Reports of HIV-1 dual infections in individual patients have been published since the early 1990s , but screening of larger cohorts suggested that the incidence of multiple HIV-1 infections is mostly low [2–5], but not always [6,7]. HIV-1 dual infections are divided into co-infections (occurring before a specific HIV immune response has been established), and superinfections (in which a second virus successfully enters the host after seroconversion) . HIV-1 dual infections are seen as problematic as they can give rise to recombinant viruses with novel properties, have instigated doubts about the efficacy of vaccine protection, and have been correlated with disease progression [1,8]. Whether in a single patient or in a larger cohort, however, it is notoriously difficult to detect HIV-1 dual infections. Not only are the methods, polymerase chain reaction amplification and heteroduplex mobility assay or sequencing, relatively time consuming, but also the sampling timepoints need to be chosen carefully. Both dramatic fluctuations of viral sequences in serum [9–11], or the outgrowth of one virus or a novel recombinant [12–15] have been seen in dual infections, suggesting that double infections can easily be missed when analysing a single sample. We report that the HIV-1 reverse transcriptase (RT) sequence routinely requested by treating physicians to assess viral drug-resistance mutations in patients failing antiretroviral therapy offers an additional way of detecting HIV-1 dual infections. In this sequence, the amount of degenerate base codes reported can be an indication of extensive virus evolution, but can also point to the presence of multiple virus strains. In our analysis of 37 patients with a degenerate base count of 34 or more (range 34–99) in the ViroSeq RT sequence, 16 were found to have an HIV-1 dual infection.
HIV-1 protease/reverse transcriptase (prot/RT) gene sequences, encompassing the complete protease sequence and the first 335 codons of the RT gene, are routinely generated in our hospital with the ViroSeq HIV-1 genotyping kit version 2 (Celera Diagnostics, Alameda, California, USA). Electrophoresis and data collection are performed on an ABI PRISM 3100 genetic analyser (Applied Biosystems, Foster City, California, USA) . A total of 1661 prot/RT gene sequence records were available for our study, obtained from 1319 HIV-infected patients. Most prot/RT sequences were generated because of therapy failure, but since 2003 the ViroSeq prot/RT sequence is part of the standard patient care for every newly HIV-1-infected patient from the Amsterdam region of the Netherlands and is determined at their first hospital visit. Therefore, approximately 400 of the 1661 prot/RT gene records are so-called baseline sequences that are not associated with therapy failure.
A total of 37 sequence records were selected for further analysis based arbitrarily upon a degenerate base count of 34 or more in the RT part of the sequence. Degenerate base codes (single-letter nucleotide codes of the International Union of Biochemistry) taken into account are R (A or G), Y (C or T), K (G or T), M (A or C), S (G or C), W (A or T), B (C, G or T), D (A, G or T), H (A, C or T), and V (A, C or G). H, V or N (any base) were never detected. Degenerate positions were randomly distributed in RT, with no special emphasis on known drug-resistance positions. The selection was found to contain a patient described earlier as having an HIV-1 triple infection . This patient was not sequenced again, but was included in the results.
Analysis of HIV-1 env-V3 and gag sequences
The V3 sequence of the HIV-1 envelope gene was amplified as described  from the selected patient samples. V3 fragments were cloned with the TOPO TA cloning kit (Invitrogen, Carlsbad, California, USA), and 16 clones per patient were selected and sequenced. Sequences were aligned with reference HIV-1 V3 sequences (from the Los Alamos National Laboratory at http://hiv-web.lanl.gov) using ClustalW available in BioEdit Sequence Alignment Editor version 7.0.1 (www.mbio.ncsu.edu/BioEdit/bioedit.html). Neighbour-joining (NJ) trees based upon Kimura two-parameter distances were constructed with the MEGA software package (www.megasoftware.net), and 1000 bootstrap replicates were analysed. Additional phylogenetic analyses were performed with the parallel version of MrBayes 3.1 (http://mrbayes.net), run at the SARA High Performance Computing facilities (www.sara.nl). SARA modified the program so that it uses the sprng library (http://sprng.cs.fsu.edu/) to generate independent streams of random numbers in the parallel processes.
For 13 patients, further analyses were performed using an amplified and cloned 720 nucleotide HIV-1 gag gene fragment , encompassing most of p17 and the first part of p24, in a similar approach as described above to confirm the presence or absence of divergent sequence clusters.
Definition of dual infection
The detection of HIV-1 dual infections was based upon phylogenetic analyses of both nucleotide and amino acid sequences of env-V3 or gag fragments with at least 15 cloned sequences per patient. Positive identification was based upon the following criteria that should be true with both the NJ method and with Bayesian inference of phylogeny: sequences of a single patient cluster independently, or sequences of a patient cluster together, but the bootstrap/posterior probability value connecting the branches is low (values under 80/0.8 are here considered insignificant). In many patients divergent sets of sequences were found that nevertheless clustered together with high confidence levels, and this was always attributed to viral evolution and not dual infection. A representative NJ tree of HIV-1 subtype B env-V3 sequences of five patients is shown in Figure 1.
The V3 fragment of the HIV-1 envelope gene was amplified, cloned and sequenced for 36 patients with a degenerate base count of 34 or more in the ViroSeq HIV-1 RT sequence. For another patient, V3 sequences were already available . The results are summarized in Table 1. Sixteen patients were found to have an HIV-1 dual infection based upon phylogenetic analysis of the cloned V3 and gag sequences (43.2% of patient records analysed, equal to 1% of the total 1661 genotyping records). Both nucleotide and translated amino acid sequences were analysed. In 12 cases, the patient's plasma contained sequences that formed at least two distinct clusters, separated by reference sequences or by sequences from other patients. In the four other cases, sequences clustered together but with low, insignificant bootstrap values (range 35–73) in both trees. Eleven of the dual infections (68.8%) occurred with B subtypes, three dual infections (18.8%) were with non-B subtypes (AG/AG, D/D, and C/C, respectively). A single mixed infection with subtype B and subtype CRF01_AE was found, apart from the triple infected patient described earlier , who was infected with CRF01_AE and two subtype B viruses. The non-B/non-AE dual infections were all in patients of African descent. Of the subtype B dual infections, two patients were of unknown origin, eight were from the Netherlands, and one was from Israel. Both patients with B/AE dual infections were Dutch.
Our study shows that a high degenerate base count in the ViroSeq HIV-1 RT sequence is an excellent indicator of an HIV-1 dual infection. We have analysed V3 and gag sequences of 37 patients with a degenerate base count in RT of 34 and more, and found support for a dual infection in 16 of them (43.2%). If we limit the analysis to those with a degenerate base count of 45 and over, 73.3% of the patients show evidence of an HIV-1 dual infection (11/15). In total, however, HIV-1 dual infections are rare: of the 1661 ViroSeq RT sequences available, only 37 had a degenerate base count of 34 or more (2.2%), of which 16 had an HIV-1 dual infection (1%).
In most cases, the treating physician ordered the genotyping because of suspected drug resistance. Approximately a quarter of the sequences available were so-called baseline sequences, however, determined at the first hospital visit of a newly recognized HIV-1 infected patient. It can be seen from Table 1 that most dual infections were found in baseline sequences (75% of the dual infections or 3% of the total available baseline sequences), and less in sequences generated because of suspected drug resistance (25% of the dual infections or 0.3% of total genotyping sequences ordered because of therapy failure).
The highest RT degenerate base counts were found to correspond to dual infections with different HIV-1 subtypes (Table 1). Genetic diversity is expected to be larger there than in double infections with a more similar virus, such as in within-subtype infections, or in re-infections by the same partner. Re-infections are estimated to have the lowest genetic diversity and could represent the detection limit of dual infections by phylogenetic analysis. Possibly, re-infections are represented in our data set as showing low confidence levels in the phylogenetic analysis. In many cases, such infections will be missed because the newly infecting strain cannot reliably be told apart from the earlier strain.
Whether the novel dual infections reported here resulted from HIV-1 co-infection or from HIV-1 superinfection cannot be estimated from the data. In at least three patients, however, one cluster of viral sequences showed significantly less sequence variation than the other cluster, suggestive of a recent superinfection.
This study shows that genotyping has an additional use in detecting HIV dual infections, especially if baseline sequence determination is part of the routine patient care. It is important for physicians to consider the possibility of a patient having a second HIV-1 infection, as this can result in a (temporary) increase in the viral load or in clinical symptoms, and is generally associated with disease progression. Vice versa it is important to consider the likelihood of an HIV-1 superinfection in an HIV-1-infected patient presenting with an increase in viral load or clinical illness, and routine genotyping could then help in the diagnosis.
The authors would like to thank Dr Willem Vermin (SARA, Amsterdam, the Netherlands) for modifying the MrBayes program.
Conflicts of interest: None.
1. Fultz PN. HIV-1 superinfections: omens for vaccine efficacy? AIDS 2004; 18:115–119.
2. Gonzales MJ, Delwart E, Rhee SY, Tsui R, Zolopa AR, Taylor J, et al
. Lack of detectable human immunodeficiency virus type 1 superinfection during 1072 person-years of observation. J Infect Dis 2003; 188:397–405.
3. Tsui R, Herring BL, Barbour JD, Grant RM, Bacchetti P, Kral A, et al
. Human immunodeficiency virus type 1 superinfection was not detected following 215 years of injection drug user exposure. J Virol 2004; 78:94–103.
4. Chakraborty B, Valer L, De Mendoza C, Soriano V, Quinones-Mateu ME. Failure to detect human immunodeficiency virus type 1 superinfection in 28 HIV-seroconcordant individuals with high risk of reexposure to the virus. AIDS Res Hum Retroviruses 2004; 20:1026–1031.
5. Diaz RS, Pardini R, Catroxo M, Operskalski EA, Mosley JW, Busch MP. HIV-1 superinfection is not a common event. J Clin Virol 2005; 33:328–330.
6. Grobler J, Gray CM, Rademeyer C, Seoighe C, Ramjee G, Karim SA, et al
. Incidence of HIV-1 dual infection and its association with increased viral load set point in a cohort of HIV-1 subtype C-infected female sex workers. J Infect Dis 2004; 190:1355–1359.
7. Chohan B, Lavreys L, Rainwater SM, Overbaugh J. Evidence for frequent reinfection with human immunodeficiency virus type 1 of a different subtype. J Virol 2005; 79:10701–10708.
8. Smith DM, Richman DD, Little SJ. HIV Superinfection. J Infect Dis 2005; 192:438–444.
9. Gerhardt M, Mloka D, Tovanabutra S, Sanders-Buell E, Hoffmann O, Maboko L, et al
. In-depth, longitudinal analysis of viral quasispecies from an individual triply infected with late-stage human immunodeficiency virus type 1, using a multiple PCR primer approach. J Virol 2005; 79:8249–8261.
10. McCutchan FE, Hoelscher M, Tovanabutra S, Piyasirisilp S, Sanders-Buell E, Ramos G, et al
. In-depth analysis of a heterosexually acquired human immunodeficiency virus type 1 superinfection: evolution, temporal fluctuation, and intercompartment dynamics from the seronegative window period through 30 months postinfection. J Virol 2005; 79:11693–11704.
11. Bernardin F, Herring BL, Peddada L, Delwart EL. Primary infection of a male plasma donor with divergent HIV variants from the same source followed by rapid fluctuations in their relative frequency and viral recombination. AIDS Res Hum Retroviruses 2003; 19:1009–1015.
12. Fang G, Weiser B, Kuiken C, Philpott SM, Rowland-Jones S, Plummer F, et al
. Recombination following superinfection by HIV-1. AIDS 2004; 18:153–159.
13. Peeters M, Liegeois F, Torimiro N, Bourgeois A, Mpoudi E, Vergne L, et al
. Characterization of a highly replicative intergroup M/O human immunodeficiency virus type 1 recombinant isolated from a Cameroonian patient. J Virol 1999; 73:7368–7375.
14. Van der Kuyl AC, Kozaczynska K, Van den Burg R, Zorgdrager F, Back N, Jurriaans S, et al
. Triple HIV-1 infection. N Engl J Med 2005; 352:2557–2559.
15. Yang OO, Daar ES, Jamieson BD, Balamurugan A, Smith DM, Pitt JA, et al
. Human immunodeficiency virus type 1 clade B superinfection: evidence for differential immune containment of distinct clade B strains. J Virol 2005; 79:860–868.
16. Eshleman SH, Crutcher G, Petrauskene O, Kunstman K, Cunningham SP, Trevino C, et al
. Sensitivity and specificity of the ViroSeq human immunodeficiency virus type 1 (HIV-1) genotyping system for detection of HIV-1 drug resistance mutations by use of an ABI PRISM 3100 genetic analyzer. J Clin Microbiol 2005; 43:813–817.
17. Cornelissen M, Mulder-Kampinga G, Veenstra J, Zorgdrager F, Kuiken C, Hartman S, et al
. Syncytium-inducing (SI) phenotype suppression at seroconversion after intramuscular inoculation of a non-syncytium-inducing/SI phenotypically mixed human immunodeficiency virus population. J Virol 1995; 69:1810–1818.
18. Cornelissen M, Kampinga G, Zorgdrager F, Goudsmit J. Human immunodeficiency virus type 1 subtypes defined by env show high frequency of recombinant gag genes. The UNAIDS Network for HIV Isolation and Characterization. J Virol 1996; 70:8209–8212.