In this work, we have performed an exhaustive bioinformatic analysis of the human genome to try to identify new serine proteases that could contain different catalytic domains within the same polypeptide chain. These bioinformatic searches led us to find a region in chromosome 16p11.2 putatively encoding a new polyprotease. After completing the cloning process using liver cDNA as template, we confirmed that the identified sequence was a new polyserine proteinase that we called polyserase-3 to underline its structural relationship with the previously described polyserases-1 and -2 [3,9]. However, the polyserase-3 architecture is less complex than the exhibited by the two other human polyproteases. Thus, this new polyserase is composed of two serine protease domains preceded by a signal peptide, whereas both polyserase-1 and polyserase-2 contain three catalytic domains in a single polypeptide chain.
A comparative structural analysis also revealed that polyserase-3 is more closely related to polyserase-2 than to polyserase-1. Thus, and similar to polyserase-2, polyserase-3 is a secreted soluble protein that lacks additional domains found in polyserase-1 such as a type II transmembrane sequence and a low-density lipoprotein receptor motif. Likewise, the serine protease domains of polyserase-2 and polyserase-3 remain as integral parts of the same molecule, whereas polyserase-1 undergoes a series of post-translational processing events that release the three protease domains from the initial translation product . The structural basis for these differences may derive from the fact that both protease domains of polyserase-3, as well as the second and third protease domains of polyserase-2, are preceded by a region lacking the consensus activation motif Arg-Ile-Val-Gly-Gly, characteristic of serine proteases. Therefore, it is unlikely that these domains can be separated from the original polypeptide chain by a trypsin-like protease. Nevertheless, the possibility that polyserase-3 can be activated under specific circumstances through alternative mechanisms such as those operating in α-tryptase [41,42], can not be ruled out.
The phylogenetic analysis of human serine proteases whose sequence is available, also revealed the relationship between polyserases-2 and -3, since the different serine protease domains of both polyproteases are grouped together and form a branch equally distant from members of the TTSP and the tryptase/prostasin families. Interestingly, analysis of the gene structure and organization of polyserases-2 and -3 showed common features with these two groups of serine proteases. Therefore, it is possible that ancestors of both TTSPs and tryptase/prostasin contributed to the formation of these polyserases through recombination or exon swapping events. Likewise, the gene encoding polyserase-3 maps very close to the polyserase-2 and prostasin genes at chromosome 16p11.2, a region linked to genetic abnormalities whose loci remain unidentified. These pathologies include paroxysmal kinesigenic choreoathetosis  and autosomal dominant myxomatous mitral valve prolapse , opening the possibility that the identified human polyserases could be implicated in the development of these diseases. Nevertheless, beyond all these similarities between human polyserases-2 and -3, clear differences were also detected between them. Thus, we have previously reported that polyserase-2 is a glycosylated protein that only shows catalytic activity in its first serine protease domain. By contrast, polyserase-3 is a secreted and non-glycosylated enzyme.
As an initial step towards the functional characterization of polyserase-3, we analyzed its expression profile in different human tissues. These studies revealed additional differences between polyserase-3 and the previously described human polyserases in their patterns of expression in different tissues and cancer cell lines. Thus, polyserase-3 is mainly expressed in adult heart, liver, intestine, ovary and testis, as well as in all analyzed fetal tissues including kidney, brain, liver and lung. By contrast, polyserase-2 is predominantly detected in adult skeletal muscle, liver, placenta, prostate and heart, as well as in fetal kidney but not in fetal lung, liver or brain. It is also noteworthy that several human tumor cell lines also express this new polyprotease, a feature shared with the previously described human polyserases, opening the possibility that these complex serine proteases could mediate proteolytic processes associated with tumor development or progression [26,45].
To evaluate the possibility that polyserase-3 is an active enzyme with ability to perform these proteolytic events, we undertook the production of the entire protein as well as its two serine protease domains as independent proteins. The activity assays showed that the complete polyserase-3 is able to degrade some substrates present in the extracellular matrix such as fibrinogen and pro-uPA. These results suggest that this enzyme could contribute to tumor progression either through the degradation of extracellular matrix proteins or the activation of other components including different tumor-associated proteases . Contrary to the situation with the entire protein, the two serine protease domains produced as independent proteins did not show any apparent proteolytic activity against the substrates indicated above. Interestingly, members of the tryptase family – that show significant degree of structural similarity with both protease domains of polyserase-3- are secreted as monomers and must form tetramers to carry out the catalysis . These facts prompted us to evaluate whether this enzyme could dimerize to generate a protein with four potential active sites, and whose quaternary structure could be similar to that formed by the tryptases. To analyze this possibility, we compared the electrophoretic mobility of the recombinant protein under non-reducing and reducing conditions. This assay would indicate that approximately half of the purified polyserase-3 may form active dimers which are likely stabilized through the formation of disulfide bonds. The information derived from the three-dimensional models generated for Spd1 and Spd2 was also consistent with this possibility. Thus, these models, together with predictions based on amino acid sequence alignments, suggest that polyserase-3 possesses a total of nine intrachain disulfide bonds. However, there are three free cysteines (Cys189, 293 and 543) that could be involved in the stabilization of the polyserase-3 dimer through the formation of disulfide bonds. The stability of the dimer would be further maintained by a series of conserved tryptophan residues (Trp 48, 50, 159 and 237 in Spd-1, and Trp 312, 314, 417 and 490 in Spd2) that have been reported to be necessary for mouse mast cell tryptase dimerization and activity . Interestingly, this protease can also cleave the α-chain of fibrinogen as demonstrated herein for polyserase-3. Due to the homology of polyserase-3 and β-II tryptase, we cannot rule out the existence of further interactions involving hydrogen bonds and salt bridges and participating in the dimer formation through the six loops (code-colored in Fig. 5A) previously described for β-II tryptase . Regarding the fact that fibrinogen cannot be cleaved by Spd1 or Spd2 produced as independent proteins, we can speculate that the presence of three aspartic acid residues (Asp164, 166 and 169) in a loop of Spd1 could form a negatively charged anchoring site that would compete with the substrate binding pocket when the protein is in a monomer state. Stabilization of this loop in the dimer state, probably by Spd2, would grant access for the substrate to the active site. These acidic residues seem also to be important for β-II tryptase, whose enzymatic activity is totally abolished in the monomer state . Nevertheless, further functional studies will be necessary to verify these predictions in the case of polyserase-3.