The extensive sequence diversity of HIV-1 poses a formidable challenge in vaccine development.1 The search for a universal HIV-1 vaccine has focused on the induction of adaptive immune responses through multiclade, polyvalent viral vectors such as the recombinant canarypox vector used in RV-144, which offered some efficacy in protection against infection.2,3 Although modest protection against HIV-1 acquisition during the RV-144 trial in Thailand mainly correlated with binding antibodies, preclinical studies in nonhuman primates and observational studies in HIV-infected adults have established that epitope-specific T-cell responses play a critical role in controlling viral replication after infection occurs.4–10 Further, simian immunodeficiency virus (SIV) vaccines that induce a greater breadth of virus-specific T-cell responses at high frequencies correlated with better viral control.11–13 A broader CD4 and CD8 T-cell response has also correlated with better markers of HIV disease progression.14–17 Thus, mitigating cell-mediated immunity through T-cell responses may be imperative for protection against disease progression in those in whom antibodies failed to protect against infection. Such optimization will likely require the identification of new targets to induce a more broadly reactive effector population.
Cryptic epitopes (CEs) are a unique set of major histocompatibility class I (MHC-I)-restricted T-cell epitopes recognized during infection with SIV, HIV, murine leukemia virus (LP-BM5), and influenza virus and provide a novel source of HIV-1 antigens for vaccine development.18–22 In contrast with traditional, protein-derived epitopes, CEs are encoded by forward (F2, F3) and reverse (R1, R2, R3) alternative reading frames (ARFs) of existing genes, corresponding to the translation of sense and antisense transcripts of messenger RNA.23–26 Thus, a double-stranded DNA vector can express traditional epitopes from the primary reading frame (F1) of the sense transcript while expressing CE in the remaining 2 frames (F2, F3) of the same transcript or the 3 reverse frames (R1–R3) of the antisense transcript. Therefore, the prevalence of epitopes could be 5 times greater in ARFs, making CEs ideal for expanding the breadth of vaccine-induced responses without expanding the current size of the vector inserts. Despite this advantage, all vector inserts to date have been designed to preserve targets encoded by the primary reading frame (F1) without considering the effects on the ARFs. Although insert sequences still contain ARFs (F2, F3, and R1–R3), neither the sequence conservation of CE from these frames nor the potential of current vaccine vectors to induce CE-specific T-cell responses has been directly evaluated.
Because of the importance of broadening T-cell response for HIV vaccine design and the potential implications of broadening these responses through CE targeting, we evaluated whether select HIV vaccines in clinical development are able to induce these types of responses. We found that recombinant vaccines with codon-optimized inserts skewed CEs to such an extent that they poorly represented naturally occurring HIV strains. Analysis of samples from the recipients of a vaccine regimen encoding noncodon-optimized HIV inserts revealed that vaccination could induce CE-specific T-cell populations but at frequencies that were low in comparison with that of protein-derived epitopes. These findings suggest that current HIV vaccines have the capacity to express CE and, if the ARFs that encode them are prioritized, vaccination could further increase the breadth of vector-induced T-cell responses.
MATERIALS AND METHODS
The gag and RT region nucleotide sequences for each full length HIV-1 clade B transmitted founder virus (TFV) genome that was available from 12 acutely infected patients in our cohort27 were aligned before determining a single consensus sequence for each reading frame (LANL's Consensus Maker tool, http://www.hiv.lanl.gov/content/sequence/CONSENSUS/SimpCon.html; Los Alamos National Laboratory, Los Alamos, NM). This reference set was then translated (ExPASy's Translate tool, http://web.expasy.org/translate/; Swiss Institute of Bioinformatics, Lausanne, CH), along with 1 noncodon-optimized vaccine insert (MVA/HIV62, Acc. AY528646), 2 codon-optimized vaccine inserts (MRKAd5, obtained through personal communication with M. Betts; VRCrAd5, Acc.VRC-4306), and the 2 wild type parent sequences that were optimized (wtMRKAd5 gag, CAM-1 Acc. D10112, pol LAV-1 Acc. K02013, and wtVRCrAd5, NL4.3 Acc. M19921). Pairwise alignment scores were computed by Geneious (BLOSUM62 cost matrix with a gap open penalty of 12.0 and gap extend penalty of 3.0; Biomatters, Auckland, New Zealand) for each frame of the TFV consensus sequences and all 5 test sequences.28 Homology was plotted (Prism6 v6.0b, GraphPad Software, La Jolla, CA) as the proportion of TFV consensus residues that were conserved in the sequence. Significance was determined using 2-tailed Fisher exact tests calculated by GraphPad QuickCalcs (http://www.graphpad.com/quickcalcs/ConfInterval1.cfm).
In a separate analysis, vaccine inserts and their wild type parent amino acid sequences were compared with individual TFV consensus sequences. Alignment scores for all amino acids translated from all 6 reading frame of gag and RT were computed by Geneious using the matrix described above and the Jukes–Cantor correction model. Unrooted, neighbor-joining phylogenetic trees were plotted to infer amino acid sequence divergence among the sequences without assuming any ancestral relationships.
Previously cryopreserved peripheral blood mononuclear cells (PBMCs) were obtained from 126 participants based on their chronological enrollment in a phase 2a trial (HVTN 205). Five individuals were excluded after their response rate for negative controls repeatedly exceeded the statistical threshold for positivity (described below), and one was excluded because the individual seroconverted 3 months post final vaccination. Statistical analyses were performed for 120 participants who received intramuscular injections of either a 0.9% sodium chloride placebo (40) or vaccine treatment (80) in a testing schema with two 3-mg DNA primes followed by two 1 mL of 1 × 108TCID50 MVA boosts at 0, 2, 4, and 6 months, respectively. Vaccine doses of the pGA/JS7 DNA prime contained JS7DNA plasmid encoding clade B HIV-1 HXB2 gag, HXB2/ADA env-tat-rev-vpu, and BH10 PR–RT with an inactivating point mutation in PR; the MVA boost, HIV/MVA62, is a modified vaccinia Ankara vector encoding the same gag, PR–RT, and env sequences.29 PBMC samples were obtained 2 weeks post last vaccination. Stratification of participants into responders and nonresponders was finalized before reporting the results to the Statistical Center for HIV/AIDS Research and Prevention for unblinding.
Consensus clade B 15mer peptides overlapping by 11 amino acids for HIV-1 Gag and Pol proteins were obtained through the NIH AIDS Reagent Program (catalog #8117, #6208, respectively). Full length Gag, Protease (PR), and Reverse Transcriptase (RT) proviral DNA regions of the MVA/HIV62 vector were used to design overlapping ARF peptides (OLPs, 8–18mers overlapping by 10 residues) for potential CE using PeptGen (http://www.hiv.lanl.gov/content/sequence/PEPTGEN/peptgen.html, Los Alamos National Laboratory) and synthesized in a 96-well array format by New England Biolabs.30–32 Individual OLPs were combined by reading frame into subpools (∼10 peptides per pool) and large pools of not >248 peptides for traditional proteins and 98 peptides for potential CE (range of 33–98 with means of 42 and 86 for gag and PR–RT regions, respectively; see Table S1, Supplemental Digital Content,http://links.lww.com/QAI/A459). Individual peptide, subpool, and large pool stocks were reconstituted at 40 mM in dimethyl sulfoxide, diluted to 100 μM in deionized water, and stored at −80°C until use.
Ex Vivo Interferon Gamma Enzyme-Linked Immunosorbent Spot Assay
Previously cryopreserved PBMCs were thawed and rested overnight at 37°C, 5% CO2 to improve recovery rates, which retained approximately 85% of the cells per vial with average viabilities of 90% as determined by trypan blue staining. Enzyme-linked immunosorbent spot (ELISpot) assays were performed as previously described using protein (NIH AIDS Reagents Program) and ARF peptide pools (New England Biolabs, Ipswitch, MA) at final concentrations of 5 μM per peptide per subpool and 2 μM per peptide per large pool. Antigens were tested in duplicate with 100,000 PBMCs per well in Millipore nitrocellulose 96-well plates while unstimulated cells were plated in quadruplicate as a negative control and phytohemagglutinin in duplicate as a positive control. Spots that developed on the nitrocellulose membrane were counted by an ELISpot reader (ImmunoSpot; Cellular Technology Limited, Shaker Heights, OH) as a measure of interferon gamma (IFNγ) cytokine production. T-cell responses to HIV-1 Gag, Pol, or their respective ARF peptide pools were reported as positive in the IFNγ ELISpot if (1) the negative control response averaged <55 SFU/106 PMBCs, (2) the average positive control response was >500 SFU/106 PBMCs, and (3) the duplicate average of the wells was greater than the mean negative control response of all samples plus 3 standard deviations and (4) ≥3 times the average response of the sample's negative control. Responses between both groups were analyzed for differences in magnitude using 2-tailed nonparametric Mann–Whitney U tests calculated by GraphPad Prism v6.0b and frequency using 2-tailed Fisher exact tests calculated by GraphPad QuickCalcs (www.graphpad.com/quickcalcs/ConfInterval1.cf).
Cultured IFNγ ELISpot Assay
In a subset of PBMC samples, a 12-day in vitro cultured assay was performed concurrently with ex vivo ELISpot assays. Cryopreserved PBMCs were prepared for overnight resting and stimulated with the same antigens as described above. In brief, the cells were suspended after resting in 10% HyClone fetal bovine serum culture medium with 25 ng/mL of interleukin-7 (#207-IL; R&D Systems, Minneapolis, MN) before being plated in 24- or 48-well culture plates at 1 million PBMCs per milliliter. The cells were stimulated with subpools (5 μM final concentration per peptide) and/or large pools (2 μM final concentration per peptide) corresponding to the previous positive large pool responses in the participant. As a control, each sample was stimulated with a previously negative ARF large pool in addition to using unstimulated cells and PHA as negative and positive controls, respectively. On days 2 and 7, interleukin-2 (#354043; Beckton Dickenson, Franklin Lakes, NJ) was added to each well at a final concentration of 40–45 BRMP units per milliliter. Cultures were incubated for an additional 2 days before washing the cells 3 times with phosphate buffer saline and resting in IL-2-free R10 media for 30–36 hours. After resting, cultured cells were tested in an IFNγ ELISpot assay as described above. Based on previous studies using cultured cells, samples with media background values greater than the media mean for all samples plus 2 standard deviations (210 SFU/1 million PBMCs) were excluded.34–38 Positive responders were identified when the magnitude of a subpool response exceeded the 3× the sample's background and media mean for all the samples plus 2 standard deviations. Groups were compared with the Mann–Whitney statistical U test using background-subtracted responses.
Codon-Optimized Vector Inserts Significantly Skew All ARF Sequences
Because CE responses are dominantly due to CD8 T cells, our strategy was to analyze HIV recombinant adenoviral serotype 5 vector vaccines as these can be potent inducers of CD8 T cells. To enhance protein expression without changing the natural amino acid sequence, some HIV-1 vaccines (including all recombinant adenoviral serotype 5 vectors) are codon optimized.39,40 However, CEs are encoded in ARFs, and their conservation is not prioritized during codon optimization of traditional genes.
To quantify the impact on ARF sequences after codon optimization, we compared the amino acid identities of synthetic constructs with ARFs of naturally occurring TFV sequences. Because TFV sequences are derived from sequencing at the earliest disease stage, each TFV sequence corresponds to the transmitted virus's genome after it established infection and before little or any experience of immune pressure in the new host.18 Thus, our use of TFV sequences ensured that epitopes encoded by vector inserts would be scored against reference sequences that accurately represent vaccine targets.41 Using 12 full length HIV-1 clade B TFV sequences that were available from our acutely infected cohort, we compiled a reference set containing 1 amino acid consensus sequence per reading frame and gene region. In a preliminary analysis of a previously recognized CE encoded by gag (Gag-CE), we found that both the Gag and CE residues encoded by a noncodon-optimized vector insert were homologous to the TFV sequence (see Figure S1, Supplemental Digital Content,http://links.lww.com/QAI/A459). In contrast, a codon-optimized vector encoded the same Gag sequence but the CE incurred an 11% loss in homology (1 of 9 residues) when compared with the TFV sequence and the wild type, noncodon-optimized sequence from which its insert was designed. Given these results at the epitope level, pairwise alignments for the full length of the inserts were then performed to measure amino acid conservation with respect to the TFV reference set for gene regions that were common (gag and reverse transcriptase, RT) to 1 noncodon-optimized vector insert (MVA/HIV62), 2 codon-optimized vector inserts (MRKAd5, VRCAd5), and both wild type sequences from which the codon-optimized inserts were derived. We found that codon optimizing the protein-coding frame (F1) amplified the effect seen at the epitope level, leading to a significant decrease (Fisher exact, P < 0.0001) in the amino acid identity for all ARFs of gag (mean, 46.3%, range of 34.0%–79.2%; Fig. 1A) and RT (mean, 40.4%; range of 27.7%–78.9%; Fig. 1B) that could encode CE (ARFs F2, F3, and R1–R328).
To further evaluate the cummulatlve effect of codon optimization on vector inserts, we used phylogenetic analyses to compare individual TFV and insert sequences from 2 vaccine vectors (MVA/HIV62 and VRCrAd5) for which we had access to clinical trial samples. Genetic distances between the TFV reference set, all vector inserts, and wild type sequences were calculated for both proteins (F1) from the gag and reverse transcriptase (RT) regions and all ARF amino acid sequences (F2, F3; R1–R3), which would include any potential CE (Fig. 2). Tight clustering of tree branches indicates that amino acid sequences for Gag and RT proteins (F1) are genetically similar (Figs. 2A, B, left panels). Noncodon-optimized sequences demonstrated a tight clustering for both proteins (F1) and potential CE (F2, F3, R1–R3; Fig. 2, all panels). Similarly, codon-optimized Gag and RT (F1) did not significantly diverge from the TFV reference set. However, codon-optimized ARFs had distances 2–5 times greater than the TFV reference set and the wild type noncodon-optimized sequences (Figs. 2A, B, right panels). The genetic disparity shown for F2 of gag and pol was representative of the codon-optimized CE encoded by the remaining ARFs (F3, R1–R3; data not shown). These results confirmed that regardless of the method used to encode the insert, there was little difference between the residues of proteins encoded by the noncodon-optimized, codon-optimized, wild type, or TFV consensus sequences. Conversely, the analysis of the sequences from each of the 5 ARFs in the gag and RT regions of the genome demonstrated that codon optimization of inserts can lead to a significant decrease in amino acid conservation for ARFs and a >2-fold decrease in homology between targets encoded by TFVs and vaccine-derived CE.
Lack of CE-Specific T-Cell Responses in Recipients of a Noncodon-Optimized Vaccine
Based on the results given above, we exclusively evaluated CE-specific T-cell responses to a noncodon-optimized vaccine vector, MVA/HIV62. Responses to proteins and CE encoded by the gag and PR–RT inserts were measured with an IFN-γ ELISpot assay using PBMCs collected from the participants enrolled in HVTN 205 who received 2 pGA2/JS7 DNA injections at 0 and 2 months followed by 2 vector boosts at 4 and 6 months or placebo (Table 1). As expected from prior studies with this product, Gag but not Pol-specific T-cell responses were present only in vaccine recipients29 (Fig. 3). CE-specific responses were occasionally detected in vaccinees, especially in PR–RT ARF F2. However, these were not significantly greater than the responses seen in the placebo controls for any of the gag or PR–RT ARFs. Further, subsequent mapping of the PR–RT ARF F2 responses failed to identify any peptide, indicating that the detected peptide pool responses were either weak or not present at all.
In Vitro Cultured Assay Detected a Higher Frequency of Responders in Vaccinees
A cultured assay was performed on a select set of positive responders to determine if low-frequency CE-specific responses were present and could be expanded after a 12-day in vitro expansion. This assay's extended period of antigenic stimulation allows for the proliferation of low-frequency HIV-1-specific T cells and therefore achieves a greater sensitivity than does an ex vivo ELISpot assay.33–36 Eighteen participant samples from both treatment groups were cultured with peptides from gag and PRT-RT ARFs but a high background seen in the samples from 7 participants excluded them from further analysis. In the 11 remaining samples with an acceptable background, the magnitude of CE-specific T cells was significantly greater in vaccinees (P < 0.0001; Fig. 4A). Further, CE-specific responses were only detected in the vaccinees and in none of the placebo controls (Fisher exact test, P = 0.06; Fig. 4B). In addition, none of the irrelevant ARF pools (those that were negative at the initial testing) were positive upon culturing. Taken together, these data suggest that CE responses may be induced by this DNA/MVA vaccine regimen, albeit at low frequencies.
Translation of CEs from ARFs highlights the immense coding capacity of HIV-1's genome and increases the demand for immunologic coverage through vaccination.7,9,10,12 These immunogens present a new source of HIV-1 antigen diversity that further challenges the development of a broadly protective HIV-1 vaccine.1,4 In retrospect, vaccine candidates in ongoing clinical trials were designed before research that established the immunological relevance of CE in HIV-1-infected patients.18–20 Consequently, current vaccines prioritize the preservation of primary reading frames rather than ARFs, which may affect the generation of CE.39,40,42 We therefore assessed the CE coding capacity of current vector designs, which supported the investigation of CE-specific T-cell responses in recipients of a noncodon-optimized HIV-1 vaccine regimen.
Our comparison of vector inserts demonstrated that codon optimization of HIV-1 genes preserves the amino acid sequences encoded by the primary reading frame (F1) but decreases amino acid conservation of potential CE from ARFs (F2, F3, R1–R3). Because of the high magnitude of responses to PR–RT–CE for ARF F2, we tested the same large pool of peptides in the participants vaccinated with the VRC DNA prime rAd5 vaccine as part of HVTN 084.43 However, no T-cell responses specific to ARF F2 were detected when evaluated ex vivo (unpublished results). Although not investigated by our group, it is also possible that codon optimization skews the normal CE production by skewing the regulatory sequences governing the synthesis of transcriptional or translational products.44,45 Similarly, recognition of optimized CE may be skewed because previous immunogenicity studies have shown that at least two-thirds of a peptide's residues must be conserved for crossrecognition by epitope-specific T cells.46 The average percent (46.3% and 40.4%, gag and PR–RT regions, respectively) of amino acid conservation for codon-optimized CE fell well below this threshold with the exception of frame R2. The higher homology in R2 results from the shared positioning of optimized codons in the primary reading frame (F1). Because codon optimization of a gene (F1) frequently changes the nucleotide in the third or wobble position of codons, its complementary base in the second reverse reading frame (R2) is also changed at the wobble position, which minimizes the decrease in homology for this reading frame.47 In comparison, all the frames of noncodon-optimized inserts were nearly identical to TFV sequences from acutely infected patients, indicating a higher probability of inducing crossreactive, CE-specific T cells during a transmission event. Moreover, these observations suggest that the native context of HIV-1's genomic architecture is more rigid than was previously suspected, where the manipulation of gene sequences may result in decreased antigen production, presentation, and/or recognition.48,49
In participants of HVTN 205 receiving a vector encoding a noncodon-optimized insert, we were unable to detect significant CE responses ex vivo. Such a finding could be due to a number of factors including the use of cryopreserved specimens and the method of resting cells overnight both of which may have contributed to enhanced cell death. We find these possibilities unlikely because both systems are used extensively by our group and the HVTN and do not significantly impact responses seen to epitopes in the protein reading frame.50–52 It seems more likely that the poor responses observed for CE lies in the fact that even with HIV infection responses to CE are low in magnitude and relatively infrequent.18,19 Further, the average OLP length (15mer, with a range of 8–18mer) exceeded the length of peptides commonly recognized by MHC-I (8–11mer), which potentially reduced the sensitivity to CD8+ T-cell responses in our assays. In addition, the higher frequency of stop codons in ARFs further restricted the size and the number of OLPs that could be artificially synthesized. Accordingly, CE generated through nonconventional translation methods, including stop codon readthrough and doublet decoding, may have been undetectable with the described set of OLPs.32,53 Further, the low immunogenicity of Gag- and PR–RT–CE suggests that MVA/HIV62 may not have been processed efficiently or effectively presented to the immune system.54,55 During infection, translation of PR–RT from the gag–pol transcript relies on a native-1 ribosomal frameshift, which reduces PR–RT expression to one-tenth the level of Gag protein. Because the vector insert maintained the same out-of-frame nucleotide code for the gag and PR–RT regions, vaccine-derived PR–RT levels may have been insufficient for priming central memory T-cell populations and limited the number of effector responses directed against Pol protein in this study. Alternatively, 90% of gag–pol reads do not result in frameshifting, such that translation could occur downstream of gag and generate CE from ARF F2 of PR–RT. Expression from this region is consistent with the trend seen for PR–RT–CE F2 responses (Fig. 3B) and our previous data, which suggested that epitopes in the pol region are predominantly cryptic.18 Nevertheless, a cultured assay was performed with the same OLPs and detected CE-specific T-cell responses exclusively in vaccinees.
Our inability to detect vaccine-induced CE responses ex vivo contrast with what has been reported in rhesus macaques immunized with SIV vaccines.26,56 In fact, CE responses have been reported in up to one-quarter of these vaccinated macaques.57 However, these macaques were vaccinated with SIV rAd5 vectors that potently induce CD8 T-cell responses unlike MVA recombinant vectors, used in our study, that induce relatively weak CD8 responses and are more skewed toward CD4. Additionally, the breadth of vaccine-induced responses is significantly greater in macaques compared with that in humans likely due to the fact that the former primates are able to encode many more MHC type I alleles than are the 6 available to humans.58 As such it seems reasonable to expect that this greater vaccine-induced breadth would extend to CE as well.
Together, our in silico and in vitro analyses suggest that codon optimization negatively impacts the breadth of CE-specific T-cell responses when compared with the repertoire generated through infection or vaccination with a noncodon-optimized vector. Vaccination with the noncodon-optimized MVA/HIV62 vector-induced cellular responses that were low in magnitude and frequency, indicating HIV-1 CE are presented but at levels that may not be adequate for recognition by peripheral T-cells. Still, the immunity afforded to animals vaccinated with naturally encoded constructs supports delivery of noncodon-optimized vectors to expand the pool of vaccine-induced epitopes.59 As future studies determine the importance of CE responses in viral control, our research may provide the direction for developing vaccine designs that increase the breadth of T-cell responses by either preserving as many natural HIV-1 targets as possible or specifically engineering vaccines to express CE as immune targets.
The authors thank all the volunteers who participated in the HVTN 205 study. The authors also thank the HIV Vaccine Trials Network and GeoVax, Inc for their collaboration in supporting this work.
1. Taylor BS, Hammer SM. The challenge of HIV-1 subtype diversity. N Engl J Med. 2008;359:1965–1966.
2. Rerks-Ngarm S, Pitisuttithum P, Nitayaphan S, et al.. Vaccination with ALVAC and AIDSVAX to prevent HIV-1 infection in Thailand. N Engl J Med. 2009;361:2209–2220.
3. Picker LJ, Hansen SG, Lifson JD. New paradigms for HIV/AIDS vaccine development. Annu Rev Med. 2012;63:95–111.
4. Haynes BF, Gilbert PB, McElrath MJ, et al.. Immune-correlates analysis of an HIV-1 vaccine efficacy trial. N Engl J Med. 2012;366:1275–1286.
5. McElrath MJ, De Rosa SC, Moodie Z, et al.. HIV-1 vaccine-induced immunity in the test-of-concept Step Study: a case-cohort analysis. Lancet. 2008;372:1894–1905.
6. Buchbinder SP, Mehrotra DV, Duerr A, et al.. Efficacy assessment of a cell-mediated immunity HIV-1 vaccine (the Step Study): a double-blind, randomised, placebo-controlled, test-of-concept trial. Lancet. 2008;372:1881–1893.
7. Letvin NL, Mascola JR, Sun Y, et al.. Preserved CD4+ central memory T cells and survival in vaccinated SIV-challenged monkeys. Science. 2006;312:1530–1533.
8. Barouch DH, Liu J, Li H, et al.. Vaccine protection against acquisition of neutralization-resistant SIV challenges in rhesus monkeys. Nature. 2012;482:89–93.
9. Hansen SG, Ford JC, Lewis MS, et al.. Profound early control of highly pathogenic SIV by an effector memory T-cell vaccine. Nature. 2011;473:523–527.
10. Fukazawa Y, Park H, Cameron MJ, et al.. Lymph node T cell responses predict the efficacy of live attenuated SIV vaccines. Nat Med. 2012;18:1673–1681.
11. Liu J, O'Brien KL, Lynch DM, et al.. Immune control of an SIV challenge by a T-cell-based vaccine in rhesus monkeys. Nature. 2009;457:87–91.
12. Wilson NA, Keele BF, Reed JS, et al.. Vaccine-induced cellular responses control simian immunodeficiency virus replication after heterologous challenge. J Virol. 2009;83:6508–6521.
13. Mudd PA, Martins MA, Ericsen AJ, et al.. Vaccine-induced CD8+ T cells control AIDS virus replication. Nature. 2012;491:129–133.
14. Edwards BH, Bansal A, Sabbaj S, et al.. Magnitude of functional CD8+ T-cell responses to the gag protein of human immunodeficiency virus type 1 correlates inversely with viral load in plasma. J Virol. 2002;76:2298–2305.
15. Goulder PJ, Watkins DI. Impact of MHC class I diversity on immune control of immunodeficiency virus replication. Nat Rev Immunol. 2008;8:619–630.
16. Brumme ZL, Brumme CJ, Carlson J, et al.. Marked epitope- and allele-specific differences in rates of mutation in human immunodeficiency type 1 (HIV-1) Gag, Pol, and Nef cytotoxic T-lymphocyte epitopes in acute/early HIV-1 infection. J Virol. 2008;82:9216–9227.
17. Chevalier MF, Julg B, Pyo A, et al.. HIV-1-specific interleukin-21+ CD4+ T cell responses contribute to durable viral control through the modulation of HIV-specific CD8+ T cell function. J Virol. 2011;85:733–741.
18. Bansal A, Carlson J, Yan J, et al.. CD8 T cell response and evolutionary pressure to HIV-1 cryptic epitopes derived from antisense transcription. J Exp Med. 2010;207:51–59.
19. Berger CT, Carlson JM, Brumme CJ, et al.. Viral adaptation to immune selection pressure by HLA class I-restricted CTL responses targeting epitopes in HIV frameshift sequences. J Exp Med. 2010;207:61–75.
20. Champiat S, Raposo RA, Maness NJ, et al.. Influence of HAART on alternative reading frame immune responses over the course of HIV-1 infection. PLoS One. 2012;7:e39311.
21. Gaur A, Green WR. Role of a cytotoxic-T-lymphocyte epitope-defined, alternative gag open reading frame in the pathogenesis of a murine retrovirus-induced immunodeficiency syndrome. J Virol. 2005;79:4308–4315.
22. Dolan BP, Li L, Takeda K, et al.. Defective ribosomal products are the major source of antigenic peptides endogenously generated from influenza A virus neuraminidase. J Immunol. 2010;184:1419–1424.
23. Michael NL, Vahey MT, d'Arcy L, et al.. Negative-strand RNA transcripts are produced in human immunodeficiency virus type 1-infected cells and patients by a novel promoter downregulated by Tat. J Virol. 1994;68:979–987.
24. Yewdell JW, Anton LC, Bennink JR. Defective ribosomal products (DRiPs): a major source of antigenic peptides for MHC class I molecules. J Immunol. 1996;157:1823–1826.
25. Cardinaud S, Moris A, Fevrier M, et al.. Identification of cryptic MHC I-restricted epitopes encoded by HIV-1 alternative reading frames. J Exp Med. 2004;199:1053–1063.
26. Maness NJ, Valentine LE, May GE, et al.. AIDS virus-specific CD8(+) T lymphocytes against an immunodominant cryptic epitope select for viral escape. J Exp Med. 2007;204:2505–2512.
27. Ochsenbauer C, Edmonds TG, Ding H, et al.. Generation of transmitted/founder HIV-1 infectious molecular clones and characterization of their replication capacity in CD4 T lymphocytes and monocyte-derived macrophages. J Virol. 2012;86:2715–2728.
28. Drummond AJ, Ashton B, Buxton S, Cheung M, et al.. Geneious. 5.6 ed. 2012.
29. Goepfert PA, Elizaga ML, Sato A, et al.. Phase 1 safety and immunogenicity testing of DNA and recombinant modified vaccinia Ankara vaccines expressing HIV-1 virus-like particles. J Infect Dis. 2011;203:610–619.
30. Draenert R, Brander C, Yu XG, et al.. Impact of intrapeptide epitope location on CD8 T cell recognition: implications for design of overlapping peptide panels. AIDS. 2004;18:871–876.
31. Yewdell JW, Hickman HD. New lane in the information highway: alternative reading frame peptides elicit T cells with potent antiretrovirus activity. J Exp Med. 2007;204:2501–2504.
32. Bullock TN, Patterson AE, Franlin LL, et al.. Initiation codon scanthrough versus termination codon readthrough demonstrates strong potential for major histocompatibility complex class I-restricted cryptic epitope expression. J Exp Med. 1997;186:1051–1058.
33. Bansal A, Yue L, Conway J, et al.. Immunological control of chronic HIV-1 infection: HLA-mediated immune function and viral evolution in adolescents. AIDS. 2007;21:2387–2397.
34. Muturi M, Gichuki C, Mutua G, et al.. Cultured enzyme-linked immunospot assay (ELISPOT) enhances detection of low HIV-specific immune responses in exposed seronegative individuals in Kenya. Retrovirology. 2006:3(suppl 1):44.
35. Goonetilleke N, Moore S, Dally L, et al.. Induction of multifunctional human immunodeficiency virus type 1 (HIV-1)-specific T cells capable of proliferation in healthy subjects by using a prime-boost regimen of DNA- and modified vaccinia virus Ankara-vectored vaccines expressing HIV-1 Gag coupled to CD8+ T-cell epitopes. J Virol. 2006;80:4717–4728.
36. Winstone N, Guimaraes-Walker A, Roberts J, et al.. Increased detection of proliferating, polyfunctional, HIV-1-specific T cells in DNA-modified vaccinia virus Ankara-vaccinated human volunteers by cultured IFN-gamma ELISPOT assay. Eur J Immunol. 2009;39:975–985.
37. Campion S, Cohen MS, McMichael AJ, et al.. Improved detection of latent Mycobacterium tuberculosis
infection in HIV-1 seropositive individuals using cultured cellular assays. Eur J Immunol. 2011;41:255–257.
38. Ritchie AJ, Campion SL, Kopycinski J, et al.. Differences in HIV-specific T cell responses between HIV-exposed and -unexposed HIV-seronegative individuals. J Virol. 2011;85:3507–3516.
39. Nagata T, Uchijima M, Yoshida A, et al.. Codon optimization effect on translational efficiency of DNA vaccine in mammalian cells: analysis of plasmid DNA encoding a CTL epitope derived from microorganisms. Biochem Biophys Res Commun. 1999;261:445–451.
40. Abdulhaqq SA, Weiner DB. DNA vaccines: developing new strategies to enhance immune responses. Immunol Res. 2008;42:219–232.
41. Keele BF, Giorgi EE, Salazar-Gonzalez JF, et al.. Identification and characterization of transmitted and early founder virus envelopes in primary HIV-1 infection. Proc Natl Acad Sci U S A. 2008;105:7552–7557.
42. Priddy FH, Brown D, Kublin J, et al.. Safety and immunogenicity of a replication-incompetent adenovirus type 5 HIV-1 clade B gag/pol/nef vaccine in healthy adults. Clin Infect Dis. 2008;46:1769–1781.
43. Graham BS, Enama ME, Nason MC, et al.. DNA vaccine delivered by a needle-free injection device improves potency of priming for antibody and CD8+ T-cell responses after rAd5 boost in a randomized clinical trial. PLoS One. 2013;8:e59340.
44. Cannarozzi G, Schraudolph NN, Faty M, et al.. A role for codon order in translation dynamics. Cell. 2010;141:355–367.
45. Smith JM, Amara RR, Campbell D, et al.. DNA/MVA vaccine for HIV type 1: effects of codon-optimization and the expression of aggregates or virus-like particles on the immunogenicity of the DNA prime. AIDS Res Hum Retroviruses. 2004;20:1335–1347.
46. Malhotra I, Wamachi AN, Mungai PL, et al.. Fine specificity of neonatal lymphocytes to an abundant malaria blood-stage antigen: epitope mapping of Plasmodium falciparum
MSP1(33). J Immunol. 2008;180:3383–3390.
47. Kotlar D, Lavner Y. The action of selection on codon bias in the human genome is related to frequency, complexity, and chronology of amino acids. BMC Genomics. 2006;7:67.
48. Watts JM, Dang KK, Gorelick RJ, et al.. Architecture and secondary structure of an entire HIV-1 RNA genome. Nature. 2009;460:711–716.
49. Stephenson KE, SanMiguel A, Simmons NL, et al.. Full-length HIV-1 immunogens induce greater magnitude and comparable breadth of T lymphocyte responses to conserved HIV-1 regions compared with conserved-region-only HIV-1 immunogens in rhesus monkeys. J Virol. 2012;86:11434–11440.
50. Bull M, Lee D, Stucky J, et al.. Defining blood processing parameters for optimal detection of cryopreserved antigen-specific responses for HIV vaccine trials. J Immunol Methods. 2007;322:57–69.
51. Filbert H, Attig S, Bidmon N, et al.. Serum-free freezing media support high cell quality and excellent ELISPOT assay performance across a wide variety of different assay protocols. Cancer Immunol Immunother. 2013;62:615–627.
52. Janetzki S, Price L, Britten CM, et al.. Performance of serum-supplemented and serum-free media in IFNgamma Elispot assays for human T cells. Cancer Immunol Immunother. 2010;59:609–618.
53. Bruce AG, Atkins JF, Gesteland RF. tRNA anticodon replacement experiments show that ribosomal frameshifting can be caused by doublet decoding. Proc Natl Acad Sci U S A. 1986;83:5062–5066.
54. Bansal A, Gough E, Sabbaj S, et al.. CD8 T-cell responses in early HIV-1 infection are skewed towards high entropy peptides. AIDS. 2005;19:241–250.
55. Harari A, Bart PA, Stohr W, et al.. An HIV-1 clade C DNA prime, NYVAC boost vaccine regimen induces reliable, polyfunctional, and long-lasting T cell responses. J Exp Med. 2008;205:63–77.
56. Maness NJ, Wilson NA, Reed JS, et al.. Robust, vaccine-induced CD8(+) T lymphocyte response against an out-of-frame epitope. J Immunol. 2010;184:67–72.
57. Maness NJ, Walsh AD, Piaskowski SM, et al.. CD8+ T cell recognition of cryptic epitopes is a ubiquitous feature of AIDS virus infection. J Virol. 2010;84:11569–11574.
58. Otting N, Heijmans CM, Noort RC, et al.. Unparalleled complexity of the MHC class I region in rhesus macaques. Proc Natl Acad Sci U S A. 2005;102:1626–1631.
59. Schirmbeck R, Riedl P, Fissolo N, et al.. Translation from cryptic reading frames of DNA vaccines generates an extended repertoire of immunogenic, MHC class I-restricted epitopes. J Immunol. 2005;174:4647–4656.