Sterile inflammation induced by Carbopol elicits robust adaptive immune responses in the absence of pathogen-associated molecular patterns

Highlights • Carbopol induces Th1/IgG2a responses without PRR activation.• Carbopol polymer morphology is changed by APC phagocytosis leading to ROS induction.• This study highlights a potentially novel mechanism for in vivo cellular activation.


Introduction
Adjuvants are essential components of vaccines in which the vaccine antigen lacks robust intrinsic immunogenicity, such as those composed of recombinant or purified subunits of pathogens, or tumor antigens that may require tolerance to be broken. In recent years the adjuvant field has made breakthroughs in understanding underlying mechanisms of adjuvant activity, with the discovery of multiple pattern recognition pathways triggering innate and adaptive immune activation [1,2]. This has led to the discovery of numerous molecules with adjuvant activity that stimulate the immune system via defined pathways, including those triggered by toll-like receptors (TLRs) [1,2], the NLRP3 inflammasome [3,4] and IRF3 [5,6]. Some of these molecules have made their way into clinical trials and have considerable promise in vaccine development. Surprisingly, however, some of the most well-known and longstanding adjuvants, including aluminum salts 'alum', oil-inwater emulsions such as Freund's adjuvants, and MF59 appear to act by mechanisms at least partially distinct from these pathways [7][8][9]. Other potential modes of adjuvant action are hypothesized to include less specific activities such as the 'depot' effect by which the adjuvant sequesters antigen and releases it into the system over time, and local tissue damage resulting in release of intracellular inflammatory mediators such as ATP, nucleic acids, uric acid, IL-25 and IL-33 [6,10].
The immune-modulating activities of polyanions were first described over 30 years ago [11,12] and more recently, polyacrylic acid polymers termed carbomers have been evaluated as adjuvants in veterinary vaccines [13][14][15][16][17][18]. These reports suggest that carbomers are not harmful in mammals and are more effective than antigen alone. Carbopols have been combined with other adjuvant formulations such as MF59 to yield additive or potentially synergistic adaptive immune responses [19,20], and Carbopol is a component of the commercially-available adjuvant Adjuplex TM (Advanced BioAdjuvants) [21] and a licensed veterinary vaccine in pigs (Suvaxyn, Wyeth). We have previously demonstrated that Carbopol elicits strong Th1-type T and B-cell responses in mice, mediating protection from otherwise lethal influenza infection, and anti-tumor responses [22]. We observed that Carbopol did not have obvious toxicity in mice [22] or non-human primates [23], and propose that this type of polymer might have utility as a human vaccine adjuvant.
Here, we establish mechanistic insight into Carbopol's adjuvant effects, identifying strong inflammatory responses, cellular recruitment and phagocyte uptake of Carbopol, and identify phagocytosis as a key checkpoint in the immune response to Carbopol, resulting in changes to the physical properties of the adjuvant and disruption of the lysosomal pathway. We conclude that Carbopol utilizes a novel mechanism of APC activation in vivo resulting in potent adaptive immune responses to co-administered antigen.

Antibody/cytokine/chemokine detection
Serological analyses for antigen-specific antibodies were performed as previously described [22]. Supernatants were separated from either peritoneal lavage or cultured cells and cytokine concentrations determined via Bio-plex array (Bio-Rad Laboratories) or ELISA (eBioscience). Analyses were performed following manufacturer's instructions and cytokine concentrations calculated using standard curves generated within the same assay. Antigen-specific T-cell IFN␥ secretion/accumulation was assessed in splenocytes cultured in the presence of gp140 (16 g/mL) over 3-5 days. Secreted IFN␥ was detected by ELISA (eBioscience) and the frequencies of antigen specific IFN␥ producing T-cells were detected by IFN␥ cytokine secretion assay (Miltenyi Biotec).

In vitro stimulation
Murine RAW-Blue and human THP-1-CD14-Blue (Invivogen) reporter cells expressing NFB/AP-1 transcription factor-induced embryonic alkaline phosphatase (SEAP) were used to detect PRR stimulation. Cells were cultured for 24 h in the presence of PBS, Carbopol or a variety of PRR ligands, supernatants harvested and co-incubated with SEAP detection reagent Quantiblue (Invivogen). Inflammasome activation was assessed after overnight incubation in the presence or absence of 25 ng/mL Escherichia coli LPS (Sigma-Aldrich, USA), followed by extensive washing to remove trace LPS prior to in vitro adjuvant stimulation (4 h, or (ATP) 30 min). Supernatants were frozen and cells lysed in RIPA buffer (Life Technologies) containing protease inhibitors. Lysates were directly loaded in Laemmli buffer and gels semi-dry transferred to nitrocellulose membranes (Amersham Biosciences, UK). Membranes were blocked with 5% milk in PBS (Oxoid, UK) and caspase-1 detected by polyclonal rabbit anti-caspase-1 antibody (Santa Cruz, USA) and anti-rabbit IgG-HRP (Serotec, USA).
Please refer to the supplementary materials for additional methods.

Carbopol triggers immune cell recruitment and potently induces pro-inflammatory chemokines and cytokines
To investigate the potential mode of action of Carbopol, we compared early innate immune responses induced by Carbopol with those induced by Alhydrogel, which elicits a strong Th2/IgG1-associated immune bias [24]. Since cellular recruitment is increasingly recognized as having a major influence on adjuvanticity and adaptive immune outcome in response to adjuvants [4,6,8,10,[25][26][27], we assessed Carbopol-induced chemokine and cytokine secretion using the intraperitoneal (i.p.) administration model. Although this does not represent a human immunization route, it nevertheless represents a useful model for mechanistic analysis [8,21]. Strong responses were detected in Carbopol recipients ( Fig. 1), with significant induction of IL-1␤, IL-6, G-CSF, KC, MIP-2, MCP-1 and RANTES ( Fig. 1A-G). These chemokine trends were shared by Alhydrogel-adjuvanted mice, which also displayed significant but less potent induction of IL-6, G-CSF, KC, and MCP-1. One notable exception was the eosinophil recruitment factor Eotaxin, secreted at high levels by Alhydrogel-treated mice, but absent in Carbopol recipients (Fig. 1H). By contrast, the pattern of cytokine secretion associated with T-cell polarization was strikingly different between Carbopol and Alhydrogel groups ( Fig. 1I-K). Alhydrogel recipients expressed significantly elevated levels of the Th2-associated cytokines IL-4 and IL-5, whereas Carbopol conversely elaborated significant quantities of the Th1-associated cytokine IL-12p40. Time course analyses revealed differing kinetics within the response to Alhydrogel and Carbopol, with chemokines/cytokine concentrations peaking at 4 h and 12 h respectively. We next analyzed cellular recruitment post i.p. administration ( Fig. 2A-H). Both Carbopol and Alhydrogel elicited a large influx of cells within 24 h, with significant increases in neutrophil, monocyte and DC numbers, but striking differences were observed between the two adjuvants. Of particular significance were the high numbers of monocytes but low numbers of eosinophils recruited into the peritoneum of Carbopol-injected mice ( Fig. 2C and H), compared to the opposing phenotype in the Alhydrogel group. Alhydrogel recipients displayed a high degree of peritoneal eosinophilia and significantly lower numbers of monocytes compared to Carbopol recipients (p < 0.001). A decline in resident B cell and macrophage numbers in the peritoneum is often associated with cellular activation (28) and this can be seen after 4 h in Alhydrogel treated mice when compared to PBS controls ( Fig. 2D-E). However, this effect was delayed in Carbopol-treated mice, in which mature macrophage numbers were transiently increased 4 h after Carbopol injection, prior to a significant reduction at 24 h when compared to PBS treated control mice. These data suggest that whilst both Carbopol and Alhydrogel drive significant leukocyte recruitment and inflammation early post-immunization, their mechanisms of action are likely to be distinct.

Carbopol promotes antigen uptake by APC
To investigate antigen uptake in the presence of Carbopol, mice were injected with fluorescently labeled ovalbumin (OVA) with or without adjuvant ( Fig. 2I-J). 24 h later, antigen was detected in association with multiple cell types, but principally with monocytes, macrophages and DC (Fig. 2I). Absolute numbers of antigen-positive cells increased approximately five-fold with Carbopol or Alhydrogel (Fig. 2J). Despite similarities in the absolute number of antigen positive cells in the presence of either adjuvant, uptake was biased toward distinct leukocyte populations: Carbopol drove antigen uptake primarily in monocytes, whereas Alhydrogel predominantly drove neutrophil uptake. Whilst differences in monocyte-associated antigen uptake may reflect the increased presence of monocytes in the Carbopol-stimulated mice, the equal frequencies of neutrophils present in both groups suggest additional bias toward antigen uptake by neutrophils in Alhydrogel-stimulated mice. These data suggest there may be differing mechanisms of antigen uptake induced by these adjuvants.

Carbopol drives Th1 responses in the absence of TLR activation
Given that Th1-biased adaptive immunity is commonly linked with adjuvants that drive pattern recognition receptor (PRR), in particular TLR [28,29] activation, we investigated whether Carbopol might drive inflammation via this route using monocyte/macrophage NFB/AP-1 reporter cell lines expressing multiple PRRs. Whilst these cells responded to a variety of PRR ligands, there was no response to overnight culture with increasing doses of Carbopol (Fig. 3A-C).
To confirm this lack of PRR activation in vivo, mice deficient in expression of the TLR downstream signaling molecules MyD88 and TRIF were immunized subcutaneously (s.c.) with HIV-1 gp140 alone or with Carbopol. Antigen-specific antibody responses in TRIF −/− mice were unchanged, demonstrating that Carbopol adjuvant activity does not require TRIF-dependent TLR activation (Fig. 3D-F). By contrast, a significant reduction in antigen-specific IgG2c titers was observed in MyD88 −/− TRIF −/− double-deficient mice immunized in the presence of Carbopol, resulting in a significant shift toward an IgG1 isotype bias. Since we had excluded a requirement for TRIF signaling, we assessed antibody responses to Carbopol in MyD88 −/− mice in comparison with the TLR9 ligand CpG and observed a similarly weak IgG2a response and IgG1 isotype bias (Fig. 3G-I). It has been reported that MyD88 is also involved in non-PRR signaling pathways including stabilization of IFN␥ receptor signaling [39] and B-cell immunoglobulin class switching [30]. We therefore directly assessed Carbopol-driven T-cell polarization in the absence of MyD88. To this end, IFN␥ secretion (Fig. 3J) and the frequency of antigen-specific IFN␥-secreting CD4 + T-cells Histogram bars represent the mean ± SEM and significance was tested by one-way ANOVA followed by Bonferroni correction (*p < 0.05, **p < 0.01, ***p < 0.001).
( Fig. 3K) were quantified in gp140-stimulated splenocytes isolated from immunized mice. Strikingly, unlike the TLR9 agonist CpG, Th1 polarization in response to Carbopol-adjuvanted immunization was intact in the absence of MyD88. Therefore, reduced IgG2abiased humoral responses to Carbopol in the absence of MyD88 are not due to defective development of antigen-specific Th1 cells, but probably result from MyD88-dependent TLR-independent signaling pathways.

Internalization into the phagolysosome promotes conformational change in Carbopol and reactive oxygen species production
Given the minimally-biodegradable nature of Carbomers, we examined the effects of Carbopol phagocytosis and the  ensuing inflammatory response. Using high-throughput imaging we assessed subcellular localization of Carbopol after phagocytic uptake by BMDC (Fig. 6A and B). Carbopol particles did not colocalize with the early-endosomal marker EEA1 [35], but were predominantly located within CD107a + lysosomal compartments [36]. Using electron microscopy we observed phagocytic cup formation and subsequent localization of Carbopol within compartments with lysosomal morphology (Fig. 6C and D). Interestingly, we observed substantial morphological modifications indicative of conformational change in Carbopol within the lysosomal compartment. Whereas extracellular Carbopol formed a very tightly packed 'closed' structure, lysosomal material adopted an 'open' mesh-like conformation ( Fig. 6Di-ii, iii-vi). It has been previously established that lysosomal accumulation of non-biodegradable particles and/or lysosomal disruption can drive cellular activation and inflammation, and a key signature of this response is ROS production [32,37]. We therefore compared Carbopol-induced ROS production in BMDC and BMDM in vitro with well-characterized particulate and non-particulate controls ( Fig. 6E and F). We detected significant BMDM-derived ROS production in response to escalating concentrations of Carbopol, which increased in a dose-and time-dependent manner (Fig. 6E). Significant ROS production was also induced by Carbopol exposure to BMDC, and this response was completely abrogated by the presence of the actin polymerization inhibitor Cytochalasin D, which prevented Carbopol uptake (Fig. 6F). Taken together, these data suggest that phagocytosis results in Carbopol accumulation and conformational modification within lysosomal compartments, triggering ROS production and inflammation.

Discussion
The successful development of vaccines against pathogens refractory to current vaccination strategies such as HIV-1, HCV and Malaria will be partly dependent upon new adjuvant technologies [38]. Although Carbopol has been used in veterinary vaccines for many years, the cellular mechanisms that drive its efficacy in adjuvant applications are yet to be characterized and therefore investigation of the mode of action of Carbopol is warranted, particularly if this adjuvant is to be taken into man.
We demonstrate that Carbopol adjuvant activity is accompanied by potent innate immune activation, comprising chemokine and pro-inflammatory cytokine secretion associated with rapid leukocyte recruitment and antigen uptake. Striking differences were observed in early immune responses to Carbopol and Alhydrogel, consistent with the adaptive immune bias reported for these adjuvants [3,22]. Similar contrasting responses have been reported between TLR and non-TLR adjuvants [39], suggesting that the innate response to Carbopol more closely resembles that of a TLR-targeted formulation. However, both in vitro and in vivo analyses demonstrated that Carbopol does not directly trigger TLR activation and the downstream signaling pathways associated with pathogen recognition. In addition, Th1 induction was intact in the absence of TLR-Myd88/TRIF signaling, demonstrating a noncanonical pathway of Carbopol-induced Th1 polarization. Whilst the TLR-Myd88/TRIF signaling axis is the best characterized mechanism of Th1 induction, other pathways of APC activation and resultant Th1-polarization have also been reported, particularly in response to infection [7,[40][41][42]. Although the alternate receptors and signaling pathways involved remain to be fully explored, some non-TLR receptors such as C-type lectins (e.g. Mincle, MCL, Mannose receptor) and scavenger receptors can drive Th1 responses via both Myd88-dependent and -independent signaling [43][44][45][46]. Given the polyanionic structure of carbomers it is possible that Carbopol is recognized via one such pathway, which will require further investigation. Carbopol and Alhydrogel yield strikingly different Th biases, with Carbopol and Alhydrogel eliciting robust Th1 and Th2 responses respectively. We demonstrate that these adaptive immune responses are mirrored by the early innate chemokine and cytokine responses that predict the adaptive immune outcome. Thus Carbopol selectively triggers MIP-2 MCP-1, RANTES and IL-12p40 release, whereas Alhydrogel drives rapid (within 4 h) Eotaxin, IL-4 and IL-5 release.
As with other immune adjuvants, Carbopol efficiently induced inflammasome formation. However since Carbopol-induced humoral responses were independent of this process, we conclude that inflammasome activation is secondary to other Carbopoltriggered mechanisms of immune activation. Since it has been reported that phagolysosomal destabilization after adjuvant phagocytosis is an important step in inflammasome activation [32], we investigated the interactions between Carbopol and phagocytes in more detail. Carbopol was avidly captured by a range of phagocytes including APC, most likely due to its particulate nature and strong negative charge which may be selectively recognized by scavenger and other receptors on phagocytes [47]. Uptake was accompanied by a conformational change in Carbopol structure similar to that described for a range of homopolymers at low pH, including carbomers, which make these compounds suitable for controlled-release drug delivery [48]. Due to the polyanionic nature of Carbopol, a reduction in pH causes the carboxylate moieties on the polymer backbone to lose charge, eliminating the repulsive electrostatic forces and collapsing the gel matrix. It is likely that the progressive acidification of the lysosome occurring during lysosomal maturation [49] drives the Carbopol conformational change. In addition, interference with lysosomal acidification has been previously reported in response to phagocytic uptake of polyanionic compounds [50]. In line with this, we found a clear correlation between Carbopol phagocytosis and ROS production by APC, which is a hallmark of lysosomal arrest and disruption. We therefore conclude that phagocytosis of the adjuvant plays an early and critical role in the ensuing inflammatory response to Carbopol administration.
The Th1 polarized response induced by Carbopol in the absence of TLR stimulation is uncommon and warrants further investigation to understand the molecular mechanisms at play. Further work on this type of adjuvant will elucidate mechanisms and lead the way to the design of adjuvants with increased potency and reduced toxicity.