A Novel Alternative Spliced Chondrolectin Isoform Lacking the Transmembrane Domain Is Expressed during T Cell Maturation*

Chondrolectin (CHODL) is a novel type I transmem- brane protein containing one carbohydrate recognition domain (CRD) of C-type lectins. Recently, data base searching revealed a variant of CHODL (AK022689) with a different 5 (cid:1) leader sequence derived from a new putative upstream alternative promoter (P 2 ). The P 2 pro- moter gives rise to at least three additional alternatively spliced isoforms, designated as CHODL f , CHODL f (cid:1) E , and CHODL (cid:1) E . Of all variants, the alternative exon E-splic- ing isoforms ( CHODL f (cid:1) E / CHODL (cid:1) E ) are expressed exclu-sively in the T lymphocyte lineage and are regulated during T lymphopoiesis. Peripheral T lymphocytes demonstrated a unique exon E-splicing pattern in compari- son with end maturation stage thymocytes, suggesting its association with the post-thymic

C-type lectins of animals are the most diverse family of animal lectins (1). They are generally multidomain proteins in which carbohydrate recognition domains (CRDs) 1 provide calcium-dependent sugar-binding activity. The C-type CRD has a common sequence module of 115-130 amino acid residues containing four cysteines that are perfectly conserved and involved in two disulfide bonds. Those C-type CRD domains have been found in various kind of proteins such as hepatic asialoglycoprotein receptor (2), lymphocyte Ig E receptor (3), mannose-binding protein (4), selectin (5), and proteoglycan core protein (6).
In general, animal lectins have a wide variety of functions. The principal function is to act as recognition molecules within the immune system and includes complement activation, recognition, and trafficking within the immune system, immune regulation (suppression or enhancement), and prevention of autoimmunity (1). Perhaps the best established functions outside the immune system are lysosomal enzyme transport by the phosphomannosyl receptors and the molecular chaperone role of calreticulin in the endoplasmic reticulum (7,8).
Recently, we have described the isolation and characterization of a novel C-type lectin, designated as chondrolectin (9). The mouse chondrolectin gene (chodl) was first isolated from a subtracted library derived from tails of 4-day-old mice and therefore referred to as chondrolectin. The mouse and human gene structures are very similar, consisting of six exons and five introns (9,10). In both species, the open reading frame for chondrolectin encodes a type I membrane-associated polypeptide of 273 amino acids, containing an N-terminal signal sequence, a single CRD, a transmembrane region (TM), and an intracellular C terminus. Chondrolectin is structurally related to hamster layilin, a hyaluronan receptor (11,12). However, chondrolectin does not contain the functional talin binding site found in layilin. Furthermore, no specific interaction between chondrolectin and hyaluronan was detected by cetylpyridinium chloride precipitation (9). Therefore, chondrolectin is not likely to have the function of a hyaluronan receptor. The distribution of chondrolectin gene expression in normal human tissues is in general very low. RT-PCR analysis revealed preferential expression of chondrolectin in testis, prostate, and spleen, whereas immunohistochemical analysis demonstrated that the expression is mainly limited to vascular muscle of testis, smooth muscle of prostate stroma, heart muscle, skeletal muscle, crypts of small intestine, and red pulp of spleen (9).
Although the physiological function of chondrolectin remains unknown at present, the identification of novel T cell-associ-ated isoforms of chondrolectin is of particular interest, because many proteins that we regard as recognition or adhesion molecules within the immune system are now known to be lectins. For example, CD44 is a hyalectin-related C-type lectin and plays a role in lymphocyte recirculation; impaired homing to lymph nodes and thymus has been found in gene-targeted CD44-deficient mice (13).
In this study, we have identified three new members of the chondrolectin family that result from alternative promotor usage. All splicing variants described in the present study are derived from the distal alternative promoter and are differentially expressed in the T lymphocyte lineage. Two variants are devoid of the transmembrane domain (exon E of the human gene) and terminate in QDEL, a short sequence that is consistent with the motif for the endoplasmic reticulum retention signal. The transmembrane-containing isoform is localized in the ER/Golgi apparatus and is associated with T lymphocyte immaturity. Our results suggest that the alternative promoter usage could be a key mechanism controlling the expression of different chondrolectin isoforms with different biological function in T cell development.

EXPERIMENTAL PROCEDURES
Cell Lines and Antibodies-U937, a human histiocytic lymphoma cell line, and K562, a human erythro-myeloid leukemia cell line, were used as nonlymphoid cell sources. RAJI, a human Burkitt's lymphoma cell line, was used as a source of B lymphocytes. JURKAT, MOLT-3, and HSB-2, human T cell lines derived from acute lymphoblastic leukemia, and SUP-T1, a human T cell line derived from a T cell lymphoblastic lymphoma, were used as a source of human T cells. Cells were obtained from the American Type Culture Collection (ATCC) (Manassas, VA) and cultured following the guidelines of the ATCC.
Isolation and Preparation of Hematopoietic Cells-Bone marrow samples were aspirated by sternal puncture from hematologically normal patients undergoing cardiac surgery. Cells were collected into sterile tubes containing Iscove's modified Dulbecco's medium (IMDM) supplemented with 10% fetal calf serum (FCS) and heparin (100 units/ml; Novo Nordisk, Bagsvaerd, Denmark), and the mononuclear cells were isolated by density gradient centrifugation over lymphocyte separation medium (ICN Biomedicals, Costa Mesa, CA). Samples were obtained after informed consent according to the guidelines of the Medical Ethics Committee of the Antwerp University Hospital. Postnatal thymus was obtained from children 0 -12 years of age undergoing corrective cardiac surgery. Those samples were obtained and used following the guidelines of the Medical Ethical Commission of the University Hospital of Ghent. Single cell suspensions were obtained by gentle disruption of the thymus in IMDM plus 10% FCS. Adult peripheral blood leukocytes were obtained after removing the red blood cells from a blood sample of a hemochromatosis patient by the NH 4 Cl lysis procedure.
Flow Cytometric Cell Sorting-Freshly prepared adult bone marrow mononuclear cells were resuspended in IMDM at 10 7 cells/ml and incubated with 43A1 supernatant in a 1:10 dilution for 20 min at 4°C, washed twice in IMDM supplemented with 10% FCS, and incubated with rabbit anti-mouse FITC (1:50 dilution) for 20 min at 4°C. After washing twice in IMDM plus 10% FCS, the cells were incubated with mouse ␥-globulins for 10 min and incubated with anti-CD38-PE for 20 min at 4°C. After washing twice in IMDM plus 10% FCS, the cells were sorted on a FACStarPlus Cell Sorter (Becton Dickinson) equipped with a water-cooled argon ion laser (Coherent Innova Enterprise Laser), tuned to 488 nm at 170-milliwatt power. Cells with a low to medium forward scatter and a low side scatter, highly positive green (CD34) fluorescence, and an orange (CD38) fluorescence signal lower than the mean fluorescence plus 2 S.D. values of cells labeled with an irrelevant isotope-matched control antibody were retained as CD34 ϩ CD38 Ϫ cells; cells with an orange fluorescence above this threshold were retained as CD34 ϩ CD38 ϩ cells.
Lymphocytes, monocytes, and granulocytes were sorted from peripheral blood based on the typical light scattering properties of the respective populations (i.e. low forward and orthogonal, intermediate forward and orthogonal, and intermediate forward and high orthogonal light scattering, respectively). T and B lymphocytes were further separated by fluorescence activated cell sorting after staining for CD3 (with anti-CD3-FITC) or CD19 (with anti-CD19-PE), respectively.
The purity of the sorted cell populations was always checked and was at least 98%.
PCR, Southern Blot, and RT-PCR-The expression from the distal and proximal promoters was analyzed using PCR and subsequent Southern blot analysis. Primer sets for transcripts of the distal and proximal promoters were designed. The S 1 primer set, derived from the human CHODL cDNA (AF257472), was used for analyzing the expression of the proximal promoter: forward, 5Ј-cttcactggctttcctg; reverse, 5Ј-tacagttcatcttgcagatc. The S 2 primer set, derived from the human CHODL homologue (AK022689), was used for analyzing the distal promoter: forward, 5Ј-aggctgcagagtcagagtc; reverse, 5Ј-tacagttcatcttgcagatc. First strand cDNAs of multiple human tissues (Clontech) were used as templates. The temperature profile for both PCRs was as follows: 95°C for 3 min and 36 cycles of 94°C for 10 s, 50°C for 30 s, and 72°C for 1.5 min. The final cycle included an extension at 72°C for 10 min. Resulting PCR products were fractionated on 1% agarose gel and transferred to a Hybond-N membrane, and subsequent Southern blot hybridization was performed according to a standard procedure (15). Probes for hybridization were PCR amplicons derived from primer sets S 1in and S 2in , which were internally located within the S 1 and S 2 PCR primer sets, respectively. S 1in primers were as follows: forward, 5Ј-gcttgaagatggcagatcgtgg; reverse, tctgatgccattccttatacagatccag. S 2in primers were as follows: forward, tcgctccacgcaacacctgc; reverse, tctgatgccattccttatacagatccag. The temperature profiles were as follows: 95°C for 3 min and 36 cycles of 94°C for 10 s, 58°C for 30 s, and 72°C for 1.5 min. The final cycle included an extension at 72°C for 10 min.
The exon E-splicing pattern in multiple human tissues was analyzed using a primary PCR and nested PCR. The primers used were derived from the exon D and exon E of human CHODL separately: forward, 5Ј-aagctgaaagatgaattatttcg; reverse, 5Ј-atgtatcaccaaccaactgcc. The primers used for the nested PCR were internally adjacent to the primary primers: forward, 5Ј-ttctatggtactatatttgtgtg; reverse, 5Ј-acaggtgtaacatgaagcac. First strand cDNAs of multiple human tissues (Clontech) were used as templates. The temperature profile of the PCR reaction was as follows: 95°C for 3 min and 36 cycles of 94°C for 10 s, 50°C for 30 s, and 72°C for 1.5 min. The final cycle included an extension at 72°C for 10 min. The exon E-splicing pattern in human hematopoietic cell lines (HSB-2, SUP-T1, JURKAT, MOLT3, K562, RAJI, and U937) was analyzed using RT-PCR and subsequent nested PCR. First strand cDNA synthesis was primed by oligo(dT) [12][13][14][15][16][17][18] and catalyzed by the RNase H derivative of Moloney murine leukemia virus reverse transcriptase (Invitrogen) after poly(A) ϩ RNA was extracted and selected using the Quickprep TM micro-mRNA purification kit (Amersham Biosciences). The PCR and subsequent nested PCR were carried out as described above.
To more precisely define the expression of exon E-splicing variants in the K562, RAJI, and U937 cell lines, an RT-PCR analysis of the human ␤-actin mRNA (as the internal reference template) was performed. The primers used were as follows: forward, 5Ј-cgtgcgtgacattaaggagaag; reverse, 5Ј-agagaagtggggtggcttttag. The temperature profile of the PCR was as follows: 95°C for 3 min and 36 cycles of 94°C for 10 s, 60°C for 30 s, and 72°C for 1 min. The final cycle included an extension at 72°C for 10 min.
All PCRs were carried out in 200-l thin-walled tubes using a PerkinElmer Life Sciences 9600 apparatus. PCR products were fractionated on a 1% agarose gel and visualized under UV light after ethidium bromide staining. Primers mentioned above were synthesized by BIOSOURCE (Nivelles, Belgium).
Cloning and Sequencing of CHODL Variants from Leukocytes-CHODL variants of leukocytes were amplified using PCR and nested PCR. Primer sets (S 2 and S 2in ) and PCR conditions were described above. Nested PCR products were fractionated on a 1% agarose gel and stained with ethidium bromide. Amplicons were visualized under UV light. Distinct bands were excised and subsequently cloned into pCRscript plasmid according to the manufacturer's instructions (Stratagene, La Jolla, CA). Clones designated as pUIA847, pUIA848, and pUIA849 were sequenced using the dye terminator method (Amersham Biosciences) by an automated ABI type 373 DNA sequencer (Applied Biosystems).
Construction of Fluorescent Protein-tagged CHODL Isoforms-Isoforms of CHODL were separately tagged with in-frame sequences of red fluorescent protein (RFP) and GFP. The coding sequences of variants were amplified from cloned splice variants. The primers used were as follows: forward, 5Ј-tcgctccacgcaacacctgc; reverse, 5Ј-actctcgagttaagcataatcaggaacatcataaggatatacttccatgccactttcttttctggtac or 5Ј-agtaaagcataatcaggaacatcataaggatacagttcatcttgcagatcctttgtg. The reverse primer contains 27 in-frame nucleotides encoding a hemagglutinin epitope (YPYDVPDYA) to facilitate the recognition of this fusion protein by the anti-hemagglutinin monoclonal antibody. The temperature profile of the PCRs was as follows: 95°C for 3 min and 36 cycles of 94°C for 10 s, 60°C for 30 s, and 72°C for 1.5 min. The final cycle included an extension at 72°C for 10 min. PCR products were fractionated on a 1% agarose gel, and the amplicons were subcloned into the pCR-Script plasmid (Stratagene, La Jolla, CA). After confirmation of sequences, inserts of pCR-Script plasmids were excised with EcoRI and XbaI and subsequently ligated into the same restriction enzyme-digested pD-sRed1-N1/pEGFP-N1 vectors (Clontech). The resulting constructs RFP-CHODL f (pUIA850), GFP-CHODL ⌬E /GFP-CHODL ⌬fE (pUIA851/852) were verified by sequencing analysis. RFP-CHODL f consists of the exon E-containing isoform (CHODL f ) followed by an in-frame RFP sequence. GFP-CHODL ⌬E and GFP-CHODL ⌬fE consist of the exon E-skipping isoform (CHODL ⌬E /CHODL ⌬fE ) followed by an in-frame GFP sequence. FLAG-tagged CHODL was subcloned from CHODL f using following primers: forward, 5Ј-agcggccgcaatggcctacttccatg; reverse, 5Ј-actcgagttatacttccatgcc. The forward primer contains a restriction site which ligates to the in-frame flag sequence of pcDNA3.1 (Invitrogen). This construct (pUIA 855) was verified by sequencing analysis.
Cell Culture, Transfection, and Fluorescence Microscopy-COS1 cells were cultured and maintained in Dulbecco's modified Eagle's medium/ nutrient mixture F-12 medium supplemented with 10% FCS and 10 mM glutamine at 37°C in a humidified 5% CO 2 incubator. After COS1 cells were passaged and subcultured on coverslips up to 50 -70% confluence, they were transfected with a mixture of both constructs (RFP-CHODL f and GFP-CHODL⌬ E ) using FUgene6 (Roche Molecular Biochemicals) and cultured up to 72 h. Cells on coverslips were washed three times with phosphate-buffered saline, fixed with methanol for 10 min, and mounted on glass slides. For the immunofluorescent staining, the coverslips were incubated with the primary antibodies (rBet1/Bcl-w/ LAMP1/KDEL, markers for ER-Golgi apparatus, mitochondrium, lysosome, and the KDEL receptor, respectively) and subsequently with FITC-conjugated goat anti-rabbit/mouse IgG antibodies. Fluorescent staining of transfected COS1 cells was viewed by fluorescence microscopy (Leica DMRA) using 488-nm/570-nm filters for excitation of the green/red fluorochrome, respectively.

RESULTS
Evidence for Transcription Initiation Upstream of the CHODL P 1 Promoter-Data base (GenBank TM /EMBL) searching with the CHODL sequence revealed several homologues (AF311699.1, AK014255, BC009418, AK022689). All of these cDNAs are derived from brain tissues. AK022689 displays high identity to the CHODL cDNA sequence but is different in its 5Ј-untranslated region. This CHODL variant codes for a pro-tein of 232 amino acids (aa), since its start codon corresponds to the second ATG of CHODL, resulting in an isoform lacking the 41 first amino-terminal aa of CHODL. Analysis of its nucleotide sequence and comparison with the human genomic DNA sequences (16) of chromosome 21 revealed that AK022689 is derived from a promoter region ϳ330 kb upstream of the CHODL P 1 promoter. The existence of this transcript provides evidence for the usage of an alternative promoter for the CHODL gene (Fig. 1). Due to the low expression of CHODL, the expression from the proximal (P 1 ) and distal (P 2 ) promoters was analyzed using PCR and subsequent Southern blotting. Specific expression from the proximal (P 1 ) promoter was detected in human adult colon, small intestine, thymus, and pancreas ( Fig. 2A); specific expression from the distal (P 2 ) promoter was detected in human adult leukocytes and placenta (Fig. 2B). Expression from both proximal (P 1 ) and distal (P 2 ) promoters was found in human adult prostate, spleen, and testis (Fig. 2, A and B). Neither expression from the distal promoter nor expression from the proximal promoter was detected in human adult ovary, brain, heart, liver, lung, and muscle. We detected significant molecular heterogeneity of the PCR products derived from the distal promoter (Fig. 2B). Subsequent cloning and sequencing of these PCR products revealed that this heterogeneity is the result of alternative splicing, suggesting the presence of variable CHODL isoforms.
Characterization of CHODL Variants Expressed in Leukocytes-As noted above, we detected molecular heterogeneity in RT-PCR amplicons, suggesting the presence of different CHODL isoforms. This hypothesis was further tested by PCR analysis using primers flanking exon E, a putative transmembrane domain of CHODL. As shown in Fig. 3, this primer set amplified a PCR product (smaller than the expected one) unique for leukocytes, suggesting the existence of an alternative exon E-splicing variant. PCR and subsequent nested-PCR using primer sets (S 2 and S 2in ) amplified three distinct amplicons from cDNA of leukocytes. Subsequent cloning and sequencing analysis revealed that they represent three novel variants of CHODL. The nucleotide sequence of three cDNAs has been submitted to GenBank TM /EMBL data base with ac- cession numbers AF523313, AF523314, and AF523315. The first novel variant, CHODL f , contains an insertion of exon (exon 3) between exon 2 and exon B (Fig. 1). The second novel variant, CHODL f⌬E , contains an insertion of exon 3 between exon 2 and exon B and skips exon E (Fig. 1). The third novel variant, CHODL ⌬E , does not contain exon 3 and skips exon E (Fig. 1). We failed to isolate 5Ј-rapid amplification of cDNA end clones of these cDNAs and are not in a position to define the precise start of transcription and to determine whether or not they contain exon 1 of AK022689. Since the 5Ј leader sequences of these novel variants contain an in-frame stop codon without additional in-frame ATG, variants with the insertion of exon 3 still share the same translation initiation site in exon B as variant AK022689 (i.e. the second ATG of the CHODL gene). Exon E is located within the coding region of CHODL and encodes the putative transmembrane region (9). Therefore, alternative splicing of this exon results in a non-transmembrane-containing isoform of CHODL with an altered carboxylterminal sequence due to a frame shift (Fig. 4, A and B). The non-transmembrane-containing isoforms or soluble isoforms (designated as CHODL ⌬E /CHODL f⌬E ) consist of 236 aa with an approximate molecular mass of 27.38 kDa and a pI of 4.8 (see, on the World Wide Web, us.expasy.org/cgi-bin/protparam) (Fig.  4B). A profile searching on ScanProsite (available on the World Wide Web at us.expasy.org/cgi-bin/scanprosite) revealed that CHODL ⌬E /CHODL f⌬E contains a CRD motif (aa 1-138), three N-glycosylation sites (aa 45-48, 199 -202, and 220 -223), two protein kinase C phosphorylation sites (aa 8 -10 and 100 -102), two casein kinase II phosphorylation sites, and one endoplasmic reticulum retention sequence (QDEL) at the very C terminus end (aa 233-236).
Alternative Exon E Splicing in Hematopoietic Cells-Leukocytes are a heterogeneous population of hematopoietic cells. In order to define more precisely the cellular site of alternative exon E splicing, we performed a more detailed RT-PCR analysis on subsets of peripheral blood leukocytes. Four subsets of peripheral blood leukocytes (peripheral T lymphocytes, B lymphocytes, monocytes, and granulocytes) were sorted by flow cytometry. Their exon E-splicing pattern was analyzed using RT-PCR. The exon E-containing variant and exon E-skipping variants were discriminated based on the size of their amplicons. As shown in Fig. 5A, B lymphocytes express the exon E-containing variants only; peripheral blood T lymphocytes (CD3 ϩ CD4 ϩ CD8 Ϫ and CD3 ϩ CD8 ϩ CD4 Ϫ cells) express exon E-skipping variants only; granulocytes and monocytes express neither exon E-containing nor exon E-skipping variants. Since the expression of the exon E-skipping variants is restricted to T lymphocytes, we investigated the expression of the exon Esplicing variants during the development of T lymphocytes. Bone marrow progenitor cells (immature CD34 ϩ CD38 Ϫ cells and more mature CD34 ϩ CD38 ϩ cells) express the exon E-containing variant only. Thymocytes at different stages of maturation (CD34 ϩ CD1 Ϫ cells, CD34 ϩ CD1 ϩ cells, immature single CD4 ϩ cells, double positive CD4 ϩ CD8 ϩ cells, and single CD8 ϩ cells) express the exon E-containing variant predominantly and the exon E skipping variants slightly. Single mature CD4 ϩ and more mature thymocytes express only the exon E-containing variant. Peripheral blood T cells (CD3 ϩ CD4 ϩ CD8 Ϫ cells or CD3 ϩ CD8 ϩ CD4 Ϫ cells) express only the exon E-skipping variants (Fig. 5B). The differential utilization of exon E-splicing variants intra-and extrathymus suggests their association with the differentiation of T lymphocytes. In accordance with its T cell origin, HSB-2, SUP-T1, JURKAT, and MOLT-3 cells express the exon E-skipping variants. The presence of the exon E-containing variant is compatible with the immature stage of those leukemic cells. The exon E-containing variant is present FIG. 2. Tissue-specific expression of CHODL mRNA. The tissue specificity of promoter utilization was analyzed using PCR. First strand cDNA of multiple human tissues was used as template. Primer sets for analyzing the expression of the proximal (P 1 ) and distal (P 2 ) promoters were indicated separately (see Fig. 1) by single and double arrows. PCR products were fractionated on a 1% agarose gel and subsequently transferred to a Hybond-N membrane. Tissue expression was detected by a Southern blot as described under "PCR, Southern Blot, and RT-PCR." Tissuespecific expression from the proximal promoter ( Fig. 2A) and from the distal promoter (Fig. 2B) are shown. A is from Ref. 9 and is used for comparison purposes with B. Notice that the size of PCR products using primer set S 2 is heterogeneous.

FIG. 3. Identification of truncated CHODL transcripts in leukocytes.
Exon E-splicing variant in human multiple tissues was detected using a primary PCR and subsequent nested PCR. The first strand cDNAs of several human tissues were used as templates, and the primers used are derived from exon D and exon F separately (see Fig. 1). The PCR products were fractionated on a 1% agarose and visualized by ethidium bromide staining. The amplicons in leukocytes were smaller than those of the other tissues.
in RAJI cells, a B-cell line, and in K562, a erythromyeloid cancer cell line. The exon E-containing variant is absent in the monocyte-derived cell line U937 (Fig. 5C).
Subcellular Localization of CHODL f /CHODL f⌬E Isoforms-To probe the biological function of CHODL variants, we analyzed the expression and subcellular localization in transfected COS1 cells. The RFP-tagged exon E-containing isoform (RFP-CHODL f ) and GFP-tagged exon E-skipping isoform (GFP-CHODL ⌬E ) were constructed separately and subsequently used for transfection of COS1 cells. Western blot analysis revealed both fusion proteins as a single band of similar molecular mass (ϳ85 kDa) (Fig. 6). To determine the subcellular localization of both isoforms, we performed immunofluorescence microscopy. In all cells in which fusion protein was detected, RFP-CHODL f (Fig. 7A) appeared as distinct ringlike structures, localized at the perinuclear region and/or along the extension of cells, whereas the GFP-CHODL f⌬E isoform revealed a diffuse distribution in the cytoplasm (Fig. 7B). In order to evaluate whether a fluorescent protein tag can interfere with the localization of proteins to which it is fused, we subsequently used a small FLAG epitope-tagged CHODL f construct in a double label immunofluorescence experiment. This construct has a FLAG epitope fused in-frame to CHODL f at its NH 2 terminus. Both constructs, FLAG-tagged CHODL f and RFP-CHODL f , were overexpressed in COS1 cells (Fig. 7C). The coincidence of both proteins in the same cells is clearly visible and indicated by yellow staining. This demonstrates that the localization of RFP-CHODL f is not affected by RFP. To more precisely map the subcellular localization of the CHODL f isoform, co-immunostaining experiments were performed with ER, Golgi, mitochondrium, and lysosome markers, respec- Poly(A) ϩ RNA was extracted from FITC-sorted hematopoietic cells and cell lines. First strand cDNA synthesis was primed with oligo(dT) [12][13][14][15][16][17][18] and catalyzed by the RNase H Ϫ derivative of Moloney murine leukemia virus reverse transcriptase. The subsequent PCR and nested PCR were performed using primers flanking exon E as described under "PCR, Southern Blot, and RT-PCR." PCR products were fractionated on a 1% agarose gel and visualized by ethidium bromide staining. The 866-bp amplicon is derived from exon E-containing variants, whereas the 764-bp amplicon is derived from exon E-skipping variants. Alternative exon E splicing patterns in subsets of leukocytes (A), during T lymphopoiesis (B), and hematopoietic cell lines (C) are shown. Images are representative for the experiment performed in duplicates.

FIG. 4. Deduced amino acid sequences of novel CHODL isoform.
A, CHODL, CHODL f, and CHODL f⌬E /CHODL ⌬E are shown schematically. SS, in-frameshift sequence. SP, signal peptide. B, the deduced amino acid sequences of CHODL f⌬E /CHODL ⌬E . Letters in italics represent the amino acid sequence (aa 1-138) that is part of the CRD of CHODL f⌬E /CHODL ⌬E . The arrowhead indicates the site of exon E skipping. Gray shading, frame-shifted sequence due to exon E skipping. Potential N-glycosylation sites in the frame-shifted sequence are underlined. The asterisks indicate the putative retention signal for the endoplasmic reticulum.
tively. The red signal of RFP-CHODL f was observed in ringlike structures and shown to co-localize with the green signal of rBet1, a marker for the ER-Golgi apparatus (Fig. 7D). DISCUSSION Previously, we reported that the human CHODL gene is transcribed from a TATA-containing promoter (designated as P 1 ), located between Ϫ32 to Ϫ26 (9). Recently, a CHODL homologous cDNA (AK022689) with different 5Ј leader sequence has been reported in GenBank TM . Sequence comparison between the cDNA and its cognate genomic DNA revealed that AK022689 is derived from an alternative promoter (P2), 330 kb upstream to the transcription initiation site P 1 . The expression pattern from P1 and P2 varies between tissues, suggesting that the promoter usage is tissue-dependent. The modulation of transcription from P2 is quite complex. Alternative splicing of transcripts renders variants that differ either in the 5Ј-untranslated region or coding region, thus resulting in a different protein translation efficiency or different proteins. In the present study, we describe the existence of at least four mRNAs derived from the putative, alternative promoter P2 with different 5Ј-untranslated regions but all sharing the second translation initiation site of the chondrolectin gene. The large sequence (ϳ330 kb) spanning between the P1 and P2 promoters also raises the possibility for the existence of other potential promoters and exons that have not been identified yet. The discrepancy of expression patterns (Figs. 2 and 3) using different primer sets probably reflects complex splicing patterns generated during the transcription of the chondrolectin gene and thus support the hypothesis mentioned above. A more extensive characterization of P2-derived transcripts will be necessary in order to understand the modulation of transcription of the chondrolectin gene.
In the present study, we have identified three novel CHODL splice variants (CHODL f , CHODL ⌬E , and CHODL ⌬fE ) derived from the P2 promoter and expressed in T cells. These variants are attributed to alternative splicing of exon 3 and exon E, either by insertion of an additional exon (exon 3 for CHODL f ) or by skipping of exon E (CHODL ⌬E ) or by a combination of both (CHODL ⌬fE ). Insertion of exon 3 does not change the protein sequence, because in-frame sequences of exon 3 do not contain an in-frame ATG. The skipping of exon E results in a nontransmembrane-containing or soluble isoform of CHODL with a frame-shifted tail ending with the sequence QDEL. All isoforms that we identified in T cells lack the signal peptide but contain the whole carbohydrate recognition domain. This indicates that these isoforms share the same ligand binding specificity of the CRD.
One of the striking findings in this study relates to the expression pattern of these variants, particularly to the expression profile of the transmembrane-containing/nontransmembrane isoforms in T lymphocytic cells. The expression of both isoforms was analyzed by RT-PCR using a common primer set flanking both sides of exon E. The exon E splicing pattern in hematopoietic cells can be categorized into three types: predominant exon E-containing cells, predominant exon E-skipping cells, and intermediate cells (expressing both exon E-containing and exon E-skipping variants). Hematopoietic progenitor cells (CD34 ϩ CD38 Ϫ and CD34 ϩ CD38 ϩ bone marrow cells) are predominantly exon E-containing cells, and subsets of thymocytes are either predominantly exon E-containing or intermediate type of cells, whereas peripheral T lymphocytes are predominantly exon E-skipping cells. Hematopoietic cell lines revealed a similar exon E splicing pattern as their correspond- ing hematopoietic cells. In general, the expression of the exon E-skipping variant is associated with the maturation of T lymphocytes, whereas the expression of the exon E-containing variant is associated with T lymphocyte immaturity. In this regard, it is interesting to note that peripheral T lymphocytes (CD4 ϩ and CD8 ϩ ) and end stage thymocytes (single CD4ϩ or single CD8ϩ) have a different exon E-splicing pattern (predominantly exon E-containing type and predominantly exon Eskipping type, respectively). This discrepancy probably reflects a fine difference during the post-thymic maturation process. It has been reported that end stage thymocytes are phenotypically immature and progressively acquire the phenotypic attributes of more mature T cells a few days after release from the thymus (17,18). This unique phenotype of end stage thymocytes could be used as a potential marker for the identification of recent thymic emigrant (RTE), a newly released population of naive T cells, since this phenotype could still be retained on RTE while disappearing during the maturation process. Recently, CD103 was reported as a feature of the RTE phenotype in humans, which is up-regulated on late CD8 ϩ CD4 Ϫ CD3 bright thymocytes. This phenotype appears to be retained on a distinct subset of naive CD8 ϩ T cells in the periphery with the expected characteristics of RTE (19). The identification of human RTE is very important for the study of the development of the naive T cell repertoire, the regulation of peripheral T cell homeostasis and naive T cell regeneration after intensive cytotoxic chemotherapy or effective antiretroviral therapy of progressive HIV infection.
Alternative splicing of the chondrolectin gene during the post-thymic maturation process might also be a predictor of certain functional attributes. This is illustrated in the example of the alternative splicing variants of CD45. Variants represent different stages of T cells during maturation, but the corresponding isoforms (CD45RO ϩ T cells and CD45RO Ϫ T cells) display different immunological functions (20,21). To elucidate the significance of differential exon E splicing during T cell development, it is necessary to understand the function of the full-size isoform CHODL f . We initiated the functional analysis of CHODL f by studying its subcellular localization. Since CHODL f is expressed in various human tissues, it could play a basic role in cells. Therefore, COS1 cells were selected for the transfection and expression of CHODL f . The fluorescent protein-tagged CHODL (RFP-CHODL f ) was used for studying the subcellular localization of CHODL f after demonstrating that the protein localization is not influenced by the fluorescent tag. RFP-CHODL f was found in ringlike structures, arrayed perinuclearly and/or along the extension of cells. A localization at the plasma membrane was never observed. Fluorescent protein-tagged CHODL f⌬E (GFP-CHODL f⌬E ) appears in granulelike structures, scattered throughout the whole cytosol. The shift in localization from ring-like structures of RFP-CHODL f to granule-like structures of GFP-CHODL f⌬E suggests that the transmembrane region (exon E) is crucial for its anchoring in intracellular membranes. Using double label immunofluorescence with several organelle markers, we demonstrated that RFP-CHODL f colocalizes with rBet1, a transmembrane protein that mediates protein transport between the ER and the Golgi apparatus. This is clearly reminiscent of calnexin, calreticulin, and VIP36. Calnexin (membrane-bound) and calreticulin (sol-uble) are homologous ER lectins that bind transiently to virtually all newly synthesized glycoproteins. They promote the correct protein folding and provide quality control by preventing incompletely folded glycoproteins from exiting the Golgi complex (7,8). CHODL f⌬E has a similar domain structure as CHODL f but lacks the transmembrane domain and terminates at the C terminus in a QDEL sequence. It is well known that proteins that possess a KDEL at the end of the C terminus reside in the lumen of the ER. Lysine can be replaced by Glutamine without affecting its retention to the ER (22). Although CHODL f⌬E contains an ER retention signal (QDEL), the GFP-CHODL f⌬E isoform was not found in the ER. Most likely, this can be explained by the fact that the QDEL-containing tail was extended in frame by the GFP protein. Therefore, it will be necessary to analyze the subcellular localization of CHODL f⌬E with a free QDEL at its carboxyl-terminal end. The development of specific antibodies for CHODL f⌬E isoform will be very helpful in defining the precise subcellular localization and permit to assess the protein distribution of the different CHODL isoforms. Furthermore, the identification of the carbohydrate ligand of chondrolectin will provide some insight into the physiological function of the novel chondrolectin isoforms in the development of T cells.
In summary, this study reports for the first time the existence of novel members of the chondrolectin family of C-type lectins associated with T cell maturation. Of all of these isoforms, the transmembrane containing isoform CHODL f is localized at the ER-Golgi apparatus.