs or NF1-associated GISTs with no standard approved therapy, or KIT-mutant GISTs that have developed resistance or relapsed after receiving multiple approved TKIs.

In this study, DS-6157a exhibited antitumor activity in GIST xenograft models including GIST harboring TKI-resistant mutations while concurrently demonstrating favorable drug-linker stability and safety profile in animals. Although DS-6157a showed in vitro ADCC activity, the antitumor activity of DS-6157a observed in tumor-bearing mouse models was thought to depend on the delivery of the DXd, because the parent anti-GPR20 antibody did not affect tumor growth in GIST xenograft models as shown in Fig. 5E. Even though contribution of the ADCC activity of DS-6157a in patients remains to be elucidated, we speculate that the targeted delivery of DXd payload is the main mechanism of action for the antitumor activity of DS-6157a, considering its antibody is not an ADCC-enhanced one.

IHC studies indicated that expression of GPR20 in normal tissues is limited to the ICCs, which function as a pacemaker in peristaltic motion of gut. The possible side effect of targeting ICCs is severe GI motor disorders. In preclinical toxicology studies in monkey, however, on-target toxicity for the ICC was not observed, whereas off-target toxicities were observed (changes were minimal or mild). For example, single-cell necrosis of crypts in the intestines and/or thickening and pigment deposition of epidermis in the skin was observed at >3 mg/kg of DS-6157a. These findings indicate that on-target toxicity is not solely by target expression. The lack of toxicity on the ICC compared with crypt cells may likely be due to the very low proliferation or low uptake of payload in the normal ICC. LOP628, a KIT-targeted ADC, exhibited antitumor activity in preclinical tumor models, including GIST cell lines (38). However, hypersensitivity reactions (HSR) were observed in all three patients dosed in a phase I clinical trial treated with LOP628 (39). Mast cell degranulation was implicated as the root cause for the HSR, because mast cells express KIT protein on the cell surface. In our IHC study, mast cells expressed KIT but not GPR20 (Fig. 2G). Therefore, it is not likely that there is concern about such adverse effects caused by DS-6157a dosing.

Our study has some limitations: the IHC study on clinical GIST samples reflected, to some extent, the distribution of driver mutations in the general population. The sample number limitation restricted the detailed characterization of some rare GIST subpopulations. None of our samples harbored mutations in RAS/RAF or PIK3CA pathways or had NTRK fusions. Future studies expanding into these subtypes are required to explore GPR20 expression in mutations that are not currently studied. Moreover, it was difficult to determine the threshold of GPR20 expression level for the efficacy, because the numbers of GIST cell lines and GIST CDX/PDX models are limited, which reflects the available resources for this tumor.

Taken together, these data support the clinical development of DS-6157a as a potential novel GIST therapy with activity in patients who are resistant, refractory, or intolerant to approved TKIs.

DNA immunization with in vivo electroporation was conducted to generate the anti-human GPR20 antibodies. Briefly, WKY/Izm rats (Japan SLC) were immunized biweekly six times with intramuscular injection of a human GPR20 expression vector followed by in vivo electroporation using ECM830 Square Wave Electroporation System and BTX531 Two-Needle Array Electrodes (BTX). Rats with high antibody titers were boosted by injection of 293T cells transiently expressing human GPR20 into muscle three days before harvesting cells from their lymph nodes. Hybridoma fusion with SP2/0-ag14 mouse myeloma cells was conducted by electrofusion using Hybrimune Hybridoma Production Systems (Cyto Pulse Sciences). Hybridoma supernatants that displayed selective reactivity with human GPR20 were screened by Cell ELISA using 293α/GPR20 cells. Clones that tested positive by flow cytometry were then tested for binding to cynomolgus monkey GPR20, antibody internalization activity using Rat-ZAP assay (ADVANCED TARGETING SYSTEMS), and binding assay against synthetic peptides consisted of N-terminal 48 amino acids of human GPR20.

Rat anti-GPR20 antibodies were purified from the culture supernatant of each hybridoma. The cDNA coding rat 04-093 mAb was sequenced and converted into a full-size rabbit IgG antibody, designated 04-093OcH1L1. The 04-093OcH1L1 antibody was produced in FreeStyle 293-F cells (Thermo Fisher Scientific) and purified by Protein-A chromatography.

The studies using human samples were conducted in accordance with ethical guidelines based on the ethical principles as defined in the Declaration of Helsinki and approved by each institutional review board (IRB). Clinical samples from patients with GIST and other sarcomas were obtained according to each IRB protocol, and FFPE tissue samples were banked, and TMAs were generated. Written informed consent was obtained from all patients or patients' guardians, except for patients with GIST whose samples were collected in NCCHE before April 2005. The utilization of these samples was announced to the public by each IRB, in accordance with the ethical guidelines. All GISTs and other sarcomas were pathologically reviewed and confirmed by sarcoma experts. Detailed sample information and analyses are described in Supplementary Materials and Methods.

IHC was performed using Autostainer Link 48 (DAKO) and PT Link (DAKO) at room temperature, unless otherwise stated. All of the primary antibodies were diluted with the DAKO antibody diluent. FFPE sections were deparaffinized and pretreated with Envision FLEX TRS High for 40 minutes at 97°C (GPR20) or with Envision FLEX TRS Low for 20 minutes at 97°C (c-Kit) followed by endogenous peroxidase blocking and protein blocking. Then, the sections were sequentially incubated with monoclonal rabbit anti-GPR20 (04-093OcH1L1, Daiichi Sankyo, final 1.0 μg/mL) or polyclonal rabbit anti-Human CD117, c-Kit (DAKO, final 30.5 μg/mL) for 60 minutes, EnVision+ system-HRP labeled polymer anti-rabbit as a secondary antibody for 20 minutes, and Liquid DAB+ (DAKO) for 10 minutes. Washes were performed after every step. Finally, the sections were counterstained with EnVision FLEX hematoxylin for 10 minutes and mounted. IHC specimens were analyzed under a masked condition according to the following scoring criteria: 3+, ≥30% tumor cells with strong membrane/cytoplasmic staining; 2+, ≥30% tumor cells with moderate or strong membrane/cytoplasmic staining, but <30% strong staining; 1+, ≥30% tumor cells with weak or higher membrane/cytoplasmic staining, but <30% moderate or strong staining; 0, no staining, or <30% tumor cells with membrane/cytoplasmic staining of any intensity (Fig. 2A). IHC specimens were analyzed and scored under a blinded condition according to the scoring criteria described in Methods.

The rat anti-GPR20 mAb 04-046 was converted into a full-size human IgG1, and its stable CHO producer cell line was generated. The expressed anti-human GPR20 antibody was collected from culture supernatants of the recombinant CHO producer cells grown in serum-free medium and purified. DS-6157a was synthesized in accordance with the published procedure (22), conjugating the DXd payload with the humanized anti-GPR20 antibody. The drug distribution was analyzed by hydrophobic interaction chromatography. Control IgG-ADC was synthesized using a human IgG1 isotype control mAb and the same drug linker as DS-6157a, resulting in a comparable DAR.

The GIST-T1 (RRID:CVCL_4976) cell line established by Dr. Takahiro Taguchi (Kochi University) was purchased from Cosmo Bio in 2017. The GIST-T1 cell line was maintained in DMEM supplemented with 10% heat-inactivated FBS. The GIST-T1/GPR20 cells (GIST-T1-derived) were established using Retro-X Universal Packaging System (Takara Bio) with a human GPR20-expressing retroviral vector pQCXIP-hGPR20 following the manufacturer's instructions. The human gastric carcinoma cell line NCI-N87 (RRID:CVCL_1603), human Burkitt lymphoma Ramos (RRID:CVCL_0597), and CHO-K1 cells were purchased from American Type Culture Collection. The GIST430/654 cell lines were obtained from Dr. Jonathan A. Fletcher (Brigham Women's Hospital) in 2013 and maintained in Iscove's Modified Dulbecco's Medium supplemented with 15% heat-inactivated FBS and 100 nmol/L of imatinib mesylate. All cell lines were cultured at 37°C and 5% CO₂ atmosphere. The GIST1 and GIST6 PDX models were provided from Dr. Makoto Moriyama (University of Toyama) in 2013. The GIST#1001 and GIST#1338 PDX models were obtained from Dr. Taisei Nomura (National Institutes of Biomedical Innovation, Health and Nutrition) in 2017. GIST-T1 and GIST430/654 cell lines have been authenticated for KIT mutations by DNA sequencing. GIST models except for GIST-T1 were tested negative for Mycoplasma contamination by Nucleic-acid Amplification Test, in 2014 (GIST430/654), 2015 (GIST1 and GIST6), 2017 (GIST#1338), and 2019 (GIST#1001). GIST-T1 was tested negative for Mycoplasma contamination according to the supplier's information. All cell lines except for 293T and 293α were used for the experiments within 3 to 15 passages after collection.

The binding domain of rat anti-GPR20 antibodies and species cross-reactivity of DS-6157a were evaluated by Cell-ELISA using 293α (HEK293 cells stably expressing integrin αvβ3) or CHO-K1 cells, respectively. For the CHO-K1, cells were transfected with expression plasmids encoding N-terminal FLAG-tagged GPR20 orthologs and a mock plasmid as a negative control.

GIST-T1/GPR20 and NCI-N87 cells were seeded to a 96-well plate at 2,000 cells/well. After overnight culture, each diluted test substance was added. Cell viability was measured after 8 days using a CellTiter-Glo Reagent (Promega). GPR20 expression of each cell was determined by flow cytometry using DS-6157a and human IgG1 isotype control (Eureka Therapeutics) as primary antibody and PE-conjugated F(ab')2 Fragment Goat Anti-Human IgG, Fcγ fragment specific (Jackson ImmunoResearch Laboratories) as secondary antibody.

GIST-T1/GPR20 cells were treated with each substance. After 72 hours, the cells were harvested and lysed with RIPA buffer (Sigma-Aldrich Co. LLC.) containing Halt Protease and Phosphatase Inhibitor Cocktail (Thermo Fisher Scientific). For GIST-T1 xenograft, tumors were disrupted in the RIPA buffer using ShakeMaster AUTO (Biomedical Science). After incubation on ice for 30 minutes, the cell or tumor lysates were collected after centrifugation at 15,000 rpm for 10 minutes at 4°C. Then, tumor lysates were homogenized using QIA shredder (QIAGEN). The protein concentration was quantitated using the BCA Protein Assay Kit (Pierce). Immunoblots were independently performed using the Simple Western system (ProteinSimple Japan K.K.). The following primary antibodies were used: anti-cleaved PARP [Asp214; Cell Signaling Technology (CST); #9541], anti-phospho-CHK1 (Ser317; D12H3; CST; #12302), anti-β-actin (8H10D10;CST; #3700), anti-CHK1 (G4; Santa Cruz Biotechnology; SC8408), anti-phospho-KAP1 (Ser824; Novus Bio; NB100-2350), anti-KAP (821912; R&D Systems; MAB7785), anti-γ(p-)H2AX (Ser139; CST; #2577), and anti-H2AX (CST; #7631). Either anti-mouse or anti-rabbit secondary antibody (Protein Simple Japan) was used for detection.

DS-6157a and IgG-ADC were labeled with a pH-sensitive dye, pHrodo, using pHrodo iFL Red Microscale Protein Labeling Kit (Thermo Fisher) by incubation for 1 hour in 10 mmol/L acetate/5% sorbitol, pH5.5 (Nacalai), and then purified with zeba-spin desalting column (Thermo Fisher) and Amicon Ultra-0.5 (Merck-Millipore). Cells were seeded in a 96-well plate (CellCarrier-96 Ultra, PerkinElmer) at 5 × 10⁴ cells/well. After overnight culture, cells were treated with each pHrodo-labeled substance (10 μg/mL), along with Hoechst 33342 (100 ng/mL). Fluorescence emission images were captured every 30 minutes up to 12 hours after treatment by Opera Phenix (Parkin Elmer). Lysosome trafficking activities were expressed by the number of dot per cell (dots/cell) and dot signal intensity, both of which were determined using Harmony analysis software (PerkinElmer). Delta dot signal intensity was calculated as a subtraction of average dot intensity values of control group at t = 0 from each intensity value. Trafficking level was expressed as a trafficking index as follows: Trafficking index = dot/cell × delta dot signal intensity

ADCC activity was evaluated using calcein-acetoxymethyl (calcein-AM; Thermo Fisher Scientific) release assay. We used human PBMCs obtained from donors as effector cells and the GIST-T1/GPR20 cells as target cells. The target cells were labeled with calcein-AM for 1 hour and then washed three times with culture medium. The effector cells (2 × 10⁵ cells) and the calcein-labeled target cells (1 × 10⁴ cells) were incubated with each substance in 96-well U-bottom plate in a humidified 5% CO₂ incubator at 37°C for 4 hours. After centrifugation of the assay plate, the supernatant was transferred and filtrated. The fluorescence intensity of calcein released from the target cells was measured using a microplate reader. ADCC activity (%) at each concentration was calculated by the following equation: ADCC activity (%) = (ER − Control mean)/(MR mean − SR mean) × 100. ER is the fluorescence intensity value of the individual effector release. Control mean, SR mean, and MR mean are the mean fluorescence intensity value of control, spontaneous release, and the maximum release well, respectively.

The GIST-T1/GPR20 and Ramos cell lines were used for DS-6157a and rituximab in CDC assays, respectively. Each tumor cell line was seeded at 5,000 cells/50 μL/well into a 96-well black clear bottom plate and incubated overnight. Then, 25 μL of the diluted ADCs or antibodies were added, and the plates were incubated for 1 hour at 4°C. Human complement was diluted and added at 25 μL/well and the plates were incubated for 1 hour at 37°C in the CO₂ incubator. After incubation, the cellular ATP level was measured using CellTiter-Glo Reagent and a multilabel counter (EnSight, PerkinElmer Japan). The cell viability (%) was calculated as follows. Cell viability (%) = 100 × (T − B)/(C − B), where T is the luminescence intensity of the test well. B and C are the mean luminescence intensity of blank and untreated wells, respectively.

All animal experiments were performed in accordance with the local guidelines of the Institutional Animal Care and Use Committee of Daiichi Sankyo Co., Ltd., and with the approval of the committee.

Briefly, each cell suspension or tumor fragment was inoculated subcutaneously into female nude mice. When the tumor had grown to an appropriate volume, the tumor-bearing mice were randomized into treatment and control groups (6–8 mice per group) based on the tumor volumes, and dosing was started. Antibodies and ADCs were administered intravenously to the mice. TKIs were administered orally. Tumor volume defined as 1/2 × length × width² was measured twice a week. TGI (%) was calculated as follows: TGI (%) =  [1–(mean of treatment group tumor volume on evaluation day)/(mean of control group tumor volume on evaluation day)] × 100.

Concentrations of DS-6157a and the total antibody in plasma were determined with a validated ligand-binding assay; the lower limit of quantitation was 0.200 μg/mL. The concentration of DXd in plasma was determined with a validated liquid chromatography–tandem mass spectrometry method; the lower limit of quantitation was 0.100 ng/mL.

DS-6157a was intravenously administered at three-week intervals over a six-week period to Crl:CD(SD) rats or cynomolgus monkeys. Clinical signs, body weight, food consumption, and clinical pathology were monitored throughout the study. A necropsy was conducted on the day after the third administration. The reversibility of the toxic changes was assessed in a subsequent two-month recovery period.

A phase I, multicenter, open-label, nonrandomized, first-in-human study of DS-6157a in patients with advanced GIST (NCT04276415) was initiated in the United States and Japan to assess the safety, maximum tolerated dose, pharmacokinetics, and preliminary antitumor activity of DS-6157a.

All statistical analysis except for toxicity studies was performed using SAS System Release 9.1.3 and 9.2 (SAS Institute Inc.). Statistical analysis in toxicity studies was performed using MUSCOT (Yukms Co., Ltd.). All IC₅₀ and ED₅₀ values were determined by a Sigmoid Emax model, and dose dependency was evaluated by a Spearman rank correlation coefficient hypothesis test. A Student t test was used to calculate differences in tumor volumes between two groups. Fisher exact test was used to determine whether or not there is a significant association between two categorical variables. GPR20 and KIT IHC data were evaluated by exploratory post hoc analyses.

K. Iida reports a patent for anti-GPR20 antibody and anti-GPR20 antibody–drug conjugate licensed and a patent for anti-GPR20 antibody issued. A.H. Abdelhamid Ahmed received a partial scholarship from Dubai Harvard Foundation for Medical Research to support his master's degree at Harvard Medical School. A.K. Nagatsuma reports personal fees from Chugai Pharmaceutical Co., Ltd. during the conduct of the study, personal fees from Chugai Pharmaceutical Co., Ltd. and Keio University outside the submitted work. M. Abe reports personal fees from Daiichi Sankyo Co., Ltd. outside the submitted work. K. Inaki reports personal fees from Daiichi Sankyo RD Novare during the conduct of the study; personal fees from Daiichi Sankyo RD Novare outside the submitted work. T. Nomura reports grants from Daichi-Sankyo during the conduct of the study. S. George reports other support from Daiichi Sankyo during the conduct of the study; personal fees and other support from Blueprint Medicines and Deciphera, personal fees from Bayer, Eli Lilly, OncLive, ResearchToPractice, Medscape, and NCCN and other support from Merck, Eisai, Springworks, and Wolters Kluwer, outside the submitted work; and Alliance for Clinical Trials in Oncology, Vice-Chair Alliance Foundation, Vice-President. T. Doi reports grants from Lilly, Novartis, Merck Serono, Eisai, IQVIA, Pfizer, grants and personal fees from Taiho, MSD, Janssen Pharma, Boehringer Ingelheim, Sumitomo Dainippon, Daiichi Sankyo, Bristol-Myers Squibb, and AbbVie, and personal fees from Rakuten Medical, Amgen, and Takeda outside the submitted work. A. Ochiai reports grants from Daiichi Sankyo during the conduct of the study. G.D. Demetri reports grants, personal fees, and other support from Daiichi Sankyo during the conduct of the study; personal fees and other support from Translate BIO, grants and nonfinancial support from AbbVie, Janssen, grants from Adaptimmune, grants, personal fees, and other support from Epizyme, grants, personal fees, nonfinancial support, and other support from Bayer, GlaxoSmithKline, Roche/Genentech, PharmaMar, and Pfizer, grants and personal fees from LOXO/Lilly, other support from Bessor Pharmaceuticals, Champions Biotechnology, ERASCA Pharmaceuticals, personal fees and nonfinancial support from Medscape, Novartis, personal fees and other support from Blueprint Medicines, Caprion/HistoGeneX, CARIS Life Sciences, G1 Therapeutics, ICON PLC, M.J. Hennessey/OncLive, McCann Health, RELAY Therapeutics, WCG/Arsenal Capital, personal fees from C4 Therapeutics, EMD-Serono, Mirati, Sanofi, Synlogic outside the submitted work; in addition, G.D. Demetri has a patent for imatinib for GIST issued, licensed, and with royalties paid from Novartis; and is AACR Science Policy and Government Affairs Committee Chair and an Alexandria Real Estate Equities Oncology Summit participant. No disclosures were reported by the other authors.

K. Iida: Conceptualization, resources, formal analysis, validation, investigation, visualization, methodology, writing–original draft, project administration, writing–review and editing. T. Iguchi: Validation, investigation, visualization, writing–original draft, writing–review and editing. T. Karibe: Resources, validation, investigation, visualization, writing–original draft, writing–review and editing.T. Nakada: Resources, investigation, visualization, methodology, writing–review and editing. K. Inaki: Data curation and investigation. R. Kamei: Investigation and visualization. Y. Abe: Conceptualization, supervision, funding acquisition, methodology, and project administration. T. Nomura: Resources, investigation, and methodology. J.L. Andersen: Resources. S. Santagata: Resources, writing–review and editing. M.L. Hemming: Resources, writing–review and editing. A.H. Abdelhamid Ahmed: Conceptualization, resources, data curation, investigation, writing–original draft, writing–review and editing. S. George: Conceptualization, supervision, writing–review and editing. T. Doi: Conceptualization, supervision, and project administration. A. Ochiai: Conceptualization, resources, supervision, methodology, and project administration. G.D. Demetri: Conceptualization, resources, supervision, and project administration. T. Agatsuma: Conceptualization, supervision, funding acquisition, methodology, and project administration. A.K. Nagatsuma: Resources, validation, investigation, and methodology. T. Shibutani: Resources, validation, investigation, visualization, methodology, writing–original draft. S. Yasuda: Data curation, formal analysis, investigation, visualization, writing–original draft. M. Kitamura: Investigation. C. Hattori: Validation, investigation, visualization, writing–review and editing. M. Abe: Resources, investigation, visualization, writing–original draft. J. Hasegawa: Investigation, visualization, writing–review and editing.

The authors thank Takahiro Taguchi (Kochi University, Japan) for providing the GIST-T1 cell line, Jonathan A. Fletcher (Brigham Women's Hospital, MA) for the GIST430/654 cell line, and Makoto Moriyama and Tsutomu Fujii (University of Toyama Japan) for the GIST1 and GIST6 PDX models. The authors thank Luigi Terracciano and Eppenberger-Castori Serenella (University of Basel) for the sarcoma tissue microarray. The authors also thank Tomoko Terauchi, Hiroko Kono, Chigusa Yoshimura, Yoko Nakano, Saori Ishida, Masanori Funabashi, Naoyuki Maeda, Megumi Miyamoto, Shota Yamazaki, Akira Okubo, and Jun Harada (Daiichi Sankyo) for their technical assistance. This study was funded by Daiichi Sankyo Co., Ltd.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.