Secondary Logo

Journal Logo

Poor Performance of Bioinformatics Programs for Genotypic Prediction of Coreceptor Usage of HIV-1 Group O Isolates

Rupp, Daniel; Geuenich, Silvia PhD; Keppler, Oliver T MD

JAIDS Journal of Acquired Immune Deficiency Syndromes: March 1st, 2010 - Volume 53 - Issue 3 - p 412-413
doi: 10.1097/QAI.0b013e3181c9f53f
Letter to the Editor
Free
SDC

Department of Infectious Diseases, Virology, University of Heidelberg, Germany

Daniel Rupp and Silvia Geuenich contributed equally to this work.

To the Editors:

Coreceptor-specific pharmacotherapy requires reliable determinations of coreceptor usage of patients' HIV-1 isolates. We examined the performance of 4 major coreceptor prediction programs (geno2pheno[coreceptor], WebPSSM, SVM-SK, WetCat), which primarily utilize the envelope-V3 loop sequence, for HIV-1 O isolates. The specificity of the sequence-based prediction was low, ranging from 0% to 28.6% relative to the biologic coreceptor phenotype. Consequently, coreceptor usage in HIV-1 O-infected individuals should at present be exclusively assessed phenotypically until prediction programs have been educated and optimized for this divergent HIV-1 group.

The reliable determination of HIV-1 coreceptor usage has recently become an issue of clinical importance because the repertoire of licensed antiretroviral drugs has been expanded by the CCR5 coreceptor antagonist maraviroc (celsentri). This new class of entry inhibitors has shown low-nanomolar activity in vitro against most R5 HIV-1 strains and primary isolates, including several HIV-1 group O viruses.1 Maraviroc has demonstrated high efficacy in HIV-1-infected individuals harboring R5 viruses.2 Coreceptor tropism testing is mandatory before prescribing coreceptor-specific drugs and useful for therapy monitoring. Thus far, treatment failure was mostly associated with lack of X4 virus detection before initiation of therapy and subsequent in vivo expansion of preexisting X4 viruses that were insensitive to the CCR5 inhibitor.3 Identification of such low-frequency X4 variants from HIV-1 group M in clinical samples can apparently be improved by “ultradeep” sequencing approaches.4

The most commonly used genotypic prediction programs utilize clonal sequence data of the third hypervariable (V3) loop of the HIV-1 envelope glycoprotein, are built around statistical learning algorithms, and are educated by data training sets. These bioinformatics prediction engines include simple amino acid charge rules, decision trees, neural networks, more advanced support vector machines (SVM), position-specific scoring matrices, and SVM-based approaches that integrate viral sequences with structural information of the envelope glycoprotein and clinical data.5 The specificity of coreceptor prediction for clinical HIV-1 M isolates by these V3 loop-based bioinformatics programs currently ranges between 83% and 97%5-8 compared with phenotypic, tissue culture-based assays which determine the biologic coreceptor tropism.9 Although such phenotypic assays are still commonly used for clinical samples, their reliability has recently been questioned based on interassay concordance analyses10 and genotypic predictions are rapidly improving in fidelity, efficiency, and cost.5

In contrast to the pandemic HIV-1 group M, infections with the highly divergent HIV-1 group O are found primarily in the Republic of Cameroon, Equatorial Guinea, and Gabon with a current seroprevalence of 1%-2%.11,12 Outside of Africa, infections are very rare and have been reported in Belgium, France, Germany, Spain, Norway, and the United States.13 Contrasting their relative contribution to the pandemic, disease progression in patients infected with HIV-1 O and M seems to be comparable.14,15

In the current study, we tested 4 web-based coreceptor prediction programs for their ability to determine the coreceptor usage of HIV-1 O primary isolates based on the amino acid sequence of the V3 loop. Viral sequences were obtained from DNA isolated from infected cells and aligned relative to the prototypic group M (clade B) strain NL4-3 (Fig. 1A). The biologic phenotype (Fig. 1B), which describes the utilization of the major coreceptors CCR5 and/or CXCR4, was recently determined for this panel of HIV-1 O isolates on TZM-bl reporter cells.16

FIGURE 1

FIGURE 1

Geno2pheno[coreceptor] identified 4 out of 7 V3 sequences (Fig. 1A) as belonging to HIV-1 O and accurately predicted the coreceptor usage of 2 isolates (28.6% specificity; Fig. 1B). In contrast, the program correctly predicted the coreceptor tropism of NL4-3, Ba-L (Fig. 1A), and several other primary M isolates (not shown), in line with recently published specificity rates.5 All 3 WebPSSM matrices performed well for the 2 clade B prototypes, but predicted all 7 O-type sequences to be X4-tropic, which was correct only for one (MVP-2901-94; 14.2% specificity) (Fig. 1B). Results from SVM-SK analyses predicted the inconsistent usage of CCR5 and CXCR4, but not dual-tropism, for all O-types (0% specificity, Fig. 1B). The fourth prediction program, WetCat, did not recognize the O-type sequences as V3 loop derived and was consequently unable to predict any coreceptor usage (Fig. 1B).

In conclusion, 4 major web-based in silico prediction programs for coreceptor usage based on bioinformatic V3 loop analyses performed below chance level for HIV-1 O viruses. In light of the increasing importance of reliable coreceptor predictions in guiding treatment selection and prognosis in HIV/AIDS patients, coreceptor usage in the small number of individuals infected or coinfected with HIV-1 O viruses should at present be exclusively assessed by phenotypic assays and, ideally, sensitivity of these highly divergent isolates to CCR5 antagonists should be predetermined in vitro. Furthermore, all online prediction algorithms should clearly indicate their data training sets, predicted group/clade of the V3 sequences, and confidence of their assignments, and need to be educated and optimized for HIV-1 group O.

Daniel Rupp

Silvia Geuenich, PhD

Oliver T. Keppler, MD

Department of Infectious Diseases, Virology, University of Heidelberg, Germany

Back to Top | Article Outline

REFERENCES

1. Dorr P, Westby M, Dobbs S, et al. Maraviroc (UK-427,857), a potent, orally bioavailable, and selective small-molecule inhibitor of chemokine receptor CCR5 with broad-spectrum anti-human immunodeficiency virus type 1 activity. Antimicrob Agents Chemother. 2005;49:4721-4732.
2. Fatkenheuer G, Nelson M, Lazzarin A, et al. Subgroup analyses of maraviroc in previously treated R5 HIV-1 infection. N Engl J Med. 2008;359:1442-1455.
3. Moore JP, Kuritzkes DR. A piece de resistance: how HIV-1 escapes small molecule CCR5 inhibitors. Curr Opin HIV AIDS. 2009;4:118-124.
4. Archer J, Braverman MS, Taillon BE, et al. Detection of low-frequency pretherapy chemokine (CXC motif) receptor 4 (CXCR4)-using HIV-1 with ultra-deep pyrosequencing. Aids. 2009,23:1209-1218.
5. Lengauer T, Sander O, Sierra S, et al. Bioinformatics prediction of HIV coreceptor usage. Nat Biotechnol. 2007,25:1407-1410.
6. Boisvert S, Marchand M, Laviolette F, et al. HIV-1 coreceptor usage prediction without multiple alignments: an application of string kernels. Retrovirology. 2008;5:110.
7. Jensen MA, Coetzer M, van 't Wout AB, et al. A reliable phenotype predictor for human immunodeficiency virus type 1 subtype C based on envelope V3 sequences. J Virol. 2006;80:4698-4704.
8. Low AJ, Dong W, Chan D, Sing T, et al. Current V3 genotyping algorithms are inadequate for predicting X4 co-receptor usage in clinical isolates. Aids. 2007;21:F17-F24.
9. Coakley E, Petropoulos CJ, Whitcomb JM. Assessing chemokine co-receptor usage in HIV. Curr Opin Infect Dis. 2005;18:9-15.
10. Skrabal K, Low AJ, Dong W, et al. Determining human immunodeficiency virus coreceptor use in a clinical setting: degree of correlation between two phenotypic assays and a bioinformatic model. J Clin Microbiol. 2007;45:279-284.
11. Brennan CA, Bodelle P, Coffey R, et al. The prevalence of diverse HIV-1 strains was stable in Cameroonian blood donors from 1996 to 2004. J Acquir Immune Defic Syndr. 2008;49:432-439.
12. Zekeng L, Obiang Sima J, Hampl H, et al. Update on HIV-1 group O infection in Equatorial Guinea, Central Africa. Aids. 1997;11:1410-1412.
13. Janssens W, Heyndrickx L, Van der Auwera G, et al. Interpatient genetic variability of HIV-1 group O. Aids. 1999;13:41-48.
14. Gurtler LG, Hauser PH, Eberle J, et al. A new subtype of human immunodeficiency virus type 1 (MVP-5180) from Cameroon. J Virol. 1994;68:1581-1585.
15. Vergne L, Bourgeois A, Mpoudi-Ngole E, et al. Biological and genetic characteristics of HIV infections in Cameroon reveals dual group M and O infections and a correlation between SI-inducing phenotype of the predominant CRF02_AG variant and disease stage. Virology. 2003;310:254-266.
16. Geuenich S, Kaderali L, Allespach I, et al. Biological signature characteristics of primary isolates from human immunodeficiency virus type 1 group O in ex vivo human tonsil histocultures. J Virol. 2009;83:10494-10503.
© 2010 Lippincott Williams & Wilkins, Inc.