Blood samples were collected from 25 HIV-positive individuals attending Italian sentinel Centers in Bologna, Parma and Naples. For all of them, HIV infection was diagnosed during the 1995 to 2005 period, and most participants were infected through sexual contact (18 of 25, 72%). HIV-1 infection was diagnosed by immunologic methods (ELISA, Western blot), and the viral load was evaluated by viral RNA quantification. The full designation of samples, according to WHO-proposed nomenclature, is CI05.00XE or CI05.00XG, where CI stands for the city of enrolment, 05 stands for the year of study, 00 for the enrolment number and E (or G) stands for env (or gag). For the sake of simplicity, however, in this paper the samples have been indicated with CI05.01 (i.e. BO05.01).

Peripheral blood mononuclear cells (PBMCs) were purified from fresh HIV-1-positive blood samples by Leucoprep density gradient centrifugation, and cellular lysates (approximately 6 × 10⁶ cells) were prepared by Proteinase K digestion at 56°C.

The quality of target DNA was verified by PCR amplification of the housekeeping p53 cellular gene. The amplification of HIV-1 V3-V5 region of the env gene and p17 region of the gag gene were performed by nested PCR analysis, using 1.5 × 10⁵ cells (corresponding to approximately 1 μg of genomic DNA) as a template. The V3-V5 region of the HIV-1 env gene (666 bp) was amplified, as previously described, using the primer pairs ED5-ED12 and ES7-ES8 for the first and the second round of amplification, respectively 3132. The p17 region (474 bp) was amplified, as previously described, using the primer pairs CL1028-CL1033 and CL1029-CL1032 for the first and the second round of amplification, respectively 1433.

Direct sequencing reactions were performed on PCR products purified with a rapid method developed in our laboratory, following the Sequenase protocol (United States Biochemical, Cleveland OH), modified in the labelling step (3 minutes on ice) 3134. The internal annealing oligonucleotides, V3B and GAG B SENSE (annealing to a 19 bp fragment of the C2-V3 region and a 17 bp of the p17 region, respectively), were used to prime sense sequence reactions. Sequences were then analyzed on 6% polyacrylamide wedge sequencing gel.

The env and gag sequences obtained were aligned. Multiple sequence alignments were performed with the MegAlign application of the Lasergene software (DNASTAR Inc., Madison, WI) using the Clustal method. Phylogenetic trees were generated by using the neighbor-joining method with the PHYLIP software package (version 3.52c; Joseph Felsenstein, University of Washington). Briefly, the SEQBOOT program was carried out to generate 100 data sets that represent randomly re-sampled versions of the input-aligned sequences, to test the reliability of the final tree topology. Evolutionary distances were estimated by the DNADIST program, using either the Kimura 2-parameter method or the maximum likelihood distance method, and the phylogenetic relationships were determined by the NEIGHBOR program. A consensus tree was constructed using the CONSENSE program with the majority rule criterion and was drawn with the NJPLOT application.

Overall, 25 HIV-1-seropositive subjects were enrolled in the study. Of them, 32% (8 of 25) were heterosexuals, 36% (9 of 25) were male homosexuals, 4% (1 of 25) was a bisexual and 28% (7 of 25) were IDUs. All but 6 subjects were native Italian, with stable or occasional sexual partners of different nationalities. In particular, the enrolled subjects included an Ethiopian (PR06.03), an Albanian (NA05.04), an Ivorian (PR06.05), a Venezuelan (PR06.06) and two Nigerians (NA05.03, PR06.07). The CD4+ counts ranged from 169 to 1387 cells/mm³ (median, 664 cells/mm³) (Table 1).

The C2-V5 region of the HIV-1 env gene and the p17 region of the gag gene were amplified by nested PCR using primers and conditions previously described 14. Overall, 60% of the samples (15 of 25) were positive at the PCR amplification reactions for both env and gag sub-genomic regions. However, 64% of the samples (16 of 25) were positive at the amplification reaction for env and 68% of the samples (17 of 25) were positive to the amplification reaction for gag. Therefore, one sample (BO07.14) was positive only in env and two samples (NA05.05 e NA05.06) were positive only in gag; the remaining 7 samples were negative to both sub-genomic regions. This result might be the consequence of either mutations in the primers' annealing regions, reducing the melting temperature, or proviral DNA quantity below the threshold of sensibility. An example of the results obtained by nested PCR for env (666 bp) and gag (474 bp) is shown in Figure 1A and 1B.

The nucleotide sequence analysis has been performed on the C2-V3 region of the env gene, directly from the PCR products without a sub-cloning step. The Italian HIV-1 nucleotide sequences, aligned with HIV-1 reference standards of different subtypes, Groups and CRF02_AG, have been aligned with Clustal method, to determine the homology values between the analyzed samples. The results, show an average divergence of 19.66% within the Italian group, with values ranging from 7.1 to 33.9% (data not shown). The Italian sequences, showed average divergences of 17.97% versus standards of B clade; 25.87% versus standards of other HIV genotypes of Group M; 43.14% and 48.4% versus standards of Groups N and O, respectively (Figure 2A). Nevertheless, specific samples show a >30% divergence versus the B clade, strongly suggesting a non-B clade classification. In particular, the sample PR06.07 shows an average divergence of 15% versus the CRF02_AG and 23.2% versus all other clades (Figure 2B); the sample PR06.03 shows an average divergence of 18% versus the C clade and 22.7% versus all other clades (Figure 2C). Therefore, the results suggest a B-clade classification for the vast majority of the analyzed Italian sequences and a C-clade and CRF02_AG classification for the latter samples.

The nucleotide sequence analysis has been performed on the p17 region of the gag gene, directly from the PCR products without a sub-cloning step. As for the env gene, the HIV-1 nucleotide sequences have been aligned with Clustal method to HIV-1 reference standards of different subtypes, Groups and CRF02_AG, in order to determine the homology values between the analyzed samples. The results show an average divergence of 15.86% within the Italian group, with values ranging from 10.81 to 29.16% (data not shown). The Italian sequences, showed average divergences of 16.88% versus standards of B clade; 25.99% versus standards of other HIV genotypes of Group M; 48.54% and 44.15% versus the standards of Groups N and O, respectively (Figure 3A). Nevertheless, specific samples show a >22% divergence versus the B clade, strongly suggesting a non-B clade classification. In particular, the sample PR06.07, also in the gag region, shows an average divergence of 13% versus the CRF02_AG and 24.34% versus all other clades (Figure 3B). Similarly, the sample PR06.03, also in the gag region, shows an average divergence of 22% versus the C clade and 32.08% versus all other clades (Figure 3D). Finally, the sample NA05.05, negative for env, shows an average divergence of 13.9% versus the CRF02_AG and 25.31% versus all other clades (Figure 3C). Therefore, also in the gag region, the results suggest a B-clade classification for the vast majority of the analyzed Italian sequences and confirm a C-clade or CRF02_AG classification for the same samples identified in env.

Nucleotide sequences have been pairwise aligned with HIV-1 reference standards of different subtypes and CRFs, discounting gaps due to nucleotide insertions/deletions. The alignments of HIV-1 env and gag nucleotide sequences have been used to generate phylogenetic trees by the neighbor-joining method. Confidence values for individual branches have been determined by a bootstrap analysis. The reference standards of different subtypes as well as CRFs included in this study cluster in the expected distinct phylogenetic branches in both env and gag sub-genomic regions, indicating that the length of sequence fragments used in this analysis is sufficient for the identification of known subtypes.

All the Italian HIV-1 isolates showing a rate of nucleotide divergence less than 30% in env and 20% in gag versus standards of the B clade, phylogenetically cluster with reference standards of this clade in both env and gag sub-genomic regions, forming several sub-clusters. Similarly, the 3 samples showing a lower nucleotide divergence both in env and gag versus standards of the C clade or the CRF02-AG, cluster with these non-B subtype standards in both sub-genomic regions, confirming the tentative subtype classification based on nucleotide divergence values (Figure 4 and 5).

Moreover, all isolates show a concordant subtype classification in both env and gag sub-genomic regions, suggesting the absence of intra-genomic recombination events.

The V3 and p17 region amino acid sequences have been deduced for each isolate by computer analysis and aligned following their subtype classification. The V3 region of the Italian B-subtype isolates identified in this study shows an amino acid variability mainly localized outside the V3 loop region. The consensus derived from the alignment, in fact, shows that the amino acid residues conserved in 100% of the aligned sequences are all localized in the V3 loop (9 of 35, 25.71%). However, besides the subtype-specific "fingerprint" GPGR sequence at the tip of the V3 loop, less-represented sequences have been identified at the tip (GPGG, GPGS, GPGQ), suggesting a constant diversification in the B clade (Figure 6A).

In regards to the non-B clade isolates identified in this study, and distributed in two different clades (C and CRF02-AG), the too limited number of samples hampers the possibility to derive a consensus sequence and to infer conclusions; nevertheless, it is worthwhile to mention that they all show the "fingerprint" GPGQ tetramer at the tip of their V3 loop sequences (data not shown).

Moreover, mutations in amino acid residues conferring resistance to fusion/binding inhibitors, in association with other mutations along the env sequence not analyzed in the present study, have been observed (i.e. Q296K), suggesting the transmission of isolates resistant to this class of antiretrovirals (Figure 6A) 353637.

The p17 region of the B clade isolates shows a low amino-acidic variability and the consensus derived from the alignment shows an overall rate of amino acid conservation of 44.44% (36 of 81 residues). None of the observed amino acid changes is found in gag residues conferring drug resistance (Figure 6B). Similar results are observed also for non-B clade isolates.