Many prokaryotic soluble PPases (pyrophosphatases) contain a pair of regulatory adenine nucleotide-binding CBS (cystathionine β-synthase) domains that act as ‘internal inhibitors’ whose effect is modulated by nucleotide binding. Although such regulatory domains are found in important enzymes and transporters, the underlying regulatory mechanism has only begun to come into focus. We reported previously that CBS domains bind nucleotides co-operatively and induce positive kinetic co-operativity (non-Michaelian behaviour) in CBS-PPases (CBS domain-containing PPases). In the present study, we demonstrate that a homodimeric ehPPase (Ethanoligenens harbinense PPase) containing an inherent mutation in an otherwise conserved asparagine residue in a loop near the active site exhibits non-co-operative hydrolysis kinetics. A similar N312S substitution in ‘co-operative’ dhPPase (Desulfitobacterium hafniense PPase) abolished kinetic co-operativity while causing only minor effects on nucleotide-binding affinity and co-operativity. However, the substitution reversed the effect of diadenosine tetraphosphate, abolishing kinetic co-operativity in wild-type dhPPase, but restoring it in the variant dhPPase. A reverse serine-to-asparagine replacement restored kinetic co-operativity in ehPPase. Molecular dynamics simulations revealed that the asparagine substitution resulted in a change in the hydrogen-bonding pattern around the asparagine residue and the subunit interface, allowing greater flexibility at the subunit interface without a marked effect on the overall structure. These findings identify this asparagine residue as lying at the ‘crossroads’ of information paths connecting catalytic and regulatory domains within a subunit and catalytic sites between subunits.

INTRODUCTION

Nearly 200 sequenced prokaryotic genomes contain genes for putative nucleotide-regulated soluble PPases (pyrophosphatases) (EC 3.6.1.1). These enzymes belong to the so-called family II of soluble PPases [1,2]: homodimeric Mn2+ or Co2+ metalloenzymes that additionally require Mg2+ for PPi hydrolysis [3]. Each subunit of a canonical family II PPase is formed by two domains connected by a flexible linker (Figure 1A), with the active site located between the domains [4,5]. This active site location makes the enzyme ideally suited for allosteric regulation. CBS-PPases [CBS (cystathionine β-synthase) domain-containing PPases] [6] differ from canonical family II PPases by having a ∼250-residue insert formed by two CBS domains (a Bateman module [7]), often intercalated by a DRTGG domain (Figures 1B and 1C). The crystal structure of the isolated regulatory insert suggests that CBS and DRTGG domains participate in the subunit interface, where the two pairs of CBS domains form a disc-like structure [8] (Figure 1D). Although each CBS domain contains a potential binding cavity for nucleotides, the Bateman module accommodates only two AMP molecules or one molecule of Ap4A (diadenosine tetraphosphate), which binds through both of its adenosine moieties [8]. Remarkably, AMP and ADP inhibit CBS-PPase, whereas ATP and Ap4A activate it [6,9].

Structural aspects of family II PPases

Figure 1
Structural aspects of family II PPases

(AC) Domain topologies of non-regulated family II PPase (A), ehPPase (B) and typical CBS-PPase, such as dhPPase (C). DHH and DHHA2 are catalytic domains; CBS1, CBS2 and DRTGG domains form a regulatory region within the DHH domain. Linker sequences are shown in black. (D) Modelled three-dimensional structure of dimeric CBS-PPase [8] based on the structures of homologous catalytic domains of non-regulated family II PPase and the regulatory region of CBS-PPase. Domain colours of one subunit are the same as in (C); the other subunit is coloured grey. Bound AMP molecule is shown in green. (E) Partial sequence alignment of six CBS-PPases. Residue numbering is for cpPPase. Complete CBS1 and CBS2 domains comprise residues 70–122 and 247–303, respectively; residues 311–319 belong to the DHH domain. Nucleotide-binding residues identified in the cpPPase structure [8] are marked with arrows, and important charged residues and motifs are presented in boxes. Presumed important differences in ehPPase sequence are marked in green. The consensus sequences based on 85% and 50% identities in 191 CBS-PPases are shown below. cnPPase, Clostridium novyi CBS-PPase; elPPase, E. lenta CBS-PPase; mtPPase, M. thermoacetica CBS-PPase.

Figure 1
Structural aspects of family II PPases

(AC) Domain topologies of non-regulated family II PPase (A), ehPPase (B) and typical CBS-PPase, such as dhPPase (C). DHH and DHHA2 are catalytic domains; CBS1, CBS2 and DRTGG domains form a regulatory region within the DHH domain. Linker sequences are shown in black. (D) Modelled three-dimensional structure of dimeric CBS-PPase [8] based on the structures of homologous catalytic domains of non-regulated family II PPase and the regulatory region of CBS-PPase. Domain colours of one subunit are the same as in (C); the other subunit is coloured grey. Bound AMP molecule is shown in green. (E) Partial sequence alignment of six CBS-PPases. Residue numbering is for cpPPase. Complete CBS1 and CBS2 domains comprise residues 70–122 and 247–303, respectively; residues 311–319 belong to the DHH domain. Nucleotide-binding residues identified in the cpPPase structure [8] are marked with arrows, and important charged residues and motifs are presented in boxes. Presumed important differences in ehPPase sequence are marked in green. The consensus sequences based on 85% and 50% identities in 191 CBS-PPases are shown below. cnPPase, Clostridium novyi CBS-PPase; elPPase, E. lenta CBS-PPase; mtPPase, M. thermoacetica CBS-PPase.

CBS domains are quite common in Nature and are found, for instance, in 75 human and eight Escherichia coli proteins. Mutations in CBS domains of human enzymes and membrane channels are associated with hereditary diseases, including homocystinuria, retinitis pigmentosa, Bartter syndrome and osteopetrosis [10]. Typically, CBS domains bind adenine nucleotides, such as S-adenosylmethionine, AMP, ADP and ATP, as regulatory ligands [1014]. Despite the physiological importance of regulation by CBS domains, details of the underlying molecular mechanisms have only begun to emerge, mainly because only limited structural data are available owing to difficulties in crystallization. Recent studies have revealed that nucleotides induce significant structural changes in CBS [1517] and the magnesium transporter CNNM2 [18]. Significant differences have also been observed in the structures of the isolated regulatory insert of cpPPase (Clostridium perfringens PPase) in complex with AMP (inhibitor) or Ap4A (activator) [8].

Much less is known about the ways in which these changes propagate from CBS domains to the catalytic site and between subunits. We reported previously that four CBS domain-containing PPases, with or without DRTGG domains, exhibit positive co-operativity both in substrate hydrolysis and nucleotide regulation [19]. Specifically, hydrolysis kinetics exhibit non-Michaelian behaviour, requiring two Michaelis constants (Km2<4Km1, positive kinetic co-operativity) for its description. Furthermore, the dependence of activity on nucleotides at fixed substrate concentrations indicates that AMP, ADP and ATP binding is also positively co-operative, whereas Ap4A binding is non-co-operative, as is the hydrolysis kinetics in its presence [9].

In the present paper, we describe ehPPase (Ethanoligenens harbinense PPase), which contains two CBS domains, but no DRTGG domain, in its structure. This enzyme is interesting because it contains inherent mutations in both regulatory and catalytic domains, in regions that are conserved in the majority of CBS-PPases. Furthermore, preliminary measurements revealed no co-operativity in ehPPase-mediated catalysis of PPi hydrolysis. The effect of the mutation in the catalytic domain on enzyme function was characterized by introducing a similar mutation in a previously characterized CBS-PPase from Desulfitobacterium hafniense (dhPPase) and a reverse mutation in ehPPase. A detailed characterization of ehPPase and mutant PPases allowed the identification of an asparagine residue responsible for modulating kinetic co-operativity and nucleotide regulation in CBS-PPases.

EXPERIMENTAL

Materials

The sources and characteristics of the adenine nucleotide preparations have been described previously [9]. The concentrations of nucleotide stock solutions were estimated by measuring absorbance at 259 nm (ε=15900 M−1·cm−1 for mononucleotides and 31800 M−1·cm−1 for dinucleotides).

Genetic manipulations

Samples of genomic DNA extracted from D. hafniense and E. harbinense (DSM numbers 10664 and 18485 respectively) were obtained from the DSMZ (Deutsche Sammlung von Microorganismen und Zellkulturen GmbH). The sequences of dhPPase (residues 1–544; GenBank® accession number YP_002458003) and ehPPase (residues 1–444; GenBank® accession number YP_004090372) were amplified by PCR and ligated into NdeI and XhoI restriction sites of the pET28bTEV plasmid (Novagen) and into NdeI and SacI sites of the pET28b plasmid (Novagen).

Site-directed mutagenesis was performed using overlap extension PCR. Mutant genes were cloned into the same plasmids to produce constructs that expressed an N312S variant of dhPPase and an S213N variant of ehPPase containing a TEV (tobacco etch virus)/thrombin-cleavable His6 tag at their N-termini. All constructs were confirmed by sequencing.

Protein expression and purification

All genetic constructs were expressed in E. coli, and the CBS-PPases thus produced were purified as described previously [19]. The final products were at least 95% pure, as estimated by SDS/PAGE using a Phast system with 8–25% gradient gels (GE Healthcare). Protein concentrations were determined with a Nanodrop spectrophotometer (Thermo Scientific) using A0.1%280 values of 0.477 and 0.478 for dhPPase and ehPPase respectively, as calculated from their amino acid compositions with ProtParam. Molar concentrations were calculated on the basis of the subunit molecular masses of 60.5 and 49.8 kDa respectively. Frozen protein solutions were stored at −40 to −80°C.

Kinetic assays

The activity assay medium contained 5 mM MgCl2, 140 μM PPi (yielding 50 μM MgPPi complex) and 0.1 M Tes/KOH (pH 7.2), except where specified otherwise. For measurements made at higher Mg2+ concentrations, buffer concentration was decreased appropriately to maintain constant ionic strength. Reactions were initiated by adding enzyme, and Pi accumulation due to PPi hydrolysis was recorded continuously for 2–3 min at 25°C using an automated Pi analyser [20]. Initial velocities of PPi hydrolysis were estimated from the slopes of the tangents to the initial portion of the Pi accumulation curves.

Isothermal calorimetry (ITC)

Heat production was measured at 25°C in a VP-iTC calorimeter (MicroCal). Enzyme and nucleotide solutions were prepared in 0.1 M Mops/KOH (pH 7.2) buffer containing 2 mM MgCl2, 0.1 mM CoCl2 and 150 mM KCl. Titrations were performed by successive 10 or 20 μl injections of 100 μM mononucleotide or 33 μM dinucleotide solution into 1.4 ml of CBS-PPase solution (10–20 μM in terms of the monomer); the interval between injections was 5 min. Measured heat values were corrected for ligand dilution effects.

Bioinformatics analysis

CBS-PPase protein sequences were retrieved by BLAST analysis of the NCBI Protein [21] and KEGG Genes [22] databases. Protein sequences were aligned with ClustalX (version 2.1) using default settings. The alignment was processed manually by eliminating incomplete and redundant sequences and sequence regions that included indels and ambiguously aligned residues.

Molecular dynamics (MD)

MD simulations were carried out using dimeric bsPPase (Bacillus subtilis PPase), whose crystal structure with bound PNP (imidodiphosphate), Mg2+ and F ions (PDB code 2HAW) was previously determined at a resolution of 1.75 Å (1 Å=0.1 nm) [23]. A virtual N77S mutation (corresponding to the dhPPase N312S mutation) was performed using the Coot program, version 0.8.1 EL CCP4-6 [24]. To account for the marked distortion in the geometry of the bsPPase-bound PNP molecule [23] in comparison with that of non-bound (Mg)2PNP [25] (for instance, the P−N−P angle is 15° greater in the bsPPase complex), the harmonic force constants for the P−N−P, N−P−O and O−P−O angles were increased by factors of 3, 5 and 5 respectively, and the torsion barrier for the rotation around the P−N bond was increased 4-fold. These factors were selected by trial calculations and were the minimal values that ensured complex stability during MD simulations. The systems were hydrogenated, electroneutralized by adding K+ ions using the program TLEAP of the AMBER 14 package (http://ambermd.org/), and solvated with an 8 Å TIP3P water molecule in a final box volume of approximately 730000 Å3.

MD simulations were performed using the PMEMD program of the AMBER 14 software suite with the ff14SB force field (http://ambermd.org/). The total number of atoms in the simulations was approximately 65000. Systems were minimized using a SANDER function with 1000 steps of steepest descent and 2000 steps of conjugate gradient with the restraint of 500 kcal·mol−1·Å−2 (1 kcal=4.184 kJ) on the protein atoms. The systems were gradually heated to 300 K under constant volume conditions with 500 kcal·mol−1·Å−2 restraints on the protein atoms that were gradually removed. Temperature was kept constant using a Langevin thermostat. The 8 Å cut-off production MD was performed for 650 ns, with snapshots saved every 10 ps. Hydrogen bonds were detected with the program HBonanza [26] using default settings. Other post-processing trajectory analyses were carried out with the program CPPTRAJ of the AMBER 14 pack.

Calculations and data analysis

The values of apparent dissociation constants for the various metal ion complexes used to calculate free metal ion concentrations at pH 7.2 were as follows: MgPPi, 112 μM [27]; Mg2PPi, 2.84 mM [27]; MgAMP, 10.5 mM (http://www.nist.gov/srd/nist46.cfm); MgADP, 0.42 mM (http://www.nist.gov/srd/nist46.cfm); and MgATP, 0.034 mM (http://www.nist.gov/srd/nist46.cfm). The values for adenine nucleotides were derived from published values of the corresponding pH-independent constants.

Most non-linear least-square fits were performed using the program Scientist (Micromath). ITC data were analysed using a MicroCal add-on routine for Origin version 7.0.

Co-operative kinetics of substrate (MgPPi) hydrolysis was analysed with eqn (1), which assumes different Michaelis constants (Km1 and Km2) and equal kcat values for two active sites in the dimer:

 
formula
1

where [E]0 and [S] are total enzyme and substrate concentrations respectively.

The dependences of Km1 and Km2 on Mg2+ concentration ([M]) were fitted to eqn (2):

 
formula
2

where (Km)0 and (Km)M are the limiting values of the respective Km at zero and infinite Mg2+ concentrations respectively, and KM is the metal-binding constant.

The dependences of hydrolysis rate on nucleotide concentration ([N]) were fitted to eqn (3):

 
formula
3

where v0 and vN are the activities of free and nucleotide-saturated enzyme respectively, and KN1 and KN2 are the apparent dissociation constants describing successive binding of nucleotide to two regulatory sites per enzyme molecule respectively. The corresponding binding schemes and details of the fitting procedure were as described previously [19].

Alternatively, rate dependences on substrate and nucleotide concentrations were fitted to the Hill-type eqn (4):

 
formula
4

where L is S or N, vL is the rate at infinite [L], and h is the Hill coefficient. The value of v0 was set to 0 when L was substrate, and the value of h was set to 1 for non-co-operative binding.

RESULTS

ehPPase contains mutations in conserved motifs

A search of the KEGG Genes database as of November 2015 revealed 183 bacterial and eight archaeal CBS-PPase sequences, 423–581 residues in length. All are expected to encode active PPases as they contain all of the polar residues found in the active sites of canonical family II PPases [4,5]. In all CBS-PPases found, the regulatory part is inserted within the catalytic DHH domain, and in the majority of CBS-PPases is formed by two CBS domains and one DRTGG domain (Figure 1C). However, four CBS-PPases found in the KEGG Genes database and 19 CBS-PPases found in the NCBI Protein database, including those of E. harbinense, Moorella thermoacetica and Eggerthella lenta, contain two CBS domains but no DRTGG domain in their structures (Figure 1B).

A comparison of the primary structure of ehPPase with those of five CBS-PPases characterized previously revealed two other important differences (Figure 1E). First, several inherent mutations in ehPPase involve residues corresponding to nucleotide-binding residues in the structures of the regulatory part of cpPPase complexed with AMP or Ap4A [8]. Thus cpPPase Lys100 (interacts with nucleotide Pα) is glycine in ehPPase. Furthermore, cpPPase Val258 and the RYRN loop, which participate in stacking, hydrophobic and hydrogen-bonding interactions with the adenine moiety, are replaced by aspartate and SAGR in ehPPase. Although CBS domains generally exhibit low conservation between species, these particular replacements involve functionally important residues and are therefore expected to severely affect nucleotide binding in ehPPase. Secondly, the highly conserved DHNE motif found in the catalytic part of five other CBS-PPases in Figure 1E and in the majority of other putative CBS-PPases is replaced by DHSE in ehPPase. A similar replacement was observed only in the CBS-PPases of Holdemania filiformis and Rubrobacter xylanophilus. Interestingly, the asparagine-to-serine mutation is much more common in canonical family II PPases that lack the CBS domain and is encountered in approximately half of the nearly 400 reported family II PPase sequences.

ehPPase lacks kinetic co-operativity and is not regulated by adenine nucleotides

As in our previous study on CBS-PPases [19], we used two types of parameters to characterize co-operativity: the Hill coefficient and macroscopic binding constants, or Km, for successive binding. The value of the Hill coefficient is 1 for non-co-operative kinetics and greater or less than 1 for the cases of positive and negative co-operativity respectively. The ratio of the macroscopic binding constants is 4 for the case of two identical and independent sites, greater than 4 for positive co-operativity, and less than 4 for negative co-operativity [28]. Because the Michaelis–Menten equation is formally analogous to the binding equation, these rules can also be used for the Michaelis constants; this type of co-operativity is usually called ‘kinetic co-operativity’ as opposed to true ‘binding co-operativity’. Given the known Mg2+ ion-dependence of co-operativity in CBS-PPases [19], kinetic characterizations were performed over a wide range of Mg2+ concentrations (0.05−20 mM).

On the basis of the above criteria, wild-type ehPPase exhibits quite low, if any, kinetic co-operativity, a result at variance with the behaviour of the four previously characterized CBS-PPases [19]. Thus the Hill coefficient for ehPPase differed insignificantly from unity over the entire range of metal cofactor concentrations (Figure 2). The Km2/Km1 ratio was slightly less than 4, but again the difference was insignificant, as this parameter is a more sensitive measure of co-operativity.

Mg2+ concentration-dependence of kinetic co-operativity in ehPPase and S213N-ehPPase

Figure 2
Mg2+ concentration-dependence of kinetic co-operativity in ehPPase and S213N-ehPPase

Shown from top to bottom are the catalytic constant kcat, the Michaelis constants Km1 and Km2, and the Hill coefficient h. Enzyme identities are indicated above the panels. Results are mean±S.E.M. computed parameter values. The lines for Km1 and Km2 show the best fits for eqn (2). Enzyme concentration was 1−100 pM.

Figure 2
Mg2+ concentration-dependence of kinetic co-operativity in ehPPase and S213N-ehPPase

Shown from top to bottom are the catalytic constant kcat, the Michaelis constants Km1 and Km2, and the Hill coefficient h. Enzyme identities are indicated above the panels. Results are mean±S.E.M. computed parameter values. The lines for Km1 and Km2 show the best fits for eqn (2). Enzyme concentration was 1−100 pM.

Remarkably, the S213N substitution conferred significant negative co-operativity on this enzyme, decreasing the Hill coefficient to 0.7, and the Km2/Km1 ratio increased to approximately 13 over the entire range of Mg2+ concentrations (Figure 2). This effect mainly resulted from a decrease in Km1. The substitution preserved the dependence of both Km values on [Mg2+], as well as the value of the Mg2+ binding constant that governs this dependence (Table 1). The catalytic constant decreased by a factor of 4.5 upon Ser213 replacement.

Table 1
Kinetic parameters for PPi hydrolysis derived from the dependences of kcat, Km1 and Km2 on Mg2+ concentration (Figures 2 and 3)

NA, not available.

 kcat dependence Km1 dependence Km2 dependence 
Enzyme kcat,0 (s−1kcat,M (s−1KM (mM) (Km1)0 (μM) (Km1)M (μM) KM (mM) (Km2)0 (μM) (Km2)M (μM) KM (nM) 
ehPPase NA 1850±50 NA 6.7±0.3 3.6±0.5 4±3 38±1 17±1 6±1 
S213N-ehPPase NA 410±20 NA 2.3±0.3 0.6±0.3 5±6 29±4 10±4 3±4 
dhPPase* NA 320±20 NA 4.7±0.5 60±20 0.6±0.2 NA 11±1 NA 
N312S-dhPPase 4.9±0.3 10.0±0.1 0.097±0.009 0.63±0.12 3.00±0.06 0.07±0.02 1.7±0.2 13±1 0.049±0.008 
dhPPase+Ap4A†‡ NA 620±20 NA NA 3.5±0.3 NA NA 14±4 NA 
N312S-dhPPase+Ap4A‡ 25±3 169±2 0.50±0.04 33±1 13.5±0.6 0.8±0.2 5.5±0.4 2.2±0.4 3±2 
N312S-dhPPase+AMP§ 4.3±0.6 6.6±0.2 0.22±0.19 2.67±0.03 0.82±0.03 4.6±0.6 9.4±0.5 2.8±0.1 11±10 
 kcat dependence Km1 dependence Km2 dependence 
Enzyme kcat,0 (s−1kcat,M (s−1KM (mM) (Km1)0 (μM) (Km1)M (μM) KM (mM) (Km2)0 (μM) (Km2)M (μM) KM (nM) 
ehPPase NA 1850±50 NA 6.7±0.3 3.6±0.5 4±3 38±1 17±1 6±1 
S213N-ehPPase NA 410±20 NA 2.3±0.3 0.6±0.3 5±6 29±4 10±4 3±4 
dhPPase* NA 320±20 NA 4.7±0.5 60±20 0.6±0.2 NA 11±1 NA 
N312S-dhPPase 4.9±0.3 10.0±0.1 0.097±0.009 0.63±0.12 3.00±0.06 0.07±0.02 1.7±0.2 13±1 0.049±0.008 
dhPPase+Ap4A†‡ NA 620±20 NA NA 3.5±0.3 NA NA 14±4 NA 
N312S-dhPPase+Ap4A‡ 25±3 169±2 0.50±0.04 33±1 13.5±0.6 0.8±0.2 5.5±0.4 2.2±0.4 3±2 
N312S-dhPPase+AMP§ 4.3±0.6 6.6±0.2 0.22±0.19 2.67±0.03 0.82±0.03 4.6±0.6 9.4±0.5 2.8±0.1 11±10 

*From Salminen et al. [19].

†From Anashkin et al. [9].

‡Ap4A concentration was maintained at 10 μM.

§AMP concentration was maintained at 50 μM.

Another unusual feature of ehPPase was its insensitivity (≤10% activity change) to adenine nucleotides (AMP, ADP, ATP and diadenosine polyphosphates, ApnA, with n=3–6) over a wide range of substrate concentrations (1–300 μM), metal cofactor concentrations (0.05–20 mM), and nucleotide concentrations (0.01–1000 μM for mononucleotides and 10−5–100 μM for dinucleotides). The activities of other CBS-PPases, including dhPPase, changed severalfold in analogous experiments with all or most nucleotides [9,19]. The S213N substitution did not restore nucleotide sensitivity in ehPPase.

N312S substitution in dhPPase abolishes kinetic co-operativity and reverses the effect of Ap4A on it

Studies of the N312S-dhPPase variant, which mimics wild-type ehPPase, provided more convincing evidence for the role of this asparagine residue in the kinetic co-operativity. This substitution reduced the catalytic constant 30-fold and made it, as well as the Km2 value, Mg2+ concentration-dependent. However, the effect on Km1 was more important in terms of co-operativity: this parameter fell by an order of magnitude while retaining its Mg2+ concentration-dependence. As a consequence, the strong positive kinetic co-operativity characteristic of wild-type dhPPase [19] disappeared completely over the entire range of Mg2+ concentrations, as indicated by the fact that the Hill coefficient and the 4Km1/Km2 ratio approached unity (Figure 3). The values of the dissociation constants governing Mg2+ binding to the site controlling kcat and both Km values were similar within the error of determination and equalled 0.05–0.1 mM (Table 1).

Mg2+ concentration-dependence of kinetic co-operativity in N312S-dhPPase in the absence and presence of 50 μM AMP or 10 μM Ap4A

Figure 3
Mg2+ concentration-dependence of kinetic co-operativity in N312S-dhPPase in the absence and presence of 50 μM AMP or 10 μM Ap4A

Nucleotide identities are indicated above the panels. The broken lines refer to the early measured dependences for wild-type dhPPase in the absence and presence of Ap4A [9,19]. Enzyme concentration was 1−4 nM. Other details are as for Figure 2. 

Figure 3
Mg2+ concentration-dependence of kinetic co-operativity in N312S-dhPPase in the absence and presence of 50 μM AMP or 10 μM Ap4A

Nucleotide identities are indicated above the panels. The broken lines refer to the early measured dependences for wild-type dhPPase in the absence and presence of Ap4A [9,19]. Enzyme concentration was 1−4 nM. Other details are as for Figure 2. 

To estimate the effects of nucleotide binding on kinetic co-operativity in the variant dhPPase, we again measured the rate-dependences on substrate concentration at a series of fixed Mg2+ concentrations, but now in the presence of Ap4A or AMP (Figure 3). The nucleotide concentrations used were saturating, as increasing them did not cause additional effects. The data shown in Figure 3 thus refer to nucleotide-bound enzyme. These data demonstrate two remarkable effects of Ap4A. First, it increased the catalytic constant severalfold, but this effect was similar to that observed previously with wild-type dhPPase [9]. Secondly, and in striking contrast with the behaviour of wild-type dhPPase, Ap4A-bound N312S-dhPPase exhibited strong positive kinetic co-operativity: Km1 exceeded Km2 over the full range of Mg2+ concentrations tested. The Hill coefficient and 4Km1/Km2 ratio changed accordingly, increasing to 1.6 and 20−25 respectively (Figure 3). This was an unexpected result because the effect of Ap4A on dhPPase and other CBS-PPases was the opposite: it abolished their kinetic co-operativity. Furthermore, both Km values decreased with increasing [Mg2+] in N312S-PPase, whereas Km1 increased and Km2 remained constant in wild-type dhPPase [19]. Importantly, the N312S substitution did not change the dhPPase far-UV CD spectrum (Supplementary Figure S1) and sedimentation pattern (Supplementary Figure S2), ruling out a large change in tertiary or quaternary structure as an explanation of these effects.

In contrast, AMP binding did not induce kinetic co-operativity in N312S-dhPPase (Figure 3). This nucleotide only decreased the catalytic constant by half and changed the dependence of Km values on Mg2+ concentration.

In the presence of 50 μM ATP and 5 mM Mg2+, the hydrolysis kinetics of N312S-dhPPase also remained non-co-operative, the catalytic constant increased 1.5-fold, and the average Km value (√Km1Km2) increased ∼5.5-fold (from 6±1 to 33±5 μM). Accordingly, ATP acted as an inhibitor of N312S-dhPPase at low substrate concentrations. With the wild-type enzyme, hydrolysis kinetics remained positively co-operative in the presence of ATP, the catalytic constant was similarly increased, but the √Km1Km2 was decreased 2-fold by ATP [19], making it an activator at all substrate concentrations [19].

N312S-dhPPase retains nucleotide-binding co-operativity

The dependence of N312S-dhPPase activity on the nucleotide concentration was measured at fixed substrate and metal cofactor concentrations. First, we estimated the effects of a series of nucleotides at their saturating concentrations (Table 2). Qualitatively, the effects were similar to those reported for wild-type dhPPase [19], but were much larger in most cases. That the effect of ATP changed its sign is explained by a larger increase in the Km values in N312S-dhPPase, which over-compensated for the small increase in the catalytic constant at non-saturating substrate concentrations.

Table 2
Nucleotide effects on N312S-dhPPase activity measured at 50 μM MgPPi and 5 mM Mg2+
  AN/A0
Nucleotide Concentration (μM) N312S-dhPPase Wild-type dhPPase 
AMP 100 0.45±0.01 0.037±0.001† 
ADP 100 0.53±0.01 0.09±0.01† 
ATP 100 0.75±0.01 2.5±0.2† 
Ap310 8.6±0.3 2.54±0.01‡ 
Ap410 13.0±0.6 3.0±0.1‡ 
Ap510 6.0±0.1 3.32±0.06‡ 
Ap610 1.92±0.03 2.58±0.06‡ 
  AN/A0
Nucleotide Concentration (μM) N312S-dhPPase Wild-type dhPPase 
AMP 100 0.45±0.01 0.037±0.001† 
ADP 100 0.53±0.01 0.09±0.01† 
ATP 100 0.75±0.01 2.5±0.2† 
Ap310 8.6±0.3 2.54±0.01‡ 
Ap410 13.0±0.6 3.0±0.1‡ 
Ap510 6.0±0.1 3.32±0.06‡ 
Ap610 1.92±0.03 2.58±0.06‡ 

*Values of Ao were 9.2 and 220 s−1 for N312S-dhPPase and wild-type dhPPase respectively.

†From Salminen et al. [19].

‡From Anashkin et al. [9].

Nucleotide-binding co-operativity was analysed using AMP and Ap4A as the most effective modulators of N312S-dhPPase. Figure 4 presents typical activity profiles measured with 50 μM substrate in the presence of 5 mM Mg2+. The parameter values derived with eqns (3) and (4) from these and analogous profiles measured at different substrate concentrations are summarized in Table 3. As indicated by the h and KN2/KN1 values, AMP binding was positively co-operative, whereas Ap4A bound non-co-operatively over the whole range of substrate concentrations, as was the case for the wild-type enzyme [9,19]. The value of the average inhibition constant for AMP (√KN1KN2) was ∼4-fold lower for the mutant enzyme (0.33 compared with 1.39 μM [19]). For Ap4A, √KN1KN2 values were similar for the two enzyme forms at low substrate levels, markedly increased with increasing substrate concentration for the mutant enzyme, and slightly decreased for the wild-type enzyme. In addition, the degree of activation by Ap4A (AN/A0) changed in opposite directions for the two enzyme forms when substrate concentration was varied.

AMP and Ap4A concentration-dependence of N312S-dhPPase activity measured at fixed concentrations of substrate (50 μM MgPPi) and Mg2+ (5 mM)

Figure 4
AMP and Ap4A concentration-dependence of N312S-dhPPase activity measured at fixed concentrations of substrate (50 μM MgPPi) and Mg2+ (5 mM)

The lines show the best fits to eqn (3). Enzyme concentration 6 nM (AMP data) or 0.6 nM (Ap4A data). Activity without nucleotides (9.2 s−1) was taken as 100%. The broken lines show the analogous dependences measured previously for wild-type dhPPase [19].

Figure 4
AMP and Ap4A concentration-dependence of N312S-dhPPase activity measured at fixed concentrations of substrate (50 μM MgPPi) and Mg2+ (5 mM)

The lines show the best fits to eqn (3). Enzyme concentration 6 nM (AMP data) or 0.6 nM (Ap4A data). Activity without nucleotides (9.2 s−1) was taken as 100%. The broken lines show the analogous dependences measured previously for wild-type dhPPase [19].

Table 3
Parameters describing nucleotide effects on N312S-dhPPase at different substrate concentrations in the presence of 5 mM Mg2+

Values in parentheses are from previous publications and describe AMP [19] and Ap4A [9] effects on wild-type dhPPase activity. NA, not available.

Nucleotide MgPPi (μM) AN/A0 (nM) KN1 (nM) KN2 (nM) KN1KN2 h 
AMP 1.00±0.02 NA NA NA NA 
 50 0.44±0.01 (0.037) 2000±2000 520±50 330±10 (1390) 2.4±0.2 
 300 0.65±0.02 (0.035) 2800±900 70±20 440±10 (1830) 1.84±0.04 
Ap44.2±0.2 (18) 7±1 28±3 14±1 (12.1) 1.02±0.04 
 50 13±1 (3.0) 92±5 370±20 186±5 (4.9) 1.00±0.02 
 300 12±1 (1.9) 140±10 550±60 270±10 (4.3) 1.00±0.04 
Nucleotide MgPPi (μM) AN/A0 (nM) KN1 (nM) KN2 (nM) KN1KN2 h 
AMP 1.00±0.02 NA NA NA NA 
 50 0.44±0.01 (0.037) 2000±2000 520±50 330±10 (1390) 2.4±0.2 
 300 0.65±0.02 (0.035) 2800±900 70±20 440±10 (1830) 1.84±0.04 
Ap44.2±0.2 (18) 7±1 28±3 14±1 (12.1) 1.02±0.04 
 50 13±1 (3.0) 92±5 370±20 186±5 (4.9) 1.00±0.02 
 300 12±1 (1.9) 140±10 550±60 270±10 (4.3) 1.00±0.04 

ITC analysis of nucleotide binding

Effects of the N312S substitution in dhPPase on AMP and Ap4A binding were also estimated by ITC (Figures 5A and 5B). These results are summarized in Table 4, along with previously published data for wild-type dhPPase [9]. KN and, accordingly, TΔS values could not be estimated with adequate precision in Ap4A titrations because the binding was very tight. Perhaps the only effect observed with AMP was approximately a 2-fold greater binding constant in the variant enzyme. The values of ΔH and TΔS, where determinable, were similar, as were binding stoichiometries. Both enzyme forms were able to bind twice as many AMP molecules as Ap4A, in accordance with the crystal structure of the regulatory part of cpPPase, indicating that one molecule of Ap4A occupies two AMP-binding sites [8].

ITC measurements of nucleotide binding to CBS-PPases

Figure 5
ITC measurements of nucleotide binding to CBS-PPases

Enzyme concentration was 16 μM. (A) Typical raw data for successive injections of AMP into an N312S-dhPPase solution. (B) Integrated heats for N312S-dhPPase titration by AMP (open circles) or Ap4A (closed circles) after correction for dilution. The lines show the best fits to a single-binding-site model. (C) Same for titration of wild-type ehPPase.

Figure 5
ITC measurements of nucleotide binding to CBS-PPases

Enzyme concentration was 16 μM. (A) Typical raw data for successive injections of AMP into an N312S-dhPPase solution. (B) Integrated heats for N312S-dhPPase titration by AMP (open circles) or Ap4A (closed circles) after correction for dilution. The lines show the best fits to a single-binding-site model. (C) Same for titration of wild-type ehPPase.

Table 4
Thermodynamic parameters for nucleotide complexes of CBS-PPases obtained by ITC
Enzyme/nucleotide KN (μM) n ΔH (kcal·mol−1TΔS (kcal·mol−1
dhPPase*     
 AMP 0.76±0.25 0.79±0.05 −5.6±0.5 −2.7±0.3 
 Ap4 0.41±0.01 −10.4±0.3  
N312S-dhPPase     
 AMP 1.43±0.19 0.75±0.06 −5.6±0.6 −2.4±0.3 
 Ap4 0.40±0.06 −10.9±0.7  
Enzyme/nucleotide KN (μM) n ΔH (kcal·mol−1TΔS (kcal·mol−1
dhPPase*     
 AMP 0.76±0.25 0.79±0.05 −5.6±0.5 −2.7±0.3 
 Ap4 0.41±0.01 −10.4±0.3  
N312S-dhPPase     
 AMP 1.43±0.19 0.75±0.06 −5.6±0.6 −2.4±0.3 
 Ap4 0.40±0.06 −10.9±0.7  

*From Anashkin et al. [9].

No heat production (ΔH <0.3 kcal/mol by absolute value) was observed in similar ITC experiments measuring AMP and Ap4A binding to both wild-type ehPPase and its S213N variant, confirming that neither is able to bind these nucleotides.

Modelling and MD simulations

To gain insight into the structural role of the asparagine residue, we analysed the three-dimensional structure of homodimeric bsPPase [23]. The subunit of this family II PPase is formed by only two domains, which are homologous with the catalytic DHH and DHHA2 domains of CBS-PPases (44% residue identity and 75% residue similarity between bsPPase and the catalytic part of dhPPase), has an identical active site, and a corresponding nearby asparagine residue (Asn77) in the conserved DHNE motif. The enzyme exhibited low, but significant, kinetic co-operativity (h=1.14±0.02). The structure of bsPPase determined at a 1.75 Å resolution (Figure 6A) contains an active-site-bound PNP, a PPi analogue that has NH instead of O in the bridge position; four Mg2+ ions; and an F ion, replacing the nucleophilic OH ion [23]. Before simulations, all organic molecules and sulfate ions were removed from both subunits, the F ion was manually replaced by OH to better mimic the productive enzyme–substrate complex, and the PNP molecule was removed from one subunit to mimic the state at which substrate binding co-operativity is manifested. The free energies of wild-type bsPPase and its virtual N77S variant were then minimized in an aqueous environment by performing MD simulations. Among other things, this procedure was expected to relieve possible structural distortions of the crystalline state. Accordingly, the simulated structures demonstrated minor variations from the crystal structure (RMSD for non-hydrogen backbone atoms of ∼2 Å and ∼3 Å for wild-type bsPPase and N77S-bsPPase respectively) (Supplementary Figure S3). More importantly, the simulated structures revealed significant differences in the subunit contact region.

Structures of B. subtilis PPase and its N77S variant

Figure 6
Structures of B. subtilis PPase and its N77S variant

(A) The crystal structure of dimeric bsPPase in complex with PNP (sienna sticks) and four Mg2+ ions (green spheres) [23]. Protein domains (shown in different colours), loop 96−109, β5 and β6 strands, and helix α5 are marked in one subunit. Asn77 is shown as red spheres in both subunits. (B and C) Loop 96−109 contacts in the simulated structures of the wild-type (B) and variant (C) bsPPase. (D) Comparison of the DHH–DHH domain pairs in the simulated structures of wild-type (green) and variant (violet) bsPPase. The structures were superposed on their left-hand DHH domains that contained bound substrate (red sticks). The curved arrow shows the direction of rotation of the right-hand DHH domain in the variant enzyme compared with that in the wild-type enzyme. (E and F) The distribution of RMSF values for side-chain residues along polypeptide chain in the simulated structures of wild-type (E) and variant (F) bsPPase for subunits A (black) and B (red). The boxed areas contain residues 72–125, which participate in the active site and subunit interface. (A)(D) were created using UCSF Chimera [29].

Figure 6
Structures of B. subtilis PPase and its N77S variant

(A) The crystal structure of dimeric bsPPase in complex with PNP (sienna sticks) and four Mg2+ ions (green spheres) [23]. Protein domains (shown in different colours), loop 96−109, β5 and β6 strands, and helix α5 are marked in one subunit. Asn77 is shown as red spheres in both subunits. (B and C) Loop 96−109 contacts in the simulated structures of the wild-type (B) and variant (C) bsPPase. (D) Comparison of the DHH–DHH domain pairs in the simulated structures of wild-type (green) and variant (violet) bsPPase. The structures were superposed on their left-hand DHH domains that contained bound substrate (red sticks). The curved arrow shows the direction of rotation of the right-hand DHH domain in the variant enzyme compared with that in the wild-type enzyme. (E and F) The distribution of RMSF values for side-chain residues along polypeptide chain in the simulated structures of wild-type (E) and variant (F) bsPPase for subunits A (black) and B (red). The boxed areas contain residues 72–125, which participate in the active site and subunit interface. (A)(D) were created using UCSF Chimera [29].

Intersubunit contacts are formed in bsPPase by the β6 strand and the loop (residues 96−109) that connects strands β5 and β6 of the DHH domains (Figure 6A). As the interacting β6 strands of two subunits are part of a continuous 12-strand β-sheet that runs through both DHH domains, the loop is a more likely candidate for the transfer of conformational changes across the subunit interface. In accordance with this prediction, the N77S replacement did not affect four hydrogen bonds connecting the non-crystallographic symmetry-related β6 strands, but severely disturbed the interactions between the loops. In the crystal structure of wild-type bsPPase [23], the loop forms the symmetrical hydrogen-bond pairs, Ile100−Thr105 (O…N and N…Oγ) and Arg99−Thr105 (Nω…O), and a single Asn102−Glu104 hydrogen bond (N…Oε) with the non-crystallographic symmetry-related loop of the other subunit and only one intrasubunit Arg79−Asn102 (N…O) hydrogen bond. In the simulated structure of wild-type bsPPase, two symmetrical intersubunit Arg79–Asn102 hydrogen bonds were replaced by an additional Asn102−Glu104 (N…Oε) hydrogen bond and by two pairs of symmetrical hydrogen bonds Arg99–Glu104 (Nω…Oε) and Phe103–Phe103 (N…O) (Figure 6B).

In sharp contrast, the loop lost most of the aforementioned intersubunit hydrogen bonds in the mutant bsPPase: only the Ile100−Thr105 (O…N and N…Oγ) and Arg99−Thr105 (Nω…O) bonds were detected (Figure 6C). Instead, enhanced intrasubunit hydrogen-bonding via Arg112 and Glu129 was evident between helix α5 and strand β6 compared with the wild-type structure. As a consequence, the DHH domains rotated in the mutant bsPPase relative to each other by ∼15° (Figure 6D).

The dramatic change in the intersubunit hydrogen-bonding clearly resulted from a changed hydrogen-bonding of residue 77. In the crystal structure of wild-type bsPPase, the Asn77 side chain of each subunit participates in three hydrogen bonds within the same subunit: with Asp96 (Nδ…Oδ2), Arg99 (Nδ…O) and Ala101 (Oδ…N). Arg99 and Ala101 lie within the loop, whereas Asp96 is part of the conserved DHH motif in the active site. All of these hydrogen bonds of Asn77 were retained in the simulated structure of wild-type bsPPase (Figure 6B), whereas the replacement residue, Ser77, formed only one hydrogen bond: with Asp96 (Oγ…Oδ2) (Figure 6C). Furthermore, the serine hydrogen bond places this residue differently with respect to Asp96 because of different sizes of serine and asparagine side chains. Consequently, the DHNE segment in the wild-type and DHSE segment in the mutant bsPPase adopted nearly identical structures (RMSD of 0.4 Å) but were moved apart by ∼4 Å in the mutant enzyme.

Reduced hydrogen-bonding at the subunit interface suggested increased flexibility in the N77S variant. This inference is supported by an analysis of residue mobility, as characterized by RMSF (root mean square fluctuation) (Figures 6E and 6F). The profiles of RMSF along the length of the polypeptide were qualitatively similar for subunits within each protein and between the proteins. However, the fluctuations were significantly smaller in the variant protein, consistent with its reduced subunit interdependence. Notably, residues 72–125 (boxed in Figures 6F and 6G) involved in the subunit interface as well as the catalytic site DHNE/DHSE loop and DHH motif displayed the smallest fluctuations.

DISCUSSION

According to the ‘autoinhibition’ concept introduced by Janosík et al. [30], the CBS domain insert acts as an ‘internal inhibitor’, sensitizing proteins with catalytic and transport functions to structural changes caused by nucleotide binding and permitting either activation or further inhibition. CBS-PPases provide strong support for this theory because nucleotide-regulated PPases are one to two orders of magnitude less active [19] than their non-regulated family II counterparts [3]. Furthermore, many mutations in CBS domains activate CBS-PPases, and some even reverse the effects of the nucleotide from inhibition to activation [31]. Similar effects of mutations have been reported for CBS [32] and AMPK (AMP-dependent protein kinase) [33,34]. In the framework of the ‘autoinhibition’ concept, the N312S mutation in dhPPase, now in the catalytic part, further inhibited PPase internally, allowing larger effects of most activators and smaller effects of inhibitors, as indicated in Table 2. Consistent with this idea, the activities of the AMP-inhibited wild-type and variant dhPPases were similar (8 and 4 s−1 respectively), despite the very large difference in their basal activities (220 and 9.2 s−1 respectively). Notably, ehPPase is more active than any other CBS-PPase [19] and in this respect resembles nucleotide-insensitive family II PPases [3], possibly indicating that ‘autoinhibition’ is absent from or decreased in ehPPase because its CBS domains are inherently corrupted.

Previous studies of CBS-PPases [8,19] have revealed added complexities in CBS domain-mediated regulation by nucleotides, showing that effector binding and catalysis proceed co-operatively in the dimeric enzyme. Accordingly, CBS-PPase regulation involves as many as three communication pathways [19]: one between the regulatory and catalytic sites (effects of nucleotides on activity), a second between the regulatory sites (nucleotide-binding co-operativity), and a third between the catalytic sites (catalytic co-operativity). The results reported above identify the asparagine residue in the conserved DHNE motif as being involved in all three pathways and emphasize its importance for communication involving catalytic sites.

That the asparagine residue is required for cross-talk between the catalytic sites is evident from the elimination of kinetic co-operativity upon its replacement by serine in dhPPase and by the lack of kinetic co-operativity in ehPPase containing a similar inherent mutation. Back replacement of serine with asparagine in ehPPase partly restored kinetic co-operativity, although the sign was reversed, providing additional support for the importance of asparagine for kinetic co-operativity. In contrast, the asparagine-to-serine replacement did not affect nucleotide-binding co-operativity, changed KN for AMP binding only ∼2-fold in substrate-free dhPPase (Table 4), and had a minor effect on the √KN1KN2 value (an average nucleotide-binding constant that equals KN in the case of non-co-operative binding) for Ap4A binding in the presence of 1 μM substrate (i.e. under conditions where substrate-free enzyme predominates) (Table 3). Notably, the role of kinetic co-operativity in CBS domain-mediated regulation has not been widely recognized. However, Labesse et al. [35] have reported that IMP dehydrogenase binds its substrate, IMP, with positive co-operativity, which is abolished by mutation of a single residue in the CBS domain.

By controlling kinetic co-operativity and hence catalytic site occupancy, the asparagine residue indirectly affects regulatory site performance, as is evident from a changed √KN1KN2 value in the enzyme–substrate complex compared with the substrate-free enzyme. The asparagine-to-serine replacement reversed the effect of substrate on Ap4A binding such that increasing substrate concentration from 1 μM to saturating levels decreased the √KN1KN2 value for Ap4A binding approximately 3-fold in the wild-type dhPPase, but increased it approximately 15-fold in the mutant dhPPase (Table 3). As a consequence, √KN1KN2 values for Ap4A binding differed 40–60-fold in their enzyme–substrate complexes (Table 3). The change in the substrate effect on AMP binding upon asparagine replacement was somewhat smaller: √KN1KN2 for the mutant dhPPase was 4-fold smaller in the enzyme–substrate complex (Table 3), but KN, an equivalent of √KN1KN2, was 2-fold greater in the free enzyme (Table 4), summing to an overall 8-fold difference. In other words, the asparagine replacement reversed heterotropic co-operativity from positive to negative with Ap4A (substrate stimulated Ap4A binding in the wild-type dhPPase, but suppressed it in the mutant enzyme) and vice versa with AMP.

Modelling and MD simulation experiments provided a structural explanation for the role of asparagine in active-site interdependence in CBS-PPases. Because the three-dimensional structure of CBS-PPase is unknown, these experiments were performed using the homologous CBS domain-lacking bsPPase, whose crystal structure is known. This choice is justified by the previous observation that deletion of the regulatory region does not eliminate the catalytic activity of dhPPase [19] or other CBS domain-containing proteins [30,36,37]. On the other hand, the use of such a ‘truncated’ PPase has as an advantage the fact that interpreting the effects of residue replacement on the communication between catalytic sites is less biased. The modelled structures indicated substantially depressed hydrogen-bonding in the specific area that separates the catalytic site and subunit interface in the asparagine-to-serine variant of bsPPase, hampering the transfer of structural changes (inherent in co-operativity) between the DHH domains. This specific area also appears to be involved in communication between the regulatory and active sites in CBS-PPases, as best indicated by the qualitative change in the Ap4A effect on kinetic co-operativity upon asparagine replacement in dhPPase. Notable in this regard, the regulatory insert is found in the DHH domain, only seven positions from the asparagine residue. The DHHA2 domain is connected to the DHH domain by a flexible linker, making its participation in the regulation phenomenon unlikely.

The central role of the asparagine residue in CBS-PPase regulation apparently stems from its closeness to the active site. One of the hydrogen bonds formed by Asn77 in wild-type bsPPase and its virtual variant is with Asp96, an important active-site residue that positions His98 to act as a general acid by protonating the leaving group phosphate [5]. Furthermore, Asn77 is separated by only one residue from Asp75, perhaps the most important residue in PPase catalysis. Asp75 bridges two metal ions that, together with a third metal ion, form a unique ‘trimetal site’ that co-ordinates the nucleophilic water and converts it into a highly reactive OH ion [23]. Two immediate neighbours of Asn77 in the conserved DHNE motif, His76 and Glu78, are equally important for catalysis, as their conservative substitutions in bsPPase diminished its catalytic activity more than 100-fold [38].

Why does Ap4A eliminate kinetic co-operativity in the wild-type CBS-PPase, but induce it in the variant enzyme, and why are mononucleotides unable to affect co-operativity? Unlike mononucleotides, Ap4A binds simultaneously through its adenosine moieties to the regulatory sites of both subunits [8], which should make the dimeric structure more rigid. The Ap4A data thus suggest that subunit interactions must be suitably balanced to allow transfer of conformational changes between them; an interaction that is neither too weak (as in the variant dhPPase in the absence of Ap4A) nor too strong (as in the wild-type dhPPase−Ap4A complex) is concordant with this type of regulatory mechanism.

In summary, we have identified an asparagine residue that is involved in regulation-associated information pathways in CBS-PPase and defined the network of hydrogen-bonding interactions that is responsible for information exchange between the catalytic sites. Our data demonstrate that CBS-PPase represents a useful model enzyme for research into the still poorly understood phenomenon of CBS domain-mediated regulation in proteins.

AUTHOR CONTRIBUTION

Viktor Anashkin designed, performed and analysed the experiments, and contributed to writing the paper. Anu Salminen discovered the unusual ehPPase, designed and constructed vectors, expressed and purified proteins, and performed their initial characterization. Natalia Vorobjeva purified proteins and performed binding experiments. Reijo Lahti designed and supervised the experiments. Alexander Baykov designed and analysed experiments and wrote the paper. All authors reviewed the results and approved the final version of the paper.

We thank Dr V.N. Orlov, Dr P.I. Semenyuk and Dr A.M. Arutyunyan for help with ITC and CD measurements.

FUNDING

This work was supported by the Academy of Finland [grant number 139031] and Russian Foundation for Basic Research [grant number 15-04-04828].

Abbreviations

     
  • Ap4A

    diadenosine tetraphosphate

  •  
  • bsPPase

    Bacillus subtilis PPase

  •  
  • CBS

    cystathionine β-synthase

  •  
  • CBS-PPase

    CBS domain-containing PPase

  •  
  • cpPPase

    Clostridium perfringens PPase

  •  
  • dhPPase

    Desulfitobacterium hafniense PPase

  •  
  • ehPPase

    Ethanoligenens harbinense PPase

  •  
  • ITC

    isothermal titration calorimetry

  •  
  • PNP

    imidodiphosphate

  •  
  • PPase

    pyrophosphatase

  •  
  • RMSF

    root mean square fluctuation

References

References
1
Young
T.
Kuhn
N.
Wadeson
A.
Ward
S.
Burges
D.
Cooke
G.
Bacillus subtilis ORF yybQ encodes a manganese-dependent inorganic pyrophosphatase with distinctive properties: the first of a new class of soluble pyrophosphatase?
Microbiology
1998
, vol. 
144
 (pg. 
2563
-
2571
)
[PubMed]
2
Shintani
T.
Uchiumi
T.
Yonezawa
T.
Salminen
A.
Baykov
A.
Lahti
R.
Hachimori
A.
Cloning and expression of a unique inorganic pyrophosphatase from Bacillus subtilis: evidence for a new family of enzymes
FEBS Lett.
1998
, vol. 
439
 (pg. 
263
-
266
)
[PubMed]
3
Parfenyev
A.N.
Salminen
A.
Halonen
P.
Hachimori
A.
Baykov
A.A.
Lahti
R.
Quaternary structure and metal-ion requirement of family II pyrophosphatases from Bacillus subtilis, Streptococcus gordonii and Streptococcus mutans
J. Biol. Chem.
2001
, vol. 
276
 (pg. 
24511
-
24518
)
[PubMed]
4
Merckel
M.C.
Fabrichniy
I.P.
Salminen
A.
Kalkkinen
N.
Baykov
A.A.
Lahti
R.
Goldman
A.
Crystal structure of Streptococcus mutans pyrophosphatase: a new fold for an old mechanism
Structure
2001
, vol. 
9
 (pg. 
289
-
297
)
[PubMed]
5
Ahn
S.
Milner
A.J.
Fütterer
K.
Konopka
M.
Ilias
M.
Young
T.W.
White
S.A.
The “open” and “closed” structures of the type-C inorganic pyrophosphatases from Bacillus subtilis and Streptococcus gordonii
J. Mol. Biol.
2001
, vol. 
313
 (pg. 
797
-
811
)
[PubMed]
6
Jämsen
J.
Tuominen
H.
Salminen
A.
Belogurov
G.A.
Magretova
N.N.
Baykov
A.A.
Lahti
R.
A CBS domain-containing pyrophosphatase of Moorella thermoacetica is regulated by adenine nucleotides
Biochem. J.
2007
, vol. 
408
 (pg. 
327
-
353
)
[PubMed]
7
Bateman
A.
The structure of a domain common to archaebacteria and the homocystinuria disease protein
Trends Biochem. Sci.
1997
, vol. 
22
 (pg. 
12
-
13
)
[PubMed]
8
Tuominen
H.
Salminen
A.
Oksanen
E.
Jämsen
J.
Heikkilä
O.
Lehtiö
L.
Magretova
N.N.
Goldman
A.
Baykov
A.A.
Lahti
R.
Crystal structures of the CBS and DRTGG domains of the regulatory region of Clostridium perfringens pyrophosphatase complexed with the inhibitor, AMP, and activator, diadenosine tetraphosphate
J. Mol. Biol.
2010
, vol. 
398
 (pg. 
400
-
413
)
[PubMed]
9
Anashkin
V.A.
Salminen
A.
Tuominen
H.K.
Lahti
R.
Baykov
A.A.
Cystathionine β-synthase (CBS) domain-containing pyrophosphatase as a target for diadenosine polyphosphates in bacteria
J. Biol. Chem.
2015
, vol. 
290
 (pg. 
27594
-
27603
)
[PubMed]
10
Ignoul
S.
Eggermont
J.
CBS domains: structure, function, and pathology in human proteins
Am. J. Physiol. Cell Physiol.
2005
, vol. 
289
 (pg. 
C1369
-
C1378
)
[PubMed]
11
Baykov
A.A.
Tuominen
H.K.
Lahti
R.
The CBS domain: a protein module with an emerging prominent role in regulation
ACS Chem. Biol.
2011
, vol. 
6
 (pg. 
1156
-
1163
)
[PubMed]
12
Ereño-Orbea
J.
Oyenarte
I.
Martìnez-Cruz
L.A.
CBS domains: ligand binding sites and conformational variability
Arch. Biochem. Biophys.
2013
, vol. 
540
 (pg. 
70
-
81
)
[PubMed]
13
Scott
J.W.
Hawley
S.A.
Green
K.A.
Anis
M.
Stewart
G.
Scullion
G.A.
Norman
D.G.
Hardie
D.G.
CBS domains form energy-sensing modules whose binding of adenosine ligands is disrupted by disease mutations
J. Clin. Invest.
2004
, vol. 
113
 (pg. 
274
-
284
)
[PubMed]
14
Kemp
B.E.
Bateman domains and adenosine derivatives form a binding contract
J. Clin. Invest.
2004
, vol. 
113
 (pg. 
182
-
184
)
[PubMed]
15
Jeong
B.-C.
Park
S.H.
Yoo
K.S.
Shin
J.S.
Song
H.K.
Change in single cystathionine β-synthase domain-containing protein from a bent to flat conformation upon adenosine monophosphate binding
J. Struct. Biol.
2013
, vol. 
183
 (pg. 
40
-
46
)
[PubMed]
16
Ereño-Orbea
J.
Majtan
T.
Oyenarte
I.
Kraus
J.P.
Martínez-Cruz
L.A.
Structural insight into the molecular mechanism of allosteric activation of human cystathionine β-synthase by S-adenosylmethionine
Proc. Natl Acad. Sci. U.S.A.
2014
, vol. 
111
 (pg. 
E3845
-
E3852
)
[PubMed]
17
McCorvie
T.J.
Kopec
J.
Hyung
S.-J.
Fitzpatrick
F.
Feng
X.
Termine
D.
Strain-Damerell
C.
Vollmar
M.
Fleming
J.
Janz
J.M.
, et al. 
Inter-domain communication of human cystathionine β-synthase: structural basis of S-adenosyl-L-methionine activation
J. Biol. Chem.
2014
, vol. 
289
 (pg. 
36018
-
36030
)
[PubMed]
18
Corral-Rodríguez
M.Á.
Stuiver
M.
Abascal-Palacios
G.
Diercks
T.
Oyenarte
I.
Ereño-Orbea
J.
de Opakua
A.I.
Blanco
F.J.
Encinar
J.A.
Spiwok
V.
, et al. 
Nucleotide binding triggers a conformational change of the CBS module of the magnesium transporter CNNM2 from a twisted towards a flat structure
Biochem. J.
2014
, vol. 
464
 (pg. 
23
-
34
)
[PubMed]
19
Salminen
A.
Anashkin
V.A.
Lahti
M.
Tuominen
H.K.
Lahti
R.
Baykov
A.A.
Cystathionine β-synthase (CBS) domains confer multiple forms of Mg2+-dependent cooperativity to family II pyrophosphatases
J. Biol. Chem.
2014
, vol. 
289
 (pg. 
22865
-
22876
)
[PubMed]
20
Baykov
A.A.
Avaeva
S.M.
A simple and sensitive apparatus for continuous monitoring of orthophosphate in the presence of acid-labile compounds
Anal. Biochem.
1981
, vol. 
116
 (pg. 
1
-
4
)
[PubMed]
21
Altschul
S.F.
Gish
W.
Miller
W.
Myers
E.W.
Lipman
D.J.
Basic local alignment search tool
J. Mol. Biol.
1990
, vol. 
215
 (pg. 
403
-
410
)
[PubMed]
22
Kanehisa
M.
Sato
Y.
Kawashima
M.
Furumichi
M.
Tanabe
M.
KEGG as a reference resource for gene and protein annotation
Nucleic Acids Res
2016
, vol. 
44
 (pg. 
D457
-
D462
)
[PubMed]
23
Fabrichniy
I.P.
Lehtio
L.
Tammenkoski
M.
Zyryanov
A.B.
Oksanen
E.
Baykov
A.A.
Lahti
R.
Goldman
A.
A trimetal site and substrate distortion in a family II inorganic pyrophosphatase
J. Biol. Chem.
2007
, vol. 
282
 (pg. 
1422
-
1431
)
[PubMed]
24
Winn
M.D.
Ballard
C.C.
Cowtan
K.D.
Dodson
E.J.
Emsley
P.
Evans
P.R.
Keegan
R.M.
Krissinel
E.B.
Leslie
A.G.W.
McCoy
A.
, et al. 
Overview of the CCP4 suite and current developments
Acta Crystallogr. D Biol. Crystallogr.
2011
, vol. 
67
 (pg. 
235
-
242
)
[PubMed]
25
Larsen
M.
Willett
M.
Yount
R.G.
Imidodiphosphate and pyrophosphate: possible biological significance of similar structures
Science
1969
, vol. 
166
 (pg. 
1510
-
1511
)
[PubMed]
26
Durrant
J.D.
McCammon
J.A.
HBonanza: a computer algorithm for molecular-dynamics-trajectory hydrogen-bond analysis
J. Mol. Graph. Model.
2011
, vol. 
31
 (pg. 
5
-
9
)
[PubMed]
27
Baykov
A.A.
Bakuleva
N.P.
Rea
P.A.
Steady-state kinetics of substrate hydrolysis by vacuolar H+-pyrophosphatase: a simple three-state model
Eur. J. Biochem.
1993
, vol. 
217
 (pg. 
755
-
762
)
[PubMed]
28
Bisswanger
H.
Enzyme Kinetics: Principles and Methods
2008
2nd edn
Weinheim
Wiley-VCH Verlag
29
Pettersen
E.F.
Goddard
T.D.
Huang
C.C.
Couch
G.S.
Greenblatt
D.M.
Meng
E.C.
Ferrin
T.E.
UCSF Chimera: a visualization system for exploratory research and analysis
J. Comput. Chem.
2004
, vol. 
25
 (pg. 
1605
-
1612
)
[PubMed]
30
Janosík
M.
Kery
V.
Gaustadnes
M.
Maclean
K.N.
Kraus
J.P.
Regulation of human cystathionine β-synthase by S-adenosyl-L-methionine: evidence for two catalytically active conformations involving an autoinhibitory domain in the C-terminal region
Biochemistry
2001
, vol. 
40
 (pg. 
10625
-
10633
)
[PubMed]
31
Jämsen
J.
Tuominen
H.
Baykov
A.A.
Lahti
R.
Mutational analysis of residues in the regulatory CBS domains of Moorella thermoacetica pyrophosphatase corresponding to disease-related residues of human proteins
Biochem. J.
2011
, vol. 
435
 (pg. 
497
-
504
)
32
Kery
V.
Poneleit
L.
Kraus
J.
Trypsin cleavage of human cystathionine β-synthase into an evolutionarily conserved active core: structural and functional consequences
Arch. Biochem. Biophys.
1998
, vol. 
355
 (pg. 
222
-
232
)
[PubMed]
33
Adams
J.
Chen
Z.P.
Van Denderen
B.J.
Morton
C.J.
Parker
M.W.
Witters
L.A.
Stapleton
D.
Kemp
B.E.
Intrasteric control of AMPK via the γ1 subunit AMP allosteric regulatory site
Protein Sci.
2004
, vol. 
13
 (pg. 
155
-
165
)
[PubMed]
34
Barnes
B.
Marklund
S.
Steiler
T.
Walter
M.
Hjälm
G.
Amarger
V.
Mahlapuu
M.
Leng
Y.
Johansson
C.
Galuska
D.
, et al. 
The 5′-AMP-activated protein kinase γ3 isoform has a key role in carbohydrate and lipid metabolism in glycolytic skeletal muscle
J. Biol. Chem.
2004
, vol. 
279
 (pg. 
38441
-
38447
)
[PubMed]
35
Labesse
G.
Alexandre
T.
Vaupré
L.
Salard-Arnaud
I.
Him
J.L.
Raynal
B.
Bron
P.
Munier-Lehmann
H.
MgATP regulates allostery and fiber formation in IMPDHs
Structure
2013
, vol. 
21
 (pg. 
975
-
985
)
[PubMed]
36
Banerjee
R.
Zou
C.G.
Redox regulation and reaction mechanism of human cystathionine β-synthase: a PLP-dependent hemesensor protein
Arch. Biochem. Biophys.
2005
, vol. 
435
 (pg. 
144
-
156
)
37
Oliveriusová
J.
Kery
V.
Maclean
K.N.
Kraus
J.P.
Deletion mutagenesis of human cystathionine β-synthase: impact on activity, oligomeric status, and S-adenosylmethionine regulation
J. Biol. Chem.
2002
, vol. 
277
 (pg. 
48386
-
48394
)
[PubMed]
38
Halonen
P.
Tammenkoski
M.
Niiranen
L.
Huopalahti
S.
Parfenyev
A.N.
Goldman
A.
Baykov
A.A.
Lahti
R.
Effects of active site mutations on the metal binding affinity, catalytic competence, and stability of the family II pyrophosphatase from Bacillus subtilis
Biochemistry
2005
, vol. 
44
 (pg. 
2088
-
2096
)
[PubMed]

Author notes

1

These authors contributed equally to this work.

Supplementary data