The Uptake of MDA-Modified Antigen Mediated by SRA is Highly Effective and Associated with M2 Macrophages
Using fluorescently labeled MOG and the MDA-modified equivalent, we have demonstrated that MDA-adducted MOG is taken up via scavenger receptor A, and the importance of SRA in mediating uptake of antigen-bearing MDA or related maleyl modifications has already been described (Weismann and Binder
2012). Reports include both binding to SRA (Willis et al.
2004; Berger et al.
2014; Shechter et al.
1981) and SRA-dependent immune responses to modified antigen (Nicoletti et al.
1999; Willis et al.
2003). Although in previous cases the binding to SRA was mostly implicated by inhibition of proliferation assays using Fucoidan, we herein extend these findings and directly demonstrate SRA dependence using competitors, small molecules, specific blocking antibodies and SRA knockout mice. Furthermore, the use of thoroughly controlled custom fluorescently labeled and MDA-modified probes enabled us to quantify the relative differences in uptake, which was in the range of ~tenfold more effective when mediated by SRA. Furthermore, the effectiveness increased with higher SRA expression, as was demonstrated to be the case for M2 macrophages. Although the exact value may differ depending on cell type, SRA expression, antigen and dose, the effect is very clear: the uptake of MDA-modified antigen is significantly more efficient compared to that of native antigen.
A novel observation in this work is the association of SRA-dependent uptake of MDA-modified antigen with M2 macrophages, the expression of SRA in particular, and scavenger receptors in general being highly elevated in alternatively activated M2-type macrophages (Canton et al.
2013; Mia et al.
2014; Mantovani et al.
2013; Kou and Babensee
2011; Wynn et al.
2013). This feature highlights the capacity of M2 macrophages to contribute to phagocytic clearance of antigen and tissue remodeling and homeostasis. Notably, this observation has important implications for the role of MDA-modified antigen in vivo. Since M2 macrophages are not expected to initiate inflammation, but rather to promote its resolution, the presence of MDA-modified antigen is therefore primarily expected to constitute a sterile ‘eat me’ clearance signal above anything else. This agrees with reports demonstrating that MDA adducts constitute a ligand for complement factor H (Weismann et al.
2011) or natural antibodies of newborns (Wang et al.
2013; Binder
2010; Chou et al.
2009), among other oxidation-specific epitopes. Secondly, SRA has several anti-inflammatory roles (reviewed in Canton et al.
2013). In this context, an adapted concept is emerging, where MDA epitopes primarily constitute clearance signals, while non-physiological accumulation may promote sterile inflammation (Busch and Binder
2017).
However, the proposal that MDA is primarily a clearance signal is challenged by reports that would suggest otherwise. Among others, MDA adducts have been suggested to: increase the antigenicity of its carrier protein (Wallberg et al.
2007; Wuttge et al.
1999); enhance T cell proliferation (Willis et al.
2002); be recognized by complement factor C3a (Veneskoski et al.
2011); entail homeostatic problems in cells (Höhn et al.
2014; Willis et al.
2004); potentially alter signaling (Ott et al.
2014); constitute a co-factor for LPS (Duryee et al.
2004); and even be immunogenic per se in the absence of adjuvant (Thiele et al.
1998). Furthermore, other studies suggest roles for anti-MDA antibodies in atherosclerosis (Antoniak et al.
2015; Duryee et al.
2010) and cardiovascular mortality (Maiolino et al.
2013).
In light of these apparently contradicting reports, it is difficult to discern an apparent consensus. Importantly, the pro-inflammatory functions of MDA adducts appear mostly to be manifest at high doses, corresponding to those that also induce cell death (Willis et al.
2002). Hence, the pro-inflammatory effects attributed to MDA adducts may instead be attributed to traces of soluble MDA remaining in the protein preparation due to insufficient dialysis, or possibly relate to endotoxins in the starting material being cross-linked by MDA to the carrier protein. Regarding the first point, we have evidence from pilot experiments using MDA-modified MOG in 2D2 T cell proliferation assays in which insufficiently dialyzed protein-induced peak proliferation before turning toxic with progressive doses (data not included). Importantly, cell death triggered in vivo at high doses may induce bystander activation due to the release of DAMPs from dying cells which may account for the pro-inflammatory activation (Thiele et al.
1998). In this context, it is important to bear in mind that the preparation of MDA-modified antigen in vitro is not necessarily representative of the modification or doses in vivo (Gutteridge
1975; Millanta et al.
2013). Great caution is advised for the preparation of MDA modifications as otherwise the effect of soluble component or adduct cannot be reliably dissociated.
In conclusion, we postulate that MDA adducts primarily constitute a clearance signal for M2-type macrophages. Importantly, however, this does not necessarily exclude ensuing pro-inflammatory activities should the clearance capacity be overwhelmed by extensive cell death or accumulation of scavenger ligands. This implies that the homeostatic clearance system for MDA adducts, consisting of SRA, natural antibodies (Wang et al.
2013; Binder
2010; Chou et al.
2009) and complement factor H (Weismann et al.
2011), may have a limiting buffer before inflammation is induced, involving, e.g., C3a (Veneskoski et al.
2011) and tissue-infiltrating monocytes. In this context, the balance between pro- or anti-inflammatory cells and stimuli may determine the final immunological outcome. The recognition of MDA adducts by complement C3 has been demonstrated using human serum (supplementary data of Veneskoski et al.
2011), but could not be detected using purified C3 in a previous study (Weismann et al.
2011), and the binding of C3 to MDA adducts could therefore be indirect via factor H. If so this would argue in favor of an anti-inflammatory role of innate recognition of MDA adducts. Furthermore, it is recognized that SRA may have differing roles depending on the involvement of co-receptors that can influence the cellular signaling and response (Canton et al.
2013).
MDA Modification Can Affect Antigen Digestion at Several Levels
Herein we have demonstrated that MDA adducts negatively affect the digestion of MOG by lysine-dependent proteases such as LysC (lysine-specific) or Trypsin (lysine or arginine), as well as Cathepsin B. A straightforward explanation for this observation is that MDA targets epsilon amines of lysines and the fact that MDA-modified protein has a lower isoelectric point, i.e., a more negative net charge. Both the physical obstruction and lack of positive charge could thus prevent the digestion of MDA-modified antigen by lysine-dependent proteases. Interestingly, compared to LysC or Trypsin, Cathepsin B has been proposed to have a more promiscuous restriction motif, not involving a lysine (Biniossek et al.
2011). However, the method applied to identify this motif in that study involves a biotinylation of lysines, a factor that may physically interfere with digestion of motifs harboring a biotinylated lysine.
In an earlier study addressing the proteosomal stability of MDA-modified antigen, it was demonstrated that the presence of MDA-modified antigen itself interferes with the co-digestion of unmodified antigen (Kaemmerer et al.
2007). This implies that the inhibition of proteolytic cleavage also potentially applies to other proteins than the modified antigen itself. When we probed the digestion pattern of fluorescently labeled MOG in cell lysates, there was no indication that the overall pattern differed (Fig.
4), although there seemed to be changes in relative abundance. Possibly the head start in uptake and effective feeding toward the cells’ proteolytic machinery may accelerate the initial rate of digestion. However, identifying the processing enzymes involved in various APCs and resulting peptides requires separate detailed study.
The fact that MDA-modified antigen potentially has resistance to proteolytic cleavage has immunological implications: (1) the antigen itself may have an extended half-life in the extracellular space due to its resistance to proteolysis; (2) depending on the antigen the proteolytic cascade may be entirely altered, as it was previously demonstrated that a single processing site may determine the ensuing processing events (Antoniou et al.
2000); (3) proteolytic cleavage may dictate whether a potential immunodominant epitope is either destroyed by ‘destructive processing’ (Manoury et al.
2002), or otherwise a cryptic epitope can be released as a consequence of non-canonical processing (Stoeckle and Tolosa
2010; Doyle et al.
2007); 4) the bulky MDA adduct may physically interfere with the presentation of a modified epitope by obstructing MHC-anchoring pockets or representing an excessive TCR-facing residue. Lastly, the overwhelming accumulation of damaged, modified and aggregated antigen in macrophages may result in a failure to maintain homeostasis (Höhn et al.
2014) or defective lysosomal integrity at excessive doses (Willis et al.
2004). For the latter scenario, we observed no evidence for MDA-adducted protein-induced toxicity at the doses we used (50 µg/mL vs. 500 µg/mL in (Willis et al.
2004)), but there is evidence that cell death and lysosomal instability can be induced by the MDA-derived chemicals (data not included).
The discussed data and implications delineate how pivotal processing events are for the generation, presentation and recognition of epitopes from a given antigen (Stoeckle and Tolosa
2010), and how a modification such as MDA adducts may interfere with this cascade. However, in the case of MDA-modified MOG our data did not support any obvious alterations of its digestion in macrophages, even though the digestion in vitro was clearly affected for LysC, Trypsin and Cathepsin B. The enzymes that digest MOG in different APCs, however, remain to be identified and possibly differ depending on polarization or activation state.
Implications for the Adaptive Immune Responses Toward MDA-Modified Antigen
We studied the role of MDA adducts in adaptive immunity using the CNS-specific antigen MOG. Using the MOG35-55 reactive TCR transgenic 2D2 mice (C57BL/6 background), we observed a significant increase in proliferation in vitro using MOG-MDA compared to native MOG and demonstrated that this was in fact attributable to degree of uptake. This in vitro observation stands in contrast to what we observed in vivo. The EAE model did not reveal significant differences between either MOG or MOG-MDA as antigen, and re-stimulation of lymph nodes yielded similar proliferation (Suppl. Fig. 4).
We can explain this discrepancy firstly by the fact that the 2D2 system has a rich abundance of antigen-specific T cells that compete for presented antigen, which is the limiting factor for proliferation. Hence, with maximized uptake the proliferation is maximized accordingly. In re-stimulation assays of ex vivo lymph nodes, however, the abundance of antigen-specific T cells is several orders of magnitudes lower, and the few antigen-specific T cells that are present do not face competition. Thus, in this case the availability of antigen is not a limiting factor, unless one would first grow T cell clones for re-stimulation or limit the antigen in a dilution series. This also raises questions about potential competition between M2-type cells and other APCs in the 2D2 in vitro system. That the EAE experiments did not display any pathological difference implies that MDA-modified MOG is equally encephalitogenic compared to native MOG. Importantly, the C57BL/6 EAE model is heavily skewed to induce autoimmune disease even with the native MOG and involves the use of complete Freund’s adjuvant and pertussis toxin in order to evoke CNS inflammation. In light of this strong stimulation, it is difficult if not impossible to detect differences that may pertain to the immunogen itself.
We have demonstrated that the uptake of MDA-modified MOG is mediated by SRA. However, SRA expression is almost exclusive to macrophages and the monocytic lineage, whereas classical dendritic cells, specifically those differentiated under FLT3L as opposed to GM-CSF monocyte-derived cells (Guilliams et al.
2014; N’diaye
2016), do not express SRA. Thus, it is questionable whether classical DCs that prime naïve T cells even have the capacity to take up MDA-modified antigen at a higher rate and promote licensing of T cells to begin with.
A previous study using DBA1 mice immunized with MDA-modified rat MOG reported it to be more encephalitogenic (Wallberg et al.
2007). Using mouse MOG, we could not reproduce this observation, even in DBA1 mice (data not included). Furthermore, in contrast to mouse MOG, rat MOG becomes highly insoluble after MDA modification and precipitates heavily. Moreover, the DBA1 epitope, MOG
79-96 (GKV
TLRIQNVRFSDEGGY) (Abdul-Majid et al.
2000) contains an amino acid substitution between rat (82 =
A) and mouse (82 =
T, underlined). A potential explanation lies within the observation that immunization of C57BL/6 mice with human MOG induces EAE through alternative mechanisms (Oliver et al.
2003), namely depending on B cell APC function (Molnarfi et al.
2013). This is attributed to a S42P substitution within the C57BL/6 MOG
35-55 epitope between rodents and humans, and hence the dependence on initiating a secondary immune response to the murine epitope. If the corresponding scenario is the case in DBA1 mice immunized with MDA-modified rat MOG, a possible explanation could be that MDA-reactive B cells potentially contribute to initiating T cell responses toward MOG. However, whether this is actually the case or instead technical or other issues account for this discrepancy remains an open question, but the scenario is simultaneously considerably artificial. Importantly, we have demonstrated herein that immunization of C57BL/6 mice with MOG-MDA induces comparable EAE, but antibody responses toward both MOG and the MDA adducts. Accordingly, it would be highly interesting to study the role of modification-specific B cells in promoting autoimmunity toward carrier antigens.
Interaction of Scavenger Receptors and Adaptive Immunity
Another previous study using SRA knockout mice in the MOG
35-55 peptide-induced EAE model claimed that SRA is important for the induction of CD4
+ T cell responses (Levy-Barazany and Frenkel
2012). In that study, MOG
35-55-immunized SRA knockout mice displayed significantly milder EAE, less demyelination and lowered cytokine production (Levy-Barazany and Frenkel
2012). A recent study used fluorescently labeled antigen to demonstrate an SRA-dependent traffic of antigen from capturing B cells to macrophages (Harvey et al.
2008). Given that SRA knockout mice developed milder EAE (Levy-Barazany and Frenkel
2012), this may imply that the SRA-dependent transfer mechanism (Harvey et al.
2008) could be involved in sustaining the immune response in EAE. It is an open question whether this mechanism requires the antigen to be an SRA ligand or not. In our case, we controlled the fluorescent labeling to achieve a 1:1 molar ratio, avoiding the antigen being fully fluorescently labeled at all lysine residues because this may actually turn the antigen into an SRA ligand. In fact the uptake of fluorescently labeled MOG itself also weakly correlated with SRA expression, although not to the same degree as the MDA-modified version.
In regard to EAE, it appears that SRA-deficient mice already display shortcomings in the primary induction (Levy-Barazany and Frenkel
2012), although it cannot be excluded that SRA may play a role at later stages such as during demyelination by CNS-infiltrating monocytes. Furthermore, deficits in SRA-dependent structural organization of the marginal zone may influence the immunization process (Canton et al.
2013). Conversely, a recent study reports the opposite, namely that SRA-deficient APCs have a higher capacity to induce CD4
+ T cell responses to ovalbumin (Yi et al.
2012). The authors attributed this to regulatory mechanisms of SRA (Yi et al.
2012), which would emphasize the notion of homeostatic clearance mechanisms mediated by SRA
+ macrophages. In our study, there was no apparent difference in the immune response or EAE, despite MOG-MDA clearly being a SRA ligand, but it is unclear whether the strong adjuvant may mask minor effects in this model of autoimmune disease.
Importantly, the involvement of co-receptors of SRA may regulate (auto)immune responses, depending on whether PAMPs (e.g., via TLR4), or instead endogenous ‘eat me’ signals (e.g., via MERTK), are co-ligated (Canton et al.
2013). It is interesting that the respective signals counteract each other, i.e., MERTK ligation inhibits TLR signaling. The latter is especially relevant for the concept of ‘sterile inflammation’ in the absence of PAMPs. Conversely, SRA
−/− mice are hyper-sensitive to TLR-mediated systemic shock and production of TNF or IL-6, implying that co-ligation of SRA modulates the capacity of TLR signaling itself and that bacterial clearance is facilitated via SRA (reviewed in Platt et al.
2002). Accordingly, we report the association of phagocytosis of MDA-modified antigen with M2 macrophages via SRA and propose that MDA adducts primarily constitute clearance signals of stress-induced modified self-antigens via SRA. However, the response may potentially be fine-tuned depending on co-ligation of additional receptors. Taken together, however, the role of SRA in regulating immune responses is still largely elusive and more studies will be required to elucidate its contribution regarding specific cell populations and (modified) antigens.