Introduction

All multicellular organisms have some form of innate immunity that functions in the identification of, and protection against, invading pathogens and parasites. The complement system, which is composed of about 35 serum and cell surface proteins (Volanakis 1998), is an important component of the more ancient innate immune system. It has been identified throughout the lineage of deuterostome animals (Smith et al. 1999; Nonaka 2001; Azumi et al. 2003) and is being discovered in the rest of the animal kingdom as well, such as in a gorgonian (accession no. AAN86548) and a squid (M. McFall-Ngai, personal communication). There are three activation pathways—classical, alternative, and lectin—that are activated by different types of molecules and lead to the activation of the terminal or lytic pathway. The activation pathways culminate in the formation of two distinct C3-convertase enzymes: the alternative pathway convertase (C3bBb) and the classical pathway convertase (C4bC2b) (reviewed by Nonaka et al. 1998; Xu et al. 2001). The formation of C3-convertases leads to the formation of C5-convertases (Pangburn and Rawal 2002) that activate the lytic pathway. C3-convertases function to cleave and activate additional C3, creating a positive feedback loop within the alternative pathway that accelerates the activation of the entire complement system (Liszewski et al. 1996), resulting in quick and efficient opsonization and lysis of foreign cells (Lambris 1988; Becherer et al. 1989).

Because complement proteins C3 and C4 can form covalent thioester bonds with amines and hydroxyls on any molecule, they have the ability to bind any surface, including self, which would lead to inappropriate cell lysis and inflammatory reactions. Covalent bond formation between C3 or C4 and non-self is an important mechanism for identifying pathogens. Therefore, a complement regulatory system is necessary to both protect self-cells against autologous complement attack and, consequently, to direct the attack towards foreign pathogens. Tight regulation of active C3-convertases is also essential to prevent unnecessary depletion of complement proteins from uncontrolled activation, as well as to inhibit complement activities after a pathogen has been cleared (Liszewski et al. 1996). One important mechanism of regulation is based on dissociating the C3-convertase complexes and by degrading C3. Some of the proteins involved are membrane cofactor protein (MCP), complement receptors 1 and 2 (CR1 and CR2), decay accelerating factor (DAF), C4 binding protein (C4BP), factor H, and factor I (Medof et al. 1987; Krych et al. 1991; Liszewski et al. 1996; Arlaud et al. 1998; Kirkitadze and Barlow 2001; Barilla-LaBarca et al. 2002).

Sea urchin immune response

The sea urchin defends itself against pathogens with its innate immune system. Coelomocytes are the immune effector cells (Boolotian and Geise 1958; Johnson 1969; Gross et al. 2000) that carry out many protective functions including encapsulation, phagocytosis, chemotaxis, and expression of putative cytotoxic agents in response to invasion of foreign cells (Smith et al. 1992; Smith and Davidson 1994). There are four morphologically distinct classes of coelomocytes (Johnson 1969; Edds 1993) of which the phagocytes appear to be the major immune effector cells. Phagocytes express Sp064, which encodes SpC3, a homologue of complement component C3 (Al-Sharif et al. 1998; Gross et al. 2000), and Sp152, which encodes SpBf, a homologue of complement component factor B (Bf) (Smith et al. 1998; unpublished data). True thioester activity, similar to that characterized for C3 in higher vertebrates, has been demonstrated chemically for SpC3 through autolytic fragmentation and thioester inactivation with methylamine (Smith 2002). In biological assays, SpC3 functions as an opsonin and augments phagocytosis of yeast by coelomocytes (Smith 2001; Clow et al. 2004). The complement system in the sea urchin has been proposed to function like an alternative pathway with the formation of a C3-convertase complex from SpC3 and SpBf, resulting in a feedback loop to augment the rate of opsonization (Smith et al. 1999; Smith 2001). This suggests that the thioester binding activity and the convertase functions would require a mechanism for protection against autologous attack by controlling or limiting the convertase activity that would avoid depletion of the components. Initial evidence for a complement regulatory system in the sea urchin was obtained from the sequence of SpC3, which has two conserved cleavage sites for factor I-like activity that are located in conserved positions (Al-Sharif et al. 1998). This predicts that additional complement components function in the sea urchin.

In the present study, we show the analysis of two cDNAs, Sp5 (accession no. AY494840) and Sp5013 (accession no. AY494841), which encode proteins SpCRL (S . p urpuratus complement related protein, long form) and SpCRS (short form), respectively. Deduced amino acid sequences from both proteins show multiple short consensus repeats (SCRs) (18 in SpCRL and four in SpCRS), two Ser/Thr/Pro (S/T/P)-rich regions, and a factor I-membrane attack complex (FIMAC) domain. In addition, SpCRL has a fucolectin domain and a Cys-rich region. Both genes are expressed in coelomocytes, gut, gonad, pharynx, esophagus, and axial organ, and the level of message accumulation in coelomocytes before and after immune challenge suggests that expression is constitutive. Northern blots for both messages show multiple bands of varying intensity, suggesting the presence of a small family of genes with similar sequences in addition to possible alternative splicing. Genome blots for both genes also show multiple bands of varying intensity consistent with gene structure of multiple exons and perhaps with cross-hybridization with other members of the gene family. Structural and phylogenetic analyses of the deduced amino acid sequences for these two proteins indicate that they share domains with a number of complement proteins from higher vertebrates including factor H, factor I, C6 and C7. These results imply functions within the complement system, possibly as complement regulatory proteins and/or within a primitive terminal pathway in the sea urchin.

Materials and methods

Animals

Purple sea urchins, Strongylocentrotus purpuratus, were obtained and housed as previously described (Gross et al. 2000; Shah et al. 2003).

RNA isolation

Total RNA from gut, gonad, esophagus, pharynx, axial organ, and coelomocytes was isolated using the RNAeasy Midi kit (Qiagen, Valencia, Calif.). Animals were sacrificed by removing Aristotle’s lantern (mouth parts), and the coelomic fluid was poured from the body cavity through sterile cheese cloth and mixed in a 1:1 ratio with ice-cold, Ca2+- and Mg2+-free sea water containing 70 mM EDTA and 50 mm imidazole (pH 7.4) (CMFSW-EI) according to Gross et al. (1999). Coelomocytes were pelleted by centrifugation at 6,500 g for 5 min at 4°C. To harvest the internal organs, the test was cut open and tissues were removed from the coelomic cavity and placed in RNAlater (Ambion, Austin, Tex.). Pelleted coelomocytes and approximately 0.24 g of each solid tissue were lysed in guanidine isothiocyanate lysis buffer (Qiagen). The lysate was bound to a silica gel-based column membrane, digested with 27 U DNaseI (Qiagen), washed, eluted in RNase-free water, and quantified with a DU 640 spectrophotometer (Beckman Instruments). RNAsin (Promega, Madison, Wis.) was added to each sample (0.8 U/μl) prior to storage at −70°C.

Reverse transcriptase polymerase chain reaction

Reverse transcriptase (RT) reactions were performed with 1–3 μg total RNA and 5 μm random hexamer primer with Superscript II reverse transcriptase (Invitrogen, Carlsbad, Calif.) according to the manufacturer’s instructions. To identify samples with contaminating genomic DNA, approximately 150 ng of each RNA sample was used directly as a template for PCR. All primer sets employed in RT-PCR reactions (see below) were used to identify amplification from genomic DNA. Samples showing bands larger than expected for cDNA amplification were assumed to contain genomic DNA and were discarded.

The cDNA generated from the RT reactions (1 μl) was mixed with 0.5 U Taq DNA polymerase (Invitrogen), 3 mm of each deoxynucleotide, 1 μm each primer, 1× company-supplied buffer (Invitrogen), and 1.5 mm MgCl2 in a volume of 20 μl. Primers included Sp5: 5for: 5′ CCC TGG ACA GTA TGT GTT GCA TGG TAG, 5rev: 5′ TAT CCC TGG TTG CAT CCT ATG AGC ACA); Sp5013: 5013for: 5′ TCG ATG GGT GTT CCG AGT GGG TCT, 5013rev: 5′ TCT ACA TCT AGC AAC TAG CAG GGT GCC; Sp056: 056for: GCA CAG CCA GCA ACC AGC ACT ACA AT, 056rev: ACG CCG ATG GGT TCT ACA GTG AAG GT; and SpL8: L8for: CAG CGT AAG GGA GCG GGA AGC GTC TT, L8rev: GTT TGC CGC AGA AGA TGA ACT GTC CCG TGT A. Reactions were heated to 95°C for 5 min; followed by 25 cycles of 94°C for 30 s, 52°C for 30 s, and 72°C for 10 min; followed by 72°C for 2 min and 4°C hold. The amplified fragments were electrophoresed on a 0.8% agarose/0.4% NuSieve gel (BioWhittaker Molecular Application, Rockland, Md.) containing 0.5 μg/ml ethidium bromide in TAE buffer (40 mm Tris base, 20 mm glacial acetic acid, 1 mm EDTA, pH 8.3). Gels were imaged with a DC120 digital camera and 1D digital software (Eastman Kodak, Rochester, N.Y.) followed by image optimization in Photoshop (Adobe Systems, Seattle, Wash.).

Cloning PCR fragments

Fragments amplified by PCR were cloned into the pCRII-TOPO vector according to the manufacturer’s instructions (Invitrogen) and transformed into TOP10 bacteria (Invitrogen).

Arrayed cDNA libraries

An arrayed cDNA library was constructed in the pBK-CMV vector (Stratagene, La Jolla, Calif.) using activated coelomocytes from five sea urchins 24 h post-injection of 1 ml of heat-killed bacteria (for details of bacterial isolation and library construction, see Pancer et al. 1999; Cameron et al. 2000; Rast et al. 2000). A similar arrayed cDNA library was constructed using non-induced coelomocytes in the pSPORT vector (Life Technologies, Rockville, Md.) using coelomocytes from non-activated sea urchins (for details, see Smith et al. 1996; Al-Sharif et al. 1998; Cameron et al. 2000). Each library was arrayed into 240 plates of 384 wells each for a total of 92,160 clones per library (Al-Sharif et al. 1998; Cameron et al. 2000; Rast et al. 2000). The insert from every clone was amplified by PCR and spotted in duplicate onto five 22×22-cm Hybond-N+ (Amersham) filters for each library, which were used for screening.

cDNA library screens

Filters were prehybridized with 10 ml of hybridization solution [0.1% BSA (w/v), 1 mm ethylenediaminetetraacetic acid (EDTA), 0.25 m phosphate buffer (pH 7.4), 7% sodium dodecylsulfate (SDS) (w/v), and 50% formamide (v/v)] at 42°C for 2 h by rotation in a hybridization oven (Robbins Scientific, Sunnyvale, Calif.). The hybridization solution was replaced, a riboprobe (see below) was added to the filters and then rotated overnight at 42°C. Filters were washed twice in 4× SSC (20× SSC is 0.3 m sodium citrate, 3 m NaCl, pH 7.0) with 1% SDS, twice in 2× SSC with 1% SDS, and twice in 1× SSC with 1% SDS at 65°C for 30 min. Each wet filter was sealed into a plastic bag and exposed to X-OMAT AR film (Eastman Kodak) without an intensifying screen. After exposure, filters were stripped by washing once in 0.4 m NaOH at 45°C for 3 min, twice in stripping buffer [0.1× SSC, 0.1% SDS, 0.2 m Tris (pH 7.5)] at 65°C for 30 min, and twice in stripping buffer with 0.5 m EDTA at room temperature for 10 min. After stripping, filters were dried and stored at −20°C.

Riboprobe synthesis

Clones that served as templates for riboprobe synthesis (see Fig. 1) were linearized with either BamH1 at the 5′ end or XbaI at the 3′ end (Promega), and 170 ng to 515 ng was labeled using 50 μCi to 70 μCi 32P rUTP (ICN, Irvine, Calif.); 0.5 mm each of rATP, rCTP, and rGTP; 15 U of T7 RNA polymerase (Promega); 1× company-supplied transcription buffer (Promega); 20 mm dithiothreitol (DTT); and 24 U of RNAsin (Promega) in a total volume of 20 μl. Samples were incubated at 37° for 1 h, followed by the addition of 46 μg yeast tRNA, 24 U of RNAsin, and digested with 1 U RQ1 DNAse (Promega) at 37°C for 15 min. Unincorporated nucleotides were removed by passing the probe through a G-50 fine Sephadex column (Amersham Pharmacia Biotech AB, Piscataway, N.J.) spun at 2,000 g for 1 min. Incorporation of 32P was analyzed with an LS6500 liquid scintillation counter (Beckman Instruments).

Clone blots

Individual colonies were dispersed in sterile water (50 μl) and 5 μl was used as the template in PCR reactions using a 9600 thermal cycler (Perkin Elmer, Wellesley, Mass). Reactions included 0.5 U Taq DNA polymerase (Invitrogen), 0.25 mm each deoxynucleotide, 1 μm T3 and T7 primers (Qiagen), 1× company-supplied buffer (Invitrogen), and 1.5 mm MgCl2 in a volume of 20 μl. Reactions were heated to 95°C for 5 min, followed by 25 cycles of 94°C for 30 s, 52°C for 30 s, and 72°C for 10 min, with a final step of 72°C for 10 min. Amplified products were electrophoresed on an agarose gel (as above), and images were captured using UV illumination with a digital camera (Eastman Kodak). Gels were double blotted by capillary action onto two Genescreen Plus membranes (NEN Life Science Products, Boston, Mass.) by standard procedure (Sambrook et al. 1989).

Non-radioactive DNA probes

Templates for Sp5 and Sp5013 were generated using clones that were amplified by PCR using T3 and T7 primers for clones matching to Sp5013 and T7 and Sp6 primers for clones matching to Sp5. Amplified products were purified using the Geneclean Turbo kit for PCR (Bio101, Carlsbad, Calif.) according to the manufacturer’s instructions, and DNA fragments (100 ng) were labeled using alkaline phosphatase non-radioactive GeneImages Alkphos Direct kit (Amersham) according to the manufacturer’s instructions. The blots were hybridized with the probe and washed according to the manufacturer’s instructions.

Clone maps

Analysis by PCR

Plasmid insert sizes were analyzed from single colonies by PCR as described above, using a combination of a gene-specific primer with either T3 or T7 primers. A touch-down thermal cycling program was used because of the significant difference in annealing temperatures between the two primers (gene specific = 62–64°C; T3 and T7 = 55°C). The PCR program (described above) with a total of 30 cycles was changed so that the initial annealing temperature of 62°C was decreased by 1°C each cycle for eight cycles to 55°C, where it was maintained for 22 cycles, followed by 72°C for 10 min and 4°C hold. Amplified products were separated on an agarose gel and imaged as above.

Restriction-enzyme digests

Digests were done with EcoR1, XhoI, and BamHI, and fragments were separated on agarose gels.

Northern blots

Poly(A)+ RNA was isolated from 100 μg total RNA from coelomocytes using the Oligotex mRNA mini kit (Qiagen) according to the manufacturer’s instructions. Poly(A)+ RNA was electrophoresed through a 1% agarose gel containing 2.2 m formaldehyde in 1× MOPS buffer [10× MOPS is 20 mm 3-(N-morpholino) propanesulfonic acid, 5 mm NaOAc, 1 mm EDTA, pH 7] and blotted by capillary action onto Genescreen Plus (NEN Life Science Products) with 10× SSC. Filters were hybridized and washed as described above for library screens, and exposed to X-OMAT AR film (Eastman Kodak). Transcript sizes were estimated from RNA standards (Ambion) as well as coelomocyte rRNA. Filters were stripped using the protocol described above for library filters.

Sequencing

Plasmids were isolated using the Wizard Plus SV Miniprep kit (Promega) according to the manufacturer’s instructions, quantified with the DU 640 spectrophotometer (Beckman Instruments), and sequenced with BigDye terminator cycle sequencing ready reaction kits (Applied Biosystems, Foster City, Calif.) using either T3, T7, or internal primers. Cycle sequencing reactions were done in a 9600 thermal cycler (Perkin Elmer) with 25 cycles of 96°C for 30 s, 50°C for 30 s, and 60°C for 4 min. Unincorporated nucleotides were removed by passing the samples through a G-50 fine Sephadex (Pharmacia) spin column and dried in a centrivap concentrator (Labconco, Kansas City, Mo.). Dried samples were either dissolved in 15 μl template suppression buffer and loaded onto an ABI prism 310 Capillary Sequencer or 3 μl sequencing loading buffer [5:1 ratio of deionized formamide to 25 mm EDTA (pH 8.0), plus bromophenol blue] and loaded onto a 377 Automated Sequencer (Applied Biosystems).

Results and discussion

Sp5 and Sp5013 clones

Two clones, Sp5 and Sp5013, were identified during a library screen using a probe designed to find sequences specific for scavenger receptors with Cys-rich (SRCR) domains (Pancer et al. 1999; Pancer 2000). However, the initial analysis of these two clones only revealed regions that encoded SCRs, and therefore they were kindly provided to us by Z. Pancer. Further analysis was pursued because many complement proteins have SCRs, and the sequences obtained did not match to Sp152, which encodes SpBf with five SCRs (Smith et al. 1998).

Messages and deduced proteins

Two arrayed cDNA libraries were screened using riboprobes made from clones generated by RT-PCR with primers specific for Sp5 and Sp5013. Sixty Sp5 clones and 30 Sp5013 clones were characterized by mixed-primer PCR analysis and restriction digests to identify overlapping regions and to optimize sequencing strategy (Fig. 1). Both standard and internal primers were used to sequence 17 Sp5 clones and 6 Sp5013 clones (Tables 1, 2). Some of the sequence near the 3′ end of Sp5 was provided by Z. Pancer. The sequenced region of the Sp5 transcript was 5.96 kb and was composed of 5′UTR [36 nucleotides (nt)], 3′UTR (432 nt), and 5,472 nt of open reading frame (ORF) (Fig. 2). The sequenced region of the Sp5013 transcript was significantly shorter, 2.751 kb, and composed of 3′UTR (42 nt), and 5′UTR (924 nt), with an ORF of 1,785 nt (Fig. 3). Although the 5′UTRs for both of these cDNAs were quite short, the entire 5′UTR for Sp5013 may have been obtained because six clones had identical sequences at the 5′ end. However, it is possible that significant secondary structure in this region of the message may have blocked complete reverse transcriptase activity.

Fig. 1a, b
figure 1

Maps and sequencing passes for Sp5013 and Sp5. a Sp5013. An overlapping series of 12 clones was analyzed by touch-down PCR reactions (see Materials and methods) using the 5013for primer and the T7 primer in addition to restriction digests. Three of the six clones that were used to generate the sequence for Sp5013 are shown. The sequencing passes are indicated as arrows. The number associated with each arrow corresponds with a primer listed in Tables 1 and 2. Clone 199N18 served as the template to generate the riboprobe for Northern-blot analysis (Fig. 9). b Sp5. An overlapping series of 39 Sp5 clones was analyzed using 5for (5′ CCC TGG ACA GTA TGT GTT GCA TGG TAG) with T7 and reverse RceR1 (5′ AAC ACA GCT TGG TTG CAC TCC TGT CC) with T3 by touch-down PCR reactions (see Materials and methods). Eleven of 17 clones that were used to generate 5.65 kb of sequence are shown. The length and overlap of the sequencing passes are indicated as arrows and the number associated with each arrow corresponds with primers listed in Tables 1 and 2. The area of the figure shown as a tiled pattern denotes the sequence provided by Z. Pancer. Clone 231K2 served as the template to generate the riboprobe for Northern-blot analysis (Fig. 9)

Full size image
Table 1 Sequencing primers for Sp5013 listed in Fig. 1a. All annealing temperatures were between and 55.3°C and 57.2°C
Full size table
Table 2 Sequencing primers for Sp5 listed in Fig. 1b. All annealing temperatures were between and 55.3°C and 57.2°C
Full size table
Fig. 2
figure 2figure 2figure 2

cDNA sequence of Sp5 and the deduced protein sequence of SpCRL (S trongylocentrotus p urpuratus complement related protein, long form). Overlapping sequencing passes were used to generate the cDNA sequence and the deduced amino acid sequence was translated using DNASIS (Hitachi, Cascade, Colo.). The signal sequence was predicted using the Expasy SignalP V1.1 program. The start codon and stop codons, including those in the 5′UTR and 3′UTR, are indicated in boldface. The Cys in the Cys-rich region and the Ser, Thr, Pro in the Ser/Thr/Pro (S/T/P)-rich regions are also indicated in boldface. All domains are labeled and/or numbered, and the limits of each SCR is indicated with an arrow. The N- and O-linked glycosylation sites are noted with CHO(N) and CHO(O), respectively

Full size image
Fig. 3
figure 3

cDNA sequence of Sp5013 and the deduced protein sequence of SpCRS (S . p urpuratus complement related protein, short form). The nucleotide and amino acid sequences were generated as described for Sp5 (see legend for Fig. 2). The leader, domains, and conserved sites are labeled as in Fig. 2. An almost-perfect Kozak sequence (GCC(A/G)CCATGG) surrounds the start codon and is underlined (Kozak 1987). Three poly(A)+ signal sequences (AATAAA) and one non-consensus-type poly(A)+ signal sequence (ATTAAA) plus one AU-rich element (ATTTA) are present in the 3′UTR and are underlined

Full size image

The deduced amino acid sequence encoded by Sp5 and called “SpCRL” consisted of 1,829 amino acids (Fig. 2). The deduced amino acid sequence encoded by Sp5013 and called “SpCRS” consisted of 595 amino acids (Fig. 3). The absence of a Kozak sequence surrounding the start codon for Sp5 and the presence of a very short 5′UTR made a reliable identification of the correct start codon difficult. However, the choice of the probable start codon was based on (1) the presence of one in-frame stop codon 21 nt upstream in the 5′UTR in addition to a second stop that was out-of-frame and (2) an ORF following the ATG that began with a leader region (underlined, Fig. 2) as predicted by the pSignal program (Nielson et al. 1997). The cleavage site for the removal of the leader was predicted to occur after Ser27, which was directly followed by Cys30 of the first SCR. Sp5013 had an imperfect Kozak sequence [GCC (A/G)CC ATG G] (Kozak 1987) (Fig. 3, underlined), two stop codons in the 5′UTR, and a hydrophobic leader putatively cleaved after Ser27, which was followed by Cys30 of the first SCR (Fig. 3). Neither sequence revealed the presence of a transmembrane region (TMPredict database, Hofmann and Stoffel 1993). Both messages had numerous stop codons in all reading frames in the 3′ UTR and both appeared to have polyadenylated stretches at the 3′ end of the sequence (however, see below). Sp5 did not have a conserved polyadenylation signal sequence, but four were identified in the 3′UTR of Sp5013. Two of these were located just 5′ of the poly(A)+ tail (Fig. 3, underlined). One AU-rich element (ATTTA, Asson-Batres et al. 1994) was identified in the 3′UTR of Sp5013 (Fig. 3, underlined).

SpCRS and SpCRL domain structure

Analysis of the deduced amino acid sequences of SpCRL and SpCRS indicated that both were mosaic proteins with a variety of domains, many of which were typically found in complement regulatory proteins and some complement components. Overall, both proteins contained SCRs (18 in SpCRL and four in SpCRS), a FIMAC domain, two S/T/P-rich regions, and numerous N-linked and O-linked glycosylation sites (Figs. 2, 3, 4). In addition, SpCRL contained a fucolectin domain and a Cys-rich region. The S/T/P-rich regions were present in similar locations of both SpCRL and SpCRS (Fig. 4). The composition of Ser and Thr plus Pro in the N-terminal region was 17% in SpCRL and 27% in SpCRS, while the same composition in the C-terminal region was 24% and 38%, respectively. In SpCRS the first S/T/P-region was located between SCRs 2 and 3, the FIMAC domain was located C-terminal of the SCR domains, and the second S/T/P-rich region was located C-terminal to the FIMAC domain. The 18 SCRs in SpCRL were spread throughout the protein and the Cys-rich region, which had eight Cys in a span of 57 amino acids, was located between SCRs 1 and 2 (Fig. 4). A fucolectin domain in SpCRL was positioned between SCRs 10 and 11, with the first S/T/P-rich region between SCRs 16 and 17. The FIMAC domain in SpCRL was located between SCR 18 and the second S/T/P-rich region.

Fig. 4
figure 4

Overview of the domain structures of SpCRL (a) and SpCRS (b). The structure for SpCRL and SpCRS are positioned to illustrate the similarities between the C-terminal end of SpCRL and the structure of the entire SpCRL. Domains are indicated as follows: The leader is shown as a jagged line at the N-terminal end of both proteins. Short consensus repeats (SCRs) are numbered and indicated with a black rectangle, and similar fill patterns indicate SCRs with similar sequences. The Cys-rich region in SpCRL is shown with a dark gray circle, and the fucolectin domain is indicated with a variegated gray rectangle. The S/T/P domains are shown in both proteins as a black diamond and the factor I-membrane attack complex (FIMAC) domain as a variegated gray oval. Inferred functions of some SCRs, based on phylogenetic analysis, are shown above the SCR as a star to indicate putative binding to SpC3, a plus sign to indicate putative binding to polyanionic structures on cell surfaces such as sialic acid or heparin, and a (Zodiac) Cancer symbol to indicate cell adhesion function. Conserved N-linked glycosylation sites are shown with a black circle/black line and O-linked sites are shown with a white circle/black line

Full size image

Glycosylation sites in each protein were identified by searching for the conserved amino acid sequences NXS or NXT, and two prediction programs for O-linked glycosylation sites were used to identify O-linked sites (Hansen et al. 1997, 1998). Both proteins had many conserved sites for both O-linked and N-linked oligosaccharides (Figs. 2, 3). There were 22 O-linked glycosylation sites in SpCRL and 17 in SpCRS, while SpCRL had 17 N-linked glycosylation sites and SpCRS had seven (Fig. 4). In both proteins, many (40% for SpCRL and 64% for SpCRS) of the conserved glycosylation sites were located in the C-terminal S/T/P-rich region (Fig. 4). The rest of the sites in both proteins were located throughout the sequence. In general, both proteins showed similar distributions of the conserved glycosylation sites, and both may be highly glycosylated, particularly at the C-terminal end.

Phylogenetic analysis of domains

An approach for predicting functions of domains and proteins as a whole when the only available data are sequences is to align the amino acids with other sequences from known proteins and to use phylogenetic analysis to identify similarities which can be used to infer function. For the SCR, FIMAC, and fucolectin domains, there were enough sequences available from other proteins that have been characterized previously, some with known function, to make this approach feasible. On the other hand, this approach was not feasible for the S/T/P-rich regions and the Cys-rich region. It is noteworthy that an S/T/P-rich region is present at the C-terminus of MCP, which is involved in protecting self-cells from autologous complement attack; however, the functions mediated by the S/T/P region of MCP are not clear (Liszewski and Atkinson 1992).

Short consensus repeats

SCRs are found in many proteins that function in the complement system; however, many are also present in immune-related proteins that are not complement components, and others that are present in proteins that are not involved in immune responses. The conserved amino acids within SCRs include four Cys, three Gly, two Pro, two Tyr or Phe, and one Trp (Chou and Heinrickson 1997). An alignment of the SCRs from both SpCRL and SpCRS was done using ClustalX (Thompson et al. 1997) and demonstrated that they were typical SCRs, with four Cys and at least 9 of the 12 conserved amino acids present in each domain (Fig. 5). When the alignment was used to generate an unrooted neighbor-joining tree (Saitou and Nei 1987), results showed that the four C-terminal SCRs in SpCRL (SCRs 15–18) were most similar to the four SCRs in SpCRS, and that they were positioned in the same relative order within the two proteins. (This similarity in SCR sequence and domain organization is indicated in Fig. 4 by fill patterns.) This region, which included the four C-terminal SCRs from SpCRL, an N-terminal S/T/P-region, and the FIMAC domain but excluded the C terminal S/T/P-region (which was quite different in sequence and length between the two deduced proteins), was 37% identical and 56% similar between the two proteins (Fig. 4). For the remaining SCRs in SpCRL, four (SCRs 1, 8, 9, and 13) clustered together, with SCR 8 being most similar to SCR 13, and SCR 1 being most similar to SCR 9. Two additional pairs of similar SCRs were identified in SpCRL; SCR 4 was most similar to SCR 14, and SCR 7 was most similar to SCR 12 (denoted with similar patterns in Fig. 4). Although the relative orders of the SCRs at the C-terminus of SpCRL were the same as those in SpCRS, the two sets of SCRs within SpCRL that show sequence similarities were not positioned in the same order (1, 4, 7, 8 vs 9, 14, 12, 13; see Fig. 4).

Fig. 5
figure 5

SCR alignment. The SCRs from SpCRS and SpCRL were aligned using ClustalX (Higgins et al. 1996) and edited in Word (Microsoft). The consensus amino acids that have been defined for SCRs in general, and most of which are found in the SCRs shown here (between 33% and 100%), are indicated in boldface and on the bottom line. Some linkers between SCRs were truncated for the purposes of improving this alignment and are indicated with an asterisk

Full size image

The phylogenetic analysis was expanded to identify sequence similarities between the SCRs from SpCRS and SpCRL with SCRs from both complement and non-complement proteins. GenBank was queried using BLASTX with the entire SpCRL and SpCRS protein sequences to identify proteins that contained SCRs that were similar to those in SpCRL and/or SpCRS. This search identified SCRs from both complement and non-complement proteins (see footnote to Table 3) and all SCRs (n=549 from 87 proteins) were aligned and used to generate a neighbor-joining tree (not shown). Clustering of SCRs implied sequence similarities, which were used to infer putative functional similarities (Table 3) and have been indicated with symbols in Fig. 4. Many SCRs from proteins not involved in complement or immune function that were included in this analysis did not cluster with the SCRs from SpCRL and SpCRS (see footnote to Table 3). The most significant result of this analysis suggested that SCRs 5, 6, 7, and 12 from SpCRL have sequence similarities with SCRs in vertebrate Bf or C4BP, which are known to bind C3b or C4b, respectively (Hourcade et al. 1995; Blom et al. 2001), and are also known to have cofactor activity (Gordon et al. 1995; Kuhn et al. 1995) (Table 3). SCRs 1, 8, 9, 13, and 17 from SpCRL clustered with SCRs in factor H that bind sialic acid (Gordon et al. 1995) and/or C3b (Zipfel et al. 1999; Jokiranta et al. 2000). Because the C-terminal SCRs in both SpCRL and SpCRS were similar to each other and were positioned in the same relative order within the proteins (see above), it was not surprising that they clustered with each other and with the same SCRs from other proteins. The similarities identified for the C-terminal SCRs in both sea urchin proteins implied functions such as maintenance of structural conformation (SCR 1 from SpCRS, SCR 15 from SpCRL), cofactor activity and binding functions for sialic acid and C3b (SCR 3 from SpCRS, SCR 17 from SpCRL; Zipfel et al. 1999), and putative cell-adhesion activities (SCR 4 from SpCRS, SCR 18 from SpCRL; Fries et al. 1993). Overall, these results suggested the possibility that some of the SCRs in SpCRL and SpCRS may be involved in functioning within the complement system of the sea urchin and perhaps through interactions with SpC3. Furthermore, because cell surface sialic acid in mammals is important in blocking the alternative and lectin pathways of complement through binding factor H (Jarvis et al. 1987; Jack et al. 2001), SCRs in the sea urchin proteins that show sequence similarities to SCRs with sialic acid-binding activities suggest that SpCRL may also associate with cell surfaces and perhaps function in complement regulation.

Table 3 Sequence similarities between short consensus repeats (SCRs) from S trongylocentrotus p urpuratus complement related protein, long form (SpCRL) and S. p urpuratus complement related protein, short form (SpCRS) and SCRs from other proteins
Full size table

FIMAC domains

Two types of proteins are known to contain FIMAC domains: the heavy chain of factor I and complement components C6 and C7 (Minta et al. 1996). The modular structure of factor I heavy chain consists of a FIMAC domain, a SRCR domain, and two LDL-receptor domains, while the light chain is a serine protease domain (reviewed by Arlaud et al. 1998). The FIMAC domain in factor I is also thought to be involved in protein-protein interactions (DiScipio 1992). The domain organization of complement components C6 and C7 have one or two thrombospondin domains, a type A LDL-receptor domain, a perforin-like domain, another thrombospondin domain, and two SCRs followed by two FIMAC domains (DiScipio and Hugli 1989; Haefliger et al. 1989). The SCR domains and the FIMAC domains in C6 are known to bind C5b (DiScipio 1992; DiScipio et al. 1999). Initial analysis of the C-terminal region in both SpCRL and SpCRS identified a single Kazal or FIMAC domain based on BLASTX searches of GenBank. This sequence has been identified previously as sea urchin coelomocyte expressed sequence tags (ESTs) [EST132 (accession no. R61984), EST351 (accession no. R62091), and EST202 (accession no. R62023), Smith et al. 1996]. Kazal domains are part of the follistatin serine protease family and function as serine protease inhibitors (Schlott et al. 2002). FIMAC domains are a subset of that family; however, they function as complement cofactors by binding C3 convertase, which promotes the cleavage of C3b by factor I (Terado et al. 2002), or by binding C5b in the terminal pathway (DiScipio 1992; DiScipio et al. 1999).

To determine whether the domains from SpCRL and SpCRS were similar to FIMAC domains from factor I and C6/C7 or if they were more like Kazal domains in protease inhibitors, they were used in a phylogenetic analysis to determine with which set of domains they clustered. The data set included 79 Kazal and FIMAC domains from 44 proteins as identified by BLASTX searches of GenBank. The sequences were analyzed by parsimony using the ratchet algorithm (Nixon 1999) in PAUP (Sikes and Lewis 2001; Swofford 2002), and a portion of the 50% strict consensus tree is shown in Fig. 6a. All of the FIMAC domains, plus those from SpCRL and SpCRS and two Kazal domains, clustered together in a clade defined by node A. Within that clade, a subclade defined by node B included all factor I FIMAC domains plus the sea urchin domains. The FIMAC domains from complement components C6 and C7, plus the zebrafish FIMAC domains (from a protein with a structure consistent with a C6 homologue, accession no. AAH57429.1), clustered in two clades defined by C (Fig. 6a). This result suggested that the SpCRL and SpCRS domains were more similar to FIMAC domains than Kazal domains, and perhaps more similar to factor I FIMAC domains than to C6/C7 FIMAC domains. Bremer support (Bremer 1988) was calculated for several of the internal nodes, and is shown below the branch near the node (for details, see legend to Fig. 6). Bremer support indicates the number of extra steps that is required in a longer tree before a clade is lost from the consensus tree (Kitching et al. 1998). Consequently, support was quite good for nodes A and B, within which SpCRL and SpCRS clustered. Bootstrapping analysis was conducted, but results were uninformative. This was due to the short length of the domain (84 informative positions) and the large number of sequences (79) that were analyzed. Bootstrapping is optimal when the data matrix has at least 1,000 informative characters (Kitching et al. 1998). Bootstrapping results may also have been confounded by the presence of several sequences, such as the FIMAC domain from carp factor I-A, which did not align well (see Fig. 6b). The carp sequence did not cluster repeatedly within a given clade but “jumped” to a variety of locations within clade A in different, but equally parsimonious, trees. An alignment of the sequences that clustered in the clades defined by nodes B and C, plus the carp sequence, is presented in Fig. 6b. For most sequences, including those from SpCRS and SpCRL, the conserved amino acids are present including 10 Cys, which are involved in forming the disulfide bonds within the domain (Terado et al. 2002), in addition to Trp, Gln/Glu, Arg/Lys, Pro, Val and Ala/Ser. The carp sequence does not align well, particularly in the C-terminal end of the domain.

Fig. 6a, b
figure 6

Phylogenetic analysis and alignment of the FIMAC domains from SpCRL and SpCRS. a An alignment using 79 FIMAC and Kazal domains from 44 proteins was done using ClustalX (Thompson et al. 1997), edited in BioEdit sequence alignment editor (Hall 1999), formatted in Winclada (obtained from http://www.cladistics.com), and analyzed in PAUP (Sikes and Lewis 2001; Swofford 2002) to generate a phylogenetic tree. Multiple FIMAC and Kazal domains from the same protein are numbered and treated as independent sequences. Accession numbers are indicated in brackets. The most parsimonious tree was calculated using the parsimony ratchet algorithm, PAUPRat, version 1 (Nixon 1999), with 1,000 iterations, each with 20 random additions of taxa and a limit of 20 trees saved and swapped for each random addition. A portion of the 50% strict consensus tree is shown that was generated from 919 equally parsimonious trees. Bremer support (Bremer 1988) was calculated for several internal nodes using PAUP and TreeRot, version 2 (Sorneson 1999), and is indicated near the relevant nodes below the branches. Kazal and FIMAC domains (54) from 18 other proteins are not shown on the tree. They include bovine acrosin inhibitor [P01000], C. elegans protease inhibitor [CAB01753.1], chick agrin [P31696], chick follistatin [Q90844], chick ovoinhibitor [P10184], chick ovomucoid [P01005], crayfish proteinase inhibitor [CAA56043.1], Dipetalogaster thrombin inhibitor [CAA10384.1], dog double-headed protease inhibitor [P01002], eel pancreatic proteinase inhibitor [P11706], Eurasian badger double-headed protease inhibitor [P16226], herring sperm activating protein [BAA14008.1], human acrosin trypsin inhibitor [P20155], human agrin [AAC39776.1], human serine protease inhibitor [CAB40839.1], human prostacyclin stimulating factor [AAB32370.1], human testican 3 [NP-058646], human transmembrane protein [AAA64622.1], Japanese quail QR1 protein [P23499], lion double-headed protease inhibitor [P08481], leech protease inhibitor [AAK58688.1], mouse testican [CAA63448.1], mouse serine protease inhibitor Kazal type 4 [NP035593.1], pig sperm-associated acrosin inhibitor [P00999], rat pancreatic trypsin inhibitor [P09655_IPK1], sea anemone elastase inhibitor [P16895], turkey ovomucoid [P01004]. b An alignment of FIMAC domains from SpCRL and SpCRS with FIMAC domains that clustered in a. The alignment was done with ClustalX (Thompson et al. 1997) and edited in Word (Microsoft). Consensus amino acids are shown in boldface and on the bottom line

Full size image

Although the domains of both SpCRL and SpCRS cluster with FIMAC domains from factor I proteins, the domain structure of the sea urchin proteins, in which the FIMAC domains were preceded by two SCRs (Figs. 2, 3, 4), was similar to the domain structure of complement proteins C6 and C7 (DiScipio 1992). Factor I, on the other hand, has the FIMAC domain located at the N-terminal end of the protein and is not associated with adjacent SCRs (Arlaud et al. 1998; Terado et al. 2002). Together, the combination of sequence similarity to factor I and the structural similarity to C6 and C7 implies that the FIMAC domains in SpCRS and SpCRL may have protein-protein binding functions similar to that known for FIMAC domains in factor I (DiScipio 1992) or for C6 (DiScipio et al. 1999), both of which interact with C3 or C5—members of the thioester-containing protein family. It is interesting that the C6 homologues identified in Amphioxus (Suzuki et al. 2002) and Ciona (Azumi et al. 2003) have a domain organization that is similar to the N-terminal half of vertebrate C6 and C7, including the thrombospondin domains, a type A LDL-receptor domain, and a perforin-like domain, but lack the SCR and FIMAC domains. Consequently, the C6-like sequences in these two lower chordates share no domains with SpCRL and SpCRS, which also appear to be C6-like. Determination of whether the chordate or echinoderm C6-like proteins might actually function in a terminal pathway will require additional analysis of complement activities from these invertebrates.

Fucolectin domain

A BLASTX search of the region between SCR 10 and SCR 11 in SpCRL revealed a significant sequence similarity with a family of fucolectins that have been characterized from the serum of the European eel, Anguilla anguilla, and have been denoted the A. anguilla agglutinins (AAA) (Bianchet et al. 2002). Fucolectins are a class of lectins that recognize fucose on the surface of cells and play an important role in the innate immune functions of both vertebrates and invertebrates as pattern-recognition receptors (Bianchet et al. 2002). An alignment between the SpCRL fucolectin domain and the seven eel fucolectin proteins revealed a number of conserved amino acids in addition to several positions where the eel and sea urchin sequences differed (Fig. 7). The fucolectin structure, which is diagnostic of the F-lectin family, is a β barrel with jellyroll topology of eight antiparallel β strands oriented in two sheets consisting of five and three strands (Bianchet et al. 2002). The domain is locked into its conformation with two disulfide bonds and two salt bridges. The disulfide bonds were conserved in both the eel fucolectins and the SpCRL fucolectin domain (Fig. 7). The salt bridges were formed between Arg41-Glu140 and Asp64-Arg131 in the eel sequences that interact with a cation—either Na+ or Ca2+. In SpCRL, neither Arg involved in salt bridges were present in the conserved positions, but were located within two to four positions of the site conserved in the eel sequences. In the AAA fucolectins, there are five loops connecting the β strands of the β barrel that protrude like complementarity-determining regions (CDRs) and surround the fucose-binding site. The CDRs in SpCRL showed conserved amino acids that defined the borders of the regions and the sizes of the regions were similar to that in AAA. The fucose-binding site, composed of the motif H(X)24RGDCC(G/E)ER, showed conservation of the significant amino acids in SpCRL except for His52, which was replaced with an Arg, and the double-Cys motif, which was missing in SpCRL. Fucolectins have been shown to have hemagglutination activity mediated by the conserved residues AIDGN located between CDR 1 and CDR 2, a motif that was partially conserved in SpCRL, ARDGN. The detailed structural analysis of fucolectins reported by Bianchet et al. (2002) has enabled a detailed analysis of the fucolectin domain of SpCRL and overall, the alignment suggests that the sea urchin protein has a fucolectin domain, but comparisons revealed a divergent structure from that of the AAA proteins (G. Vasta, personal communication). Variations in amino acids required for salt-bridge formation to stabilize the barrel structure and the missing His52 which is involved in fucose binding in AAA, make it possible that the sea urchin domain may bind a different ligand than that characterized for AAA. The location of the ligand, perhaps on self-cells or on the pathogen, may provide a clue as to how SpCRL might be involved in the complement system of the sea urchin.

Fig. 7
figure 7

Fucolectin alignment. The fucolectin domain from SpCRL was aligned with the seven fucolectins from the European eel, Anguilla anguilla [denoted A. anguilla agglutinins (AAA)], using ClustalX (Higgins et al. 1996) and edited in Word (Microsoft). The conserved amino acids are shown in boldface within the alignment. Amino acids directly involved in binding functions are shown in boldface in the top line in addition to the fucose-binding site and the hemagglutination site. The complementarity-determining regions (CDRs) are labeled and identified with brackets. The two Arg located within the fucose-binding site and the His located towards the N-terminus (in boldface, top line) are the triad of amino acids that are involved in the polar interactions with fucose. The first residue of this triad in SpCRL is Arg, which does not match the conserved His in the eel fucolectins. The Cys are identified with numbers on the top line and disulfide bonds are formed between Cys1–4 and Cys2–3 in AAA sequences (Bianchette et al. 2002). Four amino acids involved in stabilizing the domain through salt bridges in the AAA sequences are indicated with arrows. Salt bridges are formed between Arg41 and Glu149 and between Asp64 and Arg131 (numbering is based on Bianchette et al. 2002). Amino acids involved in cation binding are shown in italics on the top line

Full size image

Transcript sizes and alternative splicing

Sp5

The mRNA length for both cDNAs were determined by poly(A)+ Northern blots (Fig. 8). Results for Sp5 (Fig. 8a) showed two pairs of bands (7.4 kb and 7.9 kb, 11.9 kb and 12.7 kb) with the smaller of each pair being more intense. All major bands were larger than the sequenced Sp5 cDNA (5.965 kb) suggesting that parts of the 5′UTR and 3′UTR were missing. The 5′UTR was very short, and it is feasible that much of this region was missing. However, the blot and the cDNA sequence both suggested that the short stretch of As at the 3′ end was not the actual poly(A)+ tail, indicating that some of the 3′UTR was also missing. This was supported by the absence of a polyadenylation signal sequence located 5′ of the poly(A)+ stretch (Fig. 2). Taking this into consideration, the known sequence for Sp5 may correspond to either the 7.4-kb or 7.9-kb band. Longer exposures of the same blot showed five less prominent bands ranging in size from 1.2 kb to 3.8 kb (Fig. 8b). These bands may not correspond directly to the Sp5 sequence because they were significantly shorter than the known sequence of the cDNA and did not hybridize well to the probe (see below for a discussion of small gene family and alternative splicing).

Fig. 8a–d
figure 8

Expression of Sp5 and Sp5013 in coelomocytes. Poly(A)+ RNA was isolated from coelomocytes from an animal within 1 day of shipment from California and therefore assumed to be activated. Between 1 μg and 1.8 μg mRNA was loaded per lane on a Northern blot and analyzed with riboprobes made from Sp5 and Sp5013. Clones used as riboprobe templates for this purpose included the entire sequence for Sp5013 and approximately the 3′ half of Sp5 (see legend to Fig. 1). The blot probed for Sp5 was exposed for 30 min (a) and 5 h (b). The blot probed for Sp5013 was also exposed for 30 min (c) and 5 hr (d). Expression of SpL8, the sea urchin homologue of the human ribosomal gene L8, was used as a control, and a single band of 1.6 kb is shown. Sizes of the bands are indicated between the two sets of blots. e Differences in observed sizes between bands for Sp5013 (d) and the approximate number of deleted SCRs which might have resulted in the observed differences are indicated to the right (see text)

Full size image

Sp5013

The Northern blot for Sp5013 revealed three major bands of 6.0 kb, 4.6 kb, and 3.8 kb after a short exposure (Fig. 8c), plus six weakly hybridizing bands of 12.7 kb, 11.4 kb, 9.3 kb, 7.4 kb, 2.9 kb, and 2.2 kb that appeared after a longer exposure (Fig. 8d). The known sequence of Sp5013 (2.795 kb) was not exactly the same size as any of the bands on the Northern; however, the cDNA might correspond with either the 2.9-kb or 3.8-kb band if parts of both the 5′UTR and 3′UTR were missing from the cDNA sequence. The 3′UTR was probably complete because the poly(A)+ tail was preceded by two polyadenylation signal sequences located 100 nt to 150 nt upstream. Although we argue above that the 5′UTR was complete based on duplicate cDNAs with the same sequence at the 5′ end, the shortness of this region is not typical for sea urchin mRNA.

Similar band sizes and alternative splicing

Size comparisons between the array of bands for Sp5 and Sp5013 in Fig. 8 indicate that a few may be the same. However, bands of the same size never appeared as major bands for both probes. The best example of message sizes present on both blots was the 12.7-kb band, which was one of the major Sp5 doublet bands and was a minor band for Sp5013 (Fig. 8b, d). Another example was the 3.8-kb band, which was a major band on the blot for Sp5013 but was a minor band for Sp5. Finally, the 7.4-kb band, which appeared on both blots, was a major band for Sp5 and a minor band for Sp5013. In addition, there was a set of minor bands of 2.8 kb/2.9 kb that was present on both blots. The probes used on the Northern blots were produced from clones that included the entire Sp5013 sequence, and that spanned the 3′ half of the Sp5 sequence (the fucolectin domain to the 3′ end, see legend to Fig. 1). These were the regions of the two messages that showed the greatest similarity and may have resulted in some cross-hybridization, even at high stringency. Because the small bands did not correspond to sizes estimated from the known sequences of the cDNAs, this suggests that the probes may have cross-hybridized to messages other than Sp5013 and Sp5, implying the presence of a small gene family with similar sequences.

Multiple bands on the Northern blots may also be interpreted as evidence of alternative splicing. The pattern of size differences among the bands for Sp5013 was present in multiples of 0.7–0.8 kb (Fig. 8e), which corresponds with regions of the message that would encode pairs of SCRs including the linker between them. For example, if the length of the message encoding SCR 1 and SCR 2 plus the linker were doubled, it would be 810 nt. Similarly, if the sequence encoding SCR 3 and SCR 4 plus the linker were doubled, it would be 738 nt (Fig. 8e). These sizes correspond with regular size differences noted between the bands for Sp5013 (Fig. 8c, e) and could be interpreted as the result of alternative splicing of multiples of SCR exons. If this set of messages of varying lengths were transcribed, it would result in a set of proteins with many more SCRs and perhaps fewer SCRs than were identified from the deduced sequence of SpCRS. In support of this interpretation, several clones for Sp5013 were identified with identical sequences at the 5′ and 3′ ends to the sequence provided here, but with protein-coding regions of variable sizes (data not shown). The deletion of SCRs by alternative splicing has recently been demonstrated for another sea urchin cDNA, Sp152, which encodes the homologue of complement factor B, SpBf (Terwilliger et al. 2000; unpublished data). Splice variants of Sp152 were identified in which the first and/or the fourth SCR were/was deleted from the total of five.

Genome analysis

Genome blots using sperm DNA from two sea urchins that was digested with HindIII, EcoRI, and PstI were analyzed with probes for each gene. Results showed multiple bands of variable intensity in both blots that would be consistent with gene structure of multiple exons plus cross-hybridization among members of a small gene family (data not shown).

Expression patterns of Sp5 and Sp5013 in sea urchin tissues

Expression patterns of Sp5 and Sp5013 were characterized by RT-PCR using several sets of gene-specific primers and total RNA from several sea urchin tissues (Fig. 9). The tissues examined included coelomocytes, gut, gonad, pharynx, esophagus, and axial organ and were isolated from sea urchins that were either activated by injection of LPS (Fig. 9a), were in an immune activated state (Fig. 9b), or were immunoquiescent (Fig. 9c). Results showed that all tissues expressed both Sp5 and Sp5013, and that expression was not altered in coelomocytes from immunoquiescent animals after injection with LPS (Fig. 9d). In comparison, expression of Sp056 (accession no. AY336600), which encodes a C-type lectin that is reliably inducible with LPS (unpublished data), was restricted to LPS-activated coelomocytes. A sea urchin homologue of the human ribosomal gene L8, SpL8 (EST219, accession no. R62029; Smith et al. 1996) was used as the control to ensure that approximately equal amounts of cDNA template were used in the PCR reactions. Overall, the RT-PCR analysis indicated that Sp5 and Sp5013 were ubiquitously expressed and were not induced by immune challenge.

Fig. 9
figure 9

Expression of Sp5 and Sp5013 in sea urchin tissues. Total RNA was isolated from coelomocytes, gut, gonad, pharynx, esophagus, and axial organ and used in reverse transcriptase (RT)-PCR. The RNA was analyzed for expression of Sp5 and Sp5013, and Sp056 (an LPS-inducible C-type lectin, accession no. AY336600, unpublished data). SpL8, a constitutively expressed homologue of the human ribosomal gene L8, was used as the control. a A sea urchin was injected with LPS for 2 days and coelomocytes and other tissues were collected for RNA isolation 24 h after the last injection. b The sea urchin used for analysis was assumed to be upregulated as a result of shipping stress. c Tissues were collected from an immunoquiescent sea urchin which had been kept in a closed system aquarium for 2 years prior to being used in this experiment (see Gross et al. 1999). d Coelomocytes were taken from an immunoquiescent sea urchin 15 min prior to a single LPS injection and 24 h post-injection. C Coelomocytes, G gut, OT ovary or testis, E esophagus, P pharynx, AO axial organ. Control PCR reactions employed cloned templates with the appropriate primers (positive) or reactions in which the cloned template was omitted (negative)

Full size image

Summary

Analysis of the two mosaic proteins, SpCRL and SpCRS, indicated that they share domains with factor H, factor I, C6 and C7. Similarities among SCRs from a variety of complement proteins (Table 3) indicated that many of the SCRs in both SpCRL and SpCRS may function in interactions with other proteins such as SpC3, or with cell surfaces perhaps through binding carbohydrates (Fig. 4). The FIMAC domains in SpCRL and SpCRS show sequence similarities with FIMAC domains from complement proteins rather than with Kazal domains from protease inhibitors (Fig. 6). Functions of SpCRL might involve interactions with SpC3 as a regulatory protein to modulate complement activation, either in fluid phase or in association with self-cell surfaces, either through some of the SCRs and/or through the fucolectin domain. This is the first evidence of proteins in the sea urchin that might function in a putative complement regulatory system and/or perhaps in a putative terminal pathway. Knowledge of whether an expanded complement system exists in the sea urchin, as has been identified from the Ciona genome (Azumi et al. 2003), will also most likely come from the analysis of the sea urchin genome when it is completed. An understanding of all the components and pathways of the sea urchin complement system will provide a better understanding of the evolution of this system that has culminated in the complex complement system that is essential to immune functions in the higher vertebrates today.