|Home | About | Journals | Submit | Contact Us | Français|
The Vγ2Vδ2 T cell subset responds to Bacille Calmette-Guerin (BCG) immunization in macaques and may be a component of protective immunity against tuberculosis. We characterized the effects of BCG on the Vγ2Vδ2 T cell receptor repertoire by comparing the starting population of Vγ2 chains in cynomolgus macaques with the repertoire found after priming or booster immunization with BCG. The starting repertoire was dominated by public Vγ2 chain sequences that were found repeatedly among unrelated animals. Primary exposure to BCG triggered expansion of cells expressing public Vγ2 chains and booster immunization was often associated with contraction of these same subsets. Thus, BCG-reactive Vγ2 chains were present at high frequency in the repertoire of mycobacteria-naïve macaques and they comprised the major response to primary or booster immunization. Normal selection processes that created the naïve Vγ2 repertoire in macaques, also encoded the capacity for rapid responses to mycobacteria. The unusual composition of a normal Vγ2 repertoire helps to explain the powerful γδ T cell responses to BCG immunization.
In human and non-human primates, the major γδ T cell subset in blood has the Vγ2Vδ2 T cell receptor 1–4. Cells expressing this TCR receptor respond to stimulation with mycobacteria in vitro 5–9; responses include proliferation, production of cytokines such as IFN-γ 10 or TNF-α 11, and cytotoxicity against M. tuberculosis (Mtb) and Mtb-infected cells 12. Vγ2Vδ2 T cells also react to purified mycobacterial components including the non-peptidic phosphoantigens called TubAg1-4 8, 13. These antigens are chemically similar to isopentenyl pyrophosphate (IPP), an intermediate in cholesterol biosynthesis that is used frequently as a model phosphoantigen for γδ T cell studies 9. The γδ T cell responses to phosphoantigens are MHC unrestricted 14, 15, require expression of the Vδ2 chain 16 and depend on specific residues in the Vγ2 chain CDR3 region 16–18.
Phosphoantigen responses are positively selected during normal ontogeny. Thymocytes or cord blood lymphocytes contain low frequencies of Vγ2Vδ2 T cells, 2, 4 but within a few years after birth, the repertoire evolves because of strong selection for the Vγ2-Jγ1.2 rearrangement 4 that is mostly paired with the Vδ2 chain 4, 19. It is believed that stimulation by self- or ubiquitous non-self antigens amplifies and maintains the Vγ2-Jγ1.2Vδ2+ population, thus creating the mature repertoire 2, 4, 20–22. In healthy human adults, the Vγ2-Jγ1.2 rearrangement accounts for more than 70% of Vγ2 chains in peripheral blood 20, 23. Similar results were obtained for Macaca mulatta 24, 25 and Macaca fascicularis 26 monkeys although we know that monkeys have some cells expressing Vγ2 chains paired with Vδ127 (a combination that is infrequent for adult human beings) and that subset will influence the repertoire. The Vγ2-Jγ1.2+ subset responds to a variety of pathogens 28–33 and tumor cells 34–40.
Bacille Calmette-Guerin (BCG) inoculation in macaques stimulates the Vγ2Vδ2 T cell subset and protects against lethal challenge with M. tuberculosis, arguing that antigen specific Vγ2 T cells are an important part of the protective immunity against tuberculosis 41. However, the starting Vγ2 repertoire is selected during normal ontogeny, presumably in the absence of mycobacterial infection, and lymphocytes from unprimed individuals respond strongly to mycobacteria in vitro 6, 42. We hypothesize that the response to BCG is targeted toward a host molecule that is induced or modified during mycobacterium infection, and that the capacity for BCG recognition is already encoded at high levels in the starting Vγ2Vδ2 T cell receptor repertoire. In this study, we test whether BCG-naïve macaques already have the capacity for memory γδ T cell responses to BCG by examining the starting Vγ2 chain repertoire and its changes after primary or booster inoculation with BCG. The nature and specificity of Vγ2Vδ2 T cell responses is important for characterizing the protective immunity against M. tuberculosis, and for the design and testing of new tuberculosis vaccines.
11 juvenile (2 years old) healthy M. fascicularis were housed in the Animal Core Facility at the Institute of Human Virology. Six animals were inoculated intradermally with 5×104 CFU/animal of BCG (Aventis-Pasteur strain) freshly reconstituted in PBS (GIBCO Grand Island, NY), five control animals received an equivalent volume of PBS alone. Five weeks later, animals that received BCG were boosted with 5×105CFU/animal. For each macaque, heparinized blood specimens were collected at several time points before and after the BCG inoculations: 4 and 2 weeks before the immunization, the day of the primary immunization, and 1, 2, 3, 5, 7, 9 weeks after primary inoculation. Experimental protocols were reviewed and approved by the Institutional Animal Care and Use Committee at the University of Maryland Biotechnology Institute.
Peripheral blood mononuclear cells (PBMC) were isolated from macaque blood specimens by centrifugation over Ficoll-Hypaque (Pharmacia, Sweden) density gradient according to the manufacturer’s instructions. PBMC were frozen at 5×106-107/ml in fetal bovine serum (FBS) (GIBCO, Grand Island, NY) with 10% dimethyl-sulphoxyde (DMSO) (Sigma, St Louis, MO) and stored at −130°C.
After thawing, cells were washed twice with RPMI (GIBCO, Grand Island, NY)-15% FBS (GIBCO, Grand Island, NY) and used for phenotypic characterization and cell cultures. PBMC were cultured at 106 cells/ml in RPMI supplemented with 10% FBS, 2mM L-Glutamine (GIBCO, Grand Island, NY), penicillin (100 U/ml) (GIBCO, Grand Island, NY), streptomycin (100 μg/ml) (GIBCO, Grand Island, NY) and 100U/ml of human recombinant interleukin-2 (rhIL-2) (Tecin, Biological Resources Branch, NIH, Bethesda, MD). The phosphoantigen isopentenyl pyrophosphate (IPP) (Sigma, St Louis, MO) was added to a final concentration of 15μM, and in some cultures live BCG (Aventis-Pasteur) was used at 0.3 multiplicity of infection (MOI). Cells were incubated for 13 days at 37°C with 5% CO2. Fresh medium containing IL-2 was added every 3 days.
Thirteen days after stimulation in vitro, cells were harvested and viable counts were determined by the trypan blue dye exclusion method. Expanded Vγ2 or freshly thawed PBMC were resuspended in PBS (GIBCO, Grand Island, NY) with 10% FBS and stained at 4°C with directly conjugated monoclonal antibodies. After 20 minutes, cells were washed twice with PBS-10%FBS and resuspended in PBS containing 2% paraformaldehyde. At least 2×104 lymphocytes (gated on the basis of forward and side scatter profiles) were acquired for each sample on a FACSCalibur (BD Biosciences, San Diego, CA). Flow cytometry data were analyzed with FlowJo software (Tree Star, San Carlos, CA).
The following monoclonal antibodies cross-react with M. fascicularis surface molecules and were used in this study: Vγ9 (clone IMMU360, Beckman Coulter, Miami, FL) (Vγ9 is an alternate nomenclature for the Vγ2 chain); CD95-PE (clone DX2, Pharmingen, BD); CD45-RA (clone F8-11-13, Cimbus Biotechnology); CD8-APC (clone SK1, BD); CD27-PE (clone CLB-27/1, Caltag Laboratories, Burlingame, CA).
Thawed M. fascicularis PBMC were used to determine the frequency of circulating cells able to produce IFN-γ in response to IPP stimulation (Cairo et al., 2005). High protein binding Immobilon-P membrane 96-well plates (Millipore, Bedford, MA) were coated overnight at 4°C with anti human IFN-γ monoclonal antibody (clone GZ4, Bender MedSystem, San Bruno, CA) diluted in PBS at 5μg/ml. PBMC were plated (105 cells/well in 200μl volume) and stimulated for 24 hours with either medium alone or IPP (15μM) or PHA (10μg/ml) (Remel, Lenexa, KS); no IL-2 was added. The following day, plates were washed and the biotinylated monoclonal antibody anti-IFN-γ (clone 7-B6-1, Mabtech, DiaPharma, West Chester, OH) was added at 5μg/ml in a PBS-0.05% Tween-20–5% FBS solution. After 2 hours incubation, plates were washed and Avidin conjugated HRP (BD Biosciences, San Diego, CA) was added, diluted 1:100 in PBS-10% FBS. After 1 hour incubation, plates were washed again and 100μl of substrate solution (BD Biosciences, San Diego, CA) were added. Plates were then incubated 1 hour, rinsed with water and let air-dry over night. Plate images were resolved and recorded through a Cellular Technology Ltd ELISPOT reader; spots were counted using the instrument software (Immunospot 3.2).
Total RNA was extracted from 1–5×106cells using the RNeasy mini Kit (Qiagen, Valencia, CA), as described by the manufacturer. One μg of total RNA was then converted into cDNA using the reverse transcription system kit (Promega, Madison WI), as described previously 26. Polymerase chain reactions were performed as described 26 using the following primers: oligo Vγ2 (5′ATC AAC GCT GGC AGT CC 3′); oligo Cγ1 (5′GTT GCT CTT CTT TTC TTG CC 3′). PCR products were separated on 1.5% agarose/Tris-acetate-EDTA buffer (TAE) gels containing 0.5μg/ml ethidium bromide (Sigma, St Louis, MO).
PCR products were purified by gel extraction, using QIAquick gel extraction kits (Qiagen, Valencia, CA) according to the manufacturer’s instructions. Purified products were denatured (1 minute at 94°C), then incubated for 30 minutes at 72°C with 2mM MgCl2, 0.2mM dATPs, and 2.5 units of Amplitaq gold (Applied Biosystems, Foster City, CA), then ligated into a pCR2.1 vector (TA cloning kit, Invitrogen, Carlsbad, CA). Ligated vector was transfected into TOP 10F′ competent cells (TA cloning kit, Invitrogen, Carlsbad, CA), and bacterial colonies representing a library of Vγ2 chain sequences were grown overnight on agar plates containing 50μg/ml ampicillin (Sigma, St Louis, MO), 500 μM IPTG (Promega, Madison WI) and 80μg/ml X-Gal (Promega, Madison WI). Colonies containing recombinant plasmids were cultured overnight in LB media and plasmid DNA were purified using the REAL Minipreps DNA purification kit (Qiagen, Valencia, CA). Sequencing reactions were done with a Big Dye v3.1 fluorescent sequencing kit (Applied Biosystems, Foster City, CA), with both M13F and M13R oligonucleotide primers for each sample. Sequences were loaded on an automated sequencer ABI3700 and analyzed using Sequencher (Gene Codes Corporation, Ann Arbor, MI) and MacClade (Sinauer Associates, Sunderland, MA) softwares.
Differences among groups were analyzed by one-way ANOVA test or Student t test. p values ≤ 0.05 were considered significant.
Characterizing the peripheral blood Vγ2 T cell repertoire in healthy Macaca fascicularis. We analyzed the Vγ2 repertoire in six animals by sequencing a representative sample of Vγ2 chain cDNA clones. Macaques in this study were heterogeneous in terms of baseline Jγ1.2 segment usage (frequency of Vγ2 chains containing the Jγ1.2 segment) and Vγ2 population complexity (number of distinct Vγ2 chain sequences divided by the total number of sequences within that sample set) (Table 1). Four of the macaques had a high proportion of Jγ1.2 segments in the Vγ2 population (ranging from 50% to 88.3%). Animals 4024 and 4028 had lower Jγ1.2 segment frequencies (33% and 42%, respectively); 4028 also had high population complexity (0.8). Macaque 4024 had only a single repeated Jγ1.2+ nucleotype (referring to the nucleotide sequence of an individual Vγ2 chain), even though repeated Jγ1.2+ nucleotypes were common among all other animals (not shown).
We next tested whether the Vγ2 repertoire is stable over intervals similar to what we used in the vaccination study. For macaques 4028, 4039 and 4736 we sequenced cDNA libraries from two pre-immunization time points two to four weeks apart. Then we compared the frequency for each sequence in the two sample sets (supplemental data, S1). Clones present at ≥ 6% of the total Vγ2 population were found in both control specimens roughly at the same abundance, suggesting that our sample sizes were sufficient to detect these events accurately. For clones that never reached 6% abundance, we were unsure of our ability to discriminate real changes from fluctuation due to sampling error. This analysis helped establish a cut-off value of 6% abundance for sequences that could be analyzed reliably. Individual Vγ2 nucleotypes that did not exceed 6% in any specimen were not evaluated further.
We assumed that outbred macaques, like human beings 43, would not respond uniformly to subcutaneous BCG immunization and we developed criteria to distinguish responder and non-responder animals. We identified 4 responder macaques among 6 that were inoculated, on the basis of changing Vγ2 T cell frequencies in blood, Interferon-γ (IFN- γ) secretion after in vitro restimulation with IPP and proliferative responses to IPP in vitro (Fig. 1 and supplemental data, S2). During 3 to 7 weeks after the primary inoculation, responder animals (4028, 4039, 4087 and 4736) had significantly higher levels of Vγ2+ T cells in blood (p<0.05) compared to baseline values (Fig. 1A) or to non-responders (data not shown). There were no substantial changes in the profile of memory markers CD45RA, CD95, or CD27 expression on circulating Vγ2+ T cells (data not shown), despite the fact that IFN- γ secretion and proliferative responses to IPP were significantly higher for responder animals (Fig. 1B and 1C). Based on these characterizations, we selected 4 animals with the best Vγ2 T cell response to BCG immunization and used these macaques for our studies of repertoire and its changes following inoculation. The response rate among macaques in our study (4 of 6 or 67%) is nearly identical to the response rate among human beings vaccinated with BCG 43.
A comparison of Vγ2 sequences in the starting repertoire from each animal showed that several clonotypes (Vγ2 chain amino-acid sequences) were present in more than one animal (Table 2). Sequences found repeatedly among different animals are termed “public” in contrast to “private” sequences that are unique to individual animals. These public clonotypes accounted for a large proportion of all Vγ2-Jγ1.2+ sequences in the starting macaque repertoire being 48.6% of the Jγ1.2 sequences in macaque 4028, 62.3% in macaque 4039, 60.2% in macaque 4087, and 79.5% in macaque 4736 (the responder group). Non-responder macaques 4024 and 7646 tended to have fewer public clonotypes in their starting Vγ2-Jγ1.2 repertoire with 24.1% for 4024 and 55.9% for 7647, but they still represented a substantial portion of Vγ2 chains. By combining all Vγ2-Jγ1.2 sequences generated for every animal at all time points (approximately 1600 individual sequences), we defined 25 public clonotypes, some of which were remarkably abundant (Table 2).
The same public clonotypes could be encoded by several discrete nucleotypes. This phenomenon was particularly notable for the most common clonotypes PB1 and PB2 that were encoded by 2 to 6 discrete nucleotide sequences (not shown). Additionally, one of the nucleotypes (…GGGAGGTG CAACAGTTT…) encoding the PB1 clonotype was present in every animal, and it is similar to a “canonical” sequence found commonly in human beings 44, 45. In many cases the same Vγ2 clonotypes were encoded by distinct nucleotypes in different animals. Mechanisms that shaped the Vγ2 repertoire were selecting for T cell receptor function (related to the amino acid sequence) and not for nucleotide sequences.
A representative number of Vγ2 sequences were determined for each responder macaque from PBMC collected before the primary inoculation, after the primary inoculation, and following the booster inoculation. Three responder animals (4736, 4087, 4039) showed only small variations in both Jγ1.2 frequency and population complexity after the primary inoculation (Table 3). After the booster all three showed dramatic drops in Jγ1.2 clone frequency (to around half of the starting levels) that were accompanied by increases in the population complexity (Table 3).
The remaining responder macaque 4028 had an increased Jγ1.2 frequency and a decreased population complexity after both primary and booster inoculations (Table 3). In this macaque, the Jγ1.2 population expanded after both inoculations mainly due to a single clone, whose frequency increased from 4.8% (baseline) to 15.9% (after inoculation) and to 28% (after booster) of the Vγ2 population.
Pre-existing public clonotypes dominate the Vγ2+ T cell response to BCG Most of the clonotypes that were expanded after BCG inoculation were in the public group; private clonotypes represented only a minor part of the response. Consequently, the pool of Jγ1.2 sequences was even more biased to the pre-existing public sequences after BCG inoculation than in naïve animals (Table 4). Overall, these changes show a focusing of the Jγ1.2+ subset after primary BCG inoculation due to preferential expansion of T cell clones expressing public Vγ2 sequences. As these public sequences comprise a substantial fraction of the starting repertoire, their appearance is not uniquely due to BCG immunization.
It is common to use in vitro BCG stimulation for studying γδ T cell responses. We compared the impact on the Vγ2 repertoire of BCG stimulation in vitro with what we observed in macaques. Macaque PBMC (collected before the primary inoculation) were stimulated for 13 days with live BCG (MOI 0.3). We measured Vγ2 expansion by flow cytometry (data not shown) and sequenced approximately 90 Vγ2 cDNA clones from each sample. The Vγ2-Jγ1.2 subset was expanded in every case (4736 had the lowest Jγ1.2 frequency, 71.6%, 4087 the highest, 97.8%) with a consequent decrease in the repertoire complexity (from 0.49 for 4039 to 0.43 for 4736) (Table 3).
Responses in vivo and in vitro were clearly different in terms of the variety of responding clonotypes, their frequencies, and the changes after stimulation (Table 5). In many cases the same clone was present after both in vivo and in vitro exposure, but clones that expanded the most were not always the same in the two studies. For example, a Vγ2 clone from macaque 4028 that was very frequent in the Vγ2-Jγ1.2 pool after either primary (25%) or booster inoculation (37.9%) was only 1% abundant after in vitro stimulation. This pattern was repeated for a number of other clones (Table 5). We also observed low frequency nucleotypes that did not react to BCG in vivo (e.g. 1.1% and 1.2% after primary and booster inoculation, respectively) but were expanded after stimulation in vitro (to 11.7%). Overall, the in vivo responses tended to be oligoclonal, implying a greater selection, while the in vitro responses tended to be polyclonal (Table 5). These data argue that in vitro responses to BCG are a poor model for understanding TCR recognition and the in vivo response to mycobacteria.
Mycobacterium infection in human or nonhuman primates elicits a powerful response within the Vγ2Vδ2 T cell subset 41, 46–48. The magnitude and dynamics of Vγ2Vδ2 T cell responses after primary or booster BCG immunization suggested this was an antigen- specific, memory T cell response 41, 43. However, we showed that the response relies primarily on public Vγ2 chain sequences that exist at high frequency within the starting repertoire of BCG-naïve macaques. These same public Vγ2 chains, positively selected during normal ontogeny, encode the response to other Vγ2Vδ2 T cell antigens including alkylphosphates or tumor cells 49, and are not specific for mycobacteria. The response to BCG is either the result of cross-recognition, where the mycobacterium mimics an antigen encountered previously, or the Vγ2Vδ2 T cell receptor recognizing a host antigen that is modulated or modified during BCG infection. Interestingly, one of the common Vγ2 nucleotypes is the macaque counterpart of a “canonical” sequence that was reported previously in human beings 44, 45.
Our observations on the changes in Vγ2 repertoire showed that control mechanisms exist to avoid monoclonal or polyclonal responses in vivo. Public Vγ2 clonotypes were expanded after BCG priming, and it was reasonable to guess that booster immunization would select again on this population, expand only the most reactive clonotypes and reduce the repertoire complexity. However, booster immunization 5 weeks after the initial dose actually contracted several reactive clonotypes and tended to return the repertoire to its starting condition in 3 of 4 animals studied. This process likely reflects activation-induced cell death of reactive Vγ2Vδ2 T cell clones, because the booster was too soon after the initial immunization. In fact, to test whether the booster immunization would drive clonal depletion or responsive clones, we allowed only 5 weeks between priming and boosting compared to 3 to 5 months in another study 41. For γδ T cells, sensitivity to activation-induced cell death may be an important adaptation that prevents the development of autoimmunity. Since this cell subset reacts to common pathogens and host molecules including IPP, a mechanism for limiting the expansion of individual clonotypes will guard against amplifying the response against self even during the response to infection. These data highlight practical aspects of BCG immunization showing more than 5 weeks is needed between primary and booster immunization to preserve strong Vγ2Vδ2 T cell responses to attenuated mycobacteria.
The role for a public and pre-existing Vγ2 repertoire in the response to mycobacteria raises important questions about how these cells function in human beings at risk for tuberculosis. Tuberculosis rates are linked with the epidemic spread of HIV 50–52, partly because dual infection worsens both diseases 52–54. Part of the mechanism for enhanced disease during dual infection may relate to the HIV-mediated depletion 55 of the same Vγ2-Jγ1.2Vδ2 T cells needed for the protective response to mycobacteria. Further, HIV is not the only infectious disease that alters γδ T cell levels. Malaria is often prevalent in the same countries worst affected by HIV and tuberculosis. Malaria infection in endemic regions leads to expansion of Vγ2Vδ2 T cells 32 while non-endemic severe malaria can deplete the Vγ2-Jγ1.2+ subset 56. These considerations are important to the success of tuberculosis immunization programs. For example, there are few data available about γδ T cells in babies born to HIV+ mothers, even though many of these infants are likely to receive BCG immunization against pediatric tuberculosis. As we consider the interactions of diseases including tuberculosis, malaria and HIV, the study of γδ T cell responses is likely to be of major importance.
Vγ2-Jγ1.2Vδ2 T cells are appropriately classed within acquired immunity as they represent the products of positive selection on a cell population with rearranging T cell receptor genes. Selection and peripheral amplification of public Vγ2-Jγ1.2 clonotypes during early life, creates a subset of cells that comprise between 1 in 200 and 1 in 40 of circulating T lymphocytes in adults. Their sheer abundance and memory phenotype (due to previous selection and positive amplification) ensure a rapid, innate-like response to pathogens like mycobacteria, and promote the development of protective immunity against tuberculosis.
We are indebted to Harry Davis, Katherine Bullock, Brandy Brown and Odell Jones for invaluable assistance with animal studies. We would like to thank Dr Melvin Cohn, Dr Adrian Hayday and Dr Maria Salvato for helpful discussions.
Work was supported by PHS grant AI51212 (C.D.P.) and a contract from the United Nations Education, Science and Cultural Organization (UNESCO) that was supported by the Italian national government (V.C. and C.D.P.).