Human homologs of the Xenopus oocyte cortical granule lectin XL35

Abstract

The cDNAs encoding two human homologs of the Xenopus oocyte lectin, XL35, were isolated from a small intestine cDNA library and termed HL-1 and HL-2. The deduced amino acid sequence of each homolog is about 60% identical and 80% similar to that of XL35, and none of these sequences contains the C-type lectin motif, although it is known that XL35 requires calcium for ligand binding. By Northern analysis, HL-1 transcripts are present at relatively high levels in heart, small intestine, colon, thymus, ovary, and testis. HL-2 transcripts, by contrast, are expressed only in small intestine. Immunocytochemistry using a polyclonal antibody produced against XL35 shows HL-1 protein to be localized exclusively in endothelial cells in colon, thymus, liver, and other tissues. Primary cultures of human aortic endothelial cells are positive for HL-1 expression by immunoblotting and by PCR analysis, but several other human cell types are not. HL-1 and -2 are both encoded at chromosome 1q23, the same locus that encodes the selectins. XL35, HL-1 and -2, and another mouse homolog are members of a new family of proteins whose members most likely perform diverse functions.

Received on June 27, 2000; revised on September 8, 2000; accepted on September 12, 2000.

Introduction

Vertebrate lectins perform critically important, diverse functions including cell–cell adhesion and pathogen surveillance (Varki, 1999). A specialized function for the lectin found in Xenopus laevis oocyte cortical granules, termed XL35, is to prevent polyspermy. This oligomeric lectin is released extracellularly at fertilization and binds to its polyvalent glycoprotein ligand that is cross-linked in the jelly coat layer surrounding the oocyte (Monk and Hedrick, 1986; Nishihara et al., 1986; Roberson and Barondes, 1982). The ensuing calcium-dependent multivalent interaction participates in the raising of the fertilization membrane that is impenetrable to sperm (Quill and Hedrick, 1996). The lectin is expressed again at gastrulation and may function as well in cell–cell or cell–matrix adhesion events in the embryo (Roberson and Barondes, 1983; Outenreath et al., 1988; Lee et al., 1997).

We isolated the cDNA encoding XL35 from a Xenopus oocyte cDNA library (Lee et al., 1997) and determined that the amino acid sequence it encodes did not display the C-type lectin motif, although it does require calcium for binding (Drickamer, 1993). We have recently identified sequence homologs of XL35 and describe here the characterization of two cDNAs, each with an unusual tissue-specific expression pattern, termed HL-1 and HL-2. One of the homologs, HL-1, is expressed exclusively in the vascular endothelial cells in a unique set of tissues, as well as in the endocardium. The other, HL-2, is expressed only in small intestine. These and other results demonstrate the existence of a family of proteins with homology to XL35.

Results

Isolation of two cDNAs homologous to XL35 from a human small intestine library

The XL35 cDNA sequence was analyzed using FASTA and TBLST programs to search for DNA sequence similarity. A single entry was identified from non-Xenopus sources consisting of a 251 bp cDNA sequence from a human heart cDNA library (GenBank accession number Z36760) as an EST (expressed sequence tag) (Figure 1, underline). Using this human EST sequence to design oligonucleotide primers, an amplimer was obtained from human liver, spleen, and placenta cDNA sources that was essentially identical (>98%) to the human heart cDNA sequence in the database. This amplimer was used to probe Northern blots of various human tissue RNAs, and strong signals were detected from a ∼1.3 kb mRNA in heart and small intestine (data not shown).

The amplimer was next used to probe a human small intestine λgt 10 cDNA library, and a total of 13 positive plaques showing different sizes between 0.7 and 2.4 kb were isolated. DNA sequencing was performed on seven cDNA clones that had a size greater than 1.0 kb. Sequence data indicated that six clones had the same sequence (named HL-1), but only one clone (HL-2) showed a sequence different from the other 6 clones (Figure 1). These two cDNA sequences showed 85% identity to one another at the deduced amino acid level (Figure 2). A striking result was obtained when the deduced amino acid sequences of the two human homologs were aligned with that of XL35. The overall amino acid identity between HL-1 and XL35 was 60% (similarity, 74%) with a 56% amino acid identity (similarity, 74%) between HL-2 and XL35. The open reading frame for HL-1 was the same size as that of XL35, 313 amino acids, while HL-2 was predicted to have 325 amino acids. HL-1 and HL-2 have two and one consensus N-linked glycosylation sites, respectively. These sites are conserved with those of XL35, which has a total of three potential sites (Figure 2). Analysis of the N-termini sequence of all three proteins using a signal peptide sequence analysis program (Nielsen et al., 1997) indicated that the N-terminal portion of XL35, HL-1, and HL-2 is composed of hydrophobic amino acids, suggesting the presence of the signal peptide sequence that causes proteins to enter the secretory pathway. XL35 is known to be secreted from the oocyte’s cortical granules at fertilization (Monk and Hedrick, 1986). A different sequence analysis program, however, predicted that HL-2 would have a single transmembrane domain and short cytoplasmic tail, probably due to the multiple methionine residues at its N-terminus (Hofman and Stoffel, 1993).

Genomic DNA screening

In order to identify the chromosomal location of the two human homolog sequences, primers were designed to regions of the HL-1 and HL-2 cDNAs whose nucleotide sequences were distinct from each other (9/25 and 19/25 mismatches, respectively, dashed and underlined sequences in Figure 1). These primers produced clear, unique 161 bp PCR products with each cDNA sequence upon amplification with human genomic DNA. These primer pairs were then used for genomic DNA screening and also used directly for chromosomal mapping. Two genomic clones containing HL-1 and HL-2 were isolated by Genomic Systems, Inc. A comparison of the partial DNA sequences of these genomic clones indicated that the HL-1 and HL-2 genes have at least one intron that differs in size and location (work in progress). Moreover, PCR reactions from several combinations of synthetic primers and each genomic clone revealed products of different size. These results demonstrate that HL-1 and HL-2 cDNA clones are different gene products.

Chromosomal localization

To analyze the chromosomal localization of the HL-1 and HL-2 genes, FISH (fluorescence in situ hybridization) analysis was performed using genomic clones encoding HL-1 and HL-2 isolated by Genomic Systems, Inc. Both clones hybridized to chromosome 1q band 23 (data not shown). Recently, fragments of genomic sequence in the human genome database confirmed our results that localized both HL-1 and -2 at chromosome 1q 22–23.5. One cosmid clone of 190 kb contained both HL-1 and -2 sequences. This result suggests that the two sequences arose most likely from a gene duplication event that ultimately resulted in proteins with similar sequences, but with tissue-specific promoters. Intriguingly, the genes that encode the selectin family of lectins are found between 1q22 and 1q25 (Watson et al., 1990).

Tissue distribution of HL-1 and HL-2 mRNA

The tissue transcript levels of HL-1 and HL-2 were determined by Northern blot analysis using the 161 bp radiolabeled probe described above. The major transcripts of both HL-1 and HL-2 were 1.3 kb in length (Figure 3A,B). The transcript of HL-1 was observed in relatively very high levels in heart, colon, small intestine, and thymus, with lower levels in ovary, testis, and spleen (Figure 3B). Additional Northern analyses also showed strong signals in lymph node and stomach (data not shown). By stark contrast to HL-1, the transcript of HL-2 was detectable exclusively in small intestine (Figure 3A). The size of this transcript was 1.3 kb, similar to that of HL-1. HL-2 is most likely more abundant than HL-1 in small intestine, since 6 of 7 clones isolated from the small intestine cDNA library contained HL-2 cDNA sequences.

In an earlier experiment using nucleotide primers that did not distinguish between the two transcripts, PCR analysis also detected transcripts in human liver mRNA, although this tissue was negative by Northern analysis (Figure 3A). Subcloning and sequencing confirmed that the PCR product corresponded to an HL sequence. This result suggests that transcripts for HL-1 and -2 may be present in tissues that are negative for expression by Northern analysis; for example, if the cell type expressing them constitutes a very minor fraction of cells in a particular tissue.

Cell type specific expression of HL-1

To determine which cell types expressed HL-1, immunohistochemistry was performed on human tissue sections using rabbit polyclonal antisera prepared against affinity-purified XL35. Striking results were obtained from colon, liver and thymus sections using anti-XL35 primary antisera and peroxidase-conjugated secondary antibody. The antisera specifically and intensively stained endothelial cells lining the blood vessels in these tissues (Figure 4), while pre-immune serum staining was negative. In alltissues, the labeling appeared throughout the endothelial cells, not just adsorbed to the cell surface. In the colon, endothelial cells lining vessels in both the submucosa (arrows) and the lamina propria (arrowheads) were stained with high intensity (Figure 4B,C), consistent with the high level of transcripts observed by Northern analysis (Figure 3A). Staining of liver revealed intense immunoreactivity in both the small hepatic arteries and hepatic veins in the portal tract (Figure 4E, arrows), as well as weaker reactivity of endothelial cells lining the hepatic sunusoids (Figure 4E, arrowhead). In thymus, prominent staining of vessels was noted (Figure 4G). An identical pattern of staining was noted in tonsil, a specialized lymph node, including high endothelial venule cells (data not shown). In heart sections, the endocardium, derived from endothelium, and heart blood vessel endothelial cells showed immunoreactivity, with weak interstitial staining of myocardium (data not shown). These results show that the HL-1 protein is expressed in the endothelial cells in specific tissues, consistent with the tissue specific mRNA expression pattern we observed.

HL-1 expression is restricted to endothelial cells

Primary cultures of human aortic endothelial cells were extracted, separated by reducing SDS–PAGE, and examined by Western blotting using anti-XL35 polyclonal antibody. As shown in Figure 5, human aortic endothelial cells (HE) contained a dominant species of ∼42 kDa. An identical band was observed in human aortic endothelial cells that had been pretreated with the endothelial cell activating agent, LPS (HE/L). Fully glycosylated XL35 migrates as a broad band of 42–45 kDA on reducing SDS-gels, while N-glycanase-treated XL35 migrates as a sharp band at 35 kDa (Lee et al., 1997). Thus, these data indicate that human aortic endothelial cells express a protein similar to the apparent molecular weight of glycosylated XL35 and to the predicted molecular weight of N-glycosylated HL-1. Similar results were observed using a polyclonal antisera prepared against a synthetic peptide whose sequence is found in HL-1, showing that the XL35 protein is being recognized by the antisera (data not shown). Moreover, expression of this protein was apparently unaltered in cultured cells by the addition of an endothelial cell activating agent, suggesting that the tissue specific expression pattern we observed by both Northern analysis and immunohistochemistry did not likely relate to the activation state of the endothelia in those particular tissues. No immunoreactivity was seen in primary tissue cultures of epithelial cells from thymal stroma (TS), a tissue in which HL-1 is expressed, indicating that expression of HL-1 is restricted to endothelial cells. The identity of the band observed at 85 kDa (Figure 5) is unknown at this time, although purified XL35 can show higher molecular weight species when samples are not completely reduced prior to or during electrophoresis.

To confirm this restricted expression of HL-1, we performed RT-PCR analysis of mRNA from cultured cells arising from several tissue types. As shown in Figure 6, a specific band was amplified only from human aortic endothelial cells. No band was amplified from Hep G2 cells (hepatic epithelial cell carcinoma), T cells, thymic epithelial cells, or bone marrow stromal cells. As a control, NIH 3T3 cells transfected with cDNA encoding HL-1 were also analyzed. The amplified band was detected only in the transfected 3T3 cells (3T3tx), but not in the mock transfected cells (3T3), again demonstrating the specificity of the amplified band.

In summary, results from several experimental approaches document the discovery of human homologs of the Xenopus oocyte cortical granule lectin, termed HL-1 and HL-2. The deduced amino acid sequences of these proteins are over 60% identical and 74% similar to that of XL35, and show long stretches of identical sequences, HL-1 is localized exclusively in vascular endothelial cells present in a unique set of tissues, while HL-2 appears to be expressed only in small intestine.

Discussion

The Xenopus oocyte cortical granule lectin XL35 is a soluble protein that makes up about 70% of the protein found in the cortical granules (Nishihara et al., 1986). The lectin has an oligomeric structure with an apparent molecular weight of 500 kDa under non-reducing conditions (Roberson and Barondes, 1982; Chamow and Hedrick, 1986). Reducing SDS–PAGE reveals a monomer of about 45 kDa with size heterogeneity due to N-linked oligosaccharides (Lee et al., 1997). At fertilization, the contents of the granules are secreted from the oocyte, and the multimeric lectin binds to its oligosaccharide ligands found in the surrounding egg jelly. These ligands are expressed on ∼500 kDa mucin-like glycoproteins cross-linked by disulfide bonds, each containing hundreds of O-linked saccharides (Quill and Hedrick, 1996). The result of this binding reaction is an aggregation of lectins and large molecular weight glycoprotein ligands that participates in the formation of the fertilization membrane and block to polyspermy. XL35 mRNA and protein are also found in the zygote and the message levels show a large increase during gastrulation, implying that the lectin is expressed and functions during morphogenesis. Relatively low levels of the message are found in the tadpole stage (Lee et al., 1997).

It was quite interesting, therefore, to find homologs of XL35 in adult human tissues. The amino acid sequences of the lectins predict signal sequences for secretion, and XL35 is known to be secreted at fertilization. If the putative signal sequences, which are very divergent, are excluded from consideration, the identity between the HL-1 and XL35 amino acid sequences is 63% and the similarity is 75%. Moreover, eight conserved cysteine residues are observed between the two sequences. HL-2 shows slightly less identity and similarity to XL35 than does HL-1. Although we have no direct evidence as yet, these structural considerations suggest HL-1 and HL-2 may indeed function as lectins. The oligosaccharide binding specificity of XL35 is known in general terms; however, the location of the amino acids that constitute its binding site is as yet unknown. Western blots of partly purified rat heart homogenate using polyclonal antisera against XL35 identify several protein bands with molecular weights roughly similar to those of HL-1 in human heart aortic endothelial cells (Figure 5). To determine if the homolog in rat heart has a ligand binding specificity similar to that of XL35, the rat heart extract was applied in the presence of Ca²⁺ to an affinity columnof immobilized melibiose, an α-linked disaccharide to which XL35 binds (Roberson and Barondes, 1982). Bound proteins were eluted with EDTA, and the eluted proteins were analyzed by Western blotting using anti-XL35 antisera (Lee, 1997). No lectin-related proteins bound to the melibiose column, showing that the rat heart homolog most likely does not bind the same disaccharide as does XL35. Since only very minor differences in the amino acid sequence of lectins can yield proteins that differ in their binding specificity, no prediction can be made whether HL-1 and HL-2 bind the same ligand (Drickamer, 1992).

No C-type lectin motif is found in XL35 (Drickamer, 1993), although it has calcium-dependent oligosaccharide ligand binding activity. Likewise, HL-1 and -2 sequences do not contain this motif. A small region of XL35 and the human homologs, however, does contain a fibrinogen-like motif, Figure 7. Members of the Ficolin/Opsonin/p35 lectin family contain significant homology to fibrinogen, in addition to collagen-like domains (Kilpatrick et al., 1997; Ohashi and Erickson, 1997). Proteins in this family are found in serum, some have been shown to have lectin-like activity, and they have been postulated to function generally in pathogen opsinization (Matsushita et al., 1996). It has been hypothesized that the carbohydrate-binding domain present in this family of lectins is contained at least partially in the small domains they share with fibrinogen. As depicted in Figure 7, HL-1, HL-2, and XL35 all contain a portion of a fibrinogen-like motif found in the Ficolin/Opsonin/p35 family, suggesting that this region may encode their carbohydrate binding domains. Both ficolin and the Hakata antigen, a circulating serum lectin that is a member of the Ficolin/Opsonin/ p35 family (Sugimoto et al., 1998), however, also contain this small fibrinogen-like motif shared with HL-1, HL-2, and XL35. Neither of these proteins, however, require calcium for their carbohydrate binding activity. It is unlikely, therefore, that this domain found in XL35 is predominantly responsible for its ligand binding activity, but further experiments are obviously necessary to identify its oligosaccharide binding site and the putative binding sites in HL-1 and -2.

A brief report described a mouse cDNA sequence isolated by differential hybridization techniques that hybridizes with specialized cells in the mouse small intestine (Komiya et al., 1998), although the degree of cell specificity of expression was not clear from the data presented. The deduced amino acid sequence from this cDNA shows a 60% homology with the XL35 sequence, and because of this homology, the sequence was named intelectin. Comparing human HL-1 and HL-2 deduced amino acid sequences to that of the murine intelectin shows about a 50% homology of the murine molecule to either human sequence.

The presence of HL-1 in vascular endothelial cells, coupled with the detection of the HL-2 transcript in small intestine and a transcript similar to the human lectins in mouse small intestine, demonstrates defined, unique expression patterns for these homologs. Members of this family with sequences similar to the Xenopus cortical granule lectin clearly must have functions distinct from that of the cortical granule lectin. Experiments to define these functions are in progress.

Materials and methods

Materials

Nitrocellulose filters, DNA labeling kits, [γ-³²P] dCTP, [³⁵S]-methionine, and Amplify reagent were purchased from Amersham. Ready polyacrylamide gels and Zeta-Probe GT nylon membranes for Southern analysis were purchased from Bio-Rad (Hercules, CA). N-Glycanase, protease inhibitor cocktail tablets, and PCR reaction master mix were from Boehringer Mannheim (Indianapolis, IN). BSA, IPTG, melibiose, and immobilized melibiose were from Sigma (St. Louis, MO). Sephaglas DNA purification reagent was from Pharmacia (Uppsala, Sweden). Ni⁺⁺-NTA columns and pQE vectors were from Qiagen (Chatsworth, CA) and were used according to the manufacturer’s instructions. The pCR II cloning vector was from Invitrogen (San Diego, CA). Restriction enzymes, Thermus aquaticus DNA polymerase, agarose, and other chemicals were purchased from major chemical suppliers. Oligonucleotides were synthesized by the University of Georgia Molecular Genetics Instrumentation Facility. The BCA protein assay kit was from Pierce (Rockford, IL). Fetal calf serum was obtained from Upstate Biotechnology Inc (Lake Placid, NY). The human small intestine 5′-stretch cDNA library, λgt10 cDNA insert screening amplimer set, and human multiple tissue Northern blots were obtained from Clontech (Palo Alto, CA). Human liver, thymus, colon, tonsil, and heart were obtained from the UCLA Department of Pathology Human Tissue Research Center. Primary cultures of human aortic endothelial cells (HAEC) were the gift of Dr. Judith Berliner (UCLA, Los Angeles, CA). NIH 3T3 and Hep G2 cells were obtained from ATCC (Rockville, MD). Human peripheral blood T cells and thymic epithelial cells were obtained as described previously (Perillo et al., 1995). Human bone marrow stroma was the kind gift of Dr. Bruce Torbett (Scripps Clinic and Research Foundation, La Jolla, CA).

Detection of partial cDNA sequences of human homologs

XL-35 cDNA and peptide sequences were used to search protein and DNA sequence databases, and a single entry was identified from non-Xenopus sources. The sole sequence match was a fragmentary 251 bp cDNA sequence obtained from a human heart cDNA library (GenBank, Accession number Z36760) as an expressed sequence tag (EST) (Figure 1). Using the human EST sequence, we designed oligonucleotide primers (sense primer: 5′-CAGACCTTCTGTGACATGACCTCT-3′ and antisense primer: 5′-AAGATGCCCAGGTCCTTGGCCTGG-3′). The human placenta, spleen, and liver cDNAs were subjected directly to PCR reactions in a 25 µl reaction volume containing 1 µM of each primers, 10 ng of each cDNA, 0.2 mM of each dNTP, 10 mM Tris–HCl, 50 mM KCl, 1.5 mM MgCl₂, and 1.5 units of Taq polymerase. The thermal cycling conditions were 92°C, 1 min; 55°C, 1 min; and 72°C, 1 min, for a total of 35 cycles. The amplification products were separated on a 0.7% agarose gel and purified using Sephaglas DNA purification kit (Pharmacia). Each amplimer was subcloned into the pCR II vector using the TA cloning kit (Invitrogen) and fully sequenced. The deduced peptide sequences from the PCR products, the human heart EST sequences, and the XL-35 cDNA clone were aligned to compare their sequence identities. The cDNA amplimers from the human tissues were essentially identical to the human heart EST sequence.

cDNA library screening

A human small intestine 5′-stretch λgt10 cDNA library was purchased from Clontech and was screened by plaque hybridization using the 251 bp amplification product of the human liver cDNA as a radiolabeled probe. The cDNA amplimer was isolated on a 1% agarose gel, purified, and labeled using [γ-³²P]-dCTP and the Mega-Prime labeling kit (Amersham). Plaque lifts were screened by standard formamide/SSC hybridization conditions using Hybond-N filters (Amersham) (Sambrook et al., 1989). Positive phage clones identified by plaque hybridization were purified through four rounds of plaque purification. Lambda DNA from positive clones was isolated from 100 ml cultures by polyethylene glycol precipitation followed by chromatography over a Qiagen Tip-100 column as described by the manufacturer (Qiagen). The full-length inserts from the clones were isolated by PCR amplification using primers specific for the flanking vector sequences. A total of 13 clones were identified with insert sizes varying from 0.7 to 2.4 kb. Seven cDNA inserts >1.2 kb were subcloned into the pCR II vector and fully sequenced. Comparison of the sequences of the clones indicated that six of the clones were identical (named HL-2) and one clone (HL-1) had a different sequence (Figure 1).

DNA sequence analysis

All cDNA clones and genomic clones were sequenced by using Taq polymerase in the dideoxy dye-terminator reaction using T7 or SP6 polymerase primers, or synthetic primers. The sequencing reactions were analyzed on an Applied Biosystem 373A DNA sequencer (Molecular Genetics Instrumentation Facility, University of Georgia). DNA sequence data were assembled into a contiguous sequence database by the method of Staden (1987)). A sequence similarity search between protein or DNA sequences was performed using the Bestfit, Pileup, Fasta, and Tblast programs of the University of Wisconsin Genetics Computer Group (GCG software, version 8.0).

Isolation of genomic DNA clones

An arrayed human genomic library in the P1 vector (Genome Systems, Inc., St Louis, MO) was screened by a PCR approach (Pierce et al., 1992; Shepherd et al., 1994) using primer pairs specific for the respective genes. Pools of clones were screened by PCR amplification using primers for HL-1 (sense primer: 5′-ATACTTTCCAGAGGCCAGTCCCCAG-3′, antisense primer: 5′-AGGTCTGGGTTCCCTCCCACAAAAC-3′) or HL-2 (sense primer: 5′-GTTCTTCCCACAGGGCAAACCCCGT-3′, antisense primer: 5′-TCTGCCCTGACACCGGAGAGCTCTG3′). The PCR reactions were performed in a 25 µl reaction volume containing 1 µM of each primers, 50 ng of human genomic DNA, 0.2 mM each dNTP, 10 mM Tris–HCl, 50 mM KCl, 1.5 mM MgCl₂, and 1.5 units of Taq polymerase. The thermal cycling conditions were: 94°C, 1 min; 65°C, 1 min; and 72°C, 1 min, for a total of 35 cycles. An amplimer of ∼160 bp was obtained from PCR reactions from the respective template (dotted-line sequences in (Figure 1). The sequence of each amplimer was identical to that of each cDNA clone, which implied that the region of amplification in the lectin genes did not contain introns. The same primers and PCR reaction conditions were used for the screening of genomic DNA clones. Two genomic clones were isolated for both HL-1 and HL-2 and each was partially sequenced using synthetic primers designed based in the sequence of the cDNA clones. Sequence data and results of PCR amplification indicated that each clone contained either HL-1 or HL-2, but not both genes.

Northern blot analysis

The cDNA-specific PCR products for HL-1 or HL-2 obtained from PCR of human liver cDNA were labeled using [γ-³²P] dCTP and the Mega-Prime labeling kit (Amersham) according to the manufacturer’s instructions. The 161 bp amplimers probes for HL-1 and HL-2 obtained from PCR were only 63.9% identical in sequence (58/161 mismatches). The radiolabeled cDNA fragments were used to probe Northern blots of various human tissue poly(A+) RNAs (human multiple tissue Northern blot, Clonetech). The blots were probed using a prehybridization buffer consisting of 0.5 M Na phosphate, pH 7.0, 1 mM EDTA, 7% SDS, 10 mg/ml of fatty acid free BSA, and 100 µg/ml denatured herring sperm DNA. The hybridization was using the identical buffer, except for the addition of denatured radiolabeled probe. Prehybridization was performed for 2 h at 65°C, and hybridization was performed overnight at 65°C. The blots were washed three times at 65°C for 15 min in 50 ml of 40 mM Na phosphate, pH 7.0, 1% SDS, 1 mM EDTA, and 0.5% fraction V BSA and an additional three times at 65°C for 15 min in the same buffer excluding BSA. The blots were then washed in 50 ml of 1 M phosphate buffer, pH 7.0, at room temperature for 10 min and data collected using a Molecular Dynamics phosphoimager or autoradiography.

Immunoblot analysis

Polyclonal rabbit antisera against affinity-purified XL35 was prepared similarly to that described by Barondes’s laboratory (Roberson and Barondes, 1983). Antisera was specific for XL35 on Western blots of oocytes, and reacted with N-glycanase-treated XL35. Human thymic stroma, obtained after removal of thymocytes, was extracted by homogenization in lysis buffer (10 mM Tris, pH 7.3, 130 mM NaCl, 5 mM CaCl₂, 1 mM PMSF, 0.5% NP-40). Cultured human aortic endothelial cells (HAEC), either stimulated with LPS (Baum et al., 1995b) or mock stimulated controls, were lysed by brief vortexing in lysis buffer. Nuclei were removed by centrifugation for 2 min at top speed in a microcentrifuge. Two hundred micrograms of protein were loaded per lane, and electrophoresed under reducing conditions. Immunoblot analysis was performed as described previously (Baum et al., 1995b). The blot was probed with rabbit polyclonal antiserum against native XL35 diluted 1:1000 in 100 mM Tris, 150 mM NaCl, 0.1% Tween (TBST) with 1% nonfat dry milk. After repeated washes with TBST, the blot was incubated with horseradish peroxidase conjugated goat anti-rabbit IgG (Bio-Rad), and visualized by chemiluminescence (ECL, Amersham).

RT-PCR analysis

Total RNA was purified from the indicated cell types using Trizol reagent (Gibco BRL), according to the manufacturer’s directions. Two micrograms of total RNA was reverse transcribed (Ready to go T primed first strand kit, Pharmacia). Two microliters of the first strand reaction was used as template for PCR amplification. PCR was performed using 200 µM dNTP, 1 µM each oligonucleotide primer, and Taq polymerase and buffer (Stratagene). The sequence of the 5′ primer was 5′CAGACTACCCAGAGGGGGACGGCAACTGG3′. The sequence of the 3′ primer was 5′CTCCACCAATGCAGTGGTGCTCAGTGTTAC3′. The thermal cycling conditions were: 94°C × 45 s, followed by 72°C × 2 min for a total of 40 cycles followed by a final extension for 10 min at 72°C.

Immunohistochemical analysis

Experiments were performed on 6 mm sections of the indicated tissues as previously described (Baum et al., 1995a), using rabbit polyclonal antiserum against native XL35 diluted 1:1000, and the sections were counter-stained with hemotoxylin. Control sections were incubated with a similar dilution of preimmune serum.

Acknowledgments

We wish to acknowledge support from the American Heart Association and NIHGMS. J. S. was a trainee of the UCLA STAR Program.

Abbreviations

XL35, Xenopus laevis cortical granule lectin; LPS, lipopolysaccharide; HL-1 and HL-2, human lectins 1 and 2; HE, human aortic endothelial cells; HE/L, human aortic endothelial cells treated with lipopolysaccharide; TE, thymic epithelia; BMS, human bone marrow stroma; TE, human thymic epithelium; HAEC, human aortic endothelial cells.

References