Single Sea Urchin Phagocytes Express Messages of a Single Sequence from the Diverse Sp185/333 Gene Family in Response to Bacterial Challenge

Immune systems in most organisms are complex and include large families of highly diverse immune response genes that encode a broad array of antipathogen proteins (reviewed in Refs. 1–5). Some examples include fibrinogen-related proteins (FREPs) in molluscs (6, 7), V region–containing chitin-binding proteins in protochordates (8, 9), and R proteins in plants (1). Highly diverse immune response genes identified from several sea urchin species are members of the 185/333 family (10), which was originally identified by changes in gene expression in coelomocytes from immune-challenged compared with nonchallenged purple sea urchins (Strongylocentrotus purpuratus) (11, 12). Of the expressed sequence tags from those screens, 73% matched to two previously uncharacterized sequences, DD185 (11) and EST333 (13). In S. purpuratus the gene family is called Sp185/333 because it has been reported in another echinoid species (14).

Individual sea urchins are estimated to have 50 ± 10 Sp185/333 genes (reviewed in Ref. 10), which range in size from 1.2 to 2.0 kb with two exons, of which the second contains several tandem and interspersed repeats (15). The genes are unusual and unique because optimal alignments of the second exon require the insertion of large gaps that define blocks of similar sequence called “elements” (10, 16–18). There are 25–27 different elements (depending on the alignment) that range in length from 12 to 357 nt and are variably present or absent in different genes, which result in mosaics of recognizable element patterns (15). Sequence diversity within the members of the gene family is evident from comparisons among 171 genes cloned from three individual sea urchins, which shows that although the genes are ≥88% similar, none shares identical sequences among individuals (15). The gene sequence diversity is predicted to occur through recombination, gene conversion, deletions, duplications, and perhaps meiotic mispairing (16, 19), which is typical for other gene families. Computational predictions suggest extraordinarily swift recombination frequency for the Sp185/333 genes that lies between that of V-J somatic recombination in the TCR α-chain and that of sea urchin histone H3 genes, which do not recombine (16). Predicted gene recombination may be the outcome of the unique gene family structure consisting of small, tightly clustered genes that share sequences and are surrounded by microsatellites (19). Although all but 1 of the 171 sequenced genes have perfect full-length open reading frames (15), half of the mRNAs encode truncated proteins due to single nucleotide polymorphisms and small (1–2 nt) indels that result in early stop codons or nonsense sequence leading to stop codons (20), which are likely the result of RNA editing (21). Overall, there appears to be multiple levels of immune diversification based on gene sequence diversity and RNA editing, in addition to putative posttranslational modifications and multimerization of the encoded proteins (22) with possible synergistic activities among the Sp185/333 proteins. The extraordinary diversity of the Sp185/333 system produces a very broad array of both full-length and truncated Sp185/333 proteins in response to immune challenge (22, 23).

The coelomocytes of sea urchins mediate the innate immune response and are composed of red and colorless spherule cells, vibratile cells, and phagocytes (reviewed in Ref. 24). Types of phagocytes are based on the structural characteristics of the actin cytoskeleton (25–27) and the overall size of the cells. The large phagocytes include discoidal phagocytes that have radially arranged actin cables with a disc-like shape, and polygonal phagocytes that have actin cables that align across the cell, resulting in polygon-shaped cells. The small phagocytes are much smaller and have a perpetual filopodial morphology (23, 28). Sp185/333 genes are expressed in small and polygonal phagocytes (23) and are upregulated in response to immune challenge with pathogen-associated molecular patterns (PAMPs), including LPS, β-1,3-glucan, dsRNA, peptidoglycans, and marine bacteria (11, 12, 20, 29). Immune challenge with LPS significantly increases the numbers of Sp185/333⁺ phagocytes in the coelomic fluid (CF) and in other tissues of the sea urchin, which are likely Sp185/333⁺ phagocytes that are thought to wander through the tissues (23, 29, 30). Recent work to evaluate functions of both native and recombinant Sp185/333 proteins shows that they bind to bacteria and some PAMPs with high affinity (C.M. Lun, C.S. Schrankel, H.Y. Chou, S. Sacchi, and L.C. Smith, unpublished observations; H.Y. Chou and L.C. Smith, unpublished observations).

Although the expression of the Sp185/333 genes has been documented in populations of coelomocytes, expression in single phagocytes has not been evaluated. To address this question, we employed nested RT-PCR with samples of immune-activated single phagocytes to determine the level of Sp185/333 gene expression. We found that 70% of the single-cell samples had Sp185/333 amplicons of the same size with identical or nearly identical sequences. Furthermore, all samples for which there was a near zero probability of more than one cell per sample show amplicons of identical size and sequence. These results suggest that individual phagocytes contain Sp185/333 mRNAs of uniform sequence, which is a result that has not been identified previously with regard to the expression of immune response genes for any invertebrate. This infers an additional level of complexity for the sea urchin immune system.

Adult purple sea urchins, S. purpuratus, were purchased from Marinus Scientific (Long Beach, CA) or the Southern California Sea Urchin Company (Corona del Mar, CA). Animals were maintained as described, fed weekly with commercial rehydrated kelp (Quickspice), and those chosen for the study (n = 6) were considered immunoquiescent (28, 31).

Vibrio diazotrophicus (Gram-negative marine bacteria; no. 33466, American Type Culture Collection) was rotated at room temperature for 16–21 h in 5 ml of marine broth (3.44% marine broth, 0.3% yeast extract, and 0.5% proteose peptone; Difco). Bacteria were heat-killed at 95°C for 15 m and washed three times with artificial CF (aCF; see Ref. 23) and stored for up to 5 d in aCF at 4°C or until used.

Sea urchins were immunologically activated by one (n = 2) or three (n = 1) separate injections of 10⁶ heat-killed V. diazotrophicus/ml CF. The initial injection was administered at 0 h followed by a second injection at 24 h and a third injection at 48 h. The total volume of CF per sea urchin was estimated according to Smith et al. (32). Other animals (n = 3) received 10⁵ Vibrio/ml CF for the first injection and 10⁶ Vibrio/ml CF for the second injection.

Whole CF (wCF; CF plus cells, 0.2–0.3 ml) was withdrawn from sea urchins into a sterile 1-ml syringe after inserting the needle through the peristomial membrane. The syringe was preloaded with 0.2–0.3 ml ice-cold calcium and magnesium–free seawater with 70 mM EDTA and 50 mM imidazole (CMFSW-EI; see Ref. 23), supplemented with 20 mM DTT according to Hillier and Vacquier (33). Following the wCF withdrawal, the syringe was filled with additional cold CMFSW-EI to 1 ml and expelled into a 1.5-ml tube on ice. Samples of wCF were collected before (0 h) and after (24 h) each Vibrio injection. Coelomocytes collected before injection were pelleted by centrifugation and stored in RNAlater (Ambion) at −20°C. Samples collected after Vibrio injection(s) were held on ice until used for density gradient centrifugation (see below). Cells were counted using a TC-20 automated cell counter (Bio-Rad) or a hemocytometer.

Coelomocytes from immune-activated sea urchins were separated on a discontinuous density gradient composed of ice-cold Percoll (Amersham) that had been predialyzed using SnakeSkin dialysis tubing (3500 m.w. cutoff; Thermo Scientific) against CMFSW-EI at 4°C overnight, followed by the addition of 20 mM DTT. Percoll layers of 2.5, 20, 40, and 70% were generated by underlayering 2 ml increasing concentration into a 12-ml glass round-bottom culture tube according to published methods (34). A top layer (0.5 ml) of 2.5% Percoll served to separate the molecules in the wCF from the cells after separation. Coelomocytes (≤2 × 10⁶ or a maximum of 0.5 ml) were gently overlayered using a Pasteur pipette and the gradient was centrifuged in a swinging bucket rotor (Eppendorf, model 5804R) at 250 × g for 15 m at 4°C with a soft start and no brake. Cells were collected from each layer plus those pelleted to the bottom and diluted to 12 ml with fresh, ice-cold CMFSW-EI plus DTT, pelleted again, and resuspended in 0.2 ml fresh CMFSW-EI. Differential cell counts were done with a hemocytometer to identify enrichment of different cell types in each fraction. Dead cells were identified by 1.5 mM propidium iodide (Sigma-Aldrich) exclusion as visualized in an Axioplan fluorescence microscope (Carl Zeiss Microscopy).

Fractionated coelomocytes (either 5 × 10⁴ phagocytes, 10⁵ vibratile cells, or 10⁵ red spherule cells) were spun onto poly-l-lysine slides (Erie Scientific) followed by fixing and processing for immunofluorescence according to Brockton et al. (23). Briefly, fixed cells were incubated with a mix of three rabbit anti-Sp185/333 Abs (anti–185-66, -68, and -71 diluted 1:4000 in blocking buffer; 2% normal goat serum, 1% BSA in PBS) plus mouse monoclonal anti-human actin (diluted 1:600 in blocking buffer; MP Biomedicals). Secondary Abs were a mixture of goat anti-rabbit Ig conjugated to Alexa Fluor 568 (1:400 dilution; Invitrogen) and donkey anti-mouse Ig conjugated to Alexa Fluor 488 (1:200 dilution; Invitrogen). Cells were washed and mounted in ProLong Gold Antifade with DAPI (Invitrogen). Primary Abs were replaced with an equal volume of blocking buffer for the negative controls. Images were captured on an Axioplan fluorescence microscope (Carl Zeiss Microscopy) with a black-and-white CCD camera (Hitachi), to which false color was added using the Olympus MicroSuite B3SV program, or an LSM 710 confocal microscope (Carl Zeiss Microscopy).

Concentrations of fractionated coelomocytes were estimated with a TC-20 automated cell counter (Bio-Rad), followed by serial dilutions with ice-cold CMFSW-EI to a final estimated cell concentration of 1 cell/μl, which was defined as 1×, followed by further dilutions of 2×, 4×, and 10× with fresh CMFSW-EI. Single-cell samples in lysis buffer (see below) were held at −70°C until processed for cDNA synthesis and nested RT-PCR. Undiluted cell fractions were pelleted and stored in RNAlater (Ambion) at −20°C until cDNA synthesis, which was used for RT-PCR and quantitative RT-PCR (qRT-PCR; see below).

Total RNA from undiluted cell fractions was extracted using an RNeasy Micro kit (Qiagen) according to the manufacturer’s protocols, including an on-column treatment with DNAse I. Total RNA from unfractionated cells was reverse transcribed using SMARTScribe reverse transcriptase (Clontech) or SuperScript III reverse transcriptase (Invitrogen) with random hexamer primers (Operon) according to the manufacturers’ instructions. cDNA from single-cell samples were generated using the SuperScript III CellsDirect cDNA synthesis kit (Invitrogen) according to the manufacturer.

qRT-PCR was carried out on cDNAs from cells from three sea urchins collected before and 24 h after challenge with heat-killed Vibrio, and that were isolated from the 20 and 40% Percoll fractions. PCR reactions used 1 μl of the reverse transcription reaction (diluted 10×) with the Sp185/333 primers F5 and 3′ untranslated region (UTR), or the SpL8 primers SL8 and ASL8 (Supplemental Table I), and evaluated with ABsolute SYBR Green fluorescein (Thermo Fisher). SpL8 expression served as the housekeeping control. The qRT-PCR reactions were performed in triplicate in the iCycler iQ (Bio-Rad) with the following program: 95°C for 12 min, then 40 cycles of 95°C for 15 s and 59°C for 45 s. qRT-PCR results were calculated based on the relative gene expression ratio (R), using the equation 2^−∆∆CP (PerkinElmer) under the assumption that E_target = E_ref = 2, where E_target is represented by the Sp185/333 target sequence and E_ref is represented by the SpL8 reference sequence (35).

Single-cell samples were analyzed by nested PCR to amplify Sp185/333 and SpL8 cDNAs. The first round of PCR included 2 μM each external primer (Supplemental Table I), 0.2 mM each deoxynucleotide, 1× company supplied buffer, 0.025 U TaKaRa ExTaq DNA polymerase (Clontech), plus 1 μl reverse transcriptase reaction (cDNA template) in a total of 20 μl. Molecular-grade water was used as used in place of the template for the negative control. For the positive control reactions, the cDNA clone EST219 (GenBank accession number R62029; http://www.ncbi.nlm.nih.gov/genbank) (13) served as the SpL8 template for the external and internal SpL8 primers. Reactions were performed in either an iCycler IQ in nonquantitative mode (Bio-Rad), a PTC-200 Peltier thermal cycler (MJ Research), or a T100 thermal cycler (Bio-Rad). The cycling program was 94°C for 1 min followed by 30 cycles of 94°C for 30 s, 57°C (external 5′UTR and 3′UTR primers) or 54°C (external SpL8 primers) for 1 min, 72°C for 30 s, with a final extension of 72°C for 2 min and a 4°C hold. Nested PCR reactions included the identical reagent concentrations plus 2 μl amplicons from the first round of PCR. Both the SpL8 and Sp185/333 internal primers were designed with a higher annealing temperature to avoid amplification by the external primers carried over from the first PCR reaction in the nested reaction. When external primers were tested at the higher annealing temperature used for the internal primers, they did not amplify known cDNAs (not shown). The nested cycling program to amplify either Sp185/333 or SpL8 amplicons was 94°C for 1 min followed by 25 or 30 cycles of 94°C for 30 s, 62°C for 1 min, 72°C for 30 s, with a final extension of 72°C for 2 min and a 4°C hold. To identify Sp185/333 or SpL8 template contamination of the reagents or primers during the preparation of the nested PCR, 2 μl first round PCR negative control was reamplified using the internal primers (Supplemental Fig. 1). Equal volumes of amplified cDNAs were loaded and electrophoresed through 0.8–1% agarose (Promega) in 0.5× TAE buffer (0.04 M Tris base, 0.02 M glacial acetic acid, 1 mM EDTA). Amplicons were visualized with ethidium bromide, and images were captured on a DC120 digital camera (Eastman Kodak) with Digital Science1D software version 3.0.0 (Eastman Kodak).

Templates for PCR were either Sp185/333 nested amplicons (1 μl) from single-cell samples or a cloned Sp185/333 gene (10 ng, clone 2-073, GenBank accession number EF607742.1) (15), which served as the positive control. Reactions of 20 μl included 0.2 μM LF and R2 primers (Supplemental Table I) and 10 μl GoTaq Green ready mix (Promega). The cycling program was 94°C for 30 s, 62°C for 40 s, 72°C for 90 s, and 72°C for 5 min. Amplicons were evaluated by gel electrophoresis.

Plasmids used to evaluate the annealing characteristics of the internal primers were isolated using the standard alkaline lysis method (36), diluted either 200- or 40,000-fold in water, and 1–3 μl were used as templates in PCR of 10–20 μl. Reaction conditions replicated those used for the nested single-cell RT-PCR reactions described above, including 2 μM for each primer and 30 cycles; however, rather than ExTaq, GoTaq Green ready mix (Promega) was used according to the manufacturer’s protocol. Negative control reactions omitted templates. Amplicons were separated by gel electrophoresis and imaged.

Nested PCR amplicons from a subset of single-cell samples were reamplified using the following internal primer combinations; in5.1 + in5.2 (mixed) or in5.3 with either in3.1 or in3.2 (Supplemental Table I). The PCR conditions were the same as those for nested internal PCR described above. Amplicons were electrophoresed through 1% agarose gel for evaluation, and the DNA concentration was evaluated with a NanoDrop ND-1000 spectrophotometer (Thermo Fisher). Amplicons were sequenced directly without prior cloning (GenScript) using ether F2 and/or R9 primers (Supplemental Table I). F2 and R9 primers anneal at sites that are located between the internal primers and amplify all known Sp185/333 sequences (15, 18). Chromatograms were inspected for correct base calling, trimmed according to the chromatogram quality, and internal primer sites were identified manually for amplicons of ≤1 kb. Consensus sequences were constructed for those amplicons that were sequenced with both F2 and R9 primers. The final sequences were submitted to GenBank (accession numbers KJ408449–KJ408477) and compared with the nr database on GenBank by BLASTn to identify the element patterns for the Sp185/333 transcripts according to Terwilliger et al. (20). Correct open reading frames were identified for each sequence based on length and similarity to Sp185/333 amino acid sequences in GenBank using BLASTp. Sequences from each single-cell sample were compared with all others by global pairwise alignments (blosum62) with free end gaps using Geneious software (version 7.0.6; Biomatters) to identify the percentage similarity among sequences.

Statistical analysis of limiting dilutions was based on Poisson distributions. The average number of Sp185/333⁺ cells per sample was calculated for each dilution using the observed probabilities to equal zero cells (no amplification) in a sample [p(x = 0)] in the equation m = −ln[p(x = 0)]. The m values were used as λ in the equation p(x) = λ^xe^−λ/x! to calculate the probability to have exactly one (x = 1) cell per sample. Based on the observed probabilities of having either zero or one cell per sample, the probability to have more than one cell per sample was calculated with the equation 1 − [p(x = 0) + p(x = 1)] = p(x > 1).

Statistical analysis was performed on data acquired from qRT-PCR and cell counts from immunostaining using statistical analysis software (SAS, version 9.1.3). Coelomocyte counts from immunostaining were obtained before versus after Vibrio challenge. Statistical analyses were performed on R values and cell counts, and data were log₁₀ transformed to approximate normal overall distributions where appropriate. Statistical significance was assessed by ANOVA in general linear models regression procedures. Experimental error was addressed by including replicate measurements among model parameters.

The goal of evaluating Sp185/333 gene expression in single phagocytes required an immune challenge that would drive the activation of as many coelomocytes as possible with the induction of the maximum number of Sp185/333 genes. Although sea urchins have been evaluated for immune responses to a variety of PAMPs and uncharacterized marine bacteria (11, 12, 20, 22), most reports of sea urchin activation have employed LPS (23, 36, 37). Consequently, a putative sea urchin pathogen, V. diazotrophicus (Vibrio), that was originally isolated from an infected green sea urchin, Strongylocentrotus droebachiensis (38), was used as the immune challenge. Immunoquiescent sea urchins (n = 3) injected with heat-killed Vibrio showed a significant increase in coelomocytes in the CF after 24 h, which included increases in Sp185/333⁺ phagocytes (Fig. 1A, 1B). This was in agreement with responses to LPS (23, 29) in addition to other PAMPs (20), and therefore Vibrio was employed to activate antipathogen responses.

The phagocyte class of coelomocytes has been evaluated for Sp185/333 proteins (23, 29); however, expression from the vibratile and spherule cells is not known. Consequently, we used discontinuous Percoll density centrifugation to separate the coelomocyte types and to identify the types that express Sp185/333 genes before proceeding to enrichment for single-cell isolation (28, 34). Differential cell counts of the fractions indicated that the polygonal, discoidal, and small phagocytes were present in the 20 and 40% Percoll fractions (Table I). Coelomocytes from each fraction were analyzed for Sp185/333 proteins and the three types of phagocytes were positive, whereas the vibratile and red spherule cells were negative (Fig. 2). Although expression in the polygonal and small phagocytes is known (23), expression in the discoidal phagocytes has not been reported previously.

To ensure that the immune challenge with heat-killed Vibrio induced Sp185/333 gene expression and to verify that this was enhanced in phagocytes as previously reported for responses to LPS (23), cell fractions from three sea urchins were evaluated for Sp185/333 gene expression by qRT-PCR before and after challenge. The phagocytes not only expressed Sp185/333 genes, but the 40% Percoll fraction containing small and discoidal phagocytes showed increased expression 24 h after immune challenge with heat-killed Vibrio, and the polygonal and small phagocytes from the 20% Percoll fraction trended toward increased expression (Fig. 1C, 1D). Consequently, we employed the phagocytes enriched in the 20 and 40% Percoll fractions to analyze expression in single-cell samples.

To evaluate Sp185/333 gene expression in single phagocytes, cells from the top two Percoll fractions were diluted to an estimated 1 cell/μl (defined as 1×). To improve the likelihood that serial dilutions of cells resulted in samples containing one cell, 1× dilutions were diluted further by 2×, 4×, and 10×. Samples from each cell fraction and dilution set were first evaluated for SpL8 gene expression by two rounds of RT-PCR, which served as the positive control to evaluate whether a given sample had contained a cell and that cDNA had been produced. SpL8 encodes the sea urchin homolog of the ribosomal L8 protein (EST219; GenBank accession number R62029) (13). Of 214 single-cell samples that were analyzed from three sea urchins only 118 supported nested RT-PCR for SpL8, indicating that ∼55% of the samples contained at least one cell (Supplemental Table II). Poisson distribution analysis of the limiting dilutions indicated that samples diluted 4× and 10× were highly likely to contain single cells (Table II).

Samples that amplified SpL8 were next used in two rounds of RT-PCR with the external and internal primers for Sp185/333 sequences (Supplemental Table I). Because of the significant sequence variability among Sp185/333 cDNAs (18, 20), several partially overlapping internal primers were designed for the second round of PCR that would amplify as many of the known Sp185/333 cDNA sequences as possible. The internal primers were analyzed using a set of Sp185/333 cDNA templates of known sequence from the hundreds that are available (18, 20), which were chosen based on alignments of their 5′ and 3′ ends with the internal primers and the level of sequence matches or mismatches (Fig. 3). Results of the PCR showed that one to six pairs of internal primers could amplify individual Sp185/333 cDNA clones with varying intensity, which depended on the level of sequence match between each primer and each template in addition to the template concentration, the primer pair used in the reaction (Fig. 4), plus primer concentration, the number of PCR cycles, and the annealing temperature (not shown). Results ranged from multiple pairs of primers that generated amplicons of varying intensity for a particular template (clone aCF4-2441) to a single pair of primers that generated a single amplicon for a particular template (clone CG2-2404; Fig. 4). The two reverse primers were noted to amplify particular templates optimally (primer in3.1 amplified Lam6-2429; primer in3.2 amplified CG4-2404). In some cases, all forward primers amplified the same template (aCF4-2441), and in other cases a single forward primer amplified a particular template (primer in5.3 amplified Lam6-2429; Fig. 4). Different combinations of internal primers varied the production of amplicons from particular templates. However, amplicons of unexpected size from a given template were not observed, suggesting that priming from unexpected sites did not occur. These results indicated that multiple pairs of internal primers could variably amplify templates of known sequence, a result that was based on the similar but diverse sequences of the target cDNAs.

Next we evaluated the 118 samples that amplified SpL8 for amplification of Sp185/333 transcripts using four pairs of internal primers. We found that only 40 samples supported amplification with the internal Sp185/333 primers (Supplemental Table I), which indicated that ∼34% of the cell samples contained phagocytes that were expressing Sp185/333 genes at the time of collection. The remaining SpL8⁺ samples may have contained phagocytes, but only a subset of phagocytes are known to express the Sp185/333 genes at a given time (23). Nested PCR showed that 33 of the 40 single-cell samples had Sp185/333 amplicons of the same size, and 4 samples produced a single amplicon from a single pair of internal primers (Fig. 5). Of the remaining samples, seven showed amplicons of different sizes, although five of those were samples from the 1× dilution series with a higher probability of more than one cell per sample (Table II). In general, 82.5% of the single-cell samples produced Sp185/333 amplicons of the same size from the internal primers, suggesting that these cells may have contained Sp185/333 transcripts of identical sequence.

A possible source of artifact for our approach was genomic DNA (gDNA) contamination in the single-cell samples. The size range of the Sp185/333 mRNAs and the genes overlap even though the genes have a single intron of ∼400 nt. Consequently, Sp185/333 amplicons generated with the internal primers were evaluated in a subsequent PCR with LF and R2 primers (Supplemental Table I) that annealed to conserved regions surrounding the intron. The templates employed in the reactions were a subset of Sp185/333 nested amplicons from single-cell samples in addition to a cloned Sp185/333 gene (clone 2-073, GenBank accession number EF607742.1) (15), which was used as the positive control for gDNA contamination. The Sp185/333 nested amplicons from the single-cell samples showed a consistent fragment size of ∼120 nt, which was the expected size based on the positions of the annealing sites for the primers (Fig. 6). These were significantly smaller than the ∼520-nt amplicon from the cloned gene, which included the intron (see Supplemental Table I). This demonstrated that gDNA did not contaminate the RNA isolated from the single-cell samples that were used for nested RT-PCR and that only phagocyte cDNAs served as templates for the single-cell analysis.

Amplicons of the same size from different pairs of internal Sp185/333 primers for individual single-cell samples suggested that they could either be the same sequence or the same size with different sequences. Consequently, 32 amplicons from 11 representative samples from three sea urchins were chosen for sequencing. These included samples with one to four amplicons of the same size and other samples chosen specifically because they had amplicons of different sizes (Fig. 5; lanes with sequenced amplicons are indicated at the bottom of the gels). Templates from the first round of RT-PCR were reamplified using the same set of internal primer pairs, and the resulting amplicons were sequenced directly using F2 and/or R9 primers (Supplemental Table I), which annealed to conserved regions located between the internal primers. Most of the amplicons (29 of 32) returned an unambiguous sequence, which matched to Sp185/333 messages based on BLASTn searches of the nr database on GenBank (Supplemental Table III). Amplicons from three samples either returned ambiguous sequence or failed the sequencing reaction, likely due to technical reasons (Fig. 5, indicated with an X). The amplicon derived from single-cell sample B8 from animal 11, internal primer set 2 was shorter (Fig. 5), showed only 70–72% nucleic acid identity and 78–82% amino acid identity compared with the longer amplicons, and had a different element pattern (Fig. 7, Supplemental Table III). The amplicon derived from internal primer pair 4 for sample B8 returned ambiguous sequence, which was consistent with the presence of mixed templates originating from two different Sp185/333 messages to which the sequencing primer annealed (Fig. 5). The length of sequences ranged from 665 to 1216 nt and spanned a large part of the highly variable region encoded by the second exon. The deduced amino acid sequences that matched to Sp185/333 proteins by BLASTp searches were 275 to 385 aa in length (Supplemental Table III). When Sp185/333 amplicon sequences from 10 single-cell samples were compared within samples by global pairwise alignments, amplicons from nine samples showed 98–100% nucleotide identity and 94–100% amino acid identity (Fig. 7). Variations among amplicon sequences from within samples may have been the result of sequencing errors because these positions were mostly located in the terminal 10% of the sequencing passes. Ambiguous nucleotides at single positions were consistent with RNA editing that has been predicted for the Sp185/333 system (21). Editing of mRNAs derived from the same gene would appear as ambiguities at single positions, which were identified in a number of the sequences reported in the present study. Previous analyses have ruled out the possibilities of incorrect incorporation of nucleotides during ExTaq polymerization of Sp185/333 sequences (15, 20) and template switching that could result from shared sequence among Sp185/333 cDNAs or genes (see supplemental file 5 from Ref. 15). Consequently, we conclude that artifacts are not the source of sequence diversity reported in the present study.

A possible explanation of the near identity among the amplicons from most of the single-cell samples, besides multiple primer pairs amplifying the same target (as shown in Fig. 4), may have been mixing or contamination among the internal primers, which would result in identical amplicons. To ensure that amplicons were only generated by the primers employed in a given reaction, 24 chromatograms of unambiguous sequence and ≤1 kb were inspected for the annealing site of the internal primer. For the 19 amplicons that were sequenced with the F2 forward primer, the predicted internal primer site for either in3.1 or in3.2 was identified at the 3′ end of the sequence. Similarly, the six sequences that were generated with the R9 primer had the predicted annealing site for either internal primer in5.1, in5.2, or in5.3. This included cases in which both in5.1 and in5.2 were used in the reaction and both were identified as ambiguous chromatogram peaks at the expected positions where the primers differed (not shown). All of the 24 sequences had the expected internal primer annealing site at the 3′ end of the sequence based on the internal primers used to generate the amplicons. Annealing sites for primers that were not employed in the amplification reactions were not identified. This indicated that the amplicons from the single-cell samples were not the result of primer contamination.

The Sp185/333 genes and messages show variations in sequence, which is mostly based on the mosaic-like element patterns (15, 18, 20). BLASTp searches showed 93–100% identity to translated nucleotide sequences of known Sp185/333 element patterns (Supplemental Table III) (12, 15, 18, 20). The most abundant message type had the E2 element pattern (16 sequences from five single-cell samples from two animals), in agreement with previous results for E2 being the most common message type (20). Other element patterns included A1-like, A6, C5, G2 and G2-like, and O1 (Fig. 7, Supplemental Table III). Remarkably, these results indicated that many individual sea urchin phagocytes expressed Sp185/333 messages of highly similar or identical sequence that were the same length and had the same element pattern.

Because many of the sequences obtained from individual single-cell samples were highly similar, we evaluated whether this was also true for amplicon sequences that were generated from different single-cell samples both within and among animals. Global pairwise alignments of sequences from among single-cell samples showed that the nucleic acid identity was 67–100%, and the amino acid identity was 54–100%, corresponding to six different element patterns (Fig. 7). All sequences of >98% nucleotide identity shared the same element pattern, which was not correlated with the animal or sample from which the sequences were generated. For example, nucleotide sequences from different samples from animals 17 and 26 that displayed the E2 element pattern were 98–100% identical (Fig. 7). Alternatively, comparisons among sequences of different element patterns either within or among animals resulted in <72% nucleotide identity. These results indicated that sequences of shared element patterns were also of nearly identical sequence for the three sea urchins that were analyzed in this study. Similarity among sequences of the same element pattern was confirmed through alignment and phylogenetic analysis that showed amplicons of the same element pattern clustered into six separate clades within which the individual sequences were unresolved (not shown). The predominant expression of the E2 element pattern genes after immune challenge was in agreement with previous studies and was consistent with a core set of Sp185/333 genes that appear to be shared among all sea urchins (20).

The Sp185/333 gene family shows swift upregulation of expression in phagocytes responding to a variety of foreign challenges (11, 12, 20). The response appears to be essential for immune function and produces a highly diverse protein repertoire for protecting the host from the multitudes of potentially pathogenic microbes present in the marine environment (39, 40). It is striking that the expression patterns of the Sp185/333 gene family in single phagocytes from immune-challenged sea urchins show transcripts of a single sequence. This increases the complexity of the Sp185/333 system in the sea urchin in which sequence diversification of the immune response is to the advantage of the host over the pathogens (10). The diversification layers include 1) sequence variations among the Sp185/333 genes, 2) putative gene duplication, deletion, and recombination to increase sequence diversity, 3) evidence of mRNA editing that increases the sequence diversity of the transcripts from a given gene, and 4) likely posttranslational modifications to the proteins, all of which results in a significantly broader array of protein isoforms in the CF than is encoded in the genes (22). The results reported in the present study for Sp185/333 transcripts of a single sequence in single sea urchin phagocytes have not been identified previously in other invertebrates, and they add an intriguing new layer of complexity to this system in the sea urchin.

Early investigations of immunity concluded that invertebrates rely on simple, nonspecific recognition and response proteins that detect and combat whole classes of pathogens (41), a paradigm that was derived, in part, from the limited numbers of immune genes in Drosophila and the broad categories of pathogens that these insects detect and to which they respond (42, 43). However, significant complexity in many invertebrate innate immune systems is becoming the new paradigm, which in many cases is based on genome annotation (e.g., see Refs. 44, 45). Expanded immune response diversity in arthropods may be derived from the single copy DSCAM gene through regulated alternative splicing in hemocytes that produce many DSCAM isotypes (46, 47) that have important pathogen opsonization functions leading to augmented phagocytosis by hemocytes (48). Somatic DNA modifications that broaden the sequence diversity of the FREP gene family in freshwater snails generate diversity in the encoded proteins (6), which are involved in snail protection against digenean parasites and other pathogens (49, 50). Diversification of the FREP genes is thought to occur in individual hemocytes (51), although the expression patterns in single hemocytes are not known. The notion of a simple immune system in echinoderms changed with the annotation of the purple sea urchin genome and the discovery of many immune genes, including large and complex gene families such as those encoding TLRs and NOD-like receptors among others (45, 52, 53). However, little is known about the details of how this vast array of pathogen recognition receptors in sea urchins are expressed in coelomocytes, how they function in host protection, and whether they are connected to the Sp185/333 gene expression patterns in single phagocytes. These recent advances have shown that invertebrate immunity is not simple, is not homogeneous, and can be unexpectedly complex and sophisticated (53, 54).

The best understood cases of limited expression of immune genes in single cells include the Ig family genes in higher vertebrates that are assembled by somatic recombination (55) and the variable lymphocyte receptor genes in lower vertebrates that are assembled by a copy choice mechanism (reviewed in Ref. 56). The animal kingdom has very few examples of restricted expression of single members from multigene families in individual cells. The multiple clusters of olfactory receptor genes that encode G protein–coupled odorant receptors is one example that shows restricted expression from a single locus or a single allele in individual olfactory sensory neurons (reviewed in Ref. 57). Another example is the killer Ig-like receptor family that shows restricted expression in vertebrate NK cells (reviewed in Ref. 58). Perhaps the best example of a gene family with immune response function is the multigene IgM H chain family in nurse sharks that shows restricted expression in individual lymphocytes (59). Expression by members of these gene families is regulated by a wide variety of mechanisms, including epigenetic silencing and negative feedback to block expression, transcriptional regulation and complex promoters to restrict expression, chromatin remodeling by NFs, and short time windows for access by regulatory proteins to open euchromatin. These examples indicate that limiting the expression from large families of similar genes is likely accomplished with multiple mechanisms acting in coordination. Complex expression regulation may also be in play for the Sp185/333 gene family; however, it is not known how this might be accomplished in individual phagocytes in the sea urchin.

An alternative explanation for our results could be that multiple Sp185/333 genes are expressed at very different levels in phagocytes, and that the single gene with the highest expression produces the only cDNAs that are amplified by nested RT-PCR. We think that this possibility is unlikely for the following reason. When large batches of coelomocytes are evaluated for Sp185/333 gene expression, the most common element pattern encoded in the mRNAs is E2 or E2.1 (81%; 491 of 608 sequenced clones) (20). Similarly, Sp185/333 amplicon sequences of the E2 element pattern are the most common among the sequences reported in the present study. mRNAs encoding other element patterns from large batches of coelomocytes are identified much less often, such as the A type pattern that was found only once in 608 clones from sea urchins that where challenged with a variety of PAMPs (20). Consequently, if single cells express multiple genes at varying levels and the expression level of the A type element pattern is low, in agreement with previous results, this type of mRNA should be below detection by nested RT-PCR compared with other sequences. Based on the number of amplicons that were evaluated, we would expect that no single-cell samples would be detected with mRNAs encoding the A type element pattern. However, 3 of 20 single-cell samples have A type mRNAs, suggesting that these rare element patterns are expressed from single genes in individual cells at similar levels of expression as that from other genes. This argues against genes encoding rare element patterns being expressed at low levels among messages from multiple genes in single cells. Although our PCR-based approach to address the question of Sp185/333 gene expression in single phagocytes does not technically eliminate the possibility of very low levels of mRNAs expressed from another or multiple Sp185/333 genes, the sensitivity of nested PCR and the care required to eliminate amplification of contaminants suggest that low levels of other transcripts in single cells may not be present. If they are present, they may be functionally ineffective in supporting translation into enough additional Sp185/333 protein isoforms to have an immunological effect.

The innate immune system of the sea urchin must protect this species from a wide range of potential pathogens that are present in high concentrations in the marine environment (39, 60) during the life of the animal, which has maximum age estimates of ∼50 y (61, 62). The outcome of Sp185/333 gene expression is a broad array of Sp185/333 proteins that are produced by the phagocytes (22, 23). Based on the number of genes and numbers of protein isoforms that might be produced, it may be advantageous to the sea urchin to limit Sp185/333 protein production to only those isoforms with activities that can be effective against the pathogen or PAMP that has been detected. For example, preliminary results for a recombinant Sp185/333 protein show that it binds to Vibrio but not Bacillus, and to LPS but not peptidoglycan (C.M. Lun, C.S. Schrankel, H.Y. Chou, S. Sacchi, and L.C. Smith, unpublished observations). Depending on the pathogen detection mechanisms that function in phagocytes, only some of these cells may respond to a given challenge by activating Sp185/333 gene expression. This notion fits with our results showing that only 40 of 118 cell samples show Sp185/333 gene expression and that only some phagocytes produce Sp185/333 proteins (23, 29). Speculation on differential expression of the Sp185/333 genes in response to different PAMPs is based on a correlation with progressive changes in the size range of Sp185/333 mRNAs over the course of a response (20) and changes in the Sp185/333 protein repertoire in response to different challenges (22). Not only does this suggest that different isoforms may have different functions, but that their expression may be limited and tailored to the specific pathogen or PAMP that is detected (22) and that expression of a single Sp185/333 sequence per phagocyte may be only one aspect of regulating the response.

Limited gene expression and the presence of a single Sp185/333 mRNA sequence in individual phagocytes infers the production of a single Sp185/333 protein isoform per cell, given minor changes from possible mRNA editing (21) and putative posttranslational modifications. As an advantage, this may reduce or alleviate the tendency of the proteins to aggregate irreversibly while in transport vesicles. Aggregation has been noted when mixtures of native Sp185/333 proteins are isolated from samples of coelomocytes (22, 23). Aggregation tends to occur more slowly for isolates of a single recombinant Sp185/333 protein (23), and aggregation of the recombinant reduces activity (C.M. Lun, A. Boyd, S.D. Gillmor, and L.C. Smith, unpublished observations). Production of a limited number of Sp185/333 protein isoforms from individual phagocytes suggests that the proteins may function synergistically and combinatorially when mixed upon secretion into the CF and prior to aggregation, which may expand significantly the response capabilities of the Sp185/333 proteins. The Sp185/333 system in the sea urchin suggests diversity, flexibility, and efficiency for responding and adjusting to the wide variety of marine pathogens that are present in the habitat in which purple sea urchins have survived for millennia.

We are grateful to Preethi Golconda, Brian D’Allura, and Evelina Bertolotti for laboratory assistance, to Hung-Yen Chou for phagocyte images, to Dr. Haim Zlatokrilov for assistance with statistics, and to Drs. Sam Loker, Martin Flajnik, and Ioannis Eleftherianos for improvements to the manuscript.

The authors have no financial conflicts of interest.