1. Introduction
Immunotherapies, whose aim is to potentiate anti-tumor immune activities, are now being implemented in the treatment of melanoma and of tumors with high mutation rates and DNA mismatch repair defects [
1,
2,
3]. In breast cancer, immune checkpoint blockades (ICBs) have been introduced to the treatment of triple-negative breast cancers (TNBCs) that demonstrate relatively high genomic instability [
4,
5]. TNBC is a most aggressive subtype of breast cancer, one of the four major breast cancer subtypes that also include the prevalent luminal A subtype, and the luminal B and HER2+ subtypes [
6].
Clinical studies in the last several years have indicated that the response rates to ICBs in TNBC patients are limited [
4,
5]. These observations emphasize the need to better identify the mechanisms leading to insufficient success rates of ICBs, and to better understand their potential use in other breast cancer subtypes. In this context, we have recently addressed the possibility that PD-L1 carry out broader pro-metastatic activities than immune suppression, and explored the ability of PD-L1 to induce cell-autonomous pro-metastatic phenotypes and functions in TNBC and luminal A breast cancer cells. The relevance of this research direction was reinforced by the fact that PD-L1 is expressed by the tumor cells in patient biopsies in both TNBC and luminal A subtypes, with a more prevalent expression in the first than in the latter [
7,
8,
9].
Our published study demonstrated that by being expressed in TNBC and luminal A breast cancer cells, PD-L1 has acted in a cell-autonomous manner to elevate the pro-metastatic activities of the tumor cells [
10]. PD-L1 has promoted the growth of the cells, their ability to release the pro-metastatic chemokine CXCL8 and to invade; all these functions were elevated by exposure to PD-1 in vitro, and have led to increased tumor growth and metastasis in a T cell-deficient mouse model system [
10]. Taken together, these findings revealed pro-metastatic roles for PD-L1 that were not connected to its immune-suppressive functions, but rather to cell-autonomous activities exerted in the cancer cells themselves, leading to increased disease progression.
In additional analyses, we have demonstrated that all PD-L1 activities in the tumor cells depended on the integrity of the S283 intracellular residue of PD-L1 [
10]. The cardinal roles played by S283 were revealed at multiple tumor-promoting levels, including tumor cell proliferation, CXCL8 release and invasion, as well as in vivo [
10]. In the T cell-deficient animal model system, cells expressing the S283-mutated PD-L1 variant demonstrated poor tumorigenic and metastatic phenotypes [
10].
In the current research, we have taken a step further in deciphering the processes mediating the pro-metastatic roles of PD-L1 in breast tumor cells, at the cell-autonomous and PD-1-induced levels. PD-L1 lacks canonical protein–protein interaction motifs, thus presenting a challenge in identifying its mechanisms of action. Recent studies proposed that the mTOR/AKT pathways are the ones mainly mediating the intrinsic activities of PD-L1 in several cancer systems [
11,
12,
13,
14]. The interactions of the mTOR cascade with STAT-mediated signaling (e.g., in immunity [
15]), and the roles attributed to STAT3 and STAT1 in promoting PD-L1 expression [
16,
17,
18,
19], have led us to inquire in depth about the potential involvement of STAT3 and STAT1 in PD-L1-mediated effects in breast cancer cells. This research direction was further enforced by studies demonstrating that STAT3 and STAT1 have cardinal roles in regulating cancer progression, including in breast cancer [
18,
20].
Here, we demonstrate that the ability of PD-L1 to induce cell-autonomous- and PD-1-induced effects is mediated through STAT3 and STAT1 activation, leading to increased tumor cell growth, CXCL8 release and cancer cell invasion in luminal A and TNBC breast cancer cells. Moreover, we have extended the search for mechanisms that control PD-L1 functions and asked whether PD-L1 needs to be N-linked glycosylated in order to exert its cell-autonomous, tumor-promoting functions in these cells.
Previous studies have demonstrated that N-linked glycosylation of PD-L1 increased PD-1 binding, leading to more efficient inhibition of anti-tumor T cell activities; it was also found to interfere with the binding of antibodies used for PD-L1 diagnosis and PD-L1-directed therapy [
21,
22,
23,
24,
25,
26]. PD-L1 N-linked glycosylation was noted in breast cancer patient tumors, and following removal of N-linked glycosylation, PD-L1 levels correlated with response rates to anti-PD-L1 therapy [
23,
24].
Adding to these findings, we now demonstrate that PD-L1 has to be glycosylated at each of its N-linked glycosylation sites (N35, N192, N200 and N219; [
25,
26,
27]) in order to exert cell-autonomous tumor-promoting functions in breast tumor cells in vitro. In this context, we have also identified the N-linked glycosylation sites that are necessary for STAT3 and STAT1 activation. Furthermore, our findings indicate that specific PD-L1 N-linked glycosylation sites are required for its ability to promote tumor growth and metastasis in vivo, and shed light on the clinical implications of these findings in breast cancer patients (using the TCGA dataset).
Overall, in this study we demonstrate that the cell-autonomous pro-metastatic functions of PD-L1 in breast cancer cells are fully dependent on all four N-linked glycosylation sites of the protein, and that N-linked glycosylation of PD-L1 at specific sites leads to STAT3 and STAT1 activation. Consequently, PD-L1 promotes breast tumor cell growth, CXCL8 release and invasion, as well as tumor growth and metastasis in vivo. These PD-L1-induced processes have clinical implications, suggesting that inhibitory measures directed at N-linked glycosylation may apply in breast cancer therapy.
2. Materials and Methods
2.1. Cell Growth
Human MCF-7 Luminal A cells and human MDA-MB-231 (MDA) TNBC cells (both from ATCC) were grown in DMEM medium. Medium was supplemented with 10% fetal bovine serum (FBS), 2% L-glutamine and 1% penicillin-streptomycin-amphotericin solution (from Biological Industries, Beit Ha’emek, Israel and Sigma-Aldrich, St. Louis, MO, USA).
To determine the effects of PD-1 stimulation on the cells, a recombinant PD-1-IgG1 Fc chimera protein (endotoxin free; #1086-PD; R&D Systems, Minneapolis, MN, USA) was used at 2 μg/mL for 72 h. Recombinant human IgG1-Fc (#110-HG; R&D Systems) was used as control at similar conditions. Concentrations were selected based on preliminary experiments performed beforehand (as mentioned in [
10]).
When relevant, MCF-7 and MDA cells were treated with kifunensine (50 μM; #K1140, Sigma-Aldrich) and/or swainsonine (50 μM; #S8195, Sigma-Aldrich) dissolved in double-distilled water. Following kinetics studies (see figures), analyses were performed of MCF-7 cells and MDA cells exposed to kifunensine or its vehicle for 48 and 72 h, respectively.
2.2. Western Blot
Cells were lysed in RIPA buffer, and Western blot (WB) analyses were performed using antibodies from Cell Signaling Technology (CST, Danvers, MA, USA), unless otherwise indicated: total (T)-STAT3: #4904; Phosphorylated (P)-STAT3-Y705: #9145; (T)-STAT1: #9172; (P)-STAT1: #9167; PD-L1: (#51296S). Antibodies to Calnexin (#2679S; Abcam, Cambridge, UK) or GAPDH (#ab9485; Abcam) served as loading controls. The membranes were incubated with streptavidin-horseradish peroxidase (HRP)-conjugated goat anti-rabbit IgG (#111-035-003; Jackson ImmunoResearch Laboratories, West Grove, PA, USA). The membranes were subjected to enhanced chemiluminescence (ECL) analysis (#WBLUR0500, Merck, Darmstadt, Germany), and were visualized with Amersham ImageQuant 800 (GE Healthcare, Little Chalfont, UK).
Densitometry graphs of STAT3 and STAT1 activation were generated with the ImageJ software (Version 1.53t); they present arbitrary units of pSTAT3/GAPDH or pSTAT1/GAPDH values in cells expressing WT-PD-L1 or different PD-L1 N-linked glycosylation mutants. % inhibition of STAT3 and STAT1 activation in PD-L1 mutants compared to WT-PD-L1 were calculated as “1 minus [the value of STAT3 or STAT1 activation in a PD-L1 mutant, divided by the value of STAT3 or STAT1 activation in WT-PD-L1]”.
2.3. STAT3 and STAT1 Down-Regulation by siRNA
The expression of STAT3 and STAT1 in MCF-7 cells and MDA cells was knocked down (KD) by transient siRNA transfection (10nM; Based on titration analyses), using Lipofectamine RNAiMAX transient reagent (#13778075; Invitrogen, Grand Island, NY, USA) according to manufacturer’s instructions, in a reverse transfection protocol. The following ON-TARGET plus siRNA SMART pools were used (all from Dharmacon, Lafayette, CO, USA): STAT3 siRNA (siSTAT3): #L-002000-00; STAT1 siRNA (siSTAT1): #L-003543-00. siRNA control (siCTRL) was introduced by ON-TARGET plus non-targeting control siRNA pool (#D-001810-10). At 72 h after transfection, cells were used in assays, as necessary. KD of STAT3 and STAT1 expression was validated by WB analyses.
2.4. Flow Cytometry Analyses of PD-L1 Expression
Cell surface expression of PD-L1 was determined by flow cytometry using mouse IgG1 antibodies against human PD-L1 (#14-5983-82, Thermo Fisher Scientific, Waltham, MA, USA), followed by FITC-conjugated goat anti-mouse IgG antibodies (#115-095-003, Jackson ImmunoResearch Laboratories). Baseline staining was determined by isotype-matched control mouse IgG antibodies (#400102; Biolegend, San Diego, CA, USA). Fluorescence was determined by Flow Cytometer S100EXi (Stratedigm, San Jose, CA, USA), using CELLCAPTURE software (Version 1; Stratedigm), and analyzed by FLOWJO V10 (Version 10.7.1; BD Biosciences, Franklin Lakes, NJ, USA).
2.5. Cancer Cell Growth In Vitro
Analyses of cell growth rates were conducted by plating the cells of different treatments at equal numbers. Cell numbers were determined after 72 h in culture using trypan blue exclusion assay in ≥2 replicates/cell type. When relevant, glycosylation inhibitors or their vehicle control were added to cell cultures one day after cell culturing, and cell counts were performed 72 h later.
2.6. Invasion
Cancer cells were plated in transwells with 8 μm-pore membranes (#3422, Sigma-Aldrich) coated by 20 μg/mL matrigel (#7058006, Sigma-Aldrich). The bottom wells were filled with medium that was supplemented with 10% FBS. Invasion of MCF-7 cells and MDA cells was determined after 21.5 h or 11 h of migration, respectively.
In specific experiments, the cells were stimulated with PD-1 or its control (as above); In other assays, the cells were treated with kifunensine or its vehicle (as above); in cases in which both PD-1 and kifunensine were used, the cells were first treated by kifunensine or its vehicle (as above) for 48 h for MCF-7 cells or 72 h for MDA cells. Then, the cells were stimulated with PD-1 or its control (as above) for additional 72 h (kifunensine was added to the plates at the time of PD-1 addition, until the end of PD-1 stimulation). Following these treatments, cells were plated in transwells, as described above.
In all invasion experiments, cells that migrated to the lower part of the membranes were fixed by methanol, stained with Hemacolor (#111661; Merck, Kenilworth, NJ, USA), photographed using light microscopy and counted at multiple high-power fields (HPF).
2.7. ELISA Assays
MDA cells undergoing different treatments were plated at similar cell numbers, and 48-96 h later, cell supernatants were collected. When relevant, one day after cell plating, the cells were exposed to kifunensine or its vehicle control for 72 h, followed by exposure to PD-1 or its control for additional 72 h, ending with collection of cell supernatants.
CXCL8 levels were determined in cleared supernatants (by centrifugation) and ELISA analyses were performed, using the following antibodies and recombinant proteins (all from Peprotech, Rocky Hill, NJ, USA): CXCL8 coating antibodies: #500-P28; CXCL8 detecting antibodies: #500-P28BT; recombinant human CXCL8: #200-08. HRP-conjugated streptavidin (#016-030-084, Jackson Immunoresearch laboratories) and substrate TMB/E solution (#ES001, Millipore, Burlington, MA, USA) were added, the reaction was stopped by addition of 0.18 M H2SO4 and absorbance was measured at 450 nm.
2.8. Generation of Glycosylation Mutants
MCF-7 cells that over-express human WT-PD-L1 or PD-L1 variants mutated in N-linked glycosylation sites were generated in parallel to control vector cells (“Vector” cells) that underwent a similar procedure with a sham vector. Briefly, cells have been infected to express mCherry-pQCXIN plasmid (with a neomycin selection marker) to allow for intravital in vivo analyses. Following mCherry infection, the cells underwent retroviral infections with WT-PD-L1-pQCXIP or with S283A-PD-L1-pQCXIH [
10]. Control “Vector” cells were derived by infection of mCherry-expressing MCF-7 cells with an empty pQCXIP vector or empty pQCXIH vector [
10]. In parallel, in the current study, MCF-7 cells expressing mCherry were infected to express PD-L1 variants mutated at N-linked glycosylation sites, generated as described below.
To generate a similar system in MDA cells, the expression of endogenous PD-L1 was first knocked-out (KO) using the Alt-R CRISPR-Cas9 system (Integrated DNA Technologies, Coralville, IA, USA) to generate a pool of three clones. These cells were infected by the control vector (“Vector” cells), by WT-PD-L1-pQCXIP or by S283A-PD-L1-pQCXIH [
10]. In parallel, MDA cells expressing mCherry were infected to express the N-linked glycosylation-mutated PD-L1 variants, generated as described below.
To generate the N35A-PD-L1 mutant, we used the PD-L1 (N35A)-sense primer GTAGAGTATGGTAGCGCTATGACAATTGAATGC and the PD-L1 (N35A)-anti-sense primer GCATTCAATTGTCATAGCGCTACCATACTCTAC. To generate the N192A-PD-L1 glycosylation mutant, we used the PD-L1 (N192A)-sense primer GAGGAGAAGCTTTTCGCTGTGACCAGCACACTG and the PD-L1 (N192A)-anti-sense primer CAGTGTGCTGGTCACAGCGAAAAGCTTCTCCTC. The N200A-PD-L1 glycosylation mutant was generated using PD-L1 (N200A)-sense primer AGCACACTGAGAATCGCCACAACAACTAATGAGA and PD-L1 (N200A)-anti-sense primer TCTCATTAGTTGTTGTGGCGATTCTCAGTGTGCT. The N219A-PD-L1 glycosylation mutant was generated using PD-L1 (N219A)-sense primer TTAGATCCTGAGGAAGCCCATACAGCTGAATTG and PD-L1 (N219A)-anti-sense primer CAATTCAGCTGTATGGGCTTCCTCAGGATCTAA. We also constructed a mutant that included all four mutations above (termed 4A-PD-L1). The generated fragments were digested with Age1 and Pac1 and ligated into the corresponding sites of pQCXIP vector (
https://www.addgene.org/vector-database/3870/; accessed on 15 January 2023). By sequencing, we validated that the PCR product of WT-PD-L1 was identical to the published sequence (NM_014143) and that PD-L1 variants carried the desired mutations. Overall, five mutated N-linked glycosylation mutants were generated: N35A-PD-L1, N192A-PD-L1, N200A-PD-L1, N219A-PD-L1 and 4A-PD-L1.
Following infection with WT-PD-L1, N35A-PD-L1, N192A-PD-L1, N200A-PD-L1, N219A-PD-L1, 4A-PD-L1 vectors and sham vector, PD-L1 expression was determined in MCF-7 cells and MDA cells by flow cytometry.
2.9. Tumor Growth and Metastasis
Three types of mCherry-expressing cells were administered orthotopically to the mammary fat pads of female athymic nude mice (#NUDE242; EnvigoRMS, Jerusalem, Israel) in two independent experimental repeats that were performed in a similar manner. The mice groups included mice injected with: WT-PD-L1-MDA cells (n = 10 mice in total), N35A-PD-L1-MDA cells (n = 10 mice in total) and N219A-PD-L1-MDA cells (n = 9 mice in total). The different cell types were mixed with matrigel (#354234; Sigma-Aldrich) at a 1:1 ratio prior to injection. Mice weight and tumor volumes were determined every 3–4 days by scale and caliper, respectively.
Since the mice groups had different lag periods in appearance of primary tumors, animals in each group of mice were sacrificed when tumor volumes in that group reached the size limits dictated by the regulations of the Ethics Committee for Animal Use at Tel Aviv University. Tumor volumes were determined during the process by using caliper, and were measured ex vivo when they were removed from mice. At the time of experiment termination, metastases were determined ex vivo in excised organs (lymph nodes that were adjacent to the primary tumors, lungs, liver and femur), based on mCherry signals, using the Maestro imaging device.
Comparisons between tumor volumes were performed by Anova. Statistical analyses of kinetics of tumor appearance were performed through Gehan–Breslow–Wilcoxon test. The proportions of mice bearing metastases were compared by Chi square test for trend.
Procedures involving experimental animals were approved by the Ethics Committee for Animal Use of Tel Aviv University (Approval no. 2306-138-5) and were performed in compliance with local animal welfare laws, guidelines and policies.
2.10. Analyses of Patient Datasets
The TCGA dataset [
28] was used to perform survival analyses based on the abundance of MAN2B1 (coding for α-mannosidase I) and MAN2A1 (coding for α-mannosidase II). Kaplan–Meier analyses of overall survival and progression-free intervals were performed for each of the two genes based on their median expression levels. Disease subtypes were defined based on the dataset PAM50Call_RNAseq annotation file of the TCGA dataset, including 420 Luminal A patients and 140 basal-like patients (often overlapping with the term TNBC). Log rank tests, provided by the TCGA dataset, were used to determine
p values.
2.11. Statistical Analyses
Statistical analyses of in vitro studies were performed by two-tailed unpaired Student’s t-tests. Statistical analyses of TCGA and in vivo studies were described in the relevant sections. Values of p ≤ 0.05 were considered statistically significant.
4. Discussion
A major breakthrough in cancer therapy has been achieved by directing ICBs to the PD-L1/PD-1 axis, well recognized for its pivotal roles in shutting down potential T cell activities against cancer cells [
1,
2,
3]. In parallel, it was recently reported that PD-L1 expression by cancer cells can promote pro-metastatic phenotypes and functions of the tumor cells [
14,
31,
32,
33,
34].
Within this line of research, our recently published study has demonstrated that PD-L1 leads to pro-metastatic functions in breast cancer cells, which are further potentiated by PD-1 stimulation [
10]. These activities, as well as the in vivo tumor- and metastasis-promoting effects of PD-L1 in a T cell-independent system, demonstrated that PD-L1 acts in a cell-autonomous manner to increase aggressiveness in breast cancer. Moreover, we found that all these activities were fully dependent on the integrity of the S283 intracellular residue of PD-L1 [
10].
The roles of PD-L1 as an inducer of tumor cell properties that elevate tumor progression suggest that therapeutic strategies directed towards its immune-suppressive properties, alongside with measures inhibiting its cell-autonomous functions, should be considered in breast cancer therapy. Moreover, they emphasize the need to identify the mechanisms mediating and regulating PD-L1 functions in breast tumor cells.
Accordingly, in this study we have provided several novel findings on PD-L1 functions and regulation in breast cancer cells: (1) We have identified STAT3 and STAT1 as major molecular pathways mediating PD-L1 functions, cell-autonomous as well as PD-1-induced, in breast cancer cells; (2) we have shown that PD-L1 has to be N-linked glycosylated at all four sites in order to exert its pro-metastatic effects in breast cancer cells in vitro, at the levels of tumor cell invasion and CXCL8 release; (3) of the four different N-linked glycosylation sites, it was mainly the N219 site that induced STAT3 and STAT1 activation, with accompanying roles for the N192 and N200 N-linked glycosylation sites (depending on the cell type), leading to the cell-autonomous pro-metastatic functions of PD-L1 in vitro; (4) the malignancy potential of breast tumor cells in vivo was highly dependent on PD-L1 N-linked glycosylation; here, we are the first to show that the cell-autonomous tumor-promoting and metastatic potential of breast tumor cells was significantly reduced upon impairment of even one of the four N-linked glycosylation sites of PD-L1, e.g., N35 or N219.
To date, several studies have addressed the signaling pathways induced by PD-L1 in cancer cells, connecting mTOR/AKT activation with increased PD-L1 activities in cancer cells, including in breast tumor cells [
11,
12,
13,
14]. In this respect, it is interesting to note that interactions between the mTOR and the STAT pathways are known to take place in immune cells [
15], suggesting that our observations on the roles of STAT3 and STAT1 in mediating PD-L1 activities in breast cancer may be connected to the activation of the mTOR axis. In this context, it is interesting to note that PD-1 stimulation of MCF-7 cells has led to increased pro-metastatic effects in the cells, without further elevating STAT3 or STAT1 activation. These findings support the possibility that STAT3 and STAT1 activation cooperates with additional signaling pathways that are induced by PD-L1 stimulation of the cells.
Our findings on the roles of STAT3 and STAT1 in mediating PD-L1 activities in breast cancer indicate that STAT3 and STAT1 cannot act independently from each other, and suggest that intersections exist between these two transcription factors in mediating PD-L1-induced effects. Indeed, concomitant roles for STAT3 and STAT1 were found in MCF-7 and MDA cells, and STAT3-STAT1 dimers were identified in the latter [
17]. The binding of phosphorylated Y-705-STAT3 to PD-L1 was also observed under hypoxic conditions in MDA cells [
35].
Moreover, our findings on STAT3 and STAT1 activation add one more layer to previous observations connecting these two transcription factors to other aspects of the PD-L1/PD-1 axis. The powerful activities of interferon γ, as the prime inducer of PD-L1 expression, were found to be mediated by STAT3, and in addition, STAT1 activation was found to up-regulate the expression of PD-L1 by tumor cells [
16,
17,
18,
19]. Furthermore, pSTAT3 gene signatures were correlated with PD-L1 expression in tumor cells and immune cells of breast cancer patient tumors, and the activated forms of STAT3 and STAT1 were significantly associated with the expression of PD-L1 in breast cancer patients [
18,
19].
As noted above, we have also provided the first evidence for the cardinal roles played by N-linked glycosylation at all four PD-L1 glycosylation sites in mediating the cell-autonomous pro-metastatic functions of PD-L1 in tumor cells. First, our studies with kifunensine demonstrated that the activation of STAT3 and STAT1 was effectively down-regulated by the inhibitor, alongside with a reduced tumor cell pro-metastatic functions. The inhibitory effects of kifunensine on PD-L1 activities in the cancer cells suggest that the activity of α-mannosidase I, which is the main enzyme inhibited by kifunensine [
29,
30], is pivotal for PD-L1-induced effects in breast tumor cells; our data also imply, via the use of swainsonine, that N-linked glycosylation by α-mannosidase II [
29,
30] plays a role in this process. This possibility was supported by the TCGA patient dataset findings, demonstrating the associations between high α-mannosidase II (MAN2A1) expression levels and reduced survival in breast cancer patients.
Then, analysis of the STAT3 and STAT1 activation patterns in PD-L1 that had been mutated at each of the four glycosylation sites individually demonstrated a complex connection between N-linked glycosylation of the different PD-L1 sites and their ability to induce pro-metastatic activities through STAT3 and STAT1 activation. Our findings revealed that in both MCF-7 and MDA cells, the N219 residue was the most crucial site for inducing STAT3 and STAT1 activation. Although less prominent, N-linked glycosylation of the N192 and N200 sites (depending on the cell type), but not of N35, also regulated STAT3 and STAT1 activation. Therefore, the N35 site probably regulates other signaling pathway/s or transcription factor/s that mediate/s the pro-tumoral activities of PD-L1 in the cancer cells.
Our findings on PD-L1 variants that had been mutated in each of the four N-linked glycosylation sites of PD-L1 indicate that all sites are required for PD-L1 pro-metastatic activities, and that the lack of glycosylation of one of the sites cannot be compensated for by any of the other three glycosylation sites. Moreover, our observations in the animal model system demonstrated that single alteration of even one of the four PD-L1 N-glycosylated sites—exemplified by the N35 and N219 sites—was sufficient to reduce the cell-autonomous T cell-independent pro-tumorigenic functions of PD-L1, also in vivo. These observations add significantly to a previous study using a PD-L1 variant mutated at
all four N-linked glycosylation sites
concomitantly, which demonstrated reduced growth of mouse TNBC tumors in a
T cell-dependent system [
21]. A mechanism connecting N-linked glycosylation of PD-L1 to cancer growth and anti-tumor T cell responses was also described in a syngeneic model system of EGF-induced responses [
22]. Overall, our findings reveal that in addition to previous reports on the roles attributed to N-linked glycosylation of PD-L1 in regulating anti-tumor T cell activities [
21], this post-translational process is required also for PD-L1 to act directly on the tumor cells in a cell-autonomous manner, leading to their increased aggressiveness.
Overall, our study indicates that N-linked glycosylation of PD-L1 is required for the ability of PD-L1 to induce cell-autonomous pro-metastatic activities in breast cancer cells, mainly by regulating the activation of STAT3 and STAT1.
5. Conclusions
In this study, we have identified key roles for members of the STAT family, namely STAT3 and STAT1, in mediating PD-L1-induced cell-autonomous and pro-metastatic activities in breast cancer cells. The roles of STAT3 and STAT1 in this system agree well with the potential use of STAT proteins as targets in cancer therapy, as has been suggested and addressed in different studies and in clinical trials [
20]. However, the multifaceted roles of these two proteins in different cancer studies, and the fact that inhibition of STAT3 or STAT1 may indirectly affect many other processes, suggest that STAT3 and STAT1 inhibitors may give rise to unpredicted and undesired effects; additional limitations may also exist when STAT3 and STAT1 are targeted in clinical protocols combining ICBs [
20,
36,
37]. Similarly, blocking enzymatic steps that are involved in N-linked glycosylation may lead to considerable adverse effects, because glycosylation is essential for the activity of many cellular proteins [
38,
39].
Instead, the fact that PD-L1-induced activation of STAT3 and STAT1 requires, as a prerequisite, that PD-L1 will be N-linked glycosylated, provides novel perspectives to PD-L1-directed therapy in breast cancer patients. Here, to follow up on observations showing that PD-L1 is N-linked glycosylated in breast cancer patient samples [
23,
24], it is important to validate the connections between the STAT3 and STAT1 proteins with the N-linked glycosylation status of PD-L1 in breast cancer patients and in mice tumor samples. The associations between these parameters and the survival rates of patients could also identify the degree of association between PD-L1 N-linked glycosylation and STAT3/STAT1 and therapeutic results.
Thus, further studies along these lines may lead to improved therapy approaches, directed to the glycosylated moiety of the protein. Recently, a 2018 paper by Hung and colleagues demonstrated that an antibody directed to the N192 and N200 glycosylation sites of PD-L1 (STM108) inhibited PD-L1 interactions with PD-1, and led to PD-L1 internalization and degradation; moreover, when the antibody was conjugated to an anti-mitotic drug, it reduced tumor growth in an animal model system using mouse breast cancer cells [
21]. These findings agree well with our observations on the roles of N192 and N200 in regulating PD-L1 functions and STAT3/STAT1 activities in breast cancer cells, and offer new therapeutic opportunities that rely on targeted inhibition of the most active form of PD-L1, the one which is fully N-linked glycosylated.
Introducing antibodies to N-linked glycosylated PD-L1 in clinical use will require assessment of PD-L1 glycosylation levels in each patient biopsy beforehand. Such an analysis may serve not only therapeutic needs, but may also assist in stratifying patients for ICB treatment based on PD-L1 N-linked glycosylation levels. Recent reports indicate that the N-linked glycosylated form of PD-L1 is not well recognized by antibodies during diagnosis or by treatment with ICBs directed at PD-L1 [
23,
24,
25,
26], suggesting that such patients may have relatively lower responses to the therapy. In practice, it is possible that some of these patients do express PD-L1, and that the antibody binding site/s are hindered due to N-linked glycosylation. Such patients could benefit from diagnostic measures and therapies using antibodies to the glycosylated moieties rather than antibodies directed at the core protein itself.