Short Communication: HIV Type 1 Tropism Determination in a Novel Dried Blood Spot Membrane and the Use of a Mixture of Outer Nested Polymerase Chain Reaction Primers

Introduction

Human immunodeficiency virus type 1 (HIV-1) infects CD4-expressing cells using either one or both of the chemokine coreceptors, CCR5 and CXCR4. On the basis of these coreceptors usage, these viruses are classified as R5-tropic and X4-tropic, respectively, and those capable of using both coreceptors are called dual-tropic variants. ¹ The coreceptor usage (CCR5 or CXCR4) defines its cytopathology, growth kinetics, and tropism. X4-tropic strains replicate faster, tend to induce syncytium (multinucleated giant cells) in vitro, and has the ability to infect transformed T cells. CXCR4-using strains are associated with more rapid progression of the disease, ² whereas R5-tropic strains replicate more slowly, do not induce syncytium, and can often infect monocyte-derived macrophages in vitro. With the approval of drugs that block the entry by the CCR5 antagonist, the assessment of viral tropism in patients has become necessary for its clinical use. This work describes the use of dry blood spot (DBS) samples for the genotypic tropism test and the determination of the prevalence of R5 variants using a novel DBS membrane and a new strategy for polymerase chain reaction (PCR) amplification.

Materials and Methods

To determine the prevalence of R5 and non-R5 variants, 608 samples were collected between May 2009 and December 2011 from HIV-1-infected individuals in DBS using a Polyethersulfone (PES) membrane, a novel membrane not previously used for biological material transportation, and samples were transported and stored at room temperature. The low protein binding nature of the PES membrane makes it well suited for biological sample transportation and storage. These membranes present a variety of pores of 10 μm diameter that can adequately hold the biological material including blood obtained after finger punctures. PES was previously treated with 2-cloroacetamide 0.2% in 0.1 M citrate buffer. Previous tests showed that in our hands, the performance of PES was superior than the one found on SS903 collection cards (data on file).

Samples have been collected at physicians' offices all over Brazil, including remote places located as far as 2,000 km, and transported to a Central laboratory, Laboratorio Centro de Genomas, in the city of São Paulo, Brazil. The collection kits included a PES collection card, a silica sachet, a self-retracting needle for finger puncture, a tissue with antiseptic solution, and a prepaid, waterproof envelope to be posted in the regular mail. Collection procedures were as follows: (1) the finger was wiped with the antiseptic tissue, (2) the finger was punctured with a self-retracting needle, (3) the four available spots on the PES collection card were filled with capillary blood, (4) the card was placed into the envelope after 2 h of drying, (5) the silica sachet was placed into the envelope card, and (6) the envelope was sealed and mailed. The study was approved by the local ethical committee of the Federal University of São Paulo (reference number 1510/07).

Nucleic acids have been extracted using the BOOM method as previously described. ³ Reverse transcription polymerase chain reaction (RT-PCR) and nested PCR were used to amplify the V3 region of envelope gp120 and confirmed by electrophoresis on 2% agarose gel followed by sequencing of the V3 region. Primers used in this study have been previously described and were used during the year of 2009. ⁴ During 2010, a new PCR strategy was implemented and the sequence of primers is depicted in Table 1. The new PCR strategy consisted of the use of a cocktail of five related primers to anneal to the 5′ end of the sequence and five related primers to anneal to the 3′ end of the sequence, instead of using a regular pair of primers for the outer PCR reaction. The design of related primers was performed adding one nucleotide at the 5′ end and subtracting one nucleotide from the 3′ end from the original primer (related primer 1) in order to obtain the related primer 2. Related primer 3 was designed using the same strategy as related primer 2, and so on.

It was assumed that a mixture of viral RNA and proviral DNA was used for target amplification. Samples that did not yielded PCR results were recollected using whole blood, and the same method used for DNA purification, RT-PCR and nested PCR, has been applied. Purified PCR products were sequenced bidirectionally with an ABI PRISM BigDye Terminator Cycle Sequencing Ready Reaction Kit and Ampli Taq DNA Polymerase (Applied Biosystems, Foster City, CA). Sequences were corrected and assembled using the Sequencher 4.0 program (Genecodes, Ann Arbor, MI) (GenBank accession numbers pending). V3-loop sequences were interpreted by geno2pheno[coreceptor] with a false-positive rate of 10% for coreceptor prediction and subtype assignment (http://coreceptor.bioinf.mpi-inf.mpg.de/) and results were made available to treating physicians to guide their decision in using the CCR5 inhibitor maraviroc.

Mixtures of HIV viral populations in the same biological sample were identified visually inspecting the ambiguities present in amino acid sequences. Different sequence possibilities were separately submitted to the geno2pheno platform and results presenting mixtures of sequences harboring R5 and non-R5 strains were reported. As a quality control/assurance procedure, one out of every 40 samples was blindly tested in duplicate to confirm the results.

Results

All of these samples are from treatment experienced patients who have been exposed to the three classes of antiretrovirals. Viral subtypes according to V3 env sequence profiles revealed HIV-1 subtype B as the most prevalent subtype (88.8%), followed by subtype F (5.6%), subtype C (3.3%), subtype A (1.8%), and subtype G (0.5%). V3 genotyping results showed 59.9% of the samples were harboring R5-tropic viruses, 35% were harboring non-R5-tropic viruses, and 5.1% were predicted to be harboring mixtures of R5 and non-R5-tropic viruses. As seen in Table 2, the prevalence of non-R5 strains was higher in subtype B viruses (42.8%) as compared to non-B viruses (19.1%; Chi-square, p<0.0003).

Results from geno2pheno reveals that CCR5 antagonists would not be effective in more than 40% of our multiantiretroviral experienced patients (patients with non-R5 variants and mixtures). Our results are in line with Melby et al. ⁵ and Wilkin et al. ⁶ who reported that treatment experienced patients have lower R5 prevalence rates (around 50% or lower) than those who are treatment naive (75% and higher). Significant improvement over time of sample processing was observed after the implementation of a mixture of primers for PCR reaction (Table 1). The new PCR strategy consisted in the use of a cocktail of primers as described above. Using the original set of two pairs of PCR primers in 2009, results have not been obtained from DBSs in 39% of patients due to negative PCR results.

Once the cocktail of primers for the outer nested PCR started to be used, the percentage of negative PCR results from DBSs was reduced to 19% during the years 2010–2011. Every time a DBS sample did not yield a positive PCR, a new whole blood sample from the same patients was collected using EDTA tubes. The performance of PCR and sequencing after recollection using whole blood samples was 100%, and therefore sequence results have been obtained from all 608 tested individuals (Table 3).

A comparison of the sensitivity between these two PCR strategies was not investigated using patient's samples. The prevalence of non-B strains was higher among samples where results were obtained from whole blood (30 out of 112 samples, 21%) as compared to samples where results were readily obtained from DBS (38 out of 428 samples, 8%; Chi-square, p=0.0003, OR=0.33, 95% confidence limits of 0.19 and 0.57). Also the prevalence of non-R5 strains was lower for whole blood (44 out of 142 samples, 31%) as compared to DBS (200 out of 466 samples, 43%; Chi-square, p=0.02, OR=1.7, 95% confidence limits of 01.1 and 2.55). There was no significant correlation between the distance from the Central Laboratory to the site where the DBS samples were collected, or the time from collection and sample processing, or the prevalence of non-R5 samples or non-B subtypes, and the percentage of negative PCR results. As viral loads and CD4⁺ T cell counts were not available, we were unable to evaluate the impact of these variables on the percentage of negative PCR results.

The visual inspection of sequence results revealed that 34% of sequences presented ambiguities that led to nonsynonymous amino acid substitutions, with a mean number of ambiguities of 2.3 varying from 1 to 5. All sequence possibilities were individually submitted to geno2pheno[coreceptor], and of note, sequences with five amino acid ambiguities generate 120 sequence possibilities. As mentioned above, with this strategy we were able to predict that 5.1% of the samples could have a mixture of R5 and non-R5 strains. Nonetheless, this strategy did not increase the prevalence of predicted non-R5 strains.

We tested 15 samples blindly in duplicate as a quality control/assurance procedure, and phylogenetically confirmed using neighbor-joining trees (PHYLIP program package, version 3.57) that samples were related (data not shown), presenting a mean genetic diversity of 0.02 varying from 0.00 to 0.04, in which nine samples were R5, and tropism prediction did not change between duplicates.

Discussion

It was intriguing that the prevalence of non-R5 strains was much lower among non-subtype B-infected individuals in Brazil. This could be explained by the fact that non-B strains are newer in the Brazilian population than the subtype B strains. For instance, it has been predicted using Bayesian Markov chain Monte Carlo (MCMC) methods and the Reversible-jump MCMC method that HIV-1 subtype B emerged in 1971, subtype F emerged in 1981, BF recombinants emerged in 1989, subtype C emerged in 1987, and BC recombinants emerged in 1992. ⁷ Regrettably, we do not have the CD4-positive T cell counts for these patients to determine if there was a difference among different HIV subtypes in our set of samples. Also, it may also be conceivable that the genetic diversity of the viruses may influence the pace of HIV-1 tropism change because the emergence of X4 viruses occurs very early among subtype D-infected individuals, and it occurs late in infections caused by subtype C viruses. ^{8 –10} Alternatively, the genotypic correlates of HIV tropism among non-B strains may still not be clear. According to our results, it can also be speculated that DBS may be less efficacious than the use of whole blood among non-B HIV strains, and further studies designed to clarify this issue need to be pursued.

We recognize that the sampling of HIV genome variants may be underrepresented due to the limited amount of sample contained in a DBS. However, the genotypic assays for the sequencing of the V3 region of envelope gp120 (genotropism) can be a faster and cheaper alternative for patients in comparison to phenotypic assays. Our results demonstrated that the use of DBS for the determination of coreceptor tropism might be a cost-effective strategy, especially in resource-limited and difficult access areas, reducing the risk of transport of infectious material or the risk of injury among health professionals.