Expression of Toll-Like Receptor 2 on γδ T Cells Bearing Invariant Vγ6/Vδ1 Induced by Escherichia coli Infection in Mice¹

Most microorganisms are detected and destroyed within hours by innate immunity that preexists and is not Ag specific. Cells of innate immunity such as macrophages discriminate between self and nonself by receptors that identify molecules synthesized exclusively by microbes. LPS, a characteristic component found on the outer membrane of Gram-negative bacteria (1), is one of the most ideal targets for innate immunity. LPS typically consists of the polysaccharide region covalently bound to the lipid region, termed lipid A. Lipid A is generally regarded as a target for the LPS receptor and consequently as the bioactive center of LPS (2, 3, 4, 5, 6). CD14 Ag has been widely recognized as an LPS signaling receptor for immune cells (3, 7), and, until now, several classes of molecules on leukocytes have been recognized as receptors for LPS, such as CD11/CD18 integrins (3, 8, 9, 10, 11), P-selectin (12), and L-selectin (13). However, CD14 lacks a transmembrane domain (3, 7), and cytoplasmic domains of CD11/CD18 integrins do not appear to be necessary for signaling translocation of NF-κB in response to LPS binding (9). Thus, these receptors may function to transfer LPS to a second receptor that transduces the signal.

Several lines of evidence suggest that the Toll-like receptor (TLR)³ family is the cell-surface receptor for LPS, the prototypical activator of NF-κB and other proinflammatory responses (14, 15, 16, 17). Toll was first identified as a protein controlling dorsoventral pattern formation in the early development of Drosophila and was shown to participate in anti-microbial immune responses (18, 19). Recently, several mammalian Toll homologues have been identified (18, 19, 20, 21, 22). One of the human Toll homologues, TLR2, has been shown to be involved in LPS signaling (14, 15, 16). In mice, there is evidence for a missense mutation in the cytoplasmic domain of TLR4 in C3H/HeJ mice exhibiting impaired ability to respond to LPS (23, 24), strongly suggesting that TLR4 is the dominant receptor for at least some types of LPS. This was confirmed by experiments using TLR4 gene-knockout mice (25). More recently, it has been suggested that TLR2 functions not only as an LPS signal transducer (14, 16, 26) but also as a receptor for bacterial lipoproteins from Mycobacteria or Gram-positive bacteria (27, 28, 29).

Based on the type of TCR they express, T lymphocytes can be divided into two major groups, αβ and γδ T cells. γδ T cells are further divided into subsets, based on their expression of certain γ- and δ-chains and their prevalence in certain tissues. Most of these subsets bear, as do αβ T cells, junctionally diverse TCRs, but two γδ T cell subsets in the mouse bear invariant TCRs. These include the Vγ5/Vδ1 subset in skin and the Vγ6/Vδ1 subset that comprises most of the γδ T cells in the female reproductive tract (30, 31, 32). Under normal circumstances, these two subsets bear truly invariant TCRs, even at the nucleotide level in the TCR gene junction. These canonical sequences are very simple, with no apparent N-region contribution. Such characteristics have led to the hypothesis that γδ T cells represent a more primitive, early line of cellular defense, preprogrammed to recognize a limited set of Ags.

We and others (33, 34, 35) previously reported that i.p. infection of mice with E. coli induced a marked increase in γδ T cells in the peritoneal cavity (33, 34, 36) and that the γδ T cells had a protective role against the infection (37). In the present study, we focused on the responsiveness of γδ T cells to native lipid A. Our results demonstrated that the purified γδ T cells in both C3H/HeN and C3H/HeJ mice responded to native lipid A from not only E. coli but also Porphyromonas gingivalis in a TCR-independent manner. The LPS/lipid A-reactive γδ T cells, which used a canonical TCR repertoire encoded by Vγ6-Jγ1/Vδ1-Dδ2-Jδ2 gene segments, strongly expressed TLR2 mRNA. A TLR2 antisense oligonucleotide significantly inhibited the proliferation of γδ T cells in response to the native lipid A. TLR2-deficient mice showed an impaired increase of the γδ T cells following the injection of native lipid A. These results suggest that the canonical Vγ6/Vδ1 γδ T cells respond to natural products from E. coli via TLR2. The implications of these findings for the mechanisms whereby a significant fraction of γδ T cells are activated during E. coli infection are discussed.

C3H/HeN and C3H/HeJ mice were purchased from Japan SLC (Shizuoka, Japan). These mice were bred in our institute under specific pathogen-free conditions. Eight- to 10-wk-old female mice were used for the experiments. The mutant mouse (F₂ interbred from 129/Ola × C57BL/6) strain deficient in TLR2 was generated by gene targeting, as described previously (29). Age- and sex-matched groups of TLR2-deficient (TLR2^−/−) mice and their littermate (TLR2^+/−) mice were used for the experiments. E. coli (no. 26; American Type Culture Collection, Manassas, VA) grown in a brain-heart infusion broth (Difco Laboratories, Detroit, MI) was washed repeatedly, resuspended in PBS, and stored at −70°C in small aliquots until use. The concentration of bacteria was quantitated by plate counts.

Biotin-conjugated anti-CD3ε mAb, FITC-conjugated anti-TCR αβ mAb, PE-conjugated anti-TCR γδ mAb, purified rat anti-mouse CD11a mAb, PE-conjugated anti-mouse CD11b mAb, biotin-conjugated anti-mouse CD11c mAb, purified rat anti-mouse CD14 mAb, and PE-conjugated anti-rat IgG mAb were purchased from PharMingen (San Diego, CA). Red-613-conjugated streptavidin was purchased from Life Technologies (Gaithersburg, MD). Murine anti-TCR γδ (UC7–13D) mAb was obtained by growing hybridoma cells in serum-free medium (medium 101; Nissui Pharmaceutical, Tokyo, Japan) and collecting the supernatant. The Ab was then concentrated and purified by 50% ammonium sulfate precipitation. The purity of the preparation was confirmed by SDS-PAGE, and the concentration of Ab was determined by the Lowry method. The mAbs, diluted to 1 mg/ml in PBS, were stored at −70°C until use. LPS (E. coli, O26/B6) and lipid A from E. coli (F585 Rd mutant) were obtained from Sigma (St. Louis, MO). Lipid A from Salmonella minnesota (R595 Rd mutant) was obtained from List Biological Laboratories (Campbell, CA). Lipid A from P. gingivalis was prepared as described (6). An E. coli-type synthetic lipid A analogue with low toxicity (ONO-4007) was kindly provided by Ono Chemical (Osaka, Japan) (38, 39). LPS was dissolved in pyrogen-free water at the concentration of 1 mg/ml. Lipid A was dissolved at a concentration of 2 mg/ml in 0.1% (v/v) triethylamine aqueous solution. The solution was appropriately diluted with pyrogen-free PBS or culture medium before use for assay.

Cell lines were grown as adherent monolayers in tissue culture dishes at 37°C in 5% CO₂ with 95% air and passaged twice a week to maintain logarithmic growth. The J774A.1 cell line was obtained from the American Type Culture Collection. The cells growing as monolayers in tissue culture dishes were detached from the surface and washed twice with HBSS before experiments.

Mice were i.p. inoculated with E. coli at a dose of 1×10⁸ CFU/mouse (one-fifth the 50% lethal dose), LPS, or lipid A in 1.0 ml PBS on day 0. Peritoneal exduate cells (PEC) were harvested on day 3 after inoculation by centrifugation at 110 × g for 5 min, washed twice, and resuspended at optimal concentrations in RPMI 1640 medium (Life Technologies) supplemented with 10% serum. Smear specimens for differential counts were stained with Giemsa solution. PEC were spread on plastic plates and incubated for 1 h in a CO₂ incubator at 37°C to obtain nonadherent cells.

The γδ T cells were purified by cell sorting using a FACSVantage (Becton Dickinson, San Jose, CA) electric cell sorter from the plastic nonadherent cells on day 3 after E. coli or lipid A injection. The purity of sorted cells was >99% (data not shown).

For three-color analysis, plastic-nonadherent cells of PEC were incubated with saturating amounts of biotin-conjugated and purified Abs for 30 min at 4°C. Cells were washed twice and incubated with FITC-, PE-, and Red-613-conjugated secondary Abs for 30 min. Cells were analyzed with a FACSCalibur flow cytometer (Becton Dickinson). The cells were carefully gated by forward and side light scattering for live lymphocytes. The data were analyzed with FACSCalibur research software (Becton Dickinson).

Total RNA was extracted by the acid-guanidium-phenol-chloroform method from γδ T cells purified by cell sorting. cDNA synthesis and PCR were performed using a cDNA cycle kit (Invitrogen, San Diego, CA). RNA was primed either with 20 pmol of γ-chain C region (Cγ) primers (5′-CTTATGGAGGATTTGTTTCAGC-3′) or 6.7 pmol of δ-chain J region (Jδ) primers (5′-TTGGTTCCACAGTCACTTGG-3′) in 20-μl reaction mixtures for reverse transcription. The PCR was performed on a PCR thermal cycler (Takara, Tokyo, Japan). PCR cycles were run for 30 s at 94°C, 30 s at 54°C, and 30 s at 72°C. Before the first cycle, a denaturation step for 7 min at 94°C was included, and after 35 cycles the extension was prolonged for 4 min at 72°C. The 5′ V primers are as follows: Vγ1/2, 5′-ACACAGCTATACATTGGTAC-3′; Vγ2, 5′-CGGCAAAAAACAAATCAACAG-3′; Vγ4, 5′-TGTCCTTGCAACCCCTACCC-3′; Vγ5, 5′-TGTGCACTGGTACCAACTGA-3′; Vγ6, 5′-GGAATTCAAAAGAAAACATTGTCT-3′; Vγ7, 5′-AAGCTAGAGGGGTCCTCTGC-3′; Vδ1, 5′-ATTCAGAAGGCAACAATGAAAG-3′; Vδ2, 5′-AGTTCCCTGCAGATCCAAGC-3′; Vδ3, 5′-TTCCTGGCTATTGCCTCTGAC-3′; Vδ4, 5′-CCGCTTCTCTGTGAACTTCC-3′; Vδ5, 5′-CAGATCCTTCCAGTTCATCC-3′; Vδ6, 5′-TCAAGTCCATCAGCCTTGTC-3′; Vδ7, 5′-CGCAGAGCTGCAGTGTAACT-3′; Vδ8, 5′-AAGGAAGATGGACGATTCAC-3′.

PCR products (4 μl) were subjected to electrophoresis on a 1.5% agarose gel (Life Technologies) and transferred to a Gene Screen Plus filter (New England Nuclear, Boston, MA). The Southern blots of γ and δ PCR products were hybridized with MNG6 cDNA containing the Cγ2 gene, Jδ1 probe (oligonucleotide; 5′-TTGGTTCCACAGTCACTTGG-3′), or Jδ2 probe (oligonucleotide; 5′-CTCCACAAAGAGCTCTATGCCCA-3′). The Cγ2 probe was labeled with [α-³²P]dCTP using a Megaprime DNA labeling system (Amersham International, Amersham, U.K.) according to the manufacturer’s instructions. The Jδ1 and Jδ2 probes were labeled with [γ-³²P]ATP using a Megalabel 5′ labeling kit (Takara Shuzo, Kyoto, Japan) according to the manufacturer’s instructions. Before hybridization, the filters were incubated in 1 M NaCl, 1% SDS, 10% dextran sulfate, and 50 μg/ml heat-denatured salmon sperm DNA for 18 h at 60°C, and then the filters were washed in 2× SSC, 1% SDS for 15 min at 60°C. The radioactivity of each band of PCR product was analyzed with a Fujix BAS2000 Bio-image analyzer (Fuji, Tokyo, Japan). For nucleotide sequencing, RT-PCR products were resolved in low-melting-point agarose gels, isolated, and cloned into the TA vector PCR II (Invitrogen). Purified dsDNAs were sequenced by using a Taq Dye primer cycle sequencing kit (Perkin-Elmer, Norwalk, CT) and an Applied Biosystems 373A DNA sequencer (Applied Biosystems, Foster City, CA).

C3H/HeN mice were killed 3 days after i.p. inoculation with lipid A. Extraction of total RNA from sorted γδ T cells or αβ T cells in PEC, γδ T cells in liver, or J774A.1 (as a positive control) and cDNA synthesis were performed as described above. Serial dilutions of total RNA were primed with 20 pmol of random primer (Life Technologies) in 20-μl reaction mixture for reverse transcription. Synthesized cDNAs were amplified by PCR with primers derived from the murine cDNA. The specific primers were as follows: TLR2 sense, 5′-GGAGCGGCGGCTGCAGGACTC-3′; TLR2 antisense, 5′-CCAAAGAGCTCGTAGCATCC-3′; TLR4 sense, 5′-AGTGGGTCAAGGAACAGAAGCA-3′; TLR4 antisense, 5′-CTTTACCAGCTCATTTCTCACC-3′ (26).

Tissue culture 96-well plates were incubated overnight at 4°C with 100 μg/ml anti-TCR γδ mAb. The plates were then washed thoroughly and incubated for 1 h at 37°C with RPMI 1640 medium containing 10% FCS. The sorted γδ T cells (1 × 10⁵/well) were incubated in the 96-well plates for 48 h with or without immobilized anti-TCR γδ mAb in the presence or absence of LPS or lipid A. During the last 8 h of incubation, 1.0 μCi of [³H]TdR/well was added. The cells were then harvested, and the amount of [³H]TdR incorporated was determined by scintillation counting. In some experiments, the sorted γδ T cells (1 × 10⁵/well) were cultured with phosphorothioate-modified anti-sense oligonucleotide (A-ODN), 5′-GACCGCCTGCCCGGAGCCTAGG -3′, or sense oligonucleotide (S-ODN), 5′-CCTAGGCTCCGGGCAGGCGGTC-3′, specific for mouse TLR2 gene (5 μmol/L) in the presence of LPS for 48 h at 37°C.

The sorted γδ T cells (1 × 10⁵/well) were incubated in the anti-TCR γδ mAb-coated plates for 48 h in the presence of LPS or lipid A. IFN-γ levels in the culture supernatants were determined by ELISA (Genzyme, Cambridge, MA). ELISA for IFN-γ was performed in triplicate using Genzyme mAb according to the manufacturer’s instructions.

Data were analyzed by Student’s t test, and a Bonferroni correction was applied for multiple comparison. The value of p < 0.05 was considered statistically significant.

We have previously reported that γδ T cells significantly increased in the peritoneal cavity of C3H/HeN mice on day 3 after E. coli infection (37). To determine whether native lipid A derived from E. coli can induce an increase in γδ T cells, flow cytometry analysis for the expression of CD3, TCR αβ, and TCR γδ was conducted with nonadherent PEC of LPS-responsive C3H/HeN mice or LPS-hyporesponsive C3H/HeJ mice on day 3 after the inoculation with LPS or lipid A. A representative result from three independent experiments is shown in Fig. 1,A. The absolute numbers of peritoneal γδ T cells were calculated by multiplying the absolute number of the nonadherent PEC by the percentage of the γδ T cells, and they are shown in Fig. 1,B. The relative number of γδ T cells in the PEC of C3H/HeN mice were increased, constituting >30% of the total CD3-positive cell population after inoculation of 100 μg/mouse of LPS or 30 μg/mouse of lipid A (Fig. 1,A). The γδ T cells were also significantly increased in the peritoneal cavity of C3H/HeJ mice, albeit to a lesser degree compared with those in C3H/HeN mice (Fig. 1, A and B). There is a possibility that the native LPS and lipid A include the contaminated materials such as lipoproteins. Therefore, flow cytometry analyses for the expression of CD3, TCR αβ, and TCR γδ were conducted with nonadherent PEC of C3H/HeN mice or C3H/HeJ mice on day 3 after the inoculation with a synthetic lipid A analogue, ONO-4007. A representative result from three independent experiments is shown in Fig. 2,A. The absolute numbers of peritoneal γδ T cells were calculated by multiplying the absolute number of the total nonadherent PEC by the percentage of the γδ T cells and are shown in Fig. 2 B. Both relative and absolute numbers of the PEC γδ T cells in C3H/HeN and C3H/HeJ mice were significantly increased after the inoculation of ONO-4007 (1000 μg/mouse), although a >10 times dose of native lipid A was required.

To compare the V gene expressions of the γδ T cells induced by injection with E. coli, LPS, or lipid A in C3H/HeN and C3H/HeJ mice, total RNA was extracted from γδ T cells sorted from nonadherent PEC of mice inoculated with E. coli, LPS, or lipid A 3 days previously, and the V gene expressions analyzed by RT-PCR are shown in Fig. 3. The PEC γδ T cells from naive mice expressed Vγ1/2, 2, 4, and a diversity of Vδ genes, whereas the PEC γδ T cells of C3H/HeN mice inoculated with E. coli preferentially expressed Vγ6 and Vδ1 genes, findings that are consistent with those obtained in our previous study (37). Similarly, Vγ6 and Vδ1 genes were exclusively used by the PEC γδ T cells in both C3H/HeN and HeJ mice injected with LPS or lipid A. These results suggest that the γδ T cells expressing Vγ6/Vδ1 genes were selectively induced by the native lipid A, a natural product of E. coli.

To determine the junctional diversity of the Vγ6-Jγ1 and Vδ1-Jδ2 gene rearrangements in the γδ T cells induced by the native lipid A, we examined the nucleotide sequences of the Vγ6 and Vδ1 transcripts of the peritoneal γδ T cells in the lipid A-injected mice. As shown in Fig. 4, all 20 Vγ6-Jγ1 transcripts and 18 of 20 Vδ1-Jδ2 transcripts of the γδ T cells from the lipid A-injected mice showed no junctional diversity, resulting in in-frame invariant canonical sequences, which are preferentially expressed in fetal thymocytes at the late stage (approximately day 17) of gestation (31) and in the intraepithelial lymphocytes of reproductive organs such as the uterus (30). Taken together, these results suggest that the lipid A-induced γδ T cells in the peritoneal cavity expressed a canonical Vγ6/Vδ1 TCR, which was the same as that of E. coli-induced γδ T cells.

We previously reported that the γδ T cells induced by E. coli infection in C3H/HeN mice exhibited a strong proliferative response to LPS even in the absence of APC under TCR engagement (37). Therefore, we examined the proliferative response and cytokine production of the lipid A-induced γδ T cells from C3H/HeN and C3H/HeJ mice in response to LPS. The γδ T cells were purified by cell sorting from the nonadherent peritoneal cells on day 3 after lipid A injection, and they were incubated for 48 h with immobilized anti-TCR γδ mAb in the presence or absence of an optimum dose (10 μg/ml) of LPS. Fig. 5,A shows that the lipid A-induced γδ T cells exhibited a strong proliferative response in the presence of LPS. Notably, the γδ T cells in LPS-hyporesponsive C3H/HeJ mice proliferated more vigorously in response to LPS than did those in LPS-responsive C3H/HeN mice. Fig. 5 B shows that γδ T cells stimulated with LPS produced a large amount of IFN-γ, whereas neither IL-2 nor IL-4 was detected in the supernatant (data not shown). Moreover, the γδ T cells in LPS-hyporesponsive C3H/HeJ mice produced more IFN-γ in response to LPS than those did in LPS-responsive C3H/HeN mice.

To determine whether the responsiveness of the γδ T cells to LPS is mediated by TCR signal, sorted γδ T cells induced by lipid A were incubated for 48 h with or without LPS (10 μg/ml) in the presence or absence of immobilized anti-TCR-γδ mAb. As shown in Fig. 6,A, the γδ T cells exhibited a significant proliferative response even without TCR stimulation, and the proliferative response was augmented by TCR stimulation. In contrast, Fig. 6,B shows that γδ T cells stimulated with LPS produced a small amount of IFN-γ in the absence of TCR stimulation compared with those in the presence of TCR stimulation. These results suggested that stimulation with LPS induced γδ T cell proliferation but that TCR stimulation was required for the IFN-γ production. Dose responses of LPS for proliferation of γδ T cells induced by E. coli are shown in Fig. 6 C. In the presence of >0.1 μg/ml of the LPS, the γδ T cells exhibited a significant proliferative response.

We next examined the effect of soluble anti-TCR γδ mAb on the proliferative response of γδ T cells to LPS. Sorted γδ T cells induced by native lipid A were incubated for 48 h with the LPS (10 μg/ml) in the presence or absence of an optimum dose (10 μg/ml) of neutralizing anti-TCR γδ mAb or the same dose of control IgG. We confirmed that this concentration of anti-TCR γδ mAb could inhibit the proliferative response of the heat-killed Salmonella-specific γδ T cells (40). Fig. 7 A shows that the proliferative response to the LPS was not inhibited by the anti-TCR-γδ neutralizing mAb, confirming that the lipid A-induced γδ T cells can proliferate in response to the LPS in a TCR-independent manner.

Although macrophages from C3H/HeJ mice are hyporesponsive to many types of lipid A, they are nearly as responsive as their normal counterparts when stimulated with P. gingivalis native lipid A (2). Therefore, we next examined the proliferative response of the lipid A-induced γδ T cells to the lipid A from P. gingivalis besides the lipid A derived from E. coli or S. minnesota. When the sorted γδ T cells were incubated for 48 h with each kind of the lipid A (10 μg/ml), they exhibited a strong proliferative response with each of them (Fig. 7 B). Taken together with the findings for the γδ T cells from C3H/HeJ mice, these results suggest that the γδ T cells respond to the native lipid A using an LPS receptor other than TLR4, which is mutated in C3H/HeJ mice (23, 24).

CD14 and β₂ integrins (CD11/CD18) have been widely recognized as LPS receptors for immune cells (3, 7, 8, 9, 10, 11). Recently, two TLR family proteins, TLR2 and 4, have been identified as LPS signaling receptors (14, 15, 16, 17, 25, 26, 29). We attempted to determine which LPS receptors are used by the lipid A-induced γδ T cells. Flow cytometry analysis for the expression of β₂ integrins and CD14 was conducted with nonadherent PEC on day 3 after the inoculation with lipid A. A representative result from three independent experiments is shown in Fig. 8,A. The γδ T cells expressed CD11a but not CD11b, CD11c, or CD14. To examine whether CD11a is involved in mediating LPS response, sorted γδ T cells were incubated for 48 h with LPS derived from E. coli (10 μg/ml) in the presence or absence of a neutralizing anti-CD11a mAb (10 μg/ml) or the same dose of control IgG. As shown in Fig. 8 B, the proliferative response of the γδ T cells to LPS was not inhibited by the neutralizing anti-CD11a mAb.

To examine the TLR2 and TLR4 expressions by γδ T cells, total RNA was extracted from γδ T cells sorted from nonadherent PEC of mice inoculated with lipid A 3 days previously, and TLR2 and TLR4 expressions were analyzed by RT-PCR. As shown in Fig. 9,A, the γδ T cells in the peritoneal cavity expressed a significant level of TLR2 mRNA but not TLR4 mRNA. In contrast, both TLR2 and TLR4 mRNA were only marginally expressed by the αβ T cells in PEC and γδ T cells in the liver, which did not respond to LPS. Thus, these results suggest that the γδ T cells in the peritoneal cavity may respond to lipid A via TLR2. To test this issue, we examined the effect of a TLR2 A-ODN treatment on the proliferation response of the γδ T cells to LPS. As shown in Fig. 9,B, treatment with the A-ODN reduced the expression of TLR2 mRNA in the γδ T cells, and the proliferation of the γδ T cells in response to LPS was significantly impaired by this treatment compared with that of S-ODN treatment (Fig. 9,C). To further confirm the involvement of TLR2 in the γδ T cell-response to native lipid A, we examined the flow cytometry analysis for the expression of CD3, TCR αβ, and TCR γδ with nonadherent PEC in TLR2^−/− or TLR2^+/− mice on day 3 after the inoculation of native lipid A. A representative result from three independent experiments is shown in Fig. 10,A. The absolute numbers of the peritoneal γδ T cells were calculated by multiplying the absolute number of the nonadherent PEC by the percentage of γδ T cells and are shown in Fig. 10,B. The increase of γδ T cells in TLR2^+/− mice with C57BL/6/129 background was relatively less as compared with C3H/HeN mice. However, the increase of γδ T cells following an i.p. injection of lipid A were significantly impaired in TLR2^−/− mice compared with that of TLR2^+/− mice (p < 0.05, Fig. 10, A and B). Thus, these results indicate that TLR2 is at least partly involved in the proliferation of γδ T cells in response to the native lipid A from E. coli.

A significant fraction of γδ T cells is reported to respond to the LPS fraction through an apparently TCR-independent mechanism (41, 42, 43). Nitta et al. reported that γδ T cells in the peritoneal cavity of mice proliferate in response to TCR triggering in synergy with LPS (43). Leclercq and Plum reported that TCR Vγ5 cells, which are exclusively associated with canonical Vδ1 chain and are preferentially present in the early fetal thymus and the epidermis of mice, are activated to produce cytokines upon interaction with LPS via a TCR-independent pathway (41). Similarly, in the present study, we found that the γδ T cells expressing Vγ6/Vδ1 genes responded to native lipid A in vivo and in vitro in a TCR-independent manner. The γδ T cells in the peritoneal cavity expressed Vγ6-Jγ1 and Vδ1-Jδ2 mRNA with no N diversity, as they do in the fetal thymus and uterus. Thus, it would appear that only primitive γδ T cells with invariant TCR such as Vγ5/Vδ1 and Vγ6/Vδ1, which develop in the thymus at the early stage of gestation, respond directly to the bacterial products from Gram-negative bacteria. We have previously demonstrated that γδ T cell-deficient mice with a truncated Cδ gene are resistant to LPS-induced lethal shock with impaired TNF-α production (44). Furthermore, mice depleted of γδ T cells by TCR-δ gene mutation showed impaired host defense against E. coli (37). Like phagocytes, the primitive γδ T cells may play an important role in innate immunity against bacterial infection through rapid responses to the bacterial components via TLR2.

Bacterial LPS, a constituent of the outer membrane of the cell wall of Gram-negative bacteria, is one of the main causative agents of septic shock in humans. Recognition of LPS is a key event in host antimicrobial defense reactions. LPS is a complex glycolipid composed of a hydrophilic polysaccharide portion and a hydrophobic domain known as lipid A (45, 46). The conserved lipid A structure has been identified as the LPS component responsible for LPS-induced biological effects (45, 46, 47).

Recently, several members of the mammalian TLR family have been identified (18, 19, 20, 21, 22). Several lines of evidence suggest that one or more members of the TLR family are the cell-surface receptors for LPS, the prototypical activators of NF-κB and other proinflammatory responses (14, 15, 16, 17, 23, 24, 48). In C3H/HeJ and C57BL/10ScCr mice, mutations of the gene lps (lps^d) selectively impede LPS signal transduction, rendering them resistant to endotoxin yet highly susceptible to Gram-negative infection (23, 24). TLR4 from the C3H/HeJ mouse has a point mutation at amino acid 712 (Pro to His), and the C57BL10/ScCr mouse appears to be null for the TLR4 locus (23, 24). These observations suggested that TLR4 is a cell-surface component of the LPS signaling pathway. However, although lps^d mice are hyporesponsive to LPS, they are not unresponsive, and LPS-dependent gene transcription will occur if a very large dose of LPS is administered (15). Moreover, cells from C3H/HeJ mice are nearly as sensitive as their normal counterparts when stimulated with LPS components derived from certain bacteria such as P. gingivalis (2, 49). These observations suggested that proteins other than TLR4 may replace the function of TLR4 in signal transduction for LPS responsiveness. Transfection of cell lines with TLR2 confers them with the ability to respond to LPS with activation of NF-κB, thus directly suggesting that TLR2 serves in place of TLR4 (14, 16, 26). Our results demonstrated that not only in C3H/HeN mice but also in C3H/HeJ, the lipid A-induced γδ T cells exhibited a strong proliferative response in the presence of E. coli native lipid A as well as the lipid A fraction from P. gingivalis. Furthermore, the γδ T cells in the peritoneal cavity strongly expressed TLR2 mRNA, whereas those from the liver that did not respond to the naive LPS in vitro (37) expressed only a marginal level of TLR2 mRNA. Treatment with TLR2 A-ODN significantly inhibited the proliferation response of the γδ T cells to the lipid A. It has to be noted, though, that the inhibition of cell proliferation was partial. This was probably because the TLR2 mRNA inhibition by the A-ODN was moderate (Fig. 9 B). In serial dilutions of the samples, the inhibition was ∼75% compared with the S-ODN treatment. Moreover, the increase of γδ T cells following the i.p. injection of native lipid A was significantly impaired in TLR2^−/− mice compared with that of TLR2^+/− mice. However, we cannot exclude the possibility that a minor expression of TLR4 on the γδ T cells also acts as a LPS receptor for signal transduction in the γδ T cells in vivo. We previously reported that after E. coli infection the γδ T cells of the PEC included many Vγ6/Vδ1 γδ T cells but those in the liver did not, and the former exhibited a strong proliferative response to LPS but the latter did not (37). Taken together, these results suggested that Vγ6/Vδ1 γδ T cells responded to the lipid A and that TLR2 in γδ T cells is responsible, at least partly, for the LPS signaling in the Vγ6/Vδ1 γδ T cells.

It has recently reported that, TLR2 mediates monocyte activation by peptidoglycans, lipoteichoic acids, and microbial lipoproteins (27, 28, 29, 50). Roark et al. reported that Vγ6/Vδ1 T cells preferentially increase among γδ T cells infiltrating inflamed tissues induced by infection with Listeria monocytogenes, a Gram-positive bacteria (51). Takeuchi et al. have most recently reported that TLR4 mainly recognize lipid A and lipoteichoic acid from Gram-negative or -positive bacteria, respectively, whereas TLR2 plays a major role in recognition of peptidoglycan and lipoprotein from both Gram-negative or -positive bacteria (29). The primitive γδ T cells may play protective roles in infection with not only Gram-negative bacteria but also Gram-positive bacteria via TLR2 signal. The native LPS and lipid A we used here may contain some bacterial proteins, raising a possibility that bacterial proteins contaminated in LPS stimulate the Vγ6/Vδ1 γδ T cells via TLR2. However, the experiments using a synthetic lipid A analogue, ONO-4007, which does not contain any other bacterial materials, demonstrated that it induced an increase of γδ T cells in the peritoneal cavities in the C3H/HeJ mice that have the mutated TLR4 gene. These results indicate that TLR2 serve to function as one of the lipid A signaling receptors. The ability of ONO-4007 to induce γδ T cells in the PEC was lower than that of the naive lipid A. Hence, it is possible that the γδ T cells respond more vigorously to the naive lipid A than synthetic lipid A via TLR2-mediating signals for contaminated materials such as lipoproteins.

In the present study, the γδ T cells from C3H/HeJ mice responded more vigorously to LPS in vitro compared with those from C3H/HeN mice. Nevertheless, the γδ T cells were increased less in C3H/HeJ mice than in C3H/HeN mice after in vivo administration of lipid A, similar to E. coli infection as reported previously (37). Skeen and Ziegler reported that the peritoneal γδ T cells proliferated in response to IL-1 and IL-7 (52). It has been reported that TNF-α and IL-12 synergistically stimulate human γδ T cell proliferation (53). We have previously reported that γδ T cells proliferate in response to IL-15 in vitro (37, 54). Therefore, these cytokines derived from infected macrophages may preferentially stimulate the invariant γδ T cells to proliferate in the inflamed sites. We have previously demonstrated that the macrophages induced by E. coli infection in C3H/HeJ mice showed an impaired expression of monokine genes such as TNF-α, IL-6, IL-12, and IL-15 compared with those in C3H/HeN mice (37). Administration of anti-IL-15 mAb inhibited, albeit partially, the increase in γδ T cells after E. coli infection in C3H/HeN mice (37). Therefore, IL-15 derived from LPS-stimulated macrophages may be partly responsible for local expansion of γδ T cells in the peritoneal cavity in vivo. Furthermore, IL-15 is reported to have a strong chemotactic activity for T cells (55, 56). Thus, impaired accumulation and expansion of γδ T cells in C3H/HeJ mice after in vivo administration of the native lipid A may be attributable in part to impaired cytokine production by macrophages that may preferentially use TLR4 for LPS signaling.

In conclusion, γδ T cells expressing invariant Vγ6/Vδ1 TCR responded to the native lipid A not only from E. coli but also from P. gingivalis in a TCR-independent manner. The LPS/lipid A-reactive γδ T cells expressed TLR2 mRNA but no detectable TLR4 mRNA. Treatment with TLR2 A-ODN significantly inhibited the proliferative response of γδ T cells to the lipid A. Additionally, TLR2-deficient mice showed an impaired increase of the γδ T cells following in vivo injection of the native lipid A. These results suggest that the invariant Vγ6/Vδ1 γδ T cells respond to the lipid A fraction via TLR2. The primitive γδ T cells bearing invariant TCR may play an important role in innate immunity against microbial infection through TLR2 activation.

We thank Ono Chemical and Dr. J. A. Bluestone for providing an E. coli-type synthetic lipid A analogue (ONO-4007) and the UC7-13D hybridoma, respectively.