LAG-3 Regulates Plasmacytoid Dendritic Cell Homeostasis¹

Lymphocyte activation gene 3 (LAG-3⁴, CD223) plays an important role in negatively regulating T cell proliferation, function, and homeostasis (1, 2, 3, 4, 5, 6). In addition, LAG-3 is required for maximal natural and induced regulatory T cell (Treg) function (4, 7). LAG-3 is closely related to the T cell coreceptor CD4, having a similar genomic organization and the same chromosomal location suggesting that LAG-3 may have arisen as the result of a gene duplication event (8). Like CD4, LAG-3 also binds to MHC class II molecules but with a significantly higher affinity (9, 10, 11). Previous studies have suggested that human LAG-3 expression is restricted to activated T cells and NK cells and is not expressed by B cells, monocytes, or any other cell types tested (6, 12). Early analysis in mice also suggested that LAG-3 expression was restricted to T and NK cells (9). Although very few naive TCRαβ T cells in murine spleen and thymus express surface LAG-3 (1–2%), ∼18% TCRγδ T cells, and ∼10% NK cells are positive. In contrast, all T cells express LAG-3 2–3 days postactivation. Northern blot analysis also demonstrated that mRNA expression was most predominant in the spleen and thymus (9). Interestingly, a recent study suggested that LAG-3 is expressed on B cells when activated in the presence of T cells (13), although this has not yet been corroborated by other groups. Taken together, these data are consistent with the general notion that LAG-3 expression is restricted to lymphoid cells.

However, additional data suggest that the LAG-3 expression may not be restricted to lymphoid cells. First, in situ hybridization showed that LAG-3 mRNA is restricted to the thymic medulla and the splenic red pulp, patterns that are not consistent with the normal distribution of T cells or NK cells (9). Second, we have recently shown that LAG-3 expression on T cells is regulated by two transmembrane metalloproteases, ADAM10 and ADAM17, which cleave LAG-3 from the cell surface (14, 15). The resultant cleavage product, soluble LAG-3 (sLAG-3), is found at significant levels in normal mouse serum (∼180 ng/ml) (15). However, we were surprised to find that Rag1^−/− mice, which lack T and B cells, and Rag1^−/−;Il2rg^−/− (CD132, γ_c) mice, which lack T, B, and almost all NK cells, had near normal levels of sLAG-3 in their sera (data not shown). These data suggested that another cell type, other than T, B, and NK cells, expressed LAG-3 and was responsible for most of the sLAG-3 found in serum.

The goal of this study was to identify the unknown LAG-3-expressing cell type. Initial analysis confirmed the previously reported expression of LAG-3 in resting natural Tregs and NK cells (7, 9). However, we also found significant expression in CD11c⁺ dendritic cells (DCs). Subsequent analysis of different DC subsets clearly showed exclusive LAG-3 mRNA expression in plasmacytoid dendritic cells (pDCs) but not lymphoid or myeloid DCs. LAG-3 was shown to contribute to the homeostatic regulation of pDCs, playing a role in both intrinsic pDC physiology and possibly pDC interactions with T cells.

The following mice were used: Lag3^−/− (obtained from Yueh-Hsiu Chen, Stanford University, Palo Alto, CA, with permission from Christophe Benoist and Diane Mathis, Joslin Diabetes Center, Boston, MA; Ref. 16); Rag1^−/− (The Jackson Laboratories; Ref. 17); B6.129-H2^dlAb1-Ea/J (MHCII^Δ/Δ; lack an 80 kd segment of the MHC class II locus and thus lack expression of all four classical MHC class II chains) (The Jackson Laboratories; Ref. 18); B6.SJL-Ptprc^aPep3^b/BoyJ (CD45.1 congenic) (The Jackson Laboratories); and C57BL/6J and BALB/c mice (The Jackson Laboratories). Genome-wide microsatellite analysis demonstrated that 100% of the 88 genetic markers tested for the Lag3^−/− mice were derived from C57BL/6 mice (Charles River Laboratories). The CD45.1:Lag3^−/−, Lag3^−/−;Rag1^−/−, Rag1^−/−;MHCII^Δ/Δ and Lag3^−/−:Rag1^−/−;MHCII^Δ/Δ crosses were performed and maintained along with the Rag1^−/−, CD45.1 and Lag3^−/− colonies in the St. Jude Animal Resource Center. All animal experiments were performed in an AAALAC-accredited, Helicobacter-free, SPF facility following national, state and institutional guidelines. Animal protocols were approved by the St. Jude Institutional Animal Care and Use Committee.

Single cell suspensions were made from spleens and RBC lysed with Gey’s solution. For pDC cell staining/purification, spleens were first treated with collagenase (Worthington Biochemical) and DNase I (Sigma-Aldrich) for 1 h at 37°C, and single cell suspensions were made. Splenocytes were first stained with 10% mouse serum to block Fc (BD Pharmingen) for 10 min on ice. The cells were then stained for the following cell surface markers using various conjugated Abs from BD Pharmingen: CD45R/B220 (RA3–6B2), CD11b/Mac1 (M1/70), Gr-1 (RB6–8C5) (granulocyte marker), CD25/IL2R (7D4), CD4 (RM4–5), CD11c (N418) (DC marker) NK1.1 (PK136) (NK cell marker for B6), pan NK (DX5) (NK-selective marker for BALB/c mice which do not express NK1.1), H-2A^b (25–9-17), and LAG-3 (C9B7W). In addition, mPDCA-1 (JF05-1C2.4.1) (Miltenyi Biotech) was used to stain for pDCs throughout the article. The cells were then analyzed by flow cytometry (BD Biosciences).

The various cell populations were sorted from Rag1^−/−, C57BL/6, and BALB/c mice. The cells were immediately lysed and the RNA extracted using TRIzol reagent (Invitrogen). Reverse transcription was performed with the High Capacity cDNA Reverse Transcription kit (Applied Biosystems). cDNA levels were analyzed by real-time quantitative PCR with the Taqman system (Applied Biosystems). Each sample was assayed in triplicate for the target gene together with 18S rRNA as the internal reference in 25 μl final reaction volume with the Taqman Universal PCR Master Mix and the ABI Prism 7900 Sequence Detection system. The probe set for LAG-3 was designed using Primer Express software and then synthesized by the Hartwell Center for Biotechnology and Bioinformatics (St. Jude Children’s Research Hospital). The relative mRNA frequencies were determined by normalization to the internal control 18S RNA. In brief, we normalized each set of samples using the difference in the threshold cycles (Ct) between the target gene and the 18S RNA: ΔCt_sample = (Ct_sample − Ct_18S). All the samples were compared with mLAG-3 levels in CD4⁺/CD25⁻ T cells (ΔCt_calibration). Relative mRNA frequencies were calculated as 2^ΔΔCt where ΔΔCt = (ΔCt_calibration − ΔCt_sample). Primer and probe set used is: LAG-3 forward primer 5′-TCCGCCTGCGCGTCG-3′, reverse primer 5′-GACCCAATCAGACAGCTTGAGGAC-3′, and probe 5′- 6FAM-CCAGGCCTCGATGATTGCTAGTCCC-BHQ1–3′.

T cells, DCs, and pDCs from the spleens of aged matched (within 4 wk) Lag3^+/+ and Lag3^−/− mice (T cells) or CD45.1.Rag1^−/− and Lag3^−/−:Rag1^−/− mice (pDCs) were positively sorted by FACS. For FACS purifications of pDCs, splenocytes were stained for PDCA-1, CD11c, and B220 expression and sorted as either PDCA-1⁺/CD11c⁺ or PDCA-1⁺/CD11c⁺/B220⁺ on a MoFlo (Cytomation). For purification of CD4⁺CD25⁻ cells, splenocytes and lymph nodes were stained for CD4 and CD25 and separated by FACS using the MoFlo. Unless otherwise stated, 1 × 10⁶ pDCs were transferred into Rag1^−/−:MHCII^Δ/Δ or Lag3^−/−:Rag1^−/−:MHCII^Δ/Δ mice via the tail vein. In the dual transfers, 2 × 10⁶ T cells were transferred via the tail vein 24 h later. At the time points indicated, the mice were sacrificed and single cell suspensions made of the spleens. The cells were stained for CD4, PDCA-1, and CD45.1 expression.

pDCs and T cells were isolated from the spleens of Lag3^+/+ and Lag3^−/− mice by FACS as above and cultured in 96-well round-bottom plates at concentrations ranging from 0.35–2 × 10⁶ cells/well in complete RPMI 1640 with 10% FBS. T cells were activated with either anti-CD3ε/CD28-coupled beads or plate-bound anti-CD3ε. pDCs were stimulated with 1–10 μM CpG type 1668 (Hartwell Center, St. Jude Children’s Research Hospital). Following a 72-h incubation supernatants were collected and for some assays the cells collected, lysed and RNA isolated for quantitative PCR (qPCR) analysis.

Rag1^−/− and Lag3^−/−:Rag1^−/− mice (age and sex-matched) were injected i.v. with 100 μl of 100 μM CpG type 1668 (Hartwell Center, St. Jude Children’s Research Hospital) in PBS. Mice were sacrificed by CO₂ inhalation and the number/percentage of live, splenic, or lymph node-derived pDCs was determined by trypan blue exclusion and flow cytometric analysis following staining with anti-CD11c and anti-PDCA-1.

Rag1^−/− and Lag3^−/−:Rag1^−/− mice were injected i.p. with 200 μl of BrdU (10 μg/ml) (BD Pharmingen) on day 0. The mice received a second injection of 100 μl BrdU (10 μg/ml) 24 h later followed by an i.v. injection of a 100 μl of CpG-1668 (100 μM) 1 h later. The next day, mice were sacrificed by CO₂ inhalation and the spleens and inguinal, mesenteric, axillary, and cervical lymph nodes were removed. The lymph nodes were pooled and processed along with the spleens and stained for pDCs using anti-PDCA-1 and anti-B220. The cells were then stained for BrdU incorporation (BrdU labeling kit; BD Pharmingen).

The sLAG-3 ELISA was performed as described (15).

Statistical analysis was performed using the unpaired Student’s t tests.

To determine the cell types responsible for the high levels of sLAG-3 observed in Rag1^−/− and Rag1^−/−:Il2rg^−/− mice, we stained splenocytes from Rag1^−/− mice for MAC-1 (CD11b), GR1, CD11c, and B220 expression and cells were sorted based on the scheme depicted in Fig. 1,A. Cells were lysed and the mRNA isolated for real time PCR analysis to determine Lag3 expression (Fig. 1,B). In this and all subsequent analysis, the fold increase in Lag3 expression was determined by comparing the levels of mRNA in the various cell types to a resting, CD4⁺ T cell population devoid of Tregs (CD4⁺/CD25⁻). There was a 250-fold increase in the level of Lag3 mRNA in CD11c⁺/B220⁺ cells compared with resting T cells while expression was minimal in the other cell populations examined. Murine pDCs, the primary cell type expressing these markers, are generally characterized as B220⁺, CD11c^low, and Ly6C⁺ (19, 20, 21). Recently, PDCA-1, 440c, and 120G8 have been described as Abs specific for pDCs (22, 23, 24). To test that the high amount of LAG-3 mRNA observed in pDCs isolated from a Rag1^−/− mouse was not a consequence of the T or B cell deficiency in Rag1^−/− mice, splenocytes from wild-type C57BL/6, BALB/c, and Rag1^−/− C57BL/6 mice were gated on live cells (lymphocyte gate) and sorted into the following populations: 1) CD11c⁻/B220⁺/PDCA-1⁻/NK1.1⁻ (or the panNK marker DX5) cells (predominantly B cells), 2) CD11c⁺/B220⁻/PDCA-1⁻/NK1.1⁻ cells (conventional DCs), 3) CD11c⁺/B220⁻/PDCA-1⁻/NK1.1⁺ cells (recently identified IKDCs (25, 26), and 4) CD11c^low/B220⁺/PDCA-1⁺/NK1.1⁻ cells (pDCs). The results show that there was a 50–100-fold increase in Lag3 mRNA levels in CD11c^low/B220⁺/PDCA-1⁺/NK1.1⁻ pDCs compared with other DC subpopulations. This was consistent across all three strains of mice (Fig. 1 C).

Next, we wanted to compare the level of Lag3 mRNA in pDCs to resting and activated T cells and Tregs. As expected, there was ∼10–15-fold increase in Lag3 mRNA in activated T cells, and resting and activated Tregs compared with resting, naive T cells. Interestingly, this was much lower than the amount of Lag3 mRNA in resting pDCs which was ∼10-fold higher than any T cell population (Fig. 1,D). Unlike T cells, Lag3 mRNA expression did not appear to increase in pDCs following CpG activation suggesting that the high amount of Lag3 in pDCs was constitutive and perhaps maximal (Fig. 1 D).

Given the high amounts of Lag3 mRNA in pDCs, we next assessed LAG-3 cell surface expression. Splenic CD11c^low/B220⁺/PDCA-1⁺ C57BL/6 pDCs were found to express cell surface LAG-3 directly ex vivo (Fig. 1 E). This is noteworthy as constitutive LAG-3 cleavage makes detection of cell surface LAG-3 challenging (9, 10). Taken together, these data show that pDCs constitutively express high levels of LAG-3 and may be the primary source of sLAG-3 found in the sera of Rag1^−/− mice.

What is the role of LAG-3 expression on pDCs? Previous work has shown that LAG-3 is a negative regulator of T cell function (1, 2, 3, 4, 7). To determine whether LAG-3 plays a similar role on pDCs, we first examined the ability of LAG-3 to control pDC expansion in vivo. pDCs characteristically produce large amounts of type 1 IFNs in response to both viral stimulation and oligodeoxynucleotides containing unmethylated CpG motifs through the Toll-like receptors TLR7 and TLR9, respectively (20, 27). To assess the role of LAG-3 on pDCs in vivo, we injected mice with CpG and examined pDC expansion. We used Rag1^−/− and Lag3^−/−:Rag1^−/− mice, which are devoid of T and B cells, to eliminate the possible contribution of LAG-3 from these cells. We injected CpG i.v. into Rag1^−/− or Lag3^−/−:Rag1^−/− mice and at various time points sacrificed the mice to determine the number of pDCs present in the spleen. The data suggested that there was a greater increase in the expansion of pDCs in LAG-3-deficient Rag1^−/− mice relative to Rag1^−/− mice following activation (Fig. 2 A). Although this suggests a direct, cell intrinsic role for LAG-3 in modulating CpG-induced pDC proliferation, we cannot rule out the possibility that the lymphopenic environment disproportionately affects Lag3^−/− pDC expansion. Minimal pDC expansion and MHC class II up-regulation was seen with the TLR7 ligand imiquimod (R837), suggesting it only weakly activates pDCs (data not shown). Curiously, there was a consistent and reproducible decrease in the number of pDCs in the Lag3^−/−:Rag1^−/− mice at 168 h postinjection. It is conceivable that Lag3^−/− pDCs might be over-stimulated, resulting in increased cell death. Alternatively, Lag3^−/− pDCs might migrate out of the spleen and into the periphery at this later time point. Additional experiments would be required to test these possibilities.

We next assessed cell surface MHC class II expression on pDCs to determine whether CpG-induced activation was equivalent. The results suggest that Lag3^−/− and Lag3^+/+ pDCs were activated equivalently as both exhibited comparable increases in MHC class II expression (Fig. 2 B). TLR9 and IDO levels in Lag3^−/− and Lag3^+/+ pDCs as determined by qPCR analysis were also equivalent (data not shown). In addition, cytokine production (e.g., IFNα) following CpG activation appears equivalent (data not shown). Finally, we examined the level of CCR5 and CCR7, which have been shown to be highly expressed on pDCs (28), and found that both Lag3^−/− and Lag3^+/+ pDCs expressed similar levels of these chemokine receptors on both resting and activated pDCs (data not shown). Taken together, these data suggest that LAG-3 has a selective effect on pDC proliferation and homeostasis, synonymous with its effect on T cells.

Next, we wanted to determine whether the ∼3-fold increase in the number of LAG-3-deficient pDCs, 24 h post-CpG stimulation, was directly caused by increased proliferation by performing in vivo BrdU incorporation assays, 24 h poststimulation with CpG (Fig. 2, C and D). Interestingly, there was a >2-fold increase in the total number of BrdU⁺ pDCs in the Lag3^−/−:Rag1^−/− spleens compared with Rag1^−/− mice. Interestingly, there was no difference in the percentage of pDCs in the lymph nodes of Lag3^−/−:Rag1^−/− and Rag1^−/− mice. There was a slight, but not statistically significant increase in the percentage of BrdU⁺ cells in the lymph nodes of Lag3^−/−:Rag1^−/− mice (Fig. 2 D). Collectively, these results show that in the absence of LAG-3, pDC proliferate and expand more in vivo, particularly in the spleen, when stimulated via the TLR9 pathway.

We have previously shown that LAG-3 is up-regulated on T cells following activation (9). Therefore, we looked at LAG-3 expression on pDCs directly ex vivo 24 h post-CpG activation. LAG-3 was constitutively expressed on the cell surface at detectable levels and increased 2-fold following activation (Fig. 3, A and B). Interestingly, this did not appear to correlate with mRNA levels which were unaltered by CpG stimulation (Fig. 1,D). This may be due to the different analysis conditions (72 h in vitro vs 24 h in vivo). Alternatively, we have recently shown that LAG-3 protein is stored intracellularly in T cells and may be rapidly transported to the cell surface following activation (S.-R. Woo, N. Li, and D. Vignali, unpublished observations). It is possible this also occurs with pDCs, and CpG activation may induce LAG-3 transport to the cell surface. Lastly, we have recently shown that LAG-3 expression on T cells is regulated by two transmembrane metalloproteases, ADAM10 and ADAM17, which cleave LAG-3 from the cell surface (14, 15). This raised the possibility that CpG activation of pDCs may increase sLAG-3 production, which is found at significant levels in normal mouse serum (∼180 ng/ml) (15). We collected supernatants from in vitro cultures of pDCs stimulated with CpG and found that activated but not resting pDCs generated a substantial amount of sLAG-3 (Fig. 3 C). Equivalent numbers of activated pDCs generate ∼5 times more sLAG-3 than activated T cells. These data suggest that LAG-3 surface expression on, and generation of sLAG-3 by, pDCs is significantly increased following CpG activation.

We have previously shown LAG-3-deficient mice have a defect in the maintenance of T cell homeostasis (4). We next asked whether LAG-3 also controlled the homeostatic expansion of pDCs. We first determined the number of pDCs in the spleens of 8-wk-old Lag3^+/+:Rag1^+/+, Lag3^−/−:Rag1^+/+, Lag3^+/+:Rag1^−/−, and Lag3^−/−:Rag1^−/− mice. There was a significant increase in the number of pDCs in the Lag3^−/−:Rag1^+/+ mice compared with their wild type controls. However, the absence of LAG-3 had no effect on pDC numbers in Rag1^−/− mice (Fig. 4 A). This suggested that there was no cell intrinsic role for LAG-3 in the maintenance of pDC homeostasis, but rather that deregulation of T cells did cause an overall imbalance in pDC numbers.

These data implied an association between T cell and pDC homeostasis and raised the possibility that pDCs may in turn influence overall T cell numbers. To investigate this possibility, we transferred 5 × 10⁶ Lag3^−/− and Lag3^+/+ T cells into Rag1^−/− and Lag3^−/−:Rag1^−/− recipient mice. Seven days later, the number of T cells present in the spleens was determined. As expected, the LAG-3-deficient T cells expand more than wild-type T cells in Rag1^−/− mice (4) (Fig. 4,B). However, both Lag3^−/− and Lag3^+/+ T cells expanded significantly more in the Lag3^−/−:Rag1^−/− mice compared with the Rag1^−/− mice, with the percentage increase in homeostatic T cell expansion maintained between both backgrounds (Lag3^−/− vs Lag3^+/+ T cells in Lag3^+/+:Rag1^−/− mice = 51% cf. Lag3^−/−:Rag1^−/− mice = 63%). This suggests that the deregulation of pDC homeostasis in Rag1^−/− mice caused increased T cell expansion independent of LAG-3 expression on T cells. To determine whether this deregulated T cell homeostatic expansion was directly due to LAG-3-deficient pDCs, we performed a dual adoptive transfer experiment combining pDCs and T cells. We transferred either Lag3^−/− or Lag3^+/+ CD45.1⁺ pDCs into MHCII^Δ/Δ:Rag1^−/− (CD45.2⁺) mice followed by wild-type CD4⁺ T cells 24 h later. As the MHCII^Δ/Δ recipient mice lack all four classical MHC class II chains due to an 80 kb deletion in the MHC locus (18), the CD4⁺ T cells must interact with the MHC class II⁺ pDCs to undergo homeostatic expansion. Although the Lag3^+/+ pDCs were unable to mediate expansion 7 days post transfer (compared with recipients that did not receive pDCs), there was a significant increase in T cell homeostatic expansion in the Lag3^−/− pDC recipients (Fig. 4 C), suggesting that the deregulated Lag3^−/− pDCs may directly affect T cell expansion.

It is possible that this increased T cell expansion was due in part to increased Lag3^−/− pDC expansion. Although our experiments above suggested that LAG-3 might not influence the steady-state maintenance of pDC homeostasis (see Fig. 4,A), this did not preclude a role for LAG-3 in controlling expansion in lymphopenic hosts. Thus, we assessed pDC expansion following adoptive transfer. We sorted pDCs and bulk cDC (CD11c⁺ DCs lacking pDCs as controls) from Lag3^+/+ and Lag3^−/− CD45.1⁺ mice, and transferred them into Lag3^−/−:Rag1^−/−: MHCII^Δ/Δ mice. Seven days later, the number of pDCs and cDCs in the spleen was determined. There was a ∼3-fold increase in the expansion of Lag3^−/− pDCs compared with their wild-type counterparts (Fig. 4,D). In contrast, there was no difference in the number of Lag3^−/− vs Lag3^+/+ bulk cDCs transferred. This increase in pDCs may partially explain the increase in T cell expansion observed in the adoptive pDC/T cell transfer experiments (Fig. 4 C). However, as evident from T cell adoptive transfers into Rag1^−/− and Lag3^−/−:Rag1^−/− mice, LAG-3 expression on pDCs has a cell-extrinsic role in modulating T cell homeostasis. In summary, these data suggest that LAG-3 acts intrinsically to control pDC homeostasis as well as extrinsically to regulate T cell homeostatic expansion.

Previous studies have suggested that the expression of LAG-3 was restricted to the lymphoid compartments: activated αβ T cells, γδ T cells, Tregs, B cells, and NK cells (6, 9, 12). However, data presented in this study demonstrate that pDCs express LAG-3. It is noteworthy that LAG-3 is uniformly expressed on all pDCs and is not expressed on any other DC subset that we have examined, suggesting the LAG-3 may serve as an additional marker for pDCs. Indeed, pDCs expressed far higher levels of LAG-3 than Tregs and activated effector T cells, which was readily detectable ex vivo. Murine pDCs were identified in 2001 and are found predominantly in T cell-rich areas of secondary lymphoid organs and characteristically produce high levels of type 1 IFNs (19, 20, 21, 27). Interestingly, LAG-3 expression in in situ hybridization studies was restricted to the splenic red pulp, which is coincident with pDC localization (9, 29).

We have previously shown that LAG-3 performs both cell intrinsic and cell extrinsic function on T cells and Tregs (2, 3, 4). Therefore, it was important to determine whether LAG-3 functioned in an analogous manner on pDCs to control their proliferation and homeostasis. We first examined the role of LAG-3 in pDC effector functions. Following activation, pDCs up-regulate MHC class II and CD80/86, produce large amounts of IFNα and modest levels of IL-12 and IL-6 (29). Interestingly, the influence of LAG-3 appears somewhat selective as Lag3^−/− pDCs expand more in vivo following CpG stimulation but do not differ in their expression of MHC class II, TLR9, IDO, or the chemokine receptors CCR5 and CCR7. Furthermore, IFNα production was comparable. It was interesting to note that there was a sharp decrease in the number of pDCs in Lag3^−/−:Rag1^−/− mice 168 h post CpG treatment in vivo, which did not occur in the wild type control mice. This observation was analogous to our previous experiments that showed contraction of Lag3^−/− T cells following in vivo stimulation with Staphylococcus aureus Enterotoxin B (3). Thus, LAG-3 may serve to control the expansion and subsequent contraction of both T cells and pDC following stimulation in vivo.

It has been shown that pDCs turn over rapidly (∼2 wk) (29, 30) and our data suggest that LAG-3 may be an important regulator of this homeostatic process. What was particularly intriguing was the cell extrinsic role of LAG-3 on pDCs in modulating T cell expansion and the influence of Lag3^−/− T cell deregulation on pDC expansion, highlighting the homeostatic interplay between pDCs and T cells. Indeed, these data and our previous observations support the concept of “homeostatic reciprocity” between T cells and pDCs, where each population influences the expansion and homeostasis of the other. Given that T cells undergo greater homeostatic expansion in the presence of LAG-3-deficient pDCs, blocking this activity with anti-LAG-3 mAbs may represent a novel approach for manipulating T cell homeostasis. Indeed, we have previously reported that anti-LAG-3 blocking Ab increases the homeostatic expansion of adoptively transferred T cells in Rag1^−/− mice (4). Thus, it is conceivable that this blocks LAG-3 on both T cells and pDCs, resulting in their reciprocal deregulation and enhanced T cell expansion.

Previous studies have shown that LAG-3 function is mediated through MHC class II interaction (1, 4, 9, 10, 31). Unlike murine T cells, pDCs express MHC class II. Therefore, it is possible that LAG-3/MHC class II interaction could occur in cis- and/or trans on pDCs or in trans with other DC/APC populations. Although resolution of this issue will require further study, at the very least our pDC adoptive transfer experiments into Lag3^−/−:Rag1^−/−:MHCII^Δ/Δ mice suggest that LAG-3 on pDCs can productively ligate MHC class II expressed on pDCs (cis-or trans).

It has been suggested that pDCs have the capacity to suppress immune responses either directly or via induction of Tregs (29, 32). It will be interesting in future studies to assess how LAG-3 might influence these properties. For instance, stimulation of immature pDCs induces the production of IDO which has potent inhibitory activity (33, 34). However, preliminary analysis suggests that IDO mRNA levels in Lag3^+/+ and Lag3^−/− pDCs following activation are identical. Whether LAG-3 contributes to the toleragenic properties of pDCs, and the generation of Tregs will clearly need to be examined in future studies (35, 36).

In conclusion, our data show that LAG-3 is constitutively expressed on pDCs. This expression is the highest on any cell types studied to date regardless of activation status. This also represents the first example of LAG-3 expression outside the lymphoid compartment. It is particularly surprising that a previously defined T cell regulatory protein should be found on a very small DC population but no other DC subset. LAG-3 appears to play an important role in both the homeostatic maintenance and activation-induced expansion of pDCs, as well as a cell extrinsic role in T cell homeostatic maintenance. Our previous studies have shown that Lag3^−/− T cells exhibit delayed cell cycle arrest, so it would be interesting to determine whether a similar defect exists in Lag3^−/− pDCs (2, 3). These data combined with the role of LAG-3 on T cells, in particular Tregs, suggests that LAG-3 may function as a global regulator of both the innate and adaptive immune response. The high, uniform expression of LAG-3 on all pDCs will provide an additional functional marker for studying pDCs and may provide a useful target for therapeutic manipulation.

We are grateful to Karen Forbes for maintenance and breeding of our mouse colonies; to Richard Cross, Jennifer Smith, Greig Lennon, and Stephanie Morgan for FACS; to the staff of the Shared Animal Resource Center at St. Jude for the animal husbandry; and the Hartwell Center for Biotechnology and Bioinformatics at St. Jude for real-time PCR primer/probe synthesis and CpG synthesis and purification.

The authors have no financial conflict of interest.