Clonal amplification and maternal-infant transmission of nevirapine-resistant HIV-1 variants in breast milk following single-dose nevirapine prophylaxis

Abstract

Background: Intrapartum administration of single-dose nevirapine (sdNVP) reduces perinatal HIV-1 transmission in resource-limiting settings by half. Yet this strategy has limited effect on subsequent breast milk transmission, making the case for new treatment approaches to extend maternal/infant antiretroviral prophylaxis through the period of lactation. Maternal and transmitted infant HIV-1 variants frequently develop NVP resistance mutations following sdNVP, complicating subsequent treatment/prophylaxis regimens. However, it is not clear whether NVP-resistant viruses are transmitted via breastfeeding or arise de novo in the infant.

Findings: We performed a detailed HIV genetic analysis using single genome sequencing to identify the origin of drug-resistant variants in an sdNVP-treated postnatally-transmitting mother-infant pair. Phylogenetic analysis of HIV sequences from the child revealed low-diversity variants indicating infection by a subtype C single transmitted/founder virus that shared full-length sequence identity with a clonally-amplified maternal breast milk virus variant harboring the K103N NVP resistance mutation.

Conclusion: In this mother/child pair, clonal amplification of maternal NVP-resistant HIV variants present in systemic and mammary gland compartments following intrapartum sdNVP represents one source of transmitted NVP-resistant variants that is responsible for the acquisition of drug resistant virus by the breastfeeding infant. This finding emphasizes the need for combination antiretroviral prophylaxis to prevent mother-to-child HIV transmission.

Figure 1

Maximum-likelihood trees of env and pol genes from 4403–4419 MTCT pair. Maternal breast milk virus env(A) and pol(B) sequences (blue stars = right breast; blue circles = left breast) were obtained from a sample collected at 4 weeks postpartum and infant virus env(A) and pol(B) sequences (orange squares) were derived from a plasma sample collected at 12 weeks of age. Groups of identical/near-identical sequences (defined as sequences with four or fewer nucleotide substitutions compared to consensus sequence) in maternal breast milk (vertical blue bar) or infant plasma (vertical orange bar) are indicated. Greater than 90% of the indicated near-identical sequences had fewer than four substitutions compared to consensus. Numerals at nodes indicate approximate bootstrap support values ≥ 70%. The scale bar represents 0.001 (0.1%) nucleotide substitutions per site.

Figure 2

Highlighter analyses of HIV sequences from infant plasma and maternal breast milk. (A) Multiple 5’-half and 3’-half genome sequences derived from infant plasma at the time of HIV diagnosis are aligned below the infant consensus sequence (Con). Nucleotide differences from the inferred T/F consensus sequence are indicated by tic marks color-coded for each base. IUPAC denotes International Union of Pure and Applied Chemistry ambiguous base assignments caused by Taq polymerase-induced nucleotide misincorporation occurring in the first two cycles of the PCR amplification. Flanking gray boxes indicate regions not amplified. Gray tics indicate deletions. The arrow indicates the location of the K103N mutation in the RT gene in consensus and all SGA sequences (AAA to AAC mutation). The horizontal axis indicates nucleotide positions based on HXB2 reference sequence numbering. Consensus sequence begins at nucleotide position 484 in the 5’ long terminal repeat (LTR) U5 and extend to position 9,606 in the 3’ LTR R. (B) Two 9 Kb amplicons, and multiple pol and rev/env fragments representing low-diversity breast milk sequences collected at four weeks postpartum (shown at the left of the blue bars in Figure 1) are aligned below the consensus sequence (identical to Con from Figure 1A). The arrow represents the location of the K103N mutation. Clustered mutations enclosed in elipses reflect recombination with more divergent maternal sequences. The horizontal axis indicates HXB2 nucleotide positions. Sequences begin at nucleotide position 582 in the 5’ LTR U5 and extend to position 9,606 in the 3’ LTR R. Boxes corresponding to LTRs and 9 major protein coding regions of the HIV genome are shown at the bottom.

Figure 3

Selection of HIV-1 RT drug-resistance mutations following ARV drug exposure in maternal plasma samples. Amino acid alignment of pol gene regions indicating single point mutations associated with high-level resistance to ARV drugs in subject 4403. CON denotes amino acid consensus sequence from the earliest maternal plasma sample. Bold amino acid residues in the consensus correspond to selected drug-resistant point mutations. Point mutations numbered 1–4 are as follows: 1 for K103N, 2 for V106M, 3 for Y188C/L, 4 for G190A. Dashes denote identity with consensus, while letters correspond to amino acid changes. Asterisks indicate sequences derived from near end-point diluted PCR positive wells (50% positive PCR wells). All other HIV pol sequences were derived by SGA as described [19]. bmL denotes milk sequences derived from left breast; all other sequences are derived from blood plasma. Low-diversity clonally-amplified sequences (0–2 nucleotides away from consensus; bmL_pG5 is a recombinant virus harboring a cluster of 6 nucleotide mismatches shared with more divergent virus sequences) are indicated with brackets.

Figure 4

Neighbor-joining trees of maternal 4403 env sequences from longitudinal plasmas samples. Full-length maternal env sequences generated by SGA from third trimester pregnancy and follow-up visits up to 12 weeks postpartum were aligned and plotted as radial Neighbor-joining trees. Clusters of monotypic and low-diversity sequences are encircled; the number of nucleotides by which these sequences differ from MRCA (branch node) is indicated for each cluster. The scale bar represents 0.01 nucleotide substitutions per site (1% divergence).