In recent years data have accumulated to support a key role for immune-regulated pathways in breast cancer (BrC) development and response to (neo)adjuvant chemotherapy.1 2 These data have reignited interest in BrC immunotherapy, especially for triple negative (TN) and human epidermal growth factor receptor positive (HER+) BrC subtypes.3 4

Dendritic cells (DCs) are vital for the initiation of T cell-mediated antitumor responses as mature DCs are able to (cross) present tumor antigens to activate (cytotoxic) T cells. Unfortunately, tumor-derived factors suppress maturation of DC and instead promote the development of immature myeloid cells that in turn can expand regulatory T cells (Treg), all of which interfere with cytotoxic T cell functionality and contribute to tumor progression.5 6 Indeed, in a flow cytometry-based study, mapping the immune landscape in BrC sentinel lymph nodes (SLNs), we have recently shown that BrC-induced immune suppression coincides with selectively hampered activation of lymph node-resident (LNR) DC subsets, that is, plasmacytoid DC (pDC) and LNR conventional DC (LNR-cDC).7 The SLN represents the site where tumor-reactive T cells should first be primed, but also where immune suppression is expected to shape the premetastatic niche. As such, SLNs are an important target for therapeutic intervention, in particular for immune modulatory strategies aiming to bolster DC activation.

In two separate phase 2 trials in patients with melanoma, we have previously established the immune-potentiating effects of the CpG-B oligonucleotide CPG7909 (CpG), a synthetic Toll-like receptor-9 (TLR9) agonist.8–12 Local administration of CpG increased activation of LNR DC subsets in the SLN and enhanced local and systemic melanoma antigen (Ag)-specific CD8⁺ T cell responses, resulting in a reduction in SLN metastases and in a significant improvement of recurrence free survival. Unfortunately, CpG can down-modulate its own immune stimulatory activity through direct TLR9-mediated activation of signal transducer and activator of transcription 3 (STAT3), and indirectly by upregulating type I interferons (IFN-I) and STAT3-mediated factors (such as interleukin 6 (IL-6) and IL-10) that in turn can activate STAT3, creating a feedforward mechanism for STAT3 activation in non-TLR9 expressing cDC.13–16 Activation of STAT3, mostly through the JAK2/STAT3 pathway, acts as a master switch in (tumor-induced) immune suppression and premetastatic niche formation,14 15 17 18 explaining the collateral expansion and/or activation of Tregs on CpG treatment,12 19 which might be more detrimental in breast tumors as they generally are more immunologically quiescent. In BrC, activation of the JAK2/STAT3 pathway has been linked to the promotion of an immunosuppressive tumor microenvironment (TME), even in early stage disease.20–23 Moreover, STAT3 overexpression in BrC correlates with higher risk of lymph node metastasis and inversely correlates with response to neoadjuvant chemotherapy.24–26

Evidence for the therapeutic efficacy of combined TLR9 stimulation and STAT3 inhibition was previously provided by Kortylewski et al, showing superior immune stimulatory effects of CpG by eliminating collateral STAT3‐mediated suppressive effects in a mouse-melanoma model.13 27 28 In BrC however, similar strategies have not yet been explored.

Taken together, these observations have prompted the ex vivo functional validation of combined CpG targeting and JAK2/STAT3 inhibition as a potential treatment strategy to modulate the immune status of SLN in patients with BrC. To this end BrC SLN‐derived viable single‐cell samples were used to study the immune modulatory effects of CpG with or without JAK2/STAT3 inhibition on the tumor‐conditioned lymph node microenvironment ex vivo. Our findings provide a clear rationale for further clinical exploration of the immune potentiation of BrC SLN through local administration of a CpG+JAK2/STAT3 inhibitor (STAT3i) cocktail, possibly in conjunction with existing (neo)adjuvant systemic therapies.

DCs are central regulators of antitumor immunity and as such represent prime targets of BrC-mediated immune suppression. In a previous comparative analysis of healthy breast draining LN versus SLN of patients with BrC, we showed that BrC-induced immune suppression primarily affected LNR DC subsets (ie, pDC and LNR-cDC) rather than migratory CD1a⁺ DC subsets. Moreover, this suppression already occurred prior to lymph node involvement and preceded profoundly disabled T cell effector functions in tumor-involved BrC SLN, pinpointing these LNR DC subsets as pivotal regulators of cellular immune suppressive pathways and therefore as promising therapeutic targets to combat BrC-induced immune suppression and secure the induction of effective antitumor immunity.7 In this report we have explored an SLN-immune modulatory strategy to effectively counteract this previously described immune suppression in BrC SLN.

We cultured viable BrC SLN-derived single-cell suspensions in the absence or presence of the immune stimulatory agent CpG with or without a small-molecule inhibitor of the JAK2/STAT3 pathway, AG490 (STAT3i), and determined the effects on DC subset frequency and activation state after 2 and 7 days of culture. We believe that the 7-day cultures most accurately reflect the overall effects of the specific immune modulatory agents (ie, CpG and CpG+STAT3i), because at day 7 external effects (ie, effects of handling and of culture per se) were minimized demonstrated by the observation that DC subset activation/maturation state in the unstimulated control group was either lower or comparable to pre-culture levels as opposed to findings at day 2. Moreover, inhibition of the JAK2/STAT3 pathway results in time-dependent activation of NF-κB which is believed to be the underlying mechanism for DC activation through JAK2/STAT3 inhibition.18 33 This notion, combined with previous findings that CpG-induced pDC activation allows the indirect (and thus delayed) activation of non-TLR9 expressing cDC subsets through induction of local cytokine release from pDC, most notably IFN-I,8 34 provides additional support that the 7-day cultures also allow these delayed effects to become detectable. In keeping with this, CpG-induced pDC activation was already detectable and significant by day 2, whereas CpG-induced activation of the cDC subsets was only detectable and significant by day 7.

Based on results from the 7-day cell cultures, CpG exposure induced activation (ie, increased CD86 or CD40 expression) of all DC subsets, whereas maturation (marked by CD83 expression) was most profoundly induced in LNR DC subsets (ie, pDC and LNR-cDC). These results are in line with our previous findings that local CpG administration in patients with melanoma led to the preferential recruitment and maturation of these LNR DC subsets.8 10 In our current study, we additionally assessed the immune modulatory effect of combined STAT3 targeting. The direct antitumor effect of STAT3i alone has been established in several pre-clinical BrC studies.35–37 However, while in these studies STAT3i was assessed in the context of targeting the primary tumor itself where STAT3 activation is abundantly present, in our study design we specifically aimed to target BrC draining SLN. As the BrC SLN in our study were derived from patients in the early stage and were mostly metastases free, baseline STAT3 activation was likely less present than in these tumor models but was predominantly induced by CpG administration (as also illustrated by figure 1). Indeed, in three separate BrC SLN cultures no activating effects of STAT3i only were observed (online supplemental figure S4). Therefore, our approach of combining STAT3i with CpG administration was specifically intended to counterbalance CpG-induced STAT3 activation and associated Th2 skewing and Treg expansion.

Adding STAT3i to CpG further enhanced the maturation state (CD83 expression) of all DC subsets, but most significantly of the LNR-cDC and pDC subsets. Whereas pDCs are involved in IFN-I-driven boosting of cytotoxic T cell responses,38 the LNR-cDC subset has been linked to cross-priming of cytotoxic T cells. We previously showed that under steady state conditions this LNR-cDC subset preferentially expressed BDCA3 and CLEC9a compared with its migratory cDC counterparts10 39 and both we and other groups have linked this BDCA3⁺ DC population (designated cDC1) to cross-priming abilities, vital to the elicitation of an effective antitumor CD8⁺ T cell-mediated immune response.39–41 More recent, as yet unpublished, studies from our lab employing multi-parameter FACS panels of up to 15 markers have shown the LNR-cDC from these SLN suspensions to consist mostly (>75%) of more cDC2-like DC with co-expression of both CD1c and BDCA3, but also of a minority of CD1c⁻CD141^hi cDC1-like cells (<25%), see online supplemental figure S5; our current data suggest that both are equally activated by CpG±STAT3i. In light of this, our finding that combined CpG and STAT3i targeting can counteract (more effectively than CpG alone) BrC-induced suppression of all these LNR DC subsets would point to the likely induction of a more robust T cell response.

To further investigate this, Ag-specific effector T cell frequencies against MAM-A, a BrC-associated Ag expressed in up to 80% and not restricted to BrC subtype,42 43 were quantified in BrC SLN by means of an IFNγ ELISPOT assay. We show that CpG can enhance MAM-A-specific T cell responses, only when administered in combination with STAT3i. In line with these findings both Th1/Th2 cytokine release and intracellular T cell cytokine analyses point to a significantly skewed Th1 response on CpG+STAT3i targeting. The observed increased Th1/Th2 ratios on combined CpG and STAT3i conditioning mainly resulted from preventing CpG-enhanced levels of Th2 cytokine expression (IL-4 and IL-5) and not from an increase in Th1 proinflammatory cytokines. Whereas IL-4 induces differentiation of naive CD4 T cells into Th2 cells, IL-5 promotes the induction of Tregs.44 Findings in this study show that compared with CpG treatment alone, combined CpG+STAT3i treatment results in significantly decreased frequencies of both IL-4 producing CD4⁺ T cells and IL-5 producing CD8⁺ T cells (figure 5) and constrains Treg rates (figure 6C). Both would thus strengthen a type 1 antitumor T cell response. Similarly to the observed IFNγ increase, CpG exposure (with or without STAT3i) in BrC SLN cell samples induced significantly higher levels of the typical Th1-associated chemokines CXC9 and CXC10 which mediate CD4⁺ Th1 and CD8⁺ T cell recruitment to the TME45 46 and of the chemokines CCL2 and CCL5 which have been described to mediate DC-T cell interactions in LNs47–49 and cDC1 recruitment,50 both vital processes for efficient CD8⁺ T cell priming and the elicitation of a Th1 antitumor response. Of note, when produced by tumor-suppressed immune cells or by tumor cells themselves, CCL2 and CCL5 can also promote recruitment of immune-suppressive cells to tumor sites, such as tumor-associated macrophages and myeloid-derived suppressive cells.45 As such, the protumor or antitumor roles of CCL2 and CCL5 appear to be highly context (ie, disease and anatomic localization) dependent.45 51 As in this study chemokines were assessed in SLN draining from patients with early-stage BrC and not in the context of the tumor itself, presumably here CCL2 and CCL5 would most prominently exert their T cell priming and antitumor functions, and conceivably even more so under proinflammatory conditions induced by CpG.

Combined, these cytokine and chemokine data illustrate that while CpG in fact controls the initiation of the (Th1) proinflammatory response, the effect of adding STAT3i to CpG activation was mainly limited to counterbalancing of the CpG-induced Th2 response, ultimately promoting type 1-mediated antitumor immunity and leading to the induction of a BrC-specific T cell response. This would also fit with the observed lack of changes in IL-12 content in 2-day supernatants between the medium control and CpG±STAT3i conditioned SLN cell cultures (online supplemental figure S3), which may be explained by the fact that CpG-B directly impacts pDC but not cDC and that cDC rather than pDC are a major source of IL-12.52 Of note, in a group of four SLN samples we observed increased IFNα release on culture with CpG, most likely secreted by pDC, which also remained unaffected by the addition of STAT3i (online supplemental figure S6).

Seemingly in contrast to the observed increased IFNγ and proinflammatory chemokine levels, is the finding of decreased frequencies of IL-2 and TNFα producing CD4⁺ and CD8⁺ T cells in the CpG±STAT3i conditioned groups. However, it is known that both IL-2 and TNFα are important in the early stages of T cell activation, involving expansion, whereas more differentiated effector T cells typically produce less IL-2 and/or TNFα.53 Therefore, CD4⁺ and CD8⁺ T cells in both the CpG and CpG+STAT3i groups likely represent a more differentiated T cell population.

The finding that the combination of CpG/STAT3i did not result in lower IL-10 levels as compared with CpG was surprising, as it is known that CpG can induce IL-10 through direct TLR9-mediated STAT3 activation.54 A possible explanation as to why we did not observe lower IL-10 levels on combined targeting with CpG and STAT3i, is that the measured IL-10 levels in the supernatants were in fact cumulative. Whereas STAT3 activation by CpG occurs rapidly, inhibition of STAT3 activity by AG490 was delayed (see figure 1B and online supplemental figure S2) most likely because it occurs through an indirect mechanism (ie, by blocking JAK2-mediated pSTAT3), probably allowing early IL-10 release to occur. As a consequence, decreased real-time IL-10 release by immune cell subsets on indirect inhibition of the STAT3 pathway might be missed because of early CpG-induced and STAT3 activation-mediated IL-10 release dominating the cumulatively measured IL-10 levels. A similar explanation could also account for the lack of decreased IL-6 release in response to combined CpG and STAT3i treatment. The induction of a Th2 (and IL-10) response by CpG was further accompanied by increased Treg levels on CpG addition alone. Importantly, combining CpG administration with STAT3i did prevent this increase.

Kortylewski et al previously provided evidence for the therapeutic efficacy of CpG combined with STAT3 targeting.13 27 28 Similarly to our findings, they showed increased CpG-induced DC maturation, T cell activation and generation of tumor Ag-specific T cells by eliminating collateral STAT3‐mediated suppressive effects in a mouse-melanoma model. Importantly, they also provided evidence for a reciprocal mechanism underlying the therapeutic effects of combined CpG administration and STAT3 targeting, that is, that by inhibiting STAT3 signaling not only the brake had been lifted from CpG-mediated immune responses (by reducing the immune suppressive potential mediated by CpG), but also that the antitumor effect of STAT3i was enhanced by the induction of pro-inflammatory conditions by CpG.

To our knowledge we are the first to ascertain the therapeutic potential of combined CpG+STAT3i targeting in BrC. Furthermore, whereas BrC immunotherapy to date has mainly focused on TN and HER2+ BrC,3 4 here we show that also in HR positive BrC, immune-regulated pathways can be targeted to restore BrC-related immune suppression and induce a BrC-specific T cell response. This is highlighted by our finding that out of nine BrC SLNs, all corresponding to HR+ BrC subtypes, in four cases significantly increased MAM-A-specific T cell responses could be detected after ex vivo SLN immune modulation with CpG+STAT3i.

Due to increased presence of tumor-infiltrating lymphocytes (TIL) and their prognostic significance in TN and HER2+ BrC,55 56 these subtypes have been regarded as more immunogenic than non-T cell-inflamed HR+ BrC subtypes. Evidence is accumulating that in non-T cell-inflamed tumors, innate immune pathways required for recruitment/activation of DCs and T cell priming/homing into the tumor bed (and thus presence of TIL) might not be engaged, and that for all of these steps in the immune activation cycle to function properly, a IFN-I proinflammatory response is required.46 57 Although exact molecular mechanisms responsible for this immune exclusion are still being unraveled, activation of the STAT3 pathway has been implicated as a potential candidate.46 In our endeavor to target BrC-mediated immune suppression of LNR DC subsets in the SLN, we have shown that CpG targeting combined with STAT3i can not only efficiently reactivate these LNR DC subsets but can in fact also induce a proinflammatory response. Combined this might convert an immune-excluded BrC into an immune-inflamed tumor, independent of BrC subtype, and thus increase the efficacy of immune checkpoint blockade. Although in this study the population of BrC SLN which was tested for BrC-specific T cell responses was limited to the HR+ BrC subtype, CpG+STAT3i mediated induction of a BrC-specific T cell response will likely not be limited to HR+ BrC subtypes. Whereas JAK2/STAT3 signaling can occur in all BrC subtypes, it is most often associated with TN and HER2+ subtypes.21 58 Indeed, we previously showed that BrC-related immune suppression preferentially affected LNR DC subsets (ie, pDC and LNR-cDC) and that their suppressed state was more profound in HR− BrC subtypes.7 Moreover, we here show that the DC stimulatory effect of CpG+STAT3i was most profound in these LNR DC subsets and although sample sizes were too small for definitive subgroup analyses, when comparing HR− versus HR+ draining SLN similar increases in activation/maturation marker expression levels on CpG+STAT3i treatment (day 7) were observed (online supplemental figure S7).

In conclusion, our findings provide a clear rationale for clinical exploration of BrC SLN immune potentiation through local CpG+STAT3i administration for all BrC subtypes. In view of increasing evidence that immune-regulated pathways influence the response to (neo)adjuvant chemotherapy, we anticipate clinical benefit not only of combining this immune stimulatory cocktail with immune checkpoint blockade, but also with (neoadjuvant) chemotherapy.