If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
CYTH proteins make up a large superfamily that is conserved in all three domains of life. These enzymes have a triphosphate tunnel metalloenzyme (TTM) fold, which typically results in phosphatase functions, e.g., RNA triphosphatase, inorganic polyphosphatase, or thiamine triphosphatase. Some CYTH orthologs cyclize nucleotide triphosphates to 3′,5′-cyclic nucleotides. So far, archaeal CYTH proteins have been annotated as adenylyl cyclases, although experimental evidence to support these annotations is lacking. To address this gap, we characterized a CYTH ortholog, SaTTM, from the crenarchaeote Sulfolobus acidocaldarius. Our in silico studies derived ten major subclasses within the CYTH family implying a close relationship between these archaeal CYTH enzymes and class IV adenylyl cyclases. However, initial biochemical characterization reveals inability of SaTTM to produce any cyclic nucleotides. Instead, our structural and functional analyses show a classical TTM behavior, i.e., triphosphatase activity, where pyrophosphate causes product inhibition. The Ca2+-inhibited Michaelis complex indicates a two-metal-ion reaction mechanism analogous to other TTMs. Cocrystal structures of SaTTM further reveal conformational dynamics in SaTTM that suggest feedback inhibition in TTMs due to tunnel closure in the product state. These structural insights combined with further sequence similarity network–based in silico analyses provide a firm molecular basis for distinguishing CYTH orthologs with phosphatase activities from class IV adenylyl cyclases.
The common feature of CYTH enzymes is an eight-stranded β-barrel tunnel architecture, termed the triphosphate tunnel metalloenzyme (TTM) fold, which was first found in the crystal structure of the fungal Cet1 RNA triphosphatase (
). The opposing acidic and basic patches shape the substrate binding tunnel. Although the overall coordination of the triphosphate itself is very similar among CYTH enzymes, the directionality of asymmetric organic substrates such as triphosphate nucleotides or thiamine triphosphate within the tunnel determines catalytic specificity (
). Such enzymes catalyze the hydrolysis of the triphosphate by a two-metal ion mechanism using metal one to establish the catalytically competent binding of the triphosphate moiety and metal two to position and activate a water for nucleophilic attack onto the γ-phosphate moiety (
). Accordingly, triphosphatases generate pyrophosphate and an orthophosphate by hydrolysis of the β–γ phosphoanhydride bond. The situation for AC-IV enzymes in the CYTH superfamily is different. Here, the directionality of the organic moiety binding within the tunnel fold is inverted causing a reversed placement of the α- and γ-phosphates within the binding tunnel (
). By comparing AC-IV with already described AC classes, a general two-metal ion mechanism was suggested for the cyclization of ATP to 3′,5′-cAMP using the second metal to lower the activation energy (
The founding members of the CYTH superfamily are involved either in mammalian thiamine triphosphate metabolism or bacterial signal transduction via 3′,5′-cyclic nucleotides. Some members of the CYTH superfamily are parts of multidomain proteins by being fused to HD hydrolase, exopolyphosphatase, nucleotidyl kinase, or CHAD domains (
). Archaeal representatives encode at least one copy of a CYTH ortholog in their genomes, which are generally annotated as adenylyl cyclases. 3′,5′-Cyclic AMP has been described for only a few archaeal species with a role in starvation and cell cycle (
). In a search for a potential nucleotidyl cyclase in the crenarchaeote Sulfolobus acidocaldarius, we identified Saci_0718, a member of the CYTH superfamily and annotated as CyaB-like class IV adenylyl cyclase. Although the substrate binding mode and the hydrolysis mechanisms differ between triphosphatases and CyaB (
), at the sequence level, Saci_0718 shares ∼29% identity with the Yersinia pestis cyclase CyaB and with the inorganic triphosphatase from Clostridium thermocellum.
Here, we have performed a structure–function analysis of Saci_0718 and designated it as SaTTM due to its triphosphatase activity. Its structures in apo and substrate-bound (triphosphate and ATP) states revealed transition between open and closed states during catalysis. Trapping of the Michaelis complex in its calcium-inhibited state confirmed a two-metal ion mechanism for SaTTM. Furthermore, this archaeal TTM is subject to feedback inhibition by its reaction product pyrophosphate. Our comparative structural and bioinformatic analysis of CYTH allowed us to identify molecular determinants, which clearly discriminate the adenylyl cyclase from the different phosphatase activities. This provides now an improved rationale for correct functional annotation within the broad CYTH domain family.
Archaeal CYTH proteins are functionally diverged from CyaB-like class IV adenylyl cyclases
To date, the CYTH-like domain superfamily (IPR033469) comprises 41,000 enzymes from all three domains of life divided into two subfamilies, the smaller mRNA triphosphatase Cet1-like(IPR004206) and the bigger subfamily of CYTH domain proteins (IPR023577) including adenylyl cyclases, ThTPases, and inorganic tri- or polyphosphatases (also known as triphosphatase tunnel metalloenzymes or TTMs). The discrimination between Cet1-like and CYTH enzymes is also consistent within the Pfam database (mRNA_triPase: PF02940 and CYTH: PF01928). Hence, we aimed to obtain a detailed understanding of the CYTH family by combining biochemical analysis combined with structural snapshots of the catalytic mechanism, as the functional annotations of CYTH enzymes are still vague and misleading among TTMs and adenylyl cyclases.
To ensure unambiguous assignment, we chose the Pfam family CYTH (PF01928) as the least common denominator for our analysis. As stated above, this protein family contains enzymes with known functions as TTMs, AC-IVs, and ThTPases. We therefore performed a systematic sequence similarity network (SSN) analysis of the CYTH superfamily PF01928 using EFI-EST (
). The network generated with an E-value of 10−20 against the CYTH-domain boundaries for 38,246 sequences clustered to 4128 nodes with pairwise sequence identities of >40% within the nodes, which are connected by a total of 121,791 edges. This SSN analysis covering all members of the CYTH enzymes of PF01928 showed 10 major distinct clusters (Fig. 1). Among these clusters, three of them (clusters 1, 4, and 7) comprise members adopting a multidomain architecture, while the length histograms for the remaining seven clusters are in the expected range of ∼160 to 200 amino acids for single-domain proteins (Fig. S7A). For example, cluster 1 contains YgiF from Escherichia coli, which possesses a two-domain architecture with an N-terminal CYTH and a C-terminal CHAD domain (PF05235) (
). Similar modular architectures including the CHAD domain can be found in cluster 4, which does not contain any structurally described homologs and whose members occur mostly in actinobacteria. The additional domain in cluster 7, whose members are exclusively found in Viridiplantae and Amoebozoa, is a phosphoribulokinase domain (PF00485).
Of interest, the SSN analysis discriminates the cyaB-like class IV adenylyl cyclase (
) (cluster 10) from other triphosphatases. Cluster 10 is a relatively small cluster within the SSN by comprising only 141 orthologs of bacterial origin. The active site structure and a two-metal ion mechanism of Y. pestis AC-IV was proposed previously (
). AtTTM1 and AtTTM2 do not possess the TTM signature motifs required for catalysis (Table S1). In cluster 5, unpublished structures annotated before as adenylyl cyclase from Bacillus anthracis and Bacillus halodurans can be found. A member of this cluster from Staphylococcus aureus has further been described to be incompetent in cAMP formation and to be distinct from class IV adenylyl cyclases (
). Cluster 3 contains orthologs of mostly archaeal origin from Pyrococcus furiosus and Pyrococcus horikoshii, as well as Saci_0718 (Uniprot: Q4JAT2). The SSN analysis clearly indicates that archaeal CYTH enzymes are functionally diverge from CyaB-like class IV adenylyl cyclases.
S. acidocaldarius CYTH enzyme is a triphosphate tunnel metalloenzyme
To gain in-depth biochemical and structural insights into this protein, we heterologously overproduced a C-terminal His6-tagged variant of Saci_0718 in E. coli. The recombinant protein was purified to homogeneity using an initial heat denaturation step to remove E. coli proteins followed by IMAC and subsequent size exclusion chromatography as a polishing step. The enzyme formed a dimer in solution corresponding to 56 kDa compared with standards (Fig. S1). Dimerization behavior is typical among other CYTH proteins (
). The first ambiguity to be resolved was determination of its substrate preference and the nature of its hydrolysis products. Initially, we tested its activity against ATP using Mg2+ as cofactor at 75 °C and the reaction product analyzed by HPLC (Fig. S2A). This experiment clearly showed that the enzyme is capable of hydrolyzing ATP to ADP and Pi without any formation of cAMP. Saci_0718 showed no base specificity during its phosphohydrolase reaction and able to utilize both ATP or GTP with similar efficiency (Fig. 2A). As the reaction products are diphosphate nucleosides and not (c)AMP/GMP, we used nonhydrolyzable analogs of ATP, AMPCPP, and AMPPCP to identify the nucleophilic attack site (Fig. S2B). It shows that Saci_0718 acts on the terminal phosphate of ATP, as the enzyme was active on AMPCPP but no phosphate release was observed for AMPPCP.
Next, we tested triphosphate (PPPi) as substrate. Under the conditions outlined before Saci_0718 is ∼19 times more active compared with NTP as substrate. The result is consistent with the previous reports of tunnel metalloenzymes (
). We also observed that the enzyme is able to hydrolyze higher polyphosphates like tetraphosphate with final formation of pyrophosphate and phosphate (Fig. S3). As triphosphate seems to be the preferred substrate, we named Saci_0718 SaTTM (S. acidocaldarius triphosphate tunnel metalloenzyme). The optimal assay conditions were observed at a basic pH 8.6 at 75 °C using Mg2+ as a cofactor, whereas Zn2+, Mn2+, and Ni2+ could be used as weaker cofactor alternatives. Similar to the metal ion chelator EDTA, calcium exerted an inhibitory effect on phosphatase activity (Fig. 2, B and C). At optimal conditions, kinetic measurement revealed a classical Michaelis–Menten behavior with a Km of 41.0 μM and a Vmax of 29.7 μM s−1 (Fig. 2D). This shows a highly efficient catalysis performed by SaTTM with a kcat/Km of 1.45 × 108 s−1 M−1 close to the diffusion limit and a turnover rate, kcat, of 5934 s–1, which is in the range of untagged NeuTTM (7900 s−1) (
For structural information, we crystallized SaTTM in different catalytically relevant complexes. Initial attempts to produce diffraction quality crystals using apoenzyme were unsuccessful, but crystallization was finally possible under a high-sulfate condition. In contrast, ligand-bound states regularly produced high hit numbers for diffraction quality crystals, indicating that SaTTM is highly flexible in solution and stabilized upon ligand binding. Crystals of SaTTM in its different catalytic states were obtained after a week at 4 °C that diffracted to 1.75- to 2.3-Å resolution. For structure determination, the P. horikoshii OT3 PH1819 (Protein Data Bank [PDB] entry 2EEN, 39% identity with SaTTM)-based homology model of SaTTM was used as a search model in molecular replacement. SaTTM was crystallized as an open-state homodimer (Fig. 3A), bound with two sulfate anions instead of the ATP that was included in cocrystallization. Of note, a solution structure of mouse thiamine triphosphatase (PDB 2JMU) reported a similarly open state. The open state of SaTTM was crystallized in space group P41212 with one monomer per asymmetric unit. The polypeptide is defined by electron density for R6-A176 but lacks completely residues P44-R49 and Q73-R80 due to missing density. The general topology of SaTTM is described by an eight-stranded β-barrel comprising the antiparallel strands β1-β9/β5-β4-β3-β2-β6-β7-β8(-β1) with three interspersing α helices and the C-terminal plug helix α4. The dimerization of SaTTM was also found in the crystal structure formed by two symmetry mates along the α2 helices and parts of the preceding β5 strand. PDBePISA revealed that the dimer interface covered 875 Å2 with a CSS score of 0.32 and accounts for 8.5% of the total solvent-accessible area of one protomer (
). The dimerization is mainly established by hydrophobic interactions of residues from β3, β4, β5, and α2 and salt bridges between K99 from one protomer and E81 from the other protomer. The C-terminal α4 plug helix blocks the tunnel entry from one side. It was previously reported that this plug helix plays a major role in determining substrate preference (
Next, we aimed to trap the Michaelis complex of SaTTM, for which we cocrystallized the enzyme together with PPPi or ATP and the respective cofactors Mg2+ or Ca2+ at 4 °C. Substrate-bound states of SaTTM crystallized in a closed tunnel conformation, highlighting the structural rearrangement upon ligand binding (Fig. 3B). This supports the previous notion that CYTH enzymes are highly flexible and can breathe during their catalytic cycle (
), we observed electron density in the PPPi⋅Mg2+ state for just one metal ion octahedrally coordinated by PPPi, E7, and E135 (Fig. 4A and S4A), whereas the second ion necessary for water activation was missing (Fig. S5C). When using a complete PPPi molecule during refinements, we found higher B-factors for the γ-phosphate. Therefore, the molecule was split into PPi and γ-Pi and occupancies were separately refined. This revealed that only 76% of the substrate-binding site were occupied with γ-phosphates, suggesting a slow turnover of the triphosphate in crystallo.
The PPPi⋅Ca2+ state (Figs. 4B and S5D) revealed not only metal ion 1 (M1) but also the water-activating position of metal ion 2 site (M2) present in the electron density. The two Ca2+ ions are coordinated by PPPi, E5, E7, and E137 (Fig. S4B), thereby positioning a water molecule in an inline-attack position 3.2 Å away from the γ-phosphate phosphorous as observed for the E. coli YgiF⋅PPPi states (PDB: 5A60, 5A61). Here, no differences of γ-phosphate B-factors and occupancies were observed, indicating no hydrolysis during crystallization, which is in line with activity assays in the presence of Ca2+ (Fig. 2C). As SaTTM catalyzes asymmetrical cleavage of a symmetric substrate it is difficult to judge the directionality of the substrate, a prerequisite for firm mechanistic conclusions. Hence, we solved the ATP-bound form of SaTTM in the presence of Ca2+ ions (Figs. 4C and S5B). The SaTTM⋅ATP state revealed the adenosine base on the side of the plug helix α4, defining the orientation of the triphosphate moiety. It also showed that the catalytic mechanism of SaTTM follows a classical inorganic triphosphatase, but not an AC-IV, providing the structural base of TTM function. Of interest, the predicted inline attack by an activated water nucleophile (see Figs. 4B, 5A, and S4B) onto the γ-phosphate apparently misses a general-base support by protein side chains and is consistent with the high pH optimum of 8.6 for SaTTM. Upon the structural and biochemical confirmation of SaTTM's inability to use calcium as reaction-competent cofactor we wondered how Ca2+ competes with Mg2+. Hence, we performed a titration experiment with increasing concentrations of calcium on a magnesium-catalyzed reaction (Fig. S6A). As expected, we observed at equimolar concentrations a highly reduced activity of SaTTM using PPPi as substrate, and almost no turnover with 5-fold excess of the inhibiting metal ion, indicating that calcium is a potent inhibitor of TTM.
For obtaining structural snapshots of the product state, we determined crystal structures of SaTTM in complex with pyrophosphate (Figs. 4D and 5B). Like the Michaelis complex, the postcatalytic state adopts the closed tunnel conformation. By using the superimposed triphosphate state, the two phosphates can be assigned in the PPi⋅Ca2+ state to the α- and β-positions complexing the calcium ion at metal 1 position (Figs. 4D and S5A). Three water molecules occupy the positions corresponding to the oxygens of the substrate's γ-phosphate moiety. The product-bound states clearly suggest an inhibitory effect of pyrophosphate on TTM activity due to tunnel closure. To confirm this notion, we performed another titration-inhibition experiment (Fig. S6B). The activity of SaTTM was rapidly decreased and already halved when using a ten-thousandth of the substrate concentration. Although this effect plateaus until equimolar concentrations, the inhibition upon 10-fold excess is amplified and with 100-fold excess the enzyme activity was completely abolished.
The product inhibition of SaTTM by pyrophosphate in the closed state suggests that product release may be a rate-limiting step and that the full catalytic cycle of TTMs (Fig. 5C) requires fluctuation between substrate-accessible open states and catalytically active closed states. Of interest, the PPi⋅Mg2+ state shows release of the divalent ion M1 and a conformational change for the bound pyrophosphate group (Fig. 5B and S5E). Calcium-mediated inhibition of TTM may hence derive not only from the different coordination of the γ-phosphate and the polarization of the attacking water but also from a stabilization of the pyrophosphate-bound product state, which blocks subsequent tunnel opening.
Role of the ordered water molecule in substrate binding and catalysis
During SSN analysis we found at a less stringent alignment score of 10−15 that CYTH enzyme orthologs from cluster 3 (including SaTTM) and cluster 10 (including CyaB) belong to the same parental cluster (cluster 3 in Fig. S7B). This suggests generally higher sequence and structural similarity between CyaB and SaTTM-like enzymes compared with other TTM subfamilies, while the triphosphatase-like catalytic mechanism of SaTTM is apparently shared with CYTH enzyme orthologs already well separated at less stringent thresholds. To find a structural rationalization for this phenomenon, we compared substrate-bound states of YpCyaB and SaTTM from the same parental cluster with AtTTM3 from cluster 9 (Fig. 6, A and B). We identified two water molecules involved in substrate coordination and present in all structures of ligand-bound CYTH superfamily proteins. These waters are located opposite the metal ions and directly hydrogen bonded to the α- and β-phosphates, hence termed as α- and β-water (Fig. 6A). Although β-water in SaTTM and YpCyaB made a bidentate contact with the DxY motif, this interaction is absent in AtTTM3 owing to a slightly moved water and an NxF motif. Here, the asparagine is 4 Å away and the phenylalanine lacks the ability to establish a hydrogen bond. This β-water establishes a hydrogen bond with R52 on β3, explaining the well-ordered β-water in AtTTM3 (Fig. 6B). The deviations in substrate coordination can also be observed on the metal ion–binding side. Although SaTTM and YpCyaB share the same arrangement by shaping the binding pocket in close proximity to the α-phosphate using a combination of two conserved aromatic residues (Y168/173 and F133/134), the situation in AtTTM3 is different. Here, a salt bridge ensemble of E167 and K200 is used for direct electrostatic interaction with the γ-phosphate. As we observed this highly similar binding mode of α- and β-phosphates in SaTTM and YpCyaB, we wondered about the impact of these residues on enzyme activity. Hence, we mutated the DxY motif and the central tyrosine, Y168, as well as E135 as a nonactive control owing to its central role in M1 binding (Fig. 6C). As hypothesized, the SaTTM D38A and Y40A mutants showed reduced activity compared with the wildtype. Intuitively, less impact on the catalysis was observed for the conservative D38N and Y40F mutants compared with the alanine mutants, as similar hydrogen bond formation is still possible in the asparagine variant and in the phenylalanine variant we still have the aspartate for β-water coordination. The double mutant had an accumulating effect with less than 10% of wildtype activity, strengthening the importance of the DxY motif in coordinating properly the water–substrate complex for catalysis. Of interest, the Y168A variant showed a 2-fold turnover rate compared with the wildtype enzyme, whereas the corresponding K200A mutant in the AtTTM3 mutant showed reduced activity (
). Y168 interacts with F133 from β5, holding the plug helix in the observed conformation, while not playing a role in direct electrostatic interaction with the triphosphate substrate as known from K200 in AtTTM3 (Fig. 6C). By mutating the tyrosine, the flexibility of the plug helix was increased and tunnel back opening was favored, which results in an enhanced activity of the SaTTMY168A variant.
Structural signature that defines versatile CYTH function
Decades of study on CYTH-containing proteins from the protein family PF01978 established the catalytic mechanisms. However, owing to high sequence similarities between different isofunctional subclusters it proved hard to assign a correct functional annotation to uncharacterized family members. We aimed to integrate our newly gained structural insights with our findings from the SSN and already known CYTH enzymes to establish a detailed understanding about their versatile enzyme activities. The directionality of substrate binding is a major mechanistic determinant on the obtained reaction product when comparing TTMs with AC-IVs (
). As already pointed out, our SSN analysis with a slightly less stringent alignment scores in the SSN merged the AC-IV-specific cluster with the SaTTM cluster. Also, SaTTM shared ∼29% identity with both TTMs and AC-IVs, so it is difficult to annotate these conserved enzymes solely on sequence similarity, as the function-specific features of CYTH enzymes have not been well defined. Our cluster analysis of the CYTH superfamily SSN identifies three key motifs that enable us to predict functions for all family members (Fig. 7, Table S1). According to our analysis, a true AC-IV requires a conserved HF motif that extends the N-terminal signature ExExK motif to HFxxxxExExK, a feature that is absent in all other CYTH enzymes. In the crystal structure of YpCyaB the phenylalanine (F5) is involved in π–π stacking of the adenine base, while the histidine (H4) residue stacks to the phenylalanine via cation–π interaction, suggesting a role in binding ATP in the correct orientation. When F5 is mutated in an AC-IV family member the catalytic activity was 8-fold reduced for kcat (
). In addition, the adenine base stacks to the phenylalanine of the HF motif within a relatively hydrophobic local environment, but no clear directional contacts like hydrogen bonds are made to the adenine amino group indicating that these AC-IVs act not strictly as adenylyl cyclases. The key HF motif is absent in the sequence of all other CYTH proteins; hence, we infer that, in the absence of an N-terminal HF motif, the CYTH enzyme function is either TTM or ThTPase.
On the other hand, the W/YxRxR motif defines the ThTPases, as this subfamily has a conserved aromatic residue (W53 in case of mouse ThTPase) that is capable of stacking the aromatic thiazole moeity of thiamine. Mutation of this residue reduced the plasticity of the ThTPase (
). We find this motif only in one other subfamily (cluster 7) that contains a bi-domain architecture harboring enzymes of the plant TTM1/2 type, which are “TTMs” with a strong substrate preference for pyrophosphate (
). All other clusters have small amino acid substitution at this position (A54 for SaTTM and S59 for YpCyaB). Two arginine residues (R55 and R57 for mouse ThTPase) are required for interaction with the β- and γ-phosphates of ThTP. In AC-IVs, the first arginine residue in this motif is substituted by a small amino acid (V61 in case of YpCyaB), mostly cysteine, valine, or leucine (Fig. S8). Hence, it can be hypothesized that the conserved arginine, which interacts with the γ-phosphate in TTM/ThTPases, is a gate keeper against NTP binding in the cyclase orientation, as the guanidinium group otherwise sterically clashes with the ribose moiety (Fig. 7). The third characteristic feature is an ExE motif on strand β8 close to the C terminus of CYTH proteins (E135 and E137 for SaTTM). These conserved glutamate (E) residues are required for metal binding. In TTM/ThTPases, the second glutamate residue is required for metal 2 binding, which is strictly substituted to alanine in case of AC-IVs. Although this contradicts the generic binding of a second metal ion, a two-metal AC-IV mechanism was suggested before for this subfamily (
). Taken together, this combined analysis provides a basis to delineate proper functional annotation of all CYTH domain-containing orthologs.
Biological role of CYTH
The CYTH superfamily proteins exist in all kingdom of life and can be traced back to the last common ancestor, but the physiological function of this protein family has largely remained elusive. It has been proposed that members of the CYTH superfamily might be involved in the interface of metabolism of phosphoryl compounds and nucleotides. Deletion mutants of the founding member of the CYTH protein superfamily showed no significant cellular defects (
So far, the best characterized CYTH proteins from a functional viewpoint derive from plants. The Arabidopsis TTM3 (AtTTM3) knockout plant showed an anomalous root phenotype, indicating a potential role in root development (
). Although it has been established that plant TTMs act as a polyphosphate hydrolase, recent reports suggest that Brachypodium distachyon and Hippeastrum x hybridum TTM proteins might be able to produce a low amount of cAMP under laboratory conditions (
). In our enzyme function–based cluster analysis (Fig. 1) we identified that all of the so-far characterized plant TTMs were present in the same cluster (cluster 9). As stated above, AtTTM1/2 from cluster 7 are different from the classical TTMs, as they lack the signature motifs for triphosphate binding and γ-phosphate hydrolysis (ExExK and ExE, see Table S1). This cluster still harbors the ThTPase typical W/YxRxR motif that is involved in substrate binding from the metal-opposing site, as well as the described DxY motif required for α/β-water placement (Figs. 6 and S8). This goes along with the observation that AtTTM2 exhibits pyrophosphatase activity in vitro. Upon their deletion, plants became hypersensitive to both virulent and avirulent pathogens and accumulate elevated levels of sialic acid upon infection (
). These contrasting findings are food for future studies on the regulatory and metabolic roles of plant TTMs.
The strictly cAMP-producing CYTH enzymes, the AC-IVs, formed only a small cluster in the whole CYTH SSN (Fig. 1, cluster 10). The biological role of AC-IV is yet to be established. The seminal work discovering AC-IV in Aeromonas hydrophila showed that the cyaB gene is not expressed under standard growth conditions. However, it can complement the function of another adenylyl cyclase, cyaA, when overproduced in a cyaA deletion strain. The cyaB mutant strain had no visible defects and behaved similar to the wildtype strain. Cellular cAMP levels were also noted to be unaltered upon deletion under normal growth condition (
). In our SSN analysis we identified that the huge majority of organisms that harbor an AC-IV are (potential) pathogens. Hence it is conceivable to envision that these AC-IVs might be repressed during normal growth and play no function on cellular cAMP production, but participate during pathogenesis. AC-IV enzymes possess no similarities with other bacterial toxins and are devoid of any signal sequence (
). Clearly, future studies will be required to address how expression of AC-IV enzymes is repressed under normal growth condition and what their precise role during host infection and pathogen maintenance.
In this study we showed that the CYTH enzyme in crenarchaeote S. acidocaldarius previously annotated as adenylyl cyclase is actually a phosphohydrolase of the TTM family. SaTTM prefers inorganic triphosphate substrate over NTPs. Inorganic polyphosphates are considered as energy source readily available for converting di- or monophosphorylated nucleotides to more accessible triphosphorylated ones (like ATP), as metal chelating polymers or as buffers for cellular pH homeostasis (
). In a recent study the overproduced ppx gene in S. acidocaldarius led to a mutant strain lacking PolyP, which was impaired in biofilm formation, motility, archaellum assembly, and adhesion to glass surfaces (
). Although the concentration of triphosphates is not known in S. acidocaldarius, but considering our present finding and previous reports, we envision that TTM in archaea might be directly involved in the polyP metabolism and will indirectly take part in PolyP-mediated regulatory processes and oxidative stress response.
Protein overproduction and purification
SaTTM (Saci_0718) ORF amplified from S. acidocaldarius DSM 639 genome having NcoI and BamHI restriction enzyme sites was cloned into NcoI/BamHI-digested T7-RNA polymerase–based expression plasmid pSA4 (
) of the pSVA1028. The correct DNA sequences were verified by sequencing.
E. coli BL21-Gold (DE3) was used for heterologous overproduction of C-terminal His6-fused SaTTM. Cells were grown in lysogeny broth supplemented with 100 μg/ml ampicillin at 37 °C until an A600 of 0.5. The overproduction of SaTTM was induced by using 0.75 mM IPTG at 18 °C overnight. Cells from 2 L culture were harvested (3200g for 20 min), resuspended in 30 ml lysis buffer (200 mM NaCl, 20 mM Hepes, pH 8.0), and subsequently lysed using French Press. The lysate was incubated at 70 °C for 10 min to reduce E. coli contaminants, and the heat-stable cell-free lysate was collected using 39,000g for 20 min. The supernatant was loaded on a 5-ml Ni-NTA column (Protino) and washed with six column volumes of lysis buffer before eluting the protein with four column volumes of elution buffer (lysis buffer with 500 mM imidazole). The elution was concentrated and further purified by size exclusion chromatography using a HiLoad 26/600 Superdex 200 PG (GE Healthcare) column equilibrated with SEC buffer (200 mM NaCl, 50 mM Tris, pH 9.0). SaTTM protein purity was monitored using a Coomassie-stained 15% SDS-PAGE. The oligomerization of SaTTM was checked on an analytical Superdex 200 10/300 GL column and using the SEC buffer system. Pure protein was concentrated to 2.5 to 5 mg/ml using ultrafiltration and stored at −80 °C until activity assays or crystallization.
Crystallization of SaTTM
To obtain SaTTM crystals 10 mg/ml protein in SEC buffer was mixed with the respective precipitant in an equal volume ratio. Crystals were grown at 4 °C using sitting-drop vapor diffusion. For the ATP state, 10 mM CaCl2 and 5 mM ATP were added to the protein solution. The best diffracting crystal appeared in condition JCSG Core IE7 (0.2 M calcium acetate, 0.1 M MES pH 6.0, 20% (w/v) PEG 8000). The open state of SaTTM (PDB: 7NS8) originated from the ATP screen in a precipitant condition with 0.2 M lithium sulfate, 0.1 M CAPS pH 10.5, 2 M ammonium sulfate. High concentration of sulfate in the precipitation condition competed out the phosphate substrate from the active site. The Michaelis complex was obtained from a SaTTM mixed with 10 mM CaCl2 (or 10 mM MgCl2), 0.5 mM MgCl2, and 5 mM of hexa-polyphosphate, P6i (or tri-polyphosphate, PPPi). Both, PPPi+Ca2+ and PPPi+Mg2+, crystallized in 0.2 M NaCl, 0.1 M Na/KPO4 pH 6.2, 20% (w/v) PEG 1000. Note, the hexa-polyphosphate was highly impure as detected by malachite green assay (data not shown). The product (PPi) inhibited SaTTM in the presence of 10 mM CaCl2, and 5 mM PPi was crystallized in 0.17 M ammonium acetate, 0.085 M sodium acetate pH 4.6, 25.5% (w/v) PEG 4000, 15% (v/v) glycerol, and SaTTM in the PPi-divalent metal-free state in the presence of 10 mM MgCl2 and 5 mM PPi in 0.2 M potassium acetate, 20% (w/v) PEG 3350. Crystals were harvested at 4 °C as temperature seems to have significant effect on crystallization of SaTTM complexes. Crystals were cryoprotected in mother liquor supplemented with 30% glycerol except the PPPi–Mg2+ complex and subsequently flash frozen using liquid nitrogen for data collection.
X-ray data for different states of SaTTM were collected from single crystals at the ESRF. Crystallography data were processed using iMOSFLM in the CCP4i2 package and XDS (
). Atomic models of the other crystallized states of SaTTM were obtained by using the structure of SaTTM in complex with ATP and calcium as a search model for Phaser-MR. Iterative model building and refinements were performed with COOT and phenix.refine (
). Data collection and refinement statistics are summarized in Table S2. The resulting coordinates were deposited at the Protein Data Bank under accession numbers 7NS8, 7NS9, 7NSA, 7NSF, 7NSD, and 7OA2. The structural analysis and figure generation was performed with UCSF Chimera (
Triphosphatase activity of SaTTM was analyzed with the Malachite Green Phosphate Assay Kit (MAK307-1KT) from Sigma-Aldrich according to manufacturer's protocol. Reactions were quenched by mixing 80 μl assay mix into 20 μl working reagent using 96-well PS F-bottom microplates (Greiner). After color development for 30 min at room temperature, the absorbance was measured at 620 nm using a plate reader (spectrometer Infinite 200, Tecan). The results were analyzed with Microsoft Excel and GraphPad Prism. The pH optimum of SaTTM was determined by measuring the amount of reaction product (Pi release) from 5 nM SaTTM, 2 mM MgCl2, and 0.1 mM PPPi buffered in 50 mM Tris and 200 mM NaCl at varying pH values (6.1–9.1) after 5 min incubation. To determine the preferred metal cofactor, 5 nM SaTTM, 2 mM metal ion (Mg2+, Ca2+, Mn2+, Zn2+, Ni2+) and 0.1 mM PPPi were incubated at 75 °C for 5 min buffered in 50 mM Tris and 200 mM NaCl at pH 8.6. The preferred substrate was analyzed by using the same parameters for 10 min with Mg2+ as a cofactor and varying substrates (PPPi, ATP, GTP, AP4A, AP5A) at 0.1 mM concentration. Mutant activities were determined accordingly with 5 min incubation. In order to test the product inhibition, as well as the ability of Ca2+ to inhibit SaTTM, a titration series of 1 nM to 10 mM inhibitor (PPi or Ca2+) was performed by measuring their influence on the product formation of 5 nM SaTTM, 2 mM MgCl2, and 0.1 mM PPPi buffered in 50 mM Tris and 200 mM NaCl at pH 8.6 incubated for 5 and 10 min at 75 °C.
To analyze the enzyme kinetics of SaTTM phosphatase activity, 5 nM enzyme was incubated together with 2 mM MgCl2 and varying PPPi concentrations (0.005–0.5 mM) incubated in 50 mM Tris and 200 mM NaCl at pH 8.6 and 75 °C. To obtain kinetic parameters, the velocities at different substrate concentrations were determined by measuring the Pi formation at different time points and the resulting linear regressions of four measurements were used for the Michaelis–Menten analysis.
To check products on HPLC, the reactions were diluted 5-fold with double-distilled water and the enzyme was denatured by adding 100 μl chloroform followed by 15 s of vigorous shaking, 15 s heat denaturation at 95 °C and snap freezing in liquid nitrogen. The samples were thawed, the two phases were separated by centrifugation (17,300g, 10 min, 4 °C), and 10 μl of the aqueous phase containing the nucleotides was subjected to HPLC analysis. Nucleotides were separated with an anion exchange column (Metrosep A Supp 5 – 150/4.0) with 100 mM (NH4)2CO3 pH 9.25 at a flow rate of 0.7 ml/min as eluent and detected at 260 nm wavelength in agreement with standards.
For the activity analysis of SaTTM toward tetraphosphate, 21 μM SaTTM was incubated in a total volume of 150 μl together with 15 mM Tris pH 7.5 and 4 mM MgCl2 for 3 min at 75 °C. The reaction was initiated by the addition of 0.5 mM tetraphosphate and terminated after 10 min through incubation on ice for 10 min, followed by centrifugation in a spin filter to remove the enzyme (4 °C, 35 min, 10,000g, Amicon-Ultra 0.5, 10 kDa). The flow through was analyzed by ionic exchange chromatography using a Thermo Scientific Dionex ICS-5000+ equipped with a conductivity detector and an EGC500 eluent generator. A volume of 10 μl of the flow through was injected followed by elution with a flow rate of 0.25 ml/min and a KOH gradient starting from 5 mM KOH. The eluent was raised to 12 mM within 5 min, followed by an increase to 20 mM until 10 min and 70 mM within 15 min. KOH concentration was maintained at 70 mM until 26 min, subsequently lowered to 5 mM at 26.1 min followed by equilibration until termination of the run at 27 min.
Sequence similarity network
The protein family PF01928 (CYTH) was used as input for the Enzyme Similarity Tool provided by the Enzyme Function Initiative (
) to generate the SSN. Using the domain boundaries and an E-value of 10−5, the initial network was generated and subsequently modified by applying higher stringent cut offs (10−15, 10−20) as alignment scores for the final SSNs. Each node represents proteins with >40% identity. The resulting networks were subjected to the recently implemented cluster analysis for automatic cluster-WEBLOGO generation, and the final networks were analyzed with Cytoscape 3.5.1 (
The atomic coordinates of the crystal structures of SaTTM, SaTTM⋅PPPi⋅Ca2+, SaTTM⋅PPi⋅Ca2+, SaTTM⋅PPPi⋅Mg2+, SaTTM⋅ATP⋅Ca2+, and SaTTM⋅PPi⋅K+ have been deposited in the Protein Data Bank (PDB ID codes 7NS8, 7NS9, 7NSA, 7NSF, 7NSD, and 7OA2, respectively).
The authors declare that they have no conflicts of interest with the contents of this article.
We thank Dr Xing Ye for his guidance and advise and the group of Prof. Henning J. Jessen (Institute of Organic Chemistry, Albert-Ludwigs-Universität Freiburg) for substrate and IC system supply. We also thank Ralf Pöschke (MarXtal facility, Marburg) for technical support during crystallization and the staff of beamLines ID23-1, ID29, and ID30B at the European Synchrotron Radiation Facility (Grenoble, France) for support with data collection. Furthermore, we thank Dr Lena Hoffmann for cloning the initial SaTTM vector, Julian Grützner and Abbas Sajediabkenar for their help during their internships, Dr Wieland Steinchen for assistance with HPLC measurements, and a referee for indicating mechanistic implications.
M. S. V., L.-O. E., and A. B. conceptualization; M. S. V. and A. B. data curation; M. S. V., R. R. N. N., and A. B. formal analysis; M. S. V. validation; M. S. V., R. R. N. N., M. K. F. M., and A. B. investigation; M. S. V. visualization; M. S. V., R. R. N. N., M. K. F. M., and A. B. methodology; M. S. V., L.-O. E., and A. B. writing-original draft; M. S. V., R. R. N. N., M. K. F. M., S.-V. A., L.-O. E., and A. B. writing-review and editing; S.-V. A., L.-O. E., and A. B. supervision; S.-V. A., L.-O. E., and A. B. funding acquisition; S.-V. A. and L.-O. E. project administration; L.-O. E. resources.
Funding and additional information
This work was supported by Deutsche Forschungsgemeinschaft SFB 987 (Collaborative Research Center 987) and Volkswagen Foundation (Az96727) to L.-O. E., M. S. V., and S.-V. A. A. B. was funded by the Deutsche Forschungsgemeinschaft (447817670).