Targeting claudin-4 enhances chemosensitivity of triple negative breast cancer

Background Triple negative breast cancer (TNBC) possesses highly aggressive phenotype, treatment with limited options, and a poor prognosis. In this study, we examined the therapeutic effect of anti-claudin (CLDN)-4 extracellular domain antibody, 4D3, on TNBC. Methods The expression of CLDN4 and CLDN1 in invasive ductal carcinoma (IDC) was examined in 78 IDCs (from 2004 to 2009 in a single center). CLDN expression and the effect of 4D3 on proliferation were examined in in human IDC cell lines MCF-7 (luminal subtype) and MDA-468 (TNBC). Results In IDC cases, CLDN1 had lower expression than CLDN4 and correlated with histological grade. In contrast, expression of CLDN4 correlated with histological grade, receptor subtype, and stage. CLDN4 expression in the two cells was at the same level. In both cells, paclitaxel (PTX)-induced growth suppression was enhanced by 4D3. Furthermore, 4D3 increased both intracellular PTX concentration (in both cells) and apoptosis. In the mouse model, 4D3 promoted the antitumor effect of PTX on subcutaneous tumors and reduced lung metastasis. The combination of PTX and 4D3 reduced M2 macrophages and mesenchymal stem cells in the tumor. 4D3 also reduced stemness of the tumors in association with increase in the intratumoral pH. Moreover, concurrent treatment of 4D3, PTX, or tamoxifen; or with PTX and tamoxifen in MDA-468 also showed the same level of antitumor activity and survival as MCF-7. Furthermore, in bone metastasis model, combination of PTX and bisphosphonate with 4D3 promoted tumor growth in both cells.

In vivo imaging of tumor IDC cell were labeled with VivoTrack 680 (ParkinElmer Inc., Waltham, MA, USA). A mouse was examined by Clairvivo OPT in vivo imager (Shimazu, Kyoto, Japan) under anesthesia 16. Fluorescence intensity was calculated with software equipped in the imager.
Intratumoral pH Tumors were penetrated by 18G needle, through which lumen a fine needle probe of pH meter (Chemical Instruments, Co. LTD., Hachioji, Japan) was inserted into tumors under anesthesia. pH was monitored for 5 min to calculate the mean value in one site. The measurement was performed at five sites for each tumor. Representative pH was a mean of values in the five sites.
Immunohistochemistry Thin sections (4-μm) of paraffin-embedded specimens were immunohistochemically stained with 0.2 µg/mL of 4D3 or 2C1 by a previously described immunoperoxidase technique 32. Secondary antibodies conjugated with peroxidase (Medical and Biological Laboratories, Nagoya, Japan) were used (0.2 µg/mL). Color development were performed with diamine benzidine hydrochloride (DAKO, Glastrup, Denmark). A counterstain was done with Meyer's hematoxylin (Sigma). Immunopositive cells at the cytoplasmic membrane were counted.
Staining strength was scored from 0 to 3 (a score of 1 was used to describe the expression level in normal pancreatic duct epithelium). The staining index was calculated as the staining strength score multiplied by the staining area (%) 14-16. For a negative control, non-immunized rat IgG (Santa-Cruz Biotechnology, Santa-Cruz, CA, USA) was used as the primary antibody.
Immunoblot analysis Whole-cell lysates were prepared as previously described 33. Lysates (20 μg) were used to immunoblot analysis using sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE, 12.5%), and then transferred electronically onto nitrocellulose filters. The filters were incubated with primary antibodies and peroxidase-conjugated secondary antibodies (Medical and (Table 1). CLDN4 expression index was associated with histological differentiation, receptor subtype, nodal metastasis (pN), and pathological stage. In contrast, CLDN1 expression index was associated with histological differentiation, and nodal metastasis (pN), but not with receptor subtype. Notably CLDN4 expression in TNBC was higher than those in luminal or HER2 subtypes.

Effect of 4D3 in human IDC cell lines
Human IDC cell lines, MDA-468 (triple negative subtype) and MCF-7 (luminal subtype), were examined with the expressions of CLDN4, CLDN1, and ER with or without E2 treatment ( Fig. 2A). MDA-468 cells expressed CLDN4 at higher levels than MCF-7 cells. In contrast, CLDN1 expression was detected at lower levels in the two cell lines than that of CLDN4. ER expression was detected in only MCF-7.
CLDN4 and CLDN1 expression were not altered by E2 treatment in MDA-468 cells. In contrast, both expressions were decreased in E2-treated MCF-7 cells.
We previously established 4D3 antibody for targeting CLDN4 in cancer cells [16]. MDA-468 and MCF-7 cells were treated with 4D3 and compared with those treated with 2C1 [28] with and without PTX treatment ( Fig. 2B and C). In both cell lines, 2C1 showed no growth inhibition alone, and no enhanced PTX effect. In contrast, 4D3 enhanced PTX-induced growth inhibition at each PTX concentration in both cell lines.
Consistent with enhanced drug permeation into tumor tissues owing to impaired tight junction [16], intracellular PTX levels were found to increase in 4D3-treatment in the two cell lines (Fig. 2D).
PTX-induced apoptosis was increased by 4D3 in the treatment involving both cell lines.

Effect of 4D3 on antitumoral effect of PTX
We examined the antitumor effect of 4D3 on the two cell lines treated with PTX in nude mice (Fig. 3).
In a subcutaneous tumor model, the antitumor effect of PTX when treated in combination with 4D3 We next examined the tumor microenvironment by measuring pH of the tumor tissue. By 4D3 treatment, tumor pH was elevated from weak acidic condition to neutral condition ( Fig 3E). To elucidate the effect of circumstantial pH on stemness, expressions of CD133 and CD44 was examined in cells cultured in media of pH 7.0 and pH 7.4 (Fig. 3F). Expressions of CD133 and CD44 were higher in pH 7.0 than that in pH 7.4 in both cell lines.

Effect of 4D3 on combination treatment of PTX with TAM
We examined the effect of 4D3 on combination treatment of PTX with TAM using mouse subcutaneous tumor model (Fig. 4A-D). I n MCF-7, TAM or PTX alone showed an antitumor effect; however, the combination of both enhanced the effect (Fig. 4A). Furthermore, the combined use of 4D3 enhanced the effects of TAM or PTX and further enhanced the combined effect of TAM and PTX. On the other hand, in MDA-468, TAM alone was not effective, and the combination of TAM and PTX was not different from PTX alone (Fig. 4B). However, the combined use of 4D3 promoted the antitumor effect of PTX, and the same effect as seen in the case of TAM and PTX combination in MCF-7 was obtained.
When examining mouse survival in a similar mouse model with inoculation of MCF-7, the 50% survival periods were 26, 33, and 43 days in mice treated with none (control), PTX and TAM combination, and the three-way combination of PTX, TAM, and 4D3, respectively. Also in the case of MDA-468, 50% survival periods were not significantly prolonged at 35 and 45 days in mice treated with none (control) or PTX and TAM combination, respectively. On the other hand, 50% survival period was significantly extended to 53 days in the three-way combination. A significant prolongation of survivals were found in the three-way combination in both cell lines.

Effect of 4D3 on combination treatment of PTX with BP
Finally, the combined effect of BP (zoledronic acid) with PTX and antibody was examined using a mouse bone metastasis model (Fig. 4E, F). In both MCF-7 and MDA-468 cell lines, there was no antitumor effect with BP alone, but an additional effect with PTX was observed. Furthermore, the combination of 4D3 enhanced the antitumor effect in both PTX and the combination of PTX and BP.

Discussion
In the present study, we examined the therapeutic effect of targeting CLDN4 to IDC, particularly TNBC. When the relationship between CLDN4 expression in IDCs and clinicopathological parameters was examined, a correlation was found between stage and nodal metastasis. Furthermore, when correlations of CLDN4 expression with clinicopathological factors were compared between TNBCs and non-TNBCs (luminal and HER2 subtypes), CLDN4 expression in TNBC was higher than for non-TNBC and also correlated strongly with cancer progression. From this, CLDN4 was considered to be more suitable as a molecular target for TNBCs.
Interestingly, CLDN4 expression in non-TNBCs inversely correlated with stage and nodal metastasis, which is similar to that reported in undifferentiated type gastric cancer [14]. Contrary to the differentiated type gastric cancer, undifferentiated type gastric cancer shows increased, nonassembled (not integrated into tight junction) CLDN4, which is associated with stemness via an integrin signal [14]. Since CLDN4 is known as an epithelial marker [12, 13, 18], downregulation of CLDN4 in non-TNBCs might be associated with epithelial-mesenchymal transition (EMT) [14].
In contrast to non-TNBCs, CLDN4 expression was correlated with primary tumor expansion and nodal metastasis, which is similar to that in bladder cancer [16], colorectal cancer [15], and differentiated type gastric cancer [14]. In such tumors, CLDN4 overexpression might provide an isolated tumor microenvironment mediated through the tight junction, which retains epithelial growth factor (EGF) and vascular endothelial growth factor (VEGF) in the tumor, and inhibits the permeation of anti-cancer drugs [14][15][16]. In TNBCs, we observed a novel barrier function that the tight junction retained lactic acid in the tumor, which lowers intratumoral pH. Tissue pH was elevated by 4D3 treatment from acidic to neutral in the environment. Cancer cells release lactic acid into the cancer microenvironment by the Warburg effect and changes it into acidic [34]. Acidity in the environment affects the activity of immune cells [35,36]. We analyzed the alterations in immune cell property in cancerous tissues by 4D3 treatment, which was examined by the expression of immune cell marker genes. As a result, M2 macrophages and MSCs decreased and M1 macrophages increased by 4D3 treatment. M2 macrophages and MSCs are known to promote the metastatic potential and resistance of cancer cells against anticancer drugs [37-39]. When cancer cells were co-cultured with MSCs and pH was changed from an acidic environment to neutral environment and stemness decreased in our data, treatment with 4D3 alone showed no significant antitumor effect in in vitro but modest in mouse models. 4D3 also possesses antibody-dependent cellular cytotoxicity activity [16]. In the in vivo environment, a stronger antitumor effect than in vitro might be induced including the immunostimulatory effect by the destruction of the cancer microenvironment.
Comparing the expression of the epithelial CLDNs between CLDN4 and CLDN1, the expression level of CLDN4 was higher and correlation with clinicopathological factors was also observed. In contrast, the expression level was low in CLDN1 and less correlation with clinicopathological factors was found. From this, CLDN4 was considered to be a better molecular target than CLDN1. It was previously reported that CLDN4 is upregulated in bladder cancer, colon cancer, gastric cancer, and pancreatic ductal cancer, and correlates with cancer progression [14][15][16]40]. CLDN4 is considered to be a molecular target with broad efficacy in epithelial malignancies.
We previously reported that 4D3 promoted the antitumor effect of CDDP, 5-fluorouracil (5-FU), and FOLFIRINOX in bladder cancer, colon cancer, gastric cancer and pancreatic ductal cancer [14][15][16]. Here, we examined the effects of 4D3 in breast cancer cells, and the use of 4D3 promoted the antitumor effect of PTX in both TNBC and luminal subtype. In the luminal subtype, the anti-tumor effect of TMX was promoted by 4D3 and a synergistic effect with PTX was observed. In contrast, in TNBC, a combination with 4D3 and PTX showed an effect comparable to the PTX and TMX combination in the luminal subtype. Furthermore, when the effect of the 4D3 on the effect of BP was examined in the bone metastasis model, the growth in bone metastasis was suppressed by the 4D3 combination in both TNBC and luminal subtype. Thus, 4D3 was found to promote its effect on any of the anticancer agent, anti-hormone agent, and BP.
Here, we examined the efficacy of 4D3 using a nude mouse model. However, since this 4D3 does not recognize mouse CLDN4, its toxicity to the host could not be evaluated. It is necessary to evaluate the toxicity of 4D3 using an effective model in the future.
From the above findings, 4D3 is considered to be an effective molecular targeting therapeutic agent for IDC, particularly for TNBC. Application in future clinical trials is desired. Consent for publication: As written informed consent was not obtained, any identifying information was removed from the samples prior to analysis, in order to ensure strict privacy protection (unlinkable anonymization).

Conclusion
Availability of data and material: All data generated or analyzed during this study are included in this published article.
Competing interests: The authors declare that they have no competing interests.  1) The staining index was calculated as the staining strength score (0 to 3) multiplied by the staining area (%).
2) P value was calculated by Kuruskal-Wallis test.