Multiple independent polymerase chain reactions (PCRs) were performed on peripheral blood mononuclear cell (PBMC) DNA from six mother-infant pairs, a total of 13 patients including one mother who gave birth to HIV-1 positive twins. Eight to eighteen clones from each patient were obtained and sequenced. The phylogenetic analysis was performed using a neighbor-joining tree of the 168 NC and p6 sequences from the mother-infant pairs (Fig. 1). This neighbor-joining tree was generated by incorporating a best-fit model of evolution into PAUP 46, and the resulting tree was then bootstrapped 1000 times to ensure fidelity. Analysis of the tree demonstrated that the sequences from the six mother-infant pairs form distinct, well separated subtrees, and all pairs were separate from the lab control strain HIV-1 isolate NL4-3. Within each subtree the sequences for the mother and infant are generally well separated into subtrees, however some intermingling was observed in pairs B, D, E, and H. The intermingling of mother-infant sequences suggests that the isolates from these patients are very closely related, and had not as of yet evolved to form separate, distinct subtrees. Taken together the data indicates that epidemiologically linked (mother-infant) patient sequences are closer to each other evolutionarily than epidemiologically unlinked sequences. The separation of the mother-infant sequences from each pair and NL4-3 indicates that no PCR contamination occurred.

The multiple sequence alignment of the deduced amino acid sequences of the HIV-1 NC and p6 genes is shown in Figs. 2, 3, 4. Of the 168 sequences analyzed, 156 contained an intact open reading frame (ORF), yielding a frequency of 92.8%. This high frequency indicates that the coding potential of the NC and p6 genes was maintained in most of the sequences analyzed. Looking more closely, the frequency of an intact ORF for the mothers' sequences was 89.4%, while the infants' sequences yielded a frequency of 96.3%. Several clones within mother-infant pair H were found to be defective due to a single nucleotide substitution, insertion or deletion, which resulted in the formation of a stop codon. There were several patient and pair specific sequence patterns within the NC sequences analyzed. An insertion of proline-threonine-valine (PTV) was seen in the sequences of mother-infant pair C at position 78, and an insertion of proline-threonine-alanine-proline-proline-glutamate (PTAPPE) was observed within several sequences of mother D at position 84. This resulted in a duplication of the PTAP motif within this patient. An amino acid substitution was also present in most of the sequences when compared as a whole, a leucine (L) was replaced with a methionine (M), valine (V), histidine (H), arginine (R) or glutamine (Q) at position 116.

The nucleotide and amino acid distances, which measure the degree of genetic variability based on pairwise comparison, were calculated for the six mother-infant pairs' sequences (Table 2). The nucleotide sequences within mothers B, C, D, E, F, and H varied by 0.26, 0.53, 0.84, 1.13, 0.27, and 5.04% (median values) respectively, ranging from 0 to 6.30%. The infant (B, C, D, E, F, I1H, and I2H) sequences differed by 0, 2.59, 0.88, 1.11, 1.78, 0, 3.22% (median values) respectively, ranging from 0 to 5.03%. Moreover, the nucleotide sequence variability between epidemiologically linked mother-infant pairs (pairs B, C, D, E, F, and H) varied by 0, 3.16, 1.13, 1.12, 1.99, and 1.87% (median values) respectively, and ranged from 0 to 6.66%. In addition, the deduced amino acid sequence variability of NC and p6 within mothers (B, C, D, E, F, and H) differed by 0, 0.80, 0.81, 2.47, 0.81, and 4.12% (median values) respectively, ranging from 0 to 13.05%. Furthermore, the infants' (B, C, D, E, F, I1H, and I2H) amino acid sequences varied by 0, 4.05, 1.63, 1.63, 2.45, 0, and 2.04% (median values) respectively, and ranged from 0 to 9.31%. The amino acid sequence variability between epidemiologically linked mother-infant pairs (pairs B, C, D, E, F, and H) varied by 0, 5.74, 1.63, 2.47, 3.28, and 3.28% (median values) respectively, and ranged from 0 to 14.55%. The nucleotide and amino acid sequence variability was also calculated between epidemiologically unlinked individuals. It was determined that the nucleotide distances gave a median value of 7.68, while the amino acid distances produced a median of 14.68. A comparison revealed that the variability between epidemiologically linked mother-infant pairs was lower than the variability between epidemiologically unlinked individuals. This suggests that epidemiologically linked sequences were closer to each other evolutionarily than to unlinked sequences.

We also evaluated if the low variability of NC sequences seen in our mother-infant pair isolates was due to errors made by Platinum Pfx Taq polymerase used in our study. We did not find any errors made by the Taq polymerase when we used a known sequence of HIV-1 NL4-3 for PCR amplification and DNA sequencing of the NC gene.

Different models of evolution were suggested by Modeltest 3.06 47 based on maximum likelihood estimates and chi square tests that were performed by the program. The estimates of genetic diversity of the NC and p6 sequences obtained were determined using the Watterson model, which assumes segregated sites, and the Coalesce model, which assumes a constant population size. These estimates of genetic diversity are displayed as theta values, and represent the rate of mutation per site per generation (Table 3). The Watterson model estimated the level of genetic diversity within infected mothers to be 0.014, and within infected infants to be 0.015. Slightly greater estimates were obtained using the Coalesce method, with the genetic diversity between mothers being 0.014, and between infants 0.029. Together these data suggest that both the mother and infant populations evolved slowly and at similar rates. The difference between the estimates of genetic diversity between the mother and infant sequences, using either method, is not statistically significant.

The ratio of the accumulation of non-synonymous (dn) to synonymous substitutions (ds) was used to estimate the selection pressure on the NC and p6 gene by using a model modified by Nielson and Yang 48, which was then implemented by codeML 49. The advantage of the codeML method lies in the fact that this model views the codon as the unit of evolution, as opposed to the nucleotide which is used in other models 50. Moreover, the Nielson and Yang model does not assume that all sites within a sequence are under the same selection pressure. This gives a more realistic view of evolution because mutations, in some cases leading to only a single amino acid change, can be more advantageous or deleterious in some regions of a protein compared to others, and thus undergoes positive or purifying selection. In addition the dn/ds ratio that is calculated determines the selection pressure acting upon the changes within the codon, with a dn/ds ratio of greater than 1 indicating that positive selection pressure is present. Not only does this model determine positive selection pressure, it also calculates the percentage of mutations that are selected. The percent of mutations that are conserved fall in the p1 category, the neutral mutations are in the p2 category, and the positively selected mutations are in the p3 category. The estimations of the dn/ds ratio as well as the percentages in each category (p1, p2, and p3) for each patient sample are given in Table 4. All of the sequence populations analyzed displayed a dn/ds ratio greater than or equal to 1.

In general, the mother sequences displayed a higher percentage of positively selected p3 sites compared to the infants. Within mothers, almost 100% of the mutations in mothers B, C, and F were positively selected. Although mother D and mother H have the highest dn/ds values, less than 1% of the mutations are positively selected. Most of the mutations in mother D and mother H are neutral. When compared to the mothers, infants have less than 3% of mutations that are positively selected, with the exceptions of infant D and the second infant H twin (I2H). In contrast to the mothers, the infants have a more even distribution of conserved and neutral mutations. It is interesting to note that in four of the seven infants, over 50% of the mutations observed were neutral mutations. This higher proportion of p2 sites in infants was also seen in analysis of the nef and reverse transcriptase (RT) genes 1251. The positive selection pressure acting on these patient sequences was estimated in codeML using both neutral models and positive selection models. In patients where a substantial proportion of mutations were in the p3 category, the positive selection model was significant over the neutral model (data not shown). These data indicate that a higher percentage of mutations are positively selected in mothers as compared to infants, however positive selection pressure was observed when analyzing the NC gene sequences from both the mother and infant patient samples.

The function of the HIV-1 NC protein is to bind to viral RNA and DNA. This protein contains two zinc fingers and many basic amino acids that allow it to interact with the viral nucleic acids. The critical residues of the zinc fingers consist of three cysteines and one histidine, and have the sequence C-X₂-C-X₄-H-X₄-C, with X representing any amino acid, and are located at positions 16 to 29 and 37 to 50 within the NC protein 20. The critical residues within these zinc fingers are located at positions 16, 19, 24, and 29 in the first zinc finger and positions 37, 40, 45, and 50 in the second zinc finger. A mutation at any of these critical residues abolishes the ability of these functional domains to bind the zinc cofactor, which will lead to improper folding of the protein 2429. Analysis of the first zinc finger sequence from the six mother-infant pairs shows that of the 168 sequences acquired, only two contained mutations at the critical residues (Figs. 2, 3, 4). Infant C clone 2 (IC-2) contained the substitution C19R, and mother B clone 2 (MB-2) (Fig. 2) contained the substitution H24Y. Furthermore, the second zinc finger contained substitutions at the critical residues in only one clone; infant C clone 3 (IC-3) contained an H45Y substitution (Fig. 2). However some sequences within mother H and the second infant H twin (I2H) contain substitutions that resulted in the formation of a stop codon at position 38 within the second zinc finger (Fig. 4). These stop codons would result in a truncation in the second zinc finger, and would result in only one functional zinc finger (the first zinc finger) within the NC protein of these clones. When two zinc fingers are present, the first generally tends to play a more critical role 1820, however removal of the second zinc finger function has been shown to greatly decrease the annealing capacity of the NC protein 2029. Despite these exceptions, the critical residues of both zinc fingers within the mother-infant NC sequences were highly conserved.

There are several basic residues, arginine (R), lysine (K), or histidine (H), within the NC protein that also allow it to function. Of the 56 amino acids that make up the NC protein, 17 are basic 21. These basic residues spread throughout the protein and are responsible for interacting with the side chains on viral nucleic acids 1852. Mutations in these basic residues has been shown to reduce RNA binding and encapsidation 21. Analysis of the sequences from the mother-infant pairs shows that there are substitutions at many of the basic residues. However looking more in depth, a majority of the substitutions are from one basic amino acid to another. Furthermore, there are several substitutions from non-basic to basic residues throughout the protein sequences obtained, and some of these substitutions are compensatory mutations for changes from a basic amino acid elsewhere within the sequence (Figs. 2, 3, 4). While there are several substitutions involving basic amino acids within the NC protein sequences from the six mother-infant pairs, the presence of several basic residues throughout the protein sequences is highly conserved.

The p6 gene was also sequenced as a result of sequencing the NC gene. The p6 protein contains two major functional domains, the viral late domain located at positions 79 to 83, and the Vpr binding domains located at positions 87 to 90 and 107 to 118 303345. The late domain contains the sequence proline-threonine-alanine-proline-proline (PTAPP) and is responsible for ensuring proper budding of a newly formed virion from the host cell membrane 3253. The prolines at positions 82 and 83 have especially been shown to be critical for Tsg101 binding 32. Analysis of the p6 protein sequences from the six mother-infant pairs revealed that the late domains, especially the critical prolines, are conserved in most of the sequences obtained (Figs. 2, 3, 4). Interestingly, in several sequences from mother D there is a duplication of the late domain (Fig. 3). It has been shown that duplication of this domain could be linked to antiretroviral drug resistance 5455. However since mother D has not been exposed to antiretroviral drugs (Table 1), this duplication must have arisen naturally or was present in the virus that was initially transmitted to mother D. In general, the late domain of the p6 protein from the mother-infant pairs was highly conserved.

The Vpr binding domain could be located in two possible positions within the p6 protein sequences of the six mother-infant pairs, either positions 87 to 90 or 107 to 118 303345 (Fig. 2). The domain located at positions 87 to 107 has the sequence phenylalanine-arginine-phenylalanine-glycine (FRFG) 30, while the domain at positions 107 to 118 has the sequence leucine-XX-leucine-XX-leucine-XX-leucine-XX ((LXX)₄)45, with X representing any amino acid. These Vpr binding domains are responsible for inclusion of the viral accessory protein Vpr into newly forming virions. Analysis of the protein sequences from the mother-infant pairs revealed that while the FRFG Vpr binding domain was mostly conserved, there were some notable exceptions. There were single amino acid substitutions within the domain in every clone of mother and infant F (pair F), infant C (IC), and infant D (ID) (Figs. 2, 3, 4). It has been shown that mutations at either of the two phenylalanines within the FRFG domain, which is seen in pair F and infant D, causes a loss of Vpr packaging within virions; while a substitution at the arginine site, which is seen in infant C, seems to have little to no effect 30. In spite of these exceptions however, the FRFG Vpr binding domain within the six mother-infant pairs analyzed was mostly conserved. Analyzing the protein sequences also showed that the (LXX)₄ domain was also mostly conserved within the sequences obtained, except for the first leucine in every clone. This first leucine was substituted with either a methionine (M), a valine (V), a histidine (H), an arginine (R), or a glutamine (Q) (Figs. 2, 3, 4). A change in this first leucine has been shown to decrease Vpr binding 45. The third and fourth leucine have been shown to be critical for Vpr inclusion 3334, and these residues are highly conserved within the mother-infant sequences obtained. As with the FRFG domain, the (LXX)₄ Vpr binding was mostly conserved within the sequences of the mother-infant pairs analyzed.

The p6 gene product also contains a region, from amino acid positions 31–46 with the sequence DKELYPLASLRSLFG that is responsible for interacting with the host cell factor AIP1 31. This motif within the mother-infant pair sequences was mostly conserved, however every clone analyzed contained a substitution at the first leucine, as also seen in the (LXX)₄ domain (Figs 2, 3, 4). Mother and infant C (pair C) (Fig. 2) and mother and infant D (pair D) (Fig. 3) also contained additional substitutions within the AIP1 binding domain. It is not known at this time what effect these substitutions would have on the interaction of p6 with AIP1. Despite these exceptions, the AIP1 binding domain was mostly conserved within the six mother-infant pairs' sequences obtained.

In addition to the substitutions mentioned, there were several other substitutions that occurred outside of the functional domains. The effect that these changes would have is not known at this time.

The cytotoxic T-lymphocyte (CTL) response is known to contribute a significant portion of the body's immune response to an HIV-1 infection. In patients with a strong CTL response, there has been shown to be a decrease in the viral load within the patient and a long-term non-progressor disease status 56. It has also been shown that mothers who transmit the virus to their infants have an increased number of CTL escape variants when compared to non-transmitting mothers 57. This could demonstrate a correlation between the amount of CTL escape variants circulating in the mothers' bloodstream and the likelihood of vertical transmission of the virus. There have been several CTL epitopes identified within the NC and p6 proteins. The first epitope within the NC protein has the sequence CRAPRKKGC and is located between amino acid positions 28 and 36 58. This epitope is recognized by HLA-B14 and contains the last cystine of the first zinc finger, and the first cystine of the second zinc finger. Analysis of the NC amino acid sequences from the six mother-infant pairs revealed that this epitope was highly conserved in most of the clones that were obtained (Figs. 2, 3, 4). Another CTL epitope, KEGHQMKDCTERQANF, is located at amino acid positions 42–57 and is recognized by several HLA types 58. This epitope spans the last 14 amino acids of the NC protein and contains the histidine and final cystine of the second zinc finger. Again this epitope was mostly conserved when the sequences from the mother-infant pairs was analyzed. The next motif, CTERQANFL, is located from positions 50 to 56 and is recognized by HLA-B61 58. This epitope contains the last cystine of the second zinc finger and was highly conserved within the mother-infant sequences obtained.

The first motif within the p6 gene sequences, GNFLQSRPEPTAPPF, is located at amino acid positions 70–84 and is recognized by several HLA types 58. Analysis of the mother-infant sequences revealed that this epitope was mostly conserved, with the exception of mother and infant C (pair C). Pair C contains a PTV insertion beginning at position 78 (Fig. 2). It is not known at this time what effect on CTL recognition this insertion would have. The next epitope is located at amino acid positions 105 to 114 and has the sequence KELYPLTSL 58. This epitope is recognized by HLA-B60 and is positioned within the (LXX)₄ Vpr binding domain. Within the mother-infant sequences obtained this CTL epitope was mostly conserved, however the first lysine within the epitope was substituted within every clone analyzed (Figs. 2, 3, 4). It is not known at this time what effect this substitution would have on recognition of this epitope. Another epitope, YPLTSLRSLF, is located at positions 108 to 117 and is recognized by HLA-B7 58. Analysis of the six mother-infant pairs' sequences revealed that this epitope was mostly conserved (Figs. 2, 3, 4). Overall, analysis of the CTL recognition epitopes within the sequences of the mother-infant pairs' displayed that these epitopes which are involved in immune recognition of the virus were mostly conserved.

Blood samples were collected from six mother-infant pairs following vertical transmission, including a set of twins (I1H and I2H) in the case of mother H. The demographics, clinical and laboratory findings on these mother-infant pairs is summarized in Table 1. The Human Subjects Committee of the University of Arizona (Tucson, AZ) and the Institutional Review Board of the Children's Hospital Medical Center (Cincinnati, Ohio) approved this study. Written informed consent to participate in this study was obtained from the mothers of the mother-infant pairs.

Peripheral blood mononuclear cells (PBMCs) were isolated using a single-step Ficoll-Hypaque method (Pharmacia-LKB) from whole blood samples of HIV-1 infected mother-and-infant pairs that were involved in vertical transmission. The PBMCs DNA, which contains the integrated HIV-1 genome, was isolated as previously described 9. The HIV-1 NC gene was amplified using a two-step polymerase chain reaction (PCR) method according to the modified protocol described by Ahmad et al 9. Outer primers NC-1 (5'GAAGAAATGATGACAGCATGTCAGGGAGTGGG, 1819 to 1851, sense) and NC-2 (5'CCATCTTCCTGGCAAATTCATTTCTTCTAATACT, 2344 to 2378, antisense) were first used, followed by nested primers NC-3 (5'CACCGGCCATAAAGCAAGAGTTTTGGCTGAAGC, 1854 to 1887, sense) and NC-4 (5' CATCTGCTCCTGTATCTAATAGAGCTTCCTT, 2310 to 2341, antisense). Equal amounts of PBMC DNA (approximately 25 to 50 copies, minimum) as determined by end-point dilution was subjected to multiple (four to six) PCRs to obtain clones that were sequenced and analyzed. PCR was carried out in a 25μl reaction mixture containing 2.5 μl 10× Pfx Amplification buffer, 2.5 mM MgSO₄, 400 μM of each dATP, dCTP, dGTP, and dTTP, 0.2 μM of each outer primer, and 2.5 units (U) of Platinum Pfx DNA polymerase (Invitrogen Inc.). The reaction was initiated at 94°C for 2 minutes (min), and then cycled at 94°C for 30 seconds (sec), 50°C for 30 sec, and 72°C for 2 min, for 35 cycles, with an addition extension period of 10 min at 72°C to end the reaction. Following the first round of PCR, 5–8 μl of the first-PCR product was used for nested PCR using the same reagents and the inner primers. This nested PCR was cycled at 94°C for 30 sec, 55°C for 30 sec, and 72°C for 2 min, for 35 cycles. A negative control was included with each PCR which used sterile water in place of DNA. PCR was also performed on HIV-1 NL4-3, of which the sequence is known (GenBank accession number M19921), to assess any errors made by the Platinum Pfx DNA polymerase. To avoid contamination, all samples, reagents and PCR products were kept separately and dispensed in a separate room free of all laboratory-used DNA.

The PCR products were visualized on a 1% agarose gel and cloned into the TOPO TA cloning system (pCR 2.1-TOPO vector) as per manufacturer's instructions (Invitrogen Inc.). Positive bacterial colonies were determined by blue-white screening, and the presence of correct-sized inserts was confirmed by restriction digest. Eight to eighteen clones from each mother and infant, from multiple independent PCRs, were sequenced using the Thermosequenase Cycle Sequencing protocol (USB) and the University of Arizona Biotechnology Center automated system (ABI Prism^® 3700 DNA automated sequencing system).

The nucleotide sequences of the HIV-1 NC gene (approximately 375 bp) from six mother-infant pairs were analyzed using the Wisconsin package version 10.1 of the Genetics Computer Group (GCG) and were translated to corresponding deduced amino acid sequences (125 amino acids). Both the nucleotide and amino acid sequences were aligned, using the HIV-1 NL4-3 NC sequence as a reference, by Clustal X. A model of evolution was optimized for the entire nucleotide sequence data set using the Huelsenbeck and Crandall approach 66. Likelihood models of evolution were calculated using PAUP 46 and a chi square (χ²) test was performed using Modeltest 3.06 47. The model with the highest likelihood was incorporated into PAUP to generate a neighbor-joining tree, which was bootstrapped 1000 times to ensure fidelity. The tree was based on the nucleotide sequences from the six mother-infant pairs, as well as HIV-1 NL4-3 which was used as a reference sequence and the outgroup for the tree. Using Modeltest and the Akaike Information Criterion (AIC) 67, all the null hypotheses were rejected except the best-fit model (GTR+G). The base frequencies for the NC gene were calculated as: freq A = 0.37, freq C = 0.23, freq G = 0.21 and freq T = 0.19. Five rate categories were also figured, and were as follows: R(A- C)= 1.00, R (A-G)= 5.34, R(A-T) = 0.79, R(C-G) = 0.79, R(C-T) = 5.34, R(G-T) = 1.00. The rate heterogeneity was taken into account using a gamma distribution with a shape parameter (α) of the distribution estimated from the data via maximum likelihood. This shape parameter had a value of 0.7085. A model of evolution for each patient was also generated and optimized to estimate corrected pairwise nucleotide distances using PAUP 46. Amino acid distances were also estimated using the Jukes-Cantor model within the Wisconsin package version 10.1 of GCG. The minimum, median, and maximum distance values were calculated for both nucleotides and amino acids for each patient as well as for the linked and unlinked patient pairs. The dynamics of HIV-1 evolution was assessed using techniques of population genetics. In population genetics, genetic diversity is defined as θ = 2N_eμ, where N_e is the effective population size and μ is the per nucleotide mutation rate per generation. The Watterson model, which assumes segregated sites, and the Metropolis-Hastings model, which assumes a constant population size, was used to estimate the differences in genetic diversity using the Coalesce 1.5 program 6869. To analyze the evolutionary processes acting upon the NC gene, we estimated the ratio of nonsynonymous (dn) to synonymous (ds) substitutions by a maximum likelihood model using codeML, which is part of the PAML package 49. The Nielsen and Yang 48 model considers the codon instead of the nucleotide as the unit of evolution and thus incorporates three distinct categories of sites. The first category represents the sites that are invariable or conserved (p1, dn/ds = 0); the second category represents sites that are neutral (p2, dn/ds = l), at which dn and ds are fixed at the same rate; and the third category represents sites that are under positive selection, where dn has a higher fixation rate than ds (p3, dn/ds >1). The dn/ds was estimated for each patient using both neutral and positive selection models in codeML.

The nucleotide accession numbers of the sequences submitted to GenBank are DQ026833 through DQ027000.