Reduced placental transfer of antibodies against microbial and vaccine antigens in HIV-infected women in Mozambique

Antibody transplacental transfer is essential for conferring protection in newborns against infectious diseases. This transfer may be affected by gestational age and maternal infections, although the effects are not consistent across studies. We measured total IgG and IgG subclasses by quantitative suspension array technology against fourteen pathogens and vaccine antigens, including target of maternal immunization, in 341 delivering HIV− and HIV+ mother-infant pairs from southern Mozambique. Maternal antibody levels were the main determinant of cord antibody levels. HIV broadly reduced the placental transfer and cord levels of IgG and IgG1, but also IgG2 to half of the antigens. Plasmodium falciparum exposure and prematurity were negatively associated with cord antibody levels and placental transfer but this was antigen-subclass dependent. These findings suggest maternal infections may impact the efficacy of maternal immunization and confirm the lower transfer of antibodies as one of the causes underlying increased susceptibility to infections in HIV-exposed infants.


Introduction 37
Each year, 2.6 million deaths occur during the neonatal period, being infectious diseases the 38 leading cause of mortality, particularly in low-income countries [1,2]. Newborns are highly 39 vulnerable to pathogens due to their functional immunological differences from adults as a result 40 of living in a semi-allogeneic sterile environment, where exposure to microbial antigens is limited 41 [3-6]. For example, microorganisms such as respiratory syncytial virus (RSV) are generally 42 asymptomatic or cause mild disease in adults but induce acute bronchiolitis, viral pneumoniae 43 and croup in infants, being those between 2 and 6 months of age at the highest risk, especially in 44 low-income countries [7,8]. 45 Newborns mostly rely on the protection elicited by maternal antibodies transferred across the 46 placenta, which provide passive immunity against common pathogens [9]. Neonatal and child 47 immunization is essential for conferring protection in newborns and infants against vaccine-48 preventable diseases [10][11][12]. Vaccination is among the most cost-effective public health 49 measures worldwide [13], and regions with high rates of infant morbidity and mortality like sub-50 Saharan Africa benefit from the implementation of the Expanded Program of Immunization (EPI) 51 IgG is transferred across the placenta, being foetal IgG concentrations higher at the third trimester 58 [17], although some studies suggest that maternal IgE is also transferred to the foetus as IgG/IgE 59 complexes [18]. IgG subclasses have different affinities for the FcRn receptor leading to 60 differences in the efficiency of transfer [19]; the greatest transport occurs for IgG1, followed by 61 IgG4, IgG3, and finally IgG2 [20]. 62 significant differences in placental transfer between the two groups were found for IgG4 ( Figure  142 3-figure supplement 5). 143

Factors associated with cord blood levels of IgG and IgG subclasses 144
Maternal antibodies, HIV infection and P. falciparum exposure were the only variables with a 145 clear general impact on univariable models and were selected for multivariable models, in which 146 maternal antibody levels had the strongest positive correlation with cord antibody levels for all 147 the antigens and subclasses (Fig. 4a). However, the effect of maternal antibody levels was more 148 variable for IgG3-4 than for IgG and IgG1 subclasses. On average, a 10% increase in maternal 149 IgG levels was associated with 8.1% to 9.7% increases in IgG cord blood levels, depending on 150 the antigen. For IgG subclasses, a 10% increase in maternal antibody levels was associated with 151 increases of cord blood levels from 7.6% to 10.9% for IgG1, 5.4% to 9.6% for IgG2, 5% to 9.9% 152 for IgG3 and 5.3% to 9.3% for IgG4. intestinalis (2.9% to 7.1% reduction). For IgG2, an HIV negative effect was observed against B. 158 pertussis, S. dysenteriae, HBV and G. intestinalis (2.9% to 7.1% reduction), whereas HIV was 159 associated with an increase of 3.3% of IgG2 to RSV. Finally, we only detected a negative effect 160 of HIV infection on IgG3 levels to C. tetani (1.1% reduction) and IgG4 to P. falciparum (1.8% 161 reduction). 162 P. falciparum exposure was negatively associated with cord blood IgG levels against S. 163 dysenteriae and HBV, IgG1 against S. pneumoniae and rotavirus, IgG2 against HBV and IgG3 164 against C. diphtheriae and rotavirus (Fig. 4c). Depending on the IgG sublcass and antigen, 10% 165 increseas in P. falciparum exposure reduced the cord blood levels from 0.3% to 0.8%. 166 Previous studies suggest that PM rather than peripheral malaria affect transplacental transfer of 167 antibodies and lead to adverse outcomes due to the damaged placental tissue [35,42,43]. 168 Therefore, we explored the effect of PM on cord blood levels and placental transfer instead of P. 169 falciparum exposure despite the low number of women with any evidence of PM. When analysing 170 HIV-infected women only, PM was associated with lower B. pertussis IgG1, C. diphtheriae IgG2 171 and HBV IgG3 levels in cord blood (Figure 4-figure supplement 1). 172 Prematurity (Fig. 5a), previously shown to have a detrimental effect on placental transfer of 173 antibodies [44], increased the quality (AIC) of some of the above multivariable models. 174 Prematurity was associated with lower cord blood IgG levels against Hib (4.2% reduction 175 compared with on-term cord blood levels), V. cholerae (2.3% reduction), measles (5.8% 176 reduction) and C. parvum (3,8% reduction without statistical significance after adjusting for 177 multiple testing) 178 The rest of the variables (age, maternal anaemia, gravidity, low birth weight, IPTp treatment, 179 seasonality; and CD4 + T cell counts, ART and viral load for HIV-infected women) did not provide 180 an added value to the multivariable models. Univariable models did not show a consistent effect 181 of any variable across antigens or IgG subclasses, but some significant associations were found 182 for age and gravidity (Supplementary Material 1). 183

Factors associated with placental transfer of IgG and IgG subclasses 184
In multivariable models including HIV infection and P. falciparum exposure, HIV infection ( Fig.  185 6a) was associated with a generalized reduced placental transfer of IgG and IgG1 (from 2.1% to 186 6.7% reduction depending on the antigen). HIV infection was also associated with a reduced 187 transfer of IgG2 against B. pertussis, HBV and G. intestinalis, but was associated with an increase 188 in IgG2 RSV transfer (5.4% increase). Although adjusted p-values were not significant, a similar 189 When prematurity was added to the multivariable models, this additional covariable had a 200 negative effect on placental transfer of Hib and V. cholerae IgG antibodies (4.5% and 2.3% 201 reduction in premature vs on-term newborns, respectively) ( Fig. 5b). 202 The rest of variables were not added to the placental transfer multivariable models, because almost 203 none of the models improved when included. Similar to cord blood levels ones, the placental 204 transfer univariable models did not show a consistent effect of any variable not included in the 205 multivariable models across antigens or IgG subclasses (Supplementary Material 1). 206

Discussion 207
Our comprehensive analysis of maternal and cord plasma IgG and IgG subclasses against a wide 208 range of microbial and vaccine antigens allowed a depth immunoprofiling, that is essential to 209 decipher the mechanisms affecting antibody placental transfer and maternal and newborn 210 immunity in women chronically exposed to pathogens. We confirmed that the main determinant 211 of cord IgG and IgG subclass levels are the maternal corresponding antibody levels, and that 212 maternal HIV infection is associated with a generalized diminished IgG levels in the cord due to 213 low maternal levels but also to a broadly reduction of IgG and IgG1 placental transfer. 214 Maternal and cord blood antibody levels are usually correlated in many studies, suggesting that 215 maternal levels are the main determinant for transfer efficiency [9,45,46]. However, the effect 216 of HIV infection on placental transfer has not been consistently assessed and the few studies looking at its effect on maternal and cord blood levels mainly focussed on total IgG. Our results 218 showed that HIV infection reduced the IgG maternal levels for some antigens, the cord blood 219 levels overall, and also had a negative effect on transplacental transfer of IgG antibodies. It is 220 interesting that although we found higher maternal HBV and G. intestinalis IgG levels among 221 HIV-infected women, cord blood levels and transplacental transfer were lower than in HIV-222 uninfected women. Higher maternal antibody levels against these pathogens in HIV-infected 223 women may be due to an increased susceptibility to co-infections with these pathogens, as 224 The efficacy of IgG placental transfer also depended on the antigen. IgG1, IgG3 or IgG4 248 transferred better than IgG2, except for S. pneumoniae and G. intestinalis, for which IgG2 transfer 249 was higher. This was unexpected because it has been previously described that the greatest 250 transport occurs for IgG1, followed by IgG4, IgG3, and finally IgG2 [9, 20]. However, IgG1 251 levels were the highest for almost all antigens in cord blood, probably because the overall higher 252 levels of this IgG subclass in maternal blood. One exception was Hib that presented higher IgG2 253 cord levels than IgG1, although IgG1 transplacental transfer was higher than IgG2 consistently 254 with previous studies [56]. The mothers had a IgG2-predominant response to Hib, and 255 consequently higher IgG2 than IgG1 levels were found in cord blood as previously described [57, At the time of the study, malaria transmission intensity was very low in the area and only a few 267 women had active malaria during pregnancy. Nonetheless, we found a negative correlation 268 between P. falciparum exposure and both placental transfer and cord blood antibody levels for 269 some antigens and IgG subclasses. Previous studies are contradictory, as some found that PM led 270 to a reduction of the transplacental transfer of some pathogen-specific IgG to C. tetani [ in pregnancy has also been implemented in some countries, but not in Africa. Acellular pertussis 289 vaccine induces mainly IgG and IgG1 responses that are thought to confer protection [66,67]. 290 We found lower cord blood levels and a reduced placental transfer of IgG and IgG1 against B. 291 pertussis among HIV-infected women and those exposed to P. falciparum. These results highlight 292 the need for further studies assessing the impact of these infections on pertussis vaccine efficacy 293 and antibody placental transfer when implemented in pregnant women from African countries. A 294 current vaccine in development for maternal immunization is RSV [68]. Natural RSV infection 295 seems to elicit an IgG1 and IgG2 response against the F protein, the major target of the host's 296 immune response [69] and of some vaccines in development [70]. Antibodies binding to the F 297 protein were protective [71] and Palivizumab, an IgG1 monoclonal antibody against RSV F 298 protein with neutralizing function, has shown to be effective [72]. Here, IgG and IgG1 against 299 RSV F protein had the highest levels in cord blood compared to other subclasses, but HIV 300 infection reduced IgG cord blood levels and placental transfer in multivariable models. Instead, 301 IgG2 cord blood levels were increased by maternal HIV infection. Therefore, HIV infection could 302 compromise the levels of RSV neutralizing antibodies transferred to the newborn and, 303 consequently, diminish the effectivity of a RSV vaccine. 304 Unfortunately, we do not know what are the thresholds of antibody levels that confer protection 305 in our study, therefore it is difficult to infer the clinical relevance of the reductions in antibody 306 levels detected in cord blood from the-HIV infected women. A study in South Africa reported 307 that the frequency of HIV-infected and HIV-uninfected pregnant women with protective antibody 308 levels against pertussis, tetanus or HBV was similar, although the overall frequencies were low 309 (32%, 41% and 30%, respectively) [40]. This same study demonstrated that the proportion of 310 HIV-infected pregnant women reaching anti-Hib protective antibody levels was lower than HIV-311 uninfected women (35% vs 59%). Thus, for the implementation of maternal immunization 312 programs, the effect of HIV infection and P. falciparum exposure must be taken into account, 313 especially after demonstrating that both infections reduce the levels of antibodies in the cord blood 314 and therefore may compromise vaccines protective effect in the newborn. 315 In conclusion, our results demonstrate that maternal HIV infection was associated with reduced 316 levels of antibodies against a broad range of pathogens and vaccine antigens in cord blood. Part 317 of this reduction in antibody levels was due to altered antibody levels in the mother, which are 318 the main determinants of cord blood levels, but HIV-infection also diminished transplacental 319 transfer of antibodies. Importantly, IgG1 was the most affected by maternal HIV infection but, 320 depending on the pathogen, other subclasses were also affected. P. falciparum exposure also 321 reduced the levels and transfer of some antibodies, although overall the effect was lower than 322 optimal sample dilution for data analysis. Two blanks were added to each plate also for quality 396 control purposes. Sample distribution across plates was designed to ensure a balanced distribution 397 of groups and time-points. Single replicates of the assay were performed. 398

Statistical Analysis 399
To stabilize the variance, the analysis was done on log10-transformed values of the MFI 400 measurements. To select the sample dilution for each antigen-isotype/subclass-plate, the dilution 401 nearest to the midpoint between the two standard curve serial dilutions ranging the maximum 402 slope was chosen. If the maximum MFI value of a standard curve did not reach 15000, the 403 reference value was automatically set up at 15000, since below this point, standard curve data 404 does not seem trustworthy. If the MFI of the first sample dilution was lower than the MFI of the second dilution (hook effect), the second one was chosen. Plates were normalized using the 406 standard curve in each plate and the average standard curve from all plates -in both cases using 407 the dilution of the latter with the value closest to 15000 MFI. The MFI values of samples were 408 multiplied by the corresponding normalization factor (MFI value of the chosen dilution from the 409 average standard curve divided by the MFI value of same dilution in the plate curve). 410 The Shapiro-Wilk test of normality confirmed that most of the antibody data were not normally 411

Competing interests 455
The authors declare that they have no competing interests.