Background
Interferon regulatory factor 8 (IRF8) is a transcription factor of the interferon regulatory factor family that plays an essential role in the development and maturation of myeloid cells and in the expression of genes involved in antigen capture, processing, and presentation as well as in the activation of these cells in response to IFN-γ, IFN-β, and pro-inflammatory stimuli [
1‐
6].
IRF8 has also been found to be expressed and functional in non-hematopoietic cells such as epithelial cells [
7,
8]. Studies of IRF8 mRNA levels in normal human colon and colorectal carcinoma (CRC) revealed that IRF8 is downregulated in tumor tissues as compared to non-malignant counterpart tissues. Moreover, analysis of The Cancer Genome Atlas (TCGA) datasets in CRC revealed that the IRF8 promoter DNA is more methylated in tumors than in normal colon tissue, indicating that its expression is in part controlled by epigenetic mechanisms [
9]. Similar data have been reported for nasopharyngeal carcinoma, esophageal cancer, breast cancer, cervical cancer, and especially in lung carcinoma [
10,
11].
In a model of inflammation-induced spontaneous colon cancer, mice with IRF8 deficiency specifically in colon epithelial cells contained an increased percentage of Ki67
+ cells in the stem cell zones of the crypt as compared to wild type mice and developed more tumor nodules when subjected to azoxymethane-DSS cycles [
9]. In lung cancer cells, IRF8 negatively regulates Akt phosphorylation, inducing cellular senescence [
12]. In addition, inhibition of IRF8 expression levels has been found to increase tumor growth in lung cancer xenografts, indicating a role for IRF8 in progression of late-stage lung cancers [
12]. Thus, it has been proposed that IRF8 could be acting as a tumor suppressor via mechanisms that need to be better elucidated. Accordingly, IRF8 has been found to act as tumor suppressor in other solid tumors and hematopoietic malignancies [
13‐
15].
Less is known regarding the role of IRF8 in BC cells, but reports are consistent with a tumor suppressive role. For example, the IRF8 promoter appeared severely hypermethylated in metastatic BC patients [
16]. Induced IRF8 expression in BC cell lines in vitro suppressed cell proliferation, colony formation, and cell migration and invasion and induced apoptosis by enhancing the pro-apoptotic effect of IFN-γ and suppressing β-catenin signaling [
17]. Also, consistent with the role of IRF8 as a tumor suppressor, high expression of IRF8 has been significantly associated with longer overall survival in ER-negative BC [
17].
Since IRF8 is downstream IFN/STAT1 signaling and there are certain therapeutic regimes that require this signaling axis for its efficacy, we investigated if its expression can predict therapeutic response. In this study herein, we support utilization of IRF8 expression as a potent biomarker not only for prognosis, but also for predicting therapeutic response in ER-negative BC subtypes (HER2+ and TNBC). In particular, we report that IRF8 expression predicts the complete pathological response to monoclonal antibody therapy (trastuzumab) or to certain combinations of chemotherapy such as FAC (fluorouracil, adriamycin, and cytoxan) in these BC subtypes. Moreover, analysis of immune contexture of tumor tissues indicates a strong correlation between activated and effector CD8+ T cell infiltration and tumoral IRF8 expression.
Our results raise new questions regarding the cross-talk between immune infiltrates and IRF8 expression in neoplastic cells and demands new efforts in studies aiming to understand how IRF8 expression levels are regulated and how they can be therapeutically manipulated.
Discussion
Breast cancer is the leading cause of death from cancer in women worldwide [
26]. When breast cancer is detected in early stages, treatment efficacy is superior resulting in improved clinical outcomes. However, if the diagnosis occurs in advanced stages or the disease, and depending on the molecular characteristics of the tumor, prognosis is usually less encouraging. The molecular characterization of breast cancers through determination of biomarkers (e.g., ER, PR, HER2, Ki67) has been useful both as a predictive metric, but also for designing therapeutic course [
27‐
29]. Moreover, assessment of tumor-infiltrating lymphocytes (TILs) has an important prognostic role in TNBC [
30,
31] and in HER2+ disease [
32,
33]; however, patients stratified according to any of these attributes experience a wide range of clinical outcomes, supporting the need to identify and validate additional biomarkers easily evaluated in clinical labs and that could add precision in distinguishing clinical progression and therapeutic response of each patient.
In this study, our results support the proposition that IRF8 expression should be considered as a potent biomarker in ER-negative BC patients (TNBC and HER2+ patients). We examined expression of the IRF8 transcription factor at the protein level via immunohistochemistry where we validated, results obtained by in silico gene expression analysis, demonstrating that the percentage of tumors with high neoplastic cell IRF8 expression diminishes significantly as disease progresses. Moreover, none of the invaded sentinel nodes exhibited expression of nuclear IRF8. Our data corroborate previous results obtained by others reporting that IRF8 expression correlates inversely with IRF8 promoter methylation [
7,
10,
11,
17], indicating that epigenetic changes are likely a major cause of IRF8 downregulation in BC cells and impact disease progression and metastasis. However, among ER-negative grade 3 tumors, there is a small subgroup which still has high IRF8 expression and consequently presents a better outcome. All together, these data indicate that IRF8 can be detected in some cases without epigenetic therapy and that the methylation silencing does not occur in all samples. Curiously, we found that high expression of IRF8 correlated with longer overall and relapse-free survival only in TN and HER2+ tumors, demonstrating that IRF8 is a prognostic marker for patients harboring ER-negative tumors.
In ER+ tumor samples, the presence of IRF8 appears to have no effect in discriminating the patient outcome, indicating that in this subgroup of patients, which bear mostly grade 1–2 tumors, there are other features or factors that are leading the tumor growth and that could act as better prognostic factors. In contrast, when ER signaling is absent, expression of this transcription factor becomes important, probably as a tumor suppressor gene, with impact in OS and RFS. Indeed, Luo et al. [
17] showed that IRF8 performed as a candidate tumor suppressor by inducing G2/M phase cell cycle arrest and apoptosis in MDA-MB-231 (ER-) and T47D cells (ER+), and also by inhibiting cell migration and invasion in MDA-MB-231, but not in T47D cells. The effects of IRF8 thus seemed to be more pronounced in ER-negative breast cancer cells, supporting its prognostic value in this particular subtype.
Moreover, our analysis revealed that IRF8 expression predicts complete pathological response to trastuzumab therapy in HER2+ tumors, as well as to FAC chemotherapy in TNBC. It is known that anthracyclines stimulate rapid production of type I interferon (IFN-I) by malignant cells after activation of innate immune receptors by endogenous molecules released by dying tumor cells [
34]. Thus, FAC therapy depends on endogenous IFN-I for success and consequently requires IRF8 expression for IFN-I induction as it binds to the promoters of IFN-I genes and participates in subsequent IRF7-mediated amplifying phase of IFN transcription [
35]. These data explain why patients having low expression of IRF8 also exhibit a compromised response to FAC chemotherapeutic regimens and why success of this therapeutic approach relies on IRF8 levels. Likewise, Perez and colleagues reported a profile of 14 genes encoding various immune functions, which included IRF8, that were associated with a significantly improved relapse-free survival in patients with HER2+ BC treated with trastuzumab [
36], supporting our data assigning IRF8 a predictive role to successful response to trastuzumab. Our findings are in line with several additional reports demonstrating that an IFN-I related signature predicts clinical response to anthracycline-based chemotherapy in several independent cohorts of patients with BC [
34,
37]. Cyclophosphamide also impacts induction of antitumor immunity in vivo: it promotes expansion of CD8a
+ DCs through induction of endogenous IFN-I and induces immunogenic tumor cell death, stimulating tumor infiltration and the engulfment of apoptotic material by DC to cross-prime CD8
+ T cells [
38]. Furthermore, sustained activation of the IFN-β/IFNAR/IRF7 signaling axis in chemotherapy-treated ER-negative BC cells instigates a state in which tumor cells are likely controlled by immune-mediated mechanisms [
39].
Altogether, these data indicate that detecting IRF8 presence as a predictive factor by a simple immunohistochemistry assay could be a useful tool to be considered in clinical practice. On the other hand, presence of CD8
+ cytotoxic T cells has been associated with better clinical outcomes in patients with HER2+ and TNBC [
40,
41]. Expression of cancer-testis antigens such as MAGE-A and NY-ESO-1 are preferentially expressed in TNBC or high-grade and ER-negative BC [
42‐
44], and their presence has been correlated with high levels of CD8
+ TILs [
45]. Our data indicate that a higher abundance of activated and effector CD8
+ T cells are infiltrating tumors with higher expression of IRF8 in ER-negative BC samples. We hypothesize that patients expressing higher levels of IRF8 could present sustained levels of IFN-I within tumors, which would be needed to maintain fitness of the antitumor immune response, and consequently improved disease outcome. In colon epithelial cells, IRF8 has been shown to regulate the expression levels of osteopontin which is a potent suppressor of CD8+ T cell response and consequently proposed to be a new checkpoint. IRF8 binds to the ISRE elements at the
Spp1 promoter and represses its expression. Osteopontin levels are elevated in human colon cancer patient periphery, correlating with decreased disease-specific survival [
46]. A similar mechanism could be involved in BC patients. However, a fraction of TN and ER-negative tumors that were included in the study herein fell in the more aggressive grade 3 category which, according to data, silence IRF8 expression by epigenetic mechanisms, although at the same time, are heavily infiltrated. Thus, outcome of these patients likely reflects a balance between the lack of the tumor suppressor functions of IRF8 and the antitumor immune response, which in these tumors could arise partly from the elevated expression of tumor-associated antigens. Along these lines, IRF8 functions extrinsically as a GM-CSF repressor in T cells [
47]. Loss of IRF8 function increases secretion of both G-CSF and GM-CSF, ligands that upon high affinity cognate receptor activation on myeloid progenitor cells promote myeloid-mediated T cell suppression, in alignment with the inverse relationship between IRF8 levels and these cells observed in patients with BC and coinciding with poorer prognosis [
48].
Acknowledgements
We thank the members of the Coussens Lab (lab manager, researchers, bioinformatician) for all support. We also thank Vanesa Losi and Fernanda Ponce for their technical assistance in immunohistochemistry assays. GG received a short-term fellowship from Fundación para el progreso de la Medicina to perform research duties at the Department of Cell, Developmental, and Cancer Biology, Knight Cancer Institute, Oregon Health & Science University, Portland, USA.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.