DS-1471a shows antitumor efficacy in liver cancer PDX models

To investigate the therapeutic potential of DS-1471a for liver cancer, its antitumor efficacy was examined in HCC PDX models. Among the 18 evaluated models in this study, notable antitumor efficacy was observed in 7 models, with tumor growth inhibition (TGI) reaching 50% or higher (Table 1). In one representative model, administration of DS-1471a at 10 mg/kg resulted in significant tumor growth inhibition when analyzed at day 16, with no effect on body weight (Fig. 1a–d). The level of circulating DS-1471a in the blood 2 days after the last dose ranged from 47.6 to 217 μg/mL, and no clear relationship was found between plasma concentration of DS-1471a and TGI (Supplementary Table 3). The antitumor effects of DS-1471a were observed across both sexes, and different tumor grades (TNM stages) and hepatitis virus types (Supplementary Fig. 1a-c). Taken together, these findings support the therapeutic potential of DS-1471a against liver cancer.

Fig. 1

Antitumor activity of DS-1471a in liver cancer PDX models. (a–d) Mice bearing HCC PDX tumors were intravenously administered vehicle (PBS) or DS-1471a at 10 mg/kg (N = 6, each group). (a) Tumor volume analysis of HCC PDX model #11. Each value represents the mean tumor volume ± SE at the given timepoints. (b) Representative images of vehicle- and DS-1471a-treated tumors. (c) Analysis of tumor weights from vehicle- and DS-1471a-treated mice upon sacrifice. Each value represents the mean ± SE. For two-group comparisons, Student’s t-test was used. *, p < 0.05. (d) Body weights of HCC PDX model mice #11 treated with vehicle and DS-1471a

Correlation between DS-1471a efficacy and protein expression levels of biomarker candidates in liver cancer PDX models

To identify molecules that affect DS-1471a antitumor effects, correlations between the expression levels of multiple proteins and DS-1471a efficacy were examined in the HCC PDX models. We first focused on CD147 as the target of DS-1471a. Expression levels of CD147 protein showed a significant correlation with the efficacy of DS-1471a (r = 0.7832, p < 0.0001) (Fig. 2a). Subsequently, immunohistochemical (IHC) staining for CD147 was also performed using the same tumor tissues from the 18 models to further examine localization and expression of CD147 within the tumors, and expression levels in the PDX models were evaluated via quantification of the IHC signal intensity and expressed as the CD147 H-score. CD147 staining was mainly observed on the tumor cell membrane (Fig. 2b). CD147 H-scores were well correlated with the CD147 protein levels determined by capillary electrophoresis (Supplementary Fig.  1 d; r = 0.7079, p = 0.0022), and a significant correlation was observed between CD147 H-score and DS-1471a efficacy (Fig. 2b; r = 0.7395, p = 0.0011). In addition to CD147, further analyses were performed for candidate proteins related to CD147 regulation and DS-1471a’s proposed mechanisms of action. In a previous study, expression levels of CD147, SMAD4, and KLF5 showed good correlation with DS-1471a efficacy [18]. In this study, we added additional factors, including FBXO22, ARF6, and HOOK1, which are related to intracellular degradation and trafficking of CD147 [15, 17, 22, 26]; AFP, which is implicated in the differentiation of hepatocellular carcinoma [24, 25]; and S100A4 and ANXA2 [20, 21], which interact with the CD147 complex (Supplementary Fig. 1e). Among these proteins, CD147 expression showed the strongest correlation with DS-1471a efficacy, while expression of SMAD4, AFP, ARF6, and FBXO22 also showed a significant correlation (Fig. 2a). These results raise the possibility that CD147 and multiple CD147-related factors could be involved in the efficacy of DS-1471a in liver cancer tumor models.

Fig. 2

Correlation between antitumor effects of DS-1471a and protein expression levels of biomarker candidates in liver cancer PDX tumor samples. (a) Correlation diagram of DS-1471a-mediated tumor growth inhibition (TGI) (%) and protein expression levels of CD147, KLF5, SMAD4, AFP, HOOK1, ARF6, and FBXO22. Original data source images are presented in Supplementary Fig. 10. (b) Immunohistochemical analysis of CD147 and its correlation with the antitumor effects of DS-1471a. (left) Representative image of CD147 immunohistochemical staining of HCC tumor samples. (right) Correlation between CD147 H-score and TGI in HCC PDX models. H-score was calculated based on membrane CD147 IHC staining intensity using the following formula: H-score = 3 × (% of strongly positive tumor cells) + 2 × (% of moderately positive tumor cells) + 1 × (% of weakly positive tumor cells)

Analysis of mechanisms controlling CD147 protein levels in liver cancer cells

Given that CD147-related proteins such as FBXO22, HOOK1, and ARF6 have been reported to contribute to the degradation and recycling of CD147 [15, 17, 22, 26], their decrease may lead to increased expression of CD147 in liver cancer cells. To examine the possible involvement of these candidate proteins in CD147 expression, degradation and recycling in liver cancer cells, siRNA-mediated knockdown experiments were performed with multiple cell lines. Six days after transfection, gene knockdown was confirmed (Supplementary Figs. 2a, 3a and 4a), and expression levels of CD147 and the other related markers were examined. In contrast to the previous studies, knockdown of AFP, FBXO22, S100A4, ANXA2, and HOOK1 exerted no substantial influence on total CD147 protein abundance (Supplementary Fig. 2b, 3b, and 4b).

To further examine the possible involvement of the above candidate proteins, liver cancer cell lines overexpressing each protein were generated and the effects on CD147 expression were examined. Although overexpression of FBXO22, HOOK1, and ARF6 was expected to decrease CD147 expression by enhancing the recycling system, overexpression of these proteins exerted no substantial influence on total CD147 protein abundance in liver cancer cell lines (Supplementary Fig. 5a and 5b). Similar results were obtained for other proteins such as S100A4 and AFP, consistent with the results of knockdown experiments (Supplementary Fig. 2–4). Taken together, these results indicate that, under the experimental conditions tested in the present study including the selected liver cancer cell lines and the steady state total protein measurements employed, overexpression or knockdown of FBXO22, HOOK1, S100A4, AFP, and ARF6 did not substantially alter total CD147 protein abundance. These findings do not exclude the possibility that these proteins could regulate CD147 through other biological mechanisms.

Increased FBXO22 enhances antitumor efficacy of DS-1471a in liver cancer models

Although CD147 protein level does not seem to be regulated by the examined factors in vitro (Supplementary Figs. 2–5), correlations between DS-1471a efficacy in liver cancer PDX models and expression levels of several proteins such as FBXO22 and ARF6 were observed. This prompted us to investigate their possible involvement in the efficacy of DS-1471a. To examine this issue, cell lines overexpressing multiple CD147-related proteins were generated and the impacts of increased expression of the candidate proteins on DS-1471a efficacy were examined. Among the tumor models overexpressing FBXO22, HOOK1, ARF6, and S100A4, only the FBXO22-overexpressing model showed enhanced antitumor effects with DS-1471a treatment (Fig. 3 and Supplementary Fig. 5a and 5c). This enhanced efficacy of DS-1471a in tumors overexpressing FBXO22 is consistent with the observed correlation between FBXO22 level and DS-1471a efficacy in liver cancer PDX tumor models (Fig. 2a). Taken together, these results highlight the potential contribution of FBXO22 to the antitumor efficacy of DS-1471a in liver cancer tumor models.

Fig. 3

Antitumor efficacy of DS-1471a in liver cancer tumor models overexpressing FBXO22 or ARF6. Mice bearing SK-HEP-1 tumors overexpressing FBXO22 or ARF6 were treated with vehicle or 10 mg/kg DS-1471a on days 7, 14, and 21 after inoculation. Graphs show the mean tumor volume ± SE at the given timepoints

Potential involvement of FBXO22 and p21 in DS-1471a antitumor activity in liver cancer models

Since overexpression of FBXO22 enhances DS-1471a efficacy in liver tumors, we next examined the possible involvement of FBXO22 in regulating CD147 protein levels in tumor xenograft models. FBXO22 is highly expressed in liver cancer and has been reported to regulate the degradation of intracellular proteins including CD147 [17, 23, 30]. However, overexpression of FBXO22 did not affect CD147 protein level in the liver cancer tumor models tested (Fig. 4a). We therefore investigated a different target of FBXO22, the cyclin-dependent kinase inhibitor p21, which has been reported to be a substrate of FBXO22 [31], and found decreased levels in tumors overexpressing FBXO22 (Fig. 4a). These results led us to hypothesize that p21 could be an important factor regulating DS-1471a efficacy in liver cancer. To test this hypothesis, we generated CDKN1A/p21-knockout (KO) cells (Fig. 4b) and examined the efficacy of DS-1471a in two different p21-KO tumor models. Treatment of p21-KO tumors with DS-1471a resulted in remarkably enhanced antitumor effects when compared with the limited effects observed in control parental tumors (Fig. 4c). Based on these correlational and phenocopy data, the above results raise the possibility that reduced p21 levels, potentially mediated by FBXO22, are associated with enhanced antitumor efficacy of DS-1471a in liver cancer.

Fig. 4

Correlation between FBXO22-p21 pathway and the antitumor effects of DS-1471a in liver cancer models. (a) CD147 and p21 protein expression in SK-HEP-1 tumors overexpressing FBXO22. Each lane represents an individual tumor from a separate animal (n = 2 per group), and the data are representative of two independent experiments. GAPDH served as a loading control. (b) Loss of p21 expression in p21-KO SK-HEP-1 cells. Images show Simple Western analysis with anti-CD147, p21, and β-actin. Data are representative of two independent experiments. (c) Antitumor activity of DS-1471a in p21-KO SK-HEP-1 tumors. Graphs show the mean tumor volume ± SE at the given timepoints. (d) Increased expression of p21 in DS-1471a-resistant tumors. Tumors from untreated or 1 mg/kg DS-1471a-treated mice were subjected to Simple Western analysis with anti-p21, CD147, and GAPDH. Each lane represents an individual tumor from a separate animal, and the data are representative of two independent experiments. Original uncropped images of Figs. 4a, 4b, and 4 d are presented in Supplementary Fig. 10. Image processing (brightness and contrast adjustments) was applied uniformly to the entire image and not selectively to individual lanes or bands. (e) Correlation diagram of p21 protein expression and DS-1471a-mediated TGI (%) in the HCC PDX model. Original data source images are presented in Supplementary Fig. 10

To further understand the mechanisms regulating the efficacy of DS-1471a, we investigated the mechanism of resistance (MoR) to DS-1471a. For this purpose, we examined whether DS-1471a-resistant tumors could be generated by repeated in vivo administration of DS-1471a (Supplementary Fig. 7). When DS-1471a was administered at a dose of 3 mg/kg, complete regression was achieved by repeating the administration once a week for two doses, with no tumor regrowth observed for at least 100 days. However, when DS-1471a was administered at a dose of 1 mg/kg, resistant tumors were observed in 3 cases, while complete regression was observed in 9 cases (Supplementary Fig. 7). Among the three resistant tumors, one showed a slight reduction in CD147 level, while the other two showed elevated p21 protein levels (Fig. 4d). These results imply that p21 upregulation may represent a potential mechanism of resistance to DS-1471a in some tumors. However, given the small number of resistant tumors and the non-uniform molecular profiles observed, these findings should be considered preliminary and hypothesis-generating, and will require further validation and comprehensive molecular characterization. Since p21 is reported to contribute to mechanisms of resistance to various drugs [32], we further examined the relationship between p21 expression and DS-1471a efficacy in the HCC PDX models. Supporting the above hypothesis, an inverse correlation between p21 expression and DS-1471a efficacy was observed (Fig. 4e). Taken together, these results provide supportive evidence that FBXO22 and p21 expression levels could be key factors influencing the therapeutic effects of DS-1471a in CD147-targeted cancer therapy.