Introduction
Worldwide, breast cancer has the highest incidence rate of all cancers among women. Despite the progress of multidisciplinary therapy, it is difficult to cure advanced or recurrent cases [
1], thus necessitating the development of innovative treatment strategies. The efficacy of immune checkpoint inhibitors has been demonstrated in multiple cancer types, and tumor immunology has attracted great attention as an innovative therapeutic strategy [
2]. Therefore, elucidating the unique immunomodulatory mechanisms underlying the breast cancer microenvironment will provide significant insights toward development of novel therapeutic strategies.
Among all cancer types, androgen receptor (AR) is most highly expressed in breast cancer after prostate cancer (The Human Protein Atlas.org.
https://www.proteinatlas.org/ENSG00000169083-AR; accessed on November 07, 2022) [
3]. It is expressed in 60–80% of breast cancers with varying extents across the subtypes [
4]. AR is a nuclear transcription factor with a diverse range of biological actions, principally in the development and maintenance of the male reproductive system [
5]. Accumulating data strongly suggest that AR signaling affects the immune response in various physiological and pathological conditions including allergic disease, autoimmune disease, and cancer [
6‐
8]. Recently, we and other researchers found that AR expression is inversely correlated with immune cell infiltration in breast cancer tissues [
9‐
12]. However, there is limited information regarding its mechanisms. Secreted factors regulated by AR can act as immune-regulation mediators. In a previous study, we identified multiple AR-dependent secreted proteins produced by breast cancer cells [
13]. Alpha-2-glycoprotein 1, zinc-binding (ZAG), encoded by the
AZGP1 gene, is one of these proteins and is the second most sensitive protein to AR activity after prostate-specific antigen (so-called PSA). ZAG is a 40-kDa single-chain polypeptide secreted in various body fluids and present in high concentrations in the human seminal plasma and breast cyst fluid [
14,
15]. Moreover, ZAG is significantly higher in the serum of patients with breast cancer than in healthy controls [
16]. Owing to its structural similarity to HLA class I, ZAG is thought to be involved in immune response, although its detailed function remains unclear [
14,
17].
In this study, we focused on ZAG as a candidate protein to regulate AR-dependent immune-regulatory mechanisms in the breast cancer microenvironment. We systematically analyzed the relationship between the expression of AZGP1/ZAG and the immunological profiles of breast cancer tissues at both the gene and protein level. Furthermore, we verified the effect of ZAG on Mφ using in vitro models of primary culture of human peripheral blood mononuclear cell (PBMC)-derived Mφ.
Histological evaluation of hormone receptors, human epidermal growth factor receptor 2, Ki67, tumor immunity-related biomarkers, and ZAG
Histological evaluation of hormone receptors (ER: estrogen receptor; PgR: progesterone receptor; and androgen receptor: AR), human epidermal growth factor receptor 2 (HER-2), Ki67, and tumor immunity-related biomarkers (programmed death-ligand 1; PD-L1, tumor-infiltrating lymphocyte; TILs) of breast cancer tissues have been described previously [
9,
23]. Histologically assessed PD-L1 and TIL are indicated with “h” (i.e., hPD-L1, hTIL) to distinguish them from others. The nuclear staining of ER, PgR, Ki67, and AR in carcinoma cells was assessed, and the percentage of immunoreactive cells was determined. We determined the ER, PgR, and HER-2 status according to the relevant American Society of Clinical Oncology/College of American Pathologists guidelines [
24,
25]. The Ki67 Labeling Index (Ki67 LI) and AR status were categorized into high and low, with cutoff values set at 20% and 60%, respectively [
9]. According to the International TILs Working Group guidelines [
26], hTILs in the stromal tissue sections were evaluated and categorized into low, intermediate, and high. Tumors with ≥ 1% immune cells displaying cytoplasmic and/or membrane PD-L1 staining were determined to be hPD-L1 positive [
27]. ZAG expressions were evaluated using immunohistochemistry and scored using the semiquantitative H-score method [
28] to calculate the sum of the percentage and intensity of positively stained invasive tumor cells. Representative examples of ZAG immunostaining are shown in Supplementary Figure
S1. ZAG status (i.e., ZAG high,
n = 22; ZAG low,
n = 23) was divided by the median value of the H-score. Details of the antibodies used for immunohistochemistry are summarized in Supplementary Table
S1.
Monocyte isolation
PBMCs were separated from the whole blood sample of healthy women by density gradient centrifugation using Histopaque® 1077 (Sigma-Aldrich, St. Louis, MO, USA). Red blood cells were hemolyzed with a red blood cell lysing buffer (Sigma-Aldrich, St. Louis, MO, USA) for 10 min at 37 °C. The cell suspension was treated with a FcR blocking reagent (Miltenyi Biotech, Bergisch, Gladbach, Germany) and Cluster of Differentiation (CD) 14 MicroBeads (Miltenyi Biotech, Bergisch, Gladbach, Germany) for 15 min at 4 °C. The cells were washed, and the CD14 + monocytes were isolated using an autoMACS® Pro Separator (Miltenyi Biotech K.K., Tokyo, Japan).
Construction of M1/M2 polarization models using PBMC-derived Mφ and THP-1 cells
Monocytes isolated from PBMCs were suspended in a basal medium consisting of RPMI 1640 media (Gibco Brl, Grand Island, NY, USA), supplemented with heat-inactivated 10% fetal bovine serum (BioWest, Nuaillé, France) and 1% penicillin–streptomycin (Sigma-Aldrich, St. Louis, MO, USA), counted and seeded into Upcell
® Multi 24 well plate (CellSeed Inc, Tokyo, Japan) in 2E5 to 5E5 cells/well. To obtain Mφ [
29], we incubated the monocytes for 4 days in a basal medium supplemented with 10 ng/mL granulocyte macrophage colony-stimulating factor (GM-CSF) (Peprotech, Rocky Hill, NJ, USA) at 37 °C and 5% CO
2. Subsequently, we replaced the medium with a basal medium comprising each of the following supplement: M1 polarization model—with 10 ng/mL interferon gamma (Peprotech, Rocky Hill, NJ, USA) + 10 pg/mL lipopolysaccharides from
Escherichia coli (LPS, Sigma-Aldrich, St. Louis, MO, USA) [
30]; M2 polarization model—with 20 ng/mL interleukin 4 (Peprotech, Rocky Hill, NJ, USA) [
31]; and non-polarization model—no supplement. In all polarization models, the indicated concentrations of recombinant human-ZAG (R&D Systems, Minneapolis, MN, USA) or vehicle control were added; after 2 days of incubation, the cells were harvested and subjected to an FCM analysis (Supplementary Figure S2a–c). To harvest the cells, the medium was replaced by ice-cold phosphate buffered saline to the Upcell
® Multi 24-well plate. The plates were maintained at 25 °C for 30 min to promote cell detachment, and the cells were collected by pipetting. Furthermore, we examined human monocytic THP-1 [
32] cells for a similar experiment with minor modifications. Briefly, THP-1 cells were supplied by Dr. A. Kotani, cultured in a basal medium and culture conditions similar to that of monocytes isolated from PBMC. We used 10 ng/ml Phorbol-12-myristate-13-acetate (PMA) (Adipogen Corp. San Diego, CA, USA) for Mφ differentiation, instead of GM-CSF, and performed each experiment in triplicate (Supplementary Figure S3a–c).
FCM analysis
FCM data from breast cancer tissue samples were obtained from our previous study [
23] and used for this ad hoc analysis. According to the staining profile of the FCM-evaluated surface antigen, the cells were classified as follows: leukocytes, total T cells, CD4 + T cells (CD4 + T), CD8 + T cells (CD8 + T), B cells (B), monocytes/macrophages (Mo/Mφ), non-classical monocytes (CD16 + Mo), myeloid-derived suppressor cells (MDSCs), dendritic cells (DCs), myeloid DCs, natural killer (NK) cells, minor NK cells, and natural killer T cells. The density of each immune cell fraction was determined as the count of cells per weight of the tumor tissue (count/g) [
23]. Furthermore, we assessed the percentage of PD-L1 and CD86 positive cells in each immune cell fraction. The gating strategy for PD-L1 or CD86 positivity is described in Supplementary Figure S4a-d. We examined the action of ZAG in M1, M2, and non-polarization models based on the expression of surface antigens, such as M1/M2 polarization markers (CD86, CD80/CD163, and MRC1) [
33] and HLA class I/II, evaluated using FCM (
n = 15). The cell suspension was supplemented with FcR blocking reagent and mixed with an antibody cocktail. The cells reacted at 4 °C for 30 min and washed. Stained samples were detected using BD LSR Fortessa (BD Biosciences, Franklin Lakes, NJ, USA) and analyzed using the FlowJo software v10.8.1 (BD Biosciences). The gating strategy for single-cell detection and representative histograms of fluorescence intensity derived from each surface antigens are described in Supplementary Figure S5a, b. Antibodies used in these experiments are summarized in Supplementary Table S2.
Statistical analyses
We used the GraphPad Prism ver. 9.1.0 software for statistical analyses and graph preparation. All data were assessed using the D'Agostino–Pearson normality test; parametric or nonparametric tests were used depending on the data distribution. We performed correlation analyses between the groups using Spearman’s rank correlation coefficient. |
r-value|> 0.3, and a significant
p value was defined as a positive or negative correlation [
34]. The Fisher’s exact test was performed to compare categorical variables between the groups. Continuous variables between the two unpaired groups were compared using the unpaired t test or the Mann–Whitney U test. For multiple comparisons, we performed the Kruskal–Wallis test and Dunn's multiple comparisons test. Continuous variables between two paired groups were compared using the paired
t test or the Wilcoxon-test. A
p value < 0.05 was defined as statistically significant. In our previous study, FCM data from patients with breast cancer contained outliers [
23]. Here, all analyses were performed without omitting outliers; nonetheless, we identified the outliers using robust regression and outlier removal method, excluded them, and performed all statistical analyses to ensure the reliability of our analyses.
Discussion
This study aimed to assess the role of ZAG protein in the AR-dependent immune-regulatory mechanisms in the breast cancer microenvironment. In silico analyses using public gene expression datasets and analysis of our in-house FCM datasets demonstrated that AZGP1/ZAG is associated with immunosuppressive phenotype and reduces the infiltration of specific immune cell subsets, particularly Mφ, into breast cancer tissues. The in vitro analysis using PBMC-derived Mφ indicates that ZAG affects Mφ phenotypic change or function.
In a previous study, we identified seven AR-responsive genes that encode secreted factors in breast cancer [
13]. GSEA demonstrated that four of these seven genes were significantly associated with immune-related processes (positive association:
CFH and
FASN; negative association:
AZGP1 and
PIP) (Supplementary Figure S9). Of these, ZAG suppresses inflammatory cytokines and exhibits anti-inflammatory effects in animal models of non-alcoholic fatty liver, epilepsy, and atopic dermatitis [
35‐
37]. However, the function of ZAG in tumor immunity and its specific mechanism are still unknown.
ZAG is a well-known AR regulated protein [
14]; however, its significance in breast cancer has not been reported except in association with a lower histologic grade [
14]. Analyses of our in-house dataset further supported these existing data and generated novel findings, such as the association of ZAG with lower Ki67 and HER2 positivity (Table
1, Fig.
2). The negative association between ZAG with Ki67 was consistent with the GSEA results, suggesting a negative correlation between
AZGP1 expression and gene sets, such as "G2M CHECK POINT" and "E2F TARGETS" (Fig.
1b). Several reports have shown fewer TILs in hormone receptor-positive HER2-negative subtype compared with that in HER2-positive subtype [
38,
39]. Therefore, in this study, it was important to exclude the influence of HER2 status on the tumor immune microenvironment using subgroup or multivariate analysis. However, as mentioned later, the number of cases was small and such an analysis was not possible in the present study; nonetheless, this should be considered in future studies.
This novel study demonstrated the relationship between the immunological profile of the breast cancer microenvironment and
AZGP1/ZAG expression, with results consistent with the findings of previous reports, suggesting the immunosuppressive functions of ZAG in animal models of various diseases [
35‐
37]. ZAG expression was associated with a lower PD-L1 positive ratio in CD4 + T; however, in our previous studies, PD-L1 expression was originally low in CD4 + T cells [
40] and was expressed principally in the myeloid lineage [
23]. CD80 and CD86 are predominantly expressed on antigen-presenting cells and bind to CD28 or CLTA-4 on T cells to provide co-stimulatory signals for T-cell activation or inactivation [
41]. In addition to the inverse correlation between Mφ and
AZGP1/ZAG expression in breast cancer tissues (Fig.
1e,
3f), ZAG expression was associated with decreased CD86-positive cells in Mo/Mφ (Supplementary Figure S7d), thus suggesting that ZAG is involved in Mφ differentiation or its function. These molecules are not mere markers; they play a major role in regulating the immune system. For example, both CD80 and CD86 are used as M1 polarization markers [
29,
33], and their immunological functions have been discussed. CD163, an M2 marker, has been associated with anti-inflammatory functions [
42]. MRC1, an M2 marker [
43], is associated with antigen recognition and processing for antigen presentation on HLA class I molecules (cross-presentation); it also has an active role in the induction of T-cell tolerance [
44]. Activation of anti-tumor T cells requires the recognition of cancer antigen presented on HLA class I molecules on the tumor; the loss of HLA class I on the tumor leads to the malfunction of recognition by the CD8 + T cells [
45,
46]. Furthermore, HLA class II molecules are used as M1 markers [
33] and are involved in antigen presentation to CD4 + T helper cells.
In addition to relatively high expression in the mammary gland,
AZGP1 is highly expressed in breast cancer, demonstrating partial organ specificity (Supplementary Figures S10a and b; The Human Protein Atlas.org.
https://www.proteinatlas.org/ENSG00000160862-AZGP1; accessed on November 07, 2022) [
3]. We only verified a part of the ZAG function for the immune system. However, if the function of ZAG on the immune system is further established, ZAG may be a promising therapeutic target. Because breast cancer-derived ZAG can be detected in the serum [
16], an assessment of serum ZAG may provide a minimally invasive marker reflecting the host immune response in the breast cancer microenvironment. Thus, our results may facilitate research with more clinical aspect in the future, and we too are conducting further analyses.
This study had some limitations, namely the relatively small number of patients enrolled in the in-house FCM dataset, the selection bias that may have affected the clinicopathological factors of the enrolled cases, and no subgroup analyses by the tumor subtype because of the small sample size [
9,
23]. ZAG expression in breast cancer tissues was associated with decreased CD86 expression in Mφ (Fig. S6d); however, this association was not replicated in in vitro experiments (Fig.
4b, h, and n; Supplementary Figures S8b, h, and n). Besides, ZAG did not necessarily demonstrate a similar effect in M1/M2 and non-polarization models (Fig.
4). For instance, ZAG supplementation decreased CD80 expression in the M1 polarization model but increased it in the non-polarization model (Fig.
4a, m). Similarly, the effect of ZAG on PBMC-derived Mφ was not necessarily reproducible on THP-1 (Supplementary Figure S8). Supplementation of ZAG resulted in a statistically significant change in the expression of surface markers (Fig.
4); however, the change was so small that further investigation is required to determine whether it is clinically significant. The M1/M2 terminology has been introduced where M1s are pro-inflammatory and M2s are anti-inflammatory [
47]. These states appear as two extremes with a large spectrum of macrophages in between [
42]. Besides, some of these molecules (i.e., CD80, CD86, and MRC1) are bifunctional regarding their inflammatory or anti-inflammatory properties. Thus, further studies are required to determine the immunological role of ZAG in breast cancer. Lastly, as a future prospect of our study, validation using a mouse model is expected to provide further insights into the findings obtained in this study.
Declarations
Conflict of interest
Hiroshi Kagamu has an advisory role in ImmuniT Research Inc, received honoraria from AstraZeneca K.K., Ono Pharmaceutical Co. Ltd., Bristol-Myers Squibb Co. Ltd., and Chugai Pharmaceutical Co., Ltd. Shigehisa Kitano received honoraria from Ono Pharmaceutical Co., Bristol-Myers Squibb Co., Ltd., AstraZeneca K.K., Chugai Pharmaceutical Co., Ltd., Pfizer Japan Inc., and MSD Co. Ltd, received research funding from Astellas Pharma Inc., Gilead Sciences Inc., Eisai Co., Ltd., Regeneron Pharmaceuticals Inc., Boehringer Ingelheim GmbH, Daiichi Sankyo Co., Ltd., Ono Pharmaceutical Co., Takara Bio Inc., PACT Pharma Inc., Chugai Pharmaceutical Co., Ltd., and MSD Co., Ltd. Naoki Niikura received honoraria from AstraZeneca K.K., Daiichi Sankyo Co. Ltd., Pfizer Japan Inc., Eisai Co. Ltd., and Nippon Kayaku Co. Ltd. Sasagu Kurozumi has received honoraria from Eli Lilly and Company, Daiichi Sankyo Co. Ltd, Taiho Pharmaceutical Co. Ltd, MSD K.K., AstraZeneca K.K., Chugai Pharmaceutical, Ltd., Dinow Inc., Eisai Co., Ltd., and Novartis Japan, Takeda Pharmaceutical Co., Ltd., Kyowa Kirin Co., Ltd.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.