A nested case-control study was undertaken to investigate the association between the FCGR2C variants and HIV-1 perinatal acquisition in children, combining data from past studies of five perinatal cohorts at two hospitals in Johannesburg, South Africa (24–27). One of the five cohorts consists of 546 HIV-infected children who were recruited as part of two sequential randomized clinical trials (NEVEREST 2 and 3) (24–26). The remaining four cohorts comprised of 849 HIV-1 infected mothers and their infants who were recruited and followed prospectively, of whom 83 (10%) infants acquired HIV (27). In the present study, only samples that were found and with sufficient volume were genotyped. FCGR2C genotypic data from 99 out of 546 and 77 of 83 HIV-1-infected children (cases) from the NEVEREST and mother-infant cohorts, respectively (n = 176) were compared with 349 of the HIV-exposed uninfected children (controls).

Mode of transmission was defined according to the presence or absence of detectable HIV-1 deoxyribonucleic acid (DNA) in the infant at birth and six weeks of age. Infants that tested HIV-1 positive at six weeks of age, but who were negative at birth, were considered to be infected intrapartum (during labor and delivery) (n = 31), while infants that tested HIV-1 positive at birth were considered infected in utero (n = 19). Infants who were HIV-1 positive at six weeks, but had no birth sample, were categorized as ‘undetermined’ (n = 28). In the ‘undetermined’ category, 25/28 (89.2%) mothers received single-dose nevirapine or triple-drug combination therapy (two nucleoside reverse transcriptase inhibitors with either a protease inhibitor or non-nucleoside reverse transcriptase inhibitor) known to reduce intrapartum transmission (27–29). Genotyping generated a result for all the FCGR2C variants assessed in this study in 27 out of the 28 samples. It was thus concluded that the majority (n = 27) of infants were likely infected in utero and were combined with the in utero group to form an in utero-enriched group. For the NEVEREST cohort, there were no birth samples as the children were recruited from six weeks of life. They were therefore classified as mixed transmission since a few were breastfeeding infections and in utero infections could not be distinguished from intrapartum infections (n = 99). All study participants were black South Africans and received nevirapine for prevention of MTCT. Maternal antiretroviral therapy was not routinely used at the time.

Ethics approval for the study was obtained from the University of the Witwatersrand Human Research Ethics Committee (Reference numbers: M170585; M180575).

FCGR2C copy number and SNPs that affect gene expression – c.169T>C (p.X57Q), c.798+1A>G, and the FCGR2B/C promoter variant at position c.−386G>C and c.−120A>T – were determined using the FCGR-specific multiplex ligation-dependent probe amplification assay (MRC Holland, Amsterdam, The Netherlands) according to manufacturer’s instructions. Amplicons were separated by capillary electrophoresis on an ABI Genetic Analyser 3130 (Life Technologies, Applied Bio systems, Foster City, CA, USA) and fragments analyzed with the Coffalyzer.NET software (MRC Holland) using peak height as a measure of gene/allele copy number. We did not utilize gene-specific polymerase chain reactions (PCR) to distinguish FCGR2B and FCGR2C promoter sequences since earlier findings indicate that African individuals do not possess the promoter variant in FCGR2B, and thus any detected c.−386G>C minor alleles were in FCGR2C (12).

The FCGR2C c.134-96C>T (rs114945036) variant was genotyped through conventional PCR and Sanger nucleotide sequencing. In brief, a 6,374 base pair fragment was amplified with the Expand Long Template PCR System (Roche, Mannheim, Germany) using the FCGR2B/C sense primer (5’-ATGTATGGGGTGTCTGTGTGTC-3’) and FCGR2C-specific antisense primer (5’-CTCAAATTGGGCAGCCTTCAC-3’) (15). The PCR reaction consisted of ~20 ng genomic DNA as template, 3.75 U Expand Long Template enzyme mix, 5 µl 10× PCR buffer 3 (2.75 mM MgCl₂), 500 µM of each deoxynucleotide, 0.3 µM of each oligonucleotide primer, and molecular grade water to a final volume of 50 µl. The PCR conditions were initial denaturation at 94°C for 2 minutes, followed by 10 cycles of 94°C for 10 seconds (denaturation), 60°C for 15 seconds (annealing) and 68°C for 7 minutes (elongation). Thereafter, 25 cycles repeat process of denaturation, annealing and elongation respectively at 94°C for 15 seconds, 60°C for 15 seconds and 68°C for 7 minutes plus 20 seconds cycle elongation for each successive cycle; and a final elongation cycle at 72°C for 7 minutes. The internal antisense primer (5’-CCTCCACTGACCAGAAAGCAC-3’) was used in standard BigDye Terminator v3.1 Cycle Sequencing reactions. Sequences were analyzed in Sequencer version 4.5 (Gene Codes Corporation, Ann Arbor, MI) and area under the curve of the electropherogram used to determine allele count for individuals bearing more than two FCGR2C gene copies.

The SNP nomenclature used in this manuscript refers to the amino acid positions in the full protein, in accordance with the Human Genome Variation Society (HGVS) guidelines (30). The numbering of nucleotides is according to the Genome Reference Consortium Human Reference 38 [GRCh38 (hg38)].

Categorical data were summarized as proportions and the Fisher’s Exact test was used for comparisons. For numerical data, the t-test was used for comparison of means. Univariate and multivariate analyses were conducted to determine factors associated with perinatal HIV acquisition. Adjustment for multiple comparisons was performed using the Bonferroni correction, which considered 16 independent tests — four unrelated clinical subgroups each tested for four variants (gene copy number, c.134-96C>T, c.169T<C, and c.−386G>C). Both unadjusted and adjusted P values are reported. Analysis of an association between FCGR2C variants and HIV-1 acquisition was limited to variants whose allele frequencies were ≥ 5%. Due to the low frequencies of minor allele homozygotes, their effect was tested using dominant model approach, where participants were divided into two genotype groups: homozygous genotype of the major allele and the two genotypes containing at least one minor allele. All analyses were performed in STATA version 15.1 (StataCorp LP, Texas, USA).

Linkage disequilibrium between FCGR2C functional variants and CNRs was computed using the Haploview software package (31) and expressed as D prime (D′) and square of the correlation coefficient (r²). The closer D′ is to 1 the stronger the LD between two loci. Hardy-Weinberg equilibrium was considered for individuals with two gene copies and the statistics abstracted from the Haploview analysis output. For the analysis, genotypic data with multiple gene copies were considered homozygous if all copies carried the same allele or heterozygous when both alleles were present.

This nested case-control study investigated FCGR2C genotypic data from 525 children to determine the role of FCGR2C variants and HIV-1 acquisition in South African children born to women living with HIV-1. The cohort includes 176 HIV-1 infected (cases) and 349 HIV-exposed-uninfected (controls) children. The HIV-1 infected children comprised four transmission mode groups: in utero (n = 19), in utero-enriched (n = 46), intrapartum (n = 31) and mixed (n = 99). Overall, there was no significant difference in sex, gestation and breastfeeding status between the HIV-1 infected and HIV-1 uninfected cohort. However, the total HIV-1 infected and HIV-1 exposed-uninfected groups differed significantly in birth weight at delivery. Specifically, a higher proportion of HIV-infected children had a birth weight below 2500 g (22% vs. 11%; P = 0.001, P _Bonf = 0.016) ( Table 1 ).

The FCGR2C gene, highly homologous to FCGR2B in the first six exons and FCGR2A in the last two exons, is subject to CNV within previously described distinct regions ( Figure 1 ). We did not observe any individual with a complete absence of the FCGR2C gene. Overall, FCGR2C CNV occurred in 166/525 (32%) children, with the frequency of duplications (n = 114) 2.2-fold higher than deletions (n = 52) (69% vs. 31%). The copy number distribution was significantly different between the HIV-1 infected and HIV-1 exposed-uninfected groups but not after Bonferroni correction (P = 0.010, P _Bonf > 0.05) ( Table 2 ). Variation in copy number among the whole study cohort was observed more frequently in CNR1, which encompasses a complete FCGR2C copy, HSPA7 and FCGR3B (28%; n = 147) than CNR2, with an incomplete FCGR2C copy, HSPA6 and FCGR3A (2.7%; n = 14). In six instances (1.1%), we observed CNV for only FCGR2C in the absence of duplicated/deleted flanking genes, as previously described among the South African black population (12). A duplication in both CNR1 and CNR2 was observed in one individual. Given the differences between the CNRs, their copy number variability was determined separately.

Within CNR1, CNV was significantly different between the HIV-1 infected and HIV-1 exposed-uninfected groups (P = 0.009, P _Bonf > 0.05). This difference was primarily determined by gene deletions. There were a higher number of HIV-1 exposed-uninfected children with a single gene copy compared to HIV-1 infected children (36% vs. 14%) ( Table 2 ). Using two FCGR2C gene copies as reference, the possession of a single gene copy was independently associated with reduced odds of HIV-1 acquisition (OR = 0.29; 95% CI 0.12-0.71; P = 0.007, P _Bonf > 0.05) and retained significance after controlling for birthweight and FCGR2C genotypes (AOR = 0.37; 95% CI 0.15-0.90; P = 0.029, P _Bonf > 0.05) ( Table 3 ). The CNR2 and the novel CNR4 variability were excluded from further association analysis due to low frequencies (< 5%).

We further genotyped functional FCGR2C variants that determine the expression of FcγRIIc (c.−386G>C, c.169T>C and c.798+1A>G) and the c.134-96C>T variant that associated with risk of HIV-1 acquisition in the Thai RV144 HIV-1 vaccine trial ( Figure 1 ). To assess the role of the FCGR2C genotypes and allele distribution in perinatal HIV-1 acquisition, children with a single FCGR2C gene copy were considered homozygous; those with more than one FCGR2C gene copy were considered homozygous if all the alleles were the same or heterozygous if both alleles were present. With MLPA, we obtained genotypic data from 166 out of the 176 HIV-1 infected for c.169T>C and c.−386G>C variants.

A SNP in exon 3 (c.169T>C) that results in an open reading frame (ORF) or a stop codon determines the expression of FcγRIIc when present with the minor allele of two splice variants in intron 7 (c.798+1A>G and c.799-1G>C). While 129/515 (25%) of individuals carried at least one FCGR2C-ORF (c.169C allele), only 4/129 (3%) individuals possessed the c.798+1G minor allele that represents the classic FCGR2C-ORF and predicts FcγRIIc expression (6, 32). Of the four individuals with the c.798+1G minor allele, three were HIV-1-exposed uninfected and one HIV-1 infected. Conversely, the c.169C allele co-occurred with the c.798+1A allele in 97% (n = 125) of the participants, representing the non-classic FCGR2C-ORF that does not yield surface expression of FcγRIIc. The c.799-1G>C splice cite variant was not genotyped, since it has been shown that the c.169C and c.798+1A alleles are syntenic with the c.799-1G allele in South African population (12).

While the c.169T>C variant alone does not result in surface expression of FcγRIIc, it may have other unknown functional consequences and was therefore investigated for a possible association with HIV-1 perinatal transmission. The c.169T>C genotype distribution was significantly different between HIV-1 infected and HIV-1 exposed-uninfected children (P = 0.002, P _Bonf = 0.032) ( Figure 2 i). In a dominant model, the c.169C was overrepresented in the HIV-1 infected compared to the uninfected children (32% vs. 22%) and significantly associated with increased odds of HIV acquisition (OR = 1.68; 95% CI 1.11-2.55; P = 0.013, P _Bonf > 0.05). The strength of association increased after adjusting for birthweight and CNR1 copy number (AOR = 2.39; 95% CI 1.45-3.95; P = 0.001, P _Bonf = 0.016). For the FCGR2B/C promoter variant at position –386G>C, which modulates gene expression levels, no significant difference in genotype frequency was observed between the HIV-1 infected and HIV-1 uninfected cohort (P = 0.288) ( Figure 2 ii). The homozygous −386 CC genotype was not observed in any individual. Due to the low frequency of the splice site variant c.798+1A>G minor allele, an association analysis was not conducted.

The FCGR2C c.134-96C>T genotype distribution was significantly different between the children who acquired HIV-1 and those who did not (P = 0.002, P _Bonf = 0.032) ( Figure 2 iii). In particular, the c.134-96T-allele was overrepresented in the HIV-1-infected children compared to the exposed-uninfected children (58% vs. 42%; P = 0.001, P _Bonf = 0.016). The overrepresentation was primarily driven by the in utero-enriched (63% vs. 42%; P = 0.011, P _Bonf > 0.05) and mixed (59% vs. 42%; P = 0.004, P _Bonf > 0.05) infected groups ( Figure 2 iii). We combined the in utero-enriched and mixed transmission groups into a larger in utero-enriched group, excluding the 27 HIV-1 infected and 75 HIV-1 exposed uninfected breastfed individuals, and still observed overrepresentation of the minor allele in the HIV-1 infected children (60% vs. 43%; P = 0.002, P _Bonf = 0.048) (data not shown).

The association between FCGR2C c.134-96C>T variant and HIV acquisition was assessed in a univariate model and a multivariate model that controlled for birthweight and CNR1 copy number, which were statistically significant at univariate analysis ( Table 4 ). In a dominant model, the c.134-96C>T minor allele was associated with increased odds of perinatal HIV-1 transmission at univariate (OR = 1.89; 95% CI 1.31-2.73; P = 0.001, P _Bonf = 0.016) and multivariate analysis (AOR = 1.89; 95% CI 1.25-2.87; P = 0.003, P _Bonf = 0.048). The association was specific to the in utero-enriched (OR = 2.34; 95% CI 1.24-4.42; P = 0.009, P _Bonf > 0.05) and the mixed transmission groups (OR = 1.94; 95% CI 1.24-3.06; P = 0.004, P _Bonf > 0.05) but not the in utero (OR = 1.24; 95% CI 0.49-3.12; P = 0.653) and intrapartum groups (OR = 1.29; 95% CI 0.62-2.69; P = 0.500). Statistical significance was retained in both in utero-enriched (AOR = 2.49; 95% CI 1.31-4.76; P = 0.006, P _Bonf > 0.05) and mixed transmission groups (AOR = 2.06; 95% CI 1.28-3.30; P = 0.003, P _Bonf = 0.048) after adjusting for birthweight.

The observed genotype frequencies for FCGR2C c.–386G>C, c.134-96C>T, c.169T>C were in Hardy-Weinberg equilibrium (P > 0.05). We also observed that 71% (39/55) of children carrying a c.169C were heterozygous for the FCGR2B/C promoter variant (data not shown). We analyzed linkage disequilibrium between the FCGR2C variants and CNRs, with and without considering the CNV. It was important to determine whether the observed variants associated with HIV-1 acquisition act independently or linkage disequilibrium plays a part. Our result demonstrated high linkage disequilibrium between c.134-96C>T and c.169T>C both without considering CNV (D′ = 0.867; r² = 0.319) and when only those with two gene copies were included (D′= 0.908 and r² = 0.213) ( Figure 3 ). Both c.134-96C>T and c.169T>C independently associated with increased odds of HIV-1 acquisition, but in multivariate analysis, c.134-96C>T retained significance (AOR = 1.91; 95% CI 1.23-2.96; P = 0.004, P _Bonf > 0.05) while c.169T>C did not (AOR = 1.14; 95% CI 0.70-1.86; P = 0.590).

Maternal viral load is a key determinant of MTCT of HIV-1 infection. For the NEVEREST cohorts, the maternal viral load was not recorded. However, in the birth cohort with the defined modes of transmission, information on maternal HIV-1 viral load was available for 279 mothers of whom 70 transmitted HIV to their infants and 209 did not. The mean HIV-1 viral load was significantly higher in transmitting mothers than non-transmitting mothers (4.53 vs. 3.95 log₁₀ copies/ml; P < 0.0001). The maternal FCGR2C c.134-96C>T variant independently associated with HIV-1 transmission (OR = 1.98; 95% CI 1.16-3.37; P = 0.012, P _Bonf = 0.192) and when controlled for maternal viral load and birthweight (AOR = 2.03; 95% CI 1.14-3.62; P = 0.016, P _Bonf = 0.256). In this small subset, the infant FCGR2C c.134-96C>T variant was independently associated with HIV-1 acquisition (OR = 1.92; 95% CI 1.14-3.23; P = 0.014, P _Bonf = 0.224). The significant association remained when adjusted for maternal viral load (AOR = 2.10; 95% CI 1.13-3.87; P = 0.018, P _Bonf = 0.288), specifically in the in utero-enriched transmission group (AOR = 2.67; 95% CI 1.33-5.37; P = 0.006, P _Bonf = 0.064) ( Table 5 ).

We next evaluated mother-child FCGR2C c.134-96C>T genotype concordance and association with HIV-1 acquisition. Concordance was defined as mother and infant each bearing at least one T allele (mother-child: CT/TT-CT/TT) or both were homozygous for the C allele (CC). Discordance was defined as one individual within the dyad possessing a CC genotype and the other bearing a T allele (CT/TT). Possession of a T allele in both mother and infant was overrepresented in the HIV-1-infected children compared to the exposed-uninfected children (45% vs. 28%; P = 0.012, P _Bonf = 0.192). Independently, MTCT was more likely among the mother-child concordant CT/TT group compared to the concordant CC group (OR = 2.58; 95% CI 1.36-4.88; P = 0.004, P _Bonf = 0.064) and retained significance after adjusting for maternal viral load, birthweight and infant CNR1 copy number (AOR = 2.87; 95% CI 1.36-6.06; P = 0.006, P _Bonf = 0.096) ( Table 6 ).