Complement factors C3a and C5a mimick a proinflammatory microenvironment and increase HBV IGRA sensitivity

As a first step, cell viability in all collected samples was assessed by stimulating whole blood with SEB and CEFT, respectively. All samples showed strong IFNγ (51,922 ± 5527 pg/ml) as well as IL-2 (42,295 ± 2280 pg/ml) responses to SEB and prominent, but in comparison much weaker cytokine secretion upon CEFT stimulation (1294 ± 302.8 pg/ml IFNγ and 167.5 ± 32.01 pg/ml IL-2) (Fig. 1).

Stimulation of whole blood with synthetic HBsAg peptides induced production of both, IFNγ (146 ± 43.1 pg/ml) and IL-2 (690.3 ± 195.4 pg/ml) with the cytokine responses for IL-2 (Fig. 2b) being generally stronger than the ones for IFNγ (Fig. 2a).

Additional stimulation with C3a did not have any significant effect on the release of the mentioned cytokines (139 ± 39.6 pg/ml IFNγ and 719.6 ± 198.9 pg/ml IL-2) whereas combination of HBsAg with C5a significantly increased the HBsAg-specific production of IFNγ (336.7 ± 67.9 pg/ml, p < 0.0001) and IL-2 (789.4 ± 216.5 pg/ml, p < 0.0001) (Fig. 2).

To test if C3a and C5a synergize to further increase HBsAg-induced IFNγ and IL-2 production, whole blood was stimulated with HBsAg, C3a and C5a. Even though samples stimulated with HBsAg, C3a and C5a showed a significantly stronger cytokine production than samples stimulated with HBsAg alone, combination of both complement fragments did not lead to a significant increase of IFNγ (248.2 ± 53.8 pg/ml) (Fig. 2a) and IL-2 (803.7 ± 217.9 pg/ml) (Fig. 2b) secretion when compared to samples stimulated with HBsAg and C5a alone. Anyhow, samples stimulated with HBsAg, C3a and C5a showed stronger IFNγ (248.2 ± 53.8 pg/ml) (Fig. 2a) as well as IL-2 responses (803.7 ± 219.9 pg/ml) (Fig. 2b) than samples stimulated with HBsAg and C3a alone (IFNγ 139 ± 39.6 pg/ml, IL-2 719.6 ± 198.9 pg/ml).

Interestingly, addition of C3a seems to slightly decrease the amount of secreted IFNγ since the cytokine concentration was lower after stimulation with HBsAg, C3a and C5a than after stimulation with HBsAg and C5a alone (336.7 ± 67.9 pg/ml after stimulation with HBsAg + C5a and 248.2 ± 53.8 pg/ml after stimulation with HBsAg + C3a + C5a), but showed an opposite effect with regard to IL-2 (789.4 ± 215.5 pg/ml after stimulation with HBsAg + C5a and 803.7 ± 219.9 pg/ml after stimulation with HBsAg + C3a + C5a).

Since HBsAg is a main component of the prophylactic hepatitis B vaccine it does not surprise that the HBsAg-mediated production of the cytokines IL-2 and IFNγ correlates with the blood donors’ vaccination status as published previously [3, 6]. Within our study cohort, 59/87 individuals showed positive anti-HBs antibody titers whereas 28/87 individuals were anti-HBs negative (Table 1). In accordance with our preexisting data successful vaccination was in most cases linked to cytokine levels crossing a certain threshold as calculated by ROC analysis. With regard to IFNγ secretion, 29/59 values measured for anti-HBs positive blood donors lay above the calculated threshold of 28.5 pg/ml in response to stimulation with HBsAg (Fig. 3b). ROC analysis further revealed that at a diagnostic specificity of 90% the according assay reached a diagnostic sensitivity of only 49.2% (Fig. 3a).

The correlation between the blood donors’ vaccination status and the HBsAg-driven cytokine response was more pronounced with regard to IL2 secretion. In this case, 50/59 vaccinated blood donors crossed the calculated threshold of 3.8 pg/ml when their blood was stimulated and IL2 levels were measured (Fig. 3d). In comparison to the IFNγ release assay the IL2 release assay already reached a diagnostic sensitivity of 84.75% at a diagnostic specificity of 90%, even without combination of HBsAg with additional stimuli (Fig. 3c).

Anyhow, combination of HBsAg with the complement factors C3a and C5a positively affected the diagnostic sensitivity of the IFNγ release assay. Upon stimulation with HBsAg and C3a the diagnostic sensitivity of the assay reached 56.9% (Fig. 4a), 71.2% when HBsAg was combined with C5a (Fig. 4c) and even 78.9% when blood was stimulated with HBsAg and both complement fragments (Fig. 4e).

In line with that, the corresponding AUC value increased from 0.79 to 0.89 (Figs. 3a, c, 4a, c, e).

With regard to the IL2 release assay neither C3a nor C5a did have any positive effect on the diagnostic sensitivity of this assay (Fig. 5a, c, e).

In the present study we could further show a clear correlation between the amount of secreted cytokines and the measured anti-HBs antibody. Thus, there was a link between the anti-HBs titer and the strength of the IFNγ response, no matter if the blood was stimulated with HBsAg alone (Fig. 6a, R² = 0.20 and p < 0.0001) or with HBsAg and complement fragments (Fig. 6, (b) R² = 0.16 and p = 0.0004 for HBsAg + C3a stimulation; (c) R² = 0.15 and p = 0.0006 for HBsAg + C5a stimulation; (d) R² = 0.12 and p = 0.0096 for HBsAg + C3a and C5a stimulation).

The same was true with respect to IL2 (Fig. 7, (a) R² = 0.32 and p < 0.0001 for HBsAg stimulation; (b) R² = 0.37 and p < 0.0001 for HBsAg + C3a stimulation; (c) R² = 0.32 and p < 0.0001 for HBsAg + C5a stimulation; (d) R² = 0.29 and p < 0.0001 for HBsAg + C3a and C5a stimulation).

In a first attempt to unravel the mechanisms by which the anaphylatoxins enhance the IFNγ release assay sensitivity, we analyzed the activation state of the T cells and the APCs upon stimulation of whole blood with HBsAg and HBsAg combined with C3a and C5a, respectively. Previous studies demonstrated that the anaphylatoxins C3a and C5a are able to modulate T cell responses by altering the expression levels of for instance costimulatory molecules [11, 12].

Thus, we used a flow cytometry-based approach to analyze the expression levels of costimulatory molecules as well as activation markers on T cells as well as on APCs (Additional file 1: Figure S1). More specifically, with regard to T cells, we measured expression levels of CD28 and CD25. We could not detect any significant changes in the expression levels of CD25 (Additional file 2: Figure S2a) and CD28 (Additional file 2: Figure S2b) on T cells when stimulating whole blood with HBsAg in combination with the anaphylatoxins C3a and C5a. Even when looking at the cytotoxic and the T helper (Th) cell subsets there were no differences (Additional file 3: Figure S3).

With regard to the MHCII⁺ APCs, we analyzed the expression levels of MHCII as well as of the costimulatory molecules CD80 and CD86 using the gating strategy shown in Additional file 4: Figure S4. Interestingly, addition of C3a and C5a did not induce upregulation of CD80 (Additional file 5: Figure S5b), but we found a tendency that it might enhance expression levels of MHCII (Additional file 5: Figure S5a) and CD86 (Additional file 5: Figure S5c) since we detected an upregulation of both molecules in 4 of 5 cases.

The recall antigen pool CEFT was purchased from JPT, Germany (#PM-CEFT) and stored at − 20 °C after reconstitution (25 mg/ml in DMSO). CEFT antigen pool consisted of antigenic peptides from human Cytomegalovirus (HHV-5; CMV), Epstein–Barr virus (HHV-4; EBV), Influenza A and Clostridium tetani. This positive control pool contained 27 peptides selected from defined HLA class I and II-restricted T-cell epitopes. Considering the high vaccination frequency against Influenza and C. tetani and the high prevalence of CMV and EBV in the general population in Germany recall antigen responses were expected for all patient samples. Staphylococcus aureus enterotoxin B (SEB) was purchased from Sigma Aldrich GmbH, Germany (#S4881) and stored at − 20 °C (1 mg/ml in sterile, endotoxin-free H₂O). Synthetic HBV peptide libraries of HBcAg and HBsAg were purchased from JPT, Germany (#PM-HBV-CP and #PM-HBV-lEP, respectively) and stored at − 20 °C after reconstitution (50 mg/ml in DMSO). Synthetic HBcAg represented a mix of 44 peptides (15 amino acids each, 11 aa overlap, peptide scan 15/11) comprising the whole amino acid sequence of HBcAg, the 21.5 kDa capsid protein of HBV (genotype A2 subtype adw2, UniProt: P0C693). Synthetic HBsAg represented a mix of 98 peptides (15 amino acids each, 11 aa overlap, peptide scan 15/11) comprising the whole amino acid sequence of HBsAg, the 43.7 kDa surface protein of HBV (genotype A2 subtype adw2, UniProt: P17101). DMSO was obtained from Th. Geyer, Germany (#RO/A9941/000100). Glucose was provided by Carl Roth, Germany (#HN06.1). Human C3a and human recombinant C5a were purchased from Hycult Biotech, Germany (#HC2126, #HC2101). C3a was reconstituted in ddH₂O (0.5 mg/ml) and stored at − 80 °C whereas C5a was reconstituted in ddH₂O (0.15 mg/ml) and stored at − 20 °C. ELISA MAX Deluxe Sets for IFNγ (#430106) and IL2 (#431806) were obtained from BioLegend, Germany. PolyHRP80 streptavidin conjugate was purchased from SDT Reagents, Germany (#SP80C).

The following antibodies were used in line with this study: anti-CD3 BV605 (OKT3), anti-CD4 FITC (A161A1), anti-CD8a PE (HIT8a), anti-CD25 BV421 (BC96), anti-CD28 APC (CD28.2), anti-HLA-A, B, C BV605 (W6/32), anti-HLA-DR APC (L243), anti-CD80 BV711 (2D10), anti-CD86 PeCy7 (IT2.2) and Human TruStain FcX™ (Fc Receptor Blocking Solution). All were purchased from BioLegend. Erythrocytes were eliminated using Red Blood Cell Lysis Buffer (BioLegend) and Live/Dead staining was performed using Zombie NIR™ Fixable Viability Kit (BioLegend). DPBS with 1% FBS (Thermo Fisher Scientific) and 0.09% sodium azide (Sigma-Aldrich) was used as FACS staining buffer.

Eighty-seven healthy donors of the blood transfusion service at the University Medical Center Hamburg-Eppendorf were enrolled anonymously in the study (Table 1), which was approved by the Ethics Committee of the Hamburg Chamber of Physicians (WF-051/12). Clinical data such as gender, age, and HBV vaccination status were provided.

Venous blood from healthy donors was collected in sterile 7.5 ml Lithium Heparin Monovettes (Sarstedt, Germany). 0.5 ml of whole blood were then transferred to sterile, pyrogen-free 2 ml tubes (Sarstedt, Germany) and stimulated with HBV antigens (HBsAg 50 µg/ml). Samples stimulated with 0.9% (w/v) NaCl solution served as negative control whereas SEB (1 µg/ml) and CEFT (50 µg/ml) stimulated samples served as positive controls. Glucose (2 mg/ml final concentration; pre-diluted in sterile 0.9% (w/v) NaCl solution) was added to each tube to further enhance cytokine secretion as described previously [3]. Following titration, the complement factor C3a was used at a final concentration of 0.1 µg/ml whereas C5a was used at a final concentration of 0.75 µg/ml. All tubes were incubated at 37 °C for 24 h. Upon centrifugation plasma supernatants were aspirated, stabilized with 0.045% (w/v) NaN₃ and stored at − 20 °C until cytokine measurement by ELISA. The amount of biomaterial per donor was not always sufficient to perform all stimulations which is why for some donors conditions had to be left out. This explains variations with regard to the group size (n).

IFNγ and IL2 concentrations in human plasma were determined using ELISA MAX Deluxe Sets (BioLegend) following the manufacturer’s protocol. Anyhow, in order to increase assay sensitivity the manufacturer’s Avidin-horseradish peroxidase conjugate was substituted by a PolyHRP80 streptavidin–horseradish–peroxidase conjugate.

Samples were initially diluted as follows: negative control (NaCl) 1/5, 1st positive control (SEB) 1/2500 for IFNγ and 1/500 for IL2, 2nd positive control (CEFT) 1/50 for IFNγ and 1/25 for IL2, test samples (HBcAg and HBsAg) 1/5. If a test sample’s absorbance value fell outside the standard curve range, these samples were subsequently retested with a tenfold higher or lower dilution, respectively. A seven-point standard curve (1–64 pg/ml) was used for quantitation. All samples were measured in duplicates with a MAX 002 plate reader (Dynex Technologies) followed by data analysis using Microsoft Excel software.

HBsAg-specific antibodies were quantified in plasma samples using the Elecsys^® AntiHBs-Kit (Roche) following the manufacturer’s protocol. The measurements were performed and the according data provided by the Institute for Laboratory Medicine and Microbiology at the University Hospital Brandenburg.

Phenotypic characterization of the whole blood cells upon stimulation was performed using a LSR Fortessa cytometer (BD). For flow cytometric analysis blood was stimulated as described above. Subsequently, red blood cells were lysed by incubating whole blood with Red Blood Cell Lysis Buffer for 20 min at RT. For live/dead discrimination cells were then washed with DPBS (350 g, 5 min), re-suspended in DPBS (1 × 10⁷/ml) and stained with Zombie NIR™ dye for 20 min at RT. Cells were washed with staining buffer (350 g, 5 min) and incubated in staining buffer with Human TruStain FcX™ for 10 min at RT. Subsequently, cells were stained with different sets of antibodies. Anti-CD3 BV605, anti-CD4 FITC, anti-CD8a PE, anti-CD25 BV421 and anti-CD28 APC antibodies were used to assess the activation state of T cells whereas APCs were analyzed with anti-HLA-A, B, C BV605, anti-HLA-DR APC, anti-CD80 BV711 and anti-CD86 PeCy7 antibodies. Cells were stained for 30 min at RT, washed and finally re-suspended in fresh staining buffer.