Multitasking Immune Sp185/333 Protein, rSpTransformer-E1, and Its Recombinant Fragments Undergo Secondary Structural Transformation upon Binding Targets

This article is featured in In This Issue, p.2517

The California purple sea urchin, Strongylocentrotus purpuratus, has a complex, sophisticated, and robust innate immune system with a large number of immune gene families (1–5). The Sp185/333 gene family has an estimated ∼50 ± 10 genes per genome that encode a vastly diverse repertoire of putative immune response proteins (6–10). Each gene has two exons that encode the leader peptide and the mature protein (9–13). The deduced mature proteins have tandem and interspersed repeats, as well as subsets of 25–27 elements, which are recognizable blocks of sequence defined by large artificial gaps that optimize two possible alignments (6). The mosaic presence/absence of elements results in 51 known element patterns (Fig. 1A) (6, 10, 11). Sp185/333 proteins are expressed by the phagocyte subclass of coelomocytes (the echinoderm immune cells) that can be categorized into small, polygonal, and discoidal phagocytes based on cytoskeletal shape and cell size (7, 14–17). We have proposed previously that each phagocyte type may be subcategorized based on expression of the Sp185/333 genes and proteins and expression of the complement C3 homolog in subsets of these cell types (7, 18). The sizes of deduced Sp185/333 proteins range from 4 to 55 kDa based on cDNA sequences (19); however, the native proteins isolated from individual sea urchins show unexpectedly large sizes that contradict predictions (7, 13). Similar characteristics have been noted for the He185/333 genes and proteins from a different sea urchin species, Heliocidaris erythrogramma (20). Two-dimensional Western blots show up to 260 Sp185/333⁺ spots with a wide range of molecular masses and isoelectric points that illustrate the level of protein diversity among animals (13, 21). Despite the sequence diversity, the predicted Sp185/333 proteins share a common structure composed of an N-terminal hydrophobic leader peptide, a Gly-rich region with an arginine–glycine–aspartic acid motif, a His-rich region, and a C-terminal region (Fig. 1B) (11, 22). Bioinformatic analyses show no matches to proteins with known function, and thus there are no functional predictions based on amino acid sequence for the Sp185/333 proteins (7, 19).

The Sp185/333 proteins have been hypothesized to be antimicrobial with immune defense functions based on gene expression patterns and observations that suggest multiple levels of sequence diversity (reviewed in Ref. 10). First, the genes make up a family of tightly clustered genes that appear to undergo duplications and deletions (9, 23) plus some preliminary evidence of recombination within the tandem repeats in the second exon (Fig. 1A) (24). Second, there is significant sequence diversity among the genes and messages that is based on variations in element patterns and sequence diversity within elements (reviewed in Refs. 8, 25). Sequence diversity of the messages may be augmented by putative mRNA editing (26). Third, the Sp185/333 genes show strong upregulated expression in response to immune challenge with bacteria, LPS, peptidoglycan, β-1,3-glucan, and dsRNA (19, 27). Fourth, these multiple levels of diversification lead to highly diverse arrays of Sp185/333 proteins that vary among sea urchins and vary in response to immune challenge (13, 21). In testing the hypothesis that sequence diversity plus responses to immune challenge suggest antipathogen function, a recombinant Sp185/333 protein called rSp0032 binds specifically to the marine Gram-negative bacteria, Vibrio diazotrophicus, and to baker’s yeast, Saccharomyces cerevisiae, but it does not bind to the Gram-positive bacteria Bacillus subtilis or Bacillus cereus (22). rSp0032 also binds to multiple pathogen-associated molecular pattern (PAMPs), including LPS (from Escherichia coli), β-1,3-glucan (from S. cerevisiae), and flagellin (from Salmonella typhimurium), but it does not bind to peptidoglycan (from Bacillus) (22). The rGly-rich and rHis-rich fragments of rSp0032 (Fig. 1B) show altered or expanded binding characteristics toward bacteria and yeast compared with the full-length protein, whereas the C-terminal fragment of the Gly-rich region functions in the multimerization of rSp0032 (22). Bioinformatic predictions indicate that the mature rSp0032 minus the hydrophobic helical leader does not have a clear signature of secondary structure, is predominantly hydrophilic without any hydrophobic motifs to provide a solid structure to support protein folding, and may be an intrinsically disordered protein (IDP) (22). We have speculated that the lack of secondary structure may be a flexible platform from which rSp0032 may interact with multiple binding targets through conformational plasticity. We have suggested that rSp0032 has multitasking binding abilities and may function initially through noncovalent electrostatic interactions based on charged differences between protein and binding targets, which is followed by secondary interactions based on changes in the structural conformation of the protein resulting in specific and firm binding of rSp0032 to several targets (22). For comparison, p53 is a well understood example of a protein with an intrinsically disordered region (IDR) that engages in interactions with as many as 20 different protein partners, integrates multiple signaling networks, and functions in a wide range of cellular activities (28). Similarly, nucleoporins, which are the gatekeepers of the nuclear pore complex, have phenylalanine glycine repeats that constitute IDRs and form a meshwork of random coil structure to restrict nuclear transport for a wide range of molecules (29, 30). These examples illustrate the notion that IDPs and proteins with IDRs typically interact with many binding partners. However, there are very few reports of IDPs that function in vertebrate immunology (31).

To test whether the bioinformatic predictions that the disordered structure of rSp0032 would change conformation in response to binding targets, circular dichroism (CD) was used to evaluate the secondary structure and conformational folding properties in solution. CD spectroscopy detects differences in absorption of circularly polarized light by the backbone amide/peptide bonds in the region of 240 nm and below (32) and can be used to determine the functional range of ligand concentrations that lead to changes in protein structure, in addition to the level and speed at which the changes take place (33). CD is often used to distinguish between disordered (random coil) and ordered (α helix or β strand) protein structures (34). Results from CD confirm the bioinformatic prediction of structural conformational plasticity of rSp0032 (22) and its rGly-rich and rHis-rich fragments over a short range of temperatures. The recombinant C-terminal fragment of the Gly-rich region could not be evaluated by CD because it multimerizes upon isolation (22). Furthermore, rSp0032 transforms from disordered to α helical structure upon the addition of SDS or 2,2,2-trifluoroethanol (TFE) to the buffer, or in the presence of LPS. SDS is an anionic detergent commonly used to simulate the interactions between anionic lipid membranes and amphipathic cationic peptides, as well as conformational changes that the peptides may undergo as a result of these interactions (35). TFE is often used as a cosolvent with peptides or proteins in solution to promote secondary structures of α helices and β strands (36–38). LPS, an endotoxin from Gram-negative bacteria, is bound specifically and tightly by rSp0032 (22). The rGly-rich and rHis-rich fragments show unexpected structural flexibility and transform to secondary structure of either enhanced α helical structure or β strand depending on the environment. These results suggest significant conformational flexibility of both the full-length and recombinant fragments that may be a basis to explain the range of molecules to which rSp0032 binds with high affinity and specificity. Consequently, we propose to change the protein name from rSp0032 to rSpTransformer-E1 (rSpTrf-E1) to reflect the conformational transformation plus the E1 element pattern to distinguish it from other versions in this family (Fig. 1A). A selected set of Sp185/333 proteins representing the diverse sequence variations within this immune protein family are also predicted to be IDPs, and if the results for rSpTrf-E1 can be applied to the Sp185/333 proteins in general, many or most may also show similar structural conformational changes that may be induced by foreign cells or PAMPs. Therefore, we propose to rename the family of Sp185/333 proteins to SpTransformer proteins (SpTrf).

Protein folding predictions for several deduced SpTrf proteins, rSpTrf-E1 (GenBank accession no. DQ183168), and the recombinant fragments from SpTrf-E1 were based on the amino acid sequences and conducted using the DisMeta server (http://www-nmr.cabm.rutgers.edu/bioinformatics/disorder/) (39).

Vector construction, expression in E. coli, and purification of rSpTrf-E1 and the rGly-rich and rHis-rich fragments using bacterial expression vectors were conducted as described (22). rSpTrf-E1 and the rGly-rich and rHis-rich fragments were evaluated by SDS-PAGE stained with Bio-Safe Coomassie (Bio-Rad Laboratories) and by Western blot with anti-SpTrf Abs (formerly anti-Sp185/333 Abs) to identify elution fractions for further protein isolation (22). Protein concentration was quantified by monitoring the absorbance at 205 nm by the Scopes (40) method using a NanoDrop 2000 spectrophotometer (Thermo Fisher Scientific).

CD spectra of recombinant proteins were obtained using a J-1500 spectropolarimeter (Jasco Analytical Instruments). Samples were allowed to equilibrate at least 10 min at room temperature before spectra were collected. Samples were placed in a 1-mm quartz cell (Jasco Analytical Instruments) and evaluated using a 190–260 nm measurement range with 50 nm/min scanning speed, 1-nm bandwidth, and 8-s response time with 1.0-nm data pitch with five scans. Background baseline for CD spectra was corrected with 10 mM sodium phosphate (pH 7.4) containing the same concentrations of SDS, TFE, and LPS in the absence of the proteins. Boxcar smoothing was used to remove signal noise as implemented by convolving the raw input data with a box-shaped pulse of ΣM + Mi + Mi/3, where M stands for the mean. CD spectra were used to calculate the mean residue ellipticity, or θ, with standard units of degrees (deg) × cm² × dmol⁻¹. The helical character was calculated using the ellipticity ratio (R = θ₂₂₂/θ₂₀₇) between the two negative peaks at 222 nm (n → π* electronic transition of carbonyl compound) and 207 nm (parallel component of the split π → π* electron transition of the protein chromophore) (32, 38). For a protein of known concentration and amino acid sequence that is 100% α helical, the ellipticity ratio is R = 1 and this is interpreted as 3.6 residues per turn positioned at an angle of 100° between sequential amino acids around the helical axis. The ellipticity ratio for a 3₁₀ helix is R = 0.4 and has 3 aa per turn that are positioned at an angle of 120° between sequential amino acids (41). The 3₁₀ helix is narrower and longer than the α helix for a given number of amino acids. The percentages of α helix and β strand were deconvoluted from the CD spectra with the DichroWeb online server using several analysis programs, including CONTINLL, SELCON3, CDSSTR, VARSLC, and K2D (http://dichroweb.cryst.bbk.ac.uk/html/home.shtml) (42, 43). Two additional programs were employed to verify the percentage of secondary structures: the CDNN program uses a neural network to access reference spectra for comparisons and analyses (http://gerald-boehm.de/download/cdnn) (44), and the K2D3 Web server uses a database of theoretical spectra derived with DichroCalc (45). Protein concentrations of 0.25 μM rSpTrf-E1, 9.07 μM rGly-rich fragment, and 2.5 μM rHis-rich fragment were selected to ensure that absorbance was within the optimal range for the instrument at the chosen wavelength range. The effect of temperature on protein secondary structure was determined using CD at temperatures ranging from 4 to 44°C with steps of 5°C.

Bioinformatic analysis of rSpTrf-E1 predicted that it is highly hydrophilic and likely to be intrinsically disordered (22), which is in agreement with analyses of other SpTrf protein variants composed of various element patterns (Supplemental Fig. 1). The element patterns that were chosen for analysis included a range of full-length deduced SpTrf proteins that represent the breadth of diversity within the protein family. These included element patterns defined as A6, D1, B8, C1, E1 (slightly different sequence from rSpTrf-E1), E2, and 01 plus two truncated variants, E2.1 and 01.2. The E2.1 element pattern has a glycine codon altered to a premature stop codon, whereas the 01.2 version has a frameshift that leads to missense sequence and an early stop codon (19). Both likely result from mRNA editing because early stop codons and frameshifts are not present in any gene (26). Secondary structural prediction using PSIPRED revealed that both full-length and truncated proteins of all element patterns contained an N-terminal leader that was predicted to be α helical, and that proteins of all element patterns except E1 and 01.2 had regions at the C terminus with a low probabilities of being α helical (Supplemental Fig. 1). Similarly, the DisMeta analysis also indicated that all full-length and truncated protein variants had an α helical, hydrophobic leader, even though the leaders had slightly different sequences, and that the mature proteins (minus the leader), except for the 01.2 element pattern, were predicted to be intrinsically disordered with poorly defined secondary structure. Unlike the other SpTrf sequences, a high probability of secondary structure was predicted for the missense sequence in the 01.2 element pattern (Supplemental Fig. 1). The C-terminal regions of all the SpTrf proteins showed some probability of secondary structures except for the E1 element pattern, which was predicted to be entirely intrinsically disordered. The bioinformatic prediction for intrinsic disorder also indicated that rSpTrf-E1 was likely to be entirely disordered without any secondary structure (Fig 2A). The rGly-rich and rHis-rich fragments of rSpTrf-E1 were also predicted to be intrinsically disordered except for a low probability of α helical structure for two short N-terminal regions in the rGly-rich fragment based on the PSIPRED prediction (Fig. 2B, 2C). rSpTrf-E1 and the recombinant fragments evaluated in this study did not include the hydrophobic leader but they did include an N-terminal 6 His tag (6-His-tag) and a factor Xa cleavage site that were required for isolation by nickel affinity chromatography. The 6-His-tag was not cleaved prior to evaluation because its presence or absence had no discernable effect on target binding function (22). Bioinformatic predictions of secondary structure showed that the 6-His-tag did not change the low probability predictions of secondary structure (Supplemental Fig. 2).

CD was employed to assess experimentally the validity of the bioinformatic predictions for rSpTrf-E1 and the recombinant fragments. Alternative approaches to determine structure were not used because high concentrations of rSpTrf-E1 that would be required for nuclear magnetic resonance or x-ray crystallography would drive multimerization of the protein, which has been repeatedly noted for native SpTrf proteins (13, 21) and has been documented for rSpTrf-E1 (reported previously as Sp0032) as multimers (7) and dimers (22). The concentration of 0.25 μM rSpTrf-E1 in phosphate buffer was determined as optimal for CD (Supplemental Fig. 3A), which showed spectra with a negative peak at 199 nm indicating predominantly random coil or intrinsic disorder (see 0 mM SDS; Fig. 3A, Table I), which was consistent with our prediction (Fig. 2A; see Ref. 22). Analysis of the CD spectrum indicated that rSpTrf-E1 was ∼1.6% α helical (Fig. 3B), which was in agreement with the bioinformatic prediction. Optimization of rGly-rich and rHis-rich fragment concentrations in phosphate buffer for CD analysis (Supplemental Fig. 3B, 3C) resulted in spectra indicating that they were ∼15 and 30% α helical, respectively (see 0 mM SDS in Fig. 3C–F, Table I). These CD results confirmed that rSpTrf-E1 was an IDP; however, the α helical content observed for the rGly-rich and rHis-rich fragments did not agree with the predictions from bioinformatic analyses (Fig. 2B, 2C).

To determine whether rSpTrf-E1 and the recombinant fragments altered their secondary structure with respect to changes in temperature, CD spectra were collected over a temperature span of 4–44°C using a thermoregulated cuvette. This range was chosen because purple sea urchins live along the near-shore Pacific coastline of North America from Mexico to Alaska (46) in temperatures that range from 5 to 23°C with mortality noted at temperatures >24°C (47). CD results showed that the protein structures in phosphate buffer remained unchanged with respect to temperature (Supplemental Fig. 3D–F). Consequently, all subsequent analyses were carried out at room temperature.

SDS is a common anionic surfactant that emulates the amphipathic properties of anionic lipid environments and is commonly used for assessing the secondary structure of membrane-active peptides and proteins under such conditions (32). When SDS was added to rSpTrf-E1, increasing concentrations altered the CD spectra, indicating a change from mostly disordered toward an α helical character with a positive peak at 193 nm, a negative peak at 206.8 nm, and at weak shoulder at 222 nm (Fig. 3A). The structural transition of rSpTrf-E1 from disordered to α helical was immediate upon the addition of 10 mM SDS, and the CD spectra intensified slightly and stabilized between 20 and 30 mM SDS (Fig. 3A, 3B). Furthermore, the ellipticity ratio of R = 0.59 in 20–50 mM SDS (Table I) observed in the CD spectra indicated unique secondary structural contributions that deviate from canonical α helices. This result suggested that the rSpTrf-E1 was not a fully classical α helix in SDS (see 2Materials and Methods) (32, 38). Overall, the observed conformational changes in SpTrf-E1 as the concentration of SDS was increased from 0 to 30 mM indicated that it transformed from disordered to 79% α helical structure (Fig. 3B, Table I). These analyses, however, did not specify the region(s) of the protein that transformed to an α helical character, which was beyond the scope of this analysis.

When the rGly-rich fragment was evaluated in the presence of increasing concentrations of SDS, the α helical signal in the CD spectra intensified with a positive peak at 193 nm and two negative peaks: a major negative peak at 207.6 nm, and a weak shoulder at 221.4 nm (Fig. 3C). This resulted in an ellipticity ratio of R = 0.78 in 40 mM SDS (Table I), which was higher than the R value for rSpTrf-E1 in SDS, suggesting a closer resemblance to α helical structural properties. As the SDS concentration was increased from 0 to 30 mM, the percentage of α helical content in the rGly-rich fragment increased from ∼15 to 75% (Fig. 3D). CD spectra for the rHis-rich fragment were similar, showing an α helical structure with a strong positive peak at 196.4 nm and a strong negative peak at 207.4 nm with a weak shoulder at 221.4 nm (Fig. 3E). The calculated ellipticity ratio of R = 0.76 (Table I) was very similar to the R value of the rGly-rich fragment, suggesting a similar α helical structure. The percentage of α helical content of the rHis-rich fragment increased from ∼30 to ∼70% when SDS was increased from 0 to 10 mM (Fig. 3F). Overall, these results demonstrated that rSpTrf-E1 transformed from intrinsically disordered to α helical in the presence of SDS, whereas the rGly-rich and rHis-rich fragments had unexpected levels of α helical structure in the absence of SDS. Furthermore, as the concentration of SDS increased, the percentage of α helical content also increased for all three recombinant proteins. In all cases, rSpTrf-E1 and the rGly-rich and rHis-rich fragments appeared to be flexible and to alter their secondary structures in this anionic amphipathic environment.

TFE is employed in CD studies because it tends to stabilize and promote α helical conformation in proteins and peptides (32). When TFE was added to rSpTrf-E1, the CD spectra indicated a structural change from intrinsically disordered with a negative peak at 200 nm to α helical showing characteristic negative peaks at 222.4 and 206.8 nm at 10% TFE (Fig. 4A). Spectra of the protein in 20% TFE shifted the negative peak at 206.8 to 207 nm, and the spectra stabilized at >20% TFE with two prominent negative peaks at 222.4 and 207.6 nm (Fig. 4A). Deconvolution of the CD spectra suggested that the α helical character of rSpTrf-E1 increased from 1 to 95% as the concentration of TFE increased. With an ellipticity ratio of R = 1 (Table I), the structure of rSpTrf-E1 appeared to have shifted to a classical α helical structure in 50% TFE (Fig. 4B). The rGly-rich fragment gave very different results in increasing concentrations of TFE. In phosphate buffer the rGly-rich fragment was ∼15% α helical, but as TFE concentration increased from 20 to 40%, the observed CD spectra were consistent with a shift to β strand structure, which was evident from the intensified positive peak at 195 nm and a broad negative peak at 224 nm (Fig. 4C). This conformational shift in the presence of increasing concentrations of TFE reduced the α helical structure to ∼3% (Fig. 4D). The changes in percentage of β strand structure could not be calculated using DichroWeb because the unusual and rare peak at 224 nm did not correlate with a spectrum of any known protein. Thus, in the presence of TFE, which is generally known to stabilize α helical structures, the rGly-rich fragment underwent a conformational shift from α helical to predominantly β strand (Fig. 4C). Similarly, the CD spectra of the rHis-rich fragment also shifted from α helical to β strand with increasing concentrations of TFE (Fig. 4E). CD spectra showed a single negative peak at 214.2 nm that appeared at 10% TFE and intensified at 30% TFE, indicating the shift to a β strand conformation. In the absence of TFE, the rHis-rich fragment was ∼30% α helical, which decreased to ∼3% in TFE concentrations of ≥10% (Fig. 4F). The decrease in the α helical structure corresponded with an increase in a β strand structure of 33–46% as TFE increased from 10 to 30%. Overall, in the presence of TFE, the structure of rSpTrf-E1 shifted from disordered to α helical, whereas TFE induced the recombinant fragments to shift from α helical to predominantly β strand. These significant differences indicated that when the regions of rSpTrf-E1 were separated and expressed as recombinant fragments, they had very different and unexpected properties relative to the properties of the full-length protein.

The multitasking binding of rSpTrf-E1 toward a few PAMPs in conjunction with the bioinformatic prediction and CD evidence that it was an IDP led to the speculation that the protein might undergo conformational transformations in the presence of different binding targets (22). Therefore, to determine whether rSpTrf-E1 underwent conformational changes upon binding PAMPs, it was evaluated by CD in the presence of LPS, which was selected in part because LPS forms micelles and does not have a significant impact on observed CD spectra, unlike flagellin. In the presence of LPS, rSpTrf-E1 transformed immediately from intrinsically disordered to α helical regardless of the LPS concentration (Fig. 5A). CD spectra intensified and then stabilized as the concentration of LPS increased (0.5–5%), exhibiting a positive peak at 193 nm and two negative peaks of which the stronger was at 207 nm plus a weak negative shoulder at 222 nm (Fig. 5A), which yielded an ellipticity ratio of R = 0.66 (Table I). rSpTrf-E1 transformed from disordered (1% α helical) in the absence of LPS to a maximum of 78% α helical structure in 5% LPS, although most of the transformation occurred at 0.5% LPS, which was the lowest concentration tested (Fig. 5B). The conformational change was similar to that for rSpTrf-E1 in the presence of SDS and TFE.

The conformational properties of the recombinant fragments were also evaluated in the presence of LPS. CD spectra for the rGly-rich fragment shifted from an α helical structure toward β strand with increasing concentrations of LPS (Fig. 5C). Starting at 12.5% LPS, the spectra exhibited a negative peak at 212 nm that intensified and stabilized as the concentration of LPS increased to 32.5%. These changes corresponded with 27.5–33% α helical structure in phosphate buffer or in 2.5% LPS, respectively. In LPS concentrations of ≥12.5%, the α helical content of the rGly-rich fragment decreased from 15 to 3% (Fig. 5D). This change correlated with a structural shift of 17–78% β strand in the rGly-rich fragment as the LPS concentration increased. Unlike the rGly-rich fragment, the rHis-rich fragment remained α helical in the presence of LPS (Fig. 5E). CD spectra indicated a strong positive peak at 193 nm with two negative peaks, the stronger of which was at 207.6 nm plus a weaker negative shoulder at 221.2 nm (Fig. 5E). This indicated that the rHis-rich fragment was 30% α helical in the absence of LPS, which increased to 62% α helical in 0.5% LPS and stabilized at >65% α helical in ≥1.25% LPS (Fig. 5F). The rHis-rich fragment ellipticity ratio of R = 0.78 (Table I) was comparable to the ratio in SDS, suggesting that this fragment shared a similar helical structure in the anionic environment with the helical structure predicted for rSpTrf-E1. Overall, rSpTrf-E1 and recombinant fragments reacted quite differently in the presence of LPS. rSpTrf-E1 transformed from an IDP to α helical structure, and the rHis-rich fragment was somewhat similar as it increased its α helical content, whereas the rGly-rich fragment responded by transforming from α helical to β strand. Furthermore, the concentration of LPS at which conformational changes were observed was much higher for the rGly-rich fragment compared with rSpTrf-E1 and the rHis-rich fragment. The major difference between the rGly-rich fragment compared with rSpTrf-E1 and the rHis-rich fragment is the histidine content. There are 27 histidines in the His-rich region of the protein, whereas there are no histidines in the Gly-rich region (see supplemental table III in Ref. 22). No other amino acid has the same distribution and may be an aspect of the differences in interactions with LPS among the recombinant proteins.

The multitasking binding ability of rSpTrf-E1 toward multiple targets plus its bioinformatic prediction as an IDP led to the speculation that this protein may function through conformational plasticity (22). Indeed, we confirm that in the absence of a binding target, rSpTrf-E1 is an IDP based on CD spectra. In the presence of SDS, TFE, or LPS it shows structural flexibility by shifting from intrinsic disorder to α helical, thus supporting predictions for its structural transformation capabilities. The rHis-rich fragment is similar to rSpTrf-E1 in its conformational transformation to α helical structure in SDS and LPS, but switches to β strand in TFE. In contrast, the rGly-rich fragment intensifies the α helical structure in the presence of SDS but switches from α helical to β strand in TFE and LPS. Although rSpTrf-E1 is a rare element pattern of the deduced SpTrf protein sequences [the E1 element pattern was identified in 3 of 688 cDNA sequences (11, 19) and 1 of 171 gene sequences (6)], it provides novel predictions that many of the SpTrf protein variants are antimicrobial IDPs that may also function through conformational plasticity. This is based on bioinformatic predictions for a wide range of deduced amino acid sequences for the SpTrf proteins, including the most common element pattern, E2, suggesting that all may be IDPs (Supplemental Fig. 1; D.P. Terwilliger and L.C. Smith, unpublished observations) and we hypothesize that they may show different secondary structural changes in the presence of a wide range of binding targets.

We have hypothesized that the initial step in rSpTrf-E1 binding to multiple targets involves noncovalent electrostatic interactions that is followed by a second phase of binding based on conformational plasticity by transformation from intrinsic disorder to ordered folds (22). IDPs have distinctive characteristics that are unlike orderly folded proteins. The disordered nature of IDPs generates structural flexibility allowing conformational plasticity to create a range of complexes that provide extraordinary advantages for interactions with multiple binding targets (48, 49). Binding to multiple targets by IDPs is known as “one-to-many” interactions, with the potential for an IDP to undergo a variety of independent folding transitions that generate multidimensional complexes upon binding to different specific targets. The possible mechanism by which this occurs may be related to the amino acid sequence of the IDP, which may shift into α helices, β strands and sheets, or irregular structures upon interacting with and binding to a specific target (28).

The multitasking binding nature of IDPs represents a type of binding interaction that is not expected to occur among the classical molecular recognition mechanisms that function for folded proteins. Rather than the lock and key (50) or the induced fit (51) theories of how ordered proteins bind to specific targets, additional models have been proposed to describe the molecular folding mechanisms of IDP conformational plasticity. One of these models is the “polyelectrostatic” model of dynamic complexes (49, 52–54) that may be applicable to rSpTrf-E1. This model suggests a cumulative electrostatic interaction of charges in the IDP when in the presence of the binding target (52, 54). Initial interactions result in a highly disordered complex in which the positively charged amino acids of the IDP interact weakly with the negatively charged amino acids (or other groups) in the binding target. These transient but cumulative electrostatic interactions may contribute to strong binding affinity, which may create an electrostatic field that can drive structural changes in the IDP (55). This model may apply to rSpTrf-E1 because the amino acid sequence is rich in charged and polar amino acids (22), suggesting that a polyelectrostatic mechanism may act in initial interactions with Gram-negative bacteria and LPS. LPS has anionic phosphates on both the glucosamine sugar in lipid A and the polysaccharide core (56), which may be the basis for forming charge-based interactions or hydrogen bonds with the histidines in rSpTrf-E1 (22). These types of electrostatic interactions may constitute an initial step in the interactions between rSpTrf-E1 and a variety of targets with anionic groups.

Conformational changes in proteins have important biological functions that are essential to life. They are involved in cellular signaling pathways, including development, growth control, sensory signal mediation, pathogen infection, and host immune systems (48, 55, 57). IDPs undergo significant changes in structure upon receiving an input signal, such as ligand binding, environmental changes, or chemical modifications (57). A relevant example of well-understood IDPs is cecropins, which are small cationic antimicrobial peptides that display broad antibacterial activity (58). Cecropins are produced by the silk moth, Hyalophora cecropia, and are intrinsically disordered and inactive in noninfected hosts. However, upon contact with negatively charged microbial membranes, they transform into two short amphipathic α helices linked by a short hinge region, and they associate with membrane surfaces on microbes. Multiple amphipathic cecropins accumulate, aggregate into a “carpet” that inserts into the hydrophobic region of the membrane, and generate partially selective ion channels. This action causes membrane permeabilization and leakage of microbial cell contents that leads to killing (58, 59). An example of a protein with an IDR is the influenza hemagglutinin (HA) that mediates interactions between the viral membrane and the host cell membrane (60). HA anchors the virus to the cellular receptor and, once docked, induces cellular endocytosis into the cell. The reduction in pH within the endosome leads to proton binding to HA that triggers a conformational change to alter the random coil structure of the IDR to α helices that promote viral fusion with the target cell membrane. Although these are well-understood examples of IDP and IDR activities, they are also examples of how disordered regions of proteins are employed on both sides of the arms race: in the defense mounted by the host immune system and in pathogen virulence.

Despite the differences in sequence and functional activities of rSpTrf-E1 compared with cecropins and HA, rSpTrf-E1 shows similar flexible structural transformation upon binding to a range of targets, particularly with regard to the formation of α helices. We have speculated that the second step in rSpTrf-E1 binding is protein conformational transformation (22), and we show in this study that rSpTrf-E1 transforms from an IDP to α helical, which may be the underlying basis for the very strong affinity and specific binding to microbes and PAMPs that have vast structural differences. This type of target binding is partially reminiscent of the template theory of Ab formation (61) in which an Ab with two extended and unstable (or disordered) regions are folded and stabilized into shapes that are complementary to the Ag with which it is in contact and to which it can subsequently bind after dissociation and entrance into circulation. The aspect of Pauling’s theory that applies to the sea urchin SpTrf system is that rSpTrf-E1 is unfolded but interactions with a foreign target initiate binding and folding based on the characteristics of the foreign target. It is interesting to consider that Pauling may have been correct in one aspect of his speculation for generating diversity in some Ag-binding proteins, but that it did not apply to vertebrate antibodies.

When separated, the recombinant fragments transform differently compared with rSpTrf-E1 and relative to their bioinformatic predictions. These differences may be biologically relevant, particularly for the rGly-rich fragment, which represents a common truncated SpTrf protein variant (19) that is expressed at higher levels in sea urchins prior to immune challenge compared with the full-length SpTrf proteins that tend to be more highly expressed after immune challenge (13, 19, 21). Most truncated SpTrf proteins are essentially composed of the Gly-rich region and may serve a surveillance function in sea urchins prior to immune challenge (22). This speculation is also based on a broadened binding activity by the rGly-rich fragment to Bacillus sp. to which rSpTrf-E1 does not bind and correlates with only partial competition by rSpTrf-E1 with the rGly-rich fragment for binding to yeast (22). The CD spectra results for the rGly-rich fragment indicate that it either intensifies the α helical structure or shifts from α helical to β strand structure depending on the polar solution environment and the biochemical characteristics of the target. This result, if applied to the naturally occurring truncated SpTrf proteins, suggests that many may also shift into different secondary structural conformations compared with the full-length proteins. Alternatively, the rHis-rich fragment, which displays similar structural flexibility as rSpTrf-E1, intensifies its α helical structure in the presence of LPS and is fully competed by rSpTrf-E1 for binding to yeast (22), suggesting identical binding mechanisms. Whether the structural transformations of the recombinant fragments and rSpTrf-E1 as deduced from CD spectra are directly related to the binding activities of these proteins in vivo remains speculative. However, the more specific binding by rSpTrf-E1 to bacteria as reported by Lun et al. (22) indicates that the Gly-rich and His-rich regions in the full-length protein may interact with each other to impart this restricted range of bacterial binding.

In conclusion, purple sea urchins have a surprisingly long life span of ∼50 y (62) and depend only on innate immunity for protection against a wide array of marine pathogens. That echinoderms may only have innate immunity is based on allograft rejections that do not show accelerated rejection of second set allografts (63) and the failure to find accelerated clearance of bacteria upon second challenge (64) among other reports (reviewed in Ref. 63). Annotation of the sea urchin genome sequence reveals a number of expanded innate immune gene families with homologs in vertebrates such as TLRs, NOD-like receptors, scavenger receptors with cysteine-rich domains, small C-type lectins, and IL-17–like cytokines, all of which likely underlie at least some of the immune activities that result in successful adversarial interactions with multitudes of potential marine pathogens (1–3, 5, 65). Homologs of vertebrate genes suggesting adaptive functions have not been identified in the sea urchin genome sequence. The sequence diversity and upregulated expression by the SpTrf gene family (formerly called the Sp185/333 gene family) in response to immune challenges indicate its importance in the innate immune system of the purple sea urchin (reviewed in Refs. 8, 10). Multiple levels of diversification mechanisms that perhaps maximize the antipathogen activities of this protein family have been suggested, including putative gene diversification, transcript editing, and posttranslational modifications that generate a broad array of SpTrf proteins both within and among sea urchins (reviewed in Refs. 10, 25). The multitasking binding ability (22) and verification of rSpTrf-E1 as an IDP provide an unusual perspective for how this protein family may function as immune effectors for at least opsonizing and perhaps killing pathogens. The specificity of rSpTrf-E1 binding to certain bacterial species and to some PAMPs suggests that other rSpTrf proteins with different element patterns may have different binding specificities, which raises the question of whether a detailed analysis of expression patterns for individual SpTrf versions may reveal a level of immune response specificity similar to that reported for DSCAM in some arthropods (reviewed in Ref. 66). The deduced SpTrf protein variants are predicted bioinformatically to be IDPs, and many (or all) may have multitasking binding activity based on structural transformation that may be more or less similar to rSpTrf-E1. Individual sea urchins are known to express many SpTrf isoforms (13, 21) and if each variant has overlapping but distinct multitasking binding activities that can target simultaneously multiple PAMPs, these proteins may provide purple sea urchins with a highly capable and robust immune response for protection against the broad array of potential marine pathogens.

We are grateful to Anika Armstrong (George Mason University) for assistance with CD, and to Hung-Yen Chou for general laboratory assistance.

The authors have no financial conflicts of interest.