A major isoform of the E3 ubiquitin ligase March-I in antigen-presenting cells has regulatory sequences within its gene

Regulation of major histocompatibility complex class II (MHC-II) expression is important not only to maintain a diverse pool of MHC-II–peptide complexes but also to prevent development of autoimmunity. The membrane-associated RING-CH (March) E3 ubiquitin ligase March-I regulates ubiquitination and turnover of MHC-II–peptide complexes in resting dendritic cells (DCs) and B cells. However, activation of either cell type terminates March-I expression, thereby stabilizing MHC-II–peptide complexes. Despite March-I's important role in the biology of antigen-presenting cells (APCs), how expression of March-I mRNA is regulated remains unknown. We now show that both DCs and B cells possess a distinct isoform of March-I whose expression is regulated by a promoter located within the March-I gene. Using March-I promoter fragments to drive expression of GFP, we also identified a core promoter for expression of March-I in DCs and B cells, but not in fibroblasts, kidney cells, or epithelial cells, that contains regulatory regions that down-regulate March-I expression upon activation of DCs. Curiously, we found downstream sequence elements, present in the first coding exon of March-I in APCs, that confer regulation of March-I expression in activated APCs. In summary, our study identifies regulatory regions of the March-I gene that confer APC-specific expression and activation-induced modulation of March-I expression in DCs and B cells.

MHC-II (pMHC-II) complexes then move to the APC surface where they can be surveyed by T-cell receptors (TCRs) expressed on CD4 T cells (2,3). Resting DCs are continually generating new and different pMHC-II complexes as part of their "sentinel" function in the immune system, and under steady-state conditions pMHC-II generation and degradation rates are identical (4). Not only must the pMHC-II complexes on the APC surface present the appropriate pMHC-II to the TCR on an antigen-specific T cell, but sustained engagement by pMHC-II with the TCR is necessary for complete T-cell activation (5). DC activation suppresses the process of pMHC-II degradation, effectively "fixing" the pMHC-II repertoire on the DC surface and increasing the likelihood of effective pMHC-II-TCR interactions necessary for T cell activation (6,7).
MHC-II turnover in APCs, including DCs and B cells, is regulated by ubiquitination of MHC-II by the E3 ubiquitin ligase March-I (8). March-I is a transmembrane ubiquitin ligase that preferentially targets internalized pMHC-II for lysosomal degradation (9). Not only is degradation of "old" pMHC-II important to allow T cells to sample APCs for "new" pMHC-II, but the kinetics of pMHC-II turnover can affect affinity maturation of germinal center B cells (10), negative selection of CD4 T cells in the thymus (11), and generation of Foxp3-expressing regulatory T cells (11).
March-I is unusual among ubiquitin ligases because of its unique tissue distribution and expression pattern. March-I is expressed in resting DCs and B cells, and activation of either cell type by a variety of TLR ligands terminates expression of March-I, reduces pMHC-II ubiquitination, and prolongs the half-life of pMHC-II (12,13). March-I is not expressed in nonhematopoietic APCs, including thymic epithelial cells, highlighting lineage-specific regulation of March-I expression in hematopoietic APCs (14,15). Even among APCs, March-I expression can be up-regulated as monocytes and macrophages express almost no March-I unless the cells are treated with interferon-␥ and IL- 10 (16 -19). In addition to controlling pMHC-II expression, March-I also regulates the APC expression of CD86 (12,20,21), a costimulatory protein that is important for the generation of immunogenic, and not tolerogenic, T cells. However, it is likely that most of the biologically significant effects of March-I reside in its ability to regulate pMHC-II turnover as mutant mice expressing non-ubiquitinatable forms of MHC-II (but still containing endogenous March-I and CD86) show phenotypes that are nearly indistinguishable from those of March-I-deficient mice (22).
Despite the well-documented role of March-I in controlling pMHC-II and CD86 expression in APCs (and thereby influencing immunological consequences of pMHC-II/CD86 dysregulation), the mechanisms regulating tissue-specific expression of March-I and TLR-mediated down-regulation of March-I mRNA expression have not been addressed. March-I protein has a very short half-life (23), and for this reason it is likely that March-I expression is regulated primarily at the transcriptional level. In this study, we have examined the March-I gene, identified the March-I isoform present in APCs, and identified the regulatory sequences within the March-I gene that confer tissue-specific expression and activation-induced repression of March-I transcription in DCs.

March-I variant 2 is the primary form of March-I found in DCs
March-I was originally identified using a BLAST search of GenBank TM for human RING-CH domain-containing E3 ligases (20). Both the Vega (24) and Ensembl (25) gene annotation systems indicate that two variants of human March-I and four variants of mouse March-I exist; however, the relative abundance of these March-I variants in professional APCs has not been determined. The organization of the March-I gene as annotated in the Ensembl database is shown in Fig. 1A. The location of the E3 ligase RING domain and transmembrane domains 1 and 2 (present in exons 8, 9, and 10, respectively) are indicated and are common to each variant of March-I.
In an attempt to quantitate March-I mRNA expression in DCs and in mouse brain (a tissue in which ESTs for each March-I variant have been identified), we designed PCR primers that selectively amplify March-I variants 1/4, 2, and 3 ( Fig.  1B). It should be noted that March-I variants 1 and 3 contain a truncated form of exon 7, and variants 1 and 2 contain a truncated form of exon 9, most likely due to the recognition of internal splice-acceptor sequences in these exons (26). Mouse brain contained transcripts encoding at least three March-I isoforms (our exon 5 primer sets cannot distinguish between variants 1 and 4); however, spleen DCs and bone marrow-derived DCs (BM DCs) only contained March-I variant 2 (March-I v2) mRNA (Fig. 1C). Quantitative RT-PCR using RING-domain primers common to each variant demonstrated that the absolute amount of March-I in brain was quite low as compared with that present in spleen DCs (Fig. 1D) and that the small amount of March-I mRNA in brain consisted primarily of variants 1/4 and 3 (Fig. 1E). These variants are capable of downregulating expression of MHC-II in transfected cells; 5 however, whether their function is regulation of MHC-II expression in the brain remains to be determined. Unlike the results in brain, RT-PCR revealed that March-I v2 was the only form of March-I detected in DCs and B cells. These data demonstrate that March-I v2 is the primary, if not the only, isoform of March-I present in DCs and B cells. It is important to note that March-I v2 was the isoform originally identified by Bartee et al. (20) and is the variant that has been used in all March-I overexpression studies published to date.

LPS signaling does not affect March-I v2 mRNA stability
March-I mRNA expression in a variety of APCs is rapidly reduced upon exposure of the cells to the TLR4 ligand LPS (12,13). We explored the possibility that reduced expression of March-I v2 mRNA in DCs exposed to LPS was a consequence of enhanced degradation of pre-existing March-I v2 mRNA. Actinomycin D is an inhibitor of RNA synthesis when used at low concentrations (27), and for this reason actinomycin D treatment has been used as a method to measure mRNA half-life. DCs were left untreated or pretreated with LPS for 1 h before addition of actinomycin D, and the amount of March-I v2 mRNA present over time was determined by quantitative RT-PCR. In agreement with previous findings (12,19), 1-h pretreatment of DCs resulted in a modest but detectable reduction in March-I v2 mRNA present at the time of actinomycin D addition ( Fig. 2A). However, when normalizing the amount of March-I v2 mRNA present in each sample to that present at time 0, the rate of March-I v2 mRNA degradation was unaffected by LPS pretreatment of DCs (Fig. 2B). These data strongly suggest that down-regulation of March-I v2 mRNA expression is regulated transcriptionally and not by post-transcriptional degradation of March-I v2 mRNA.

Analysis of the March-I v2 promoter
We performed 5Ј RACE and EST database searches to identify the transcription start site(s) (TSS) of March-I v2. 5Ј RACE analysis revealed that the 5Ј-UTR of mouse March-I v2 was extremely short and was located only 8.9 Ϯ 4.2 bp upstream of the translation start site in exon 7 ( Table 1). Examination of the mouse EST database confirmed our 5Ј RACE results and confirmed that the 5Ј-UTR of mouse March-I v2 was less than 20 bp upstream of the translation start site.
Based on our identification of the March-I v2 TSS, we generated March-I v2 promoter constructs containing up to 1 kb upstream of the TSS driving expression of GFP in lentiviral vectors (Fig. 3A). Because regulatory elements are often present in the first coding exon of genes (28), we included the first 125 bp of exon 7 in some of our promoter constructs. As a negative control, we used the lentiviral vector containing no promoter elements (but still retaining the GFP open reading frame). Because March-I v2 is only expressed in professional APCs (including DCs) and March-I v2 expression is down-regulated in activated APC lines, bone marrow cells were transduced with lentivirus prior to differentiation of the cells into DCs in vitro using GM-CSF. Because we were primarily interested in monitoring March-I v2 promoter activity in our studies, we simultaneously monitored expression of mRNA for endogenous March-I v2 and March-I v2 promoter-dependent expression of GFP in each sample. Each lentivirus encoded puromycin N-acetyltransferase (puromycin; whose expression was dependent on the phosphoglycerate kinase promoter), and for that reason transduction efficiency in each condition was normalized to the amount of puromycin mRNA present in the sample. Each of the promoter constructs expressed GFP mRNA in DCs well above promoterless control lentivirus levels (Fig. 3B). Sequences in the first 125 bp of exon 7 were not required for basal expression of GFP as expression of the March-I v2 (Ϫ663

Expression of March-I in antigen-presenting cells
to ϩ125)-GFP construct was nearly indistinguishable from that of the shorter March-I v2 (Ϫ663 to ϩ2)-GFP construct. Importantly, deletion of 71 bp upstream of the TSS almost completely prevented March-I v2 promoter activity, highlighting an important role for sequences present within this region for March-I v2 core promoter activity. From these data, we conclude that March-I v2 promoter constructs containing as little as 175 bp upstream of the TSS allow March-I v2 expression in DCs and that sequences present in exon 7 are not required for basal March-I v2 promoter activity.

TLR signaling regulates March-I v2 promoter activity
Activation of APCs leads to down-regulation of March-I expression (12,13,29), and we have shown that, after as little as 2 h of stimulation with LPS, March-I mRNA is reduced by ϳ50% (19). To directly examine whether activation of DCs affects the function of our March-I v2 promoter-GFP reporter constructs, we stimulated immature DCs transduced with March-I v2 promoter-GFP lentivirus with the TLR4 ligand LPS or the TLR9 ligand CpG for 2 h prior to isolating mRNA and measuring expression of GFP in our March-I v2 promoter-GFP constructs. Both LPS and CpG down-regulated expression of the March-I v2 (Ϫ663 to ϩ125), March-I v2 (Ϫ385 to ϩ125), and March-I v2 (Ϫ175 to ϩ125) promoter constructs (Fig. 4A), demonstrating that these promoter constructs contained sequences that respond to DC activation signals. Curiously, GFP expression was not altered by LPS or CpG treatment of cells expressing the March-I v2 (Ϫ663 to ϩ2) promoter construct. Monitoring endogenous March-I v2 expression in these transduced cells confirmed that the cells responded to these activation stimuli like non-transduced cells (Fig. 4B). These data reveal a role for sequences present in March-I v2 exon 7 in the response of the March-I v2 promoter to DC activation by LPS or CpG.

The March-I v2 promoter is only active in APCs
March-I v2 is expressed primarily in APCs (14,15), so we set out to determine whether the March-I v2 (Ϫ663 to ϩ125) promoter construct we identified resulted in cell type-specific expression of our GFP reporter. Whereas we could reliably detect CMV promoter-driven expression of GFP in epithelial cells, kidney cells, and fibroblasts, we were unable to detect any activity of the March-I v2 promoter construct in these cells types above background levels (Fig. 5). By contrast, both the CMV and March-I v2 promoter constructs resulted in expression of March-I v2 mRNA in DCs, although the March-I v2 promoter was significantly weaker than the CMV promoter. The March-I v2 promoter construct was also expressed well in spleen B cells. (Human CMV promoter activity in lentivirus is weak in mouse B cells (30), and for this reason March-I v2 promoter activity seems high as compared with CMV promoter activity.) Although it is not possible to examine March-I v2 promoter activity in every cell type in the mouse, these data strongly suggest that the March-I v2 promoter identified in this study contains the DNA elements necessary for cell-specific expression of March-I v2 in APCs.
In this report, we show that, of the isoforms of March-I annotated in Vega and Ensembl databases, only March-I v2 is expressed in DCs and B cells. Whereas other March-I variants are expressed in brain, RT-PCR demonstrated that the overall expression of March-I in brain is quite low as compared with spleen DCs. Having said that, the amount of March-I v2 mRNA expressed in spleen DCs is also quite small as we have quantitated the amount of March-I v2 mRNA present in resting spleen DCs (using plasmid DNA as a standard) and found that

March-I v2 has a very short 5-UTR
A, 5Ј RACE was performed using mRNA isolated from mouse spleen DCs. The length of the 5Ј-UTR (relative to the initiation codon) for nine independent clones was identified by DNA sequence analysis. B, 12 different March-I ESTs were aligned to the March-I v2 coding sequence, and the length of the 5Ј-UTR (relative to the initiation codon) was determined. The GenBank ID for each EST is indicated. The average Ϯ S.D. for each set of analyses is shown.

Expression of March-I in antigen-presenting cells
in three independent spleen DC samples there is ϳ1 molecule of March-I v2 mRNA per cell. 6 Such a finding is consistent with the nearly undetectable expression of endogenous March-I protein in APCs reported by others (8,12,23).
Because March-I v2 cannot, by definition, be the product of alternative splicing, we reasoned that nucleotide sequences regulating expression of March-I in APCs would be present upstream of the variant 2 TSS. Within only a few hundred bp of the TSS were intragenic promoter sequences that could drive expression of a GFP reporter in DCs but not in kidney cells, epithelial cells, or fibroblasts, suggesting that this region of the

Experimental procedures
Cell isolation and culture C57BL/6 mice were bred and maintained in house at the NCI-Frederick animal facility. All mice were cared for in accordance with National Institutes of Health guidelines and approved by the National Cancer Institute Animal Care and Use Committee. HEK-293 cells, HeLa cells, and mouse embryonic fibroblasts (MEFs) were cultured in DMEM containing 10% fetal bovine serum and 10 mM Hepes, pH 7.4. Cells were subcultured every 2nd or 3rd day and maintained at subconfluent levels. DCs and B cells from mouse spleens were purified by positive selection or negative selection, respectively, using Miltenyi Biotec kits. BM DCs were generated by differentiating mouse bone marrow cells in BM DC medium (RPMI 1640 medium containing 10% fetal bovine serum, 50 M ␤-mercaptoethanol, 10 mM Hepes, pH 7.4, 25 g/ml gentamycin, and 20 ng/ml GM-CSF) as described previously (29). 6 S. Kaul, unpublished observation.

Expression of March-I in antigen-presenting cells Lentiviral promoter constructs
The various DNA fragments spanning the putative March-I v2 promoter were cloned into the promoterless lentiviral plasmid pLV-unsGFP-PGK-Puro (Cellomics). This plasmid encodes destabilized GFP (whose expression is regulated by an experimentally determined promoter) as well as puromycin (whose expression is regulated by the phosphoglycerate kinase promoter). The bp numbering for each promoter construct is shown relative to the March-I v2 transcription start site identified by 5Ј RACE (as described in this report). LV-102 (Ϫ1064 to ϩ125), LV-115 (Ϫ663 to ϩ125), LV-309 (Ϫ317 to ϩ125), and LV-314 (Ϫ175 to ϩ125) were constructed first by generating the PCR fragments using C57BL/6 mouse genomic DNA as template and different forward primers, BIG-S-F, SM-S-F, SMSF-400, and SMSF-300, respectively, and the common reverse primer P4S-R. The antisense construct LV-123 (ϩ125 to Ϫ663) was constructed in a similar fashion by amplifying the DNA fragment using the forward primer SM-AS-F and the reverse primer P4AS-R. The resultant PCR fragment was digested with BamHI and XhoI and cloned into pLV-unsGFP-PGK-Puro. LV-26 (Ϫ385 to ϩ125) and LV-27 (ϩ125 to Ϫ385) were constructed by first PCR-amplifying a 758-bp fragment using PR4-F3 and PR4-R1 as primers and mouse genomic DNA as template and then cloning the PCR product in TA cloning vector pCR4-TOPO (Invitrogen). This plasmid was digested with EcoRI, and a 513-bp EcoRI fragment from this digestion was cloned into pLV-unsGFP-PGK-Puro. LV-306 (Ϫ663 to Ϫ71) and LV-665 (Ϫ663 to ϩ7) were constructed using the common forward primer SM-S-F and reverse primers P4SR-120 and P4SR-ATG, respectively. The resultant PCR fragments were digested with XhoI and BamHI and cloned in these sites in pLV-unsGFP-PGK-Puro. The nucleotide sequences of all March-I promoter lentiviral constructs were confirmed by DNA sequence analysis.

RT-PCR primer sequences
The oligonucleotides used for measuring relative transcript levels in this study were: GFP, 5Ј-caacagccacaacgtctatatcat-3Ј and 5Јatgttgtggcggatcttgaag-3Ј; puromycin, 5Ј-gttcgccgactaccccg-3Ј and 5Ј-agagttcttgcagctcggtg; and March-I RING domain, 5Ј-aagagagcccactcatcacacc-3Ј and 5Ј-atctggagcttttccc-  March-I v2 (Ϫ663 to ϩ125) "forward" promoter, or a negative "control promoter." Expression of GFP, puromycin, and GAPDH mRNA was determined by RT-PCR. Data using HEK-293 cells, HeLa cells, BM DCs, and spleen B cells were normalized to the amount of puromycin mRNA present in the sample, and the control was the promoterless lentivirus. Data using MEFs were normalized to the amount of GAPDH mRNA present in the sample, and the control was lentivirus with the March-I v2 (Ϫ663 to ϩ125) promoter inserted into the expression vector in the backward orientation. In each sample, the amount of GPF mRNA was expressed relative to the amount of mRNA detected using the CMV promoter construct (arbitrarily designated a value of 1). Control experiments using RNA from MEFs, HEK-293 cells, HeLa cells, or BM DCs that were not reverse transcribed revealed negligible contamination of cDNA with host cell genomic DNA or lentivirus DNA. Spleen B-cell cDNA did contain contaminating lentivirus DNA, and therefore RT-PCR data were calculated as 2 ⌬Ct (⌬Ct ϭ Ct plus RT Ϫ Ct minus RT ). The results shown are the average (error bars represent S.D.) of three independent experiments. The expression of GFP driven by the March-I v2 promoter in each cell type was compared with that using the control promoter. *, p Ͻ 0.05; ns, not significant.

Generation of lentiviral stocks
Lentiviral stocks were generated by transfecting subconfluent HEK-293T cells with lentiviral promoter plasmid, EZ-LentiPACK Packaging plasmid, and EZ-Transfx reagent as described in the EZ-LentiPACK Packaging System kit (Cellomics Technology). Lentiviral supernatants were harvested 48 and 72 h after transfection. Viral supernatants were centrifuged briefly (500 ϫ g, 10 min) to remove cellular debris. The approximate titer of viral supernatants was quantitated using Lentivirus Titration XpressCards (Cellomics Technology).

Lentiviral transduction
Lentivirus (10 -100 l of HEK-293 viral supernatant) was added to bone marrow cultures on day 2 of culture in BM DC medium containing 10 g/ml protamine sulfate (Sigma). BM DC medium was changed every other day. Day 7 BM DC cultures were cultured in the presence or absence of 1 g/ml LPS (Sigma) and 2 g/ml CpG-1668 (Invivogen) for 2 h. BM DCs were then washed and harvested at 4°C, and the cell pellet was immediately resuspended in TRIzol (Ambion) and stored at Ϫ80°C.
Primary spleen B cells were isolated by negative selection using Miltenyi Biotec kits and transduced with lentivirus by spinoculation. Lentivirus (250 l of HEK-293 viral supernatant) was added to 8 ϫ 10 6 B cells in 2 ml of RPMI 1640 medium containing 10% fetal bovine serum, non-essential amino acids, the B cell-activating factor BAFF (10 ng/ml), 10 mM Hepes, pH 7.4, and 10 g/ml Polybrene (Sigma) in a tissue culture plate. After 15-min incubation at room temperature, the cultures were subjected to centrifugation at 1100 ϫ g for 90 min and then placed in a 37°C, 5% CO 2 incubator. After 2 days, cells were harvested at 4°C and washed, and the cell pellet was immediately resuspended in TRIzol (Ambion) and stored at Ϫ80°C.
Adherent MEFs, HEK-293 cells, and HeLa cells were cultured in RPMI 1640 medium containing 10% fetal bovine serum and 10 mM Hepes, pH 7.4. Cells were transduced using 10 -100 l of HEK-293 viral supernatant in the above medium containing 10 g/ml protamine sulfate, and the medium was changed after 1 day. The next day cells were harvested at 4°C and washed, and the cell pellet was immediately resuspended in TRIzol (Ambion) and stored at Ϫ80°C.

RNA Isolation, cDNA synthesis, and RT-PCR
Total RNA was isolated from the TRIzol suspensions according to the manufacturer's instructions. cDNA was synthesized using 1-2 g of total RNA and the Superscript III First-Strand Synthesis kit (Invitrogen). RT-PCR was performed using a QuantiStudio6 Flex sequence detection system (Applied Biosystems) and the Quantitect SYBR Green PCR kit (Qiagen).

Quantitation of March-I v2 promoter activity
In experiments monitoring expression of March-I v2 promoter activity in lentivirus-transduced BM DCs, RT-PCR was used to quantitate the amount of GFP and puromycin mRNA present in the sample. Ct values for GFP and puromycin were determined, and data are shown as 2 ⌬Ct (⌬Ct ϭ Ct puromycin Ϫ Ct GFP ). 2 ⌬Ct for the promoterless (blank) vector control was arbitrarily set at 1.0, and all values are expressed relative to the blank vector control.
In experiments monitoring regulation of March-I v2 promoter activity in transduced BM DCs treated with either PBS, LPS, or CpG, RT-PCR was used to quantitate the amount of GFP and GAPDH mRNA present in the sample. Ct values for GFP and GAPDH were determined, and data are shown as 2 ⌬Ct (⌬Ct ϭ Ct GAPDH Ϫ Ct GFP ). In experiments using the March-I v2 (Ϫ663 to ϩ125) and March-I v2 (Ϫ663 to ϩ2) sense constructs, the control virus was a March-I v2 (ϩ125 to Ϫ663) antisense construct, and in experiments using the March-I v2 (Ϫ385 to ϩ125) and March-I v2 (Ϫ175 to ϩ125) sense constructs, the control virus was a March-I v2 (ϩ125 to Ϫ385) antisense construct. The data shown are 2 ⌬Ct for the sense construct/2 ⌬Ct for the antisense construct. The amount of GFP mRNA present after treatment of each sample with LPS or CpG was expressed as a fraction of that present in the PBS-treated control sample.

Statistical analyses
Results were analyzed using two-tailed Student's t tests. p values Ͻ0.05 were considered statistically significant. p values are indicated by * for p Ͻ 0.05, and non-significant differences are indicated by ns.
Author contributions-S. K. and S. K. M. data curation; S. K. and S. K. M. formal analysis; S. K. investigation; S. K. and P. A. R. writingoriginal draft; S. K., S. K. M., and P. A. R. writing-review and editing; S. K. M. and P. A. R. conceptualization; P. A. R. supervision; P. A. R. validation; P. A. R. project administration.