Advertisement

Three‐dimensional structure of the human transglutaminase 3 enzyme: binding of calcium ions changes structure for activation

Bijan Ahvazi, Hee Chul Kim, Sun‐Ho Kee, Zoltan Nemes, Peter M. Steinert

Author Affiliations

  1. Bijan Ahvazi*,1,
  2. Hee Chul Kim1,
  3. Sun‐Ho Kee1,
  4. Zoltan Nemes1 and
  5. Peter M. Steinert1
  1. 1 Laboratory of Skin Biology, National Institute of Arthritis and Musculoskeletal and Skin Diseases, National Institutes of Health, Bethesda, MD, 20892‐8023, USA
  1. *Corresponding author. E‐mail: Bijan{at}discus.niams.nih.gov
View Full Text

Abstract

Transglutaminase (TGase) enzymes catalyze the formation of covalent cross‐links between protein‐bound glutamines and lysines in a calcium‐dependent manner, but the role of Ca2+ ions remains unclear. The TGase 3 isoform is widely expressed and is important for epithelial barrier formation. It is a zymogen, requiring proteolysis for activity. We have solved the three‐dimensional structures of the zymogen and the activated forms at 2.2 and 2.1 Å resolution, respectively, and examined the role of Ca2+ ions. The zymogen binds one ion tightly that cannot be exchanged. Upon proteolysis, the enzyme exothermally acquires two more Ca2+ ions that activate the enzyme, are exchangeable and are functionally replaceable by other lanthanide trivalent cations. Binding of a Ca2+ ion at one of these sites opens a channel which exposes the key Trp236 and Trp327 residues that control substrate access to the active site. Together, these biochemical and structural data reveal for the first time in a TGase enzyme that Ca2+ ions induce structural changes which at least in part dictate activity and, moreover, may confer substrate specificity.

Introduction

Transglutaminases (TGases; protein–glutamine:amine γ‐glutamyl‐transferases) are a family of calcium‐dependent acyl‐transfer enzymes that are ubiquitously expressed in mammalian cells. The common reaction performed by each of the nine TGase enzyme isoforms known in the human genome involves the activation of a target or donor protein‐bound glutamine residue to form a thiol‐acyl enzyme intermediate. This can be accepted by a nucleophilic group to accomplish the reaction. Of several potential nucleophiles, the most common is water, which results in the net deamidation of the target glutamine. However, if the nucleophile is the ϵ‐amino group of a protein‐bound lysine, an Nϵ‐(γ‐glutamyl)lysine intra‐ or interchain isopeptide cross‐link is formed (Lorand and Conrad, 1984; Folk and Chung, 1985; Greenberg et al., 1991; Melino et al., 1998; Nemes and Steinert, 1999), leading to the formation of a stable, permanent and insoluble macromolecular complex.

TGases are expressed abundantly in stratified squamous epithelia where they are required for formation of the cell envelope barrier structure. The cell envelope of specialized epithelia such as the epidermis involves the accretion of several precursor structural proteins (Rice and Green, 1977; Kim et al., 1990; Steinert and Marekov, 1995; Steinert et al., 1998; Nemes et al., 1999). To date, at least two different TGase enzymes are known to be required for effective barrier formation (Melino et al., 1998; Nemes and Steinert, 1999) and include: the TGase 1 enzyme, which is usually membrane bound (Rice and Green, 1977; Steinert et al., 1996); and the TGase 3 enzyme, which is cytosolic (Kim et al., 1990, 1993). Both enzymes require activation by proteolysis at specific sites, which increases their specific activities by 2–3 orders of magnitude. Both of these TGases are also widely expressed in other tissues, including brain (Kim et al., 1999) and muscle (Choi et al., 2000).

While several TGase isoforms that have been well characterized to date preferentially use the sequence motif Gln–Gln–Val for the first step of the reaction (Lorand and Conrad, 1984; Folk and Chung, 1985), there nevertheless is a high degree of isoform substrate specificity. However, the features that specify which glutamine donor and lysine acceptor substrates may approach an enzyme isoform to engage in reaction remain unknown. For example, the human blood clotting factor XIIIa (hfXIIIa) enzyme uses fibrin almost exclusively for both reaction steps, resulting in its polymerization (Pisano et al., 1968; Gladner and Nossal, 1983; Shen and Lorand, 1983). On the other hand, TGases 1 and 3 preferentially use different glutamines and lysines on the same substrates, such as loricrin and small proline‐rich proteins 1, 2 and 3 (Candi et al., 1995, 1999; Tarcsa et al., 1998; Steinert et al., 1999). Likewise, trichohyalin is recognized preferentially by TGase 3 (Tarcsa et al., 1997). However, the bases of these substrate preferences remain unresolved.

The requirement for Ca2+ ions for TGase reactions in in vitro solution assays and in situ on tissue sections has been known for decades (Folk and Chung, 1985; Lorand et al., 1987; Melino et al., 1998). However, the numbers of ions, their location in the three‐dimensional structures of TGases and precise role(s) in the activation and function are not yet understood. The three‐dimensional structures of the zymogen and activated forms of the hfXIIIa enzyme isoform have been solved (Yee et al., 1994, 1996; Weiss et al., 1998). The activated enzyme possesses a single Ca2+ ion located in the vicinity of residues 473–490 in the catalytic core domain (Fox et al., 1999). However, binding of a Ca2+ ion in this site does not significantly change the structure and does not result in activation per se, but may somehow contribute to the subsequent binding of substrates (Fox et al., 1999). The human TGase 2 enzyme can bind as many as six Ca2+ ions with an apparent average affinity constant of 90 μM (Bergamini, 1988), but their properties and location in the structure remain unknown. In addition, the structure of a fish TGase enzyme (fTG, perhaps equivalent to mammalian TGase 2) has been presented (Noguchi et al., 2001). The structure does not show a Ca2+ ion (Noguchi et al., 2001), but in vitro activity assays with it have routinely incorporated Ca2+ (Yasueda et al., 1994). Thus, the role of Ca2+ ions in TGase structure and function remains unresolved, but a role in activation and/or substrate specificity is plausible.

In this study, we have explored the functional and structural requirements of Ca2+ ions in the TGase 3 system. We demonstrate that binding of three Ca2+ ions is needed for activity. Furthermore, comparisons of the solved three‐dimensional structures of the zymogen and activated forms demonstrate that on binding of Ca2+ ions to the latter, a channel opens to expose two key residues near the active site. Thus, we conclude that binding of Ca2+ ions in the TGase 3 system facilitates substrate binding and perhaps specificity.

Results and discussion

Human TGase 3 binds three Ca2+ ions

In initial studies, we explored the Ca2+ requirements for the zymogen and activated forms of human TGase 3. First, by exhaustive equilibrium dialysis into buffer containing increasing concentrations of 45Ca2+, we found that the zymogen does not acquire label. However, direct quantitative biophysical measurements employing the inductively coupled plasma–mass spectrometry dynamic reaction cell (ICPMS‐DRC) (West Coast Analytical Service, Santa Fe Springs, CA) demonstrates the existence of 1.0 mol of Ca2+ion/mol and the absence of any other metals (Figure 1A). When we treated the zymogen with a solution of Chelex and EGTA, we were unable to remove significant amounts of it. However, denaturation in 8 M urea or 1% SDS released the Ca2+ ion, but the enzyme did not re‐acquire a Ca2+ ion on removal of the denaturants. This is probably due to improper refolding as the protein subsequently could not be activated. To identify the source of this ion, we expressed a mini‐culture of baculovirus (Kim et al., 2001) in the presence of 45Ca2+ and discovered that the zymogen sequesters a Ca2+ ion from the insect cells during expression. Next, we proteolyzed the TGase 3 zymogen with dispase and repeated the equilibrium dialyses. This TGase 3 enzyme is active and has acquired two additional Ca2+ ions (Figure 1A).

Figure 1.

The Ca2+ content of TGase 3. (A) Binding of Ca2+ to the zymogen and the activated form in the presence of Ca2+ ions was measured by equilibrium dialysis at 4°C. (B) Isothermal calorimetric titration of Ca2+ binding to TGase 3. To solutions of the zymogen and activated forms were added a series of injections of 5, 10 and 20 μl of CaCl2 and the heat absorbed or liberated was recorded (upper panel). The lower panel records the net heat changes. The heat of dilution determined from injections of a solution without protein were subtracted to determine heat changes due solely to Ca2+ binding. The data shown are representative; the SD of replicates was <5%.

The thermodynamics of the acquisition of the two Ca2+ ions were determined by isothermal titration calorimetry (Figure 1B). When the proteolyzed TGase 3 was assayed in the presence of CaCl2, the binding data indicate 2.0 low affinity sites (average Kd = 4 μM) with ΔH = −4.639 ± 0.15 kcal/mol. Such an exothermic reaction implies significant stabilization of the enzyme on binding. However, it has not been possible to determine the properties of the Ca2+‐binding site in the zymogen. Based on the avidity with which it acquires a Ca2+ ion during expression, we suspect that the Kd of this site should be in the nM range, typical of the [Cai2+] of eukaryotic cells.

The three Ca2+ ions are located in the 50 kDa portion of TGase 3

Activation of the zymogen requires specific cleavage at Ser469 (Kim et al., 1990), located in a hinge region that separates the catalytic core and β‐barrel 1 domains. Typically, the resultant 50 and 27 kDa fragments remain associated together in the active enzyme. This cleavage is replicated in vitro by treatment with dispase (Figure 2A). In addition, we have found that some cleavage can occur by physical manipulation of protein solutions, perhaps by autocatalysis, or contaminating proteases in buffers, so that the purified zymogen retains some activity (H.‐C.Kim and P.M.Steinert, unpublished observations).

Figure 2.

Location of Ca2+ ions in TGase 3. (A) SDS–PAGE. Lane 1, proteolytic activation of TGase 3, yielding 50 and 27 kDa fragments that normally remain associated together; lane 2, purified 50 kDa portion; lane 3, 77 kDa zymogen. (B and C) TGase 3 activities of the zymogen (lanes 1 and 2), the 77 kDa dispase‐proteolyzed form (lanes 7 and 8) and the separated 50 (lanes 5 and 6) and 27 (lanes 3 and 4) kDa forms were measured by incorporation of [14C]putrescine into casein (Folk and Cole, 1966). The − and + indicate the absence or presence of 5 mM Ca2+ in the standard assay reaction. (B) The proteins were repurified through a monoQ column and were not pre‐treated with 1 mM CaCl2. (C) The proteins were pre‐treated with 1 mM CaCl2 prior to repurification. In the controls, 5 mM EDTA or EGTA was added to the standard assays. (D) TGase 3 is active in the presence of lanthanides. The Ca2+ + and − row indicate the presence or absence of calcium in the reaction. The M3+ row denotes the presence of 1 mM lanthanide: lanes 1–3, Er3+; lanes 4 and 5, Sm3+; lanes 6 and 7, Tb3+; and lanes 8 and 9, Lu3+. In one experiment, 5 mM ErCl3 was used instead of CaCl2.

The following sets of experiments were performed to clarify the precise roles of Ca2+ ions. The dispase‐proteolyzed zymogen form was freed from dispase by one step through a monoQ column. The intact zymogen and the proteolyzed protein were also reacted with 1 mM CaCl2 for 1 h at 23°C and then rechromatographed. Similarly, the separated 50 and 27 kDa pieces were either treated or not treated with 1 mM CaCl2, and separated on the same column. Under these conditions, following ICPMS‐DRC analyses, the proteolyzed 77 kDa form and the 50 kDa fragment each contained one Ca2+ ion when not exposed to CaCl2, and three Ca2+ ions after exposure to CaCl2. However, the intact zymogen retained only one Ca2+ ion even if it had been treated with CaCl2. The 27 kDa fragment contained no Ca2+ ion (data not shown). Thus all three Ca2+ions are located in the 50 kDa part of TGase 3, consisting of the N‐terminal β‐sandwich and the catalytic core domains (see below).

The binding of two additional Ca2+ ions activates TGase 3

Each of these proteins that had not (Figure 2B) or had (Figure 2C) been pre‐treated with 1 mM CaCl2 was then assayed in the absence or presence of 5 mM CaCl2 in the standard TGase assay. The 27 kDa form had no activity under any conditions (Figure 2B and C, columns 3 and 4). The intact zymogen and the 50 and 77 kDa proteolyzed proteins possessed no activity until exposed to Ca2+: without pre‐treatment with 1 mM CaCl2, they were inactive (Figure 2B, columns 1, 5 and 7, respectively) until 5 mM CaCl2 was added to the assay (Figure 2B, columns 2, 6 and 8). The zymogen possessed 10% maximal activity, possibly due to random cleavage. However, the 50 and 77 kDa proteolyzed forms which had been pre‐treated with 1 mM CaCl2 possessed substantial activity (40–50% of maximum) in the absence of 5 mM CaCl2 in the assay (Figure 2C, columns 5 and 7). This activity increased toward 100% with 5 mM CaCl2. This increase may be due to an equilibrium exchange of either or both of the two Ca2+ ions: in the absence of CaCl2 in solution, the bound ions may partly vacate, resulting in net partial loss of activity.

Together, the data indicate that the tightly bound Ca2+ ion in the zymogen is necessary but not sufficient for enzyme activity. However, the acquisition of two additional Ca2+ ions in a pre‐treatment regimen, and in the absence of added CaCl2 in the assay, is sufficient to activate the enzyme or at least allow the enzyme reaction to proceed in the presence of substrate. Historically, therefore, use of the standard TGase 3 assay conditions with 2–5 mM CaCl2 (Kim et al., 1990) allows the binding of these two additional activating Ca2+ ions.

Other lanthanide trivalent cations can substitute for Ca2+ to retain TGase 3 activity

Other lanthanide trivalent cations can bind to hfXIIIa in the general vicinity of its sole Ca2+ ion site (Fox et al., 1999). Accordingly, the above experiments with the TGase 3 forms were repeated with four different lanthanide ions, Er3+, Sm3+, Tb3+ and Lu3+. First, we found that neither could replace the tightly bound Ca2+ ion of the zymogen. Secondly, pre‐treatment of the activated enzyme with 1 mM lanthanide, followed by purification through the monoQ column, and then assay for TGase activity in the absence of 5 mM CaCl2, revealed significant activity in each case, with somewhat higher activity with Sm3+ (Figure 2D). Thirdly, when we added standard 5 mM CaCl2 to the TGase assay, activities increased, with somewhat higher activity with Sm3+ (Figure 2D). Direct measurement and the X‐ray data (see below) of the Er3+ form of activated TGase 3 revealed the presence of a net of two Er3+ ions and one Ca2+ ion before addition of 5 mM CaCl2, and only three Ca2+ ions afterward. Thus, Ca2+ ions are the preferred cation for the TGase 3 enzyme and will displace other ions. Further, we incubated the Er3+ form in the presence of 5 mM ErCl3, and found high activity comparable with that for CaCl2 (Figure 2D), i.e. the one Ca2+/two Er3+ ion form is about as active as the three Ca2+ ion form. In addition, we have repeated these experiments with Yb3+, and found that TGase 3 acquires a net of one Yb3+ ion/mol, and the two Ca2+/one Yb3+ ion form is comparably active (data not shown). In contrast, incubation of hfXIIIa with >40 μM lanthanide inhibited this enzyme (Achyuthan et al., 1989), due to binding of a Yb3+ ion in a novel site (Fox et al., 1999).

Overall topology and domain structure of zymogen and activated forms of TGase 3

The X‐ray diffraction crystal structures of the zymogen and activated form (Kim et al., 2001) were solved in order to understand better the structural and functional significance of the Ca2+ activation properties of TGase 3. The refinement details are shown in Table I.

View this table:
Table 1. X‐ray diffraction data collection and refinement statistics

While the zymogen and activated enzyme function in solution as monomers (Kim et al., 1990), the X‐ray models of the human TGase 3 zymogen and the activated forms consist of two crystallographically independent monomers of 692 residues per asymmetric unit, with overall resolution of 2.2 and 2.1 Å, respectively (Figure 3). In the zymogen, one monomer has missing density for a flexible loop between residues 461 and 479, and the second monomer has missing density for residues 460–472. In the active enzyme, residues 460–480 are missing in both monomers. The r.m.s. differences between the Cα carbon atoms of the two monomers in the zymogen and the activated form are 0.39 and 0.30 Å, indicating that the Cα backbone structures are almost identical. The dimensions of each monomer are ∼110 × 66 × 51 Å, and the protein molecule is reminiscent of a flattened triangle. The activated form has one β‐octylglucoside molecule in the barrel 1 domain (Figure 3B), acquired during crystallization to prevent twinning (Kim et al., 2001).

Figure 3.

Stereo images of the solved structures. Solved structures of dimers of (A) human TGase 3 zymogen. (B) The activated form in the presence of Ca2+. (C) Superposition of Cα traces. Each is a pair of stereo diagrams showing the overall secondary structures. The four domains are the β‐sandwich (red), the catalytic core (green), barrel 1 (magenta) and barrel 2 (orange). The Ca2+ ions are shown in yellow. In (A), the Cl ion (cyan) between the β‐sandwich and core, and the Br ion (blue) in barrel 1 domains are shown. In (B), a single molecule of β‐octylglucoside is visible. In (C), the gray color represents the zymogen and the magenta color is the activated form. The three Ca2+ ion‐binding sites are numbered 1–3. The three changed loop regions of the activated form, Ile223–Val231, Leu319–Ser325 and Tyr439–Asp446, are shown in green. The Ca2+ ion in site 1 superimposes in both structures. (D) Stereo overlay images of the Cα traces of the solved structures of fXIIIa (red), fTG (green) and activated TGase 3 (yellow) enzymes. Calcium ions are also included. The cleavage site is marked by an arrow in (A).

The zymogen and activated forms have well‐organized structures consisting of four folded domains that are similar in organization and structure to the hfXIIIa (Yee et al., 1994) and fTG enzymes (Noguchi et al., 2001). The N‐terminal β‐sandwich domain of TGase 3 contains the first 134 amino acids and consists of nine strands of β‐sheets interspersed by three α‐helices. The catalytic core domain extends from residues Asn135 to Gly472 and consists of 15 β‐sheets interspersed with 15 α‐helical segments. The longest α‐helix of 16 residues is located in the center of the molecule and harbors the active site Cys272 residue. Other members of the catalytic triad, His330 and Asp353, are located on adjacent strands of β‐sheets. This catalytic triad is structurally similar to that of other TGases (Yee et al., 1994; Noguchi et al., 2001). The barrel 1 and barrel 2 domains span residues 473–592 and 593–692, respectively. The active site is shielded from contact with the solvent because it is buried within a narrow cleft with walls formed by two β‐sheet strands of the catalytic core domain and the C‐terminus of barrel 1. Residues 462–471 form a flexible solvent‐exposed loop that links the last α‐helical segment of the catalytic core domain to the first β‐strand of the barrel 1 domain. This hinge region harbors Ser469, the cleavage site for proteolytic activation of the zymogen. This residue is flanked by polar residues, predicted to lie near the surface of the protein, which may be involved in recognition by an activating protease.

The arrangement of the two monomers in the asymmetric unit of the zymogen and the activated form crystal structures are different from those of hfXIIIa (Yee et al., 1994; Weiss et al., 1998) or fTG (Noguchi et al., 2001). Notably, in TGase 3, three of the strands of the β‐pleated sheet of the barrel 2 domain of one monomer form intermolecular contacts with four strands of the β‐pleated sheet of the β‐sandwich domain of the second monomer. Also, there are salt bridge interactions between residue Glu65 of one monomer and Arg653 of the second monomer, hydrogen bonding of Asn50 with Glu621, and other intermolecular contacts that are mediated through water molecules. Accessible surface area calculations for the interface between the monomers in the asymmetric unit give a value of 5375 Å2 for the area buried upon formation of the two crystallographically independent monomers. As the protein is a monomer in solution (Kim et al., 1990), this interface may not be functionally relevant and is imposed by crystal contacts within the unit cells. In contrast, there is only one molecule of the fTG enzyme in its asymmetric unit (Noguchi et al., 2001).

There are two non‐proline cis peptide bonds in the zymogen and active forms that have been conserved in the hfXIIIa (Weiss et al., 1998) and fTG (Noguchi et al., 2001) enzymes. One is Arg268–Tyr269 in the loop prior to the strand of α‐helix that contains the active site Cys272 residue, and the other is Asn382–Phe383 in a loop adjoining two α‐helices of the core domain. The latter appears to stabilize a Gly367–Pro368 cis peptide bond.

Consistent with the above biochemical data, the crystal structure of the TGase 3 zymogen has one Ca2+ ion/monomer in the catalytic core domain (Figure 3A). This is retained in the activated form, which also has two additional ions in the catalytic core domain (Figure 3B). These sites are numbered 1–3 in Figure 3C. We found no evidence for other partially occupied cation‐binding sites.

The binding of three Ca2+ ions induces changes in the structure of activated TGase 3

Comparisons of the Cα atomic alignments of the solved TGase 3 zymogen and activated forms reveal high degrees of identity (Figure 3C), but there are four changes. One is located in a flexible surface loop in the N‐terminal β‐sandwich domain. Three other changes are observed in the catalytic core domain in the vicinity of each of the three Ca2+‐binding sites (colored green in Figure 3C). There are also high degrees of similarity in the Cα alignments of the solved hfXIIIa, fTG and activated TGase 3 enzymes (Figure 3D, colored red, green and yellow, respectively). A detailed analysis of these four solved structures is in progress.

The Ca2+ ion‐binding site 1 of the zymogen

This site encompasses residues Asn224–Asn229. The Ca2+ ion is positioned 13.7 Å above the Cα carbon atom of Cys272 in the active site (Figure 4A). The Ca2+ ion adopts an octahedral coordination by forming direct contacts with the main chain carbonyl oxygen atoms of Ala221, Asn224, Asn226 and Asn229, the carbonyl side chain oxygen of Asn224, and a water molecule. A summary of oxygen atoms that are potentially involved in the coordination of this (and the other two Ca2+) ions is presented in Table II. Interestingly, the location of this Ca2+ ion site is retained in the activated TGase 3 (Figure 4B), but there are changes. The Ca2+ ion is heptacoordinated in an environment that can be best described as a distorted pentagonal bipyramid by forming direct contacts with the main chain carbonyl oxygen atoms of Ala221, Asn224 and Asn226, the carbonyl side chain oxygen of Asn224 and Asp228, and a water molecule. The loop Ile223–Val231 containing Asn229 has shifted away and Asp228 instead coordinates with the Ca2+ ion. In this way, the Ca2+ ion in the activated TGase 3 is shielded by the carbonyl side chain oxygen of Asp228, which is now located in a tight turn between β‐strands and α‐helix and is an outlier in the Ramachandran plot. The Asp228 side chain in the zymogen form (φ = 47.6, ψ = 40.6) is exposed, while in the activated form it is buried (φ = −144.9, ψ = 19.7). In addition, the <B> value is decreased from 34.52 to 24.36 Å2. Indeed, the <B> value of this Ca2+‐binding site has been reduced from 31.30 to 19.11 Å2 (Table II). Together, these data show that the Ca2+ ion has been buried by the movement of the Asp228 side chain.

Figure 4.

Stereo views of the key residues involved in the Ca2+ ion‐binding sites in the zymogen and activated forms of TGase 3. (A and B) Site 1. The Ca2+ ion in the zymogen (A) forms direct contacts with the main chain carbonyl oxygen atoms of Ala221, Asn224, Asn226 and Asn229, the carbonyl side chain oxygen of Asn224, and a water molecule, but, in the activated form (B), the Ca2+ ion forms direct contacts with the main chain carbonyl oxygen atoms of Ala221, Asn224 and Asn226, the carbonyl side chains oxygen of Asn224 and Asp228 and a water molecule. In this case, the Ca2+ ion is shielded further by the loop Ile223–Val231 bearing Asp228. Asp228 in the activated form is buried, while in the zymogen it is exposed. (C) Sites 2 and 3. Site 2 in the activated TGase 3 exists near the end of the catalytic core domain by the α‐helical segment leading to the loop which connects to the barrel 1 domain, encompassing residues Asn430–Glu448. The Ca2+ ion forms contacts with the carbonyl side chain oxygen groups of Asn393, Glu443 and Glu448, the main chain carbonyl oxygen atoms of Ser415 and two water molecules. The third Ca2+‐binding site in the activated TGase 3 occurs in the vicinity of the loop segment leading to the catalytic His330 active site. This Ca2+ ion forms contacts with the carbonyl side chain oxygens of Asp301, Asp303 and Asp324, the side chain atoms of Asn305, the main chain carbonyl oxygen atoms of Ser307 and a water molecule. The loop bearing residues Leu319–Ser325 has moved from its position in the zymogen, thereby creating a channel through the molecule. In all images, gray represents the water oxygen atom.

View this table:
Table 2. The distance and B‐values of all the residues within 3.5 Å of the specified calcium ion site

There is a high probability of calcium being the metal present in the crystal structure of the zymogen in this position. First, the zymogen acquires a Ca2+ ion during expression, and the ICPMS‐DRC analyses indicate 1.0 Ca2+ ion/mol and the absence of any other metals. Secondly, the mean peak size in the |Fo| − |Fc| difference electron density map was at the 10σ level (large positive difference peak) for each Ca2+ ion in each independent monomer in the zymogen and the activated forms. Thirdly, this Ca2+ ion site was investigated further by close inspection of anomalous electron density maps. The clarity of the 2|Fo| − |Fc| electron density maps, the observed bond distances and the coordination geometry strongly support the assignment of this Ca2+‐binding site. In refinement modeling analyses, when one water molecule replaced the Ca2+ ion at this point, the <B> value refined to a value of ∼3 Å2, implying that a water molecule in this location does not have enough electrons to match the X‐ray diffraction data adequately. Finally, the compared <B> values as well as bond–valence calculations using parameters from Brese and O'Keeffe (1991) result in a valence of 2.3 for a Ca2+ ion in the zymogen and the activated form, supporting this assignment as a calcium site.

Site 1 of TGase 3 is generally similar to a novel Yb3+ site found at the dimer 2‐fold axis near residues Asp270 and Glu272 of hfXIIIa (Figure 3D), where binding results in inhibition of this enzyme. Asp270 is conserved as Asp227 in TGase 3, but Glu272 is replaced by Asn229. The residue Asn229 coordinates with the Ca2+ ion in the zymogen, but, in activated TGase 3, the loop bearing this residue moves away so that a stronger coordination forms with Asp228 instead. As the equivalent loop bearing Glu272 does not change in hfXIIIa on activation (Fox et al., 1999), this may explain why a lanthanide ion in this site inhibits hfXIIIa but does not affect activated TGase 3.

This Ca2+ ion may be required for maintenance of the correct three‐dimensional structure of the active site region of TGase 3. Further, its enhanced coordination after enzyme activation suggests that it is essential for enzyme activity.

The Ca2+ ion‐binding site 2 in activated TGase 3 is homologous to that in hfXIIIa

One of the two new Ca2+‐binding sites acquired in the activated TGase 3 exists near the end of the catalytic core domain near the α‐helical segment leading to the loop which connects to the barrel 1 domain, encompassing residues Asn430–Glu448. This Ca2+ ion is 23 Å below the Cα carbon atom of the active site Cys272 residue. The site resembles that of the sole known Ca2+‐binding site of hfXIIIa and is thought to adopt an EF‐hand‐like conformation (Fox et al., 1999). The Ca2+ ion adopts a heptacoordinated conformation forming direct contacts with the carbonyl side chain oxygen groups of Asn393, Glu443 and Glu448, the main chain carbonyl oxygen atoms of Ser415, and direct ligation to two water molecules (Figure 4C; Table II). The residues involved in the coordination with Ca2+ ion binding are different from those of hfXIIIa. In hfXIIIa, the Ca2+ ion lies in an acidic pocket formed by Asp438, Glu485 and Glu490, but the Ca2+ ion coordinates only with the main chain carbonyl oxygen of Ala457 and four water molecules. However, in both enzymes, binding of a Ca2+ ion promotes no (hfXIIIa) or only minor (TGase 3) changes in shape that do not affect in any obvious way the conformations of residues near the active site.

We have also crystallized the activated TGase 3 with Er3+ and Yb3+ complexes in the absence of Ca2+ ions. The refinement analyses revealed that the activated form of TGase 3 contained a net of two Er3+ ions or one Yb3+ ion. The mean peak sizes in the |Fo| − |Fc| difference electron density maps were found to be at the 15σ level (large positive difference peak) for both metals. The anomalous electron density maps indicate that the Yb3+ ion is located in site 2, and the two Er3+ ions are located in sites 2 and 3 of TGase 3. In each case, they engage the same oxygen atoms as the Ca2+ ions, as described above (data not shown: refinement analyses in progress). These data imply that Ca2+ ions in site 2 can exchange with other cations more easily than site 3.

The binding of Ca2+ ion in the third binding site in activated TGase 3 opens a channel toward the active site

The third Ca2+ ion‐binding site acquired in the activated TGase 3 enzyme is novel. It occurs in the vicinity of a loop segment consisting of residues Asp320–Ser325 leading to the catalytic residue His330 in the active site. This Ca2+ ion is 22 Å from the Cα carbon atom of the active site Cys272 residue and 17 Å from the Cα carbon atom of His330. The precise location of this loop is different in the zymogen when compared with the activated TGase 3. In the latter, the loop has moved and becomes directly bonded to the Ca2+ ion via the carbonyl side chain oxygen of Asp324, and thereby opens a channel through the protein. The Ca2+ ion adopts a heptacoordinated stack with direct contacts with the carbonyl side chain oxygens of Asp301, Asp303 and Asp324, the side chain atoms of Asn305, the main chain carbonyl oxygen atoms of Ser307 and a water molecule (Figure 4C, Table II).

The channel is clearly visible on the electrostatic surface potential image (Figure 5). Figure 6A shows an enlargement of this zone revealing that on one side the channel is conical in shape, extending from the surface of the protein toward the catalytic triad. In addition, the guanidinium side chain of Arg396 changes its orientation to form a hydrogen bond with Glu586 in the activated form, which allows the hole to pass through to the reverse side (Figure 6B). The movement of the loop and opening of the channel expose Trp236 and Trp327 on the upper and outer surface of the channel (Figure 6A).

Figure 5.

Structural comparison of the zymogen and activated form. The upper and lower rows represent images rotated 180° with respect to each other, to show the channel in the active form. The electrostatic surface potential maps are shown in the center. The acidic and basic regions are colored red and blue, respectively. The arrow denotes a channel that opens on binding of a Ca2+ ion in site 3. On the left and right are secondary structure images of the zymogen and the activated TGase 3 in the same orientations. The electrostatic potential, including calcium ions, has been mapped onto the surface plan from −15.0 kT (deep red) to +15.0 kT (deep blue).

Figure 6.

Stereo view images of the novel channel opened by Ca2+ ion binding in site 3. Protein domains are colored as in Figure 3. (A) On one side, the electrostatic surface potential (black transparent) shows that the active site triad residues Cys272, His330 and Asp353 are buried and inaccessible. The movement of the loop bearing residues Asp320–Ser325 opens the channel in the activated TGase 3, and exposes the side chains of the Trp236 and Trp327 residues. (B) On the opposite side, the guanidinium group of Arg396 has moved to form a salt bridge with Glu586, allowing the hole to extend through the protein.

Notably, in the hfXIIIa and fTG enzymes, the equivalents of these tryptophan residues are predicted to play multiple important roles in the enzyme reactions. First, they are buried and must be exposed. Secondly, movements of their side chains would allow direct access of substrates to the buried active site. However, how these two residues become exposed so the conformations of their side chains can change to allow enzyme reaction has not been shown. Thirdly, these two tryptophan residues are thought to participate in oxyanion intermediate formation with first the acyl glutamine donor and then the acceptor lysine substrates (Pedersen et al., 1994). Because of near identities of sequences, we suggest that these predicted events should also occur in the TGase 3 system. Thus, based on our new data, we can predict specifically that the Ca2+ ion‐induced formation of the channel and exposure of the two key tryptophan residues will facilitate the movements of their side chains as appropriate substrates approach. A complementary possibility is that the opening of this channel in the TGase 3 system confers specificity on which substrates can approach the trytophan residues to engage the active site.

Conclusions

We have successfully solved the structures of two forms of a second mammalian TGase enzyme, the inactive TGase 3 zymogen and its activated form. Our X‐ray data reveal the presence of three Ca2+ ions that are required for enzyme activity. The presence and function of the three ions have been confirmed by biochemical and biophysical assays. Our new data provide the first insights on the structural role that Ca2+ ions play in any of the TGase isoforms. The Ca2+ ion‐binding site 1 in the zymogen is tightly bound and cannot be removed without protein denaturation. Because its coordination with residue side chains changes following activation, the data suggest that its presence is essential for activity. However, it is not sufficient for activity. For activation, two additional Ca2+ ions are required that change the structure of the enzyme. Further, their exothermic binding properties imply significant net enzyme stabilization. Our data imply that the intact loop of residues 462–471 in the zymogen (not visualized in Figure 3A) occludes binding of Ca2+ ions in sites 2 and 3, which is removed on cleavage at Ser469. Of these, site 2 is similar to the sole site of hfXIIIa, and binding of a Ca2+ ion causes only a minor change in structure that has no discernible effect on access to the active site. However, binding of a Ca2+ ion in site 3 creates a channel which exposes the two key Trp236 and Trp327 residues that control access of substrates to the active site of TGase enzymes. We therefore suggest that the binding of the three Ca2+ ions, and especially at site 3, facilitates approach of substrates and/or directly activates the enzyme. Indeed, it is also possible that these changes reflect or regulate the unique substrate specificity properties of TGase 3. Finally, detailed comparisons of the high resolution structures of these two TGase 3 enzyme isoforms with those of the solved hfXIIIa and fTG enzymes are now in progress (see also preliminary comparisons in Figure 3C and D). Coupled with the new insights into the critical role of Ca2+ ions in the activation process of the TGase 3 system, such analyses may reveal other changes that reflect TGase isoform substrate specificity and the mechanism of action.

Materials and methods

Protein expression, crystallization and data collection

Recombinant human TGase 3 zymogen was expressed in the baculovirus system and purified as described previously (Kim et al., 2001). To prepare the activated form, a 12 ml solution of zymogen (17 mg/ml in a buffer of 20 mM Tris–HCl pH 8.0 containing 1 mM EDTA and 125 mM NaCl) was treated with 3 mg of dispase (Roche) at 37°C for 30 min. The active TGase 3 form was then recovered by passage through a monoQ column with a linear gradient of 0–500 mM NaCl, and eluted at 100 mM NaCl (Kim et al., 2001). Whenever the zymogen is digested with dispase, most stays as the 77 kDa (activated) form, but a minority of the protein falls apart as the 27 and 50 kDa fragments. The fragments were eluted from the column at 125 and 350 mM NaCl, respectively.

Crystallization trials of the zymogen were performed using the hanging drop vapor diffusion method by adding the non‐ionic detergent β‐octylglucoside (0.1 mM) to 100 μl of protein in 20 mM Tris–HCl pH 8.0, 1 mM EDTA and 125 mM NaCl at a concentration of 17 mg/ml. To a fresh 2 μl of protein, 2 μl of precipitant solution was added and equilibrated over a well solution containing 4% (w/v) PEG 20K and 100 mM Tris–HCl pH 8.5 at 15°C. The condition used for the activated TGase 3 in the presence of Ca2+ ions is somewhat different from that previously reported for the activated TGase 3 in the absence of calcium (Kim et al., 2001). The crystallization of TGase 3 (15 mg/ml) was done using 0.1 mM β‐octylglucoside in 100 μl of protein in 20 mM Tris–HCl pH 8.0, 1 mM EDTA, 125 mM NaCl and 3.0 mM CaCl2. To a fresh 4 μl of protein, 2 μl of reservoir solution was added and equilibrated over a well solution containing 4–12% (w/v) PEG 6K, 100 mM bicine pH 9.0 and 1% dioxane at 21°C.

Thin, plate‐like crystals were harvested into mother liquors containing 20% (v/v) 2‐methyl‐2,4‐pentanediol as the cryoprotectant, and were then flash‐cooled into liquid nitrogen with a rayon mounting loop (Oxford Cryosystems, Oxford, UK) prior to data collection. All data were collected at the National Synchrotron Light Source, Brookhaven National Laboratories, using beamline X9B, the ADSC Quantum 4 detector with a wavelength of 0.92 Å, crystal to detector distance of 205 mm and oscillation range of 1.0°, and exposure for 60 s. The Matthew's coefficient is 2.88 Å3/Da for the zymogen and 2.68 Å3/Da for the activated TGase 3, suggesting two molecules per asymmetric unit, with solvent content of 58 and 56%, respectively (Matthews, 1968). The diffraction data from sets of crystals were indexed, processed, scaled and merged using the HKL2000 suite of programs (Otwinowski and Minor, 1997). An attempt was made to collect multiwavelength anomalous dispersion data in which halide ions are introduced as anomalous scatterers in the cryoprotectant during a short soak (Dauter et al., 2000). The crystal was soaked in 1 M NaBr for 45 s, but no anomalous scatterers could be obtained. However, in the zymogen structure, one bromide ion was detected.

Structure solution and refinement

The crystal structure of the zymogen was solved by molecular replacement using AmoRe (Navaza and Saludjian, 1997) and a resolution range of 10–3.5 Å for the rotation search and between 10 and 3.0 Å for the translation search. There were two solutions to both the rotation and translation searches. These solutions gave rise to two molecules within each asymmetric unit. The polyalanine model of the hfXIIIa enzyme (PDB 1GGT; Yee et al., 1994) was used as the initial search model. The initial R‐factor after correctly orienting and positioning the two molecules was 44.60% for all the data from 10 to 3.0 Å resolution. The rigid‐body refinement using CNS software (Brünger et al., 1998), followed by simulated annealing applying strict 2‐fold non‐crystallographic symmetry (NCS) constraints (Kleywegt, 1996) to 3.0 Å resolution, improved the solutions, yielding a final R‐factor of 39.8%. Subsequently, the polyalanine model was converted to the TGase 3 sequence using the program O (Jones et al., 1991). Initial phases were improved using solvent flattening and histogram matching (CCP4, 1994) with 2‐fold NCS molecular averaging (Kleywegt and Read, 1997). For final refinement, the NCS restraints were released. After many cycles of manual model rebuilding into SIGMAA‐weighted 3|Fo| − 2|Fc| and |Fo| − |Fc| maps, the refinement converged at an R‐factor of 18.20% and an Rfree of 22.50%. This included all low‐resolution data from 25 Å resolution as bulk solvent correction. In order to avoid overfitting of the model, each step of the rebuilding procedure was monitored using the free R‐factor and a residue real space correlation coefficient as a guide. For the activated TGase 3 data set with the additional Ca2+ ions, starting with one of the monomers of the preliminary zymogen model as the search probe, molecular replacement in AmoRe (Navaza and Saludjian, 1997) was used to search for the location of monomer in this crystal form in P1 space group. A translation search yielded the correct solution (CC = 38.2%). The model was refined in CNS (Brünger et al., 1998) and with the simulated annealing, positional and B‐factor refinement protocols using a maximum‐likelihood target (Brünger et al., 1990). Anisotropic scaling and a bulk solvent correction were used, and the individual B‐factors were refined isotropically. Except for the Rfree set (a random sampling consisting of 8.6% of the data set), all data between 20 and 2.1 Å (with no σ cut‐off) were included in the refinement which was converged at an R‐factor of 18.87% and Rfree of 23.34% (Table I). The stereochemical quality of the structure was analyzed with the programs PROCHECK (Laskowski et al., 1992) and CNS (Brünger et al., 1998). Figures were generated with Molscript (Kraulis, 1991) and Raster3D (Merritt and Bacon, 1997).

Microcalorimetric titration studies of TGase 3

The thermodynamic properties of Ca2+ binding to the activated TGase 3 were measured by isothermal titration calorimetry using a MicroCal VP‐ITC calorimeter. Prior to measurements, the TGase 3 (1 μM) was dialyzed three times against 20 mM Tris–HCl pH 8.0 and 2 mM EGTA. The EGTA solution was removed by extensive dialysis against 20 mM Tris–HCl pH 8.0 and treated with Chelex 100. This TGase 3 was placed in a 1.38 ml sample cell. A 250 μl syringe loaded with 0.5 mM standard CaCl2 solution (Fisher Scientific) was used for a series of automatic injections of 10 μl each into the protein solution. After each injection, a 5 min pause was allowed for reaching the baseline. Heat produced due to dilution was measured by injecting the Ca2+ solution into the sample cells from which TGase 3 protein was omitted. For each titration step, the heat of dilution was subtracted. Data were fitted to appropriate binding models, and thermodynamic parameters were determined from non‐linear least‐squares fits, using ORIGIN software.

TGase 3 activity assay

TGase activity was measured by incorporation of [14C]putrescine into casein (Folk and Cole, 1966). Assays were done at 37°C in 0.5 ml portions of 0.1 M Tris–HCl pH 7.5, containing 1% N,N‐dimethylated casein (Sigma), 0.5 μCi putrescine (118 mCi/mmol), 1 mM dithiothreitol, 5 mM CaCl2 and 1 mM EDTA. In some experiments, the assays contained 1 mM lanthanide, ErCl3, SmCl3, TbCl3 or LuCl3, instead of CaCl2.

TGase 3 calcium‐binding assay

Calcium binding was determined by equilibrium dialysis (Bergamini, 1988) allowing equilibration of the enzyme (0.5–0.8 mg/ml) against 100 vols of buffer containing 20 mM Tris–HCl pH 8.0 supplemented with increasing amounts of 45CaCl2 (300 c.p.m./nmol) up to 10 mM. After 48 h at 4°C, aliquots of the enzyme solution and of the dialysate were withdrawn for liquid scintillation counting. The results were normalized to the protein concentration determined by absorbance at 280 nm.

TGase 3 quantitative metal analysis

All plastic and glassware was incubated in 5 M HCl for 36 h and rinsed with double‐distilled water that had been passed over a Chelex 100 column. For buffer and transition metal solutions, Chelex 100‐treated water was used throughout. Prior to adjustment of the pH, Chelex 100 resin was added (0.5 g/100 ml) and the solution was stirred for 45 min; after adjustment of the pH, the solution was passed through a 0.2 μm filter and stored in metal‐free plastic vials at −20°C. Prior to measurements, the TGase 3 (10 mg/ml) was dialyzed five times against 20 mM Tris–HCl pH 8.0 and 5 mM EGTA. The EGTA solution was removed by dialysis against 20 mM Tris–HCl pH 8.0. The bound metal ions were detected by use of a Perkin‐Elmer ELAN 6100 ICPMS‐DRC apparatus. The concentration of zymogen was 1.5 mg/ml.

Accession numbers

The structure factors and the coordinates of zymogen and the activated TGase 3 have been deposited in the Protein Data Bank with the accession number 1L9M for zymogen and 1L9N for the activated TGase 3.

Acknowledgements

We thank Craig Hyde for his initial support, William Idler and Karen Boeshans for expert assistance, and Zbigniew Dauter at Brookhaven NSLS for assistance with the data collection at X9 Beamline.

References

View Abstract