The CD83 reporter mouse elucidates the activity of the CD83 promoter in B, T, and dendritic cell populations in vivo

Abstract

CD83 is the major surface marker identifying mature dendritic cells (DCs). In this study, we report the generation of reporter mice expressing EGFP under the control of the CD83 promoter. We have used these mice to characterize CD83 expression by various immune system cell types both in vivo and ex vivo and under steady-state conditions and in response to stimulation with a Toll-like receptor (TLR) ligand. With those mice we could prove in vivo that the CD83 promoter is highly active in all DCs and B cells in lymphoid organs. Interestingly, this promoter activity in B cells mainly depended on the stage of development, is up-regulated in the late pre-B cell stage, and was maintained on a high level in all peripheral B cells. We also confirmed that CD83 in those cells is mainly intracellular but is up-regulated after TLR stimulation. Otherwise, CD83 promoter activity in T cells seemed to depend on stimulation and could be found mainly in CD4⁺CD25⁺ and CD8⁺CD25⁺ T cells and in CD4⁺ and CD8⁺ memory cells. In addition, we identified the murine homologues of the human CD83 splice variants. In contrast to those in human, those extremely rare short transcripts were never found without the expression of the highly dominant full-length form. So, the murine CD83 surface expression is mainly regulated posttranslationally in vivo. Our CD83 reporter mice represent a useful mouse model for monitoring the activation status and migration of DCs and lymphocytes under various conditions, and our results provide much needed clarification of the true nature of CD83 promoter activity.

The primary cells mediating adaptive responses are T and B lymphocytes and dendritic cells (DCs). A DC's ontogeny, state of differentiation, and degree of maturation governs its expression of a plethora of membrane-bound and soluble molecules that can induce immunostimulatory and immunosuppressive responses (1–3). Maturing DCs up-regulate certain cell surface molecules that greatly enhance their ability to present antigen and activate naive CD4⁺ and CD8⁺ T cells. Among these molecules is CD83, first described by Zhou et al. (4), CD83 is one of the most useful markers for identifying mature DCs capable of activating naïve T cells (5–8). CD83 expression also occurs on certain T cell subsets (9, 10), B cells (10–12), and murine thymic epithelial cells (13, 14). Studies of CD83 transcription have shown that it is mediated by NF-κB during the induction of adaptive responses (15, 16).

CD83 is conserved from fish species to mammals (17), with mouse CD83 sharing 63% amino acid identity with human CD83 (18, 19). To date, two protein isoforms of CD83 have been reported in humans: a membrane-bound form (mCD83) (5) and a soluble form (sCD83) (20). mCD83 is a highly glycosylated surface protein of the Ig superfamily with a molecular mass of 40–45 kDa (5, 21). mCD83 contains an extracellular Ig-like V domain at the N terminus, a short intracellular cytoplasmic domain of 39 aa, and one transmembrane domain (5). In contrast, sCD83 may contain only the extracellular Ig-like domain (20). But the origin of sCD83 is not yet clear. In humans, four different splice variants of CD83 have been sequenced. The largest variant encodes mCD83, whereas all of the smaller transcripts encode putative soluble forms of CD83 (22). However, truncated splice forms of CD83 have yet to be detected in human serum. At least some sCD83 may be generated by proteolytic cleavage of mCD83 (20).

Although the CD83 ligand remains a mystery, analyses of gene-targeted CD83-deficient mice have revealed that thymic CD83 expression is crucial for the maturation of CD4⁺CD8⁺ thymocytes into CD4⁺ T cells (13, 14). In addition, CD83 may regulate the intercellular interactions between DCs and peripheral T and B cells (12, 23–25). In vitro culture of either human or murine lymphocytes in the presence of sCD83 inhibits their proliferation (26, 27). More remarkably, the administration of recombinant human CD83 protein can prevent the onset of experimental autoimmune encephalomyelitis (EAE; a mouse model for multiple sclerosis) and even to cure established EAE disease in vivo (28). Another group has demonstrated that administration of soluble human Ig-conjugated CD83 can delay acute rejection of MHC-mismatched mouse skin allografts (29). To identify the physiological signaling pathways underlying these effects, it will be necessary to conduct studies using CD83 that has been correctly folded and posttranslationally modified in a living organism. Here, we describe a CD83 knockin mouse generated by positioning a reporter cassette consisting of EGFP linked to an internal ribosomal entry site (IRES2; ref. 30) right after the CD83 stop codon. We have examined EGFP expression in various tissues of these CD83 reporter mice and have observed strong CD83 promoter activity early during the differentiation of B cells and DCs. Moreover, this activity is enhanced by inflammatory stimuli. In contrast, CD83 promoter activity is weak in naïve CD4⁺ peripheral T cells and very weak in naïve CD8⁺ peripheral T cells. Our CD83 reporter mouse model is suitable for use in any immunological experiment in which the nature of CD83 signaling is examined or in which the generation, migration, and/or suppression of DC, T, or B cell activation must be followed.

Results

Generation of CD83-IRES2-EGFP Mice.

We created CD83-IRES2-EGFP mice by using standard methods of homologous DNA recombination in embryonic stem (ES) cells. The targeting vector inserted a reporter cassette consisting of EGFP linked to a viral IRES2 sequence positioned right after the stop codon located in exon 5 of the genomic CD83 gene (Fig. 1A). A neomycin (neo) resistance gene introduced after the EGFP stop codon in the targeting vector enabled selection. The entire coding region of the CD83 gene remained intact in the targeting construct. Expression of the targeted CD83 gene generated a bicistronic transcript encoding both CD83 and EGFP. It is unknown whether splicing of murine CD83 mRNA occurs; however, our EGFP reporter is still likely to be expressed because, in humans, all splice forms include exon 5. Targeted IRES2-EGFP-neo insertion was confirmed by Southern blot analysis of genomic DNA from ES cell clones (Fig. 1B). The genotypes of offspring generated by crossing heterozygous CD83-IRES2-EGFP^+/− mice to C57BL/6 mice were confirmed by PCR analysis (Fig. 1C). CD83-IRES2-EGFP^+/− and CD83-IRES2-EGFP^+/+ mice were born at the expected Mendelian frequency and have thrived and reproduced as well as their WT (−/−) littermates. No obvious morphological and developmental abnormalities have been detected in CD83-IRES2-EGFP^+/− or CD83-IRES2-EGFP^+/+ mice of up to 12 months of age.

Fig. 1.

Generation of CD83-IRES2-EGFP reporter mice. (A) CD83 knockin gene targeting strategy. Untranslated regions, filled boxes; ORFs, open boxes; PCR primers, arrowheads; DT-A, diphtheria toxin subunit A. The Southern blot probes a, b, and neo are indicated by underlining. (B) Southern blot analysis of BamHI-digested DNA from ES cell clones. het, heterozygous; wt, wild type; wt band, filled arrow; mutant band, open arrow. (C) PCR screening of tail DNA of four littermate progeny of CD83-IRES2-EGFP^+/− mice backcrossed to C57Bl6 (Bl6) mice. Control, water.

Analysis of Bone Marrow (BM)-Derived DCs (BMDCs) from CD83-IRES2-EGFP Mice.

To confirm the activity of the CD83 promoter and the specificity of EGFP expression in our reporter mice, we generated DCs in vitro from BM precursors of CD83-IRES2-EGFP^+/− mice. Only 5% or fewer of cells in freshly isolated BM expressed EGFP. In contrast, ex vivo fluorescence microscope analysis of spleens from those mice showed a strong EGFP expression pattern (Fig. 2D). CD11c⁻ Gr1⁺ granulocytes and CD11c⁻ CD11b⁺ monocytes from freshly extracted BM did not express EGFP, whether or not they were treated with LPS (1 μg/ml) for 24 h (Fig. 2A). However, after 8–10 days of in vitro culture, large numbers of proliferating and differentiating CD11c⁺ BMDCs were observed that showed an up-regulation of EGFP in a particular subpopulation (Fig. 2B). These EGFP^hiCD11c⁺ cells also showed high levels of the DC activation markers CD80, CD83, CD86, and MHC class II (MHCII), whereas EGFP^low+intCD11c⁺ cells showed low to intermediate expression of these markers (Fig. 2C). To activate the cultured BMDCs, we treated them on culture days 8–10 with LPS (1 μg/ml) for 16–24 h and observed (as expected) that EGFP, CD80, CD86, and MHCII were all up-regulated on all BMDC populations (Fig. 2C). However, EGFP^hiCD11c⁺ BMDCs showed a much greater increase in surface levels of CD80, CD86, and MHCII than did EGFP^low and EGFP^int CD11c⁺ cells. No cells showing CD83 expression in the absence of EGFP expression were detected. Thus, EGFP expression increases with DC maturation and the normal up-regulation of other costimulatory markers in our knockin mice.

Fig. 2.

EGFP expression in untreated and LPS-stimulated BMDCs. (A) Lack of EGFP expression by monocytes or granulocytes isolated from freshly extracted BM from either CD83-IRES2-EGFP^+/− mice (light line) or WT littermates (^−/−, dark line). This result did not change after stimulation with LPS for 24 h. (B) Increasing numbers of CD11c⁺ EGFP⁺ cells in CD83-IRES2-EGFP^+/− BMDC cultures treated with GM-CSF for 4, 6, or 10 days. Gray, CD83EGFP^lo+int; green, CD83EGFP^hi. (C) Surface expression of the DC activation markers CD80, CD83, CD86, and MHCII (I-A^b) on CD11c⁺ BMDCs from CD83-IRES2-EGFP mice was compared with the EGFP^lo population on day 8 (gray line) versus EGFP^hi cells at 22 h post-LPS stimulation (green line). (D) EGFP expression in bright field plus fluorescence microscopic image overlays (×20) of CD83-IRES2-EGFP^+/− (a and b) and WT (c) BMDCs untreated (a) or after 24 h LPS stimulation (a and c). (d) Fluorescence image of WT (Upper) and CD83-IRES2-EGFP^+/− (Lower) spleens ex vivo. (E) Kinetics of EGFP versus mCD83 expression as determined by FACS of CD11c⁺ and CD11c⁻ populations of CD83-IRES2-EGFP^+/− and CD83-IRES2-EGFP^−/− BMDCs that were either left untreated or stimulated with LPS for the indicated times. (F) Comparison of surface CD83 and intracellular (+surface) CD83 after saponin treatment with the kinetics of EGFP expression after LPS stimulation. Results shown are one analysis representative of at least three independent littermate pairs.

We next examined the kinetics of DC activation marker induction in CD83-IRES2-EGFP^+/− BMDCs. The induction patterns of CD80, CD86, and MHCII expression on CD83-IRES2-EGFP^+/− BMDCs were comparable to those observed for BMDCs obtained from WT littermates. However, the kinetics of CD80, CD86, and MHCII induction in the knockin BMDCs were slightly different from those observed for EGFP expression, in that 21% of CD86^hiCD11c⁺ cells and 35% of MHCII^hiCD11c⁺ cells did not express EGFP (Fig. 2C). Nevertheless, after LPS stimulation for 24 h, fluorescence microscopy revealed bright EGFP fluorescence in the cytoplasm of >90–95% of CD83-IRES2-EGFP^+/− BMDCs (Fig. 2D). We then compared the kinetics of EGFP versus CD83 surface expression in LPS-stimulated CD11c⁺ and CD11c⁻ BM cells. Significant increases in EGFP and CD83 expression were detected in knockin CD11c⁺ BMDCs by 2 h post-LPS stimulation (Fig. 2E). A small population of EGFP⁺CD83⁻ cells was present among untreated knockin CD11c⁺ BMDCs and in CD11c⁺ BMDCs that had been LPS-stimulated for only 45 min, but these latter cells very rapidly became CD83⁺ as the LPS stimulation progressed. After 2 h of LPS treatment, the up-regulation of both EGFP and CD83 in knockin CD11c⁺ BMDCs appeared to be linear (Fig. 2E). This up-regulation pattern was confirmed by using real-time PCR and Western blotting [supporting information (SI) Fig. S1 A and B]. Surface CD83 or EGFP expression was detected in only a small number of unstimulated CD11c⁺ BMDCs, but never in unstimulated or LPS-stimulated CD11c⁻ populations (Fig. 2E). Further FACS analysis investigating surface versus intracellular CD83 expression in permeabilized cells confirmed that the EGFP expression correlates the total CD83 expression (Fig. 2F). To demonstrate that the immune system was functionally normal in our CD83-IRES2-EGFP knockin mice, we performed mixed lymphocyte reactions and measured T cell proliferation. LPS-stimulated BMDCs from CD83-IRES2-EGFP^+/− mice showed the same capacity as LPS-stimulated BMDCs from control littermates to stimulate BALB/c splenocytes in co-cultures (Fig. S1C). In addition, cultures of WT and CD83-IRES2-EGFP^+/− BMDCs showed identical yields and equivalent viability, as determined by trypan blue or Annexin V staining (data not shown). We also analyzed cDNAs of BMDCs, splenocytes, splenic B cells, thymocytes, and BM cells from WT and CD83-IRES2-EGFP^+/− mice. We identified very low expression levels of spliced variants of the CD83 mRNAs homologous to the described human transcripts in those cell types (Fig. S2). In our hands, LPS activation of BMDCs did not show a relevant shift in the expression profile of those splice forms (Fig. S2B).

CD83 Promoter Activity in Vivo in the Absence of Stimulation.

We next used flow cytometry to analyze CD83 promoter activity in the BM, thymus, spleen, and lymph nodes (LNs) of CD83-IRES2-EGFP^+/− mice under steady-state conditions in vivo. About 5% of total BM cells expressed EGFP (Fig. 3A), whereas <5% of thymic cells were EGFP⁺ (Fig. 3B). In the spleen, 35–50% of total splenocytes expressed EGFP (Fig. 3C). About 50% of LN cells were also EGFP⁺ (Fig. 3D). These high levels of EGFP expression in the spleen and LNs were unexpected but consistent in all animals tested. Further investigation of the BM revealed that <1% of CD8⁺ T cells but ≈20% of CD4⁺ T cells expressed EGFP (Fig. 3A). As well, <1% of B220^intCD43⁺ and up to 3% of B220^intCD43⁻ BM cells showed CD83 promoter activity (Fig. 3A). All EGFP⁺B220^intCD43⁻ BM cells were also IgM⁺ (data not shown). More than 90% of B220^hi cells in the BM expressed EGFP and >95% of B220^hiIgM⁺ IgD⁺ B cells were EGFP^hi (Fig. 3A). The low EGFP expression in the thymus was mainly caused by a population of CD4⁻CD8⁻TCRγ δ⁻TCRβ⁻ cells; ≈40% of these cells were EGFP⁺ (Fig. 3B). Further analysis revealed that this thymic EGFP⁺ cell population was composed of 15–35% B220⁺CD11c⁻ B cells, 5% CD11c⁺ DCs, and 60% CD11c⁻DEC-205⁺ thymic epithelial cells (data not shown). CD4⁺CD8⁺ thymocytes were EGFP⁻ but up to 5% of CD4⁺CD8⁻ and CD4⁻CD8⁺ thymocytes were EGFP⁺ (Fig. 3B). Similarly, ≈5% of CD4⁺CD25⁺CD127⁻ regulatory T cells (Tregs) present in the thymus expressed EGFP (Fig. 3B). Only very low numbers (1% or less) of thymic CD4⁻CD8⁻TCRαβ⁺ cells and thymic γδ⁺ T cells were EGFP⁺ (Fig. 3B).

Fig. 3.

CD83 promoter activity in murine tissues. Samples of BM, thymus, spleen, and LNs were assessed for the presence of EGFP-expressing cells using flow cytometric analysis of immunostained cells. Results shown are representative of the levels of EGFP expression detected in the indicated gated subpopulations obtained from four or more littermate pairs of CD83-IRES2-EGFP^+/− and CD83-IRES2-EGFP^−/− mice. Monocyte, DC, B cell, and T cell subsets and their activation status were determined based on the expression of various cell surface markers. The dot blots show total EGFP⁺ cells in BM (A), thymus (B), spleen (C), and LNs (D) as related to forward scatter. The histograms show EGFP expression in the indicated gated cell subpopulations from CD83-IRES2-EGFP^+/− mice (dark line) versus their WT littermates (−/−, light line) in %max.

The high level of EGFP expression in the spleen and LNs was largely caused by the B220⁺ B cell population. In the spleen, 65% of IgM⁺IgD⁻ and almost 90% of IgM⁺IgD⁺ B cells showed EGFP expression. Almost all CD23^hiCD21^loIgM^lo follicular B cells and CD23^hiCD21^hiIgM^hi marginal zone precursor cells (transitional 2, T2) and >80% of CD23^loCD21^loIgM^hi transitional 1 (T1) and CD23^loCD21^hiIgM^hi marginal zone B cells were EGFP⁺ (Fig. 3C). About 20% of CD3⁺CD4⁺ splenic T cells were EGFP⁺ but only 3% of CD3⁺CD8⁺ splenic T cells expressed EGFP at all and these were EGFP^lo. Further examination of T cell subsets revealed that EGFP was expressed by only 5–10% of CD62L⁺CD44⁻ naïve CD4⁺ and CD8⁺ T cells, but that almost 60% of central and >30% of effector memory CD4⁺ T cells, and 20% of central and ≈10% of effector memory CD8⁺ T cells, were EGFP⁺ (Fig. 3C). In general in the spleen, the total EGFP expression in the T cell compartment was always lower than in the B cell compartment. With respect to splenic DCs, 70% of CD11b⁺CD8⁻ DCs, >80% of CD11b⁻CD8⁺ DCs and 88% of CD11c⁺B220⁺ Gr-1⁺ plasmacytoid DCs (pDCs) were EGFP⁺ under steady-state conditions (Fig. 3C). In contrast, only 2–3% of CD11c⁻CD11b^hi splenocytes were EGFP⁺ (Fig. 3C). In the LNs, the dominant EGFP-expressing population was again the B220⁺ B cells. About 80–90% of IgM⁺IgD⁻ LN B cells and >90% of IgM⁺IgD⁺ LN B cells were EGFP⁺ (Fig. 3D). Among LN T cells, ≈9–10% of CD3⁺CD4⁺CD25⁻CD69⁻ naïve T cells but <2% of CD3⁺CD8⁺CD25⁻CD69⁻ naïve T cells expressed EGFP (Fig. 3D). About 78% of CD4⁺CD25⁺CD69⁺ activated T cells were EGFP⁺ but only 58% of CD8⁺CD25⁺ activated T cells showed some level of EGFP expression (Fig. 3D). With respect to LN DCs, almost all (87–100%) were EGFP⁺. Indeed, EGFP expression by CD8⁺ DCs and pDCs in LNs was up to one log greater than EGFP expression by LN B cells or LN CD8⁻ DCs (Fig. 3D). Less than 4% of CD11c⁻CD11b^hi LN cells expressed EGFP (Fig. 3D).

Our experiments with CD83-IRES2-EGFP^+/− BMDCs in vitro showed a clear expression pattern in which mCD83 was present on the surface of EGFP⁺ cells. However, we were unable to replicate this pattern in vivo when we examined freshly isolated cell populations with a commercially available anti-CD83 antibody (Michel-17) (31). Consistent with data published by other groups (10, 32), we found that neither freshly isolated splenic DCs nor CD3⁺CD8⁺ T cells expressed substantial amounts of mCD83 (Fig. 4A). Furthermore, only 10–15% of freshly isolated CD3⁺CD4⁺EGFP⁺ T cells and EGFP⁺B220⁺ B cells expressed mCD83. EGFP⁻ DCs, T cells, and B cells showed no anti-CD83 binding (Fig. 4A).

Fig. 4.

Correlation between CD83-EGFP and CD83 surface expression and CD83-EGFP up-regulation after LPS challenge in vivo. (A) Flow cytometric analysis of CD83 expression (blue) by EGFP⁺ versus EGFP⁻ splenocytes from untreated CD83-IRES2-EGFP^+/− mice versus isotype control (black). PE, phycoerythrin. (B) EGFP expression in mouse spleens at 22 h after i.p. injection of LPS (numbers show percentage of EGFP⁺ cells; gray line, PBS-treated −/− mice; black line, LPS-treated −/− mice; blue line, PBS-treated +/− mice; red line, LPS-treated +/− mice). Results shown are one analysis representative of two independent littermate pairs.

CD83 Promoter Activity After LPS Challenge in Vivo.

We next investigated whether CD83 promoter activity in our reporter mice could be further induced by treatment with a Toll-like receptor (TLR) ligand. To this end, we i.p.-injected groups of CD83-IRES2-EGFP^+/− mice with LPS (100 μg/ml PBS) and analyzed their splenocytes after 22 h. As controls, we i.p-injected CD83-IRES2-EGFP^+/− mice with PBS and CD83-IRES2-EGFP^−/− littermates of the knockin mice with LPS or PBS (Fig. 4B). Flow cytometric analysis revealed that, despite the high base level of CD83 promoter activity in the spleen of unstimulated (PBS) control mice, the number of EGFP⁺ cells in this organ almost doubled after LPS injection (Fig. 4B). More detailed screening demonstrated that EGFP expression by B220⁺IgM⁺ splenic B cells of LPS-stimulated knockin mice increased by ≈20% such that >90% of splenic B cells were EGFP⁺. In contrast, only ≈70% of B cells in PBS-treated knockin mice were EGFP⁺. In addition, the total EGFP fluorescence in LPS-treated knockin splenic B cells was elevated. Thus, not only the number of EGFP⁺ cells but also the level of EGFP expression per individual cell were up-regulated in response to LPS (Fig. 4B). CD11c⁺ splenic DC populations of LPS-treated knockin mice demonstrated a similar increase in CD83 promoter activity compared with controls. Elevated EGFP was observed in >90% of total CD11c⁺ splenic DCs in the mutants (Fig. 4B). However, LPS stimulation did not significantly increase EGFP expression in either CD4⁺ or CD8⁺ T cells of knockin mice (Fig. 4B). No EGFP or CD83 expression was detected in either neutrophils or BM monocytes either under steady-state conditions or after LPS injection (data not shown).

Discussion

The murine CD83 promoter has not been precisely defined, which obviated the creation of a transgenic animal. To elucidate CD83 promoter activity in immune system cells, we generated CD83 knockin reporter mice by using a cloning strategy that minimized interference with CD83 function (Fig. 1). Expression of the targeted CD83 gene successfully produced a bicistronic transcript encoding both CD83 and EGFP.

We also confirmed next to the by far dominant CD83-TM RNA the murine homologues to the human CD83-a, CD83-b, and CD83-c forms (Fig. S2). As in human, those shorter splice forms were extremely rare transcripts. We could not find a significant difference in ratios of expression profiles before and after stimulation with LPS (Fig. S2). So, all EGFP-expressing cells reflected mainly the expression of CD83-TM, because of the very low RNA levels of the short types, whereas no analyzed cell type showed only the shorter splice forms.

After confirming the specificity of our reporter gene under controlled conditions of BMDC culture, we showed that all EGFP^hi BMDCs derived from our knockin mice were also positive for CD80, CD86, and MHCII expression, as expected for mature, fully activated DCs (Fig. 2B). We also identified an EGFP⁺mCD83⁻ DC population that represents immature DCs, consistent with previous reports in which intracellular pools of CD83 were found in the Golgi complex and endocytic vesicles of immature and mature DCs (33, 34). However, in contrast to the findings of Cao et al. (34), we did not detect CD83 promoter activity in CD11c⁻ monocytes/macrophages (Fig. 2A). With respect to TLR engagement, LPS stimulation of BMDCs significantly induced EGFP and mCD83 to levels that further increased over time in parallel with CD80, CD86, and MHCII (Fig. 2 B, E, and F). Interestingly, although these DC activation markers were all highly induced by TLR signaling, they differed in their expression kinetics (Fig. 2C). BMDCs could be CD86^hi and MHCII^hi at the same time as they were EGFP⁻, but CD80⁺ and CD83⁺ cells were invariably EGFP⁺. Functionally, our knockin BMDCs showed normal viability and T cell activation capacity.

When we screened our knockin mice for in vivo CD83-EGFP expression, we noted weak CD83 promoter activity in BM (Fig. 3A) and thymus (Fig. 3B) but surprisingly strong CD83 promoter activity in spleen (Fig. 3C) and LNs (Fig. 3D). The main sources of CD83-EGFP expression in the latter tissues were the resident B cell populations. In contrast to CD83 expression by BMDCs and T cells, CD83 expression by B cells appeared to depend on differentiation stage rather than activation status because all B cell populations in the knockin spleen and LNs were highly positive for EGFP. In BM, pro-B and early pre-B cells showed no EGFP expression (Fig. 3A) but CD83 promoter activity was up-regulated in late pre-B cells. The majority of naïve immature B cells were EGFP⁺, as were nearly 100% of recirculating B cells (Fig. 3A). With respect to the thymus, most thymocytes did not express CD83 during thymic selection (Fig. 3B), but B220⁻CD11c⁻DEC-205⁺ thymic stromal cells, which have been previously shown to be CD83⁺ (13), recirculating B cells, and DCs were EGFP⁺ (data not shown).

In contrast to DCs and B cells, the knockin mice demonstrated an activation-dependent promoter activity in T cells. Only 5–10% of naïve CD4⁺ and CD8⁺ T cells were EGFP⁺, but up to 80% of CD4⁺ and 45% of CD8⁺-activated CD25⁺ T cells showed increased EGFP expression. In addition, 30% of effector memory cells, 60% of central memory CD4⁺ T cells, and 10–20% of memory CD8⁺ T cells activated the CD83 promoter (Fig. 3 C and D). Lastly, ≈5% of thymic Tregs and 30–40% of splenic and LN Tregs were EGFP⁺, implying that CD83 alone cannot be used to distinguish these cells from the large numbers of EGFP⁺-activated conventional T cells in these organs.

The unexpectedly high level of CD83-EGFP expression we found in BMDCs stands in contrast to the much lower levels of in vivo mCD83 expression reported by other groups (6, 9, 11, 13) and confirmed by us (Fig. 4A). Only ≈10% of freshly isolated EGFP⁺ B cells and EGFP⁺CD4⁺ T cells, and no EGFP⁺ DCs, showed mCD83 expression. Thus, CD83 induction occurs early during an immune response but the protein appears to be confined intracellularly in a much wider spectrum of cells than previously appreciated, with major concentrations being found in B cells, DCs, and CD4⁺ T cells. Recent reports have demonstrated that CD83 not only colocalizes with MHCII molecules but also modifies their turnover rate in B cells, T cells, and DCs (10, 35). In addition, CD83 is required for the viability of B and CD4⁺ T cells (10, 35). Thus, the high levels of CD83 promoter activity that we observed in both immature and mature B cells may not only prepare these cells for activation but also provide a survival signal. Furthermore, the number of CD83⁺ B cells and DCs and their relative expression of this marker were increased by LPS treatment of the knockin mice (Fig. 4B). (As expected, there was no further CD83 induction in T cells, which lack the relevant TLR.) The up-regulation of CD83 promoter activity in LPS-treated CD11c⁺ DCs was massive compared with that in B cells (Fig. 4B). If CD83 indeed controls the MHCII turnover rate, this dramatic elevation in CD83 activity could explain why mature DCs are so much more efficient than immature DCs in presenting antigens. Almost all splenic CD8⁻ DCs were EGFP⁺ but expressed much less CD83 per cell than did CD8⁺ DCs and pDCs. Thus, it may be that CD83 is involved in the efficient cross-presentation of antigens to T cells, a capacity unique to CD8⁺ DCs (36).

Prazma et al. (10) have reported the up-regulation of CD83 on CD4⁺ and CD8⁺ T cells. Their observation that CD83 up-regulation does not occur until 4–6 h after activation can be easily explained by results showing that neither CD83 promoter activity nor preformed intracellular CD83 molecules are present in naïve T cells. Prazma et al. further described a loss of viability of CD83-deficient CD4⁺ T cells. Regulation of MHCII expression is thought to limit T cell clonal expansion and thus control T cell responses (37). However, whether CD83 truly regulates MHCII expression and thus T cell responses requires additional investigation. Another unresolved issue is the definition of the differences between the CD83⁺ and CD83⁻ memory T cell populations. With our CD83-IRES2-EGFP reporter mice, we should be able to answer these questions and monitor the activation and migration of DCs and T and B cell subpopulations during immune responses in vivo.

Materials and Methods

Generation of CD83-IRES2-EGFP Reporter Mice.

To generate the CD83-IRES2-EGFP targeting vector, we used the Expand High Fidelity PCR System (Roche) to amplify two DNA fragments from E14K ES cell (129/Ola) genomic DNA. The long homology arm was a 6-kb segment starting upstream of CD83 exon 3 and containing exon 4 and the ORF region of exon 5, including the CD83 stop codon (Fig. 1A). The short homology arm was a 2-kb segment containing the 3′ terminal part of exon 5 and its downstream sequence. The PCR products were cloned into a modified pSPUC_LFneoDTA vector containing the diphtheria toxin subunit A cassette, an IRES2-EGFP sequence (derived from pIRES2-EGFP; Clontech), followed by a neo resistance cassette. We electroporated E14K ES cells with the targeting construct and selected transfectants in G418. Five ES cell clones that had undergone homologous recombination (as confirmed by Southern blot analysis) were used to generate CD83-IRES2-EGFP knockin mice using a standard protocol (32). CD83-IRES2-EGFP^+/− mice were then backcrossed extensively to C57BL/6 mice (Jackson). CD83-IRES2-EGFP^+/− mice and control littermates from backcross generations F₃–F₆ were used for all experiments. All animal experiments were performed according to protocols approved by the Animal Care Committee of the University Health Network (Toronto). For Southern blot analyses, we isolated genomic DNA from E14K ES cells and PCR-amplified DNA segments representing parts of CD83 intron 2 (probe c, 455 bp) and the 3′ UTR following exon 5 (probe a, 674 bp). A neo-specific probe (603 bp) was also used for Southern analysis. Germ-line transmission of the mutant allele was determined by PCR screening of the mice using DNA extracted from tails via ethanol precipitation (WT, 160 bp; knockin, 280 bp).

Generation of BMDCs.

BMDCs were generated from CD83-IRES2-EGFP^+/− and CD83-IRES2-EGFP^−/− knockin mice and C57BL/6 controls as described (38). BMDCs were either left untreated or incubated with 1 μg/ml LPS (Sigma) in RPMI medium 1640/10% FCS medium for 22–24 h. CD83 expression was determined by flow cytometry and fluorescence microscopy as described below.

Viability Assays.

Cell viability assay was done by using trypan blue with Vi-CELL XR cell viability analyzer with Vi-CELL XR 2.03 software (Beckman Coulter). Cell apoptosis has been measured by translocation of the membrane phospholipid phosphatidylserine by using phycoerythrin-conjugated Annexin-V following the protocols of the manufacturer (BD Bioscience).

Lymphocyte Activation Assays.

For mixed leukocyte reactions, titrated numbers (3,800–15,000) of day-11 cultured BMDCs were cocultured for 48–72 h in a 96-well flat-bottom plate (Falcon) with 2 × 10⁵ splenocytes isolated from BALB/c mice (Jackson). Cultures were then pulsed with 1 μCi per well [³H]methyl-thymidine (PerkinElmer) overnight for 12–16 h to measure proliferation. Plates were harvested with a Filtermate harvester (PerkinElmer), and thymidine incorporation was determined by using Top Count NXT (PerkinElmer). All samples were analyzed in triplicate.

Cloning and Sequencing of CD83 mRNA Splice Products.

cDNAs from total RNA of isolated murine cells were synthesized as described (22). CD83 cDNA was amplified with primers binding to exon 1 (5′-GCCTCCAGCTCCTGTTTCTA-3′) and exon 5 (5′-TGGGAAAATGCTTTGTAGTCG-3′) as follows: 15 min 94°C, 36–40 cycles of 30 s 94°C, 30 s 55°C, 90 s 72°C, and a final 10-min 72°C step. The amplified fragments were subcloned into pCR2.1 (Invitrogen) and sequenced.

Flow Cytometric Analysis and Fluorescence Microscopy.

Flow cytometry was carried out on samples of BM, spleen, thymus, LNs, peritoneal fluid, or BMDCs according to standard protocols. All antibodies were purchased from BD PharMingen, except anti-CD83 (Michel-17; Ebioscience). Analyses of immunostained cells were performed by using a FACS Canto with CellQuest (BD Bioscience) software. For fluorescence microscopy, BMDCs derived from CD83-IRES2-EGFP^+/− mice were stimulated in vitro with LPS as described above and analyzed 22 h later for EGFP expression with a Zeiss Axiovert 200M inverted fluorescence microscope with deconvolution software (Zeiss).