with Mycobacterial Lipid Antigens Responses in Guinea Pigs by ...

The Journal of Immunology

Induction of CD1-Restricted Immune Responses in Guinea Pigs by Immunization with Mycobacterial Lipid Antigens1 Kenji Hiromatsu,2* Christopher C. Dascher,* Kenneth P. LeClair,† Masahiko Sugita,* Stephen T. Furlong,3* Michael B. Brenner,* and Steven A. Porcelli4* Group 1 CD1 molecules have been shown to present lipid and glycolipid Ags of mycobacteria to human T cells. However, a suitable animal model for the investigation of this component of antimycobacterial immunity has not yet been established. Previously, we found that guinea pigs express multiple isoforms of group 1 CD1 proteins that are homologous to human CD1b and CD1c. In this study, we show that CD1-restricted T cell responses can be generated in guinea pigs following immunization with lipid Ags from Mycobacterium tuberculosis. Splenic T cells from lipid Ag-immunized guinea pigs showed strong proliferative responses to total lipid Ags and partially purified glycolipid fractions from M. tuberculosis. These lipid Ag-reactive T cells were enriched in CD4negative T cell fractions and showed cytotoxic activity against CD1-expressing guinea pig bone marrow-derived dendritic cells pulsed with M. tuberculosis lipid Ags. Using guinea pig cell lines transfected with individual CD1 isoforms as target cells in cytotoxic T cell assays, we found that guinea pig CD1b and CD1c molecules presented M. tuberculosis glycolipid Ags to T cells raised by mycobacterial lipid immunization. These results were confirmed using a T cell line derived from M. tuberculosis lipid Ag-immunized guinea pigs, which also showed CD1-restricted responses and cytolytic activity. Our results demonstrate that CD1-restricted responses against microbial glycolipid Ags can be generated in vivo by specific immunization and provide support for the use of the guinea pig as a relevant small animal model for the study of CD1-restricted immune responses to mycobacterial pathogens. The Journal of Immunology, 2002, 169: 330 –339.

T

he human CD1 gene family consists of five nonpolymorphic genes, CD1A, B, C, D, and E, which map to a locus on chromosome 1 (1). The proteins encoded by this gene family are transmembrane glycoproteins expressed in association with ␤2-microglobulin on the surface of dendritic cells and other APCs. The CD1 proteins show limited but significant homology to MHC class I and II Ag-presenting molecules that present peptide Ags to T cells (2). This homology of structure provided a clue to the possible function of CD1 proteins in the presentation of foreign Ags to T cells. However, CD1 proteins are distinct from MHCencoded molecules by the marked abundance of hydrophobic amino acids in the membrane distal domains. This finding suggested that CD1 proteins might have a different type of Ag binding site than the peptide binding groove found in MHC molecules (2). This prediction has been confirmed by the crystal structure of a mouse CD1 protein, which shows a deep internal cavity formed by

*Division of Rheumatology, Immunology, and Allergy, Brigham and Women’s Hospital, and Harvard Medical School, Boston, MA 02115; and †Antigenics Inc., Woburn, MA 01801 Received for publication February 26, 2002. Accepted for publication April 26, 2002. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. 1 This work was supported by grants from the Irene Diamond Foundation and National Institutes of Health/National Institute of Allergy and Infectious Diseases (AI48933 and AI45889) to S.A.P. and National Institutes of Health/Small Business Innovation Research (AI40798) to K.P.L. K.H. was partially supported by a fellowship from the Medical Foundation. 2 Current address: Department of Microbiology and Immunology, Albert Einstein College of Medicine, Bronx, NY 10461. 3 Current address: Astra-Zeneca Pharmaceuticals, 1800 Concord Pike, Wilmington, DE 19850. 4 Address correspondence and reprint requests to Dr. Steven A. Porcelli at the current address: Department of Microbiology and Immunology, Albert Einstein College of Medicine, Forchheimer Building, Room 416, 1300 Morris Park Avenue, Bronx, NY 10461. E-mail address: [email protected]

Copyright © 2002 by The American Association of Immunologists, Inc.

the ␣1 and ␣2 domains which is lined almost entirely by hydrophobic amino acid side chains (3) The unique structure of the Ag binding groove of CD1 suggested that CD1 binds and presents something other than peptides. Direct evidence now confirms this hypothesis, as several studies have shown that the human group 1 CD1 proteins (i.e., the group composed of CD1a, CD1b, and CD1c) mediate specific T cell recognition of mycobacterial lipid and glycolipid Ags (4). Ags identified to date include mycolic acid, glucose monomycolate, lipoarabinomannan, and other phosphoglycolipids (5– 8). T cells recognizing mycobacterial lipid Ags presented by CD1 can release IFN-␥ and show cytolytic activity upon contact with lipid Agpulsed target cells and mycobacteria-infected monocytes. Furthermore, CD8⫹ CD1-restricted T cells have also been recently shown to mediate a direct microbicidal effect on intracellular microorganisms by granulysin-mediated killing (9). These activities suggest a proinflammatory role for these T cells in type 1 cellular immune responses that are critical for the elimination of intracellular pathogens such as Mycobacterium tuberculosis (MTb)5 and Mycobacterium leprae (10). Whereas previous work establishes the function of CD1 in Ag recognition by T cells, the critical question of the role of this system in host defense against infection in vivo has not been examined. Evaluation of the in vivo role of the CD1 system in the protective immune response to pathogens such as mycobacteria will require the development of a suitable animal model in which such responses are generated and can be studied. In the case of MTb infection, such an animal model would ideally be one that accurately replicates the important immunopathological and clinical features of human tuberculosis and also must have a CD1 system comparable to that which is present in humans. Mice and 5 Abbreviations used in this paper: MTb, Mycobacterium tuberculosis; MPL, monophosphoryl lipid A; BM-DC, bone marrow-derived dendritic cell; DN, double negative.

0022-1767/02/$02.00

The Journal of Immunology rats are probably not suitable animal models in this case because they express only CD1d-like molecules and lack obvious homologs of the human group1 CD1 proteins (i.e., CD1a, b, and c) which are all known to present microbial lipid Ags (4). In contrast, our previous studies suggested that the guinea pig may prove to be an excellent animal model for studying group1 CD1-dependent immune responses in MTb infection (11). We have already reported that the guinea pig genome contains an extended family of CD1 genes, with at least eight distinct CD1 homologs (11). Based on analyses of predicted amino acid sequences, it was possible to classify each guinea pig CD1 gene as a homolog of one specific human CD1 isoform. This revealed that the guinea pig has genes that encode multiple homologs of human CD1b (i.e., guinea pig CD1B1, B2, B3, and B4) and CD1c (CD1C1, C2, and C3), in addition to at least one gene encoding a CD1e-like protein. In the current study, we demonstrate the existence of CD1-restricted nonpeptide lipid Ag-reactive T cells in guinea pigs. These CD1-restricted T cell responses were shown to be induced in vivo following immunization with MTb lipid Ags and led to the generation of cytotoxic T cell responses restricted by several different members of the guinea pig group 1 CD1 family. These findings provide further justification for the use of the guinea pig as a relevant small animal model for study of the in vivo role of CD1-mediated immune responses to lipid Ags in host defense against MTb and other microbial pathogens.

Materials and Methods Animals Outbred female Hartley guinea pigs were purchased from Charles River Breeding Laboratories (Wilmington, MA). Inbred strain 2 guinea pigs were obtained from Tenovus Research Laboratory (Southampton, U.K.) and bred in our animal facility at the Dana-Farber Cancer Institute. Animals were housed under specific pathogen-free conditions and all animal experiments were performed according to the Guidelines on Animal Welfare approved by the committee on animal use at Harvard Medical School and the Brigham and Women’s Hospital.

Monoclonal Abs CT5 (anti-pan guinea pig T cell), CT6 (anti-guinea pig CD8), and CT7 (antiguinea pig CD4) were obtained from Serotec (Raleigh, NC). CD1F2/6B5 (mouse IgG1), reactive with all of the currently known isoforms of guinea pig group 1 CD1, was produced in our laboratories from a hybridoma cell line generated from splenocytes of a mouse hyperimmunized with guinea pig CD1 proteins (12). A hybridoma line producing mAb IVA12 (antihuman MHC class II monomorphic determinant and cross-reactive to guinea pig MHC class II) was obtained from the American Type Culture Collection (Manassas, VA).

Purification of lipid Ags from MTb Lipid Ags were purified as previously described (13). Briefly, total sonicates (10 mg dry weight bacteria/ml) of MTb (strain H37Ra) in PBS were extracted with four volumes of chloroform/methanol (2:1, v/v) according to the method of Folch et al. (14). After phase separation, the organic phase was collected, dried by rotary evaporation, and resuspended in the original starting volume in chloroform for storage. For further subfractionation of MTb total lipids, silica column chromatography was used (5). After evaporation of the solvent from the MTb total lipid extract, the lipids were resuspended in hexane and passed over open silicic acid columns to which all major lipid classes bind. The columns were then eluted with solvents of increasing polarity (chloroform followed by acetone and finally methanol) to elute the major lipid fractions in order of increasing polarity (neutral lipids followed by glycolipids and phospholipids in sequential fashion). After elution from silicic acid columns, all lipid fractions were dried under a nitrogen stream and then redissolved in chloroform for storage. Mycolic acids were released from dried MTb cell walls by saponification and purified using C18 reverse-phase HPLC as previously described (5). The composition of lipid preparations was assessed using analytical TLC with development of TLC plates by charring in sulfuric acid to reveal all lipid species. Further characterization of lipid preparations was conducted using special stains of TLC plates that revealed the presence of carbohydrate (orcinol), amino groups (ninhydrin), and phosphorous (molybdenum blue).

331 In addition, to assess the possibility of protein contaminants, amino acid composition using standard methods of acid hydrolysis followed by reverse-phase HPLC was conducted by the Harvard Microchemistry Facility (Cambridge, MA). Based on this analysis, the maximum possible levels of protein contamination in lipid Ag fractions used in this study were as follows: MTb total lipid, 4.05%; neutral lipid fraction, 0.06%; glycolipid fraction, 0.19%; phospholipid fraction, 6.41%; and mycolic acid, 0.33%. These values most likely represent overestimates of protein content since the majority of amino acids detected (⬃70%) were alanine and glutamate/ glutamine, probably originating from mycobacterial peptidoglycan rather than proteins (15). Consistent with this, silver staining of SDS-PAGE gels (sensitivity of ⬍10 ng for a single protein band) revealed no visible band in a 100-␮g sample of the MTb total lipid Ag preparation, indicating protein contamination of ⬍0.01% for any single protein constituent (data not shown). Each of the lipid fractions was dried under a stream of nitrogen and then resuspended by sonication into complete medium (RPMI 1640; Life Technologies, Grand Island, NY) with 10% FCS (HyClone Laboratories, Logan, UT) and additional supplements as previously described for in vitro use in T cell recognition assays (16).

Liposome preparation for immunization of guinea pigs with MTb lipid Ags Liposomes containing MTb lipid Ags with or without adjuvants were prepared using the dehydration/rehydration method (17, 18). Briefly, 50 mg of 1,2-distearoyl-sn-glycero-3-phosphocholine (Avanti, Alabaster, AL) and 25 mg of cholesterol (Avanti) were dissolved in 5 ml of chloroform and added to a clean round-bottom vacuum flask. For those preparations that contained monophosphoryl lipid A (MPL) as adjuvant, 1 ml of chloroform containing 1 mg of MPL from Salmonella minnesota R595 (Ribi, Hamilton, MT) was added (19). Approximately 2 mg of each of the MTb lipid Ag fractions (glycolipids, neutral lipids, phospholipids, and mycolic acids as indicated) dissolved in chloroform were added, and the entire lipid mixture was dried to a thin film using a Rotovap flash-evaporator (Buchler Instruments, Fort Lee, NJ). Five milliliters of distilled water was then added to the flask, and the lipid was suspended into the solution as liposomes by sonicating in a water bath sonicator (Fisher Scientific, Pittsburgh, PA). For those preparations that contained QS-21 as adjuvant, 150 ␮g of QS-21 (Antigenics, Woburn, MA) was added during the rehydration step. In some experiments, liposomes containing OVA were made by adding soluble OVA (900 ␮g) during the rehydration step. Liposome suspensions were lyophilized for storage at ⫺20°C and immediately before use were resuspended by adding an appropriate volume of PBS and passing three times through a 25-gauge needle.

Immunization of guinea pigs with lipid Ags Specific pathogen-free outbred female Hartley strain guinea pigs (250 –300 g) or inbred strain 2 guinea pigs (250 –300 g) were immunized s.c. with purified lipid Ags of MTb in liposomes with or without adjuvant (QS-21 or MPL). For each immunization, animals received liposomes containing a mixture of some or all of the MTb lipid Ags (glycolipids, neutral lipids, phospholipids, and mycolic acids) as indicated. In each case, the amount of immunogen used per animal contained 500 – 600 ␮g of each lipid component, and the dose of QS-21 or MPL applied as adjuvant was 50 or 300 ␮g per immunization, respectively (19, 20). In some experiments, guinea pigs were immunized with bone marrow-derived dendritic cells (BM-DCs) pulsed with MTb total lipid Ags. BM-DCs were obtained from bone marrow cultures supplemented with human GM-CSF (300 U/ml; PeproTech, Rocky Hill, NJ) as described elsewhere (12). BM-DCs were pulsed overnight with MTb total lipid extract (30 ␮g/ml) and, after washing three times with PBS, Ag-pulsed BM-DCs (2 ⫻ 107 cells total per immunization per animal) were injected s.c. into two or three sites on the abdomen. For CD1-restricted CTL induction (see Figs. 3 and 4), guinea pigs were immunized one, two, or three times, as specified, with a mixture of liposomes reconstituted with MTb lipid Ags (600 ␮g each of glycolipids, neutral lipids, and phospholipids) along with QS-21 (50 ␮g) and MPL (300 ␮g). Animals were sacrificed at the indicated time points, and primary ex vivo CTL responses were analyzed as described below.

T cell proliferation assays T cells were enriched from the spleens of immunized guinea pigs by passage through nylon wool fiber columns. T cells (5–10 ⫻ 104/well) were cultured in complete medium as described in triplicate in 96-well flatbottom microtiter plates with or without irradiated (5000 rad) CD1⫹ BMDCs or autologous nylon wool-adherent spleen cells (1–5 ⫻ 104/well) in the presence or absence of Ag (16). To assess CD1 restriction, purified mAb CD1F2/6B5 (reactive with guinea pig CD1b1, b2, b3, b4, c1, c2, and

332 c3) or isotype-matched nonbinding control mAb (P3, mouse IgG1) was added at a final concentration of 20 ␮g/ml. Cultures were incubated for 3– 4 days at 37°C in a 5% CO2 incubator, pulsed with [3H]thymidine (1 ␮Ci/ well, 6.7 Ci/mmol; NEN Life Science Products, Boston, MA), and incubated for an additional 5– 6 h. The plates were harvested on a Tomtec 96-well plate harvester (Wallac, Turku, Finland) and thymidine incorporation was measured with a Betaplate liquid scintillation counter (Wallac).

Cytotoxic assays Guinea pig cell lines transfected with expression vector constructs encoding the guinea pig CD1b1, b2, b3, b4, c2, or c3 proteins (12) or mocktransfected cells (i.e., stably transfected with empty vector) were labeled with 51Cr for 2 h. The target cells were cultured overnight at 37°C in a 5% CO2 incubator in complete medium with or without MTb total lipid extract (10 or 20 ␮g/ml), and then harvested and extensively washed. The labeled target cells were plated in quadruplicate in 96-well V-bottom microtiter plates with effector T cells at a range of different E:T ratios in a total volume of 150 ␮l of complete medium and incubated at 37oC for 4 h. Samples of supernatant (25 ␮l) were harvested and counted in a gamma counter. Total release of 51Cr was determined by addition of 1% Triton X-100 to wells containing target cells, and spontaneous release was determined from wells in which Ag-pulsed or Ag-unpulsed target cells were cultured without effector T cells. The data were calculated as percent specific lysis according to the following formula: (100 ⫻ (cpm release from experimental sample ⫺ cpm spontaneous release)/(cpm maximum release ⫺ cpm spontaneous release)). For the source of splenic effector T cells for primary CTL responses, guinea pigs (strain 2 or Hartley as specified) were immunized with MTb lipid Ag mixture (600 ␮g each of glycolipids, neutral lipids, and phospholipids) reconstituted in liposomes along with 50 ␮g of QS-21 and 300 ␮g of MPL as adjuvants. Strain 2 guinea pigs immunized once were sacrificed at 2 wk after the primary immunization (S2-005, Fig. 3A) or sacrificed at 4 wk after the primary immunization (S2-007/8, Fig. 3C). In some experiments, strain 2 guinea pigs were immunized twice (primary and boosted at week 4) and sacrificed 12 wk after the second immunization (S2-003, Fig. 3C); Hartley strain guinea pigs were boosted twice after the primary immunization at 4 and 8 wk and sacrificed 24 wk after the last immunization (H0408, Fig. 3C). To prepare effector T cells, spleen T cells from MTb lipid Ag-immunized guinea pigs were enriched by nylon wool columns and then cultured for 6 days in complete medium containing a mixture of MTb total lipid (10 ␮g/ml) and glycolipid fraction (30 ␮g/ml) in the presence of mitomycin C (Sigma-Aldrich, St. Louis, MO)-treated splenic adherent cells as APCs and 0.5 nM human rIL-2. In preliminary experiments, we found that in vitro stimulation with MTb glycolipid Ag in combination with MTb total lipid Ags induced better CTL responses than stimulation with glycolipid Ag alone, possibly due to enhancement of bystander help from CD4⫹ T cells activated by the broader antigenic challenge or some other adjuvant activity of the total lipid preparation. At the end of the 6-day culture period, viable T cells were purified by Ficoll-Hypaque density centrifugation (specific gravity, 1.107) and used as effector T cells in CTL assays with MTb lipid Ag-pulsed CD1-expressing target cells.

Generation of CD1-restricted T cell lines An inbred strain 2 guinea pig was s.c. immunized twice with MTb lipid Ags (mixture of 600 ␮g each of glycolipids, neutral lipids, and phospholipids) in liposomes along with QS-21 (50 ␮g) and MPL (300 ␮g) at 4-wk intervals. The animal was further boosted with 2 ⫻ 107 syngeneic BM-DCs that had been pulsed overnight with MTb total lipid extract (30 ␮g/ml). The spleen was harvested 2 wk after the last immunization and splenic T cells were first enriched using a nylon wool column (Polysciences, Warrington, PA). CD4⫹ T cells were then depleted by staining with anti-CD4 mAb (CT7) followed by magnetic separation using goat anti-mouse Ig-coupled Dynabeads (Dynal Biotech, Lake Success, NY) according to the manufacturer’s instructions. The resulting CD4-depleted T cells were cultured with MTb total lipid extract (20 ␮g/ml) in the presence of irradiated (5000 rad) syngeneic BM-DCs in 1 ml of complete medium (3 ⫻ 106 T cells with 3 ⫻ 105 BM-DCs in 1 ml of medium/well of a 24-well plate). Four days after the start of culture, half of the medium was replaced with fresh complete medium containing 0.25 nM human rIL-2 (PeproTech). Subsequently, cultures were fed every 3– 4 days by replacing half of the medium with fresh complete medium containing 0.25 nM IL-2. Live cells were collected after ⬃2 wk of expansion and restimulated using BM-DCs plus MTb total lipid extract. Cultures were carried through four similar cycles of Ag stimulation (in some cases preceded by immunomagnetic depletion of residual CD4⫹ T cells) followed by expansion in medium containing IL-2 to generate line S2-1031, which showed CD1-restricted reactivity to MTb lipids.

CD1-RESTRICTED T CELL RESPONSES IN GUINEA PIG Statistical analyses Data were analyzed by unpaired t test using GraphPad Prism (GraphPad Software, San Diego, CA).

Results Induction of ex vivo T cell responses in guinea pigs immunized with mycobacterial lipid Ags Our initial experiments focused on the induction of specific T cell proliferative responses following the immunization of guinea pigs with suitable preparations of MTb lipid Ags. To produce uniform suspensions of the lipid Ags in a form that was likely to be immunogenic, we incorporated a mixture of three MTb lipid Ag fractions (glycolipids, neutral lipids, and mycolic acids) into liposomes. To provide a suitable adjuvant, we incorporated either MPL or QS-21 into the immunogen. Four weeks after immunization with the liposome-entrapped MTb lipid Ag mixture with or without adjuvant, guinea pigs were sacrificed and T cells were purified from their spleens by passage over nylon wool columns. The resulting T cell-enriched populations were ⬎95% T cells based on FACS analysis using mAb CT5 and were ⬃70% CD4⫹ (mAb CT7) and 20% CD8⫹ (mAb CT6); data not shown). These cells were tested for proliferative responses to the MTb lipid Ag fractions using irradiated nylon wool column-adherent spleen cells as APCs (60% positive by FACS for CD1 expression with mAb CD1F2/6B5; data not shown). As shown in Fig. 1A, nylon wool column-passed splenic T cells from guinea pigs immunized twice with liposome-entrapped MTb lipid Ags without adjuvant showed very weak proliferative responses to total MTb lipid Ags and no responses to the glycolipid, neutral lipid, or mycolic acid fractions. In contrast, immunization of guinea pigs with liposomes containing MTb lipids plus either QS-21 (Fig. 1B) or MPL (Fig. 1, C and D) as an adjuvant induced significant proliferative responses by spleen T cells to the MTb total lipid and glycolipid preparations. Splenic T cells purified from naive unimmunized guinea pigs showed only minimal, if any, responses to mycobacterial lipid Ags in vitro (Fig. 1D and additional data not shown). We could not raise significant T cell responses to neutral lipids or mycolic acids from these guinea pigs using these immunization protocols, suggesting either that these lipid Ag fractions are less immunogenic or that the frequency of T cells specific to them is very low. In vivo immunization with BM-DCs pulsed with MTb lipid Ags As an alternative method for inducing T cell responses against MTb lipid Ags in vivo, we immunized guinea pigs with lipid Agpulsed DCs without the addition of chemical adjuvants. DCs expressing CD1 molecules were generated from guinea pig bone marrow cells using in vitro cultivation with human rGM-CSF (12). Fig. 2A shows a representative FACS profile of guinea pig BMDCs at day 10 after culture of bone marrow cells. When gated on the high forward scatter population, nearly 90% of the BM in these cultures expressed high levels of MHC class II and CD1 proteins on their surfaces. Given the known properties of DCs in the stimulation of primary T cell responses in vivo in humans and mice, we tested the ability of guinea pig BM-DCs to prime lipid Ag-specific responses following MTb lipid Ag-loading ex vivo and injection back into animals (21). Naive guinea pigs received two s.c. injections 2 wk apart of BM-DCs pulsed with MTb total lipid Ag extract. At 2 wk after the last immunization, proliferative responses of purified splenic T cells against different MTb lipid Ag fractions in the presence of irradiated BM-DCs as APCs were examined (Fig. 2, B and C). As shown in Fig. 2B, an outbred Hartley guinea pig that was immunized with BM-DCs pulsed with MTb lipid Ag showed strong


333

FIGURE 1. Primary ex vivo proliferative T cell responses from guinea pigs immunized with liposome containing MTb lipid Ags. Proliferative responses of splenic T cells from Hartley strain guinea pigs immunized with MTb lipid Ags (mixture of glycolipids, neutral lipids, and mycolic acids) entrapped in liposomes without adjuvant (animal A025, A), with QS-21 (animal A029, B), or MPL (animal A032, C and D). Four weeks after immunization, guinea pigs were sacrificed and spleen T cells were purified and tested for proliferative responses to the indicated lipid Ags in the presence of irradiated autologous splenic nylon wool fiber-adherent cells as APCs. A range of concentrations was tested for each of the lipid Ag fractions, and the results shown in A–C represent the optimal dose for each lipid Ag preparation shown (12.5 ␮g/ml for MTb total lipids, 25 ␮g/ml for all others). D, One representative example of a dose titration of the lipid Ag fractions is shown for the proliferative responses of T cells from a guinea pig immunized with MTb lipid Ags plus MPL in liposomes (animal A032, f) and for T cells from an age-matched naive guinea pig (Œ). Asterisks indicate statistically significant difference compared with medium only control (ⴱ, p ⬍ 0.01; ⴱⴱ, p ⬍ 0.001).

proliferative responses to MTb glycolipid and phospholipid fractions. The high level of background proliferation observed in this case (medium alone) was most likely due to the simultaneous sensitization of alloreactive T cells, since Hartley guinea pigs are not fully inbred. To rectify this potential problem, we conducted analogous experiments using inbred strain 2 guinea pigs as the BM-DC donor and as the recipient immunization with BM-DCs pulsed with MTb lipid Ag. As shown in Fig. 2C, in vivo immunization of an inbred strain 2 guinea pig with MTb lipid Ag-pulsed BM-DCs of strain 2 origin also induced significant proliferative responses of splenic T cells to total MTb-extractable lipids in vitro, as well as a strong response to the phospholipid fraction and a weak but significant response to the glycolipid fraction. In vivo sensitization of strain 2 guinea pigs using MTb lipid Ag-pulsed BM-DCs was repeated three times, showing consistent levels of T cell proliferation to total MTb lipid Ags and phospholipids (stimulation index of 3.2–5.5 compared with medium alone) and a lesser level of proliferation to glycolipids (stimulation index, 1.7–2.6), with minimal or no responses to neutral lipids. In similar experiments conducted in parallel, we also observed that guinea pigs immunized with splenic DCs pulsed with the total MTb lipid Ag extract showed proliferative responses to MTb lipid Ags upon in vitro challenge (data not shown). These results indicated that BM-DCs and splenic DCs expressing CD1 could present mycobacterial lipid

Ags to T cells in vivo, thus leading to lipid Ag-specific recall responses that could be detected by T cell proliferation ex vivo. Ex vivo CTL responses from guinea pigs immunized with mycobacterial lipid Ags In humans, the majority of group 1 CD1-restricted T cell lines reactive with mycobacterial lipid Ags that have been studied to date exhibit cytotoxic activity against lipid Ag-pulsed DCs or CD1-expressing transfectant cells along with proliferative responses and IFN-␥ production. Therefore, we next examined whether the MTb lipid Ag-reactive T cells arising after immunization of guinea pigs showed cytotoxic activity against MTb lipid Ag-pulsed target cells. Based on the results described in Fig. 1, we immunized naive strain 2 or Hartley guinea pigs with MTb lipid Ags (mixture of glycolipid, neutral lipid, and phospholipid fractions) entrapped in liposomes with the combination of QS-21 and MPL as adjuvants. Effector T cells were prepared from immunized guinea pigs by in vitro stimulation of splenic T cells with MTb total lipid extract (10 ␮g/ml) and glycolipid-enriched fraction (30 ␮g/ml) for 6 days. We first tested CTL responses against lipid Ag-pulsed syngeneic strain 2 BM-DCs, which we have shown to express high levels of multiple different guinea pig CD1 proteins (12). As shown in Fig. 3A, the effector T cells derived from the MTb lipid Ag-immunized guinea pigs lysed the target BM-DCs

334

CD1-RESTRICTED T CELL RESPONSES IN GUINEA PIG

FIGURE 2. Primary ex vivo proliferative T cell responses from guinea pigs immunized with MTb lipid Ag-pulsed BM-DCs. A, FACS analysis of BM-DCs obtained from bone marrow cells cultured in the presence of human GM-CSF. Anti-guinea pig MHC class II (mAb IVA12), anti-pan guinea pig CD1 (mAb CD1F2/6B5), and isotype-matched, nonbinding control mAb (mAb P3) were used as primary Abs and detected with a FITC-conjugated goat anti-mouse IgG/M F(ab⬘)2. The histograms shown are representative of FACS profiles obtained on more than 10 preparations of BM-DCs. B, BM-DCs from a naive Hartley guinea pig were pulsed overnight with MTb total lipid extract (30 ␮g/ml). After extensive washing with PBS, the Ag-pulsed BM-DCs were injected s.c. into a naive Hartley guinea pig (day 0). On day 14, the guinea pig was boosted with total MTb lipid Ag-pulsed BM-DCs and on day 28 the animal was sacrificed and T cell proliferative responses to lipid Ag were tested in the presence of the same irradiated BM-DC preparation that was used for the immunizations. C, BM-DCs obtained from naive inbred strain 2 guinea pigs were pulsed with MTb total lipid Ags overnight and injected s.c. into naive strain 2 guinea pigs. The animals were boosted with a second injection of Ag-pulsed syngeneic BM-DCs 2 wk later and then sacrificed after an additional 2 wk. Purified splenic T cells were tested for proliferative responses to MTb lipid Ags in the presence of syngeneic BM-DCs as APCs. Asterisks indicate statistically significant difference compared with medium-only control (ⴱ, p ⬍ 0.01; ⴱⴱ, p ⬍ 0.001).

pulsed with MTb glycolipids or total lipids significantly above the background level of target cells that were not pulsed with MTb lipids. Similar to the proliferative responses observed after MTb lipid Ag immunization using the liposome vehicle with QS-21 or MPL as adjuvant (Fig. 1), effector T cells in cytolytic assays showed the strongest reactivity to the MTb total lipid Ag preparation and the glycolipid-enriched fraction, but also showed a trend toward weak recognition of BM-DCs pulsed with the neutral lipidenriched fraction. Next, we examined these cytotoxic T cell responses for their dependence on the expression of particular CD1 molecules by the target cells. As described elsewhere, we generated a new mAb (CD1F2/6B5) against guinea pig CD1 proteins which recognized all seven of the currently identified isoforms of guinea pig group 1 CD1 proteins (12). As shown in Fig. 3B, the cytolytic activity of effector T cells obtained from lipid Ag-immunized guinea pigs against BM-DCs pulsed with glycolipids or total lipids were significantly inhibited by anti-CD1 mAb, but not by isotype-matched control mAb (P3, mouse IgG1). These results strongly suggested that mycobacterial lipid Ag-reactive cytotoxic T cells elicited in guinea pigs by immunization with MTb lipid Ags were at least partially CD1 restricted. To address which of the multiple different group 1 CD1 proteins of the guinea pig were involved in CTL responses evoked by immunization with MTb lipid Ags, we used a panel of guinea pig cell lines stably transfected with one of six different forms of guinea pig CD1 as target cells in CTL assays. Subclones of the guinea pig cell line 104C1 each transfected with a construct encoding one of six different guinea pig CD1 proteins (i.e., guinea pig CD1b1, b2, b3, b4, c2, or c3) or with empty vector alone (mock transfectant cells) (12) were pulsed with the MTb glycolipid fraction and used as target cells for CTL assay (note that guinea pig CD1c1 was not included in this analysis because we were unable to achieve stable high levels of expression of this isoform by transfecting 104C1 cells). The MTb glycolipid fraction was used for this analysis because this fraction was consistently active in preliminary CTL ex-

periments (Fig. 3A and additional data not shown) and represented a more purified form of lipid Ag without significant protein or peptide contamination. As shown in Fig. 3C, the CTL responses of either inbred strain 2 or Hartley guinea pigs immunized with MTb lipid Ags in liposomes plus MPL and QS-21 showed consistent recognition of MTb glycolipid-pulsed transfectants expressing either guinea pig CD1b1 or guinea pig CD1c2. Note that one animal (Hartley guinea pig H0408) was immunized with lipid Ag three times and sacrificed 6 mo after the final immunization to investigate the long-term memory responses of lipid Ag-reactive CD1restricted CTLs in this system. The appearance of similar levels of CTL activity in this animal that was rested for an extended period of time compared with the animals that had been more recently boosted strongly suggested that a long-term CD1-restricted memory CTL response can be induced following MTb lipid Ag immunization. A more detailed analysis of the effects of key parameters of the CTL reaction (glycolipid Ag concentration used for pulsing targets, and E:T ratio) was conducted for CTL recognition of targets expressing guinea pig CD1b1 and guinea pig CD1c2 (Fig. 4). As shown in Fig. 4, guinea pig CD1b1 and c2 but not mock-transfected 104C1 were recognized by effector T cells from strain 2 guinea pigs immunized twice with lipid Ags in liposomes with QS-21 plus MPL, and the level of lysis varied directly with the MTb glycolipid Ag concentration and the E:T ratio. Furthermore, to exclude the possibility of random variations between different 104C1 transfectant clones accounting for the observed CTL lysis, we analyzed the CTL responses using multiple independently derived subclones of 104C1-expressing guinea pig CD1b1 (two clones: clone 16 and clone 18) or guinea pig CD1c2 (three clones: clone 5, clone 11, and clone 12). All of these subclones behaved similarly as target cells in these assays, yielding specific lysis values that were significantly greater than the background level obtained with the mock transfectant cells (data not shown). Taken together, these results indicated that immunization with MTb lipid Ags elicited MTb glycolipid-specific CTL responses in vivo that


335

FIGURE 3. Primary ex vivo cytotoxic T cell responses to mycobacterial lipid Ags. A and B, CTL activity against MTb lipid Ag-pulsed BM-DCs. A strain 2 guinea pig (S2-005) was immunized with MTb lipid Ags (mixture of glycolipid, neutral lipid, and phospholipid fractions) entrapped in liposomes with QS-21 and MPL. Two weeks after the immunization, splenic T cells were purified and further stimulated with MTb total lipids (10 ␮g/ml) and glycolipidenriched fraction (30 ␮g/ml) for 6 days in the presence of irradiated splenic adherent cells as APCs and subsequently tested for cytotoxic activity against lipid-pulsed BM-DCs as target cells. In the experiment shown in B, mAb CD1F2/6B5 against guinea pig CD1 proteins or isotype-matched control mAb (P3) were tested for inhibition of the CTL responses. Asterisks indicate statistically significant difference compared with medium-only control (A) or to isotype-matched control mAb (B) (ⴱ, p ⫽ 0.035; ⴱⴱ, p ⬍ 0.001). C, CTL responses of guinea pig T cells from MTb lipid Ag-immunized guinea pigs against guinea pig CD1 transfectant cells. Guinea pigs were immunized with MTb lipid Ag (mixture of glycolipid, neutral lipid, and phospholipid fractions) in liposomes along with QS-21 and MPL. Guinea pig fibroblast cell line 104C1 transfected with each isoform of guinea pig CD1 gene were labeled with 51Cr, pulsed with glycolipids, and used as target cells in CTL assay. Effector T cells were prepared by 6 days of in vitro stimulation of splenic T cells with a mixture of MTb glycolipids (30 ␮g/ml) and total lipids (10 ␮g/ml) in the presence of autologous APCs. Effector splenic T cells were from strain 2 guinea pigs immunized once (animal S2007/8, sacrificed 4 wk after a single immunization) or twice with MTb lipids (animal S2-003, sacrificed12 wk after the second immunization) and from a Hartley guinea pig immunized three times (animal H0408, sacrificed 6 mo after the third immunization). Asterisks indicate statistically significant difference compared with lysis of mock transfectant (ⴱ, p ⬍ 0.001).

were restricted by at least two different group 1 guinea pig CD1 proteins. Phenotype of MTb lipid Ag-specific T cells elicited in immunized guinea pigs To gain further insight into the types of CD1-restricted T cells that arise following MTb lipid Ag immunization in vivo, we examined the phenotype of the lipid-reactive T cells with respect to expression of CD4 and CD8 coreceptor molecules. Since the T cells showed CTL activity, we reasoned that at least a fraction of them would belong to the CD8 single-positive fraction or possibly to the CD4⫺CD8⫺ double-negative (DN) fraction which has also been associated with CTL effector activity (13, 22). Thus, to confirm that MTb lipid Ag-reactive T cells reside in the CD4-negative fraction (i.e., among CD8 single-positive or DN T cells), we performed the following experiments. Hartley guinea pigs were s.c. immunized with OVA and a mixture of MTb lipid Ags (glycolip-

ids, neutral lipids, and phospholipids) incorporated into liposomes along with the QS-21 and MPL. OVA was added concomitantly as a control, representing a foreign protein Ag that should elicit peptide-specific CD4⫹ T cell responses that could be used to monitor the effectiveness of the immunization. At 4 wk after immunization, T cells were enriched from the spleens of sacrificed animals by nylon wool column (purity ⬎90%) and then further fractionated by immunomagnetic depletion of CD4⫹ T cells (residual CD4⫹ T cell contamination ⬍5%). The proliferative responses to MTb lipid Ags and OVA of the T cell-enriched fraction and the CD4-depleted T cell-enriched fraction were then examined in the presence of irradiated autologous BM-DCs as APC. As shown in Fig. 5A, splenic T cells isolated from an immunized guinea pig 4 wk after MTb lipid Ag priming showed significant dose-dependent proliferation to MTb total lipid Ags and to OVA. Depletion of CD4⫹ T cells led to a marked reduction in the OVA-specific T cell proliferation, but had little effect or even slightly increased the

336


FIGURE 4. Guinea pig CD1b1 and CD1c2 proteins present glycolipid Ags to T cells from MTb lipid-immunized guinea pigs. Strain 2 guinea pigs immunized as described in Fig. 3 were sacrificed 3 wk after the second immunization and splenic T cells were further stimulated in vitro with total lipid and glycolipid as described. Effector cells were pooled from two identically treated animals (animals S2-009 and S2-010). Two guinea pig CD1b1 transfectant clones and three guinea pig CD1c2 transfectant clones along with mock transfectant cells were used as target cells for CTL assay. Representative results obtained with one clone of each transfectant are shown. Target cells were pulsed with two different concentrations of MTb glycolipidenriched fraction in these CTL assays (50 ␮g/ml (f) or 25 ␮g/ml (Œ)) or with medium only (F).

proliferative responses to the MTb total lipid Ag (Fig. 5B). This suggested that the T cell fraction responding to MTb lipid Ags was mainly or exclusively composed of CD4⫺CD8⫺ and/or CD8⫹ T cells. This finding was confirmed by preliminary experiments designed to determine optimal conditions for establishing CD1-restricted MTb lipid Ag-specific T cell lines using CD4-depleted splenic T cells from inbred strain 2 animals similarly immunized with MTb lipid Ags. As shown in Fig. 5C, CD4-depleted T cell fractions from immunized strain 2 guinea pigs again responded to MTb whole lipid extract and to the glycolipid-enriched Ag fraction. These results provided further support for the conclusion that CD4-negative T cells were involved in the proliferative responses to these Ags. Establishment of a CD1-restricted MTb lipid Ag-reactive T cell line from an immunized guinea pig The ability to detect proliferative and cytotoxic T cell responses to MTb lipid Ags prompted us to establish CD1-restricted MTb lipid Ag-specific T cell lines to facilitate more detailed in vitro analyses. Since the CD4 depletion studies indicated that CD1-restricted T cells primed by MTb lipid immunization of guinea pigs may be predominantly CD4 negative, we used CD4-depleted splenic T cells from strain 2 guinea pigs immunized with liposome-entrapped purified MTb lipid Ags along with QS-21 and MPL and boosted with lipid Ag-pulsed BM-DCs to initiate cultured T cell lines. These T cells were cultivated in the presence of CD1⫹ BMDCs from inbred strain 2 guinea pigs and pulsed with an MTb whole lipid extract. To prevent possible overgrowth of any residual CD4⫹ T cells during the culture, the CD4 depletion was repeated three times during the course of culture. After two stimulations using syngeneic CD1⫹ BM-DCs, the third and fourth stimulations were conducted with CD1⫹ BM-DCs from allogeneic outbred Hartley guinea pigs to favor presentation of lipid Ags by nonpolymorphic Ag-presenting molecules such as CD1. After the sixth restimulation, one T cell line was obtained (designated as S2-1031) that had expanded sufficiently to permit detailed analysis. FACS analysis revealed that line S2-1031 was composed predominantly of CD4⫺CD8⫺ and CD8⫹ T cells, with only 5.2% CD4⫹ cells (Fig. 6A). The specificity and restriction of T cell line S2-1031 was examined by [3H]thymidine incorporation assay and mAb blocking. The T cell line proliferated strongly to CD1⫹ BM-DCs pulsed with MTb total lipid Ags. Attempts to characterize the specific lipid Ag recognized by line S2-1031 were unsuccessful, as this T cell line

responded to MTb total lipid extract but not to any of the standard subfractions derived from it (i.e., purified mycolic acids or silicic acid column fractions enriched in MTb glycolipids, neutral lipids, or phospholipids; data not shown). The CD1 restriction of this proliferative response to MTb lipids was shown by adding the CD1-specific blocking mAb (CD1F2/6B5.1) to the proliferation assay. The mAb (CD1F2/6B5.1) blocked the MTb lipid Ag-specific response of line S2-1031 by ⬎90%, whereas the same mAb did not affect the proliferative responses of an alloreactive CD4⫹ guinea pig T cell line (Fig. 6B). Isotype-matched control mAb (mAb P3, mouse IgG1) had no effect on these proliferative responses (data not shown). To further dissect the CD1 restriction of this MTb lipid Agreactive T cell line, CTL responses were determined using transfectants of cell line 104C1 that selectively expressed each of the cloned guinea pig CD1 isoforms. T cell line S2-1031 lysed MTb lipid Ag-pulsed 104C1 cells stably transfected with guinea pig CD1b1, -b2, and -c3, but not mock transfectants in an Ag dosedependent manner, as shown in Fig. 7. Consistent with the findings obtained in the proliferation assays with this T cell line, we could not clearly detect significant CTL responses (i.e., ⬎10% specific lysis) to more highly purified lipid Ag fractions (purified mycolates or the MTb glycolipid, neutral lipid, and phospholipid-enriched fraction; data not shown). Thus, although the precise lipid Ag specificity of line S2-1031 cannot yet be determined, the results provided clear evidence for the restriction of its MTb-specific CTL activity by at least three members of the guinea pig group 1 CD1 family.

Discussion To test the possible role of group 1 CD1 molecules in host defense, it is necessary to develop an appropriate animal model for assessment of the functions of group 1 CD1-restricted T cell responses in vivo. Mice and rats are inadequate for this purpose since the genomes of these rodents lack all group 1 CD1 genes. As previously reported, we have analyzed the guinea pig CD1 gene family and proposed that guinea pigs could serve as a suitable small animal model to test the role of group 1 CD1 proteins in host defense (11). In the current study, we have extended our previous findings on this animal model by demonstrating that guinea pigs can mount CD1-restricted T cell responses following in vivo immunization with MTb lipid Ags. This is the first demonstration to our knowledge of group 1 CD1-restricted lipid Ag-specific T cell responses in a nonhuman species and also the first successful demonstration


337

FIGURE 5. Enrichment of MTb lipid Ag-specific cells in CD4-depleted T cell fractions. A, Proliferative responses over a range of Ag doses to OVA (Œ) or MTb total lipids (f) of unfractionated splenic T cells from a Hartley guinea pig immunized with a mixture of MTb lipid Ags and OVA in liposomes with QS-21 plus MPL. APCs were irradiated autologous nylon wool-adherent splenocytes. B, Effect of CD4 T cell depletion on reactivity to MTb lipid Ags and OVA. A Hartley strain guinea pig was immunized with MTb lipid Ags (mixture of glycolipid, neutral lipid, and phospholipid fractions) with OVA in liposomes along with QS-21 and MPL. The animal was sacrificed 4 wk after immunization, and splenic T cells were purified and used as responders in proliferation assays with irradiated autologous splenic nylon wool-adherent cells as APCs. Some portions of splenic T cells were further treated with anti-CD4 mAb, followed by depletion of CD4⫹ cells using magnetic beads. Whole splenic T cells and CD4-depleted splenic T cells were compared for reactivity to 12.5 ␮g/ml MTb total lipid Ag extract or OVA. C, CD4-depleted splenic T cells obtained from inbred strain 2 guinea pigs immunized as in B were tested for proliferative responses to MTb total lipid extract (12.5 ␮g/ml) or glycolipid-enriched fraction (12.5 ␮g/ml) using irradiated syngeneic splenic nylon wool-adherent cells as APCs. Two representative experiments obtained from separate individual guinea pigs are shown. Asterisks indicate statistically significant difference compared with medium-only control (ⴱ, p ⬍ 0.001).

of CD1-restricted recall responses following in vivo immunization with lipid Ags. Evidence supporting the hypothesis that MTb lipid Ag-specific CD1-restricted T cells play a role in host defense against microbial pathogens comes from the recent finding that human CD1c-restricted T cells specific for a mycobacterial isoprenoid glycolipid Ag are seen in individuals previously infected with MTb, but not in naive donors (8). CD1-restricted T cells specific for mycobacterial Ags have also been derived from the active skin lesions of leprosy patients, and effective cell-mediated immunity is correlated with expression of group 1 CD1 molecules by the DCs in leprosy skin lesions (23, 24). Furthermore, these CD1-restricted mycobacterial lipid Ag-specific T cells have been found to be biased toward secretion of Th1 type cytokines and can kill CD1⫹ target cells pulsed with lipid Ags while concurrently producing the bactericidal protein granulysin (25). Together, these data provide strong circumstantial evidence that CD1-restricted T cells specific for foreign lipid Ags may play a role in host defense against bacterial infections. In humans, CD1-restricted responses were initially found among CD4⫺CD8⫺ DN T cells, although subsequent reports also con-

firmed the existence of CD1-restricted T cells that were CD4 (23) or CD8 single positive (13). In our analysis of MTb lipid-immunized guinea pigs, we found that CD1-restricted lipid Ag-specific T cells were prominent in the CD4-negative fraction of splenic T cells, consistent with CD1 restriction being mainly a feature of CD8⫹ or DN T cells in this experimental setting. These CD4negative CD1-restricted T cells were associated with both proliferative and cytotoxic activities, similar to what has been described for human CD1-restricted T cells reactive with mycobacterial lipid Ags. In the case of human MTb lipid-reactive T cells, Ag recognition has typically been associated with production of IFN-␥, a key cytokine in the immune response to pathogenic mycobacteria. We are not yet able to test whether this is a property of the CD1restricted T cells that develop after MTb lipid immunization in the guinea pig because of the current lack of reagents for the specific detection of IFN-␥ or other cytokines in this species. Although our current study established the priming of Ag-specific T cell responses following immunization with suitably adjuvanted and reconstituted mycobacterial lipid preparations, it is also of interest to determine whether such responses develop during

338


FIGURE 6. Generation of a CD1-restricted T cell line specific for MTb lipid Ag from an immunized guinea pig. A, FACS analysis showing phenotype of lipid Ag-specific guinea pig T cell line S2-1031. The T cell line was stained with anti-guinea pig pan-T cell (mAb CT5), anti-guinea pig CD4 (mAb CT7), anti-guinea pig CD8 (mAb CT6), or isotype-matched control Ab (P3). B, CD1 restriction of T cell line S2-1031. T cells were stimulated with MTb total lipid Ags in the presence of syngeneic BM-DCs. Anti-CD1 mAb (CD1F2/6B5) was added to a final concentration of 20 ␮g/ml. An alloreactive CD4⫹ T cell line derived from a Hartley guinea pig by stimulating with BM-DCs from a strain 2 guinea pig was also stimulated with strain 2 BM-DCs with or without anti-CD1 mAb. Results shown are representative of three independent experiments. Asterisk indicates statistically significant difference (ⴱ, p ⫽ 0.0066).

actual infections with mycobacterial pathogens. Our previous studies on PBL from human donors with previous mycobacterial infection strongly support the view that CD1-restricted T cells responses to specific lipid Ags do develop as a result of natural infections with MTb (8). In the guinea pig model, preliminary

FIGURE 7. Restriction of MTb lipid Ag-specific T cell line by individual guinea pig CD1 molecules. Guinea pig fibroblast cell line 104C1 cells transfected with different isoforms of guinea pig CD1 were used as target cells in CTL assay. Transfectant cells were first labeled with 51Cr and pulsed for 18 h with either 10 or 20 ␮g/ml MTb total lipid Ags or in medium alone. The CD1-restricted lipid Ag-specific T cell line S2-1031 was used for the effector cells. CTL activity at an E:T ratio of 40:1 is shown. Asterisks indicate statistically significant difference compared with lysis of cells not pulsed with MTb lipid Ag (ⴱ, p ⬍ 0.05; ⴱⴱ, p ⬍ 0.01).

experiments to determine whether such T cell responses are primed in animals infected with MTb strain H37Ra or bacillus Calmette-Gue´ rin have demonstrated lymphocyte proliferative responses to mycobacterial lipid Ags similar to those elicited upon immunization with MTb lipids (K. Hiromatsu, unpublished data). However, we have thus far failed to detect CD1-restricted CTL activity in lymphocytes isolated from guinea pigs infected with MTb strain H37Ra. This raises the possibility that active mycobacterial infection may be associated with suppression of certain aspects of the immune response that can be elicited by isolated lipid vaccination, possibly reflecting a significant immune evasion strategy of the pathogen, and suggests that vaccination by live attenuated mycobacteria could actually be less effective than vaccination with purified subunit formulations in some cases. This is a potentially important issue and its resolution will require extensive further studies using the guinea pig model and other animal models of MTb infection. The conservation of multiple homologs of human CD1b and CD1c in the guinea pig strongly suggests that the group 1 CD1 molecules may have evolved specialized functions early in the course of mammalian evolution (11). The guinea pig model may also provide additional insights into the selective pressures that may have led to the diversification of CD1 into its multiple distinct isoforms. From studies of the subcellular localization and Ag-presenting properties of human CD1a, b, and c molecules, it has been postulated that these separate CD1 isoforms may each be specialized to efficiently survey different intracellular compartments for lipid Ags (26 –28). If this is correct, then it would appear that the different isoforms of group 1 CD1 are not functionally redundant, but are specialized to complement each other in providing detection of foreign or abnormal self-lipids. Interestingly, in our studies of the subcellular localization of the various guinea pig CD1 isoforms, we have found that guinea pig CD1 proteins show at least

The Journal of Immunology three distinct patterns of intracellular localization (12). Given our current findings on the ability of several guinea pig CD1 proteins to present MTb lipid Ags to T cells, it will be of interest to determine in future studies whether the different guinea pig CD1 molecules involved in these responses acquire their lipid Ags in distinct subcellular compartments of APCs. Considerable work remains to be done to identify the specific MTb lipids that are recognized by guinea pig CD1-restricted T cells. As shown in the current study, MTb lipid Ag-immunized guinea pigs frequently gave positive T cell responses to the MTb glycolipid-enriched fraction after immunization with a broad mixture of mycobacterial lipid Ags, whereas highly purified mycolic acids and a neutral lipid-enriched fraction appeared not to be well recognized in most cases. Analysis of amino acid content of the various lipid preparations that we have used does indicate a possibility of low levels of residual protein contamination in some of these, since up to 6.4% amino acid content on a weight basis was detected in the total lipid extract and phospholipid-enriched fractions. However, analysis by SDS-PAGE and ultrasensitive silver staining failed to reveal any detectable protein contaminants in our lipid Ag preparations, and the profile of amino acids present suggest that most of these are derived from peptidoglycan rather than from proteins. Thus, although we cannot completely rule out at this point the possibility that trace amounts of highly immunogenic protein Ags in our lipid preparations could contribute to some of the T cell responses we have observed, this appears to be unlikely. It is also noteworthy that in several cases we demonstrated guinea pig CD1-restricted recognition of the MTb glycolipid-enriched fraction, which represents material eluted with acetone from a silicic acid column and contains only trace amounts of detectable amino acids. Finally, the restriction of these responses by guinea pig CD1 proteins provides further evidence in favor of lipid rather than protein Ag targets, since work on human and murine CD1 has strongly linked these proteins to the binding and presentation of lipids rather than proteins or peptides. Our current data provide further support for the possibility of using the guinea pig as an animal model in which to test the hypothesis that augmentation of group 1 CD1-restricted mycobacterial lipid Ag-specific T cell responses can be used as a component of an effective vaccination strategy against MTb infection (29). This appears to us to be an appealing strategy, given that CD1 genes are largely nonpolymorphic and that the CD1 target ligands (i.e, complex lipids and glycolipids) are essential components of bacterial cells that may be less variable in their structures than the more highly mutable peptide Ags that are targets of MHC-restricted T cells. Currently, Mycobacterium bovis bacillus Calmette-Gue´ rin is the only available vaccine against tuberculosis, and its efficacy had been disputed despite extensive use in humans and testing in the field. CD1-mediated presentation of lipid Ags may provide a completely new vaccination strategy that could offer advantages to new protein subunit vaccines or live attenuated mycobacterial vaccines that are currently under development. Thus, the guinea pig model may provide valuable opportunities for testing the efficacy of lipid Ag vaccination in a small animal model that includes the potential for a prominent group 1 CD1-restricted T cell response.

Acknowledgments We thank Daphney Frederique for technical assistance in preparing liposomes, Gerald Watts for purification of mAb, and William Lane at the Harvard Microchemistry Facility for performing analyses of amino acid content of lipid preparations.

339

References 1. Albertson, D. G., R. Fishpool, P. Sherrington, E. Nacheva, and C. Milstein. 1988. Sensitive and high resolution in situ hybridization to human chromosomes using biotin labelled probes: assignment of the human thymocyte CD1 antigen genes to chromosome 1. EMBO J. 7:2801. 2. Porcelli, S. 1995. The CD1 family: A third lineage of antigen presenting molecules. Adv. Immunol. 59:1. 3. Zeng, Z., A. R. Castano, B. W. Segelke, E. A. Stura, P. A. Peterson, and I. A. Wilson. 1997. Crystal structure of mouse CD1: An MHC-like fold with a large hydrophobic binding groove. Science 277:339. 4. Porcelli, S. A., and R. L. Modlin. 1999. The CD1 system: antigen-presenting molecules for T cell recognition of lipids and glycolipids. Annu. Rev. Immunol. 17:297. 5. Beckman, E. M., S. A. Porcelli, C. T. Morita, S. M. Behar, S. T. Furlong, and M. B. Brenner. 1994. Recognition of a lipid antigen by CD1-restricted ␣␤⫹ T cells. Nature 372:691. 6. Moody, D. B., B. B. Reinhold, M. R. Guy, E. M. Beckman, D. E. Frederique, S. T. Furlong, S. Ye, V. N. Reinhold, P. A. Sieling, R. L. Modlin, et al. 1997. Structural requirements for glycolipid antigen recognition by CD1b-restricted T cells. Science 278:283. 7. Sieling, P. A., D. Chatterjee, S. A. Porcelli, T. I. Prigozy, R. J. Mazzaccaro, T. Soriano, B. R. Bloom, M. B. Brenner, M. Kronenberg, P. J. Brennan, and R. L. Modlin. 1995. CD1-restricted T cell recognition of microbial lipoglycan antigens. Science 269:227. 8. Moody, D. B., T. Ulrichs, W. Muhlecker, D. C. Young, S. S. Gurcha, E. Grant, J. P. Rosat, M. B. Brenner, C. E. Costello, G. S. Besra, and S. A. Porcelli. 2000. CD1c-mediated T-cell recognition of isoprenoid glycolipids in Mycobacterium tuberculosis infection. Nature 404:884. 9. Stenger, S., D. A. Hanson, R. Teitelbaum, P. Dewan, K. R. Niazi, C. J. Froelich, T. Ganz, S. Thoma-Uszynski, A. Melian, C. Bogdan, et al. 1998. An antimicrobial activity of cytolytic T cells mediated by granulysin. Science 282:121. 10. Kaufmann, S. H. 1999. Cell-mediated immunity: dealing a direct blow to pathogens. Curr. Biol. 9:R97. 11. Dascher, C. C., K. Hiromatsu, J. W. Naylor, P. P. Brauer, K. A. Brown, J. R. Storey, S. M. Behar, E. S. Kawasaki, S. A. Porcelli, M. B. Brenner, and K. P. LeClair. 1999. Conservation of a CD1 multigene family in the guinea pig. J. Immunol. 163:5478. 12. Hiromatsu, K., C. C. Dascher, M. Sugita, C. Gingrich-Baker, S. Behar, K. LeClair, M. B. Brenner, and S. A. Porcelli. 2002. Characterization of guineapig group1 CD1 proteins. Immunology. In press. 13. Rosat, J. P., E. P. Grant, E. M. Beckman, C. C. Dascher, P. A. Sieling, D. E. Frederique, R. L. Modlin, S. A. Porcelli, S. T. Furlong, and M. B. Brenner. 1999. CD1-Restricted microbial lipid antigen-specific recognition found in the CD8⫹ ␣␤ T cell pool. J. Immunol. 162:366. 14. Folch, J., M. Lees, and G. H. Sloane Stanley. 1957. A simple method for the isolation and purification of total lipids from animal tissues. J. Biol. Chem. 226:497. 15. Goren, M. B. 1982. Immunoreactive substances of mycobacteria. Am. Rev. Respir. Dis. 125:50. 16. Morita, C. T., S. Verma, P. Aparicio, C. Martinez-A, H. Spits, and M. B. Brenner. 1991. Functionally distinct subsets of human ␥/␦ T cells. Eur. J. Immunol. 21:2999. 17. New, R. R.C. 1990. Preparation of liposomes. In Liposomes: A Practical Approach. R. R. C. New, ed. Oxford Univ. Press, New York, pp. 33–104. 18. Gregoriadis, G., B. McCormack, M. Obrenovic, R. Saffie, B. Zadi, and Y. Perrie. 1999. Vaccine entrapment in liposomes. Methods 19:156. 19. Zhou, F., and L. Huang. 1993. Monophosphoryl lipid A enhances specific CTL induction by a soluble protein antigen entrapped in liposomes. Vaccine 11:1139. 20. Powell, M. F., D. J. Eastman, A. Lim, C. Lucas, M. Peterson, J. Vennari, R. P. Weissburg, T. Wrin, C. R. Kensil, M. J. Newman, et al. 1995. Effect of adjuvants on immunogenicity of MN recombinant glycoprotein 120 in guinea pigs. AIDS Res. Hum. Retroviruses 11:203. 21. Inaba, K., J. P. Metlay, M. T. Crowley, and R. M. Steinman. 1990. Dendritic cells pulsed with protein antigens in vitro can prime antigen-specific, MHC-restricted T cells in situ. J. Exp. Med. 172:631. 22. Porcelli, S., C. T. Morita, and M. B. Brenner. 1992. CD1b restricts the response of human CD4⫺8⫺ T lymphocytes to a microbial antigen. Nature 360:593. 23. Sieling, P. A., M. T. Ochoa, D. Jullien, D. S. Leslie, S. Sabet, J. P. Rosat, A. E. Burdick, T. H. Rea, M. B. Brenner, S. A. Porcelli, and R. L. Modlin. 2000. Evidence for human CD4⫹ T cells in the CD1-restricted repertoire: derivation of mycobacteria-reactive T cells from leprosy lesions. J. Immunol. 164:4790. 24. Sieling, P. A., D. Jullien, M. Dahlem, T. F. Tedder, T. H. Rea, R. L. Modlin, and S. A. Porcelli. 1999. CD1 expression by dendritic cells in human leprosy lesions: correlation with effective host immunity. J. Immunol. 162:1851. 25. Stenger, S., R. J. Mazzaccaro, K. Uyemura, S. Cho, P. F. Barnes, J. P. Rosat, A. Sette, M. B. Brenner, S. A. Porcelli, B. R. Bloom, and R. L. Modlin. 1997. Differential effects of cytolytic T cell subsets on intracellular infection. Science 276:1684. 26. Sugita, M., R. M. Jackman, E. van Donselaar, S. M. Behar, R. A. Rogers, P. J. Peters, M. B. Brenner, and S. A. Porcelli. 1996. Cytoplasmic tail-dependent localization of CD1b antigen-presenting molecules to MIICs. Science 273:349. 27. Sugita, M., E. P. Grant, E. van Donselaar, V. W. Hsu, R. A. Rogers, P. J. Peters, and M. B. Brenner. 1999. Separate pathways for antigen presentation by CD1 molecules. Immunity 11:743. 28. Briken, V., D. B. Moody, and S. A. Porcelli. 2000. Diversification of CD1 proteins: sampling the lipid content of different cellular compartments. Semin. Immunol. 12:517. 29. Porcelli, S. A., and R. L. Modlin. 1995. CD1 and the expanding universe of T cell antigens. J. Immunol. 155:3709.