The third lysozyme of Crassostrea virginica, cv-lysozyme 3, possesses a number of characteristics that mark it as an intermediary between two previously reported i-type lysozymes from this species. Cv-lysozyme 3 has a molecular mass and distinctive N-terminal amino acid sequence similar to cv-lysozyme 1, the lysozyme involved in the host defense [56, 58]. Its chemical characteristics and major sites of gene expression, however, are more like those of cv-lysozyme 2, a lysozyme that functions in digestion . In addition, the optimal muramidase activity of cv-lysozyme 3 was detected under pH and ionic strength conditions that favor the activity of neither cv-lysozyme 1 nor c-lysozyme 2. Consistent with this biochemical and expression characterization, our phylogenetic tree also placed the clade including cv-lysozyme 3 as the sister to a clade that includes cv-lysozyme 2. Moreover, an episode of positive selection was associated with the transition from defensive to digestive function.
Cv-lysozyme 3 shared properties with both cv-lysozyme 1 and cv-lysozyme 2 (Table 1). Specifically, cv-lysozyme 3 was composed of a similar number of amino acid residues and had a similar molecular mass as cv-lysozyme 1. It, along with its apparent ortholog cg-lysozyme 4, also shared a unique N-terminal domain with cv-lysozyme 1 that is absent in cv-lysozyme 2 and all other bivalve i-type lysozymes (Figure 5). On the other hand, the theoretical isoelectric point (pI), arginine residue number, and protease cutting sites in the amino acid sequence of cv-lysozyme 3 were closer to those of cv-lysozyme 2 [56–58]. Moreover, cv-lysozyme 3 and cv-lysozyme 2 are both expressed primarily in the digestive glands (Figure 4; ) whereas cv-lysozyme 3 minor expression sites overlapped with that of cv-lysozyme 1 (Table 1; ). These characteristics of cv-lysozyme 3 are consistent with the intermediate status of the new lysozyme between the previously reported cv-lysozyme 1 and cv-lysozyme 2.
The topology of our phylogenetic analysis is also consistent with the transitional status of cv-lysozyme 3 (Figure 6): cv-lysozyme 3 and its C. gigas ortholog branch off after the defensive cv-lysozyme 1 and as sister to a clade that includes the digestive cv-lysozyme 2. This suggests that the split between defensive lysozymes, like cv-lysozyme 1, and other lysozymes, including the digestive cv-lysozyme 2 and the potentially transitional cv-lysozyme 3, preceded the split of mytiloid and ostreid bivalves, a view consistent with taxonomic views that place these families (along with Chlamys of the Pectinidae) in the Pteriomorpha . The sister group relationship of the digestive lysozymes (the clade containing cv-lysozyme 2 and the other digestive lysozymes, cv2C, of Figure 6) and the potentially transitional cv-lysozyme 3/cg-lysozyme 4 clade is consistent with an evolutionary history in which the latter are transitional forms that have been lost (or not yet found) in lineages outside of Crassostrea. An alternative in which cv-lysozyme 3 and cg-lysozyme 4 are not transitional forms but have instead arisen within the lineage leading to Crassostrea is not supported by the topology of phylogenetic tree in Figure 6, which places digestive lysozymes from Mytilus and Chlamys as closer to the digestive lysozymes from Crassostrea than the cv3/cg4 clade. However, the branch leading to the cv3/cg4 clade is not a long one, as would be expected for an early origin, so the possibility of a more recent genesis within Crassostrea cannot be entirely excluded. Finding a cv3 ortholog within mytiliods would resolve this issue. Expression levels of cv-lysozyme 3 appear to be low, suggesting that orthologs to this lysozyme may not yet have been found due to the rarity of their transcripts. We purified 18.6 mg of cv-lysozyme 1 from the first HIC peak in this research (not shown). Similarly, 2.2 mg of cv-lysozyme 2 can be purified from 5.5 g of crystalline style total proteins . These qualitative data suggest that cv-lysozyme 3 expression level is far lower than the two other C. virginica lysozymes, such that the transcipts for this transitional form are rare. We predict that deeper sequencing of other bivalves with digestive lysozymes will reveal orthologs to cv-lysozyme 3. Alternatively, orthologs to cv-lysozyme 3 may have gone extinct in some bivalve lineages that once possessed them, including Mytilus.
Results from studies on c-type lysozymes in vertebrates and insects indicate that the shift from a defensive function to a digestive function is accompanied by changes from proteins that are basic (i.e., high pI) and sensitive to proteases (i.e., more protease cut sites in amino acid sequence) to acidic and more resistant to protease lysis [10, 34–36, 40]. This is also true of the two i-type oyster lysozymes, cv-lysozyme 1 and cv-lysozyme 2 [56–58]. Interestingly, the two non-oyster lysozymes (T. japonica in [51, 64]; Mytilus edulis 1 in ) biochemically determined to have high pI and numbers of trypsin cut sites (and thus inferred to have defensive function) occur basally in the phylogenetic tree, whereas the sole non-oyster lysozyme determined to be digestive based on low pI and few trypsin cut sites (Chlamys islandica in ) falls in the clade that includes the digestive cv-lysozyme 2. The single exceptional sequence within the digestive clade (with predicted high pI and many trypsin cut sites), cg-lysozyme 3, was generated from mantle tissue , and may represent an instance of reversal to the ancestral defensive function.
Despite their non-sister relationship, cv-lysozyme 1, cv-lysozyme 3, and cg-lysozyme 4 share a unique N-terminal domain. This suggests that this N-terminal domain may have been introduced on to the ancestral core of bivalve i-type lysozyme relatively recently. Itoh and Takahashi  identified a repeat domain that shows high sequence identity (> 50%) with the cv-lysozyme 1 N-terminal region at the N-terminus of some peptidoglycan recognition proteins (PGRPs) from the Pacific oyster, C. gigas. Our alignments (Figure 7) show an even higher level of sequence similarity between these PGRP repeats and the cv-lysozyme 3 and cg-lysozyme 4 N-terminal region. Because peptidoglycans constitute a major proportion of the bacterial cell wall, it may be reasonable to assume that this shared stretch of sequence is involved with the recognition and binding of bacteria in both PGRPs and lysozymes. A PGRP presumably involved in host defense exhibits the molecular signature of positive selection in ants . Regardless of function, the unique N-terminal domain uniting cv-lysozyme 1, cv-lysozyme 3, and cg-lysozyme 4 apparently represents a modular sequence element that can be integrated into different proteins via domain shuffling . Exon structure of bivalve lysozymes supports this possibility. An intron, which could facilitate domain shuffling, occurs just after the hard-to-align N-termini in the lysozymes of both Mytilus edulis 1 and Chlamys [69, 70].
Studies on c-type lysozymes of ruminant artiodactyls indicate that they include intermediate transitional forms between lysozymes that function in host defense and lysozymes that function in the stomach for digestion [71, 72]. Ito et al , for example, reported that lysozymes purified from the kidney of cow and sheep have highest sequence identity with conventional defensive lysozymes, but show greater chemical similarity with stomach lysozymes (i.e., lowered pI and high enzymatic activity at low ionic strength). Some of these intermediates may have been retained because their chemical properties allow them to function under the pH and ionic strength conditions that are unfavorable for the other paralogs . Interestingly, these intermediates of c-type lysozymes are also reported to have optimal pH and ionic strength conditions for lysozyme activity that differ from that of the lysozymes functioning in immunity . Cv-lysozyme 3 showed a higher optimal pH for lysozyme activity compared to cv-lysozyme 1 and cv-lysozyme 2 (Figure 3; Table 1). Purified cv-lysozyme 3 also formed polymers, a phenomenon reported in different lysozymes [14, 59–62], although cv-lysozyme 1 or cv-lysozyme 2 do not (not shown). Thus, cv-lysozyme 3 responds differently to pH and perhaps ionic strength than do other C. virginica lysozymes. The three C. viriginica lysozymes together would thus allow relatively high lysozyme activity to be maintained over a broad range of pH (5.5-9.0) and ionic strength (I = 0.005-0.26) (Figure 3B). The coexistence of three, or perhaps more, lysozymes with different chemical properties would thus help maintain a relatively high lysozyme activity level within tissues under the variable physiochemical conditions faced by oysters, which are osmoconformers and poikilotherms.
Cv-lysozyme 3 could potentially function both in immunity and digestion because it shares the properties of both lysozymes (i.e., cv-lysozyme 1 and cv-lysozyme 3). Previously, we purified just one lysozyme (i.e., cv-lysozyme 2) from the oyster crystalline style , indicating cv-lysozyme 3 is not secreted in quantity into the oyster digestive tubules, which will impede its function as a digestive enzyme. On the other hand, its amino acid sequence similarity with cv-lysozyme 1, including a shared unique amino-terminus (Figure 7), suggests a function in host defense, as the digestive gland can be an important portal of entry of pathogens.
The i-type lysozymes of bivalves share another similarity with c-type lysozymes in some vertebrate lineages: the evolution of digestive lysozymes from a defensive progenitor is accompanied by an episode of positive selection. Jollès et al  noted an elevated rate of nonsynonymous substitutions on the branch leading to a monophyletic clade of ruminant stomach lysozymes. Messier and Stewart  likewise found a burst of nonsynonymous substitutions on the branch leading to digestive lysozymes in Colobine monkeys. That burst involved an inferred 9 nonsynonymous changes, less than the estimated 14 for the branch leading to bivalve digestive lysozymes. Identifying which particular residues changed in bivalves, however, would be a far more speculative exercise than for Colobines, which diverged far more recently (15 My) than bivalve digestive lysozymes (which, given divergence times for the lineages leading to Chlamys and Mytilus, and Crassostrea, should have occurred in the Early Ordovician, about 480 My ago; ).
Considering together lysozymes that appear to be functional and phylogenetic intermediates in both ruminants and bivalves suggests an evolutionary pathway between ancestral defensive and derived digestive forms. The genes encoding lysozyme appear to have undergone multiple duplications within bivalves. This pattern is consist with that seen in the nematode genus Caenorhabditis , where different numbers of lysozyme paralogs among three congeners, along with varied phylogenetic relationships among them, suggest repeated cycles of gene duplication, divergence, and extinction. Duplication of an ancestral defensive paralog of lysozyme in bivalves would allow one copy to explore and adapt to physiological conditions from which the ancestral form was excluded. This specialization may be abetted by more tissue-specific expression of the duplicates . Duplicates that were expressed in acidic environments, even while retaining their defensive function, would be fortuitously preadapted for digestive function, which would evolve following subsequent duplication of the intermediate form. This adaptive scenario makes specific predictions about the breadth of expression and physiological range of defensive, digestive, and intermediate forms (in those species that have the latter two). For example, physiological range should be broader and tissue-specificity should be lower in species having just a single (defensive) lysozyme. Lysozymes have already taught us much about adaptive protein evolution, and their further exploration should continue to inform how new protein functions evolve.