Synthetic oligodeoxynucleotide CpG motifs activate human complement through their backbone structure and induce complement-dependent cytokine release

ABSTRACT


INTRODUCTION
Mitochondrial DNA (mtDNA) is released after ischemia/reperfusion injury, surgery, severe trauma, organ failure and sepsis (1)(2)(3)(4)(5).Mitochondria evolved from α-proteobacteria via symbiosis (6,7).Due to its structural similarity to bacterial DNA, mtDNA triggers inflammation (8).The pro-inflammatory effect of bacterial and mtDNA is attributed to their particular CpG sequence, wherein cytosine is unmethylated (9).Synthetic oligodeoxynucleotides (ODN) containing CpG motifs have been used to mimic immunostimulatory effects of bacterial and mtDNA (10).There are three classes of CpG motifs: CpG-A, CpG-B, and CpG-C.They are categorized basing on their lengths, position and spacing between CpG motifs as well as their backbone modifications (11).
CpG ODNs have been shown to activate the innate immune system via pattern recognition receptor (PRR) TLR9, which is primarily expressed in the endoplasmic reticulum of B cells and plasmacytoid dendritic cells (12)(13)(14).CpG-TLR9 interaction activates MyD-88 and NFκB signaling pathways, leading to maturation, differentiation, and proliferation of both innate and adaptive immune cells and release of pro-inflammatory cytokines (11,12,15).
Many studies have focused on characterization and modulation of the CpG-TLR9 response while little is known about CpG ODNs effect on other PRRs of innate immune responses.
The complement system represents a crucial arm within innate immunity and consists of more than 50 soluble and membrane-bound proteins, with pattern recognition, immune clearance and inflammatory actions (16,17).The complement system is activated through three distinct pathways: the classical (CP), the lectin (LP), and the alternative (AP) pathway.
All three pathways converge at the formation of the C3 convertase (i.e., C4bC2a from CP and LP and C3bBb from AP), which cleaves C3 into C3a and C3b.C3b fragments can attach to pathogen surfaces or bind to C3 convertases to form the C5 convertases, which cleave C5, leading to the release of C5a and the formation of the terminal complement complex C5b-9.
Mangsbo and colleagues have described that in human whole blood CpG-B ODN induced the generation of both C3a and C5a at significant levels and reduced the responsiveness of TLR9 signaling when C3 function was blocked (18).Yet, the exact mechanism of complement activation by CpG remains unclear.Therefore, the aim of this study was to investigate the effect of different classes of CpG ODNs on activation of the complement system in human whole blood and plasma and to identify the mechanism of complement activation by CpGs.

Whole blood and plasma experiments
Human whole blood was collected from healthy adult volunteers into NUNC cryotubes (Nalgene NUNC, Roskilde, Denmark) containing 50 μg/ml lepirudin (Refludan, Pharmion, Copenhagen, Denmark).Plasma was prepared by centrifugation at 3000g for 15 min at 4°C.Aliquots of whole blood and plasma were incubated with increasing doses of the three classes of CpG ODNs (between 10 and 100 μg/ml) for up to 4 h at 37°C.GpC ODNs and CpG ODNs were used at equimolar concentration (10 μM).To assess the complement pathway involvement, mAbs blocking function of C2, MBL, and factor D were added to plasma and incubated for 5 min at room temperature prior to activation by CpG-B ODN (50 μg/ml) for up to 10 min.Cytokine release was determined in whole blood pre-incubated with TLR9 inhibitor ODN 4191, C5 inhibitor eculizumab, C5aR1 antagonist PMX-53 or anti-CD14 prior to CpG-B ODN incubation.In all experiments, complement activation was stopped by adding 10 mM EDTA, and aliquots were stored at -70°C until further analyses.

Enzyme immunoassays
C3bc, C3bBbP and fluid phase C5b-9 (sC5b-9) concentrations in plasma were measured by in-house ELISAs as previously described in detail (28)(29)(30)(31).Briefly, the capture mAbs clone bH6, anti-factor P (clone #2, Quidel, San Diego, CA) and clone aE11 were used, they react with neoepitopes C3bc (bH6) and C9 (aE11) respectively, which are exposed after complement activation and not present in the native components.After background OD values were subtracted, standard curves were fitted by a sigmoidal four parameter (logistic) equation using GraphPad Prism version 9 (GraphPad Software) (31).Inter-and intra-assay coefficient of variability below 20% were accepted.When original ODs were inaccurate (>20% differences or above maximum range) and could be responsible for a change in trend, remeasurements were conducted.Cytokines were measured in EDTA plasma samples with a 27-plex kit (Bio-Rad Laboratories, Hercules, CA) according to the manufacturer´s instructions.Only the cytokines in which the concentrations were significantly statistically increased (p < 0.05) compared to untreated samples were selected for further analyses.
Fluorescence was expressed as median fluorescence intensity.To correct for non-specific binding, negative controls were implemented at each step in which PBS was used as substitute for biotinylated ODN-2006, plasma, capture antibody or detecting antibody.Each experiment was performed in triplicate and repeated three times at different days.

Ethical statement
This study was designed and performed according to the ethical guidelines from the declaration of Helsinki.Informed written consent was obtained from the blood donors.This study was approved by the Regional Ethic Committee (REK S-04114).

Statistical analysis
Complement levels in whole blood or plasma over time were compared with their control group using generalized linear mixed model analyses.Time, group and time-by-group interaction was handled as fixed intercept and donor as random intercept to calculate complement trends within each donor over time.All treatment groups were compared with non-inhibited CpG ODN groups using paired t-test.As this is an exploratory in vitro study with small sample sizes, no post hoc testing was performed.No statistical analysis was performed with the gathered data from the C2 and MASP-2 deficient experiments since the plasma and serum was collected from only one individual.Pearson coefficients were calculated for assessing the correlation between CpG ODN/bead ratio and fluorescence emitted upon complement protein binding.All statistical analyses were conducted using GraphPad Prism 9 (GraphPad Software, San Diego, CA) and IBM SPSS Statistics for Macintosh (IBM Cooperation, Armonk, NY).P values less than 0.05 were considered statistically significant.

Complement activation by different classes of CpG ODNs in whole blood and plasma.
CpG-B and CpG-C ODN led to a significant dose-dependent increase in C3bc, C3bBbP and sC5b-9 levels in whole blood (p < 0.01 for all, Fig. 2) and plasma (Fig. S1), whereas CpG-A ODN had no effect on complement activation (p > 0.05) in neither whole blood nor plasma.

Backbone mediated effect accountable for complement activation by CpG ODNs.
Similar to CpG-A, equimolar concentrations of sequence-modified GpC-A did not result in complement activation (Fig. 3).GpC-B ODN resulted in a similar extent of complement activation as measured by C3bc, C3bBbP and sC5b-9 and were not different from CpG-B ODN.A similar trend in C3bc, C3bBbP and sCb-9 levels was observed for sequencemodified GpC-C ODN (Fig. S2).In further experiments, only CpG-B ODNs were used as the backbone is identical to CpG-C ODNs.

Classical and alternative complement pathways are activated by CpG-B ODN.
Contribution from the C2-dependent CP and LP in complement activation was assessed in C2 deficient plasma.CpG-B ODN led to increased levels of C3bc and sC5b-9 over time (Fig. 4A).Reconstitution of C2 led to a 2-to 3-fold increase in complement activation, indicative for contribution of CP and/or LP.The contribution of the lectin pathway was examined in MASP-2 deficient serum.A time-dependent increase of C3bc and sC5b-9 levels was observed, yet reconsitution with MASP-2 did not increase complement activation (Fig. 4B).These data suggest that the lectin pathway is not required for CpG ODN induced complement activation.The degree of contribution of each complement pathway was further investigated by pre-incubation of plasma from healthy individuals with specific inhibitory antibodies against C2 (inhibition of both CP and LP), MBL (LP inhibition), or factor D (AP inhibition).

Direct binding of C1q and factor H to CpG-B ODN.
Binding of CpG-B ODN to various complement components in EDTA-plasma was investigated by flow cytometry.C1q and factor H showed a significant positive correlation between the concentration of CpG-B and either C1q or factor H (p = 0.006, p < 0.001, respectively) (Fig. 6).No binding of C2, properdin, MBL or MASP-2 to CpG-B ODN could be detected (Fig. S3).

DISCUSSION
In this study, we showed that CpG ODN class-B and -C induced dose-and time-dependent complement activation both in human whole blood and plasma of healthy donors.These complement effects were dependent on the backbone of CpG ODNs, but independent of the CpG sequence.CpG ODNs were capable of binding to C1q and factor H, indicative of initial CP and direct AP driven complement activation.CpG ODN induced proinflammatory cytokines and chemokines were significantly suppressed by complement inhibition at the level of C5-C5aR1, demonstrating that the biological effect of CpG was mediated through this complement axis.
CpG-B and CpG-C ODN led to significant dose-and time-dependent increase in the activation products C3bc, C3bBbP and sC5b-9 in whole blood and plasma, whereas CpG-A ODN had no effect on complement activation.To our knowledge, this is the first study that demonstrates a similar dose-dependent increase of complement in whole blood and plasma upon CpG stimulation, and in particular the plasma results indicate that CpG ODN induced complement activation did not depend on CpG deposition on activating cell surfaces.While other studies have focused on the CpG sequence and its association to complement and synergistic TLR9 activation, the present study highlights that the CpG ODN structure itself, i.e. the backbone, is responsible for the complement activation which might include multiple levels of the complement system cascade (18,32,33).Unlike CpG-A, both CpG-B and CpG-C share a complete phosphorothioate (P-S) backbone, explaining the differential effects on complement activation.These findings are also consistent with previous studies assessing P-S ODNs in primates and showing ODN backbone-dependent complement activation (34)(35)(36).A different magnitude of complement activation was observed for CpG-B and CpG-C, and this might be explained by the palindrome sequence comprised in CpG-C ODN.In the presence of a palindrome sequence, secondary hairpin structures with a stem and loop could form which may contribute to additional nuclease stabilization (37).This may have positively affected complement protein binding affinity.In contrast to our results, one prior study described that CpG-A ODN incubation led to increased C3a and C5a levels, although notably this was only conducted in the plasma from one individual (18).The lack of immunostimulatory potential in our model may be related to the phosphodiester (P-O)/P-S chimeric backbone of CpG-A ODNs.Compared to P-O ODNs, P-S ODNs are more likely to bind to DNA-binding proteins, i.e. complement products (38).Additionally, P-S ODNprotein complexes dissociate at a much lower rate (38).We argue that these functional differences between the backbone structures may explain that the apparent threshold needed for CpG-A ODN induced complement activation was not reached in our model.
C3bc and sC5b-9 activation substantially increased upon C2-reconstitution in C2-deficient plasma but not in MASP-2 deficient serum upon MASP-2 reconstitution.A CP-mediated complement activation by CpG ODNs is thus plausible.CpG ODNs were only found to form direct complexes with C1q, a recognition molecule of the CP, suggesting that this initial activation was through CP.In a study that supports our findings, Jiang et al., observed that both single and double stranded DNA fragments were capable of binding to C1q, resulting in the activation of the complement system via CP (39).DNA motifs that support C1q interaction have not been characterized yet, but our results imply that the activation is dependent on physical properties of DNA rather than its sequence (40).Contrary to our data, Tang et al. documented that CpG-B ODNs were able to bind to MBL, although excess mannan interfered with this binding (41).Unfortunately, the use of purified components precludes interactions within physiological conditions, by contrast, our model comprises close to physiologic conditions (28).
Further, CpG ODNs were found to bind directly to factor H, a negative regulator of the AP.This pathway is constitutively activated in the fluid phase by low-rate spontaneous hydrolysis of C3 to C3(H2O).To regulate this mechanism, C3(H2O) is rapidly inactivated to iC3(H2O) by factor H in conjunction with factor I (17).We speculate that when factor H binds to CpG ODN, its interaction with C3b could be impaired, leaving the C3(H2O)Bb to split more C3 leading to spontaneous self-amplification and extensive sC5b-9 formation.This presumption would explain the observed complement activation in C2 deficient plasma and absence of complement products when AP function was counteracted by the inhibition of factor D.
However, further research is needed to confirm this hypothesis.In support of this, studies investigating toxicity of P-S ODNs in primates observed complement activation together with decreased plasma levels of factor H.Moreover, the addition of purified factor H prevented ODN induced complement activation (35).Overall, our results show that CpG ODNs induce initial activation of the CP as well as direct activation of the AP.
We found complement-dependent cytokine release upon CpG ODN incubation, which was independent of TLR9 and CD14.The lepirudin based whole blood model used here is physiologically relevant to assess cross-talk between complement and TLR signaling pathways since lepirudin does not interfere with the complement cascade (28,42,43).
Inhibition of C5 reduced release of all cytokines induced by CpG ODN, indicating that these cytokine responses were complement-dependent.This complement-dependent effect appeared to be mediated entirely through C5aR1 since specific C5aR1 antagonism led to reduction of cytokine levels comparable to C5 inhibition.Similar interactions between complement and TLR signaling have been reported; decay-accelerating factor deficient mice showed higher levels of IL-12 than wild-type mice in response to CpG ODNs (44).Of interest, the lack of C5aR1 on dendritic cells led to significant lower levels of IL-12 in the presence of CpG ODNs (45).In contrast, others have found evidence for C3/C3a dependent CpG ODN signaling as the cellular uptake of CpG ODNs in monocytes was markedly reduced upon C3 inhibition as well as the CpG ODN induced cytokine responses, while C5aR1 blockage had only a minor effect (18,44).As discrepancies to aforementioned studies exists, future studies should account for TLR-9 expression levels on monocytes, anticoagulants used and the role of the involved complement products upon CpG ODN induced TLR-9 signaling.
Given that synthetic CpG-B and -C ODN share common structural features with bacterial DNA such as backbone P-S modification, we propose that the mechanism by which bacterial DNA or mtDNA activate the complement system could be similar (46).Future studies should assess the effect of isolated mtDNA on the complement system.A recent study concluded that the complement system mediated undesired inflammatory responses elicited by cell-free mtDNA in COVID-19 patients (47).These future findings can be of great relevance, as they might form the basis for a novel therapeutic strategy, i.e., complement inhibition in a broad spectrum of diseases in which mtDNA release contributes to pathology (1)(2)(3)(4)8).In addition, our findings contribute to the current understanding on the mechanism of immunogenicity induced by CpG ODNs, when used as adjuvants in vaccines against life-threatening diseases such as: cancer, malaria, hepatitis and COVID-19 (48,49).
A limitation of this study is that we did not measure complement activation products specific for the CP and LP.Unfortunately, there is scarcity of reliable assays that reflect CP and in particular LP activation.However, by using pathway-specific antibodies e.g., C1q, MBL, MASP-2 and deficient-C2 plasma and -MASP-2 serum we could distinguish between initial activation of CP and LP.Purified C2 itself has been reported to induce low-grade spontaneous complement activation (50).Therefore, the levels of complement activation in reconstituted C2 deficient plasma might be partially attributed to the purified protein.
In conclusion, we found that CpG ODN class-B and -C activated the complement system in a dose-and time-dependent manner.The effect appears to be CpG ODN class-dependent and triggered by physical properties of the ODN backbone.Complement activation was classical and alternative pathway mediated through interactions with C1q and factor H, respectively.
CpG ODN induced cytokines release appeared to be secondary to complement activation, mediated through C5aR1 and independent of TLR9 stimulation.Further elucidation on underlying immunological mechanisms behind sterile inflammation by CpG ODNs may offer unexplored opportunities to modulate the innate immunity in pathogen induced and sterile inflammatory processes.Plasma and serum from one C2-deficient and MASP-2-deficient individual was used to assess the route of complement activation by CpG-ODN.C2-deficient plasma was stimulated with CpG-B ODN (50 µg/ml) alone or together with purified C2 (A).A time-dependent increase of C3bc and sC5b-9 levels was observed, although these levels were more markedly increased upon the reconstitution with C2.MASP-2-deficient serum was stimulated with CpG-B ODN (100 µg/ml) alone or together with purified MASP-2 (B).CpG-B ODN led to increased levels of C3bc and sC5b-9 over time, no differences were found in the level of complement formation upon the reconstitution with MASP-2.As these experiments include only one donor, statistical analyses were not conducted.

Figure 1 .
Figure 1.Backbone and sequence characteristics of CpG-A, -B and -C ODNs.

Figure 2 .
Figure 2. CpG-B and CpG-C ODNs induced complement activation in whole blood.

Figure 3 .
Figure 3. Immunostimulatory effect of CpG ODNs on the complement system is

Figure 4 .
Figure 4. CpG-B ODN induced complement activation in C2-deficient plasma and

Figure 5 .
Figure 5. Interplay between classical and alternative complement pathways upon CpG

Figure 7 .
Figure 7. CpG-B ODN leads to complement dependent cytokine release.

Table I .
Sequence list of oligonucleotides1Capital letters indicate phosphodiester and those in lower case phosphorothioate bonds.Underlining stands for palindromic sequence.