Extensive repertoire of membrane-bound and soluble dendritic cell-specific ICAM-3-grabbing nonintegrin 1 (DC-SIGN1) and DC-SIGN2 isoforms. Inter-individual variation in expression of DC-SIGN transcripts.

Expression in dendritic cells (DCs) of DC-SIGN, a type II membrane protein with a C-type lectin ectodomain, is thought to play an important role in establishing the initial contact between DCs and resting T cells. DC-SIGN is also a unique type of human immunodeficiency virus-1 (HIV-1) attachment factor and promotes efficient infection in trans of cells that express CD4 and chemokine receptors. We have identified another gene, designated here as DC-SIGN2, that exhibits high sequence homology with DC-SIGN. Here we demonstrate that alternative splicing of DC-SIGN1 (original version) and DC-SIGN2 pre-mRNA generates a large repertoire of DC-SIGN-like transcripts that are predicted to encode membrane-associated and soluble isoforms. The range of DC-SIGN1 mRNA expression was significantly broader than previously reported and included THP-1 monocytic cells, placenta, and peripheral blood mononuclear cells (PBMCs), and there was cell maturation/activation-induced differences in mRNA expression levels. Immunostaining of term placenta with a DC-SIGN1-specific antiserum showed that DC-SIGN1 is expressed on endothelial cells and CC chemokine receptor 5 (CCR5)-positive macrophage-like cells in the villi. DC-SIGN2 mRNA expression was high in the placenta and not detectable in PBMCs. In DCs, the expression of DC-SIGN2 transcripts was significantly lower than that of DC-SIGN1. Notably, there was significant inter-individual heterogeneity in the repertoire of DC-SIGN1 and DC-SIGN2 transcripts expressed. The genes for DC-SIGN1, DC-SIGN2, and CD23, another Type II lectin, colocalize to an approximately 85 kilobase pair region on chromosome 19p13.3, forming a cluster of related genes that undergo highly complex alternative splicing events. The molecular diversity of DC-SIGN-1 and -2 is reminiscent of that observed for certain other adhesive cell surface proteins involved in cell-cell connectivity. The generation of this large collection of polymorphic cell surface and soluble variants that exhibit inter-individual variation in expression levels has important implications for the pathogenesis of HIV-1 infection, as well as for the molecular code required to establish complex interactions between antigen-presenting cells and T cells, i.e. the immunological synapse.

The dissemination of human immunodeficiency virus-1 (HIV-1) 1 and establishment of infection within an individual involve the transfer of virus from mucosal sites of infection to T cell zones in secondary lymphoid organs. How this happens is not precisely known. However, there is growing support for the notion that dendritic cells (DCs) present within the mucosal sites may play a central role in this process (1)(2)(3)(4)(5)(6)(7)(8)(9)(10)(11)(12)(13)(14)(15). The normal function of DCs is to survey mucosal surfaces for antigens, capture the antigens, process captured proteins into immunogenic peptides, emigrate from tissues to the paracortex of draining lymph nodes, and present peptides in the context of MHC (major histocompatibility complex) molecules to T cells (1). It is now generally believed that HIV-1 may subvert this normal trafficking process to gain entry into lymph nodes and access to CD4 ϩ T cells. There is also evidence demonstrating that productive infection of DCs and the ability of DCs to capture virus with subsequent transmission to T cells is mediated through two separate pathways (Refs. 5 and 8; reviewed in Refs. 3 and 15). Thus, strategies designed to block mucosal transmission of HIV will require a clear understanding of the molecular determinants of not only virus infection but also of virus capture by DCs or other cell types that can subserve a similar function.
Two recent reports by Geijtenbeek et al. (16,17) demonstrated that a mannose-binding, C-type lectin designated as DC-SIGN (DC-specific, ICAM-3 grabbing, nonintegrin) may play a key role in DC-T cell interactions as well as in HIV pathogenesis. First, by binding to ICAM-3 expressed on T cells, DC-SIGN is thought to facilitate the initial interaction between DCs and naive T cells (17), setting the stage for subsequent critical events that lead to antigen recognition and the formation of a contact zone termed the immunological synapse (15,18). Second, HIV-1 may exploit DC-SIGN for its transport via DCs from mucosal surfaces to secondary lymphoid organs rich in activated memory CD4 ϩ T cells that express CC chemokine receptor 5 (CCR5). Unlike CCR5, the major coreceptor for HIV-1 cell entry (19), DC-SIGN is not a coreceptor for viral entry. Geijtenbeek et al. (16) confirmed an earlier observation that DC-SIGN is an HIV-1 envelope (gp120)-binding lectin (20) and extended significantly this finding by showing that it promotes efficient infection in trans of cells that express CD4 and CCR5. This delivery and subsequent transmission of HIV in a DC-SIGN-dependent manner to viral replication-permissive T cells may play a major role in viral replication, especially at low concentrations of HIV (16).
Our interest in DC-SIGN stems from our studies that focus on understanding the host genetic determinants of HIV-1 pathogenesis. For example, we have demonstrated that polymorphisms in the gene for CCR5 influence the rate of disease progression in infected adults and children and in mother-tochild transmission (21,22). Because of the apparent role of DC-SIGN in HIV-1 pathogenesis and DC-T cell interactions, we hypothesized that mutations influencing the gene expression of this molecule and/or its interactions with HIV-1 gp120 or ICAM-3 could have an impact on the pathogenesis of HIV-1 infection. As a first step in testing this hypothesis, we elucidated the gene and mRNA structure as well as the expression pattern of DC-SIGN.
In this study, we identified another highly homologous gene designated here as DC-SIGN2 and made the surprising observation that plasticity of the DC-SIGN1 (original version) and DC-SIGN2 gene generates a wide repertoire of DC-SIGN-1 and -2 transcripts. Interestingly, in addition to DC-SIGN1 (CD209) and DC-SIGN2 (CD209L), the low affinity immunoglobulin ⑀ Fc receptor (CD23) also maps to chromosome 19p13.3 forming a cluster of highly related genes that all undergo complex alternative splicing events (23,24). In contrast to previous reports (16,17), we show that the mRNA expression of DC-SIGN1 (original version) is not restricted to DCs but is broader and includes placenta, PBMCs, and THP-1 monocytes. We also found that there was cell maturation and/or activation-induced differences in DC-SIGN1 mRNA expression levels. By using a DC-SIGN1-specific antiserum, we found that DC-SIGN1 was expressed on the endothelial cells of the placental vascular channels and also coexpressed with CCR5 in the placental macrophages. Abundant DC-SIGN2 mRNA expression was detected in the placenta, but significantly less in THP-1 monocytic cells and DCs, whereas mRNA expression in resting or activated PBMCs was not detected. Notably, there was interindividual variation in the expression levels as well as the repertoires of DC-SIGN1 and DC-SIGN2 transcripts expressed. While this paper was being prepared for submission and was in review, Soilleux et al. (25) described a DC-SIGN homologue designated as DC-SIGNR that is identical to the prototypic membrane-associated DC-SIGN2 described herein, and Pohlmann et al. (26) showed that DC-SIGNR binds to HIV/SIV and activates infection in trans. Thus, our discovery of an extensive repertoire of DC-SIGN-1 and -2 transcripts with variable expression levels may have important implications for the pathogenesis of HIV-1 infection and the generation of T cell immune responses.

MATERIALS AND METHODS
Cells, Cytokine Differentiation of DCs, and RNA-CD34 ϩ peripheral hematopoietic progenitor cells (PBHP) and peripheral blood mononuclear cells (PBMCs) were isolated from healthy adult normal volunteers treated with granulocyte colony-stimulating factor (G-CSF, Amgen, CA) as described previously (27). The CD34 ϩ PBHP cells were cultured in medium supplemented with 20 ng/ml of stem cell factor and 50 ng/ml granulocyte-macrophage colony-stimulating factor (R&D Systems, Minneapolis, MN). Tumor necrosis factor-␣ (10 ng/ml) was added on day 7, and on day 11 of culture IL-4 (10 ng/ml) was added to one-half of the cells. The cytokine-differentiated CD34 ϩ PBHP cells were kept in culture for a total of 15 days. By day 14 of culture more than 99% of cells were CD33 ϩ indicating that the predominant cell population was of the myeloid series (27). The proportion of cells that stained for T/B lymphocyte markers (CD3/CD19) was less than 1-3%. PBMCs were also isolated from 20 ml of blood obtained from normal donors who did not receive granulocyte colony-stimulating factor. An aliquot of these PBMCs were stimulated with PHA (5 g/ml, Sigma) for 4 days. In some experiments IL-2 (50 units/ml, Life Technologies) was added to the culture medium after day 4. CD3 and CD28 monoclonal antibodies (PharMingen) were coated on tosyl-activated Dynal beads (Dynal, Lake Success, NY) and used to stimulate PBMCs (1:1 concentration). The placenta samples were from anonymous normal donors. mRNA from highly purified leukocyte subsets, including CD14 ϩ monocytes, was also  Horizontal lines indicate exons I-VI, and dashed lines illustrate the splicing events that lead to the formation of the prototypic exon II-containing DC-SIGN1A mRNA transcript that was originally described by Curtis et al. (20) and designated herein as mDC-SIGN1A Type I. ϩ1 indicates the translational start site in this prototypic DC-SIGN1 mRNA. Exon II is predicted to encode the TM domain (see Fig. 2) and the exon II-containing DC-SIGN1A mRNA transcripts are predicted to encode membrane-bound or mDC-SIGN1A isoforms, whereas mRNAs that lack this TM-encoding exon II are predicted to encode sDC-SIGN1A isoforms. Alternative splicing events that lead to the generation of mRNA transcripts that contain or lack the TM-encoding exon II can be deduced by joining the various exonic sequences indicated. The starting and ending nucleotide number of each exonic segment is separated by dots (e.g. join 1.. 46), and exonic segments are separated from each other by a comma (e.g. 1.. 46, 147..206, 981..1052). Note that we did not determine the length of the 5Ј untranslated region of DC-SIGN1. sDC-SIGN1A Type I represents the prototypic exon II-lacking DC-SIGN1A mRNA. The translation initiation codon for all DC-SIGN1A mRNA transcripts resides in exon Ia. Positions shown in bold denote splicing sites that are distinct from those found in the prototypic mDC-SIGN1A (Type I) or sDC-SIGN1A (Type I) mRNA transcripts. An asterisk indicates the stop codon used by the DC-SIGN1A transcripts shown in this panel. The numbering system is based on the nucleotide sequence deposited under GenBank™ accession number AC008812, with the first nucleotide of the initiation Met codon of the prototypic mDC-SIGN1A (Type I) mRNA considered as ϩ1. This nucleotide corresponds to residue 50622 in the nucleotide sequence in GenBank™ accession number AC008812. (The nucleotide numbers in AC008812 that correspond to each of the DC-SIGN1 nucleotide numbers shown in this figure can be found in the notes that accompany GenBank™/EBI Data Bank accession numbers AY042221-AY042233.) Note that the DC-SIGN1 is in the reverse orientation in this genomic contig. b, molecular basis for generation of DC-SIGN1B mRNA transcripts. The top panel is a schematic illustration of DC-SIGN1 gene.
Horizontal lines indicate exons I-VI. Exon Ib is the exonic sequence that separates exon Ia and Ic sequences, and all DC-SIGN1 mRNAs that obtained from a commercial source (CLONTECH). Cell lines were obtained from ATCC and the National Institutes of Health AIDS repository. Total RNA was extracted from cells using Trizol ® reagent (Life Technologies, Inc.) and first strand cDNA was generated using reverse transcriptase (RT) and random hexamers or oligo(dT) primers (Super-script™ Preamplification System, Life Technologies). The local institutional review board approved the studies conducted.
Primers, PCR Amplification, and Sequencing-The sequences of the oligonucleotides used in PCR and for hybridization experiments are shown in Table I. The cycling condition for PCR amplification of DC-SIGN1 cDNAs was 94°C for 10 s, 52°C for 30 s, and 72°C for 60 s. The cycling condition for amplification of DC-SIGN2 cDNAs was 94°C for 30 s, 65°C for 30 s, and 72°C for 90 s. A total of 35 cycles was used. The PCR products were cloned into TOPO vectors 2.1 or II (Invitrogen) and sequenced on both strands. To determine the genomic structure of DC-SIGN1, a series of sense and antisense orientation primers based on the cDNA sequence described by Curtis et al. (20) were designed (sequences not shown). Two expressed sequence tags (ESTs) that had homology to DC-SIGN2 were purchased from Research Genetics (Huntsville, AL, Image Clones 146996 and 240697) and sequenced on both strands.
Southern Blot Hybridization-One g of total RNA was used for synthesizing cDNA by random primers (Superscript Preamplification System, Life Technologies, Inc.). One-tenth of the cDNA product was used for PCR amplification. The PCR amplification profile consisted of 30 cycles of 94°C for 10 s, 55°C for 30 s, and 72°C for 60 s. PCR amplification was performed in a 100-l reaction volume in the presence of 20 mM Tris-HCl, 50 mM KCl, 1.5 mM MgCl 2 , 0.1 mM of each dNTP, 0.2 M of each primer, and 2.5 units of Taq DNA polymerase (Life Technologies, Inc.). The primers used for amplification were oligonucleotides 1-1 and 1-2 for DC-SIGN1 and oligonucleotides 2-3 and 2-4 for DC-SIGN2 (Table I). An oligonucleotide that is DC-SIGN1 exon Ib-specific (oligonucleotide 1-3, Table I) was used to amplify exon Ib-containing cDNAs. The amplified products were size-fractionated by electrophoresis on a 1.5% agarose gel. After denaturation in alkaline solution, the DNA was transferred to a nylon membrane (Amersham Pharmacia Biotech) by capillary action. Hybridization was performed with the following end-labeled oligonucleotide probes: (i) oligonucleotides derived from DC-SIGN1 sequences in exon Ib, exon Ic, exon II, and exon VI (oligonucleotides 1-4, 1-5, 1-6, and 1-8, respectively in Table I); (ii) an oligonucleotide that had 11 nucleotides of the 3Ј-end of exon Ic and 11 nucleotides of the 5Ј-end of exon III of DC-SIGN1 (oligonucleotide 1-7, Table I); (iii) an oligonucleotide that had identity with DC-SIGN2-specific exon II sequences (oligonucleotide 2-6, Table I). The membranes were hybridized with the radiolabeled probes at 42°C for 12 h and washed under the following conditions: 2ϫ SSC, 0.1% SDS at 42°C for 5 min (twice); 0.1ϫ SSC, 0.l% SDS at 45°C for 15 min (twice). The filters were exposed to Biomax (MR) film (Kodak) at Ϫ80°C in a Quanta III cassette for 15 h.
Polyacrylamide Gel Electrophoresis-DC-SIGN1 and DC-SIGN2 cDNAs were amplified using the PCR conditions described above in a 50-l reaction. The primers used for amplification were oligonucleotides 1-1 and 1-2 for DC-SIGN1 and oligonucleotides 1-1 and 2-5 for DC-SIGN2 (Table I). One of the primers used for amplification was endlabeled with 32 P to facilitate detection of the PCR products by autoradiography. Five l of the PCR product was mixed with 15 l of formamide dye (95% formamide, 10 mM EDTA, 0.02% bromphenol blue, 0.02% xylene cyanol) and boiled for 5 min. The mixture was then chilled and loaded on a 3 or 4% polyacrylamide gel containing 8 M urea and electrophoresed for 12 h at 200 V in a Protean II xi cell (Bio-Rad). The polyacrylamide gels were dried, and autoradiography was performed as described above.
In Vitro Translation-The TNT ® -coupled Reticulocyte Lysate System (Promega) was used to translate in vitro DC-SIGN1 cDNAs cloned into pcDNA4/HisMax TOPO vector (Invitrogen). The 35 S-labeled translated products were fractionated in a 9% acrylamide gel and were exposed to XAR-2 film (Kodak) in a Quanta III cassette.
Antibodies and Peptides-A synthetic peptide (NH 2 -CSRDEEQFLS-PAPATPNPPPA-COOH) derived from the C-terminal region of DC-SIGN1 was KLH-conjugated and used to immunize rabbits. The corresponding peptide sequence is absent in the DC-SIGN2. Rabbits were bled after 6 weeks to obtain polyclonal antiserum and were subsequently affinity-purified. Goat polyclonal antibodies (Ab) for CCR5 (sc-6128), PECAM-1 (sc-1505), and the corresponding blocking peptides were purchased from Santa Cruz Biotechnology (Santa Cruz, CA). The DC-SIGN1 blocking peptide was synthesized by Zymed Laboratories Inc. (San Francisco, CA).
Immunohistochemistry-OCT (Sakura Finetek USA, Inc., Torrance, CA)-embedded frozen term placental sections were air-dried for 30 min, washed in PBS (pH 7.4), and fixed in 4% cold paraformaldehyde for 10 min. The fixed sections were washed in Tris-buffered saline for 5 min, and were permeabilized with 0.05% PBS-Tween® (Sigma Chemical Co. St Louis, MO) for 5 min. All of the subsequent washes were in PBS-Tween®. The sections were blocked using an avidin-biotin blocking kit (Vector Laboratories, Burlingame, CA) according to the manufacturer's instructions. Subsequently the sections were blocked with 5% bovine serum albumin for 30 min, washed, and incubated with either DC-SIGN1 antiserum or DC-SIGN1 antiserum plus DC-SIGN1 blocking peptide for 1 h. The sections were washed for 5 min, incubated with a 1:100 dilution of biotinylated goat anti-rabbit antibody (Dako, Carpinteria, CA) for 30 min, washed, and then stained for 30 min with the avidin-biotin complex-glucose oxidase system (Vector Laboratories). Color development was achieved using the glucose oxidase substrate kit (Vector Laboratories). Distilled water was used to block additional color development. For double staining, the sections were incubated in PBS for 5 min, and endogenous peroxidases were inhibited using a peroxidase block (Santa Cruz) for 5 min. Slides were then washed in PBS-Tween for 5 min, blocked with 5% bovine serum albumin for 30 min, and then incubated with one of the following: (i) PECAM-1 Ab, (ii) PECAM-1 Ab and its blocking peptide, (iii) CCR5 Ab, or (iv) CCR5 Ab and its blocking peptide. Subsequent steps for detection of goat primary antibodies was performed using the goat Immunocruz staining system according to the manufacturer's directions (Santa Cruz). Sections were incubated with diaminobenzidine for 10 min, and the reaction was stopped with distilled water. The sections were then dehydrated with graded alcohols and two washes in xylene and were mounted with Vectamount™ (Vector Laboratories).

RESULTS
Genomic Organization of DC-SIGN1-In the course of identifying polymorphisms in DC-SIGN1, we identified several alternatively spliced DC-SIGN1 cDNAs (see below). To identify the genomic sequences homologous to these cDNAs, we determined the gene structure for human DC-SIGN1. Genomic DNA was subjected to PCR using primers corresponding to the known cDNA sequence (20); GenBank™ accession no. M98457), and the PCR products were cloned and sequenced. In addition, while this work was in progress, as part of the Human Genome Sequence Project, a ϳ143,619-bp contig of human chromosome 19p that contained DC-SIGN1 became available (GenBank™ accession no. AC008812). Other than a few polymorphisms, there was complete homology between the DC-SIGN1 genomic sequences that we had identified and those found in this contig (data not shown). Comparisons of the cDNA and genomic sequences revealed that the coding region of the previously described prototypic DC-SIGN1 cDNA (GenBank™ accession no. M98457) was encoded by six exons (Fig. 1a, top  panel). The nomenclature for the exons was based on the alcontain exon Ib are designated as DC-SIGN1B mRNA transcripts. Sequence analysis of exon Ib-containing transcripts revealed two potential translation initiation sites (ϩ1 or ϩ101). Transcripts predicted to initiate translation at ϩ101 in exon Ib may contain or lack the TM-encoding exon II. Dashed lines illustrate the splicing events that lead to formation of the prototypic exon II-containing DC-SIGN1B mRNA transcript (mDC-SIGN1B Type I). Splicing out of the TM-encoding exon II generates transcripts designated as sDC-SIGN1B Types I-IV (Type I is the prototype). The positions shown in bold denote splicing sites that are distinct from those found in the prototypic mDC-SIGN1B (Type I) or sDC-SIGN1B (Type I) mRNA transcripts. An asterisk indicates the stop codon used by the prototypic m-and sDC-SIGN1B mRNAs. The stop codons utilized by sDC-SIGN1B types III and IV at position 4335-4337 and 4492-4494, respectively, are indicated by daggers, and positions 4334 and 4491 are underlined. The DC-SIGN1B transcripts are also predicted to initiate translation at ϩ1 in exon Ia and have an in-frame stop codon (TGA; ϩ124 -126); these transcripts are predicted to generate a short polypeptide of 41 amino acids (see Figs. 2 and 3). c, non-canonical splice donor and/or acceptor sites used in generation of some DC-SIGN1 mRNA transcripts.
FIG. 2. Predicted structure and molecular diversity of membrane-bound and soluble DC-SIGN1 gene products with novel intraand/or extracellular domains. a, gene organization of DC-SIGN1 and alternative splicing events that lead to the generation of the prototypic DC-SIGN1 protein product described by Curtis et al. (20). Boxes signify exons (I-VI) and dashed lines introns (I-V in black circles). The nucleotide length of the introns are shown in parentheses. The first nucleotide of the initiation Met codon of the prototypic DC-SIGN1A transcript is considered as ϩ1. The stop codon used by the prototypic DC-SIGN1A isoform is denoted by an asterisk. The box with vertical hatch lines represents a small portion of the predicted 3Ј-untranslated region, and some DC-SIGN1B transcripts are predicted to terminate translation at position 4491 in this region (Fig. 1). The prototypic DC-SIGN1 protein product is predicted to have a short cytoplasmic (CYT; open boxes) and TM (box with forward slash) domain. Exons III-VI encode the predicted extracellular (EC) domain of the prototypic DC-SIGN1, which includes a short stretch of sequence just proximal to the repeats (box with horizontal lines), the seven full repeats and one half-repeat (numbered black boxes), and the lectin-binding domain (backward slash). The green box represents the alternatively spliced exon Ib. b-e, schematic illustrations of the molecular diversity and predicted structures of DC-SIGN1A and DC-SIGN1B isoforms generated by alternative splicing events. DC-SIGN1 variants that lack or contain exon Ib sequences (green box) are designated as DC-SIGN1A (b and c) or DC-SIGN1B (d and e) isoforms, respectively (Fig. 1). Predicted amino acid ternatively spliced exons identified in the DC-SIGN1 cDNAs (see below). Exons Ia and Ic encoded the majority of the cytoplasmic domain of the prototypic DC-SIGN1 cDNA (20). Exon II encoded 5 amino acids of the cytoplasmic domain and the entire transmembrane (TM) domain. Exon III encoded the repeats as well as a short stretch of amino acids that preceded the seven full repeats and the one-half repeat. Exons IV, V, and VI together encoded the predicted extracellular lectin-binding domain of DC-SIGN1.
Extensive Structural Diversity of DC-SIGN1 Transcripts-RT-PCR was used to amplify DC-SIGN1 cDNAs from PHAactivated PBMCs derived from normal human donors, human CD34 ϩ PBHP-derived mature DCs, and THP-1 monocytic cells. Sequence analyses of these PCR amplicons revealed several distinct cDNAs that shared homology to the previously reported prototypic DC-SIGN1 cDNA ( Fig. 1, a and These novel DC-SIGN1 transcripts differed from the originally reported cDNA sequence (GenBank™ accession no. M98457) by the presence or absence of stretches of sequences, indicating that they had arisen by a complex pattern of alternative splicing events in the exons encoding the intra-or extracellular domains and/or by splicing out of exon II, the exon that encodes the predicted TM domain ( Fig. 1, a and b). The predicted translation products of these transcripts are illustrated in Fig For example, in mDC-SIGN1A Type II, the first 6 amino acids encoded by exon V are spliced out, whereas in mDC-SIGN1A Type III, some of the repeats encoded by exon III are spliced out (Figs. 1a and 2b).
The second group of transcripts was designated as sDC-SIGN1A. sDC-SIGN1A transcripts also had a Met (ATG) translation initiation codon within exon Ia, but the exon predicted to encode the TM domain (exon II) was spliced out, suggesting the synthesis of soluble forms of DC-SIGN1A (Figs. 1a, 2c, and 3a). The prototypic version of this class of transcripts, designated as sDC-SIGN1A Type I, lacked only the TM-containing exon II, whereas additional splicing events resulted in sDC-SIGN1A Types II-IV (Figs. 1a and 2c).
The exon Ib-containing DC-SIGN1 cDNAs were collectively designated as DC-SIGN1B transcripts and are predicted to encode the third (mDC-SIGN1B), fourth (sDC-SIGN1B), and fifth (truncated DC-SIGN1B) category of DC-SIGN1 isoforms (Figs. 1b and 2, d and e). Notably, exons Ia, Ib, and Ic are sequences that are not interrupted by an intron and that col-lectively comprise exon I. There is a Met (ATG) translation initiation codon within exon Ib, and thus DC-SIGN1B transcripts can potentially initiate translation at two sites: ϩ1 or ϩ101 (Figs. 1b, 2, d and e, and 3b). The sequence flanking the ϩ101 position has a strong Kozak consensus sequence for initiation of translation (GCCATGG). The deduced amino acid sequence of transcripts that commence translation at the downstream Met codon (i.e. ϩ101) in exon Ib differed from mDC-SIGN1A or sDC-SIGN1A isoforms only in the predicted cytoplasmic domain. These transcripts could be further categorized into those that had (DC-SIGN1B) or lacked (sDC-SIGN1B) the TM-encoding exon II (Figs. 1b, 2, d and e, 3c). Notably, prototypic m-or sDC-SIGN1 differed from m-or sDC-SIGN1B (Type I) by only 14 amino acids in the predicted N terminus encoded by exon Ib (Figs. 2 and 3c; Supplementary Figs. 1 and 2). Finally, usage of the Met codon in exon Ia in DC-SIGN1B transcripts predicted the production of a truncated protein of 41 amino acids (nucleotides ϩ1 to ϩ123; Figs. 1b, 2, d and e, and 3b), and these isoforms were designated as truncated DC-SIGN1B isoforms (tDC-SIGN1B). To minimize the possibility that the exon Ib-containing DC-SIGN1 transcripts (i.e. DC-SIGN1B mRNAs) reflected PCR amplification of pre-mRNA contaminating the mRNA preparations, we confirmed the presence of these transcripts in poly(A) ϩ RNA (see below, and data not shown).
Splicing events generated sDC-SIGN1-A or -B transcripts that are predicted to encode novel C termini (Figs. 1, a and b, 2, c and e, and 3, d-f). In some instances, the splice junctions for the DC-SIGN1 mRNAs did not obey the consensus rules for 5Ј-intron/exon boundaries (Fig. 1c). Based on the splicing events in exons III-VI, these exons could be further subdivided (e.g. exon IIIa, IIIb, etc.). However, we refrained from doing so, recognizing that based on mRNA expression analyses there are probably additional splice variants that have not been discovered as of yet (see below). The model shown in Fig. 4 summarizes the predicted DC-SIGN1 gene structure, the primary transcript, mature mDC-SIGN1 (A or B) and sDC-SIGN1 (A or B) mRNAs, and a schema of the potential processing events underlying the formation of the mature messages. Collectively, the findings illustrated in Figs. 1-4 demonstrate that the DC-SIGN1 gene is subject to highly complex alternative splicing events, generating a wide array of transcripts that are predicted to encode for an extensive repertoire of membrane-associated as well as soluble DC-SIGN1 isoforms with variable intra-and/or extra-cellular regions.
DC-SIGN2, a Gene with Structural Homology to DC-SIGN1 That Is Also Subject to Alternative Splicing-By searching the GenBank™ data bases, we found a cDNA (28) and two ESTs (Image Clones 146996 and 240697) that had high overall sequence homology with the DC-SIGN1 transcripts that we had identified. The cDNA and ESTs differed from each other by the presence or absence of additional stretches of sequences. To determine whether the cDNA and ESTs represented allelic versions of the DC-SIGN1 gene or products of a novel gene, RT-PCR was performed on human placenta mRNA using primers specific to those found in the cDNA and ESTs. Sequence analyses of the PCR products revealed additional novel cDNAs differences among the isoforms, the source(s) from which their transcripts were cloned, and the length of the message (in nucleotides (nt)) and the predicted translated product (amino acids (aa)) are indicated to the right of the schema depicting the structural domains present in a given variant. Panels c and e depict the transcripts that encode the isoforms predicted to lack the TM domain (i.e. isoforms lacking exon II). An in-frame initiation codon present at ϩ101 in the exon Ib is predicted to commence translation of an intact open reading frame but with a novel cytoplasmic tail (see Figs. 1b and 3b). The splicing out of exon V (panel e, sDC-SIGN1B Type III) is predicted to generate a soluble variant with a novel C-terminal sequence (blue box). Similarly, splicing events in sDC-SIGN1 Type IV are predicted to result in a novel C terminus (Fig. 3e). † and † †, denote the amino acid lengths of the DC-SIGN1B translated sequences that are predicted to initiate translation at either ϩ1 in exon Ia (41 amino acids) or ϩ 101 in exon Ib (varying lengths). ⌬, denotes skipping of the indicated exons/sequences. Because of splicing events in exons III-VI, these exons can be further subdivided (e.g. exon IIIA, IIIB, etc.); however, these demarcations are not indicated.  1b (ϩ124 -126). The open reading frame initiated from a start codon at ϩ101-103 in Exon Ib is predicted to encode DC-SIGN1B products that except for the N terminus (MASACPGSDFTSIHS, amino acid sequence in green) are identical to the prototypic DC-SIGN1A isoforms (see panel c). c, alignment of the deduced amino acid sequences of prototypic mDC-SIGN1A, sDC-SIGN1A, mDC-SIGN1B, and mDC-SIGN2 isoforms. Dots and dashes represent sequence identities and gaps, respectively. The cytoplasmic (CYT) domain, TM domain, the extracellular (EC) domain, which includes the repeats and the lectin binding domain, are indicated along with the deduced amino acid sequences encoded by the six exons of the prototypic mDC-SIGN1A (20) and mDC-SIGN1B. The first amino acid (I) of each of the repeats is in red. Note that the predicted amino acid sequences of mDC-SIGN1A Type I and mDC-SIGN1B Type I are identical beyond the first 14 amino acids of the cytoplasmic domain. The underlined sequence denotes the peptide sequence that was used for raising antiserum and is not found in DC-SIGN2. d-f, splicing patterns of exon II-lacking transcripts that are predicted to encode sDC-SIGN1 isoforms with novel C termini. Stop codons are boxed. The antisense orientation primer used for PCR amplification is underlined in panel f. with sequences identical to the previously described cDNA/ESTs but distinct from DC-SIGN1 (A or B) transcripts, suggesting that they were alternatively spliced products of a distinct gene and not allelic variants of DC-SIGN1 (Fig. 5). The predicted translation products of these transcripts are illustrated in Fig. 6 and are also shown in Supplementary Fig. 3.
Genomic sequences identical to the novel DC-SIGN-like mRNAs that we had discovered as well as the previously identified cDNA (28) and ESTs were found 15.8 kb centromeric to DC-SIGN1 on chromosome 19p13.3, and these two genes were arranged in a head-to-head manner (Fig. 7). Based on their close sequence homology, their colocalization on chromosome 19p13.3, and their order of discovery, we designated the previously described DC-SIGN as DC-SIGN1 and this related gene that we had identified as DC-SIGN2. The coding region of the prototypic full-length DC-SIGN1 and DC-SIGN2 shared 84 and ϳ80% identity (29) at the nucleotide and protein levels, respectively (Fig. 3c, and data not shown). Comparison of the DC-SIGN2 mRNA and gene sequences revealed that the coding region of DC-SIGN2 was encoded by seven exons (Figs. 5a and 6a).
Similar to the alternative splicing events observed in DC-SIGN1, DC-SIGN2 transcripts in which the exon predicted to encode the TM domain (exon III) was spliced in or out were found and were designated mDC-SIGN2 or sDC-SIGN2 isoforms, respectively (Figs. 5a and 6, a-c). Additional alternative splicing events generated mDC-SIGN2 or sDC-SIGN2 transcripts, which are predicted to encode isoforms with varied extracellular domains (Fig. 5, c and d and 6, b and c). Notably, of the Ͼ30 DC-SIGN2 transcripts that we cloned and sequenced from the placenta of a normal donor, 21 cDNAs were found that contained sequences corresponding to intron IV, and in this particular placenta sample, we were unable to identify a prototypic mDC-SIGN2 transcript (Fig. 5). These findings provided the first clue that there might be significant interindividual variability in the repertoire of DC-SIGN2 tran-scripts expressed in term placenta. The discovery of DC-SIGN2 transcripts with distinct splicing patterns that contained intron IV and/or lacked exon VI from multiple sources ( Fig. 6; e.g. ESTs and this study) indicated that the splicing patterns that we found were not aberrant or random events but rather may represent fairly common processing events.
A unique differential splicing event was observed that distinguished DC-SIGN2 mRNAs that contained (mDC-SIGN2) or lacked (sDC-SIGN2) the TM-encoding exon III. Among the DC-SIGN2 cDNAs that we cloned and sequenced, all sDC-SIGN2 transcripts contained sequences corresponding to exon IVa, but none of the transcripts that had the TM-encoding exon III, i.e. mDC-SIGN2 transcripts contained exon IVa sequence (Figs. 5, a and e and 6c). Exon IVa is predicted to encode a short hydrophobic stretch of amino acids ( Fig. 5e and Supplementary  Fig. 3).
It should be noted that intron I of the DC-SIGN2 gene corresponds to exon Ib of the DC-SIGN1 gene. Similar to the scenario observed in DC-SIGN1B, the use of an alternative translational start site at position 111 of intron I in DC-SIGN2 is predicted to encode isoforms with a novel intracellular domain. However, DC-SIGN2 transcripts that contained intron I sequences were not found in the cDNA clones that we have sequenced thus far.
Although these findings provided evidence for extensive alternative splicing events within the region preceding the lectinbinding domain (i.e. region that encodes the repeats) of DC-SIGN2, it was conceivable that in some individuals the variation in the number of repeats could be because of allelic variation. For example, it was conceivable that one allele could encode for eight repeats whereas the other allele could encode for seven repeats. To determine this, we amplified the genomic DNA that spanned the region between exon III and intron IV from normal donors. We found that in some instances, one allele encoded seven repeats whereas the other allele encoded FIG. 4. Schema of predicted alternative splicing events that lead to the generation of membrane-bound or soluble DC-SIGN1 gene products. Splicing event 1 links the end of exon Ia to the beginning of exon Ic and is predicted to generate the previously described prototypic DC-SIGN1A message (mDC-SIGN1A Type I; Fig. 1a). Additional splicing events in this primary mDC-SIGN1A mRNA generate exon II-retaining mDC-SIGN1A Types II-IV mRNAs. Splicing event 2 links the end of exon Ic to exon III generating the prototypic exon II-lacking sDC-SIGN1A message (sDC-SIGN1A Type I mRNA), and additional splicing events in this message leads to exon II-lacking sDC-SIGN1A Types II-IV mRNAs. In contrast, transcripts in which splicing event #1 does not occur are predicted to generate the prototypic exon II-retaining mDC-SIGN1B message (mDC-SIGN1B Type I mRNA) and/or tDC-SIGN1B (exon Ia ϩ partial exon Ib). Splicing event 2 in the mDC-SIGN1B primary transcript links the end of exon Ic to exon III generating the prototypic exon II-lacking sDC-SIGN1B Type I mRNA, and additional splicing events in this message lead to exon II-lacking sDC-SIGN1B Types II-IV mRNAs. The structure of the translation products of the mRNAs generated by these splicing events is shown in Fig. 3  Exon III is predicted to encode the TM domain (Fig. 6), and the exon III-retaining DC-SIGN2 mRNAs are predicted to encode membrane-bound or mDC-SIGN2 isoforms, whereas mRNAs that lack this TM-encoding exon III are predicted to encode soluble or sDC-SIGN2 isoforms. Alternative splicing events that lead to the generation of DC-SIGN2 mRNAs that contain or lack the TM-encoding exon III can be deduced by joining the various exonic sequences indicated. The starting and ending nucleotide number of each exonic segment is separated by dots (e.g. join 1.. 46), and exonic segments are separated from each other by a comma (e.g. 1.. 46, 127..210, 1919..2002). Asterisks indicates the stop codon used by most m-or sDC-SIGN2 transcripts, whereas the daggers indicate the stop codon utilized by mDC-SIGN2 mRNAs Type V and VI at positions 5608 -5610. Sequences corresponding to exon IVa were found only in the sDC-SIGN2 transcripts. Note that repeats 3-5 cannot be distinguished from each other; hence the splice junctions for mDC-SIGN2 type VI and sDC-SIGN2 Type I transcripts cannot be inferred. b, DC-SIGN2 mRNA transcripts that use non-canonical splice donor and/or acceptor sites. c and d, alternative splicing events that lead to the generation of sDC-SIGN2 isoforms with novel C termini. ⌬, denotes skipping of the indicated exons/sequences. Stop codons are boxed. e, deduced amino acid sequences encoded by exon IVa and alignment of the region bridging exon II and exon IVb in sDC-SIGN2 and mDC-SIGN2 isoforms.  (Fig. 5a). An asterisk denotes the stop codon found in the prototypic mDC-SIGN2 transcript. The box with vertical hatch lines represents the 3Ј-untranslated region. The predicted structure of the prototypic mDC-SIGN2 protein product is shown in panel b. Exons I, II, and a portion of exon III encode a short cytoplasmic domain (Cyt, open boxes); the TM domain (boxes with forward slash) is encoded by sequences in exon III. Exons IVb-VII encode the predicted extracellular domain of the prototypic mDC-SIGN2, and this includes a short stretch of sequence just proximal to the repeats (box with horizontal lines), the seven full repeats and one half-repeat (numbered black boxes), and the lectin-binding domain (box with backward slash). The green box represents the alternatively spliced exon IVa that is found only in those isoforms that lack the TM-encoding exon III. Image Clone 240607 was a partial cDNA clone that contained exons V, VI, and VII (data not shown). The alignment of the deduced amino acid sequences of the DC-SIGN2 isoforms depicted in this figure is shown in supplementary Fig. 3. b and c, schematic illustration of the molecular diversity and predicted structures of DC-SIGN2 isoforms generated by alternative splicing events. Panels b and c depict the transcripts that encode the isoforms that are predicted to contain the TM domain (mDC-SIGN2 isoforms) and isoforms that lack the TM domain (sDC-SIGN2 isoforms), respectively. Predicted amino acid (aa) differences among the isoforms and the source(s) from which their transcripts were cloned are indicated to the right of the schema depicting the structural domains present in a given variant. Retention of intron IV leads to formation of a novel C terminus in mDC-SIGN2 types II and IV and sDC-SIGN2 type II (blue box). Because of splicing out of exon VI, a novel C terminus is predicted to form in mDC-SIGN2 types V and VI (red box). The sDC-SIGN2 isoforms exclusively contain a short hydrophobic stretch of amino acids because of the presence of exon IVa (green box). The yellow box represents intron VII. ⌬, denotes skipping of the indicated exons/sequences. eight repeats. These findings suggested that in addition to alternative splicing, a variation in the number of repeats encoded in the DC-SIGN2 gene could be another source for variability in generating the DC-SIGN2 mRNA repertoire. Additional studies are under way to characterize the nature and frequency of this genetic polymorphism (i.e. variability in number of repeats) in different ethnic populations. Studies are also underway to determine whether there is variability in the number of repeats in the DC-SIGN1 gene.
Additional inspection of the genomic contig from chromosome 19p13.3, demonstrated that the gene for the low affinity immunoglobulin ⑀ Fc receptor (CD23), another Type II lectin (23, 24), was situated ϳ43.3 kb telomeric to DC-SIGN1 (Fig. 7). Thus, DC-SIGN1 (CD209), DC-SIGN2 (CD209L), and CD23 form a cluster of highly related genes, suggesting that they may have arisen by gene duplication of an ancestral gene, and notably alternative splicing events in all three genes lead to the generation of multiple transcripts (Figs. 1-7) (28,30,31).
Expression of DC-SIGN1 Is Not Restricted to DCs-Given the aforementioned findings, we asked whether the DC-SIGN1 transcripts were expressed in a complementary manner. That is, does a given cell type express only one DC-SIGN1 transcript, similar to the exclusive expression of odorant receptors in olfactory neurons (32), or are different DC-SIGN1 variants expressed in a combinatorial manner? In the first scenario, a given cell type could potentially be classified into one of five groups depending on which DC-SIGN1 transcript it expressed. In the second scenario, distinct transcripts could be coex-

FIG. 7. Colocalization of DC-SIGN1 (CD209), DC-SIGN2 (CD209L), and
CD23 to within ϳ85kbp of chromosome 19p13.3. All three genes are subject to highly complex splicing events (23,30,31). Fig. 9 are shown. Total RNA (1 g) isolated from DCs derived from cytokine-differentiated CD34 ϩ PBHPs, PBMCs, placenta, THP-1 cell line, or other cell lines (data not shown) was reverse-transcribed with oligo(dT) primers. The resulting cDNA was PCR-amplified using DC-SIGN1A (primers 1-1 and 1-2)-or DC-SIGN1B (primers 1-3 and 1-2)-specific primers. The PCR amplicons were fractionated by agarose gel (1.5%-) electrophoresis, transferred to nylon membrane, and hybridized with the indicated radiolabeled probes. The blots were washed and then exposed for 15 h. b-e, specificity of the radiolabeled oligomers used. Nylon membranes spotted with the indicated DNA listed on the top of each blot were hybridized with the radiolabeled probe: b, Exon VI (Ex VI); c, Ex II; d, Ex Ic-Ex III; e, Ex Ib. Ex VI probe hybridizes all DC-SIGN1 (A or B) transcripts; Ex II probe hybridizes all DC-SIGN1 (A or B) transcripts that contain the TM-encoding exon II; Ex Ic-Ex III probe hybridizes DC-SIGN1 (A or B) transcripts that lack the TM-encoding exon II. (f) DC-SIGN1 (A or B) expression in CD34 ϩ PBHP-differentiating DCs cultured in the presence or absence of IL-4. Note, activation-induced differences in the levels of DC-SIGN1 expression (compare hybridizing signal in DCs Ϯ IL-4). The probes used are indicated to the right of each blot. g, cDNAs amplified using DC-SIGN1B-specific primers from DCs derived from cytokine-differentiated CD34 ϩ PBHPs or THP-1 cells were fractionated by gel electrophoresis and Southern blot-hybridized with the radiolabeled Ex VI oligomer or a radiolabeled oligomer that is specific to DC-SIGN1B. Note, the additional hybridizing signal observed in the THP-1 lane. h, DC-SIGN1 (A or B) expression in THP-1 cells obtained from ATCC. The molecular weight makers (in bp) are indicated to the left of the autoradiographs. Shorter exposures of panels f and g revealed a ladder of hybridizing signals, but they could not be completely fractionated by agarose gel electrophoresis (data not shown). pressed in variable patterns to confer specific properties onto the expressing cells, with the variability being dependent on the ratio of expression of the different DC-SIGN1 mRNAs. An additional level of complexity could be that the expression patterns varied depending on the activation state and/or maturation stage of the cell.

FIG. 8. Expression of DC-SIGN1 transcripts that lack or contain the TM-encoding exon in DCs and THP-1 cells. a, the overall experimental strategy for the findings shown in panels b-h and in
To address the aforementioned question, a RT-PCR-based strategy that included Southern blot hybridization was used to determine the expression of DC-SIGN1 mRNAs in primary human cells and human cell lines (Fig. 8a). To perform semiquantitative RT-PCR, in initial experiments we determined the number of PCR cycles wherein the hybridizing signal for DC-SIGN1 cDNAs were in the linear range (30 cycles), and PCR was performed using equal (1 g) amounts of mRNA from each cell/tissue type.
To increase the specificity and to estimate the relative amounts of DC-SIGN1 mRNAs that had or lacked the TMencoding exon II, five procedures were adopted. First, PCR was performed using unlabeled oligonucleotides specific for DC-SIGN1A or DC-SIGN1B (Table I), and the PCR products containing the DC-SIGN1 cDNAs were transferred to a membrane, and hybridized using DC-SIGN1 (A or B)-specific internal 32 P-labeled oligomers. This strategy assured that the hybridizing signal contained the DC-SIGN-specific sequence and not nonspecific amplification.
Second, because DC-SIGN1 and DC-SIGN2 transcripts shared high sequence homology, the specificity of the nested radiolabeled DC-SIGN1 probes and washing conditions were optimized in control experiments using cloned DC-SIGN1 and DC-SIGN2 cDNAs (Fig. 8, b-e). Four nested radiolabeled oligomers were used in these hybridization studies (Fig. 8a and Table I). (i) The exon VI oligomer was designed to hybridize DC-SIGN1A and DC-SIGN1B transcripts regardless of whether they contained or lacked the TM-encoding exon II. This oligomer hybridized specifically to mDC-SIGN1, and a very faint cross-hybridizing signal was detected in mDC-SIGN2 cDNAs (Fig. 8b). (ii) The exon II oligomer was designed to hybridize transcripts that contained the TM-encoding exon II, i.e. mDC-SIGN1 (A or 1B) mRNAs. This probe specifically hybridized mDC-SIGN1 but not sDC-SIGN1, mDC-SIGN2, or sDC-SIGN2 cDNAs ( Fig. 8c and data not shown). (iii) The exon Ic-exon III oligomer is specific for sDC-SIGN1 (A or B) DNA, i.e. transcripts that lacked exon II. Notably, this probe did not hybridize to DC-SIGN-1 or -2 transcripts that contained the exon II-encoding TM domain or to sDC-SIGN2 DNA ( Fig. 8d and data not shown). (iv) The exon Ib oligomer was designed from a region that is not found in DC-SIGN1A transcripts, and in hybridization studies it was specific to m-or sDC-SIGN1B cDNAs ( Fig. 8e and data not shown).
Third, to confirm that the DC-SIGN1 PCR primers used to generate the cDNAs were specific, the Southern blots shown in Figs. 8, f-h and 9 were stripped of radioactivity and reprobed with primers specific to DC-SIGN2. On rehybridization, DC-SIGN2 cross-hybridizing signals were not detected.
Fourth, because of the very faint cross-hybridization signals observed with the exon VI probe (Fig. 8b), we designed oligomers specific to DC-SIGN1 exon Ic (oligomer 1-5; Table I) and DC-SIGN2 exon II (oligomer 2-6; Table I). A set of Southern blots identical to those shown in Figs. 8 and 9 were hybridized with either a radiolabeled DC-SIGN1 exon Ic or DC-SIGN2 exon II probe. Hybridizing signals obtained with the DC-SIGN1 exon Ic probe were identical to those observed previously with the DC-SIGN1 exon VI probe. In contrast, a hybridizing signal was not detected with the DC-SIGN2 exon II probe, indicating that the mRNA expression patterns observed using the strategy outlined was specific for DC-SIGN1. As a final step to increase specificity and validate the expression pattern of DC-SIGN1 and DC-SIGN2 transcripts, cDNAs were synthesized from multiple different normal donors and cell lines.
An example from four separate experiments demonstrating the cell and tissue expression of DC-SIGN1 transcripts is shown in Figs. 8 and 9. We first focused on the expression of DC-SIGN1 mRNA in CD34 ϩ PBHP cells cytokine-differentiated toward the DC lineage (Fig. 8f). m-and sDC-SIGN1 (A or B) cDNAs were abundantly expressed in mature DCs, i.e. CD34 ϩ PBHPs cytokine-differentiated for 15 days but not at earlier time points (Fig. 8, f and g). In addition to the prominent hybridizing signals of ϳ1-1.3 kb in length, several hybridizing bands that were Ͻ1 kb in length were also detected (see below and data not shown). Notably, the hybridizing signal in CD34 ϩ PBHPs differentiated with IL-4 was stronger than that observed in DCs cultivated without IL-4 (day 15 ϮIL-4; Fig. 8, f and g), suggesting that the expression level of DC-SIGN1 mRNA may be dependent on the maturational/activation state of DCs. On longer exposures, faint hybridizing signals were evident at day 8 and 12 cytokine-differentiated CD34 ϩ PBHPs, suggesting that the expression of DC-SIGN1 in immature DCs was significantly lower than that in mature DCs derived from CD34 ϩ PBHPs.
In addition to DCs, m-and sDC-SIGN1 (A or B) transcripts were expressed in other antigen-presenting cells such as highly purified resting CD14 ϩ monocytes (data not shown) as well as THP-1 and U937 cells, two monocytic cell lines (Fig. 8, g and h, and data not shown). Expression of DC-SIGN1 transcripts was confirmed in two independent sources of THP-1 cells (ATCC and National Institutes of Health AIDS repository; data not shown). Because it was difficult to control for differences in the labeling and hybridizing efficiencies of the different probes required to differentiate between the exon II-containing or -lacking DC-SIGN1 transcripts, it was not possible to assess in a quantitative manner their relative abundance in DCs or THP-1 cells. Nevertheless, the findings shown in Fig. 8 indicated that both m-and sDC-SIGN1 (A or B) transcripts are abundantly expressed in DCs and THP-1 cells.
Weak expression of DC-SIGN1 mRNA was detected in resting PBMCs obtained from eight normal donors (Fig. 9a and data not shown). In contrast, abundant expression for m-and sDC-SIGN1 (A or B) transcripts was detected in all eight PBMC samples after stimulation with PHA ( Fig. 9, b-d and data not shown) as well as in PBMCs activated with CD3/CD28 (Fig. 9e). DC-SIGN1-specific hybridizing signals were evident in PBMCs activated with PHA for 4 days but not in PBMCs cultured in PHA (days 1-4) plus IL-2 (days 5-12; Fig. 9e). Notably, there was inter-individual variation in the expression of DC-SIGN1 transcripts in PHA-activated PBMCs (Fig. 9, b  and d; e.g. compare hybridizing signals in donors 2 and 4 versus donors 1, 3, and 5 in panel d).
Because of our interest in the potential role of HIV attachment factors such as DC-SIGN1 in mother-to-child transmission of the virus, we also determined whether DC-SIGN1 is expressed in the placenta. Notably, we detected both interindividual variation in the levels of DC-SIGN1 expression as well as heterogeneity in the repertoire of transcripts expressed (Fig. 9f; compare pattern of hybridizing signals in donor 2 versus 1 and 3). The expression of DC-SIGN1 in placenta was confirmed by immunohistochemical staining of term placentae. DC-SIGN1 expression colocalized with that of PECAM, an endothelial cell marker, as well with CCR5 (Fig. 10). The double immunostaining in Fig. 10C indicates that DC-SIGN1 is coexpressed along with CCR5 in placental villi, and the distribution pattern of CCR5 ϩ DC Ϫ SIGN1 ϩ cells is consistent with their expression in villous macrophages.
Weak DC-SIGN1-specific hybridizing signals of ϳ1.2 kb in length were also observed in MG63 (osteoblast) cells, HSB-2 (T cells), and MC116 cells, a B-cell line (data not shown). DC-SIGN1 expression was observed in the T cell line, HUT78; however, only an ϳ300 and ϳ600 bp hybridizing signal was detected in this cell type (data not shown). The presence or absence of hybridizing signals of varying sizes in T cells might reflect differences in the activation states of these cell lines. A ladder of hybridizing bands was also observed in HL-60 cells, a granulocytic cell line (data not shown).
The strongest hybridizing signals for DC-SIGN1 in mature DCs, PBMCs, placenta, and THP-1 cells were in the 1,000 -1,300-bp range (Figs. 8 and 9), which was concordant with the large number of transcripts identified in this size range by direct cDNA sequencing ( Figs. 1 and 2). However, the strong intensity of the hybridizing signals at ϳ1-1.3 kb masked the ladder of hybridizing bands that was evident on shorter exposures (data not shown) or as seen in Fig. 8g (Ex VI probe) as well as in Fig. 9, d and f. Furthermore, we found it difficult to resolve this ladder of hybridizing bands using horizontal gel electrophoresis. To circumvent this limitation, the DC-SIGN1 sense orientation oligomer used in the aforementioned experiments was radiolabeled and used in PCR, and the amplicons were resolved on a polyacrylamide gel (Fig. 11a). The findings of these experiments revealed a ladder of PCR amplicons in all cell types having size ranges concordant with the lengths of the DC-SIGN1A or DC-SIGN1B cDNAs that we had identified by direct sequencing (Fig. 11a and data not shown). This ladder of PCR amplicons is consistent with the notion that DC-SIGN1 undergoes extensive splicing events to generate a large repertoire of transcripts of varying lengths. Fig. 11b illustrates that on gel electrophoresis, the lengths of the transcripts in the 1-1.3-kb size range may appear deceptively similar, and direct sequencing may be necessary to distinguish their unique sequence characteristics.
We next determined whether the transcripts predicted to encode membrane-associated and soluble DC-SIGN1 isoforms are translated in vitro (Fig. 11c). The in vitro translated products of the predicted sizes (epitope tag plus coding region) for both DC-SIGN1A and DC-SIGN1B products confirmed the integrity of the Expression Pattern of DC-SIGN2 Transcripts-To determine the expression of DC-SIGN2 transcripts, a strategy similar to that used to examine the expression of DC-SIGN1 was adopted ( Fig. 12 and data not shown). In initial experiments, we observed that akin to DC-SIGN1, DC-SIGN2 transcripts were expressed in the placenta, and concordant with our isolation of cDNAs of varied lengths from this tissue, a ladder of amplicons were observed in some placental samples (Fig. 12a). However, in these initial experiments, we found that there was extensive inter-individual heterogeneity in not only the expression levels but also the repertoire of transcripts expressed. For example, we found that placenta from donor 3 lacked a transcript in the size range for the prototypic mDC-SIGN2 mRNA (Fig. 12a). This finding was notable because it may explain, in part, why we were unable to directly clone mDC-SIGN2 Type I transcripts from mRNA derived from this placenta sample. In agreement with our cDNA cloning studies, all four placenta samples had transcripts in the size ranges that were consistent for expression of intron IV-containing mRNAs, suggesting that these transcripts may comprise a major proportion of the DC-SIGN2 mRNA repertoire. As indicated earlier, variability in the length of the DC-SIGN2 transcripts may be accounted for, in part, by the variation in the number of repeats present on a given allele.
To extend and confirm these findings, we determined the expression of DC-SIGN2 transcripts in 10 additional term placentae from normal donors. Consistent with our initial studies (Fig. 12a), we found that there was striking heterogeneity in both the levels of expression of DC-SIGN2 transcripts as well as in the repertoire of transcripts expressed. Notably, despite equal expression for actin in all placental samples, we were unable to detect transcripts for DC-SIGN2 in 4 of the 10 placental samples, and only 2 of 10 placenta mRNA samples (samples 11 and 12) had transcripts with lengths corresponding to the prototypic mDC-SIGN2 mRNA.
DC-SIGN2 amplicons were also found in mature DCs (day 15 cytokine differentiated CD34 ϩ PBHPs; Fig. 12c). However, using a RT-PCR Southern blot hybridization strategy similar to that shown in Fig. 8a, the expression of DC-SIGN2 in CD34 ϩ PBHP-derived mature DCs was found to be significantly lower than that of DC-SIGN1 (data not shown). Using the Southern blot hybridization strategy, weak expression for DC-SIGN2 was also detected in THP-1 monocytic cells, whereas expression was not detected in CaCo2 (colorectal adenocarcinoma), RD (rhabdomyosarcoma), HUT 78 (T cell), MC116 (B cell) cells, or resting or activated PBMCs (data not shown).

DISCUSSION
DCs are thought to act as "Trojan horses," capturing HIV in the mucosal surfaces for transport to the T cell areas of draining lymphoid tissues. The proficiency of DCs in interacting with T cells makes them prime candidates for enhancing viral infection. Recent reports indicate that DC-SIGN, a surface receptor with high expression in DCs, may play an important role in DC-T cell as well as DC-HIV interactions (16,17). We have significantly extended these initial reports by (i) discovering that complex alternative splicing events in DC-SIGN (designated here as DC-SIGN1) pre-mRNA generates a wide repertoire of DC-SIGN1 transcripts. These DC-SIGN1 transcripts are predicted to encode both membrane-associated (mDC-SIGN1-A or -B) as well as soluble (sDC-SIGN1-A or -B) isoforms with varied intracellular and/or extracellular ligand (gp120/ICAM-3) binding domains. (ii) We have identified another highly homologous gene designated here as DC-SIGN2. Similar to DC-SIGN1, alternative splicing of DC-SIGN2 pre-mRNA also generates a wide repertoire of DC-SIGN2 transcripts that are predicted to encode membrane-associated and soluble isoforms. (iii) Interestingly, in addition to DC-SIGN1 (original version) and DC-SIGN2, we found that the low affinity immunoglobulin ⑀ Fc receptor (CD23) also maps to chromosome 19p13.3, forming a cluster of highly related genes that all undergo highly complex alternative splicing events (23,24). (iv) In contrast to previous reports (16,17), we found that DC-SIGN1 mRNA expression is not restricted to DCs but is significantly broader and includes THP-1 monocytic cells, resting CD14 ϩ monocytes, PBMCs, and placenta. Immunostaining with a DC-SIGN1-specific antibody indicated that DC-SIGN1 is expressed on placental endothelium as well as on CCR5 ϩ cells. The distribution of these CCR5 ϩ DC-SIGN1 ϩ cells is con- FIG. 11. Extensive repertoire of DC-SIGN1 mRNA transcripts in DCs, THP-1 cells, and PBMCs and in vitro translation of DC-SIGN1 cDNAs. a, oligo(dT)-primed cDNAs were PCR-amplified with DC-SIGN1-specific primers. The sense orientation primer was 32 P end-labeled, and the resulting PCR amplicons were fractionated on a 4% polyacrylamide gel. The molecular weight makers (in bp) are indicated to the left of the gels. b, PCR-amplified products of 11 cDNAs shown in Fig. 1 (Fig. 1, b- sistent with that of placental macrophages. (v) DC-SIGN2 transcripts were also detected in placenta but not in PBMCs. In contrast to DC-SIGN1, expression of DC-SIGN2 mRNA in DCs and THP-1 monocytic cells was significantly lower. (vi) Notably, we found that there was inter-individual variation in the repertoire of DC-SIGN1 and DC-SIGN2 transcripts expressed and that there were cell maturation stage and/or activation state differences in the expression levels of DC-SIGN1 mRNA.
Is this discovery of a wide array of alternatively spliced DC-SIGN1 and DC-SIGN2 transcripts unusual? On the contrary, alternative splicing of the precursor for mRNA (pre-mRNA) is a powerful and versatile regulatory mechanism utilized by higher eukaryotes for generating functionally different proteins from the same gene and accounts for a considerable proportion of proteomic complexity (33)(34)(35). Indeed, there are remarkable examples of hundreds and even thousands of functionally distinct mRNAs and proteins being produced from a single gene. In the human genome, such protein-rich genes include neurexins (36, 37), n-cadherins (38 -41), calcium-activated potassium channels (42,43), and others (34,44,45).
Alternative splicing is often tightly regulated in a cell typeor developmental stage-specific manner. Coordinated changes in alternative splicing patterns of multiple pre-mRNAs are an integral component of gene expression programs, like those involved in nervous system differentiation (46) and apoptotic cell death (47). Similar programs are also likely to exist during T cell and DC differentiation (48 -50). In addition to cellular differentiation, the pattern of splicing can be influenced by the activation of particular signaling pathways (51)(52)(53)(54)(55)(56)(57). Notably, in our studies we found that the expression pattern of DC-SIGN1 transcripts may depend, in part, on the cell maturation/ activation state (Figs. 8 and 9).
It is known that alternative splicing can generate mRNA structures that can take many different forms (33)(34)(35). Exons can be spliced into mRNA or skipped. Introns that are normally excised can be retained in the mRNA. The positions of either 5Ј or 3Ј splice sites can shift to make exons longer or shorter. In addition to these changes in splicing, alterations in transcriptional start site or polyadenylation site also allow the production of multiple mRNAs from a single gene. It is remarkable that nearly all of these variations in mRNA structure were observed in DC-SIGN1 and DC-SIGN 2 transcripts (Figs. 1-6).
An emerging paradigm is the observation that proteins involved in cell-cell contact or recognition often exhibit a high degree of molecular diversity. Examples include genes for cadherins, cadherin-related neuronal receptors, olfactory receptors, and neurexins in the nervous system (36, 38 -41, 58) and for immunoglobulin and T cell receptor genes in the immune system (59 -61). In this context, it is notable that DC-SIGN1mediated binding of DCs to ICAM-3 on resting T cells is thought to be a key initial adhesion step in the multistep process that leads to the formation of the immunological synapse and the activation of resting T cells (17). Thus, DC-SIGN1 (and potentially DC-SIGN2) demonstrates the generality of the features found in certain other genes involved in cell-cell adhesion/recognition. These common features include extensive alternative splicing events, cell type-and activation-specific expression, and a similar domain structure with distinct patterns of shared and divergent sequences.
In this report, we have demonstrated the genomic basis for the generation of not only several membrane-associated but also potentially soluble forms of DC-SIGN1 and DC-SIGN2. Furthermore, our studies suggest that the expression levels of DC-SIGN1 transcripts that lack the TM-coding exon are not minor variants of the overall pool of DC-SIGN1 mRNAs. Remarkably, the skipping of the TM-coding exon is observed in several type II membrane proteins that belong to the C-type animal lectin family (30,31,(62)(63)(64), suggesting that this is an evolutionarily conserved property.
Because DC-SIGN-1 and -2 lack a leader sequence, it is not clear whether loss of the hydrophobic TM-encoding exon would limit the ability of these molecules to traverse across the endoplasmic reticulum membrane, resulting in their retention in the cytoplasm. However, there are examples among the lectin family wherein molecules lacking the secretory signal are externalized by mechanisms other than the classical secretory pathway (65). Notably, certain other cytoplasmic proteins lacking a signal sequence are externalized and function extracellularly. These include IL-1 (66), fibroblast growth factor (67,68), and others (69,70). Alternatively, these TM-lacking DC-SIGN isoforms may function as intracellular molecules. For example, the invariant or ␥ chain, another type II membrane protein, is responsible for targeting the Class II ␣␤ dimers to the endocytic pathway that influences the delivery of antigens (71).
We found that the mRNA expression pattern for DC-SIGN1 was broader than reported previously (17). For example, we Oligo(dT)-primed placenta cDNAs were PCR-amplified with DC-SIGN2-specific primers (primers 1-1 and 2-5). The sense orientation primer was 32 P end-labeled, and the resulting PCR amplicons were fractionated on a 3% polyacrylamide gel. Note that the size markers (in bp) are indicated to the left of the gel, and the degree of separation of radiolabeled amplicons in lanes 1-3 is different from that in lane 4. ࡗ, indicates transcripts for which sizes correspond to intron IV-containing DC-SIGN2 cDNAs. The arrows indicate transcripts unique to placenta sample 4. b, inter-individual variation in the expression of DC-SIGN2 transcripts in placenta. DC-SIGN2-specific primers were used to PCR amplify oligo(dT)-primed cDNAs from 10 different individuals (donors [5][6][7][8][9][10][11][12][13][14], and the products were analyzed as described above. Note the actin controls for the cDNA synthesized from each donor. c, mRNA expression of DC-SIGN2 in DCs derived from cytokine-differentiated CD34 ϩ PBHPs. cloned the transcripts of DC-SIGN1 from THP-1 cells and PBMCs. Expression of DC-SIGN1 mRNA, albeit low was detected in resting PBMCs. In contrast, in PBMCs stimulated with PHA or CD3/CD28 (stimulation of the T cell receptor) there was an increase in DC-SIGN1 mRNA expression. In studies not shown, DC-SIGN1 mRNA expression in PBMCs also increased significantly after stimulation with PMA and ionomycin, a calcium ionophore; this form of stimulation is known to activate the PKC pathway in T cells by bypassing the T-cell receptor. In ongoing studies we are investigating the precise cell types in resting as well as in PHA-, CD3/C28-, and PMA/ionomycin-activated PBMC cultures that express DC-SIGN1 mRNA. It is difficult at the present moment to reconcile the differences between our findings and those of Geijtenbeek et al. (17) whose studies indicated that the expression of DC-SIGN1 is DC-specific. By using a PCR-based strategy they found no mRNA expression for DC-SIGN1 on THP-1 cells, granulocytes, PBMCs activated for 2 days with PHA and IL-2, or peripheral blood leukocytes (17). The reasons for this discrepancy remain unclear but could be related to differences in PCR conditions or primer design. We are currently in the process of generating monoclonal antibodies to determine whether there is a discordance between the levels of DC-SIGN1 mRNA and protein expression. Notably, there are several examples of tissue-or cell type-specific regulation of translation, including that for IL-2 (72)(73)(74)(75)(76)(77)(78)(79)(80).
We found that the genes for DC-SIGN1 (CD209), DC-SIGN2 (CD209L), and CD23 colocalize to an ϳ85-kb region of chromosome 19p13.3. Alternative splicing events in CD23 generates several transcripts including two isoforms (Fc⑀RIIa/CD23a and Fc⑀RIIb/CD23b) that differ only at the N-terminal cytoplasmic region (23,30,31). Interestingly, Fc⑀RIIa (CD23a) and Fc⑀RIIb (CD23b) exhibit differences in their tissue expression, and IL-4 differentially regulates their expression (30,81). These two CD23 isoforms also have differential functions in allergic reactions, immunity to parasitic infections, and B cell development (30,81). As a corollary, we found that alternative splicing of DC-SIGN1 pre-mRNA also leads to the generation of transcripts that are predicted to encode distinct N-terminal regions (DC-SIGN1-A and -B) and that IL-4 differentially regulates the expression of DC-SIGN1 in DCs. There is growing evidence that lectins, including CD23, can serve as cell surface transducers of signals from the outside to the inside of the cell (82,83); in this context, we are currently investigating whether DC-SIGN1-A and -B isoforms activate distinct intracellular signaling pathways.
The biological properties of this large repertoire of DC-SIGN1 and -2 isoforms with respect to their roles in HIV pathogenesis and DC-T cell interactions remains unknown. Changes in splicing have been shown to determine the ligand binding of growth factor receptors and cell adhesion molecules (33,35). The mDC-SIGN1 and mDC-SIGN2 isoforms with varied extracellular domains may bind ligands, including gp120, with varied avidity. Furthermore, in addition to ICAM-3, this extensive array of membrane-associated DC-SIGN1s (and potentially mDC-SIGN2s) may mediate cell-cell contact via interactions with a larger number of specific ligands or adhesion molecules of different protein families. Studies are currently underway to determine whether, similar to the findings in other gene systems, an alternative splice variant of DC-SIGN-1 or -2 cross-regulates or antagonizes the biological activities of the other isoforms (47, 84 -91). For example, an alternatively spliced isoform of CD40 influences the function of the prototypic full-length CD40 isoform (91). An intriguing possibility is that the DC-SIGN-1 and -2 isoforms lacking the transmembrane domain if secreted may act as natural competitive inhib-itors of DC-SIGN/ICAM-3/HIV binding interactions in vivo, or alternatively, they may function in regulating the expression of the membrane forms of DC-SIGN. Furthermore, lectin-binding domains can oligomerize (92)(93)(94)(95)(96), and potentially this oligomerization among the varied membrane forms of DC-SIGN1 or between DC-SIGN1 and DC-SIGN2 isoforms in cell types in which they are coexpressed may further increase the repertoire and specificity of DC-SIGN-like surface proteins available for mediating cell-cell contact.
The prototypic membrane-associated DC-SIGN1 (mDC-SIGN1 Type I) and DC-SIGN2 (mDC-SIGN2 Type I) isoforms have been shown to mediate gp120 adhesivity and potentiate in trans the infection of T lymphocytes by HIV (16,26). By mRNA expression studies and immunostaining, expression for DC-SIGN1 was detected in both placental endothelial cells and CCR5-expressing cells in which distribution was consistent with placental macrophages (Hofbauer cells), a cell type that can support HIV infection (97). We also detected DC-SIGN2 transcripts in the placenta; and while this manuscript was in review, using a DC-SIGN2-specific antiserum, Pohlmann et al. (26) documented expression for DC-SIGN2 in the placental endothelium but not macrophages. The expression of both DC-SIGN1 and DC-SIGN2 in the placenta has important implications for vertical transmission of HIV-1. However, pertinent to our search for genetic determinants that account for the significant inter-individual variability in susceptibility to HIV infection, our studies indicate that DC-SIGN1 and DC-SIGN2 gene expression in the placenta and other cell types may be highly variable. We examined a large panel of placenta samples, and found inter-individual variation with respect to both the levels of expression as well as the repertoire of transcripts expressed. Notably, in some instances, we were unable to detect expression for the prototypic mDC-SIGN2 transcripts in placenta, and transcripts that contained intron IV appeared to be more abundant than the prototypic isoform. Conceivably, inter-individual variation in the generation of DC-SIGN isoforms could account, in part, for host differences in susceptibility to HIV-1 infection, especially vertical transmission.
In summary, while searching for polymorphisms in the gene for DC-SIGN1, we identified another homologous gene designated here as DC-SIGN2 that recently has been shown to serve also as an HIV attachment factor. Notably, we found that alternative splicing of DC-SIGN1 and DC-SIGN2 generates a wide array of transcripts that are predicted to encode both membrane-associated and soluble isoforms. Determining the functional properties of this extensive repertoire of DC-SIGN1 and DC-SIGN2 isoforms in vivo is likely to pose a daunting task, and in this respect it will be important to develop reagents that can discriminate between the different isoforms. In addition, the inter-individual heterogeneity in DC-SIGN expression, especially DC-SIGN2 in placenta, introduces an unanticipated degree of complexity with regard to dissecting the determinants of HIV susceptibility. Nevertheless, this plethora of DC-SIGN-like molecules will serve as a powerful tool to probe HIV-host cell interactions as well as DC-T cell interactions and as a potential target for a novel means to block these interactions. Based on the striking parallels between DC-SIGN-1 and -2 and other alternatively spliced type II membrane proteins such as CD23, we hypothesize that the diverse DC-SIGN isoforms have pleiotropic activities and that they may interact with additional, as yet undiscovered molecules.