The isolation of naturally-occurring functional Tregs will be a prerequisite for successful adoptive immunotherapy techniques in humans. Murine adoptive immunotherapy models have demonstrated proof of concept that isolated Tregs can be used as therapy for several autoimmune disorders [29–31]. The isolation of functional Tregs in humans, however, is problematic, in part due to a paucity of specific surface markers. In this report, we demonstrate direct evidence of the surface expression of LRRC32 on freshly-isolated, naturally-occurring, and non-expanded CD4+ CD25hi human Tregs following TCR activation. We furthermore have characterized LRRC32 processing and demonstrate that sorted subsets of freshly isolated Tregs bearing this marker appear more suppressive than subsets lacking this marker.
Previous studies have shown that constitutive over-expression of Lrrc32 in CD25- Teffs can lead to Foxp3 upregulation and that these cells subsequently acquire a suppressive phenotype [8, 25]. Similarly, overexpression of Foxp3 has been shown to result in increased mRNA levels of Lrrc32, suggesting positive feedback between FoxP3 and LRRC32 [8, 25]. Although other groups have shown that surface LRRC32 is highly elevated in expanded activated Tregs compared to CD25- Teffs, we demonstrate here that low levels of intracellular LRRC32 are detectable in naturally-occurring freshly-isolated unstimulated Tregs (Figure 1d) [8, 25].
Previous studies utilizing an antibody generated against amino acids 296-308 of LRRC32 failed to detect LRRC32 on transfected Jurkat cells or on native CD4+CD25hiTregs . However, as shown here, a commercially available anti-LRRC32 monoclonal antibody does recognize surface LRRC32 on transfected HEKs and naturally-occuring freshly-derived Tregs that have undergone stimulation. Furthermore, it detects low intracellular expression of LRRC32 in naturally-occurring freshly-derived Tregs. Failure to detect surface LRRC32 by the antibody raised against peptide 296-308 may be due to competitive occupation by a ligand as this region of LRRC32 corresponds to a loop and has been hypothesized to correspond to a ligand binding site . One proposed ligand that could occupy this site may be LAP, as recently published work has demonstrated an interaction between LAP and LRRC32 [11, 13, 32]. If residues 296-308 of LRRC32 act as a binding site for LAP, occupation of this site may account for the failure of previous LRRC32-specific antibodies to recognize surface LRRC32 expression.
To determine if LRRC32 is sequestered in Tregs, we examined the intracellular expression of LRRC32 in naturally-occurring freshly-isolated unstimulated Tregs by flow cytometry. We show that unstimulated Tregs contain low levels of intracellular LRRC32 protein. However, coupled with our RT-PCR studies showing high levels of Lrrc32 mRNA in unstimulated Tregs relative to Teffs, these data suggest that post-transcriptional mechanisms may play a role in controlling LRRC32 production and expression in non-activated Tregs. Such post-transcriptional controls may be diminished upon stimulation via TCR/CD28 signaling, as indeed, upon stimulation, evidence of increased intracellular LRRC32 protein was evident in Tregs as assessed by flow cytometry (Figure 1d).
In addition, our signal peptide deletion construct studies reveal that the putative signal peptide in LRRC32 is critical for the cell surface expression of LRRC32, consistent with other reports that signal peptides are necessary for surface protein expression . Our data showing that LRRC32ΔSP is transcribed (Figure 4c) but not detected intracellularly (Figure 4b) suggest that LRRC32ΔSP is rapidly broken down in the cytosol or is not translated at detectable levels following transcription. However, mechanisms for the rapid degradation of misfolded proteins exist to maintain cell viability, and as such, cytosolic LRRC32ΔSP, unable to enter the endoplasmic reticulum owing to the lack of a signal peptide, may be translated but rapidly degraded afterwards by processes such as ubiquitination .
Finally, using freshly isolated, non-expanded CD4+CD25hiLRRC32+ Tregs, we show that such cells expressing surface LRRC32 appear to be functionally more suppressive than CD4+CD25hiLRRC32- Tregs. Previous reports have shown that upon activation of Teffs, surface CD62L is usually decreased [35–37]. However, Tregs normally maintain CD62L expression and functional phenotype . Furthermore, previous reports have shown that CD62L+CD4+CD25hi Tregs are more suppressive than their CD62L- counterparts [39, 40].We show here that expression of surface CD62L appears to decrease significantly on LRRC32+ Tregs compared to LRRC32- Treg populations.
Differences in CD62L processing may be responsible for the observed difference in CD62L expression between LRRC32+ and LRRC32- activated Tregs. It has been shown that 90% of CD62L is rapidly cleaved from the surface within 4 hours of T cell activation prior to increasing over the next 48 hours, due to enhanced message stability, before ultimately decreasing due to downregulation of gene transcription . Furthermore, it has been reported that CD62L is rapidly shed in T cells, including Tregs, after activation [42, 43]. In accordance, our data show that unstimulated LRRC32- CD4+CD25hiFoxP3+ Tregs expressed more surface CD62L than stimulated LRRC32+ or LRRC32- CD4+CD25hiFoxP3+ Tregs. However, upon activation, decreases in surface CD62L expression of LRRC32+ versus LRRC32- cells were noted, suggesting that LRRC32+ cells are more activated compared to LRRC32- Tregs. Given that an overnight stimulation is sufficient to induce LRRC32 expression on the cell surface of Tregs, we chose this as our timepoint for phenotypic analysis. However, altering the time course of stimulation may also alter surface Treg marker expression.
Activated Tregs that express LRRC32 may also represent a distinct population of more highly activatable Tregs compared to LRRC32- Tregs . Indeed, our phenotypic studies using LRRC32+ and LRRC32- subsets of Tregs in the context of CD62L expression would appear to support the interpretation that LRRC32+ Tregs, relative to LRRC32- Tregs, are more prone to activation, as shown by increased cleavage of surface CD62L, and that this more highly activated state may translate into increased suppressive activity. Notably, although only a fraction of Tregs expressed LRRC32 upon activation overnight, these cells appeared to be more functionally suppressive than their LRRC32- counterparts.
Most natural FoxP3+ adult Tregs are CD45RO+, and the expression of CD45RO is typically a marker of T cell activation [44, 45]. As LRRC32+ Tregs appear to be more suppressive than LRRC32- Tregs, it is possible that lower expression of CD45RO on LRRC32+ Tregs relative to LRRC32- Tregs may be due to increased auto-suppressive activity by LRRC32+ Tregs compared to LRRC32- Tregs. As noted above, stimulated Tregs demonstrated expected increases in the surface expression of GITR, CD69, and CTLA4. GITR, or glucocorticoid-induced tumor necrosis factor receptor, was originally shown to be highly expressed on unactivated Tregs but relative to Teffs, and its expression was increased upon cell activation [46–48]. It appears that GITR is a co-stimulatory molecule, and although it is preferentially expressed on CD25hi cells, it is also expressed at lower levels on Teffs, and upon activation, Teffs can also upregulate GITR [48, 49]. Hence, the use of GITR as a specific marker for Tregs appears to be limited. CD69 has been described in the context of a CD69+CD4+CD25- Treg subset that does not express Foxp3 but does express surface-bound TGF-β1 in an ERK-dependent manner . Normally, CD69 is upregulated upon T cell activation, and thus expected on our Tregs [50–52]. Since LRRC32 also binds LAP, thereby helping to concentrate TGF-β1 at the cell surface, it is interesting to speculate whether CD69 in Tregs may play a role in helping to upregulate TGF-β1 surface expression in the context of LRRC32 when the latter is available. Although our data did not find any significant differences in the CD69 expression in stimulated LRRC32+ and LRRC32- Treg subsets (Figure 5c), it is possible that part of the reason for the observed difference in suppressive activity between the LRRC32+ and LRRC- Treg subsets may be in part due to synergy between CD69 and LRRC32 via increased surface expression of TGF-β1. CTLA4, or cytotoxic T lymphocyte antigen-4, can inhibit Teff activation via 1) binding B7.1 and B7.2, thereby depriving CD28 on Teffs of the ability to bind these co-stimulatory ligands, 2) inhibiting IL-2 transcription and progression of cells through the cell cycle via inhibition of cyclin D3, cdk4, and cdk6 production, and 3) decreasing the amount of time the TCR is engaged [53–57]. As we did not see any significant differences in CTLA4 expression in the LRRC32+ and LRRC32- Treg subsets, we do not have data to suggest that differences in CTLA4 expression might have contributed to the observed differences in suppressive function in the LRRC32+ and LRRC32- Treg subsets. Clearly, the results in this set of experiments raise many more interesting questions and suggest that the role of LRRC32 in the context of these other cell activation markers is complex.
Previous studies examining LRRC32 and T cell regulation have utilized Teffs transfected with constructs containing either wildtype Lrrc32 or Lrrc32 lacking leucine rich repeat regions, the signal peptide, the cytoplasmic domain, or Lrrc32 with a mutated cytoplasmic residue postulated to be part of a PDZ domain and thus thought to bind an intracellular protein [8, 25]. These experiments were performed to address how LRRC32 may be processed and ultimately function in Treg cells. PDZ domain mutation studies have suggested that the intracellular portion of LRRC32 is critical for surface expression, and other studies examined the LRRC32 deletion mutants in the context of downstream effector molecules such as FoxP3 [8, 25]. These studies concluded that because FoxP3 expression was markedly decreased upon deletion of the leucine rich regions or the signal peptide, these regions were critical for LRRC32 function [8, 25]. However, these studies never demonstrated actual cleavage of the putative signal peptide . Here, we demonstrate via immunoprecipitation and confocal studies that LRRC32 encodes a signal peptide that is cleaved, and upon cleavage, allows mature LRRC32 to reach the cell surface.
It is likely that the LRRC32 signal peptide is cleaved from the newly translocated preprotein by type I eukaryotic endoplasmic reticulum signal peptidase, based upon the amino acid sequence of LRRC32 . The initial amino acid sequence of LRRC32 incorporating a charged N-terminal domain followed by a hydrophobic domain is consistent with published reports of the consensus motif for eukaryotic type I endoplasmic reticulum signal peptidase [58, 59]. Furthermore, the sequence G-L-A at positions 15 though 17 of the preprotein is consistent with the -3, -1 rule, stating that residues at the -3 and -1 positions, relative to the cleavage site, must be neutral and have small side chains .