Mast Cell α and β Tryptases Changed Rapidly during Primate Speciation and Evolved from γ-Like Transmembrane Peptidases in Ancestral Vertebrates

Abstract

Human mast cell tryptases vary strikingly in secretion, catalytic competence, and inheritance. To explore the basis of variation, we compared genes from a range of primates, including humans, great apes (chimpanzee, gorilla, orangutan), Old- and New-World monkeys (macaque and marmoset), and a prosimian (galago), tracking key changes. Our analysis reveals that extant soluble tryptase-like proteins, includingα- andβ-like tryptases, mastins, and implantation serine proteases, likely evolved from membrane-anchored ancestors because their more deeply rooted relatives (γtryptases, pancreasins, prostasins) are type I transmembrane peptidases. Function-altering mutations appeared at widely separated times during primate speciation, with tryptases evolving by duplication, gene conversion, and point mutation. Theα-tryptase Gly²¹⁶Asp catalytic domain mutation, which diminishes activity, is present in macaque tryptases, and thus arose before great apes and Old World monkeys shared an ancestor, and before theαβsplit. However, the Arg–³Gln processing mutation appeared recently, affecting only humanα. By comparison, the transmembraneγ-tryptase gene, which anchors the telomeric end of the multigene tryptase locus, changed little during primate evolution. Related transmembrane peptidase genes were found in reptiles, amphibians, and fish. We identified soluble tryptase-like genes in the full spectrum of mammals, including marsupial (opossum) and monotreme (platypus), but not in nonmammalian vertebrates. Overall, our analysis suggests that soluble tryptases evolved rapidly from membrane-anchored, two-chain peptidases in ancestral vertebrates into soluble, single-chain, self-compartmentalizing, inhibitor-resistant oligomers expressed primarily by mast cells, and that much of present numerical, behavioral, and genetic diversity ofα- andβ-like tryptases was acquired during primate evolution.

Human mast cell tryptases exhibit impressively diverse processing, secretion, solubility, catalytic activity, and inheritance. As a group, tryptases are implicated in allergic and other types of inflammation. All mast cell tryptase genes cluster tightly on chromosome 16p13.3 (1). Products of these genes fall into two general categories: membrane-anchored and soluble. The γ gene, TPSG1, encodes a type I transmembrane peptidase anchored to the cell surface after secretion (2, 3). Behavior of peptidase chimeras containing part of mouse γ tryptase suggests that the anchor is a C-terminal hydrophobic peptide, which is not exchanged for a lipid anchor, as occurs in some related enzymes, including prostasin (4). In humans, γ is the sole membrane-anchored tryptase and is hypothesized to be proinflammatory (5). The other three transcribed tryptase genes encode soluble enzymes lacking an anchor. These are TPSAB1 (encoding α and βI tryptases), TPSB2 (encoding βII and βIII), and TPSD1 (encoding δ tryptases) (1, 6, 7).

α Tryptase, the first human tryptase to have its full primary structure determined (8), features a propeptide mutation causing apparent failure of autocatalytic maturation (9). Also, α appears to have a critical catalytic domain mutation, which greatly reduces activity toward substrates readily cleaved by β tryptases. This mutation perturbs the active site in a manner that is unproductive for substrate binding (10–12). In contrast to β tryptase, α fails to evoke neutrophilic inflammation (13) and most or all of it appears to be secreted (probably as a proenzyme) and not stored in mast cell granules (14). α Genes are absent in a substantial portion of most humans (6). This appears to be because they are alleles at a locus that also accepts βI genes (1). Humans without α genes have diminished circulating levels of immunoreactive tryptase compared with those with α genes (15). Thus, α and β tryptases differ in important ways.

β Tryptases are products of two adjacent loci. The major alleles (I–III) are highly similar (16, 17). β Tryptases encode soluble, active enzymes that are stored in secretory granules and released in response to allergen-bound IgE and other stimuli (18). They self-assemble into tetramers, which shield the active site from inactivation by circulating inhibitors (19). β Tryptases are the dominant forms isolated from tissue extracts and are targets for therapeutic inhibition because of postulated importance in diseases such as asthma, anaphylaxis, and inflammatory bowel disease (7, 20, 21). Human δ tryptases differ in that they are C-terminally truncated (1). They also feature the propeptide mutation that blocks processing in α tryptases. Consequently, δ tryptase has little catalytic activity (22) and may have defective propeptide processing. Although the δ gene TPSD1 was first hypothesized to be a pseudogene (1, 23), transcripts and immunoreactivity were later detected in multiple organs (22). However, the targets and roles, if any, of δ tryptases remain to be determined.

The origins of human tryptase isoforms and of mammalian tryptases in general are obscure. Unlike many other mammalian peptidases, tryptases lack obvious orthologs in nonmammalian vertebrates. To understand origins and consequences of the diversity of expressed human tryptases, the present study probes the evolution of human α, β, and γ tryptases. The data acquired for this study suggest that tryptases proliferated and changed rapidly during mammalian evolution, arising from ancestral membrane-anchored peptidases, which are present in a variety of vertebrate genomes. The pace of change accelerated during evolution of primates. This work's comparison of primate tryptases suggests that several idiosyncratic features of human enzymes are recent developments. The resulting analysis reveals human tryptase origins and highlights peptidases of likely functional importance.

Materials and Methods

Data mining

Full primary sequences of mammalian tryptases not already published or annotated were obtained by data mining, including basic local alignment search tool (BLAST)³ and BLAST-like alignment tool searches of high throughput genome sequence and whole genome shotgun databases at the National Center for Biotechnology Information using mammalian tryptase genes and cDNAs as query sequences. Marmoset sequences were obtained from BLAST searches of genomic sequence archived at the Washington University School of Medicine Genome Sequencing Center (http://genome.wustl.edu). Amino acid sequences of previously unreported tryptases were extracted from genomic DNA using existing tryptase gene structures as a guide, following standard rules for placement of intron-exon boundaries. To avoid connecting two closely related but separate genes, only genomic fragments containing complete coding sequence of the catalytic domain were subjected to this analysis. Most tryptase genes are small compared with other serine protease genes, so that shotgun-derived fragments can be long enough to contain a complete gene.

Sequencing of primate tryptase genes

Primate genomic DNA was obtained from the Primate Cell Repositories of the Coriell Institute for Medical Research (Camden, NJ), which provides genomic DNA from species-specific cultured fibroblasts. DNA encoding primate α- and β-like genes was amplified by PCR using Advantage 2 (BD Clontech). For α- and β-like genes, primers bracketed full protein-coding sequence and were based on highly conserved portions of the 3′ and 5′ flanks of human α and β tryptases. Individual amplimers were ligated into pCR2.1-TOPO (Invitrogen Life Technologies). Resulting recombinant plasmid DNA was purified and the inserted portion was sequenced.

Sequencing of the human implantation serine protease (ISP) gene

We obtained full sequence of the human ISP2 gene from a fragment of human chromosome 16p13.3 in bacterial artificial chromosome (BAC) 48D21 (Invitrogen Life Technologies). A fragment generated by digestion of the pBeloBAC11 clone with HindIII was subcloned as described (3). This ∼8-kb subclone (48K), which contains the 5′ half and flank of the δII-tryptase gene, was further sequenced by gene walking to obtain the adjacent ISP2 gene, which is oriented tail-to-tail with respect to the δII gene (24).

Phylogenetic comparisons

Preprotryptase amino acid sequences were compared using Geneious software (Biomatters). Aligned sequences were subjected to phylogenetic analysis, including tree preparation, using unweighted pair group method with arithmetic mean (UPGMA) and neighbor-joining techniques with bootstrap resampling.

Results

Nucleotide sequence, size, and intron/exon structure of primate α-like/β-like genes

Genomic sequence bracketing the full protein coding sequence of primate tryptases was obtained by sequencing α10 cloned amplimers each from gorilla, chimpanzee, orangutan, and crab-eating macaque (cynomolgus monkey) genomic DNA. Nucleotide sequence of genomic amplimers yielding unique tryptase sequences was deposited into GenBank (see accession numbers in Table I). The length of sequenced genes was similar to that of related human genes. Exonic sequence was predicted by aligning human tryptase cDNAs. In all cases, introns began with dinucleotide GT and ended with AG. Intron phase (0, 1, or 2) was identical with that of homologous introns in other soluble mammalian tryptases.

Table I.

Primate tryptase sequences: nomenclature, sources, and identifiers

Primate	Protein	GenBank Identifier
Homo sapiens (human)	Tryptase, αI, αII	M30038; AF098328
	Tryptase, βI	M33494
	Tryptase, βII	M33492 and AF099145
	Tryptase, βIII	AF099143
	Tryptase, γI, γII	NM_012467; AAF76458
	Tryptase, δI	NM_012217
	ISP2	AF529082a
Pan troglodytes (chimpanzee)	Tryptase, α	XM_001158624
	Tryptase, β	EF206351a
Gorilla gorilla (gorilla)	Tryptase, β1	EF208020a
	Tryptase, β2	EF208021a
	Tryptase, β3	EF208022a
Pongo pygmaeus abelii (Sumatran orangutan)	Tryptase, β1	EF452227a
	Tryptase, β2	EF452228a
	Tryptase, β3	EF452229a
	Tryptase, β4	EF452230a
Macaca fascicularis (crab-eating macaque)	Tryptase, αβ1	EF212445a
	Tryptase, αβ2	EF212444a
Macaca mulatta (rhesus macaque)	Tryptase, αβ	XM_001088289
	Tryptase, γ	AANU01106415b
	ISP2	XP_001118570
Callithrix jacchus (common marmoset)	Tryptase	Contig12371.9c
	Tryptase, γ	Contig12371.8c
Otolemur garnettii (small-eared galago)	Tryptase, γ	AAQR01303204b

^aCloned and sequenced, this work.

^bDeduced from unannotated whole genome shotgun sequence.

^cDeduced from unannotated sequences from http://blast.wustl.edu.

Primary structure of primate α-like/β-like tryptases: general features

The preprotryptase amino acid sequence predicted from each of the cloned genes is 275 aa, comprised of a −30 residue leader and a +245 residue catalytic portion (Fig. 1). Each possesses “catalytic triad” residues His⁴⁴, Asp⁹¹, and Ser¹⁹⁵ (corresponding to His⁵⁷, Asp¹⁰², and Ser¹⁹⁵ of chymotrypsinogen) common to active peptidases of the chymotrypsin family. Each has the Gly-1 characteristic of mast cell tryptases at the site of propeptide removal as well as most of the residues forming the hydrophobic interfaces between subunits in crystallized human βII tryptase tetramer (19). These interface-forming residues include Pro⁴⁸, Tyr⁶⁶, Tyr⁶⁷, Tyr⁸⁴, Pro¹⁴⁰, Pro¹⁴¹, and His¹⁶³, which are absolutely conserved in known soluble primate and murine tryptases (Fig. 1). Of these residues, only Pro¹⁴¹ (Pro¹³⁸ in γ tryptase) is conserved in nonoligomerizing γ tryptases. The six surface loops involved in making intersubunit contacts are underlined in Fig. 1. None of the deduced tryptases, including the chimpanzee enzyme otherwise similar to α, has the Gln⁻³ found in human α and δ tryptases (1, 8). Consensus glycosylation sites at Asn¹⁰² and Asn²⁰³ are conserved in this group, although human βII, which lacks the Asn¹⁰² consensus site, is aberrant in this regard. Two of the residues forming the “specificity triad,” which shapes the pocket accommodating the substrate P1 side chain at the site of hydrolysis, are absolutely conserved. One of these is Asp¹⁸⁸ (189 in chymotrypsinogen), as expected of enzymes that are primarily tryptic in specificity. The other entirely conserved residue is Gly²²⁵ (226 in chymotrypsinogen). However, residue 215 (216 in chymotrypsinogen) varies, being Asp in human and chimpanzee α and in all macaque αβ tryptases, but Gly in all others, including human β tryptases and classic chymotrypsin-family peptidases of tryptic specificity. This is somewhat surprising because this residue in human α distorts the substrate binding site and limits peptidase activity (11, 12). Although rhesus and crab-eating macaque tryptases feature this α-like mutation, overall they are nearly as similar to human β as to α, as revealed by Fig. 2, therefore, we refer to them here as “αβ” tryptases. Marmoset tryptase, in contrast, possesses the classic specificity triad but is nearly as similar to α as to β (Table II), hence, we refer to this enzyme simply as tryptase. The tryptases most similar to human are from gorilla (Table II, Fig. 2). All three gorilla tryptases are more nearly identical with human than to chimpanzee β. Although orangutan tryptases share some residues with human and chimpanzee α and others with human and gorilla β; overall, they are more β-like, especially in regard to functionally important residues like Gly²²⁵. Therefore, we apply the β label to these orangutan tryptases.

FIGURE 1.

Alignment of primate α-like/β-like/δ-like tryptases. Examples of tryptases sequenced or deduced for this work are aligned as indicated and compared with human βI, αII, and δ tryptases and mouse mast cell protease (MMCP) 6 and 7 tryptases. See Table I for GenBank identifiers. Hyphens (-) indicate amino acid identity compared with corresponding residues in human βI tryptase. Asterisks (*) identify gaps in the alignment. Symbol meanings: #, “catalytic triad” residues common to all active serine peptidases; %, conserved “specificity triad” residues typical of most serine peptidases of tryptic specificity; +, the predicted sites of N-glycosylation conserved in all of these enzymes; underlined residues indicate loops involved in subunit contacts.

FIGURE 2.

Tree of α-like/β-like tryptases. Potential phylogenetic relationships between classic murine and primate soluble mast cell tryptases, including proteins newly sequenced or deduced for this study, are probed with this rooted dendrogram generated by UPGMA with 500 iterations of bootstrap resampling. The tree also is a template for tracking other tryptases with the propeptide processing mutation of human α (R-3Q, heavy solid line) as well as the catalytic domain mutation (G215D, hatched line). Proposed branch points for α and β clades are identified by the appropriate symbol. Sources of primate tryptase sequences are in Table I. Nonprimate sources are as follows: opossum (M. domestica, deduced from whole genome shotgun sequence AAFR03046243), rat Mcpt6 and 7 (R. norvegicus, AAB48262 and AAB48263), mouse Mcpt6 and 7 (M. musculus, P21845 and Q02844). Note that the αβ primate tryptases share ancestry more recently with rodent Mcpt6 tryptases than with Mcpt7, that key α mutations appeared at times widely separated in primate evolution, and that the αβ dichotomy is recent and only evident among great apes.

Table II.

Pairwise comparisons of percentage identity of mouse and primate soluble tryptase catalytic domains^a

	Go	Ch	Or	Hu	Ch	Rh	Mf	Ma	MP	MP
	β1	β	β4	αII	α	αβ	αβ1		6	7
Hu βI	99	97	97	94	94	91	90	86	78	77
Go β1		98	96	94	94	91	90	86	78	77
Ch β			96	95	95	90	88	84	77	77
Or β4				93	93	93	91	87	78	77
Hu αII					98	91	90	85	76	77
Ch α						91	91	84	76	76
Rh αβ							98	86	76	76
Mf αβ1								85	76	75
Ma									73	76
MP 6										71

^aHu, Human; Go, gorilla; Ch, chimpanzee; Or, orangutan; Rh, rhesus; Mf, M. fascicularis macaque; Ma, marmoset; MP, mouse mast cell protease.

Genomic and amino acid sequence of primate γ tryptases

As shown in Fig. 3, we deduced γ-tryptase sequence from three nonhuman primates, including a prosimian (the small-eared galago, Otolemur garnettii), an Old World monkey (rhesus macaque), a New World monkey (common marmoset, Callithrix jacchus), and a great ape (orangutan). Table III demonstrates the percentage identity of pairs of primate and rodent γ tryptases. We predict that mature versions of these tryptases are two-chain, type I transmembrane, tryptic peptidases, like human γ. Hydropathy analysis (http://gcat.davidson.edu/rakarnik/kyte-doolittle.htm) using a 19-residue window reveals C termini typical of membrane-spanning regions (data not shown). Inspection of these sequences for glycosylphosphatidylinositol anchor attachment sites using the Big-PI algorithm (http://mendel.imp.ac.at/sat/gpi/gpi_server.html) reveals no consensus sites. Therefore, these primate γ tryptases, like mouse (as revealed by γ/prostasin chimeras; Ref. 4), are not likely to have C-terminal peptide anchors swapped for lipids.

FIGURE 3.

Alignment of γ tryptases. Hyphens (-) indicate amino acid identity compared with the corresponding residues in human γI tryptase. See Table I for identifiers of primate GenBank files, which contain full genomic sequence. Mouse (M. musculus) and rat (R. norvegicus) sequences are from NP_036164 and NM_175593, respectively. Asterisks (*) identify gaps in the alignment, the largest of which is in mouse and rat prosequence, corresponding to an exon that is transcribed in human and, by homology, other primate γ tryptases but not in murine γ genes. Symbol meanings: @, an absolutely conserved cysteine via which the propeptide remains linked to the catalytic domain after cleavage-activation between Arg-1 and Ile +1 to yield two-chain mature enzymes; underlined residues comprise a predicted transmembrane segment present in γ but not soluble tryptases; the meanings of #, %, and + symbols are as in Fig. 1.

Table III.

Pairwise comparisons of percentage identity of rodent and primate γ-tryptase catalytic domains

	Orangutan	Rhesus	Marmoset	Galago	Mouse	Rat
Human	92	93	86	79	77	77
Orangutan		93	88	80	77	78
Rhesus			87	79	76	77
Marmoset				82	78	78
Galago					80	79
Mouse						90

Primate ISP2 “implantation tryptase” genes

The full sequence of human ISP2 was obtained from subclone 48K of a human BAC (as noted in Materials and Methods), which also contains tryptase genes (3). The sequence of the human ISP2 gene (which appears to encode a nonfunctional protein because the coding sequence goes out of frame compared with the mouse ISP2 sequence) was deposited in GenBank (Table I). Its presence on the same subclone as δ tryptase confirms that it lies close to classic tryptase genes and belongs to the tryptase locus. Because it appears to be encoded by a pseudogene, human ISP2 is not included in the tree in Fig. 4. Also, mining of available chimpanzee and orangutan genome sequence yielded only flawed ISP2 genes, which therefore are excluded from the tree. However, amino acid sequence representing apparently intact ISP2 was deduced from whole genome shotgun sequence from rhesus macaque and is included in the phylogenetic analysis. Thus, it appears that the ISP2 gene may have remained functional in primates at least until macaques and great apes shared a common ancestor.

FIGURE 4.

Tree of vertebrate tryptase-like proteases. This rooted dendrogram was prepared by UPGMA with 500 iterations of bootstrap resampling. Refer to Table I for GenBank sources of galago, rhesus, and human γ tryptases and rhesus ISP2, to Fig. 3 for GenBank sources of mouse and rat γ tryptase, and to Fig. 2 for GenBank sources of opossum tryptase and rat and mouse Mcpt7 and Mcpt6. Other sequences were obtained as follows: frog channel-activating peptidase (CAP) 1 (Xenopus laevis, AAB96905), lizard, opossum, dog, cat, mouse, rat, rhesus, orangutan, chimpanzee, and human prostasin (Anolis carolinensis DS229345; M. domestica XP_001372224; C. familiaris XP_848861; Felis catus AANG01151030; M. musculus NM_133351; R. norvegicus AAG32641; Macaca mulatta XP_001112376; Pongo pygmaeus abelii CAH90878; Pan troglodytes XM_001157456; Homo sapiens NM_002773), lizard, mouse, rat, horse, dog, marmoset, rhesus, chimpanzee, and human pancreasin (A. carolinensis AAWZ01023110; M. musculus NM_175440; R. norvegicus NM_182949; and E. caballus AAWR01029521; C. familiaris AAEX02025250; C. jacchus contig 2675.45 (http://genome.wustl.edu); M. mulatta XM_001086389; Pan troglodytes XP_510751, Homo sapiens NM_031948), horse γ tryptase (E. caballus AAWR01029469), platypus tryptase-like protein 1 and 2 (O. anatinus AAPN01127233 and AAPN01196394), mouse, rat, and squirrel ISP2 (M. musculus AAK15264, R. norvegicus XP_220240, and Spermophilus tridecemlineatus AAQQ01728300), mouse, dog, cattle, and pig mastin (M. musculus AAS21652; C. familiaris P19236; B. taurus XM_869964; S. scrofa NP_998959), bat, hedgehog, horse, dog, pig, gerbil, sheep 1, sheep 2, cattle 1, and cattle 2 tryptase (M. lucifugus AAPE01472347; E. europaeus AANN01723492; E. caballus AJ515902; C. familiaris M24664; S. scrofa NP_999356; M. unguiculatus D31789; O. aries Y18223 and Y18224; B. taurus NP_776627 and AAFC03119556).

Nonprimate classical soluble tryptase sequences

The cDNA and corresponding amino acid sequence were deduced for this work from GenBank-deposited, unannotated whole genome shotgun sequences from the following nonprimate mammals: little brown bat (Myotis lucifugus), AAPE01472347; European hedgehog (Erinaceus europaeus), AANN01723492; short gray-tailed opossum (Monodelphis domestica), AAFR03046243; and duck-billed platypus (Ornithorhynchus anatinus), AAPN01127233, AAPN01196394, AAPN01292231. The platypus sequence provides evidence of a complete tryptase sequence in a monotreme, which is believed to be a survivor of an early branch in the mammalian tree. Similarly, the marsupial opossum sequence is an example of a tryptase sequence from a nonplacental mammal.

Also included in the analysis are tryptases already deduced and searchable in GenBank as follows: mouse (Mus musculus), P21845 and Q02844; rat (Rattus norvegicus), U67909 and U67910; dog (Canis familiaris), M24664; horse (Equus caballus), AJ515902; gerbil (Meriones unguiculatus), D31789; sheep (Ovis aries), Y18223 and Y18224; cattle (Bos taurus), NP_776627; and pig (Sus scrofa), NP_999356.

Nonprimate tryptase-like mastins, ISPs, γ-like prostasins, and related genes

See Fig. 4 legend for sources of newly or previously deduced full-length protein sequences used to construct a master tree, including tryptase-related type I transmembrane peptidases in non-mammalian vertebrates.

Discussion

This work illuminates origins of human α, β, and γ tryptases. In particular, it reveals that much of the diversity of form and function among expressed human tryptases was generated during primate evolution. The data further suggest that the human genome harbors tryptase-like pseudogenes, some of which are expressed and active in other mammals, including nonhuman primates. Several genetic mechanisms contributed to creation, alteration, and inactivation of primate tryptase-like genes. These include segmental duplication, gene conversion, chimera formation, and point mutation. Overall, our analysis charts an evolutionary path from γ-like, membrane-anchored, two-chain, inhibitor-sensitive peptidases in ancestral vertebrates to soluble, single-chain, self-compartmentalizing, inhibitor-resistant oligomers.

Compared with opossum and dog tryptase loci, the murine Mcpt6/Mcpt7 tryptase locus is duplicated and the human α/human β/human δ (TPSAB1/TPSB2/TPSD1) locus is triplicated. As shown in prior work from this laboratory, the δ (TPSD1) gene product is a truncated tryptase chimera created by gene conversion events (1). The present analysis, focusing on α- and β-like tryptases, indicates that αβ dichotomization and key mutations affecting activation, storage, secretion, and activity of α tryptase occurred during evolution of primates. As indicated by branching in Figs. 2 and 4, the αβ dichotomy is unrelated to the Mcpt6/Mcpt7 split in rodents. Indeed, human and chimpanzee TPSAB1 (α and βI) and TPSB2 (βII/βIII) genes are duplicated in relation to the corresponding mouse locus (24). This duplication likely arose from an ancestor shared with Mcpt6, given that the mouse MCP-6 sequences aligns more closely than MCP-7 with the αβ clade, as shown in Fig. 2. The duplication probably occurred during primate evolution, because comparable duplications have not been noted in nonprimates. The rodent Mcpt7 gene shares mixed ancestry with the chimeric human TPSD1 (δ) gene, which appears to have been created by a conversion event involving a recent ancestor of a primate αβ-like gene and a more distantly related tryptase similar to rodent Mcpt7 (1). The exact origin of the Mcpt6/Mcpt7 duplication extant in rodents and further multiplied and modified in primates is unclear. However, some mammals, including dogs, possess only one orthologous tryptase gene, suggesting that the duplication of the putative ancestral soluble tryptase gene occurred after dogs, rodents, and primates shared a common ancestor, but before ancestors of rodents and primates split from the tree.

As demonstrated by Fig. 1 and Table II, there are numerous differences between α and β tryptases (e.g., 19 mismatches between human αII and βI and 15 between chimpanzee α and β). Of the variant residues, two in particular cause unconventional behavior in human α vs β: 1) Gln⁻³ substituting for Arg in the α propeptide apparently precludes autolytic processing, activation, oligomerization, and storage in secretory granules (9, 14), and 2) Asp²¹⁵ substituting for Gly in the catalytic domain disorders the active site and reduces activity of any α that manages to be correctly processed and activated from its zymogen form (10–12). Our analysis reveals that origins of these mutations differ. The Arg⁻³Gln processing mutation is nearly new, being present in human but not chimpanzee α. We did not detect this mutation in other primate tryptases, with the exception of human δ, where its presence may be due to a recent partial conversion involving α and δ genes, whose tandem orientation can facilitate such an event, much as δ itself is a chimera generated by more remote conversion events (1). Although the −3 residue is Arg in most nonprimate tryptase propeptides, this is not universal. In a gerbil tryptase, for example, the corresponding residue is Glu (25). Whether activation of this tryptase is impaired is not known. It is worth noting that mastins possess propeptides very similar to those of tryptases; they, too, lack a basic amino acid in the −3 position, yet are processed to active, oligomeric, granule-sequestered forms (26–28). Thus, for some tryptase-like enzymes, autolytic processing at Arg⁻³ is not essential for maturation.

Surprisingly, we found that the Asp²¹⁵Gly catalytic domain mutation is present in several tryptases in Old World monkeys (i.e., macaques) but we found no evidence of this mutation in tryptase from a New World monkey (i.e., marmoset) (Figs. 1 and 2). Indeed, all α-like/β-like sequences identified in rhesus and crab-eating macaques contain α-like Asp²¹⁵, although marmoset contains conventional Gly²¹⁵ tryptase. The macaque and marmoset tryptases are no more closely related to human/chimpanzee α than to β in overall structure, as revealed by Figs. 2 and 4. Thus, these data suggest that the catalytic domain mutation appeared after New and Old World monkeys diverged from the tree, but well before the split into α and β, which occurred after Old World monkeys and great apes shared a common ancestor. Although it is possible that the Asp²¹⁵Gly mutation occurred once in ancestors of macaque αβ tryptases and a second time in ancestral chimpanzee/ human α, it is more likely that the Asp²¹⁵Gly mutation arose once in an ancestor shared by macaque αβ and chimpanzee/human α tryptases. The fact that we have not encountered this mutation in any tryptase-related protease in a nonprimate further supports the conclusion that it arose during evolution of primates. It remains to be seen whether macaque Asp²¹⁵ tryptases are catalytically impaired, like human α.

The trees in Figs. 2 and 4 suggest that classic soluble mast cell tryptases were present early in mammalian evolution, existing now in all major groups of mammals, including monotreme (platypus), marsupial (opossum), and placental mammals (many examples). The Fig. 4 dendrogram reveals clearly that soluble tryptases form a clade separate from tryptase-like mastins and ISPs. At present, there is no evidence that mastin and ISP are functional in humans. However, mastins are expressed and active in several nonprimates, including dog (28, 29), pig, and mouse (in which mastin is also known as TC30 tryptase and MCP-11/Prss34, respectively (27, 30). GenBank-deposited mastin expressed sequence tag EC335262, which encodes most of the protein from an opossum (silver-gray brushtail, Trichosurus vulpecula), is evidence that mastins were present in early mammals—at least back to the time that placental and marsupial mammals shared a common ancestor. Another expressed sequence tag originates from pig-tailed macaque (DY760362), in which it may be a transcribed pseudogene, as also predicted of rhesus mastin. There is no evidence of transcription of the human mastin gene. These sequences are not included in the Fig. 4 dendrogram because they are incomplete or contain early stop codons or other major flaws.

Although the sequence deduced from the human ISP2 gene suggests that it is likely to be a pseudogene (even if transcribed), our database screens suggest that it is potentially functional in rhesus macaques. This is worth noting, because ISPs (and mastins) are the proteins most closely related to classic soluble mast cell tryptases. Curiously, we do not find clearly identifiable soluble tryptase or tryptase-like proteases (including mastins and ISP2) in nonmammalian vertebrates (or for that matter, in invertebrates). As demonstrated by Fig. 4, the α and β tryptases, mastins, and ISPs are related to γ tryptases, which differ in key ways summarized in Table IV. The differences include 1) an extra exon encoding the propeptide, 2) Arg rather than Gly at the site of zymogen activation, 3) a two-chain mature form, and 4) a C-terminal transmembrane anchor. Nonetheless, several features link γ to soluble tryptases. These include overall structural/phylogenetic similarity to soluble tryptases and tryptase-like enzymes (Fig. 4), conservation of structural motifs (such as the polyproline tract LPPPF/Y, 139−143 in αβ and 136 −140 in γ; see Figs. 1 and 3) shared only by γ and soluble tryptases; the position of γ in the tryptase gene cluster as the nearest neighbor of soluble tryptase genes TPSB2 and Mcpt6 in human and murine genomes, respectively, tryptic specificity, and expression in mast cells. Rat and mouse γ genes lack the small second exon featured in primate γ tryptases (see Table IV), which accounts for the propeptide gap in the murine γ propeptide sequences aligned in Fig. 3. Inasmuch as this small exon is a feature of primate γ tryptases as well as of related type I transmembrane peptidases, like prostasin (3), rodent γ tryptases can be seen as transitional. In light of these considerations, and because the γ stem is basal to that of all known soluble tryptase-like enzymes, extant γ tryptases may be similar to tryptases in their ancestral form.

Table IV.

Features of tryptases and closest relatives

Protease	Specificity	Transmembrane Anchor	No. of Chains	Activation Site	No. of Prepro Exons	Mast Cell Expression	5′ UTR Introna	Inhibitor Resistance	216 aa
Testisins	Tryptic	+	2	Arg	3	−	−	−	Gly
Prostasins	Tryptic	+	2	Arg	3	−	−	−	Gly
Pancreasins	Tryptic	+/−	2	Arg	3	−	−	−	Gly
γ Tryptases	Tryptic	+	2	Arg	2−3	+	−	−	Gly
ISP2s	Tryptic	−	1	Gly	2	?	+	+/−	Gly
Mastins	Tryptic	−	1	Gly	2	+	−	+	Gly
β Tryptases	Tryptic	−	1	Gly	2	+	+	+	Gly
δ Tryptases	Tryptic	−	1	Gly	2	+	+	?	X
α Tryptases	Tryptic	−	1	Gly	2	+	+	+	Asp

^aUTR, Untranslated region.

Like soluble tryptase-like enzymes, γ tryptases have no clear counterparts in nonmammalian vertebrates. In mammals, some of their closest relatives are pancreasins (also known as marapsins) (31), which are tryptic serine peptidases of uncertain function, not known to be expressed in mast cells. Except in humans and chimpanzees, pancreasins (like γ tryptases) are predicted to be type I transmembrane peptidases (Ref. 31 and data not shown). The closest relatives of pancreasins are other transmembrane peptidases, including prostasins, testisins, and various nonmammalian, vertebrate transmembrane peptidases, some of which are shown in the tree in Fig. 4. Table IV summarizes attributes of these enzymes and supports an evolutionary path to soluble tryptases. Losing the transmembrane portion of a type I peptidase is relatively simple as exemplified by pancreasin, which is soluble in humans and membrane-anchored in mice (31). A change of just two nucleotides compared with mouse pancreasin caused tail loss in the human enzyme. These considerations lead us to propose that premammalian vertebrate ancestors of extant mammalian soluble mast cell tryptases were type I transmembrane peptidases, which lost their tails.

Loss of the transmembrane segment in ancestral tryptases apparently triggered a quite dramatic increase in the pace of evolutionary change. In the Fig. 4 tree, this is evident by comparing the extent of sequence conservation in membrane-anchored, tryptase-related serine peptidases (γ tryptases, pancreasins, and prostasins) with the level of conservation among soluble, tryptase-related peptidases (ISPs, mastins, and tryptases themselves). As a group, soluble tryptase-like enzymes diverge much more dramatically than their transmembrane relatives. For example, aligned human and opossum tryptase catalytic domains have more than twice as many mismatches as the same pairing of prostasins, and pairings within the ISP2 or mastin groups vary to an even greater extent. The soluble tryptase-like genes evolved not only by accumulating mutations but by duplication and reduplication, so that some now appear to be redundant in some mammalian genomes, including human. Indeed, we calculate that a minimum of five sequential duplications (see Fig. 5), starting from an ancestral γ-like gene, occurred on the evolutionary path to the present human tryptase locus, including the multiallelic αβ loci, TPSAB1 and TPSB2. In contrast, genes encoding the transmembrane group, including prostasin, pancreasin, and γ tryptase itself, tend to be solitary. If the transmembrane peptidases are encoded by single-copy genes and, like prostasin, are essential for life and specialized to cleave specific endogenous substrates, they will tend to resist evolutionary change and exaggerate differences in rates of change between transmembrane and soluble tryptase-like enzymes.

FIGURE 5.

Proposed origins of human tryptases. Pseudogenes or catalytically flawed tryptases are shown in gray. Numbers 1−5 identify an ordered sequence of duplications, starting from an ancestral γ-like gene in early mammals, and giving rise to the current diversity of soluble human tryptase and tryptase-like genes. Upward arrows indicate proposed gene conversion events leading to the creation of the δ chimera and to allelic disparity of α and βI genes at the TPSAB1 locus.

Our analysis indicates that the transmembrane enzymes are more deeply rooted in vertebrate evolution than the soluble tryptase-like peptidases, because close relatives and putative orthologs of the transmembrane group are present in major groups of premammalian vertebrates, including amphibian, reptile, and fish, but nonmammals examined to date lack obvious orthologs of soluble tryptase-like peptidases. Liberation from the membrane would provide access to a broader range of extracellular targets and the potential to participate in a wider range of functions. We speculate that this created new opportunities to respond to selective evolutionary pressures, with coevolving pathogen-derived substrates and antipeptidases accelerating the pace of change. The major expansion of families of vertebrate genes encoding other soluble immune peptidases, like mast cell chymases and lymphocyte granzymes (32–34), also may be a response to such pressures.