An AGM model for changes in complement during pregnancy: neutralization of influenza virus by serum is diminished in late third trimester

Abstract

Pregnant women in the third trimester are at increased risk of severe influenza disease relative to the general population, though mechanisms behind this are not completely understood. The immune response to influenza infection employs both complement (C') and antibody (Ab). The relative contributions of these components to the anti-viral response are difficult to dissect because most humans have pre-existing influenza-specific Abs. We developed the African green monkey (AGM) as a tractable nonhuman primate model to study changes in systemic innate immunity to influenza during pregnancy. Because the AGMs were influenza-naïve, we were able to examine the role of C' in influenza virus neutralization using serum from non-pregnant animals before and after influenza infection. We determined that serum from naïve AGMs neutralized influenza via C', while post-infection neutralization did not require C', suggesting an Ab-mediated mechanism. The latter mimicked neutralization using human serum. Further, we found that ex vivo neutralization of influenza with both naïve and influenza-immune AGM serum occurred by virus particle aggregation and lysis, with immune serum lysing virus at a much higher rate than naïve serum. We hypothesized that the anti-influenza C' response would diminish late in AGM pregnancy, corresponding with the time when pregnant women suffer increased influenza severity. We found that influenza neutralization capacity is significantly diminished in serum collected late in the third trimester. Strikingly, we found that circulating levels of C3, C3a, and C4 are diminished late in gestation relative to nonpregnant animals, and while neutralization capacity and serum C3a return to normal shortly after parturition, C3 and C4 levels do not. This AGM model system will enable further studies of the role of physiologic and hormonal changes in downregulating C'-mediated anti-viral immunity during pregnancy, and it will permit the identification of therapeutic targets to improve outcomes of influenza virus infection in pregnant women.

Figure 1. Comparison of C3 concentrations in NHS and NAGS.

NAGS and NHS were used to compare C3 concentrations by western blot, relative to purified Hu C3 controls. Controls of 50 ng, 100 ng, and 200 ng are in lanes 1, 2, and 3, respectively, on each of the three blots. 1∶500, 1∶250, and 1∶50 dilutions of NAGS are in lanes 4, 5, and 6, respectively and of NHS are in lanes 7, 8, and 9, respectively. One NAGS donor and one NHS donor were on each blot, as indicated in the figure by NAGS or NHS number. C3 β- and C3 α-chain and their MW in kilodaltons are indicated on each blot.

Figure 2. Flow cytometry-based PR8-GFP neutralization assay.

PR8-GFP was mixed with a range of NAGS dilutions for 40 min at 37°C, and then used to infect U937 cells for 48 hr. Flow cytometry was used to determine the percentage of the cell population that was positive for GFP expression. Controls were: (A) Virus with no serum; (B) Cells only—no serum and no virus; (C) Virus plus 1∶40 HI AGM serum. AGM serum dilutions used to treat virus were: (D) 1∶40 NAGS; (E) 1∶80 NAGS; (F) 1∶160 NAGS; (G) 1∶320 NAGS. (H) Percent of cells that were GFP-positive following neutralization with indicated dilutions of NAGS from one example animal, expressed relative to that found for the no serum control.

Figure 3. Comparison of C′-dependent neutralization of parainfluenza and influenza viruses by NHS and NAGS.

(A) Influenza neutralization. PR8-GFP was incubated with the indicated fold dilutions of pooled Hu or AGM HI sera or varying dilutions of pooled NHS or NAGS for infection of U937 cells and analyzed as described in Fig. 2 (pooled samples are comprised of 4 individuals, Hu or AGM). Bars indicate mean of percent infected over triplicate experiments plus SD. ** p<0.001; *** p<0.0001 based on t-tests on log-transformed percentages for matching serum treatments for NHS and NAGS; treatments with no differences shown were not significantly different. (B) Variability in PR8 neutralization among samples from individual AGMs. PR8-GFP was incubated with the indicated fold dilutions of NAGS from four representative AGMs and analyzed as described in (A). There was no significant difference among animals based on a one-way ANOVA of log-transformed neutralization percentages. (C) Variability in PR8 neutralization among samples from individual Hus. Performed as in panel B, but with three individual Hus. Based on a one-way ANOVA of log transformed neutralization percentages, Hu subjects were significantly different from one another; p = 0.0045.

Figure 4. Effect of HI on the neutralization capacity of pre- and post-immune sera from AGM experimentally infected with PR8.

Two AGMs were inoculated with PR8 as described in the Materials and Methods. Pre-immune sera (d 0) and sera collected at d 14 post-infection were tested at the indicated fold dilutions for in vitro neutralization of PR8-GFP as described in Fig. 2. Results are expressed as the mean plus SD from triplicate experiments to determine the percentage of GFP-positive cells compared to in vitro control infections (which lacked serum treatment). *** p<0.0001 as compared to matching serum treatment for the same animal at day 0 (shown on day 14 values) based on t-tests on log transformed neutralization percentages.

Figure 5. EM visualization of PR8 neutralization mechanism and kinetics by pre-immune, post-immune, and HI AGM sera.

Sucrose-purified PR8 was incubated with a 1∶20 dilution of normal or HI (A) pre-immune or (B) post-immune AGM serum at 37°C for 0, 5, or 30 min, as indicated. These mixtures were fixed and negatively stained on a carbon-coated mesh grid and analyzed by EM. (C) Zoom of indicated area of post-immune 5 min time point with arrows indicating nucleocapsid leakage from lysed virus particles. Bar indicates 100 nm in each image, and magnification is as follows: (A) 0 min- 68,000x; 5 min- 68,000x; 30 min- 49,000× (B) 0 min- 49,000x; 5 min- 49,000x; 30 min- 49,000× (left); 98,000× (right).

Figure 6. Capacity of T3 and postpartum serum C′ to neutralize PR8-GFP.

C′ neutralization capacity was quantified as a function of the serum concentration effective in achieving 50% neutralization (ED₅₀) for each animal. (A) Logistic regressions for two example animals, where dashed line crossing the x-axis (a or b) represents the ED50 for each animal; R² is the coefficient of determination as an indication of how well the data points fit the regression model. ED50s were determined for AGMs at a range of gestation and postpartum dates: (B) T3 = 110–165 days; Shapiro-Wilks normality test indicated T3 population was not normally distributed; difference between T3 and control group tested with Mann-Whitney compare ranks two-tailed non-parametric test. (C) Early T3 = 110–128 days, Mid T3 = 129–146 days, Late T3 = 147–165 days gestation; (D) PC =  parturition check-in, 5–10 days after parturition; PP =  postpartum, 2–6 months after parturition. Differences among groups in panels C and D were tested with one-way ANOVA with Tukey correction for multiple comparisons. Lines within data point distribution represent mean +/−SD in panels B–D; * p<0.05; ** p<0.001; *** p<0.0001.

Figure 7. C′ factor quantification in T3 and postpartum AGMs.

AGM serum C3 β-chain was quantified by western blot relative to purified Hu C3 standards. (A) Representative western blot with 1∶200 (lanes 1, 3, and 5) and 1∶400 (lanes 2, 4, and 6) serum dilutions shown sequentially for each of three different animals (1217, 1246, and 1247). Standards of 200 ng, 100 ng, and 50 ng of human C3 are in lanes 7, 8, and 9, respectively. (B) Quantification of serum C3 β-chain in T3 vs control animals, based on C3 β-chain band associated with 200 ng of standard Hu C3; difference between groups tested with unpaired two-tailed t-test. (C) Quantification of serum C3 β-chain across T3 tertiles (times defined in legend for Fig. 6C). ELISAs were used to quantify (D) C3, (E) C4, and (F) C3a, relative to purified human standards for each protein, during late T3 and at postpartum timepoints, as defined in the legend for Fig. 6D. Differences among groups in panels C–F were tested with one-way ANOVA with Tukey correction for multiple comparisons. Lines within data point distribution represent mean +/− SD in panels B–F; * p<0.05; ** p<0.001; *** p<0.0001.