The Structure and Nucleotide-Binding Characteristics of Regulated Cystathionine β-Synthase Domain-Containing Pyrophosphatase without One Catalytic Domain

Abstract

Regulatory adenine nucleotide-binding cystathionine β-synthase (CBS) domains are widespread in proteins; however, information on the mechanism of their modulating effects on protein function is scarce. The difficulty in obtaining structural data for such proteins is ascribed to their unusual flexibility and propensity to form higher-order oligomeric structures. In this study, we deleted the most movable domain from the catalytic part of a CBS domain-containing bacterial inorganic pyrophosphatase (CBS-PPase) and characterized the deletion variant both structurally and functionally. The truncated CBS-PPase was inactive but retained the homotetrameric structure of the full-size enzyme and its ability to bind a fluorescent AMP analog (inhibitor) and diadenosine tetraphosphate (activator) with the same or greater affinity. The deletion stabilized the protein structure against thermal unfolding, suggesting that the deleted domain destabilizes the structure in the full-size protein. A “linear” 3D structure with an unusual type of domain swapping predicted for the truncated CBS-PPase by Alphafold2 was confirmed by single-particle electron microscopy. The results suggest a dual role for the CBS domains in CBS-PPase regulation: they allow for enzyme tetramerization, which impedes the motion of one catalytic domain, and bind adenine nucleotides to mitigate or aggravate this effect.

1. Introduction

Numerous biosynthetic reactions critically depend on the ability of cells to degrade pyrophosphate formed from ATP as a byproduct [1]. In most cell types, this function is performed by specific soluble enzymes, pyrophosphatases (PPases; E.C. 3.6.1.1), belonging to three nonhomologous families [2,3,4]. Family II PPases, found in prokaryotes and archaea, surpass other PPases by catalytic efficiency (k_cat of up to 10⁴ s⁻¹), and their prevailing form is a dimer of identical subunits, each formed by two catalytic domains, DHH and DHHA2 [3]. The active site is located between the domains in each subunit and binds one transition metal ion (Mn²⁺ or Co²⁺) and 1–2 Mg²⁺ ions as essential activators and the MgPP_i complex as a substrate.

In some organisms, the two catalytic domains of family II PPases are supplemented by a pair of cystathionine β-synthase (CBS) domains and one DRTGG domain that together form a regulatory insert within the DHH domain [3,5] (Figure 1). Such CBS-PPases are less efficient catalysts (k_cat is less by 1–2 orders of magnitude) because the regulatory insert acts as an internal inhibitor. AMP and ADP binding to the CBS domains further increases the inhibition [5], whereas ATP and diadenosine polyphosphate binding alleviates it [5,6], allowing efficient regulation of enzyme activity and, hence, cellular PP_i level and the intensity of biosynthesis by the energy status of the cell. Quite important in this respect is the unprecedented affinity of CBS-PPase for diadenosine tetraphosphate (Ap₄A), a cellular stress alarmone [7]. Its accumulation under stress conditions and concomitant CBS-PPase activation provide the host organism with a way to escape stress by decreasing cellular PP_i levels to restore biosynthesis [6]. No other protein with or without the CBS domain demonstrates similar affinity for Ap₄A, supporting the key role of CBS-PPase in combatting stress in host bacteria.

CBS domains are relatively small (approximately 60 amino acid residues) units that are usually present in pairs with a characteristic interaction between them [8]. They are found in thousands of soluble and membrane-associated proteins of all kingdoms of life [9] and have a regulatory function in all characterized cases [8,10]. The regulatory ligands are commonly, but not exclusively, various adenosine phosphates. Mutations in CBS domains are associated with several hereditary diseases in humans [11]. CBS domains can be excised from proteins in a functionally competent form; furthermore, some proteins with similar structure and function either have or do not have CBS domains, depending on the host organism [12,13,14,15]. Crystallization of CBS domain-containing proteins is difficult, likely because of their structural flexibility, and most available structures are for CBS domain-only proteins or fragments of multidomain proteins. This explains why only limited information is available on the mechanism of CBS domain-mediated regulation.

In this context, CBS-PPase provides a useful model for studying this mechanism. CBS-PPase regulation has been thoroughly characterized using kinetic and site-directed mutagenesis approaches. The regulation depends on multiple effects, including changes in k_cat and substrate binding cooperativity [15]; regulatory ligand binding is also cooperative [15]. Therefore, the regulatory phenomenon should depend on a branched network of interactions that link the regulatory and active sites within a subunit and with other subunits in the tetrameric protein molecule.

Further advances in understanding CBS-PPase regulation critically depend on the knowledge of its 3D structure. Previous X-ray crystallographic studies have elucidated the structures of the DHH and DHHA2 domains in dimeric CBS domain-lacking “canonical” Family II PPase [16,17] and the structure of a dimeric separate regulatory part of CBS-PPase [18]. The first proposed model of the entire CBS-PPase molecule based on these structures also assumed that it is homodimeric [18]. However, more recent studies have demonstrated that CBS-PPase is tetrameric [19,20]. We have subsequently proposed two possible structural models of the domain organization in tetramers [19], but neither of them has been verified experimentally.

In the present study, we combined in silico and in vitro approaches to determine the 3D structure of CBS-PPase without the DHHA2 domain, the expected most movable part of the protein molecule. The inactive truncated enzyme variant (ΔdhPPase) retained the oligomeric structure of the full-size enzyme and its ability to bind adenosine phosphates, highlighting the role of the deleted domain in the regulatory mechanism.

2. Results

The DHHA2 domain is the most movable element in unregulated Family II PPases [16,17]. Therefore, we deleted this domain from CBS-PPase in order to obtain a more stable structure. Because the deleted DHHA2 domain contains an important part of the active site, the ΔdhPPase variant did not catalyze PP_i hydrolysis. Therefore, to evaluate the functional competence of ΔdhPPase and, hence, its structural integrity, we measured its interaction with regulatory ligands. We also determined the oligomeric structure of ΔdhPPase and analyzed its thermal unfolding in the absence and presence of the ligands. Finally, we determined the shape of the deletion variant and computationally predicted its detailed 3D structure.

2.1. Nucleoside Phosphate (Ap₄A and AMP) Binding

Binding measurements were attempted with two regulating adenosine phosphates—diadenosine tetraphosphate (Ap₄A, activator) and AMP (inhibitor)—using isothermal titration calorimetry (ITC). Figure 2 presents the ITC data for Ap₄A.The titration curve obtained for the full-size dhPPase (Figure 2B, black symbols) was quite similar to that reported earlier [6] and demonstrated tight binding with a stoichiometry of approximately one Ap₄A molecule per two subunits (

Table 1. Thermodynamic parameters for the complexes of dhPPase with Ap₄A and AMP, as obtained by isothermal calorimetry.

Enzyme Variant	Nucleotide	ΔH, kcal/mol	n	KN, µM
ΔdhPPase	Ap4A	−17.0 ± 0.3	0.45 ± 0.01	<<0.01
dhPPase	Ap4A	−10.8 ± 0.2 (−10.4 ± 0.3) a	0.42 ± 0.01 (0.41 ± 0.01) a	0.01–0.1
dhPPase a	AMP	−5.6 ± 0.5	0.79 ± 0.05	0.8 ± 0.3

^a From Ref. [6].

) or two Ap₄A molecules per tetramer. Consistent with this stoichiometry, the two adenosine groups of the Ap₄A molecule form contacts with both subunits in the structure of a separate dimeric regulatory part (CDC in Figure 1) of C. perfringens CBS-PPase [18]. Notably, the stoichiometry was twice as large, and the enthalpy change ΔH was two times smaller in a similar titration of dhPPase with AMP (Table 1). The variant ΔdhPPase exhibited the same Ap₄A-binding stoichiometry but a much sharper transition (i.e., stronger binding affinity) and a larger enthalpy change ΔH (Table 1). However, the binding constants could only be roughly estimated from Figure 2B because of the very tight binding (Table 1). Nevertheless, the ΔdhPPase affinity was undoubtedly significantly higher, and the difference was reproducible in replicate titrations.

Similar titrations attempted with AMP were unsuccessful because of the incipient precipitation of AMP-bound ΔdhPPase. Precipitation (loss of solution transparency) started after two or three AMP additions, resulting in an unusual titration curve with an endothermic and exothermic signals upon each addition. Notably, precipitation was not observed with full-size dhPPase, which is consistent with earlier data [6]. The heat effects of the first two AMP additions, which did not cause ΔdhPPase precipitation, were two times less and similar to those previously reported for full-size dhPPase [6]. This finding and the precipitation effect indicated that ΔdhPPase can bind AMP.

Quantitative characteristics of inhibitor binding were instead obtained for Mant-AMP, an AMP derivative containing a fluorescent N-methyl-anthraniloyl substituent in the ribose moiety. Mant-AMP binding to the full-size dhPPase was first characterized by measuring its effect on enzymatic activity (Figure 3A). The inhibition curve was clearly sigmoidal and was, therefore, analyzed in terms of Scheme 1, which assumes the presence of two types of interacting binding sites for Mant-AMP on the enzyme. A similar analysis of dhPPase inhibition by AMP has been performed previously under identical conditions [15], and

Table 2. Comparison of Mant-AMP and AMP as inhibitors of dhPPase.

Nucleotide	Residual Activity, %	h	Ki1, µM	Ki2, µM
Mant-AMP	<1	1.75 ± 0.04	41 ± 7	0.8 ± 0.1
AMP a	3.7 ± 0.1	1.68 ± 0.04	9.6 ± 1.4	0.21 ± 0.08

^a From Ref. [15].

compiles the parameter values for both nucleotides. Their comparison indicates that both nucleotides inhibited dhPPase almost completely and exhibited strongly positive binding cooperativity (the Hill coefficient of approximately 1.7, the K_i1/K_i2 ratio of approximately 50). The major difference was a fourfold greater K_i value for Mant-AMP, indicating its lower binding affinity. These results demonstrated a similar behavior of the two nucleotides as dhPPase ligands.

Mant-AMP binding to the inactive ΔdhPPase variant was measured fluorimetrically in the absence of the substrate, and parallel data were obtained for the full-size dhPPase. To suppress background fluorescence, bound Mant-AMP was excited by energy transfer (FRET) from a nearby Trp residue that is excited at a much lower wavelength. FRET titration could be performed at a lower protein concentration and faster than ITC titration (1.5 versus 5 min per point) to diminish precipitation. To further decrease the overall time of ΔdhPPase contact with Mant-AMP, a titration curve was constructed from two overlapping titrations performed separately with identical protein samples but using different Mant-AMP concentration ranges (0.2–10 and 10–100 µM). Together with the lower propensity of Mant-AMP to induce protein aggregation, these precautions eliminated the protein precipitation problem with ΔdhPPase.

The titration curves for dhPPase and ΔdhPPase (Figure 3B) look similar, yielding very similar parameter values (

Table 3. The parameter values for Mant-AMP binding derived from the data in Figure 3B.

Enzyme Variant	ΔFmax	KN1, µM	KN2, µM
dhPPase	10.3 ± 0.4	22 ± 2	6.5 ± 1
ΔdhPPase	8.8 ± 0.4	21 ± 6	6 ± 2

). Mant-AMP binding to both dhPPase forms demonstrated significant positive cooperativity (K_N1 > K_N2). Of note, AMP binding to dhPPase is also a positively cooperative process [15]. Considering that both the K_i values in Table 2 and K_N values in Table 3 are true binding constants, the differences between them for dhPPase suggested that substrate bound to active sites in K_i measurements significantly altered nucleotide binding to regulatory sites. Specifically, the bound substrate increased the positive binding cooperativity (K_i1/K_i2 > K_N1/K_N2), mainly via the binding of the second ligand molecule (K_i2).

2.2. Effects of Deletion and Nucleotide Binding on Protein Thermal Stability

The effects of domain deletion and nucleotide binding on the stability of the protein molecule were assessed by measuring its thermal unfolding profile. In the method often called ThermoFluor [21], protein unfolding under the action of an applied temperature gradient is monitored with a dye whose fluorescence increases upon binding to the hydrophobic groups that become exposed in the unfolded or partially unfolded protein (Figure 4A). The principal characteristic derived from a ThermoFluor profile is the melting temperature (T_m), which is determined as the peak temperature on the first derivative plot dF/dT (Figure 4B–D). In a multidomain protein, such as CBS-PPase, separate structural units unfold sequentially, resulting in several peaks on the dF/dT plot. For reliable peak assignments to specific domains, two other available deletion variants of dhPPase, comprising the separate catalytic or regulatory part, were used (ΔΔdhPPase and CDC in Table 1). Importantly, no protein precipitation was observed in the ThermoFluor measurements because of the 10-fold shorter duration of the experiment and lower protein concentration compared with the ITC titration.

The first-derivative plot for the full-size dhPPase demonstrated three major peaks with T_m values of 57.0, 64.0, and 77.5 °C (Figure 4B), likely corresponding to three structural units—the DHHA2 domain, the regulatory part (two CBS and one DRTGG domains), and DHH domain, respectively. All these units preserve their structures when separated from the protein [15,17,18] and are, therefore, self-contained and should unfold independently. The deletion of the DHHA2 domain eliminated peak I (Figure 4B), suggesting that it corresponds to this domain. This assignment is consistent with the presence of peak I in the melting profile of the separate catalytic part (ΔΔdhPPase) (

Table 4. The melting temperatures of full-size dhPPase and its deletion variants and the effects of adenosine phosphates. The values are precise within 0.5 °C.

Nucleotide	Tm, °C
	dhPPase	ΔdhPPase	CDC	ΔΔdhPPase
None	57.0, 64.0, 77.5	68.3, 83.5	58.0	56.5, 77.5
AMP	57.1, 67.4, 80.8	68.6, 85.5	66.5
Ap4A	52.9, 68.8, 79.7	71.0, 86.5	61.5

).

Peak II likely belongs to the regulatory part, as it is found in the melting profiles of both ΔdhPPase and CDC but not ΔΔdhPPase (Table 4). This assignment is also supported by the observations that AMP and Ap₄A universally increased T_m for separate CDC and for peak II of dhPPase and ΔdhPPase (Table 4). In contrast, ligand effects on peak I in dhPPase were different—AMP had little effect on T_m, whereas Ap₄A decreased it appreciably (Table 4). The different effects of AMP (inhibitor) and Ap₄A (activator) on DHHA2 stability correlated with their opposite effects on enzymatic activity, and this observation may provide a clue to the mechanism of activity regulation. Peak III is absent only in CDC and likely belongs to the DHH domain. Consistent with this assignment, the melting profile of homologous Family II PPase from Streptococcus gordonii, formed by only the DHHA2 and DHH domains [22], exhibited only two peaks with T_m values of 56.5 and 82.3 °C. These values clearly match peaks I and III of dhPPase and the two peaks of ΔΔdhPPase.

2.3. Oligomeric Structure of ΔdhPPase

Wild-type CBS-PPases, including dhPPase, are homotetrameric [19,20]. The deletion variant, which lacked the catalytic DHHA2 domain, retained the tetrameric structure, as demonstrated by two different approaches—size-exclusion chromatography and analytical ultracentrifugation. The elution volumes of native dhPPase and ΔdhPPase in size-exclusion chromatography differed by 0.35 mL (Figure 5A), corresponding to a mass difference of ~38 kDa. This value matches reasonably well the molecular mass of the deleted domain (13.9 kDa) multiplied by four. The deletion decreased the sedimentation coefficient, s_20,w, of dhPPase by only 0.8 S, again consistent with the tetrameric structure of its deletion variant. The molecular masses estimated from these measurements are summarized in

Table 5. The molecular masses of dhPPase and its deletion variant.

Enzyme Variant	Molecular Mass, kDa
	Sedimentation a	SEC	Theory b
ΔdhPPase	171–192	~176	186
dhPPase c	203–234	~214	241

^a The ranges of the masses shown were estimated assuming frictional ratios of 1.25–1.35, typical for globular proteins [23]. ^b Calculated from the amino acid sequence. ^c From Ref. [19].

.

2.4. Single-Particle Electron Microscopy of ΔdhPPase

Cryo-electron microscopy (cryo-EM) was used to determine the domain organization in ΔdhPPase. Although a recent attempt to use this approach with full-size dhPPase was unsuccessful [24], the deletion variant, lacking the movable DHHA2 domain, demonstrated more promising results.

Before cryo-EM grid preparation, sample monodispersity and aggregation were assessed using negative-stain transmission electron microscopy. The resulting micrographs showed good particle distribution (Figure S1), and the resulting reference-free class averages clearly revealed a symmetrical linear structure, despite some heterogeneity. The small deviations from linearity seen in some images may be an artifact of the molecule binding to the solid support.

To assess the spatial organization of the ΔdhPPase tetramer, cryo-EM was applied to the same enzyme sample. The 2D classification yielded approximately 15,000 single particle projections that produced good quality class averages (Figure 6A). The angular distribution of particle projections reveals that the majority of projections depict views perpendicular to the longest dimension of the protein (Figure S2). The 3D reconstruction and non-uniform refinement led to the density map with 16 Å resolution (Figure 6B).

2.5. The Modeled 3D Structure of ΔdhPPase

We previously used Modeller to propose two possible models of tetrameric full-size apo-dhPPase [19]. Here, we predicted the ΔdhPPase apo-structure using an advanced tool, Alphafold2 [25]. The principal feature of the modeled ΔdhPPase structure is that four DHH domains form a core, whereas the regulatory parts are positioned in the outmost region from the core region (Figure 7A). The modeled structure displays D2 symmetry (Figure S3).

According to the modeled structure, each subunit contacts two other subunits in tetrameric ΔdhPPase, forming four contacts with one of them (through DRTGG, CBS1, CBS2, and DHH domains, marked orange) and one contact with the other (through DHH domain, marked violet) (Figure 6B). The DHH domains of subunits a1 and a4 and, similarly, subunits a2 and a3 interact by forming common β-sheets with a 790-Ų buried surface area per subunit in each pair. This type of DHH domain interaction with a similar buried surface area of 805 Ų per subunit was previously observed in the crystal structure of canonical dimeric Family II PPases, which lack regulatory domains [16,17]. Each DHH domain also forms an alternative interaction in the pairs a2–a4 and a1–a3, but these alternative contacts appear to be much weaker, based on the buried surface areas of only 112 Ų per subunit.

However, the principal tetramer-stabilizing interaction occurs via the regulatory domains within the pairs a1–a3 and a2–a4, as in the crystal structure of the dimerized regulatory part of Clostridium perfringens CBS-PPase [18]. The buried surface area for the entire regulatory part comprising CBS1, CBS2, and DRTGG domains in ΔdhPPase is 2110 Ų per subunit, by far surpassing all other contacts. In the dimerized separate regulatory part of C. perfringens CBS-PPase, this contact area was of similar size (1770–2190 Ų per subunit) [18]. The modeled structure was accommodated with a correlation factor of 0.86 within the density map obtained by cryo-EM (Figure 7C).

3. Discussion

3.1. 3D Structure of dhPPase

The unusual flexibility of the CBS-PPase structure apparently explains the failure of previous attempts by several groups to crystallize the full-size enzyme for X-ray structural analysis. A recent breakthrough in the 3D structure prediction of oligomeric proteins [25] allowed us to construct a reliable model of its less flexible deletion variant. The validity of the predicted structure is supported by the fact that it is grounded on the known crystal structures of its two parts derived from highly homologous proteins—the two-domain “canonical” Family II PPase and the regulatory part of C. perfringens CBS-PPase. Moreover, the predicted nontrivial shape of the ΔdhPPase molecule was directly confirmed by single-particle electron microscopy. Both types of evidence refer to the deletion variant of CBS-PPase, which lack one catalytic domain, but is likely also applicable to the full-size enzyme. This premise is supported by data showing that the structures of both enzyme forms are tetrameric and fully functional in their regulatory parts (capable of high-affinity AMP and Ap₄A binding). The tentative model of the full-size dhPPase deduced from small-angle X-ray scattering data [24] is entirely different from that in Figure 6, perhaps, because of the limitations of this method.

The major subunit contacts stabilizing the ΔdhPPase tetramer are those between the regulatory parts (two such interactions per tetramer) and between the common β-sheet-forming DHH domains (two more interactions per tetramer). These interactions are intertwined to form tetramers, although they form only dimers separately [16,17,18]. The ΔdhPPase structure presents an unusual type of domain swapping [27], wherein the domains are partially exchanged not between monomers, as in the classical case, but between dimers. To the best of our knowledge, identical architecture has not been observed in other tetrameric CBS-proteins. An “inverted” type of linear homotetramer was found in Mycobacterium tuberculosis cystathionine β-synthase [28] (7XNZ) and the protein THA0829 from Thermus thermophilus HB8 [29] (5AWE) with four CBS modules in the central part and two peripheral catalytic domain dimers on both sides. The basal form of IMP dehydrogenase is a planar tetramer in which only catalytic domains interact with one another [30] (1ZFJ), but ATP and GTP induce the combination of two planar tetramers into an octamer along the fourfold axis via intertwined interactions of both catalytic and CBS domains [31,32,33] (3DQW, 7OJ1, and 7PJI, respectively). GMP reductase forms a similar octamer [34].

3.2. Mechanism of Regulation

The differential effects of adenosine phosphates on the activities of CBS domain-containing enzymes are usually interpreted in terms of internal inhibition caused by the CBS domains [11,12,15,35]. According to this concept, evolutionarily acquired CBS domains distort the protein structure, thereby suppressing its functionality, and CBS-domain ligands either mitigate or aggravate these effects by appropriately tuning the protein conformation. In both cases, the involved conformational change consumes part of the ligand-binding energy and creates structural tension. Because the catalytic and regulatory sites are separated in space, signal (tension) transfer between them should involve the intermediate element(s) of the protein structure. Accordingly, deletion of the involved structural element(s) should ease the tension and stimulate ligand binding. In the absence of regulatory ligands, substrate binding induces a catalytically competent active site conformation, causing structural tension that consumes part of the binding energy. This logic may explain, with some reservations, why AMP increases and Ap₄A decreases the accessibility of the dhPPase active site for substrate, as characterized by the Michaelis constant [36], and why this parameter is greater for dhPPase in the absence of adenosine phosphates than for its CBS domain-lacking ΔΔdhPPase variant [15].

The tetrameric structure appears to be another important feature of CBS-PPase, allowing for its regulation by adenosine phosphates. The catalytic DHH and DHHA2 domains are connected by a flexible linker in CBS-PPase, permitting closure and opening of the active site, located in between, during the catalytic cycle. In dimeric canonical Family II PPase, which lacks the regulatory part, subunits interact via their DHH domains, whereas DHHA2 domains have a large freedom to move [16,17]. In tetrameric CBS-PPase, whose catalytic core is formed by four DHH and four DHHA2 domains, the motion of the latter domain is more constrained, resulting in internal inhibition. We speculate that the activity of CBS-PPase is regulated by changes in DHHA2 domain mobility, and the effect of DHHA2 deletion on Ap₄A binding provides support for this assertion. The increased affinity of ΔdhPPase to Ap₄A and greater ΔH versus the full-size enzyme (Table 1) suggest that the DHHA2 domain is indeed part of the mechanism that generates tension in CBS-PPase upon Ap₄A binding. The observation that DHHA2 deletion increased the thermal stability of the remaining domains (Table 4) suggests that the CBS-PPase structure is more relaxed in the absence of the DHHA2 domain than in its presence, which is consistent with the concept of internal inhibition.

The fact that AMP/Mant-AMP binding is not similarly strengthened by the deletion may mean that AMP induces tension in the dhPPase structure through its different element, not the DHHA2 domain. This explanation seems likely in view of the different modes of AMP and Ap₄A binding—the former ligand binds to one subunit, whereas the latter interacts with two subunits [18]. The fact that the AMP effect is not associated with the DHHA2 domain is evidenced by the unchanged T_m for this domain (peak I) upon AMP binding to the full-size enzyme (Table 4). However, the identity of the structural element that mediates the AMP signal can only be conjectured. Earlier mutagenesis studies of dhPPase have identified two interacting elements, helix α1 of CBS2 domain and the loop formed by residues 310–323 in the DHH domain, among the best candidates for signal transmission from the regulatory site to the catalytic DHH domain [37]. Notably, an Ala substitution for the highly conserved Asn312 in this loop reversed the effect of ATP on dhPPase from activation to inhibition [37]. Furthermore, replacement of Arg168 in the highly conserved RyR motif of Moorella thermoacetica CBS-PPase (corresponding to Arg276 of dhPPase) weakened AMP binding and reversed its effect from inhibition to activation [38]. Therefore, both residues may control tension generation in dhPPase upon AMP/Mant-AMP binding, and their presence in ΔdhPPase may explain the unchanged Mant-AMP binding to it.

Further studies are clearly needed to test these intriguing possibilities and delineate the regulation mechanism of CBS-PPase and other CBS domain-containing enzymes and transporters. These studies will necessarily include the determination and analysis of the 3D structure of full-size CBS-PPase, both empty and with bound regulatory ligands.

4. Materials and Methods

4.1. Materials

Wild-type dhPPase (UniProtKB: B8FP42) and its deletion variant were produced in E. coli BL21 cells transformed with the pET-42b vector (Novagen, Merck Group, Darmstadt, Germany) carrying the corresponding genes [19]. The DHHA2 domain (residues 425–544) was deleted in the dhPPase gene using overlap extension PCR with Phusion DNA polymerase with the forward/reverse primers TATACATATGTCAAAAAAAATCCATGTCG/TAATCTCGAGTTAAGCTGCCTTGAACATTTG. The isolation procedure included cell disruption by freezing/thawing, ion exchange, and size exclusion chromatography [19]. The purity of the isolated proteins, as estimated by SDS-PAGE [39] with Coomassie staining, was >90%. Protein concentrations in milligrams per milliliter were determined spectrophotometrically, using an A^0.1%₂₈₀ value of 0.477 for dhPPase and 0.556 for ΔdhPPase, as calculated from the amino acid composition with ProtParam (https://web.expasy.org/protparam/ (accessed on 26 November 2023)). Molar concentrations were calculated in terms of the subunit using subunit molecular masses of 60.5 and 46.6 kDa for dhPPase and ΔdhPPase, respectively.

P1,P4-Diadenosine 5′-polyphosphate (Ap₄A, ammonium salt) and AMP (free acid) were obtained from Sigma-Aldrich (St. Louis, MO, USA). 2′/3′-(N-Methyl-anthraniloyl)-adenosine-5´-monophosphate (Mant-AMP, triethylammonium salt) was a Jena Bioscience GmbH (Jena, Germany) product. The concentrations of nucleotide stock solutions were estimated by measuring absorbance at 259 nm (ε = 15,900 M⁻¹ cm⁻¹ for the monoadenosine phosphates and 31,800 M⁻¹ cm⁻¹ for Ap₄A).

4.2. Enzyme Activity Assay

The initial rates of PP_i hydrolysis were measured using a continuous P_i assay [40]. The assay medium contained 0.1 M Tes-KOH, pH 7.2, 5.23 mM MgCl₂, 140 μM PP_i, (corresponding to 50 µM MgPP_i complex). The reaction was initiated by adding 0.3 nM enzyme and continued for 2–3 min at 25 °C. Rate values were obtained from the initial slopes of the P_i accumulation curves.

4.3. Isothermal Titration Calorimetry (ITC)

Heat production upon nucleotide binding to dhPPase and ΔdhPPase was measured at 25 °C using a VP-iTC calorimeter (MicroCal Ltd, Piscataway, NJ, USA). Enzyme and adenine nucleotide solutions were prepared on 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-μL injections of 100 μM AMP or 33 μM Ap₄A solution into 1.4 mL of 8–12 μM protein solutions. The interval between injections was 5 min. The measured heat values were corrected for ligand dilution effects.

4.4. Förster Resonance Energy Transfer (FRET) Measurements

The increase in Mant-AMP fluorescence due to FRET from excited protein tryptophanes was recorded in a Cary Eclipse spectrofluorometer (Varian, Inc., Palo Alto, CA, USA) using a 3 × 3 mm optical cuvette. The excitation and emission wavelengths were 295 and 455 nm, respectively, and the slits were set at 5 nm. Portions of 1 or 10 mM Mant-AMP solution were added to 0.3 mL of 5 μM enzyme solution in 0.1 M Mops-KOH buffer (pH 7.2) containing 150 mM KCl, 2 mM MgCl₂, and 0.1 mM CoCl₂. The mixture was equilibrated for 1 min after each addition before fluorescence was read. The fluorescence values averaged over a 0.5 min time span were corrected for nucleotide and inner filter effects.

4.5. Binding Data Treatment

The ITC data were analyzed using a MicroCal ITC subroutine in Origin 7.0 software (OriginLab, Northampton, MA, USA) using a single binding site model.

The FRET titration curves were analyzed in terms of Scheme 1, which describes the cooperative binding of ligand N to two initially identical sets of binding sites on protein E. K_N1 and K_N2 are the “microscopic” nucleotide-binding constants.

A system of Equations (1)–(5), derived from Scheme 1, was fit to the binding data using SCIENTIST (MicroMath Scientific Software, Salt Lake City, UT, USA). ΔF is the fluorescence change, ΔF_max is its limiting value at an infinite concentration of N, K_N1 and K_N2 are the dissociation constants, and [N] is the free (unbound) ligand concentration. Equations (2) and (3) define the binding constants, and Equations (4) and (5) describe the mass balance for the protein and ligand. This treatment considers a decrease in free ligand concentration because of complex formation. The factor ½ in Equation (1) signifies that all binding sites contribute equally to the fluorescence change upon ligand binding. The total enzyme concentration, [E]₀, and the concentrations of various enzyme species in Equations (1)–(5) refer to the subunit concentrations; [N]₀ is the total ligand concentration.

ΔF = ΔF_max([EN]/2 + [EN₂])/[E]₀

(1)

[E][N]/[EN] = K_N1

(2)

[EN][N]/[EN₂] = K_N2

(3)

[E] + 2[EN] + [EN₂] = [E]₀

(4)

[N] + 2[EN] + 2[EN₂] = [N]₀

(5)

Enzyme inhibition data were analyzed in a similar way but using Equation (6) instead of Equation (1) and the respective inhibition constants K_i1 and K_i2 instead of K_N1 and K_N2.

A = {A₀[E] + (A₀ + A_lim)[EN]/2 +A_lim[EN₂]}/[E]₀

(6)

4.6. Protein Thermal Stability Measurements (ThermoFluor)

The instrumental setup consisted of a C1000 thermal cycler with a CFX96 real-time PCR detection system (Bio-Rad, Hercules, CA, USA). The assay mixtures (50 μL volume), placed in transparent low-profile 96-well multiplate PCR plates (SSIbio, Lodi, CA, USA), contained 5–10 μM protein, 0.02% SYPRO Orange dye (Invitrogen-Thermo Fisher, Waltham, MS, USA), 0.1 M Mops-KOH buffer (pH 7.2, measured at 25 °C), 2 mM MgCl₂, 0.1 mM CoCl₂, and 150 mM KCl. The indicated adenine nucleotides were added at the following concentrations: Ap₄A, 20 µM; AMP, 200 µM. These concentrations exceeded 100–1000-fold the dissociation constants for the nucleotide complexes of dhPPase [6,15]. The plates were covered with an UltraFlux Standard PCR sealing film (SSIbio) and heated from 30 to 95 °C with stepwise increments of 0.5 °C and a 40 s hold step for every point. Fluorescence was excited at 515–535 nm and monitored at 560–580 nm (second channel, “VIC” setting), and melting temperatures, T_m, were derived from the resulting curves using the BioRad CFX Manager 3.1 software.

4.7. Size-Exclusion Chromatography

Protein samples (0.5 mL, 5–10 mg protein) in elution buffer (0.1 M Mops-KOH, pH 7.2, containing 150 mM KCl, 2 mM MgCl₂, and 0.1 mM CoCl₂), were applied onto a calibrated Superdex 200 Increase 10/300 GL column (GE HealthCare, Chicago, IL, USA) equilibrated with the same buffer. The elution was performed on a GE HealthCare AKTA Purifier 100 FPLC System at a flow rate of 0.4 mL/min, with absorbance monitoring at 280 nm. Fractions (0.5 mL) were collected.

4.8. Sedimentation Velocity Analysis

Analytical ultracentrifugation was performed at 25 °C in a Spinco E instrument (Beckman Instruments, Fullerton, CA, USA) equipped with a computerized data collection unit, with scanning at 280 nm. Samples containing 10 μM protein in the isolation buffer were preincubated for 6 h at 25 °C. The sedimentation velocity was measured at 60,000 rpm, and the sedimentation coefficients (s_20,w) and molecular masses were estimated using the program SedFit [23].

4.9. Structure Modeling and Refinement

The three-dimensional structure of the dhPPase deletion variant was predicted from its amino acid sequence using AlphaFold2 (version 2.3.0) [25] with default parameter settings. All calculations were performed using Nvidia (Santa Clara, CA, USA) RTX A5000 graphical card. Twenty five models generated (five models per prediction mod) were ranked according to their iptm + ptm score, and the best model (score = 0.807) was selected.

4.10. Single-Particle Electron Microscopy

Samples for negatively stained transmission electron microscopy (TEM) were prepared from 3 μL of 2.5 mg/mL (52 µM) ΔdhPPase solution in 0.1 M MOPS-KOH buffer, pH 7.2, containing 150 mM KCl, 2 mM MgCl₂, and 0.1 mM CoCl₂. The protein samples were placed on a glow-discharged carbon film (Ted Pella, Redding, CA, USA), blotted after 60 s, and stained with two droplets of 1% uranyl acetate solution for 10 and 20 s. The data for single particle analysis were collected using a JEOL JEM-2100 200 keV transmission electron microscope equipped with a Direct Electron DE20 camera. Approximately 70,000 particle projections were collected and subjected to 2D classification in cryoSPARC [41].

Samples for cryo-EM were prepared from 3 μL of 2.5 mg/mL ΔdhPPase solution in 0.1 M MOPS-KOH buffer, pH 7.2, containing 150 mM KCl, 2 mM MgCl₂, and 0.1 mM CoCl₂. The protein samples were placed on a glow-discharged 300 mesh lacey carbon film on Copper TEM Grids (Ted Pella, Redding, CA, USA), blotted, and vitrified using an EM GP2 Automatic Plunge Freezer from Leica Microsystems CMS GmbH (Wetzlar, Germany) according to the standard procedure [42]. The vitrification parameters were as follows: 4 °C, 95% humidity in the specimen chamber, 30 s pre-blot time, 12 s blot time with Whatman #1 filter paper, and immediate vitrification in liquid ethane at -180 °C. A Gatan (Pleasanton, CA, USA) Elsa Cryotransfer Holder was used to acquire the cryoEM dataset using a JEOL (Akishima, Tokyo, Japan) JEM-2100 200 keV electron microscope. SerialEM [43] was used to acquire 800 stacks with a 2 µm defocus and 60e/Ų total electron dose.

Further data processing was performed in cryoSPARC [41]. Initially, 737 image stacks were subjected to motion correction and patch contrast transfer function (CTF) estimation. The images were manually inspected to exclude views with aggregates and crystalline ice. Blob-based auto picking was performed, followed by 2D classification and a subsequent run of template-based autopicking. A total of 52,000 particle projections were picked, and 15,000 were left after the 2D classification. An initial reference was generated with a cryoSPARC, version 4.4.0+231114 (https://cryosparc.com/ (accessed on 26 November 2023)) ab-initio approach, and final reconstruction was performed using non-uniform refinement. Reprocessing and blob autopicking were performed using the standard procedure.

The AlphaFold2 model was flexibly fitted into the density map using coot of the CCP4 Software suit, version 8.0.016 (https://www.ccp4.ac.uk/ (accessed on 26 November 2023)). A set of interatomic distance restraints with 5.0 Å radius was derived from the initial model. The model underwent real-space refinement for all chains with the Geman-McClure alpha parameter set to 0.01.