The effects of T-DXd on the expression of HLA class I and chemokines CXCL9/10/11 in HER2-overexpressing gastric cancer cells

Trastuzumab deruxtecan (T-DXd), a HER2-targeting antibody–drug conjugate with a topoisomerase I inhibitor deruxtecan (DXd), exhibits an excellent anti-tumor effect in previously treated HER2-positive tumors. A recent study demonstrated that T-DXd not only suppressed tumor growth but also enhanced anti-tumor immunity through increasing the number of tumor-infiltrating CD8+ T cells and enhancement of major-histocompatibility-complex class I expression on tumor cells in a mouse model. However, the effect of T-DXd on anti-tumor immune responses in human cancers is largely unknown. We investigated the effect of T-DXd on the expression of HLA class I and CXCL9/10/11, T-cell chemoattractants, in HER2-positive human gastric cancer (GC) cells. We found that T-DXd significantly inhibited GC cell proliferation in a HER2-dependent manner, while it slightly increased the expression of HLA class I in HER2-positive GC cells. Moreover, we revealed that T-DXd significantly induced mRNA expression of CXCL9/10/11 in HER2-positive GC cells. T-DXd-triggered up-regulation of these chemokines was mediated through the activation of DNA damage signaling pathways. These results suggest that T-DXd triggers anti-tumor immune responses at least in part through induction of the expression of HLA class I and CXCL9/10/11 on HER2-positive GC cells, resulting in the enhancement of anti-tumor immunity in human GC.


HER2-dependent inhibition of GC cell proliferation by T-DXd.
To first assess the effect of T-DXd on cell proliferation of HER2-positive and HER2-negative GC cells, three HER2-amplified GC cell lines (NCI-N87, OE19, and MKN7) and two HER2-non-amplified GC cell lines (AGS and NUGC3) were used in this study (Fig. 1A). Although deep deletions (Deep del) and missense mutations (Missense) in the molecules of HER2 signaling including MAPK and AKT signaling pathways have been reported in the five GC cell lines, their biological significances (oncogenicities) are unknown (Fig. 1A). Cell surface overexpression of HER2 was confirmed by flow cytometry in NCI-N87, OE19, and MKN7 cells but not in AGS and NUGC3 cells (Fig. 1B). Using these five GC cell lines, we tested cell growth inhibitory activity by T-DXd in vitro. As shown in Fig. 1C, concentrations of more than 0.1 μg/ml T-DXd significantly suppressed cell proliferation of HER2-positive NCI-N87 and OE19 cells, and even in MKN7 cells, which are known as trastuzumab-resistant HER2-positive GC cells. On the other hand, T-DXd did not inhibit cell proliferation of HER2-negative AGS and NUGC3 cells (Fig. 1C). However, 10 μg/ml T-DXd markedly suppressed cell proliferation in both HER2-positive and HER2negative GC cells (Fig. 1C). A previous report demonstrated that the higher concentration (10 μg/ml) of control IgG-ADC-conjugated with DXd had cell growth inhibition activity in several cancer cell lines, and the antibodyindependent cytotoxicity occurred at higher concentrations of ADC-conjugated with DXd 28 . Based on the previous report and our present result, we strongly suggested that HER2-independent cell growth inhibition might have occurred in HER2-negative GC cell lines treated with 10 μg/ml T-DXd.

Up-regulation of HLA class I expression by T-DXd in HER2-positive GC cells. To next examine
the effect of T-DXd on the expression of HLA class I in HER2-positive GC cells, NCI-N87, OE-19, and MKN7 cells were treated with several concentrations of T-DXd and subjected to flow cytometry and western blot of HLA class I as well as HER2. The cell surface expression of HER2 was increased in NCI-N87 and OE19 cells by T-DXd, but not trastuzumab ( Fig. 2A and Supplementary Fig. S1), while it was dramatically decreased by T-DXd and trastuzumab in MKN7 cells ( Fig. 2A and Supplementary Fig. S1 www.nature.com/scientificreports/ expression of HER2 might be one of the main mechanisms of resistance to trastuzumab in MKN7 cells due to low binding rate of trastuzumab to the cells. Although IFN-γ markedly increased both cell surface and protein expressions of HLA class I in HER2-positive GC cells ( Fig. 2A,B), T-DXd caused a slight increase of cell surface expression of HLA class I ( Fig. 2A). Because increased expression of HLA class I by T-DXd was not observed by western blot analysis (Fig. 2B),  Fig. S2). These results suggest that T-DXd slightly increases cell surface expression of HLA class I in HER2-positive GC cells via the effect of DXd. We also examined the effect of T-DXd on cell surface expression of HER2 and HLA class I in HER2-negative NUGC3 and AGS cells. As a result, the relative percentages of HER2 expression were slightly increased at 10 μg/ ml T-DXd in NUGC3 cells or decreased at 0.1-10 μg/ml T-DXd in AGS cells ( Supplementary Fig. S3, left). However, because the basal expression levels of HER2 were very low in both cell lines, T-DXd might have little effect on HER2 expression in NUGC3 and AGS cells. In addition, the expression of HLA class I in AGS cells was not increased by T-DXd, while only 10 μg/ml T-DXd slightly increased cell surface expression of HLA class I in NUGC3 cells ( Supplementary Fig. S3, right), which might depend on HER2-independent cytotoxic effects of T-DXd. Because a previous report suggested that AGS cells were defective in MHC class I inducibility compared with other human cell lines 29 , the different expression of HLA class I by T-DXd might be observed between the two cell lines.
Induction of mRNA expression of T-cell chemoattractants CXCL9/10/11 by T-DXd. In the previous report, T-DXd increased the number of tumor-infiltrating CD8 + T cells in a mouse model 6 . We, therefore, tested whether T-DXd triggered the expression of T-cell chemoattractants CXCL9/10/11 in HER2-positive GC cells. T-DXd significantly induced mRNA expression of CXCL9/10/11 in a dose-and time-dependent manner in NCI-N87 cells (Fig. 3A,B). We found that mRNA expression of CXCL9/10/11 induced by T-DXd was significantly higher than that induced by Trastuzumab (Fig. 3C). www.nature.com/scientificreports/ We also investigated the underlying mechanism of how T-DXd up-regulated mRNA expression of CXCL9/10/11 in HER2-positive GC cells. Because trastuzumab slightly increased mRNA expression of CXCL9/10/11 in HER2-positive GC cells (Fig. 3C), there is probably a link between HER2 activity and the expression of CXCL9/10/11. We, therefore, examined the association between HER2 level and CXCL9/10/11 expression in GC tissues. TCGA dataset showed that mRNA expression of CXCL9/10/11 was decreased in HER2-amplified GC tissues as compared with that in HER2-non amplified GC tissues (Fig. 4A), suggesting that mRNA expression of CXCL9/10/11 might be inversely associated with the activity of HER2 signaling in GC. However, our present data suggest that T-DXd has little effect on the downstream signaling of HER2 including Akt and ERK pathways (Fig. 2B). Thus, additional mechanisms must exist to up-regulate CXCL9/10/11 by T-DXd. Previous reports suggested that anti-cancer drugs such as cisplatin and hydroxyurea induced mRNA expression of CXCL9/10/11 through cell cycle-specific DNA damage in human cancer cells 26,27 , and the DNA topoisomerase I inhibitor, irinotecan, is also known to cause cell cycle-specific DNA damage 16,30 . Indeed, we found that irinotecan significantly induced mRNA expression of CXCL9/10/11 in NCI-N87 cells (Supplementary Fig. S4). Because DNA damage-induced checkpoint activation is partially regulated by ataxia telangiectasia-and-rad3-related (ATR) and ataxia-telangiectasia mutated (ATM) pathways 31 , we also examined the involvement of DNA damage signalings such as ATR and ATM pathways in the induction of mRNA expression of CXCL9/10/11 by T-DXd in HER2positive GC cells. As a result, we found that weak but significant correlations exist between mRNA expression of the signaling molecules involved in ATR and ATM pathways such as chk1 (CHEK1) and chk2 (CHEK2), and chemokines CXCL9/10/11 in GC tissues (Fig. 4B). We also demonstrated that T-DXd-triggered mRNA expression of CXCL9/10/11 was significantly attenuated by an ATM inhibitor (KU), but not ATR inhibitor (VE) and chk1 inhibitor (UCN) in HER2-positive GC cells (Fig. 4C), suggesting that T-DXd increased mRNA expression of CXCL9/10/11 through the activation of ATM-but not ATR/chk1-mediated DNA damage signaling pathway in HER2-positive GC cells. These results suggest that T-DXd might increase mRNA expression of CXCL9/10/11 through DXd-mediated activation of DNA damage signaling in HER2-positive GC cells.
Because T-DXd increased the expression of CXCL9/10/11 in HER2-positive GC cells, recruited immune cells by the chemokines might also affect HLA class I expression on GC cells through the production of cytokines such as IFN-γ. Therefore, we finally examined the effect of immune cells including T cells on HLA class I expression of T-DXd-treated HER2-positive GC cells. NCI-N87 cells were co-cultured with or without human peripheral blood mononuclear cells (PBMC) in the absence or presence of T-DXd and subjected to flow cytometry to analyze the expression of HLA class I on NCI-N87 cells. The cell surface expression of HLA class I on NCI-N87 cells was slightly increased by the treatment with T-DXd or the co-culture with PBMC, while T-DXd dramatically enhanced the expression of HLA class I on NCI-N87 cells co-cultured with PBMC (Fig. 5A), suggesting that immune cells including T cells further enhance the expression of HLA class I on HER2-positive GC cells in the presence of T-DXd. We also found that the level of IFN-γ production was not changed in NCI-N87 cells by the treatment with T-DXd or co-cultured with PBMC, while it was significantly increased in NCI-N87 cells co-cultured with PBMC in the presence of T-DXd (Fig. 5B). This result suggests that IFN-γ might be one of the crucial factors that is involved in the up-regulation of HLA class I expression in NCI-N87 cells co-cultured with PBMC in the presence of T-DXd.
Taken together, our data suggest that T-DXd slightly enhances the expression of HLA class I in HER2-positive GC cells, and it also increases the expression of CXCL9/10/11 through the activation of ATM-mediated DNA damage signaling pathway in HER2-positive GC cells.

Discussion
In the current study, we identify for the first time the effect of T-DXd on the expressions of HLA class I and chemokines, CXCL9/10/11, in HER2-positive GC cells. Although T-DXd slightly up-regulated HLA class I expression, it markedly increased mRNA expression of CXCL9/10/11 in HER2-positive GC cells. We also found that T-DXd-induced mRNA expression of CXCL9/10/11 was mediated through the activation of the DNA damage signaling pathway regulated by ATM and not ATR or Chk1 in HER2-positive GC cells.
We found that T-DXd inhibited cell proliferation of not only trastuzumab-sensitive HER2-positive GC cells (NCI-N87 and OE19 cells) but also trastuzumab-resistant HER2-positive GC cells (MKN7 cells) (Fig. 1C). Because trastuzumab markedly decreased the expression of HER2 in MKN7 cells ( Fig. 2 and Supplementary  Fig. S1), downregulated HER2 expression and lower binding of trastuzumab to MKN7 cells might be the main mechanism of resistance to trastuzumab. However, T-DXd significantly suppressed cell proliferation of MKN7 cells even in such trastuzumab-resistant HER-2 positive GC cells. Indeed, Ogitani et al. revealed that T-DXd has the potential to deliver sufficient DXd into cancer cells regardless of HER2 levels, that is, T-DXd can inhibit cell proliferation of not only HER2-strong positive cells but also HER2-weak positive cells. Therefore, T-DXd significantly suppressed cell proliferation of HER2-downregulated MKN7 cells 1 , which is consistent with the previous report that T-DXd has a clinical benefit for HER2-positive patients who had failed to trastuzumab therapy 5 .
In the present study, moderately increased expression of HLA class I by T-DXd was observed in HER2-positive GC cells. Although we evaluated the effect of T-DXd on the expression of HLA class I in HER2-positive GC cells by both flow cytometry and western blot analyses, we could only detect a slight increase in cell surface expression of HLA class I (Fig. 2) At present, molecular mechanisms by which T-DXd up-regulates cell surface expression of HER2 in NCI-N87 and OE19 cells are largely unknown. Because T-DXd and irinotecan but not trastuzumab significantly increased HER2 expression (Fig. 2A, Supplementary Figs. S1, S2), the cytotoxic effect of topoisomerase I inhibitor such as DNA strand breaks might be involved in the up-regulation of cell surface expression of HER2 in NCI-N87 and OE19 cells. ATM kinase plays a central role in sensing DNA double-stranded breaks and coordinating their repair, and DXd activates DNA-damage response pathways including ATM pathway 1,28 . Interestingly, a recent report suggested that ATM functions as a modulator of HER2 receptor levels and stability. Indeed, suppression of ATM kinase activity by a specific inhibitor significantly reduced HER2 receptor levels in vivo 32 . Therefore, T-DXd possibly up-regulates cell surface HER2 levels in NCI-N87 and OE19 cells through the activation of the ATM pathway.
Our data suggest that the ATM-mediated DNA damage signaling pathway might be involved in T-DXdinduced mRNA expression of CXCL9/10/11 in HER2-positive GC cells, because an ATM inhibitor (KU-55933), but not ATR and Chk1 inhibitors (VE-821 and UCN-01), selectively inhibited T-DXd-induced mRNA expression of CXCL9/10/11 (Fig. 4B). ATM and its downstream molecule Chk2 are the primary kinases responsible for G1/S phase arrest of the cell cycle 33,34 , and S-phase-specific DNA damage has been reported to activate the immune response including induction of CXCL9, -10 expression in cancer cells 26,27 . T-DXd contains the DNA topoisomerase I inhibitor (DXd), and DNA topoisomerase I inhibitors such as irinotecan act on the S and G2 phases of the cell cycles 16,30 . Therefore, T-DXd especially DXd might cause S-phase-specific DNA damage and consequently trigger mRNA expression of CXCL9/10/11 in HER2-positive GC cells.
In summary, our present study revealed that T-DXd increased surface expression of HLA class I and upregulated T cell chemoattractants, CXCL9/10/11, in HER2-positive GC cells. These results suggest a possibility that CXCL9/10/11 released from HER2-positive GC cells by T-DXd might attract tumor-infiltrating lymphocytes, which produce a large amount of IFN-γ 35,36 , leading to activation of immune cells and further enhancement of HLA class I expression on cancer cells (Fig. 5 and Supplementary Fig. S5). Therefore, T-DXd might enhance anti-tumor immune responses in human GC. However, in the current study, we performed only in vitro experiments. Further investigations using GC tissues from clinical trials of T-DXd are required.  . For two-group comparisons, statistical analyses were performed using the unpaired t-test. For multigroup comparisons, we applied one-way ANOVA with post hoc Tukey-Kramer test. The Spearman correlation test was used to analyze the association in each experiment. A value of p < 0.05 was considered to be significant.