IL26 is a unique amphipathic member of the IL10 family of cytokines that participates in inflammatory signaling through a canonical receptor pathway. It also directly binds DNA to facilitate cellular transduction and intracellular inflammatory signaling. Although IL26 has almost no described role in cancer, our in vivo screen of inflammatory and cytokine pathway genes revealed IL26 to be one of the most significant inflammatory mediators of mammary engraftment and lung metastatic growth in triple-negative breast cancer (TNBC). Examination of human breast cancers demonstrated elevated IL26 transcripts in TNBC specimens, specifically in tumor cells as well as in Th17 CD4+ T cells within clinical TNBC specimens. IL26 did not have an autocrine effect on human TNBC cells, but rather its effect on engraftment and growth in vivo required neutrophils. IL26 enhanced mouse-derived DNA induction of inflammatory cytokines, which were collectively important for mammary and metastatic lung engraftment. To neutralize this effect, we developed a novel IL26 vaccine to stimulate antibody production and suppress IL26-enhanced engraftment in vivo, suggesting that targeting this inflammatory amplifier could be a unique means to control cancer-promoting inflammation in TNBC and other autoimmune diseases. Thus, we identified IL26 as a novel key modulator of TNBC metastasis and a potential therapeutic target in TNBC as well as other diseases reliant upon IL26-mediated inflammatory stimulation.
These findings identify IL26 as a unique, clinically relevant, inflammatory amplifier that enhances TNBC engraftment and dissemination in association with neutrophils, which has potential as a therapeutic target.
Triple-negative breast cancers (TNBC) are a heterogeneous group of cancers that lack activating mutations in typical proto-oncogenes, making targeted therapy difficult compared with other breast cancers (1, 2). However, many TNBC tumors display an inflammatory signature composed of cytokines and chemokines that directly stimulate tumor growth at the primary site, protect tumor initiating cells, enhance dissemination and engraftment in new niches, and ultimately promote overt metastasis (3–6). Cytokines and chemokines produced by TNBC cells also recruit tumor-enhancing innate immune cells to the primary site while systemically mobilizing innate immune cells to distant niches. We and others have previously shown that various cytokine networks are essential for tumor progression, both clinically and in animal models, particularly through a combination of TNBC cell–derived IL6 and IL8 (3). Furthermore, animal studies have concluded that inflammation from infiltrating innate immune cells plays a key role in enhancing tumor growth and metastasis (4). Once in the tumor microenvironment (TME), these immune cells are activated to produce even greater inflammation, which directly supports tumor growth, invasion, and subsequent metastasis (4, 6–9). These data are backed by numerous clinical studies of human TNBC that describe them as highly infiltrated with immune cells—including T cells, neutrophils, and monocyte/macrophage lineage cells—especially when compared with other subtypes of breast cancer (10–12). Nevertheless, which cytokine networks are most important for overall regulation of TNBC progression, or if there are any central regulators that have an outsized impact on patient survival, remain unknown.
In order to investigate the role of inflammation in TNBC and narrow down the list of candidates involved in TNBC progression, dissemination, and engraftment, we conducted a focused inflammatory screen of TNBC in vivo, similar to our previous in vivo screen of immune modulator genes (13). Specifically, we modeled multiple stages of breast cancer progression by direct injection into the mammary fat pad (MFP) for engraftment and growth and i.v. injection via lateral tail vein for dissemination, engraftment, and growth at a metastatic site (lung). These screens revealed IL26 as a significant and novel mediator of progression in both primary and disseminated niches and was validated in multiple TNBC cell lines as well as in breast cancer samples.
IL26 is a 19-kDa α-helical cytokine that belongs to the IL10 cytokine family (IL10, IL19, IL20, IL22, IL24, and IL26) and has no known homolog in mice (14). IL26 mRNA has most widely been detected in activated lymphoid cells including Th17 and natural killer cells (15), but reports of expression in a variety of other cell types are emerging, including monocytes, bronchial epithelial cells, fibroblast-like synoviocytes, and smooth muscle cells (14). IL26 canonically activates the Jak–Stat3 pathway through binding a heterodimeric complex of IL10RB and IL20RA (16, 17). This triggers expression of proinflammatory cytokines IL1β, IL6, GM-CSF, and TNFα in human monocytes and activates type I (β) and type II (γ) IFN, as well as induces IL8 and IL10 expression, in human epithelial cells (14). Recent evidence also suggests that the highly cationic amphipathic properties of IL26 result in novel functions, such as direct bactericidal activity through pore formation and bacterial-membrane disruption (18). In addition, the amphipathic nature of IL26 allows it to act as a cell-penetrating carrier molecule for DNA, particularly neutrophil extracellular traps (NET), giving extracellular DNA access to intracellular receptors such as STING and TLR9 (14, 18, 19). Though the lack of a mouse homolog has made mechanistic studies difficult, mounting clinical data suggests IL26 is involved in a host of human diseases. High serum levels of IL26 are observed in patients with contact dermatitis, rheumatoid arthritis, Crohn's disease, chronic hepatitis C infection, and severe pediatric asthma (14, 20). Finally, emerging evidence implicates a role of IL26 in various cancers. IL26 mRNA is elevated in biopsies of cutaneous T-cell lymphomas, IL26 directly promotes proliferation and survival of gastric cancer cells, and elevated IL26 expression is a poor prognostic indicator of both recurrence-free survival and overall survival of hepatocellular carcinoma after surgical resection (21–23).
The appearance of IL26 as a significant target in both shRNA screens, along with the apparent ability of human IL26 to modulate tumor progression in mice with no native homolog, warranted further investigation. Herein, we demonstrate the role of IL26 in TNBC engraftment and progression in mice and the conserved ability of IL26 to elicit DNA-mediated inflammation in mouse cells. Further, our study implicates the effect primarily depends upon neutrophils in the TME, potentially serving as an microenvironmental regulator in orchestrating tumor inflammation by enhancing neutrophil NET DNA stimulation of multiple inflammatory factors, which are collectively important for TNBC. Due to this critical role, we also investigated its therapeutic potential through IL26-specific vaccine-induced antibodies. In sum, our studies reveal a potentially significant role for IL26 in TNBC and suggest it as an actionable therapeutic target to suppress inflammation and inhibit TNBC progression.
Materials and Methods
Cell lines and reagents
MDA-MB-231, HEK293TW, SUM159, SUM149, MDA-MB-468, and 32DC3 cells were acquired from the ATCC and through the Duke Cell Culture Facility. Mouse E0771 cells were purchased from CH3 Biosystems (940001). MDA-MB-231-LM2 cells were a kind gift from Joan Massague (Memorial Sloan Kettering, New York, NY; ref. 24). All cells were cultured according to the ATCC and vendor specifications and tested to be free of Mycoplasma and other rodent pathogens (RADIL IMPACT III Test). For human cell lines, short tandem repeat DNA profiling was performed on parental stocks to verify their identity through the Duke DNA Analysis Sequencing Facility. Individual lines were expanded upon receipt for frozen stocks, and cells were used for experiments at ≤20 passages. Recombinant human IL26 (rhIL26) was acquired from either R&D Systems (cat 1870-IL) or produced and validated in collaboration with Genscript.
Library deconvolution and analysis
DNA from MDA-MB-231 shRNA containing library cells was isolated using QIAamp DNA mini kit and sent to Sigma for shRNA quantification. Log-transformed expression values were compared between lung tumors, MFP tumors, and initial (pre-engraftment) tumor cells using LIMMA version 3.14.4 on the software R. shRNAs with an absolute Log2 fold change greater than 1 and FDR values for empirically Bayes moderated statistics below 0.05 were considered to have statistically significant differential expression.
Lentiviral and adenoviral techniques
All lentiviral vectors were produced in 293T cells, using second-generation packaging plasmids and our previously described techniques (3). Viral stocks were concentrated by ultracentrifugation and utilized with 5 μg/mL polybrene to generate stable cell lines. Inducible shRNA vectors were generated as previously described (3), and CRISPR lentiviral vectors were produced using the plentiCRISPRv2 backbone (obtained from Addgene; ref. 25) and produced using standard methods. The chemokine–cytokine shRNA library was generated by combining a human cytokine–chemokine pLKO.1 library (Sigma SH0811, 106 genes targeted by 528 shRNA constructs) with a custom second library (78 genes targeted by 642 shRNA constructs) to cover all genes listed in the Kyoto Encyclopedia of Genes and Genomes cytokine-chemokine set. Viral shRNA library was tittered on MDA-MB-231 cells and utilized at a multiplicity of infection (MOI) of 1 with puromycin selection beginning 48 hours after infection. After 5 more days, MDA-MB-231 library cells were utilized in the in vivo screen. The OVA and IL26 adenoviral vectors were generated using Gateway cloning techniques. We first generated OVA and IL26 ENTRY plasmid clones, then recombined them with pAd-CMV5 vectors (Invitrogen) using LR clonase II. Vectors were linearized using PacI and adenoviral stock amplified using previously described techniques (26).
NETs were isolated from SCID-beige mouse neutrophils after enriching with the EasySep Mouse Neutrophil Enrichment Kit (StemCell Technologies). Purity of isolated neutrophils was >90% as analyzed by LY6G+ and CD11b+ FACS staining. NETosis was induced via 100 nmol/L PMA (Sigma) for 4 hours at 37°C. Plates were gently washed with PBS, followed by strong pipetting by centrifugation.
Peripheral blood mononuclear cell/splenocyte/neutrophil stimulation
Human peripheral blood mononuclear cells (PBMC) isolated from healthy donors, mouse SCID-beige splenocytes, or mouse 32DC3 neutrophil cells were plated in a 96-well plate with RPMI-1640 medium (Invitrogen) at 37°C with 5% CO2. rhIL26 (50 ng/mL) was incubated with 1 or 10 μg/mL of sheared DNA or isolated NETs in nuclease-free H2O for 30 minutes at 37°C for the IL26 to bind to the DNA. In some conditions, DNase (30 IU/mL) was preincubated with IL26 and/or DNA/NETs. DNA was isolated from MDA-MB-231, Sum159, and MDA-MB-468 cultured cells, combined, and sheared using a 30-second sonication pulse. The resulting solutions were added to cells in a 96-well plate and allowed to incubate for 24 hours, after which, conditioned medium was collected for ELISA.
Elispot assays were performed using our previously described technique (27). Briefly, splenocytes (500,000 cells/well) were plated in RPMI-1640 medium with 10% heat-inactivated FBS and stimulated with OVA peptide (SIINFEKL; 1 μg/mL; Sigma), lysed E0771-WT cells (10 k/well), or lysed E0771-IL26+ cells (10 k/well). PMA (50 ng/mL) and ionomycin (1 μg/mL; Sigma) were used as positive controls. Irrelevant HIV-gag peptide mix (2.6 μg/mL; JPT) was used as a negative control.
RNA ISH (RNAscope)
In situ detection of IL26 mRNA transcripts was performed on formalin-fixed, paraffin-embedded Cancer Diagnosis Program (CDP) Breast Cancer Progression Tissue Microarray sets using the RNAscope Multiplex Fluorescent Kit (323100; Advanced Cell Diagnostics; ref. 28) following the manufacturer's protocol. A more detailed description can be found in the Supplementary Methods.
qRT-PCR breast cancer cDNA arrays
Commercially available TissueScan Breast Cancer cDNA arrays were obtained from OriGene (Rockville). Each array (BCRT101, BCRT102, BCRT103, and BCRT104) contains cDNA from 48 samples comprised of normal breast tissue or stage I, IIA, IIB, IIIA, IIIB, IIIC, and IV breast cancer tissue, along with designations of tumor molecular subtype. Reactions were performed with iTaq Universal SYBR Green Supermix (BioRad) and normalized to β-actin using primers provided by the manufacturer.
In vivo tumor growth and metastasis
Female, 6- to 8-week-old SCID-beige (Taconic Biosciences) or C57Bl/6J (Jackson Labs) mice were implanted with tumor cells into the fourth inguinal MFP [MDA-MB-231 = 106 cells in PBS; MDA-MB-468 = 3 × 106 cells in 1:1 PBS with Matrigel (Corning); SUM159 = 5 × 106 cells in 1:1 PBS with Matrigel; SUM149 = 4 × 106 cells 1:1 PBS with Matrigel]. For E0771 cells, animals were either injected with 105 cells in the MFP or with 105 cells in 100 μL PBS directly into the lateral tail vein. Tumor growth was monitored biweekly by caliper to track tumor growth for up to 6 weeks or when reaching a terminal endpoint of 2 cm3, and volume was calculated as (length × width2)/2. For bioluminescence imaging, tumor-bearing animals were injected intraperitoneally with D-luciferin (100 mg/kg) and live, whole-body bioluminescence intensity was measured using an IVIS Kinetic (Perkin Elmer). Mice were then immediately sacrificed via CO2 inhalation, and lungs were isolated for further ex vivo imaging. For vaccine experiments, animals were vaccinated with 5 × 1010 viral particles of described adenoviral vectors via foot pad as described previously (28). To deplete neutrophils in vivo, 300 μg of anti-Ly6G (1A8) or isotype control antibodies (2A3; BioXCell) were injected i.p. into mice 24 hours before engraftment of tumor cells and again twice weekly i.p. for a total of six treatments. All studies with animals were approved by the Duke University Institutional Animal Care and Use Committee.
Single-cell RNA sequencing analyses
FACS-sorted CD3+ single-cell (scRNA-seq) data as unique molecular identifier (UMI) count matrix from two individual TNBC primary tumor samples (project approval number is “SEGMENT” 13/123) were obtained from Gene Expression Omnibus repository under the accession id GSE110686. The UMI count matrix was previously generated by the authors using Cell Ranger software (version 1.3.1) as provided by the 10xGenomics pipeline. A more detailed description of data processing and analysis can be found in the Supplementary Methods.
Human breast cancer dataset analysis
METABRIC and The Cancer Genome Atlas (TCGA) data were accessed and queried using the web-based cBioPortal software (2, 29–31), and data were visualized and statistics were performed using Graphpad Prism software. According to cBioPortal, copy-number data are derived by algorithms such as GISTIC or RAE, and the copy-number level per gene is reported as: −2 = deep deletion/possible homozygous deletion; −1 = shallow deletion/possible heterozygous deletion; 0 = diploid; +1 = low-level gain/few additional copies; and +2 = amplification.
All statistical analyses, unless otherwise noted, were performed using GraphPad Prism version 8 (GraphPad Software). Marks for significance include: *, P ≤ 0.05; **, P ≤ 0.01; ***, P ≤ 0.001; and ****, P ≤ 0.0001. Error bars represent SEM, tests between two groups were performed via t test, and tests between three or more groups were performed via ANOVA with Tukey post hoc correction. Survival analysis for all mouse xenograft studies was performed using Prism 8 (GraphPad). Survival plots were considered significantly different if the log-rank (Mantel–Cox) test resulted in a P ≤ 0.05.
An in vivo functional genomics screen using a custom-designed shRNA library comprised of 1,170 shRNAs (targeting all annotated human cytokines, chemokines, and their receptors) was initially used to identify inflammatory genes that contribute to TNBC cell engraftment and experimental metastasis in vivo using a well-defined metastatic human TNBC cell line (186 genes, Supplementary Table S1; ref. 32). Cells were infected with the library at an MOI = 1 to limit multiple integrations and selected using puromycin (1 week) to generate a stable library. MDA-MB-231 shRNA library–infected cells (1 × 106 to ensure ∼800× coverage of the library) were either implanted orthotopically into the MFP or injected i.v. into female SCID-beige mice and allowed to form tumors (Fig. 1A). Excised tumors from MFP (3 weeks) and lungs (4 weeks) revealed significantly selected shRNAs compared with the parental library (Fig. 1B); 120 shRNAs affecting TNBC cell progression in the MFP and 166 shRNAs in the lung (Fig. 1C). Further analysis established 35 shRNAs that were significant in both microenvironments (Fig. 1D; Supplementary Tables S2 and S3). In comparing shRNA candidates between the lung and MFP environments, multiple significant shRNAs targeting IL26 were identified, which was determined to be one of the most significant genes influencing tumor cell engraftment and colonization in both niches. IL26 is a human-specific gene without a homolog in mice and has no reported connection with breast cancer. Therefore, to verify that IL26 is a clinically relevant hit, the METABRIC dataset was analyzed (2), which revealed that IL26 DNA is amplified in approximately 3% of breast cancers (60 of 1,981 profiled), and amplification of IL26 is a poor prognostic indicator of breast cancer as a whole (Fig. 1E). This data suggested that IL26 could play a role in breast cancer progression and prompted further investigation.
The relative protein expression in multiple human breast cancer cell lines was also assessed for IL26 expression based on molecular subtype (Supplementary Fig. S1A). IL26 protein was detectable in TNBC lines compared with HER2+ BT474, ER+ MCF7 cell lines, and a 293T control, indicating that IL26 expression may be most abundant in TNBC. Next, mRNA expression in the TCGA dataset (n = 981) was interrogated by the clinically relevant molecular subtypes (31), which revealed IL26 was most highly expressed in Basal-like breast cancer (which is highly enriched for TNBC; ref. 33). This was statistically significant compared with both luminal A and luminal B subtypes, and narrowly missed the P ≤ 0.05 cutoff against Normal-like (n = 981 samples profiled, Fig 2A). Although not significant, HER2-enriched tumors appeared to also contain elevated IL26 expression. To validate these findings, an independent panel of breast cancer mRNAs was interrogated for IL26 expression by qRT-PCR. Significantly elevated IL26 was observed in all breast cancers (n = 160, Supplementary Fig S1B), although unlike the TCGA results, the expression was not specifically higher in TNBC compared with ER+ or HER2+ breast cancer (Supplementary Fig. S1B and S1C). As these and TCGA data are comprised of bulk sequencing data, tumor-specific expression of IL26 in formalin-fixed, paraffin-embedded CDP Breast Cancer Progression Tissue Microarrays was analyzed. Critically, these histologically defined assessments revealed significant upregulation of IL26 in TNBC tumors compared with other breast cancer tumors (Fig. 2B), which was observed directly in tumor cells (non-TNBC, n = 106; TNBC, n = 23; Fig 2C). The microarrays also contained a small number of matched, tumor-adjacent normal mammary epithelium, and pairwise assessment indicated a specific and significant increase of IL26 in tumor cells compared with the normal cells (Supplementary Fig. S1D). To identify other cells expressing IL26 in TNBC, single-cell RNA sequencing of CD45+ immune cells from two patient TNBCs was utilized. The analyses revealed that IL26 was expressed in a CD3+CD4+IL17+ T-cell cluster in TNBC (Fig. 2D and E; Suppl ementary Fig. S2), directly consistent with recently published reports that associate Th17 T cells with IL26, neutrophils, and metastasis (18, 34). Collectively, these data confirmed IL26 expression in breast cancer cells specifically, indicated that TNBC expresses significantly more IL26 than other molecular subtypes, and implicated Th17 cells within tumors as another source of IL26 in TNBC.
After confirming that the shRNAs selected in the screen could all significantly suppress IL26 expression (Supplementary Fig. S3), IL26 knockdown (KD) MDA-MB-231 cells were generated to determine the functional effect of IL26 in TNBC (16, 22). Although IL26 suppression did not alter proliferation or anchorage-independent growth rates in vitro (Supplementary Fig. S4A and S4B), it did significantly delay tumor formation and growth in the MFP (Fig. 3A and B). In addition, mice injected with control cells had a greater proportion of lungs positive for luciferase (4/10) compared with mice injected with IL26 KD cells (1/10; Supplementary Fig. S4C), though this did not reach statistical significance. In addition, i.v. injections of control or IL26 KD cells into SCID-beige mice revealed that this effect was not niche specific as IL26 KD reduced experimental metastasis to the lungs (Supplementary Fig. S4D and S4E). Although some measure of IL26 suppression persisted in vivo, IL26 expression was strongly selected for in both control and shRNA KD tumors compared with cell lines before injection, further suggesting that IL26 is a tumor-promoting cytokine (Supplementary Fig. S5A and S5B). Notably, IL26 KD tumors expressed significantly more inflammatory cytokine mRNAs known to be downstream of IL26, such as IL6 and IL8 (Supplementary Fig. S5C). Whether this reflects compensatory selection for inflammatory cytokines downstream of IL26 or the temporal result of IL26 expression returning by the end of the experiment due to the limitations of shRNA KD, or possibly both, is unclear. IL26 KD cells were also generated for two additional TNBC lines (SUM159 and MDA-MB-468 cells) and one TNBC inflammatory breast cancer (IBC) cell line (SUM149). As before, suppression of IL26 did not alter proliferation in vitro (Supplementary Fig. S6) but did suppress engraftment and growth of all types of TNBC tumors in vivo (Fig. 3C–E). Finally, the effect of IL26 in murine tumors was assessed by overexpression in mouse TNBC E0771 cells (Supplementary Fig. S7). In this setting, IL26 conferred a significant growth advantage in both SCID-beige (Fig. 3F) and immune competent mice (Supplementary Fig. S8), strongly suggesting that IL26 was acting as a paracrine factor through a mechanism conserved in mice.
The reported canonical signaling pathways for IL26 were initially investigated to determine the possible mechanism behind the effects of IL26 in vivo. Surprisingly, canonical receptors (IL20RA and IL10RB) were not detectable on MDA-MB-231 cells by flow cytometry, and only minimal expression was observed by qRT-PCR (Supplementary Fig. S9A; ref. 16). However, artificial expression of IL20RA and IL10RB in MDA-MB-231 cells resulted in Stat3 signaling upon stimulation with rhIL26, thus demonstrating that this pathway is functionally intact and that the lack of canonical signaling is likely due to a lack of receptor expression (Supplementary Fig. S9B). Furthermore, Stat3 activity was undetectable when mouse IL10RB and IL20RA counterparts were artificially expressed in 293T cells (Supplementary Fig. S9C and S9D), suggesting the in vivo effects observed were not likely due to interaction with mouse IL10RB/IL20RA receptors.
Recent studies suggest that IL26 is an amphipathic, cationic cytokine that directly binds extracellular DNA (and human NET DNA) and acts as a carrier molecule to mediate DNA entry into cells for intracellular DNA stimulation of inflammation (18, 19). Because this function of IL26 would not necessarily be dependent on species-specific receptors, we hypothesized that IL26 could be functioning as a mediator of DNA-induced inflammatory signaling in the TME. To test this hypothesis, primary murine monocytes/granulocytes (SCID-beige splenocytes, Supplementary Fig. S10A) or mouse 32DC3 neutrophil cells were stimulated overnight with rhIL26, DNA, or DNA + rhIL26. In these experiments, rhIL26 or DNA alone were unable to induce significant cytokine production; however, the combination of rhIL26 and DNA significantly induced IL6, CXCL1, and IL1β in mouse splenocytes (Fig. 4A) and IL6 and CXCL1 in mouse neutrophils (Fig. 4B). Similar effects were seen in human PBMCs, although there was some stimulatory effect of IL26 alone (Supplementary Fig. S10B). Importantly, IL26 DNA binding was confirmed similarly to previous groups (Supplementary Fig. S10C) and preincubation of DNase to IL26+DNA tempered mouse splenocyte stimulation, indicating that these effects are due to DNA-bound IL26 (Supplementary Fig. S10D; ref. 19).
Extracellular DNA alone is not known to induce inflammatory responses in vivo, but DNA NETs have recently been described to promote TNBC metastasis (7, 35, 36). Binding of rhIL26 to NETs produced by mouse neutrophils (Supplementary Fig. S11) was first confirmed by confocal microscopy (Fig. 4C). In addition, stimulation of mouse splenocytes with rhIL26 alone, rhIL26+NETs, NETs alone, or rhIL26+NETs pretreated with DNAse demonstrated that IL26 induces a strong cytokine response (IL6, Cxcl1) when combined with NETs, which could be diminished with pretreatment of DNase (Fig. 4D). These data reveal that human IL26 interacts with mouse NETs, and that IL26+NETs are a potent inducer of inflammatory signaling in mouse cells. Finally, to confirm that neutrophils (and thereby NETs) mediate the effects of IL26 in vivo, neutrophils were depleted by anti-Ly6G antibody treatment of SCID-beige mice harboring MDA-MB-231 tumors. Antibody treatment significantly depleted neutrophils (Supplementary Fig. S12) and significantly suppressed tumor growth relative to isotype-treated mice (Fig. 4E). Notably, this treatment did not alter the growth of shIL26 cells (Fig. 4E), suggesting that neutrophils have little effect in the absence of IL26. Collectively, these data suggest that tumor-derived IL26 binds DNA to stimulate proinflammatory cytokines in innate immune cells to promote an inflammatory TME.
The ability of IL26/DNA to elicit inflammatory cytokines in vitro, along with the selection of these cytokines/chemokines over time in tumors in vivo (especially when IL26 was suppressed), suggested that these IL26-induced cytokines could play a key role in early engraftment of TNBC (14). Notably, the highly metastatic MDA-MB-231 LM2 subclone (24) express significantly elevated levels of these IL26-inducible cytokines (IL6, IL8, and CXCL1), supporting this hypothesis (Supplementary Fig. S13A–S13C). To formally test this, several of these cytokines/chemokines (IL6/IL8/CXCL1) were directly suppressed using inducible shRNA vectors in TNBC cells, which we have previously published (Supplementary Fig. S13D–S13F). Although single-gene suppression did not alter engraftment to the lung after i.v. injection, tandem suppression significantly reduced TNBC dissemination and growth in the lung (Fig. 5A and B). This effect was temporary, however, and tumors with coordinately suppressed cytokines eventually expanded in the lungs, potentially indicating insufficient gene suppression. These limitations led to generation of CRISPR-based knockouts (KO) completely deficient in expression of IL6, IL8, and CXCL1 (Crispr3x; Supplementary Fig. S14A). Again, mammary engraftment and experimental metastasis was suppressed (Fig. 5C and D) and was comparable with the single suppression of IL26. Importantly, isolation of these outgrowths revealed that tumor cells did not reacquire expression of these genes (Supplementary Fig. S14B). To determine if IL26 expression and stimulation of other nontumor cells could rescue this deficit, IL26 was overexpressed in MDA-MB-231 Crispr3x cells. Notably, IL26 overexpressing Crispr3x tumors grew significantly faster in the MFP compared with both control cell lines (Fig. 5E). Furthermore, lungs from mice bearing Crispr3x + IL26 tumors had significantly more de novo metastases compared with controls, and the level of metastasis was not correlated with the size of the tumor (Fig. 5F and G; Supplementary Fig. S14C). Collectively, these data indicate that the impact of IL26 is not limited to these specific cytokines produced by tumor cells and suggests that its enhancement of tumor engraftment and dissemination may be mediated through other cell types in the TME.
Assessment of inflammatory signaling was also expanded to patient samples by analyzing the METABRIC dataset for IL26 and associated cytokines/chemokines (IL6, IL8, and CXCL1). This IL26 signature, consisting of the IL26, IL6, IL8, and CXCL1 genes, was significantly elevated in TNBC, elevated in HER2+/ER– breast cancer, and low in ER+ and HER2+ breast cancer (Fig. 6A). Because neutrophils were critical for IL26 function in vivo, if an established neutrophil signature consisting of 23 documented neutrophil selective genes (37–40) would also be elevated in TNBC and associate with this IL26 network in human breast cancer (METABRIC) was determined. These analyses revealed that the neutrophil gene signature was also significantly enriched in TNBC (Fig. 6B), and that these signatures are significantly correlated in patients (Supplementary Fig. S15). Furthermore, there was an inverse relationship between 10-year survival and expression of both the IL26 network (P = 0.033) and neutrophil signature (P < 0.00038), suggesting that these networks may contribute to metastasis and poor survival in human TNBC (Fig. 6C and D). In sum, these data demonstrate that an IL26 inflammatory network is associated with TNBC, that a neutrophil signature is enriched in TNBC in association with the IL26 network, and that both correlate with breast cancer survival.
Finally, the therapeutic potential of targeting IL26 in TNBC was tested. We were unable to identify a specific inhibitor or antibody to block IL26 in vivo; therefore, we generated an adenoviral vaccine to immunologically target IL26, as done previously (28). To determine the immunologic potential of this vaccine, C57Bl/6 mice were vaccinated and assessed for IL26-specific antibody and T-cell responses. Although Ad-IL26 vaccination elicited significant IL26-specific antibody responses compared with control vaccination (Fig. 7A), only modest (nonsignificant) T-cell responses to IL26+ cells were observed (Supplementary Fig. S16A). Importantly, both groups of mice produced equivalent anti-Ad antibody responses, indicating equivalent injection (Supplementary Fig. S16B). Based on these results, vaccine-induced antitumor responses to IL26 versus a control Ad vector (Ad-OVA) were tested. In this study, one cohort of mice was vaccinated 2 weeks prior to tumor implantation (Prevention group), and another cohort was vaccinated after implantation (Treatment group). Both prevention and treatment administration of the Ad-IL26 vaccine significantly suppressed the engraftment and growth of IL26 + E0771 cells compared with control vector (Fig. 7B). Notably, IL26 expression was elevated after implantation in vivo (Fig. 7C; Supplementary Fig. S6) compared with in vitro passaged controls, again suggesting selection for IL26 expression or the stimulation of IL26 from factors in the TME as with MDA-MB-231 cells. In addition, modest IL26 expression from the largest Ad-IL26 vaccinated tumor that yielded sufficient protein to analyze was noted by Western blot (Fig. 7C; Supplementary Fig. S6). Collectively, these studies suggest that antibodies and/or vaccines targeting IL26 may have a therapeutic benefit in TNBC treatment.
Although inflammation is known to play a key role in the progression and growth of TNBC, the identities of the most critical cytokines and chemokines remain unclear (3, 41, 42). To decipher the impact of inflammatory genes important for TNBC progression in different native and metastatic microenvironments, we utilized an in vivo–focused shRNA screen of cytokine/chemokine pathways that identified IL26 as one of the most significant tumor-expressed cytokines in both orthotopic and disseminated tumors. As a unique human gene without a murine homolog and little association with cancer, we first confirmed the clinical relevance and expression of IL26 in TNBC by multiple means, including analysis of TCGA and METABRIC datasets, qRT-PCR of breast cancer tissue and through RNAscope analysis of human breast cancer tissue microarrays. These studies revealed that human breast cancer cells directly express IL26, which is elevated in TNBC compared with other subtypes. In addition, using scRNA-seq data, we also identified significant IL26 expression in a Th17 subpopulation based on human breast cancer single-cell RNA sequencing, thus replicating similar results from other groups (18). To our knowledge, this is the first report demonstrating IL26 expression in breast cancer or TNBC and suggests that this cytokine may also play a role in prompting other types of breast cancer.
The conserved ability of IL26 to promote tumor progression using multiple TNBC and IBC cell lines in vivo was in some regards surprising, especially in light of these cells lacking canonical IL26 receptors and no discernable direct effect of IL26 in vitro (based on Stat3 reporter assays). Although we found that IL26 did not elicit Stat3 signaling through the mouse homologs of IL26 receptors (IL20RA and IL10RB), it is possible that IL26 is acting in part through other noncanonical receptors. Alternatively, the ability of IL26 to bind DNA/NETs and induce inflammation was intriguing, especially because NETs have been shown to promote breast cancer metastasis and exit from dormancy (7, 43). We confirmed that this mechanism is indeed conserved, as human IL26 and mouse NETs stimulated the secretion of multiple inflammatory cytokines from mouse (and human) immune cells. Using a depletion strategy, we also found that neutrophils mediate the IL26 effect in vivo and that their depletion only suppresses tumor growth of human TNBC tumors in the presence of IL26, whereas growth was unaffected in tumors with suppressed IL26. These data support that IL26 is acting through NETs in vivo to promote an inflammatory TME, and future studies to define the exact mechanisms of this process are ongoing.
Our observation that IL26 is associated with TNBC, typified by inflammation in the TME, along with the induction of inflammatory cytokines in mouse cells by IL26, led us to hypothesize that these downstream cytokines also contribute to tumor dissemination and growth. In vivo experiments revealed that IL26 induced cytokines (such as IL6, IL8, and CXCL1), collectively contributed to TNBC metastatic dissemination to the lungs, comparable with IL26 suppression alone. Moreover, IL26 overexpression was able to rescue (and even promote metastasis of) IL6, IL8, and CXCL1 triple KOs cells, indicating that IL26 plays a more central, individual role in mediating local inflammation through the stimulation of multiple inflammatory genes. This could suggest that IL26 may be more significant in human TNBC, where multiple cells express IL26, such as Th17 CD4+ T cells. Notably, we found that elevated expression of an IL26 inflammatory axis or neutrophil signature inversely correlates with survival in breast cancer and was positively associated, supporting a potential link with clinical metastatic progression. Furthermore, both the IL26 inflammatory axis and neutrophil signature are highest in TNBC, which matches the TCGA data for IL26 mRNA expression in TNBC and to a lesser extent, in HER2+ breast cancer. As several studies have demonstrated that IL26 is highly expressed in inflammatory environments (such as those found in primary and metastatic tumor lesions), as well as in leukemias where it was originally cloned (15, 44, 45), it may play a significant role in dissemination, survival, and growth in metastatic niches of other cancers as well.
Although many studies have implicated neutrophils in breast cancer metastasis and progression (8, 46–48), it is unclear how these effects are mediated. The elegant study by Park and colleagues demonstrated that DNA NETs from neutrophils served as the dominant mechanism through which neutrophils exerted their prometastatic effect on tumors (7). However, it was not clear how these NETs elicited changes in tumor metastasis, although the authors speculated that a chemotactic factor, such as HMGB1 (which binds DNA), may be associated with the NETs. Our work suggests that IL26 produced by tumor cells (as well as possibly Th17 T cells) may be the critical factor through which DNA NETs, as well as other DNA in the TME, stimulate inflammatory cytokines/chemokines (such as IL6, IL8, CXCL1, and others). This is supported by other studies of IL26 in infection, where IL26 was demonstrated to be a principal mediator of inflammation and innate immune recruitment in the lungs. Collectively, these data support a role for IL26 as a major regulator of DNA-elicited inflammation and may explain why suppression of this single cytokine had a significant effect in TNBC engraftment screens in two different and distinct microenvironments.
Therapeutically, our studies suggest a key role for IL26 in tumor progression through its amplification and stimulation of multiple cytokines, in comparison with the contribution of other individual inflammatory cytokines. As TNBC cannot be treated with conventional targeted therapy, we investigated the potential of blocking IL26 through vaccination to suppress metastasis and tumor growth. Ad-IL26 vaccination was well-tolerated and elicited significant levels of anti-IL26 antibodies, which significantly reduced IL26+ tumor growth and prolonged survival. Our findings, together with previous reports demonstrating that neutrophil NETs enhance metastatic seeding and awakening of dormant cells in TNBC (7, 43), raise the intriguing possibility that targeting IL26 could be a safe and effective means of reducing metastasis and controlling outgrowth of TNBC without compromising the essential functions of neutrophils in patients. This identification of a conserved IL26-DNA/NET axis that regulates inflammation could offer new potential approaches to combat oncogenic inflammation in cancer, as well as inflammation responsible for autoimmunity and other diseases. Few pharmacologic means exist to prevent DNA deposition by neutrophils, especially as their depletion is problematic in chemotherapy treatments where G-CSF is often administered to combat infections resulting from neutropenia. In comparison with systemic reduction of DNA via i.v. delivery of DNAse enzymes, targeting IL26 (a highly regulated nonenzymatic protein) for transient blockade/neutralization may be a more tractable approach. Indeed, if blockade of IL26 can be achieved, it may also be possible to limit inflammation in other solid cancers, leukemias, and in various autoimmune diseases, where IL26 has been implicated (34, 44).
In conclusion, our studies reveal that IL26 is highly expressed in TNBC and mediates a proinflammatory TME, at least partially through binding extracellular DNA (particularly NETs) to stimulate the expression of multiple proinflammatory cytokines (IL6, IL8, and CXCL1). This phenotype appears to be dependent upon neutrophils, supporting and providing mechanistic evidence for how DNA excreted from these cells supports tumor growth and metastasis (7, 8, 46–48). Interestingly, our data also suggest that individual inflammatory cytokines are not essential for growth and metastasis of TNBC cells, but would rather be collectively targeted through IL26 as a more manageable method of extinguishing proinflammatory responses in tumors. Thus, this unique, noncanonical avenue of cytokine-driven inflammation presents a distinct cofactor target that may be exploited therapeutically for multiple types of cancer, as well as in autoimmune disease settings.
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
Conception and design: T.N. Trotter, C.W. Shuptrine, H.K. Lyerly, Z.C. Hartman
Development of methodology: T.N. Trotter, C.W. Shuptrine, L.-C. Tsao, Z.C. Hartman
Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): T.N. Trotter, C.W. Shuptrine, L.-C. Tsao, R.D. Marek, J.-P. Wei, X.-Y. Yang, G. Lei, T. Wang, Z.C. Hartman
Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): T.N. Trotter, C.W. Shuptrine, L.-C. Tsao, R.D. Marek, C. Acharya, T. Wang, Z.C. Hartman
Writing, review, and/or revision of the manuscript: T.N. Trotter, C.W. Shuptrine, L.-C. Tsao, R.D. Marek, C. Acharya, H.K. Lyerly, Z.C. Hartman
Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): C.W. Shuptrine, T. Wang, Z.C. Hartman
Study supervision: H.K. Lyerly, Z.C. Hartman
This research was supported by grants from the NIH (T32-CA009111 to T.N. Trotter; 5K12CA100639-09 and R01 CA238217-01A1 to Z.C. Hartman), Department of Defense (DOD; BC113107 to H.K. Lyerly), and Susan G. Komen (CCR14299200 to Z.C. Hartman).
We would like to especially thank the members of the Duke Surgery Center for Applied Therapeutics for their technical assistance and stimulating discussion regarding this project. We also thank Dr. Josh Snyder and Josh Ginzel for their assistance in performing RNA scope experiments and Lauren Halligan, MSMI, who created the illustration in Fig. 1 and edited the graphical abstract.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.