γδ T Cells Respond Directly to Pathogen-Associated Molecular Patterns ¹²

The first T cells to develop, γδ T cells are evolutionarily highly conserved (1) and likely are ancestral to modern αβ T cells and B cells (2). γδ T cells localize to epithelial barriers and traffic to inflamed tissue (3, 4). In these locations, macrophages, B cells and other specialized T cell subsets (regulatory T cells; Ref.5) respond to microbial surface-derived compounds, including (glyco)proteins, peptides, carbohydrates, and lipids (pathogen associated molecular patterns (PAMPs) ⁴; Ref.6), using a variety of well-defined PAMP receptors. Our previous studies defined the global genomic expression of bovine γδ T cell subsets and indicated that these cells express mRNAs for multiple myeloid and B cell proteins (7, 8). These studies suggested that expression of PAMP receptors or related myeloid proteins might be one mechanism through which γδ T cells participate in rapid innate defense responses.

γδ T cells have been shown to respond to non-peptide Ag independent of MHC presentation. Human γδ T cells respond to nonpeptidic phosphorylated molecules through a TCR-dependent but MHC-independent mechanism by proliferation (9), increased cytotoxicity and expression of TNF-α (10), IFN-γ (11), and chemokines (12). Human γδ T cells also respond to microbial alkylamines by expressing IL-2 (13) and IFN-γ and TNF-α (14). Stimulation with crude LPS has been used as a negative control, as it did not up-regulate TNF-α and IFN-γ in γδ T cells to the same extent as did bacterial alkylamines (14). However, Lahn et al. (15) measured a profound and very early indirect stimulatory effect of LPS on unpurified γδ T cells, whereas others have demonstrated a direct stimulatory effect of LPS on γδ T cells that is TCR-independent (16, 17). These latter results suggest that γδ T cells may respond to some PAMPs, but the detailed global direct response in the absence of APCs, and their expression of PAMP receptors has not been described.

We hypothesized that γδ T cells respond directly to PAMPs by quickly up-regulating genes with important innate functions. To discover the global gene complement expressed after PAMP stimulation, we have analyzed the transcriptional profile of purified expanded human γδ T cells after exposure to LPS using microarray analysis. Several genes were rapidly up-regulated indicating that γδ T cells were activated and increased expression of several chemokines upon stimulation with LPS. We have confirmed a similar, yet more robust, response to PAMPs from neonatal bovine γδ T cells isolated from peripheral blood. Differential gene and protein expression following stimulation with LPS and peptidoglycan (PGN) in both expanded human and primary bovine γδ T cells were detected. Both human and bovine γδ T cells expressed and differentially regulated genes that encode innate receptors thought to be expressed only in myeloid cells. Although few innate receptor transcripts were detected in the human γδ T cells, those encoding nucleotide-binding oligomerization domain (NOD) 1 and NOD2 were identified. mRNAs encoding innate receptors, including Toll-like and scavenger receptors, were more readily detected in γδ T cells isolated from bovine peripheral blood, suggesting that expansion in culture diminishes their expression. Our findings indicate that purified γδ T cells are directly stimulated by PAMPs, but the response differs from that following stimulation with known γδ T cell Ags and the inflammatory response expected from cells of the monocyte lineage.

Blood was collected from healthy adult human donors and lymphocytes purified using Histopaque 1077 (Sigma-Aldrich) according to manufacturer’s instructions. Lymphocytes were cultured in RPMI 1640 media (10% FBS in RPMI 1640 supplemented with 1% each essential amino acids, penicillin/streptomycin, l-glutamine, and 10 mM HEPES) supplemented with IL-2 (1 ng/ml), IL-15 (10 ng/ml; PeproTech), and isopentyl pyrophosphate (IPP; 1 μg/ml; Sigma-Aldrich) (18). Bovine lymphocytes were Histopaque-purified from fresh calf blood as previously described (7).

Bovine γδ T cells were sorted using the MACS magnetic bead system (Miltenyi Biotec) to >95.5% purity. Briefly, Histopaque-purified bovine lymphocytes were labeled with GD3.8, a pan-bovine γδ T cell marker that when cross-linked does not activate the cells, and then washed and labeled with MACS bead-anti-mouse IgG-conjugated Ab. Cells were then washed and purified on a magnetic column according to the manufacturer’s instructions. The cells were labeled with PE-conjugated goat anti-mouse Abs (Jackson ImmunoResearch Laboratories) and analyzed using a FACSCalibur (BD Biosciences), as described (7), to confirm purity.

Human γδ T cells were sorted after 2–3 wk expansion in culture on a VANTAGE SE cell sorter (BD Immunocytometry Systems) to >95% purity, as previously described (7) after labeling with a pan γδ Ab (Clone IMMU510; BD Immunocytometry Systems) conjugated to FITC. For analysis of innate receptor mRNA expression and stimulation with ultra-pure LPS and IPP, bovine cells were labeled with GD3.8 conjugated to FITC and similarly VANTAGE sorted.

Human or bovine sorted cells were plated at ∼2 × 10⁶ cells/ml and stimulated with either phenol-extracted LPS (phLPS, catalog no. L2630; Sigma-Aldrich), ion-exchange column-purified LPS (ionLPS, catalog no. L3024; Sigma-Aldrich), ultra-pure LPS from Escherichia coli 0111:B4 (ultraLPS; Invivogen) or PGN (catalog no. 77140; Invivogen) (each at 10 μg/ml) each suspended in sterile PBS, or PBS alone. RNA was extracted from sorted γδ T cells with TRIzol Reagent (Invitrogen Life Technologies) according to the manufacturer’s instructions. RNA was treated with RNase-free DNase, extracted again with phenol chloroform and ethanol precipitated.

Sorted cells from four human γδ T cell cultures were treated with either PBS or phLPS for 4 h, RNA was extracted, pooled, and used to probe Affymetrix Genechip Human Genome U133A 2.0 Arrays (catalog no. 900471; Affymetrix) that represent 14,500 human genes. cDNA amplification and synthesis of biotin-labeled cRNA was performed with the Alternative One-cycle target labeling protocol with 10 μg of total RNA as described in the GeneChip Expression Analysis Technical Manual (March 2004). Hybridization was performed with 10 μg of cRNA. Washing and staining was performed in the GeneChip Fluidics Station 450 using the Midi_euk2v3 protocol. Chip scans were performed on the Affymetrix GeneChip Scanner 3000. GeneChip Operating Software (GCOS version 1.1; Affymetrix) (19, 20) was used for data collection and analysis with scaling to an arbitrary target intensity of 500 and further analysis was done using Microsoft Excel.

Real-time RT-PCR was performed as previously described (7). Primers were designed with Primer Express, primer design software from Applied Biosystems. The reverse transcription reactions were performed with Superscript II (Invitrogen Life Technologies) and ∼700 ng of RNA according to the manufacturer’s protocol. Relative-specific mRNA in the γδ T cells was quantified by measuring SYBR Green incorporation during real-time quantitative RT-PCR using the relative standard curve method. Primers specific for 18S RNA were used as the endogenous control. The PCR was set up in triplicate, cycled, and data collected on the Applied Biosystems GeneAmp 5700 Sequence Detection System and calculations were performed as described in the manufacturer’s protocol and in User Bulletin no. 2 for the ABI PRISM 7700 Sequence Detection System.

Purified human γδ T cells were cultured and sorted as described above. The remainder of the same purified human γδ T cells used in microarray analysis were plated more densely (5 × 10⁶ cells per ml) and stimulated for 22 h. The supernatant fluids from these cultures were collected, frozen at −80°C, and sent to Pierce for Searchlight Proteome array analysis to measure representative protein concentrations as suggested by the microarray. The proteome array was run in triplicate and the protein concentrations for cells isolated from a given individual and stimulated with PBS or LPS were compared using the paired t test.

MACS bead-sorted bovine γδ T cells were equally divided into separate wells at 4 × 10⁶ to 1 × 10⁷ (depending on the sort yield) cells/ml in complete RPMI 1640 media, and stimulated with either phLPS, PGN, or PBS for 4 h. To remove the PAMP, the cells were then centrifuged at 500 × g for 5 min, washed with cold PBS, recentrifuged, then resuspended in 700 μl of RPMI 1640 media and incubated for 18–20 h at 37°C. Supernatant fluids were collected after centrifugation and 600 μl of each supernatant was placed into a well of a 24-well plate. Transwell inserts with 3-μm filters (Costar) were placed into the wells. Fresh bovine blood was collected and RBC were water-lysed two or three times. The resulting fresh bovine leukocytes were added to the transwell inserts at 1 × 10⁶ cells per well. Controls included medium only in the lower well (passive), cells added directly to the medium (no transwell insert, 100% migration) and recombinant human chemokines RANTES (CCL5) and MIP-1α (CCL3) (PeproTech) at 5 or 10 ng/ml each in media. The plate was then incubated at 37°C for 6 h to measure potential migration of all cell populations. The inserts were then removed and ∼50,000 polystyrene microbeads (Polybead Polystyrene, 15 μm; Polysciences) were added to each well. The immigrant cells were collected and the wells rinsed with Hank’s buffer with 2 mM EDTA. The side and forward scatter profiles were collected on a FACSCalibur cytometer (BD Immunocytometry Systems). Thirty thousand events were collected and gates were set to contain the lymphocytes, monocytes, neutrophils, cells with a phenotype consistent with activated neutrophils, and the polystyrene beads. Cell numbers in each gate were calculated relative to the number of beads present.

Based on their presumed innate functions, localization to mucosal surfaces and the observed expression of several myeloid genes, we hypothesized that γδ T cells respond directly to PAMPs. To begin to assess this response, the transcriptional profile induced in human γδ T cells after LPS stimulation was elucidated. After expansion in culture, human γδ T cells were sorted to 95.87–99.49% purity, rested overnight, and stimulated with either a crude phLPS preparation or PBS for 4 h. RNA from cultures derived from four different subjects was extracted and pooled to minimize individual variation. The differential gene expression was analyzed on Affymetrix Human Genome U133A 2.0 Arrays (GEO accession numbers: GSM45116, GSM45117, GSE2396).

The number of genes demonstrating robust changes was minimal, yet revealed a specific cytokine response of γδ T cells to crude LPS. Eighty-one genes had significant increases in LPS-stimulated compared with resting γδ T cells (Table I). The genes that revealed the highest signals after PAMP stimulation were those encoding chemokines, such as the lymphotactin (XCL1) precursor, MIP-1β (CCL4), and MIP-1α, (Table I). Other cytokines, such as lymphotactin, TNF-α, and GM-CSF also showed considerable fold increases following stimulation. An increase was detected in IFN-γ mRNA, which, along with TNF-α, is considered a principal response in γδ T cells, but the IFN-γ increase was relatively minor.

Other genes related to activation and inflammation increased after LPS stimulation. Of note is a 13-fold increase in the IL-2R mRNA, which suggests that LPS stimulation of γδ T cells might prime γδ T cells by rendering them more susceptible to cytokine stimulation. Other than TNF-α, none of the other genes typically associated with NFκB activation in myeloid cells following LPS stimulation were detected in the array. Interestingly, a gene associated with inhibition of NFκB (NF of κ light polypeptide gene enhancer in B cells inhibitor, α) was up-regulated and may be in part responsible for this difference between the response to PAMP stimulation in γδ T cells compared with myeloid cells. Thus, PAMP stimulation of γδ T cells clearly resulted in up-regulation of several cytokine and activation-related genes.

Only 12 genes had significant decreases in transcription in LPS-treated γδ T cells compared with resting cells (data not shown). These genes had a very low average signal of 134.7, compared with an average signal of 2885 for the up-regulated genes. Furthermore, their signals were only detected in the PBS-treated γδ T cells, and all were absent in the LPS-treated cells. These data suggest that the genes that decreased in transcription had a very minor role relative to the overall profile of the PAMP-stimulated γδ T cells.

Differential gene expression was confirmed in γδ T cells purified from cultures established from three donors that were not included in the microarray analysis. Human γδ T cells, expanded in culture for up to 3 wk, and sorted to 97.99–99.46% purity, were stimulated with PBS, phLPS, or PGN and RNA was extracted from the cells. Real-time RT-PCR was used to detect mRNAs coding for the chemokines MIP-1α, MIP-1β and RANTES. Fig. 1 shows the results with RNA pooled to overcome individual variation. Stimulation of these cells with phLPS caused a consistent increase in expression in MIP-1α and MIP-1β mRNA, whereas RANTES expression at the mRNA level remained unchanged, confirming the result of the array. PGN had a minimal effect on the human cells.

To demonstrate that the response to PAMPs occurred with nonexpanded primary cells, bovine peripheral blood γδ T cells were isolated at >95.5% purity using magnetic beads (MACS). Cells were stimulated with PBS alone, phLPS, or PGN, RNA was extracted and real-time RT-PCR was performed to measure MIP-1α and RANTES transcripts. Little to no signal for the mRNA encoding the chemokine MCP-1 was detected (data not shown). Fig. 2 shows real-time RT-PCR performed on RNA from six different calves. γδ T cells from some calves were more sensitive to PGN, whereas others showed greater chemokine mRNA change after LPS stimulation. RANTES was up-regulated 2- to 5-fold and MIP-1α 2- to 240-fold, depending on the PAMP and individual calf. Expression of chemokines varied greatly between individual calves, as they are not genetically alike, and have variable exposure to environmental agents. However, the up-regulation of MIP-1α and RANTES after treatment of the purified cells with phLPS and PGN was consistent, and generally much more robust than the up-regulation that occurred in cultured human γδ T cells, suggesting that expansion in culture may diminish the response of the human γδ T cells to PAMP stimulation.

The LPS preparation used in the initial characterizations of γδ T cell PAMP responses was phenol-extracted, and therefore contained contaminants potentially responsible for chemokine regulation. To confirm that the observed response of bovine γδ T cells was a response, at least in part, to LPS, and not contaminating factors, such as IPP, additional experiments were performed. γδ T cells were Vantage FACS-sorted, rested overnight, and stimulated with phLPS, a more pure ionLPS, or ultraLPS, or IPP followed by RNA extraction and real-time RT-PCR analysis. Cells from the majority of calves responded to ionLPS, or ultraLPS, and there was no statistically significant chemokine induction detected in response to stimulation with IPP (Fig. 3). Great variation between individuals was again noted. Bovine γδ T cells are therefore capable of responding to ultraLPS, but the increase in chemokine transcripts did not occur in response to IPP.

We also attempted to amplify mRNAs encoding proteins that are up-regulated by the NF-κB pathway in myeloid cells after interaction with PAMPs by quantitative real-time RT-PCR. Primer pairs were designed to measure the human and bovine (when sequence was available) genes: IP-10, Cox-2 (21), iNOS (22), LPS-induced TNF factor (LITAF) (23), and Bcl-x_L. These genes either could not be detected, or were not altered in purified human and bovine γδ T cells after stimulation with phLPS or PGN (data not shown), consistent with the microarray, thus the response of γδ T cells to PAMPs is distinct from that of cells of the monocyte lineage.

To confirm that differential mRNA expression was indicative of change of protein concentration in the supernatant fluids, the concentrations of 11 different proteins secreted from human γδ T cells after stimulation with PBS or phLPS were determined. Purified γδ T cells from four individual human cultures were stimulated with PBS or phLPS, and either lysed for RNA extraction and microarray analysis after 4 h, or stimulated for 22 h after which time the tissue culture supernatant fluids were collected. Chemokine protein concentrations in the tissue culture supernatant fluids were measured in Searchlight proteome arrays in triplicate (Table II). Variation of protein concentrations between subject samples was great, however, differences between treated and untreated cells from individuals were largely statistically significant. The cell purity varied from 95.87–99.49% yet the protein expression profiles were similar, suggesting that contamination with other cell types did not significantly contribute to the expression profiles. Notably, concentrations of MIP-1α and MIP-1β in some supernatant fluid samples approached 10 ng/ml after LPS stimulation. Expression of IFN-γ and TNF-α proteins were low, relative to the levels of chemokine protein in the supernatant fluids. Consistent with bovine real-time RT-PCR data, the chemokine MCP-1 was not detected in human γδ T cell tissue culture supernatant fluids. Overall, the Searchlight proteome array analyses were consistent with the real-time RT-PCR and microarray data indicating that mRNA levels accurately reflected expression levels of several secreted proteins in the PAMP-stimulated γδ T cells.

Because bovine-specific Ab reagents are not available, the functional capacity of chemokines expressed by bovine γδ T cells was characterized in chemotactic assays. Purified bovine γδ T cells were stimulated for 4 h with PBS, phLPS, or PGN, the cells were washed, and then incubated for 18–20 h. The collected supernatant fluids were then used in migration assays with recombinant MIP-1α and RANTES as controls. As shown in Fig. 4, both the recombinant chemokines (a representative control experiment) and the γδ T cell supernatant fluids from six different calves stimulated the migration of cells in the R4 gate with a phenotype consistent with activated neutrophils. The immigrant cells stained positive for BN1–15 (a bovine neutrophil marker, data not shown), and had increased side and forward scatter phenotypes relative to control unactivated neutrophils. Surprisingly, there was no change in any of the other gated cell populations, suggesting that neonatal bovine lymphocytes and monocytes are not responsive to these chemokines. Similar to the real-time RT-PCR results, migration efficiency varied between animals. In these assays differences in both the animal source of γδ T cells and the leukocyte donors must be considered. The migration of the same cell population toward recombinant chemokines and the γδ T cell supernatant wells indicated that, consistent with the RNA data, stimulation of γδ T cells with PAMPs results in expression of functional chemokine proteins.

Given that human and bovine γδ T cells responded to PAMPs in the absence of APCs, the array data was mined for signals from innate receptors potentially responsible for the response to PAMPs. The most obvious candidates, the TLRs, were below detection in the microarray analysis (Table III). Likewise, commercial Abs directed against human TLR2 and TLR4 proteins did not consistently label human γδ T cells. Though these mAbs did label PAMP responsive γδ T cells repeatedly from one donor, most PAMP responsive donor cells were not stained with TLR Abs (data not shown). The array data suggested expression of the proteins NOD1, NOD2, and NALP1, members of a family involved in intracellular recognition of PAMPs (24). NOD1 and NOD2 are receptors for specific PGN motifs (25, 26, 27) and are thought to be expressed in epithelial and monocytic cells, respectively (24, 28). In the microarray experiment, NOD1 was down-regulated with LPS treatment, but this was not significant, whereas there was a moderate increase in NOD2 expression (Tables I and III). In addition, expression of mRNAs encoding NOD1 and NOD2 in expanded human γδ T cells was consistently detected in real-time RT-PCR assays (data not shown). Two genes peripheral to TLR4 function were also detected in the human γδ T cells: MyD88 and CD11b (Table III). Although a potential receptor for LPS was not identified, the expression of NOD transcripts in γδ T cells supports the hypothesis that γδ T cells express innate receptors thought to be restricted to myeloid cells.

Cells isolated directly from peripheral blood are more likely to reflect in vivo phenotypes, thus, primers were designed specific for available sequences and used to identify mRNA expression for a number of potential molecular pattern receptors in bovine γδ T cells. Blood was collected from three different calves, γδ T cells were labeled, sorted by FACS to 96.1–99.27% purity, and mRNA was isolated for real-time RT-PCR analysis. Consistent with the more robust response to PAMPs, bovine γδ T cells had detectable mRNA signals for nearly every innate receptor tested by real-time RT- PCR (Fig. 5). TLRs are innate receptors for a variety of bacterial products such as Tri-acyl lipopeptides (bacteria, mycobacteria) (29) and soluble factors from Neisseria meningitides (Ref.30 ; TLR1 and 2), lipoteichoic acids (LTAs) (TLR2; Refs.31 and 32), dsRNA (TLR3; Ref.33), LPS (TLR4; Ref.34), bacterial flagellin (TLR5; Ref.35), single-stranded viral RNA (TLRs 7 and 8; Ref.36), and CpG DNA (TLR9; Ref.37). Specific signals for each of these genes were amplified in the bovine γδ T cells (Fig. 5 a). Expression of the related genes CD14, that can deliver LPS to TLR4, and adaptor proteins, MyD88, TRAM, and SARM were also identified. Although subtle, expression of mRNAs encoding TLRs 1, 2, 4, CD14, and TRAM were higher in the resting cells.

Scavenger receptors (SR) expressed on myeloid cells can also recognize PAMPs and have a role in recognition and uptake of host molecules and apoptotic cells (38). Bovine γδ T cells expressed some scavenger receptors as demonstrated by real-time RT-PCR (Fig. 5 b). Primers specific for mannose binding lectin (MBL), CD36, and SR-BI all produced strong signals, whereas SR-AI and II transcripts were not detected in γδ T cell mRNA. SR-AI and II have been reported to recognize both LTA from Gram-positive bacteria (39) and LPS from Gram-negative bacteria (40). CD36 has been implicated in the uptake of apoptotic cells (41). Differential expression was more apparent in this group of proteins; MBL transcripts were down-regulated by LPS treatment, whereas CD36 expression increased. Consistent with our hypothesis, the bovine γδ T cells expressed mRNA encoding SRs thought to be expressed primarily in monocytes.

Specific signals were also detected for the mRNAs encoding protein kinase regulated by RNA (protein kinase regulated by RNA (PKR)), CD38, and CD11b in bovine γδ T cells (Fig. 5 c). PKR is responsible for early cellular response to viral infection, through recognition of dsRNA (42). CD38 can play an important role in regulating innate immunity and inflammatory responses to bacterial pathogens (43). Along with CD18, CD11b is part of the myeloid compliment receptor CR3, which can bind a promiscuous range of ligands (44). While expression of CD11b mRNA decreased slightly with LPS stimulation, signals for the proteins PKR and CD38 increased. Thus, mRNAs encoding several proteins thought to be expressed exclusively in myeloid cells were readily detected in highly purified bovine γδ T cells.

The goal of this investigation was to characterize the direct response of γδ T cells to PAMPs in the absence of APCs. Several studies have characterized a response of γδ T cells to PAMPs, but, with the exception of Leclercq and Plum (16), they have largely described the response in the context of PBMC and have not detected a direct response in purified cells (15, 17, 45, 46). In several cases it is reported that APCs are necessary for γδ T cell responses to known bacterial Ags (47, 48, 49). We have shown that purified human and bovine γδ T cells up-regulate several chemokines and genes involved in activation in direct response to a crude phLPS preparation. Purified γδ T cells from bovine peripheral blood similarly expressed the chemokines MIP-1α and RANTES in direct response to both phLPS and PGN. The γδ T cells in this study were >95.5% pure, and frequently >99% pure, suggesting that PAMPs are able to directly stimulate γδ T cells and APCs are not required. Contamination with monocytes or other cells is unlikely to result in the rapid and robust response that is consistent between cell preparations of variable purities.

Although contaminants of the crude LPS preparation may be, in part, responsible for the response of γδ T cells, bovine γδ T cells likely express functional receptors specific for LPS. Bovine γδ T cells from most individuals had altered chemokine gene transcription after stimulation with phLPS, ionLPS, and ultraLPS, whereas IPP did not alter transcription of chemokine genes. Cipriani et al. (12) measured an increase in MIP-1α, MIP-1β, RANTES, and IFN-γ in supernatant fluids of sorted human Vδ2⁺ γδ T cells after stimulation with IPP. Human γδ T cells expressed chemokines in response to stimulation with phLPS, but TNF-α and IFN-γ were present at relatively low concentrations in stimulated human γδ T cell supernatant fluids. Others have reported that IPP concentrations in E. coli are insufficient to elicit a response (50) and bacterial-derived phosphoantigens drive overt production of TNF-α (14, 50) which was not detected after phLPS stimulation of human γδ T cells in our experiments. Thus, IPP is probably not responsible for the results we obtain with phLPS stimulation of human γδ T cells. Consistent with our findings are the results of Wang et al. (14) who did not detect TNF-α and IFN-γ production in response to a phLPS preparation. A corollary to our results is the observation of Leclercq and Plum (16) who detected a direct response to LPS involving up-regulation of GM-CSF in a subset of murine γδ T cells through a TCR-independent mechanism. Our data clearly indicates that γδ T cells have a unique response to microbial PAMPs that is distinct from that of γδ T cells after stimulation with phosphorylated Ags or alkylamines.

The differential chemokine protein levels expressed by PAMP-stimulated human and bovine γδ T cells were functionally relevant. Five million human γδ T cells produced nanogram levels of MIP-1α and MIP-1β within 22 h following LPS stimulation. In bovine γδ T cells these and/or other factors in supernatant fluids from purified, PAMP-stimulated γδ T cells had in vitro chemotactic activity specific for cells with a phenotype consistent with activated neutrophils. This immigrant population was not sensitive to IL-8 (data not shown), consistent with the finding that activated human neutrophils down-regulate the IL-8R, and increase expression of CCR1, a receptor for MIP-1α and RANTES (51). Therefore, while IL-8 is responsible for chemotaxis of circulating neutrophils to sites of inflammation, our data suggests the interesting possibility that tissue resident γδ T cells respond to PAMPs by expressing chemokines responsible for recruiting activated neutrophils from within tissue to a specific site of infection in the epithelium. However, the intent of these experiments was to characterize expression of chemokine proteins from bovine γδ T cells, and not detailed characterization of the immigrant cells.

Notably, the bovine γδ T cells showed a more robust up-regulation of chemokines in response to PAMPs, than did the human cells. Thus, expansion in culture may dampen the reaction to PAMPs, potentially by down-regulating innate receptors. Alternatively, the differences may be species or age of donor specific (adult humans vs neonatal calves). Time course experiments for peak chemokine expression were not performed. Prior stimulation of the human cells is likely to alter the timing of mRNA expression relative to cells in vivo. This may be the reason for detection of an increase in expression of RANTES mRNA in bovine, but not in human γδ T cells. However, RANTES protein was detected in human γδ T cell supernatant fluids. Thus, RANTES mRNA is likely up-regulated at an interval different from that of mRNA encoding MIP-1α and MIP-1β. The bovine system is one in which ex vivo cells can be used as a valuable confirmation of data obtained with human γδ T cells that must either be cloned or expanded in culture for sufficient cell numbers.

Consistent with their more robust response to PAMPs, freshly isolated bovine γδ T cells had more consistent expression of a variety of mRNAs encoding innate receptors. Similarly, Mokuno et al. (17) have detected expression of TLR2 mRNA in a murine γδ T cell subset and demonstrated its role in proliferation in vivo in response to (likely a contaminant of) LPS (17, 52). The innate receptor or receptors in humans cells that mediate the responses described here is unclear and may be below the detection level of the microarray. Our analyses of the predicted TLRs, TLR2 and TLR4 in human cells have been inconsistent. SRs and CD11b may have a role. CD11b/CD18 (Mac-1, CR3) acts in concert with CD14 and TLR4 for maximal signaling after LPS stimulation (21). CD11b was detected on the microarray, and Abs specific for CD11b consistently stain human γδ T cells expanded in culture (data not shown). Thus, this protein, generally considered to be monocyte/macrophage-restricted, may play a role in LPS signaling in γδ T cells. The responses in human and bovine cells, although similar, may be due to use of different PAMP receptors. Expression of innate receptors on γδ T cells is also clearly different from that on myeloid cells, and, logically, signaling in γδ T cells in response to microbial products is also distinct.

The presence of the NOD genes in the microarray analysis of LPS-stimulated human γδ T cells was intriguing. These receptors may be responsible, at least in part, for the observed response of γδ T cells to stimulation with the crude LPS preparation. It has recently been shown that NOD2 rather than TLR2 is likely responsible for internal sensing of PGN, whereas TLR2 responds to contaminants in commercial PGN preparations (32). Mutations in NOD2 that disrupt its signaling through NFκB have been associated with Crohn’s disease, a type of inflammatory bowel disease (53, 54). An increase in expression of lymphotactin in Crohn’s disease was also reported, but the specific role of γδ T cells was not addressed in this study (55). Our data suggest that NOD2 mRNA is expressed in human γδ T cells and may be in part responsible for the reaction to PAMPs that is distinct from that of monocytes. Given that the largest populations of γδ T cells in adult humans resides in the gut mucosa, disruption of NOD2 function in these cells may disrupt the balance of chemokines expressed and warrants further investigation.

To better elucidate the innate functions of γδ T cells, their specific responses to PAMPs in the absence of APCs was investigated. The possibility of direct γδ T cell responses to PAMPs through the expression of innate receptors has not been previously presented. Our data indicates that the response of γδ T cells to PAMPs is distinct from that of characterized γδ T cell Ags, and unlike the response of cells of the myeloid lineage. Given their localization to mucosal sites, and early evolution, direct response to microbial PAMPs may be an important, yet underappreciated, role of γδ T cells in innate immunity.

We thank Dr. Jean Starky for microarray assistance, Larissa Jackiw for cell sorting, and Jill Graff and Ellen Kress for critical review and several helpful suggestions.

M. A. Jutila holds shares in LigoCyte Pharmaceuticals, which together with Montana State University, holds a National Institutes of Health contract that partially funded the work presented in this article.