Structural analysis of human soluble adenylyl cyclase and crystal structures of its nucleotide complexes – implications for cyclase catalysis and evolution Silke Kleinboelting1, Joop van den Heuvel2 and Clemens Steegborn1 1 Department of Biochemistry, University of Bayreuth, Germany €r Infektionsforschung, Braunschweig, Germany 2 Helmholtz-Zentrum fu

Keywords class III; guanylyl cyclases; nucleotidyl cyclases; structure-based comparison; substrate specificity Correspondence C. Steegborn, University of Bayreuth, Department of Biochemistry, Universit€ atsstr. 30, 95447 Bayreuth, Germany Fax: +49 921 557832 Tel: +49 921 557831 E-mail: [email protected] (Received 1 April 2014, revised 29 June 2014, accepted 4 July 2014) doi:10.1111/febs.12913

The ubiquitous second messenger cAMP regulates a wide array of functions, from bacterial transcription to mammalian memory. It is synthesized by six evolutionarily distinct adenylyl cyclase (AC) families. In mammals, there are two AC types: nine transmembrane ACs (tmACs) and one soluble AC (sAC). Both AC types belong to the widespread cyclase class III, which has members in numerous organisms from archaeons to mammals. Class III also contains all known guanylyl cyclases (GCs), which synthesize the cAMP-related messenger cGMP in many eukaryotes and possibly some prokaryotes. Among mammalian ACs, sAC is uniquely regulated by bicarbonate, and has been proposed to be more closely related to a bacterial AC subfamily than to mammalian ACs, on the basis of sequence comparisons. Here, we used crystal structures of human sAC catalytic domains to analyze its relationships with other class III ACs and GCs, and to study its substrate selection mechanisms. Structural comparisons revealed a similarity within an sAC-like subfamily but no family-specific structure elements, and an unexpected sAC similarity to eukaryotic GCs and a potential bacterial GC. We further solved novel crystal structures of sAC catalytic domains in complex with a substrate analog, unprocessed ATP substrate, and product after soaking with ATP or GTP. The structures show a novel ATP-binding conformation, and suggest mechanisms for substrate association and recognition. Our results could explain the limited substrate specificity of sAC, suggest how specificity is increased in other cyclases, and indicate evolutionary relationships among class III enzymes, with sAC being close to a putative ‘ancestor’ cyclase. Database Coordinates and structure factors for the novel sAC-cat structures described have been deposited with the Worldwide PDB (www.pdb.org): ApCpp soak (entry 4usu), ATP + Ca2+ soak (entry 4usv), GTP + Mg2+ soak (entry 4ust), ATP soak (entry 4usw).

Abbreviations AC, adenylyl cyclase; ApCpp, a,b-methylene-ATP; GC, guanylyl cyclase; PDB, Protein Data Bank; sAC-cat, soluble adenylyl cyclase catalytic domain; sAC, soluble adenylyl cyclase; sGC, soluble guanylyl cyclase; tmAC, transmembrane adenylyl cyclase.

FEBS Journal 281 (2014) 4151–4164 ª 2014 FEBS

4151

Crystal structures of sAC nucleotide complexes

Introduction cAMP is a ubiquitous second messenger that regulates functions from bacterial transcription and fungal virulence to mammalian pH homeostasis and memory [1–3]. The formation of cAMP from ATP is catalyzed by adenylyl cyclases (ACs), an enzyme family that can be divided into six classes on the basis of sequence homologies in their catalytic domains [4,5]. All eukaryotic ACs belong to class III, which is the largest class, with members from almost all organisms. This class also contains all known guanylyl cyclases (GCs). In mammals, eight GC isoforms and 10 AC isoforms, GC-A to GC-H and AC1 to AC10, mediate distinct physiological functions. The ACs can be divided in nine transmembrane ACs (tmACs) (AC1–AC9) and one soluble AC (sAC) (AC10) [5]. tmACs are mainly regulated by heterotrimeric G-proteins, thereby mediating intracellular propagation of extracellular signals recognized by G-protein-coupled receptors [6]. sAC has been detected at various intracellular locations, such as the nucleus, mitochondria, and centrioles [7], and is regulated by calcium, bicarbonate, pH, and the ATP concentration [8,9]. Overall sequence comparisons previously suggested sAC to be an ‘evolutionarily’ old mammalian AC, being more closely related to bacterial ACs than to mammalian tmACs and GCs [5,10]. However, the general structure and features relevant for class III catalysis are identical or very similar between sAC, tmACs, and other class III cyclases. Several crystal structures of class III catalytic domains show a conserved dimeric architecture [11]. In mammalian receptor GCs and many bacterial ACs, homodimerization of the protein with a single catalytic domain creates two symmetrical active sites at the dimerization interface. Mammalian tmACs and sAC, in contrast, have, on one polypeptide chain, two structurally similar catalytic domains (C1 and C2), which form a pseudoheterodimer (from now on referred to as heterodimer). As C1 contains a specific set of catalytic residues, whereas C2 contributes another, complementary set of catalytic residues, only one of the pseudosymmetric sites at the dimer interface is active [5]. The inactive site apparently serves as a regulator-binding site [5,12,13]. Each catalytic domain monomer features the conserved class III cyclase architecture, with a central sevenstranded b-sheet shielded from solvent by helices a2, a3, and a5. A small helix a1 and a peripheral b7/b8module undergo an active site closure movement during catalysis [14]. Within the active site, two conserved Asp residues (Asp47 and Asp99; numbers refers to human sAC if not stated otherwise) coordinate two 4152

€lting et al. S. Kleinbo

divalent cations, ion A and ion B [12,14,15]. Ion B interacts with the b-phosphate and c-phosphate, and supports substrate binding, whereas ion A appears to bind transiently to the a-phosphate to support turnover [12,14]. The ATP base inserts in a hydrophobic cleft, and is recognized through hydrogen bonds between the adenine ring N1 and a conserved Lys (Lys334), and between the 6-amino group and a conserved Thr (Thr405). Thr405 is functionally replaced by an Asp in tmACs [5,12,15]. Lys and Thr/Asp are replaced in GCs by Glu and Cys, respectively, to adapt to the hydrogen bond donor/acceptor pattern of guanine. However, additional factors contribute to substrate selection, as AC specificity is not easily switched to GTP through mutagenesis [5]. Also, ATP competes efficiently with the substrate GTP in a bacterial GC [16], and GTP competes with ATP in human sAC [12], suggesting that NTP discrimination occurs only after binding. Consistently, direct interactions between base and specificity-mediating residues have been observed in some AC–nucleotide complex structures, but are missing in others [12,17]. However, the details and dynamics of interactions relevant for substrate recognition and turnover remain to be fully understood. Minor structural differences between sAC and tmAC isoforms appear to be also responsible for their different physiological regulation and drug sensitivity. Structures of human sAC [12] show, for example, a small C2 b2–b3 insertion that prevents binding of the tmAC-activating diterpene forskolin. It converts this site, together with the conserved sAC residue Arg176, into a binding site for the sAC activator bicarbonate [12]. sAC is inhibited specifically by KH7, and preferentially by catechol estrogen and other ligands for a central binding site that features subtle sequence variations between isoforms [18–20]. The unusually low apparent ATP affinity of sAC [8] and its insensitivity to the pharmacological tmAC ‘p-site’ inhibitors, which resemble the ATP/cAMP substrate/product [9], again indicate subtle differences between sAC and tmACs. Such differences are often not recognizable from sequence comparisons, however, and are only partly understood. Here, we show, through detailed structure and sequence comparisons, that sAC-like enzymes, defined on the basis of sequence homology, are recognizable structurally but lack larger subfamily-specific structural elements. We also find hints regarding a close relationship between sAC and GCs, and describe crystal structures of sAC complexes resulting from soaking with GTP and the substrate ATP, respectively. The structures reveal a novel ATP-binding conformation and yield FEBS Journal 281 (2014) 4151–4164 ª 2014 FEBS

€lting et al. S. Kleinbo

insights into nucleotide binding and base recognition, which improve our understanding of class III catalysis and suggest pathways of class III evolution.

Results Structural comparison of sAC and other class III ACs and GCs sAC represents a uniquely regulated class III AC subfamily. On the basis of sequence comparisons, mammalian sAC is also assumed to be more closely related to bacterial ACs than to tmACs [5,10]. We used the recently solved human sAC catalytic domain (sAC-cat) crystal structures [12] for improved similarity analyses to reveal class III subfamily-specific structural features and to refine their evolutionary relationships. We first searched the Protein Data Bank (PDB) for structural sAC homologs by using the DALI server [21] and structures of apo-sAC and of an sAC–a,b-methylene-ATP (ApCpp) complex to identify the order of structural similarity with other class III cyclases. A GC [the heterodimeric human soluble GC (sGC)1 b3subunit] was the top hit for sAC–ApCpp as the search structure, and a tmAC was found in position 7, after several bacterial ACs/GCs. In the apo-sAC search, the bacterial sAC-like AC CyaB from Pseudomonas aeruginosa was the top hit, and the first tmAC was ranked at position 20 after human sGC and several bacterial ACs/GCs (Table 1; Table S1). Only after all of the ~ 40 different class III cyclases (or conformations thereof) of the PDB did remotely related diguanylyl cyclases [22] and DNA-polymerases [23] appear, which share a core fold of five b-strands flanked by three a-helices with class III cyclases. PDB structural similarity searches with the 3D-BLAST algorithm [24,25], which is based on translating small local structural elements into a ‘module sequence’, yielded similar results. With the sAC–ApCpp complex as the search structure, seven of the top 10 hits were different conformations of the sAC-like CyaC from Spirulina platensis [14]. The GC Cya2 from Synechocystis [16] was found at position 3, mycobacterial Rv1900c [26] as the first non-sAC-like AC at position 6, and the first tmAC at position 10 (Table 1). Surprisingly, the P. aeruginosa sAC-like cyclase CyaB [27] was not identified by 3D-BLAST (Table S1). Similar results were obtained with apo-sAC as the search structure (Table 1). The highest ranked mammalian cyclase was human sGC at position 10, and the first tmAC appeared at position 14 (Table S1). Taken together, the structure search results confirm class III cyclases as a structurally defined protein family, suggest a closer FEBS Journal 281 (2014) 4151–4164 ª 2014 FEBS

Crystal structures of sAC nucleotide complexes

structural relationship of sAC with bacterial ACs, and in particular with a bicarbonate-activated sAC-like subfamily, and indicate that even GCs might be more closely related to sAC than tmACs. We then compared sAC-like and tmAC-like cyclases in detail to identify structural features typical for one of the subfamilies. Comparison of sAC-cat (PDB ID 4CLF) and a tmAC heterodimer (PDB ID 1ASZ) showed a lack of C2 helix a1 in sAC (Fig. 1A). This feature is unique to mammalian sAC, however, as a1 is important for catalysis [5], and it is thus always present in the symmetric sites of homodimeric cyclases, including bacterial sAC-like homodimers. Only in heterodimeric cyclases such as mammalian sAC and tmACs could this site evolve to adapt to different regulators [12]. In sAC C1, the b7–b8 loop is shortened, and a four-residue insertion in the C2 b2–b3 loop leads to insensitivity of sAC to the tmAC-specific activator forskolin [12]. In tmAC C1, a1 is shortened as compared with other cyclases, and the linker to a2 is extended in sAC C1 and tmAC C2. However, the structural differences between core elements are small and mainly found in heterodimeric cyclases and regions not directly involved in catalysis. Even comparison of the most deviating tmAC C2 with sAC C2 and CyaC results only in few slightly shifted helices (Fig. 1A), indicating that they represent different conformational states rather than structurally distinct subfamilies. Thus, there appear to be no pronounced structural elements typical for sAC-like versus tmAClike subfamilies. Despite a lack of larger, subfamily-specific structure differences, our class III structure comparisons enable a significant improvement in sequence alignments. Comparison of a structure-based alignment (Fig. 1B) with a representative previous sequence-based alignment [5] shows that b1 and a2 of sAC C2 had been assigned wrongly, owing to the hitherto unrecognized lack of a1 in sAC C2 [5,14]. Furthermore, the sAC C2 b2–b3 linker was slightly misaligned before (two-residue shift to the N-terminus), causing a misalignment of sAC a3 [14]. A significantly shortened sAC C1 b7– b8 loop previously caused an erroneous shift of b8 residues 211–218 in alignments [5,14]. The structure-based alignment shows that bacterial cyclases (e.g. GRESAG4.1 from Trypanosoma brucei) often have an additional or longer helix between a3 and b4, and the C2 domains of sGC and tmACs have a three-residue loop extension between a2 and b2 at the protein surface. Interestingly, our structure-based alignment reveals that sAC C1 Arg176, which plays a central role as a trigger arm in bicarbonate-dependent sAC activation [12], is conserved in mammalian sACs but not in sAC-

4153

€lting et al. S. Kleinbo

Crystal structures of sAC nucleotide complexes

Table 1. Structural comparison of sAC structures against the PDB using DALI and 3D-BLAST (top 10 hits). Multiple hits resulting from comparison with separate chains of class III cyclase PDB entries were removed. sAC–ApCpp structure (4CLK)

Apo-sAC structure (4CLL) 3D-BLAST

DALI

2WZ1-A

Human sGC-1b3

1WC4-B

2W01-A

GC Cya2 from Synechocystis

1WC3-Ba

3R5G-Ba

AC CyaB from P. aeruginosa

2W01-B

3UVJ-A

Catalytic domain of the heterodimeric human sGC-1

1WC5-Ba

3ET6-A

Catalytic domain of a eukaryotic GC-b from C. reinhardtii

2BW7-Ba

1YBT-B

M. tuberculosis AC RV1900c CHD

1YBU-D

1AZS-A

Complex of Gs-a with the catalytic domains of mammalian AC (forskolin) Mycobacterial AC RV1264, holoenzyme, active state AC/GC from Si. pomeroyi

AC CyaC from S. platensis in complex with ATPaS

3C15-A

1Y11-A

3MR7-A

1WC1-Ca

a

3D-BLAST

DALI a

AC CyaC from S. platensis in complex with ApCpp and europium AC CyaC from S. platensis in complex with ApCpp and strontium GC Cya2 from Synechocystis

AC CyaB from P. aeruginosa

1WC5-Ba

AC CyaC from S. platensis in complex with ApCpp in the presence of bicarbonate

1y10-C

Mycobacterial AC RV1264, holoenzyme, inhibited state

2W01-D

GC Cya2 from Synechocystis

1YBU-C

M. tuberculosis AC RV1900c CHD in complex with ApCpp GC Cya2 from Synechocystis

1WC3-Ba

AC CyaC from S. platensis in complex with ApCpp and strontium AC CyaC from S. platensis in complex with inhibitor catechol estrogen

1YBT-A

M. tuberculosis AC RV1900c CHD

1WC0-Ba

AC CyaC from S. platensis in complex with ApCpp

2BW7-Ba

AC CyaC from S. platensis in complex with inhibitor catechol estrogen AC CyaC from S. platensis in complex with ApCpp and strontium Human sGC-1b3

1WC4-Ba

AC CyaC from S. platensis in complex with ApCpp and europium

2BW7-Aa

AC CyaC from S. platensis in complex with inhibitor catechol estrogen

1WC1-Ca

AC CyaC from S. platensis in complex with ATPaS

1WC6-Aa

AC CyaC from S. platensis in complex with ATPaS in the presence of bicarbonate

2WZ1-B

Human sGC-1b3

3R5G-A

a

AC CyaC from S. platensis in complex with ApCpp in the presence of bicarbonate AC CyaC from S. platensis in complex with inhibitor catechol estrogen M. tuberculosis AC RV1900c CHD in complex with ApCpp

2W01-B

1WC0-Ba

AC CyaC from S. platensis in complex with ApCpp

1WC3-Aa

1WC1-Ca

AC CyaC from S. platensis in complex with ATPaS AC CyaC from S. platensis in complex with ATPaS in the presence of bicarbonate Complex of Gs-a with the catalytic domains of mammalian AC: complex with pyrophosphate and Mg2+

2WZ1-A

1WC6-Ca

1WC5-Da

1WC1-Ca

AC CyaC from S. platensis in complex with ApCpp in the presence of bicarbonate AC CyaC from S. platensis in complex with ATPaS

2BW7-Ba

Bicarbonate-activated sAC-like AC.

4154

FEBS Journal 281 (2014) 4151–4164 ª 2014 FEBS

€lting et al. S. Kleinbo

A

Crystal structures of sAC nucleotide complexes

C

B

Fig. 1. Structure-based comparison of class III cyclase catalytic domains. (A) Overlay of the human sAC C2 domain (cyan), a tmACII C2 domain (yellow) and the sAC-like AC CyaC (red). Helices are shown as cylinders. (B) Structure-based alignment of class III catalytic domains whose crystal structures are available, including members from five subclasses: mammalian sAC and tmACs; sAC-like bacterial ACs; mammalian GC; and bacterial ACs and putative GCs. The catalytic residues discussed are highlighted in yellow. Human sAC C1 secondary structure elements are shown at the top. (C) Phylogenetic relationships between various class III ACs and GCs based on the structurebased alignment in (B). *Rv1625c is an AC converted to GTP-selectivity through three single-site mutations.

FEBS Journal 281 (2014) 4151–4164 ª 2014 FEBS

4155

€lting et al. S. Kleinbo

Crystal structures of sAC nucleotide complexes

like AC from bacteria (Fig. 1B). In tmACs or sGC, the Arg is replaced by an Asp, consistent with their insensitivity to bicarbonate. sAC Arg176 was previously aligned with the catalytic Arg416 in sAC C2, and with the corresponding catalytic Arg in other cyclases, e.g. Arg1150 in the homodimeric sAC-like AC CyaC [5]. The CyaC residue that is structurally equivalent to sAC Arg176 is Ala1149, however, next to the catalytic Arg1150. In homodimeric sAC-like enzymes, sAC Arg176 might be functionally replaced in bicarbonate-dependent activation by the slightly shifted catalytic Arg. During evolution, heterodimers apparently enabled optimization of bicarbonate recognition through an Arg at position 176, while the no longer needed catalytic Arg at position 177 was lost. Another residue that is important for catalysis in homodimeric AC, and also for bicarbonate recognition in sAC, is Lys95/334. sAC C2 Lys334 recognizes the substrate base, and the corresponding sAC C1 residue Lys95 contributes to bicarbonate binding [12]. In homodimeric sAC-like ACs, this residue thus combines both functions. Consistently, a homodimeric sAC-like AC mutated at this position appeared to be bicarbonate-

insensitive, but its activity was low and hard to evaluate [28]. Generating a phylogenetic tree from the structurebased alignment resulted in three branches (Fig. 1C): a main branch consisting of all eukaryotic cyclases except T. brucei GRESAG4.1 and mycobacterial Rv1625c and Rv1264, and two branches containing all other bacterial cyclases and GRESAG4.1. The overall classification is similar to former phylogenetic analyses [5,10,29], except for the positions for sAC C1 and C2. A former tree classified sAC C1 as a bacterial AC domain and sAC C2 as a tmAC-C1-like domain [5]. In our tree, both sAC domains form an extra branch, which separated early from mammalian tmGC/sGC, but also from bacterial cyclases, including sAC-like ACs. sAC thus seems to connect these branches, and it is tempting to speculate that sAC, and in particular homodimeric bacterial sAC-like AC, are evolutionarily ‘old’ and close to a putative ‘ancestor cyclase’. Separated from the eukaryotic GCs, the putative bacterial GC Cya2 [16] appears in the sAC-like branch, suggesting that GCs might have separated from ACs twice during evolution.

Table 2. Diffraction data and refinement statistics.

Data processing Space group Cell dimensions ( A) Resolution range ( A) Unique reflectionsa Completeness (%) Rmeasb (%) CC1/2 Multiplicity Refinement Refinement resolution ( A) Reflections used Number of protein atoms Number of solvent atoms Number of ligand atoms rmsd bond lengths ( A) rmsd bond angles (°) Average B-factor ( A2) Final R/Rfreec,d (%)

ApCpp

ATP/Ca2+ soak

ATP

GTP/Mg2+ soak

P63 a = b = 100.8, c = 96.8 64.83–1.95 40 732 (2961) 14.2 (1.8) 100 (100) 9.1 (100.0) 100 (72) 6.9 (7.1)

P63 a = b = 100.9, c = 97.2 87.36–2.00 37 919 (2681) 18.1 (2.2) 100 (100) 7.4 (86.5) 100 (76) 6.3 (6.5)

P63 a = b = 101.1, c = 96.8 87.55–2.05 35 253 (2448) 13.3 (1.8) 100 (100) 10.9 (99.2) 100 (64) 5.7 (5.6)

P63 a = b = 100.3, c = 97.4 86.90–1.90 43 826 (3281) 11.4 (1.9) 100 (100) 8.1 (71.8) 100 (68) 4.0 (3.8)

1.95 38 816 3716 203 58 0.019 2.0 44.0 15.9/21.2

2.00 35 990 3629 185 29 0.019 2.0 43.5 16.0/20.9

2.05 33 478 3644 202 46 0.019 2.0 36.8 16.4/21.2

1.90 41 625 3654 221 30 0.019 2.0 38.6 16.1/20.2

a

NumbersP in parentheses are for the outermost shell. P pffiffiffiffiffi n

b

Rmeas =

n1

h

PP

jhI iI j

I

where I is the intensity of an individual measurement, hI i is the corresponding mean value, and h and n are the

h

indices and redundancies of the reflections. P jjFobs jkjFcalc jj P R  factor ¼ where |Fobs| is the observed and |Fcalc| the calculated structure factor amplitude. jF j

c

obs

d

Rfree was calculated from 5% of measured reflections omitted from refinement.

4156

FEBS Journal 281 (2014) 4151–4164 ª 2014 FEBS

€lting et al. S. Kleinbo

Crystal structures of sAC nucleotide complexes

apo-sAC-cat crystals thus can take up nucleotides, and small differences in the cocrystallized complex are seen only in loop regions 132–142 (C1 a3–b3 loop) and 452–458 (C2 b7–b8 loop) and for the very C-terminal residues not visible in the cocrystal structure [12]. The C2 b7–b8 loop, which appears to close upon substrate binding or turnover [14,15], is in an even more closed conformation than the closed form of CyaC, the most closed conformation observed so far (Fig. 2B) [14]. However, apo-sAC already assumes a closed state with C1 a1 and C2 b7–b8 tilted even more towards the center than in this closed CyaC state [12]. Binding of ApCpp thus induces no significant further movement of C2 b7–b8. This finding indicates that the open state, which was previously suggested to be the enzyme conformation binding the substrate [14], is not required as

Substrate binding and recognition Mammalian sAC appears to be connected to roots of phylogenetic branches that evolved into both ACs and GCs. sAC shows particular structural similarity to GCs (see above) and substantial GC side activity (7.5%) [12], but the responsible factors are unknown. We therefore used human sAC, as a well-crystallizing AC [12,30], for crystallographic studies on sAC and general class III features involved in substrate recognition and turnover. We first solved a crystal structure of sAC-cat soaked with the substrate analog ApCpp (Table 2), resulting in an sAC-cat–ApCpp complex almost identical to an sAC-cat–ApCpp structure obtained through ApCpp cocrystallization (PDB ID 4CLK [12]; rmsd  Fig. 2A). The for 451 Ca positions of 0.22 A;

A

B

E

C

F

D

G

Fig. 2. Structural studies on adenine nucleotide binding to human sAC. (A) Comparison of the sAC–ApCpp complex obtained by soaking (cyan) with a cocrystallized sAC–ApCpp complex (PDB ID 4CLK) (orange). Only the soaked-in ApCpp ligand is shown (stick representation colored according to atom type), and is overlaid with 2Fo – Fc electron density (blue; contoured at 1r). (B) Overlay of the sAC–ApCpp complex (cyan) and CyaC in an open conformation (gray) and a closed (red) conformation. The sAC–ApCpp ligand is shown in stick representation, colored according to atom type, with Ca2+ as a yellow sphere. (C) AC activity of sAC with various divalent cations: 10 mM Ca2+ and 10 mM Mg2+, or 20 mM of either Ca2+ or Mg2+ (error bars: standard deviation; N = 2). (D) Overlay of the sAC–ATP–Ca2+ soak, which resulted in a pyrophosphate complex (yellow; water as a red sphere) and the sAC–ApCpp complex (cyan; water and Ca2+ labels underlined) containing a six-fold-coordinated Ca2+. (E) Active site of the sAC–ATP structure overlaid with 2Fo – Fc electron density (blue; contoured at 1r) for the ligand, shown in stick representation. The Na+ is shown as a purple sphere. (F) Interaction scheme for ATP in the syn conformation, which is apparently slightly preferred in this complex with sAC. Interactions with side chains are indicated by black dots, and interactions with the backbone by gray lines. (G) Comparison of the two different conformations for the adenine in the sAC–ATP structure (green) and the sAC–ApCpp complex (cyan). Ligands are shown in stick representation, colored according to atom type.

FEBS Journal 281 (2014) 4151–4164 ª 2014 FEBS

4157

Crystal structures of sAC nucleotide complexes

a stable conformation for substrate accommodation. However, the b7–b8 loop, and in particular its Lys451, severely obstructs active site access and possesses relatively high B-factors, consistent with a dynamic function. We thus assume that the open state is formed transiently for nucleotide entry. Substrate binding to class III cyclases was proposed to involve a nucleotide distortion facilitating turnover [12,16], and we therefore tried to obtain an sAC complex with unmodified ATP. In a previous attempt, soaking apo-sAC-cat crystals with ATP and MgCl2 resulted in turnover and an sAC–PPi complex without bound cAMP [12]. We now soaked apo-sAC-cat crystals with ATP and CaCl2 instead of Mg2+. Ca2+ occupies the ion B site, where it acts as a better anchor for the ATP phosphates, but it was suggested not to function as well as ion A, which supports turnover [5,14,31]. Testing sAC activity did indeed show slow turnover (~ 1 nmol cAMP min1mg1) in the presence of Ca2+, ~ 3500-fold and ~ 7000-fold slower than with Mg2+ and with a mixture of both divalent ions, respectively (Fig. 2C). Nevertheless, solving the  structure of sAC soaked with ATP/Ca2+ at 2.00 A resolution also revealed mainly an sAC-cat–PPi complex (Table 2; Fig. 2D). The crystallized sAC thus can undergo all conformational changes essential for turnover, and we assume that, in fact, the closed conformation stabilized in the sAC crystals corresponds to an active state, accelerating the reaction as compared with solution even in the presence of Ca2+ alone. The obtained sAC–PPi complex shows no significant differences from the sAC–PPi complex resulting from soaking with ATP/Mg2+ [12]. Ion and phosphate arrangements are also similar to those in the sAC–ApCpp–Ca2+ complex, except that the Ca2+ is not coordinated by the Ile48 carbonyl oxygen, resulting in a reduced, two-fold coordination by the protein that might precede product dissociation (Fig. 2D). Additional density for the ribose at very low contour levels might indicate a low, perhaps ~ 10%, occupancy of the substrate ATP. However, reducing the soaking time did not increase the occupancy of ATP. The reaction with Ca2+ appears to be too fast in the crystal for trapping of an sAC–ATP–cation Michaelis complex. Cellular ATP mainly exists as a complex with a divalent cation, and normally binds together with this partner to enzymes. As even the less active Ca2+ led to ATP turnover, however, we soaked apo-sAC crystals with ATP without any divalent cation (ATP sodium salt, 98.5% purity) to analyze binding contributions from ATP moieties other than the phosphates.  resolution structure of an sAC– The resulting 2.05 A 4158

€lting et al. S. Kleinbo

ATP complex (Table 2; Fig. 2E) featured well-defined electron density for all three phosphates and the ribose. Thus, in contrast to the sAC–PPi product complex obtained from ATP soaks in the presence of divalent cations, the unprocessed substrate is bound in this structure. Spherical density indicated a cation occupying roughly the ion B site, with six-fold coordination (Fig. 2E,F), and serving as binding partner for the ATP phosphates. The distance of the ion to Asp47 is  and and its distance to Asp99 is 2.4 A.  Phos2.5 A,   phate oxygens have distances of 2.1 A and 2.4 A, respectively, and weakly defined water molecules serve as the fifth and sixth coordination partners (2.4 and  Fig. 2F). Both these distances and valence cal2.0 A; culations would be compatible with Na+, which is present in high concentration in the crystallization solution, but also with Ca2+, which might be present as an impurity [32]. Building either metal ion into the structure and using the density as an indicator did not give a clear result (Fig. S1A,B). However, the B-factor 2) shows the expected similarity refined for Na+ (47 A 2) [32]. Ca2+ refined to the coordinating atoms (51 A 2  to a higher value (75 A ), which likely constitutes an artificial compensation effect blurring its larger diffraction power. Considering B-factors, the different coordination from that of the available sAC–ApCpp–Ca2+ complex, and the high Na+ concentration in the crystallization drop, it appears highly likely that the ion is an Na+. In contrast to the ion-bound phosphates, density for the ribose and, in particular, for the purine base is present but very weak (Fig. 2E), indicating conformational freedom for these substrate parts. It appears that the interactions with the phosphate tail dominate nucleotide binding even in the presence of nonoptimal cations. The sAC–ApCpp and sAC–ATP complexes show  the same overall protein conformation (rmsd of 0.2 A for 452 Ca atoms), but the nucleotides are bound sig for all ligand nificantly differently (rmsd of 1.9 A  Fig. 2G). In the atoms, maximum deviation of 4.5 A; sAC–ApCpp complex, Arg176 points towards the substrate and might contribute to sugar binding, indicating that this regulatory residue might have a catalytic function [12]. In the ATP complex, Arg176 points to the regulatory site, but does not assume the conformation observed in bicarbonate-bound sAC [12], and it does not contact the substrate, possibly explaining the weakly defined sugar position (Fig. 2E). The phosphate groups of ATP are coordinated, similarly to  and the backbone amides of ApCpp, by Thr52 (2.9 A)  and Gly50 (2.9 A)  (Fig. 2F). Further Phe51 (2.9 A)  and Lys144 interactions are provided by Ser49 (2.7 A)  (2.3 A). The conserved ions coordinating Asp47 FEBS Journal 281 (2014) 4151–4164 ª 2014 FEBS

€lting et al. S. Kleinbo

Crystal structures of sAC nucleotide complexes

 and Asp99 (2.6 A),  besides coordinating the (2.5 A) + Na , also interact directly with the phosphates, but this interaction could be an artefact caused by the low pH in the crystallization drop. Asn412 forms a hydro with the ring oxygen of the ribose. gen bond (3.4 A) The catalytic Arg416 interacts with the 20 OH of the  The adenine base is not well defined by ribose (3.8 A). electron density, indicating a flexible position (Fig. 2E; Fig. S1C–F). ATP appears to mainly assume a syn conformation, with the adenine rotated 180° around the N-glycosidic bond as compared with nucleotides in the sAC–ApCpp complex and other class III adenine nucleotide complexes. In this conformation, the purine is inserted between Phe296, Phe336, and Phe338, and  coordinated by the carboxyl oxygen of Val406 (2.7 A)  and the side chain of Thr405 (3.5 A). Lys334, which is necessary for the recognition of the adenine, is rotated away and interacts only via a water molecule with the purine, as observed in the ApCpp structure. The observed ATP position might correspond to an initial substrate interaction pose (see Discussion). sAC can bind and convert GTP, resulting in ~ 7.5% GC side activity [12,16]. sAC AC activity is also inhibited by GTP, with a potency resembling the Km for A

B

ATP (Ki ~ 1 mM) [12], suggesting similar binding affinities for both nucleotides. The higher AC activity at high substrate concentrations must thus be a result of more efficient turnover (higher kcat), indicating that the substrate base is recognized only during turnover. To confirm that GTP has a comparable sAC affinity to that of ATP, we determined the Kd for GTP in the presence of Ca2 +. The Kd(GTP) of 1.22  0.07 mM (Fig. 3A) is comparable with the apparent ATP affinity (2.03  0.42 mM [12]) and the Ki(GTP) previously calculated on the basis of competitive inhibition (Ki ~ 1 mM). This result supports the proposed catalytic mechanism with base discrimination during turnover rather than initial substrate binding (see [12] and Discussion). To further analyze nucleotide binding and recognition, we tried to obtain an sAC–GTP complex by soaking apo-sAC crystals with GTP and Mg2 +. The  resolution structure (Table 2) reveals resulting 1.90 A only density for the product PPi coordinated via an Mg2+ (Fig. 3B), showing that even the low GC activity is sufficient for efficient GTP turnover. The structure looks identical to the PPi complex resulting from  (356 soaking with ATP/Mg2+, with an rmsd of 0.2 A C

D

Fig. 3. GTP binding to human sAC and a model for base discrimination during turnover. (A) Binding affinity of GTP for sAC-cat in the presence of Ca2+ yields a Kd of 1.22  0.07 mM (error bars: standard deviation; N = 2). (B) Active site of the GTP/Mg2+-soaked sAC crystals resulting in a PPi–Mg2+ complex. The ligand is shown in stick representation, and overlaid with 2Fo – Fc electron density contoured at 1r. (C) GTP ligands, shown in stick representation, were modeled in the sAC structure based on either of the two adenine conformations observed for ATP (green ligand) and ApCpp (cyan ligand). (D) Model for substrate recognition during catalysis. ATP first binds with ion B in a loose conformation or an ensemble of conformations, including the syn orientation seen in our sAC–ATP complex. Adjustment of ion B and binding of ion A initiate the SN2 reaction with in-line geometry. The a-phosphate moves towards the 30 OH, and elongation of the bond to the leaving group is supported by a shift of the AMP/cAMP molecule towards Lys334, which transiently interacts with the base. Once the bond to PPi is broken, the concomitantly formed cAMP can shift the ribose and base back to their previous positions.

FEBS Journal 281 (2014) 4151–4164 ª 2014 FEBS

4159

Crystal structures of sAC nucleotide complexes

Ca atoms) and the same PPi position. We thus generated potential sAC–GTP complexes by modeling the GTP ligand on the basis of the two adenine conformations that appear to exist in sAC structures (Fig. 3C). Both conformations result in a short distance between  In addition, guanine O6 and Thr405 (3.7–3.8 A).  and N3 Asn412 is close to the 2-amino group (2.9 A)  (2.6 A) in the ‘ATP conformation’. N1, which is supposed to be recognized by Lys334, is oriented away and interacts with the carbonyl oxygen of Val406. In the ‘ApCpp’ conformation, the ring nitrogen N7 might  but the amino group interact with Asn412 (3.7 A),  which would have to comes close to Phe336 (2.2 A), adjust for such a complex. O6 is well coordinated by  backbone the backbone nitrogen of Gly98 (3.2 A),  oxygen of Val406 (2.9 A), and side chain of Thr405  O6 and the nearby N1 are accessible for (3.8 A). Lys334, but, because of its protonation, N1 cannot form a hydrogen bond with Lys334. Thus, only the ATP-like syn-oriented guanine has a hydrogen-bonding pattern that is fully compatible with the adeninerecognizing residues Thr405 and Lys334. Interestingly, most ACs have an Asp at the position of sAC Thr405, which can be a hydrogen bond acceptor (for the adenine amino group) or donor (for the guanine oxo function). The Asp can only act as a hydrogen bond acceptor, excluding guanine in both orientations, syn and anti, and thereby probably contributing to the higher ATP specificity of these ACs.

Discussion The ubiquitous nucleotidyl cyclase class III comprises enzymes differing in function, regulation, and substrate specificity [2,5,33]. Their catalytic cores show a conserved structure, whereas their sequences and neighboring domains have largely diversified during evolution [5,33]. Only a small set of catalytically or structurally important residues are fully conserved [5]. Structural comparisons can provide evolutionary information that is hard to obtain at a sequence level [24]. Two different structure search tools did indeed first identify other class III cyclases, and then remotely related diguanylyl cylases and DNA-polymerases. Surprisingly, 3D-BLAST [24], which one might expect to be less sensitive for conformational differences of similar folds, showed the only failure to recognize a class III AC (CyaB) and ranked some class III cyclases lower than more remotely related proteins. Slightly different results were also obtained according to the search structure conformation. It is unclear whether they are mainly caused by differences of the algorithms in recognizing subtle structure relationships or differences in recognizing different conformations of a 4160

€lting et al. S. Kleinbo

fold. 3D-BLAST first recognized a block of different CyaC conformations, e.g. independent of the search structure, whereas DALI yielded a block of different sAC-like cyclases. A systematic evaluation of the algorithms will be required to clarify their different sensitivities regarding related structures and conformations. However, all searches indicated a closer relationship of sAC to other bicarbonate-activated ACs, which is thus independent of specific algorithm preferences. However, no major structural differences between the sAC-like and tmAC subfamilies were found, indicating that the subfamily characteristics recognized in these structural comparisons are subtle and remain to be further studied. Integrating the structural comparisons with sequence analyses yielded an improved, structure-based alignment, and even revealed subtle sequence features such as the lack of equivalence of sAC Arg176 and the catalytic Arg [14,15]. The alignment suggests that sAC might directly belong to neither the mammalian class III cyclases nor to the bacterial subclass, but might rather be the evolutionarily oldest mammalian ‘ancestor’ cylase linking both families. Such a phylogenetic position might also explain the particular closeness of sAC to GCs, mammalian sGC and bacterial Cya2, suggested by its substantial GC side activity and by both structure search algorithms. Cya2, positioned within the bacterial sAC-like family, might represent a bacterial class III branch that evolved to GCs, but its GTP specificity remains to be confirmed in its physiological environment [16,34]. However, the closeness to the confirmed eukaryotic GCs might support a close resemblance of sAC to an ‘ancestor’ cyclase, possibly of low selectivity, that evolved into ACs and GCs. The possibly closer relationship of sAC to eukaryotic GCs than to tmACs is not apparent from our improved sequence alignment. However, in this and many previous studies, the C1 and C2 domains were analyzed together and with the homodimeric catalytic domains, which have to provide all essential features of both C1 and C2 (i.e. function as C1/C2 domains). This approach ignores the fact that subtle changes in structure and sequence, despite the conserved general architecture, influence regulation and specificity. Further crystal structures of class III cyclase domains will be required for a separate, structure-assisted C1 and C2 analysis to further refine phylogenetic trees. sAC is a good model system for studying class III substrate selection. It has all of the canonical class III AC residues and significant substrate specificity, but also recognizes GTP sufficiently, possibly reflecting an evolutionary closeness to a potential GC and AC ancestor cyclase. Interestingly, our sAC complex with ATP shows a flexible base that appears to significantly popuFEBS Journal 281 (2014) 4151–4164 ª 2014 FEBS

€lting et al. S. Kleinbo

late the syn state above the ribose, an unusual conformation for a class III nucleotide complex. ATP and ApCpp differ only subtly in bond angles [14], and the different bridging groups between the a-phosphate and b-phosphate are not involved in initial binding, suggesting that these difference are not responsible for the different base binding modes. The ATP complex could only be obtained in the absence of divalent cations, however, and the Na+ substituting for ion B might influence the nucleotide pose. However, the ribose and b/c-phosphate positions are similar to those in other sAC complexes with nucleotides or pyrophosphate, and the sAC–ATP complex thus might indicate a state that occurs physiologically, fitting into our model of a coupling between substrate binding and turnover [12,14] (Fig. 3D). We speculate that the substrate adenine can fluctuate between syn and anti, as observed in the sAC– ATP complex, during initial binding. In the subsequent step, and possibly supported by rearrangements resulting from binding of the catalytic ion A, the base freezes in the ApCpp orientation. The base interaction with Lys334 as one of the specificity-mediating residues is eventually formed through the stretching of ATP in the transition state, and exists only transiently (Fig. 3D). Consistently, the active site contains no cAMP, but PPi, after turnover (see above and [12]), indicating that the b/c-phosphate interactions generally dominate the binding of active site ligands. The direct and stable Lys–base contact in some of the previous class III AC complexes with ATP analogs, which are resistant to the nucleotide distortions required for transition state formation and, according to our model, for formation of the Lys– base interaction, might be artefactual. The nucleotide position might be shifted as compared with a Michaelis complex with regular substrate, enabling the Lys–base contact without distortion. Significant coupling between binding and distortion of the substrate ATP is supported by the observation that the sAC activator bicarbonate induces conformational changes that are also observed upon nucleotide binding but affects mainly kcat, rather than Km [8,12,16]. A weak and transient hydrogen bond with the base might explain the low selectivity of sAC and some other class III cyclases. Alternatively, the syn conformation observed in the sAC–ATP complex could explain the substantial sAC activity against GTP, as the base’s hydrogen-bonding groups fit to the sAC residues in this orientation. We speculate that only later during evolution did further features such as the Asp replacing sAC Thr405 evolve to better exclude productive GTP accommodation in AC active sites. Interestingly, no GC complex with a guanine nucleotide has yet been reported, but it has been speculated that these enzymes FEBS Journal 281 (2014) 4151–4164 ª 2014 FEBS

Crystal structures of sAC nucleotide complexes

acquired other active site features such as shape complementarity with the nucleotide base to distinguish their substrate GTP from ATP and other nucleotides [16]. In our sAC–ATP complex, the ribose is defined but shows weak electron density and few interactions with the enzyme, suggesting that it might move during separation of the a-phosphate and b-phosphate. Comparison with an sAC complex with PPi and cAMP [12] shows that the ribose ultimately occupies a similar position as in ATP or ApCpp complexes, the a-phosphate of cAMP is shifted, and the b/c-phosphates are in their initial position. These observations suggest that, during catalysis, the b/c-phosphates act as ‘anchors’, while the a-phosphate swings towards the 30 OH. Resolution of the bond to the b-phosphate might be supported by a transient shift of the AMP moiety that also results in the substrate specificitymediating interaction with Lys334 (Fig. 3D). Once the bond to the b-phosphate is broken and the a-phosphate has completed ring formation, the cAMP can shift ribose and adenine back to the original position, releasing the interaction with Lys334.

Experimental procedures Chemicals All chemicals were obtained from Sigma (Saint Louis, USA) unless stated otherwise.

Protein production and purification N-terminally His-tagged human sAC-cat (residues 1–469) was expressed in Hi5 insect cells with the BIIC method [35], and purified with nickel affinity chromatography followed by anion exchange and size exclusion chromatography, as described previously [30].

Crystal structure determination of sAC-cat complexes Human sAC-cat was crystallized in the apo form with the hanging drop method at 4 °C, as described previously [12]. After 5 days, crystals were transferred to drops of proteinfree cryoprotection solution containing 150–200 mM of the respective ligand for 24 h (ApCpp) or 4 h (all other ligands) of soaking at 4 °C. The cryoprotection solution contained 100 mM sodium acetate (pH 4.8), 200 mM trisodium citrate, 15% (w/v) poly(ethylene glycol) 4000, and 10% (v/v) glycerol. Crystals were then flash-frozen in liquid nitrogen. Complete datasets for the crystals with space group P63 were collected at 100 K at the Berlin Electron Storage Ring Society for Synchrotron Radiation beamline 14.1 operated

4161

€lting et al. S. Kleinbo

Crystal structures of sAC nucleotide complexes

by Helmholtz-Zentrum Berlin [36]. All diffraction data were processed with XDS [37]. The resolution limit of each dataset was defined according to the CC1/2 value [38]. Molecular replacement was performed with MOLREP [39]. Ligand occupancies were fixed at 1. Manual model building was performed in COOT [40] and refinement in REFMAC [41]. Model quality was analyzed in COOT, and structure figures were generated in PYMOL (www.pymol.org).

Binding measurements GTP binding affinity was determined by measuring microscale thermophoresis on a NanoTemper Monolith NT.label-free instrument (NanoTemper Technologies, Munich, Germany) with a 25% UV-LED setting and 30% IR-laser power. The assay buffer contained 50 mM Tris/HCl (pH 8), 50 mM NaCl, and 15 mM CaCl2. The Kd value was determined twice in independent experiments. Binding transitions were fitted with a single-site equation in GRAFIT (Erithacus Software, East Grinstead, UK).

For the structure-based alignment, the following structures from the PDB were used: ID 4CLK (human sAC-C1 and sAC-C2), ID 1AZS (human tmAC-C1 and tmAC-C2), ID 3R5G (CyaB from P. aeruginosa), ID 1WC0 (CyaC from S. platensis), ID 2W01 (Cya2 from Synechocystis), ID 1YK9 (Rv1625c mutant from Mycobacterium tuberculosis), ID 1YBT (Rv1900c from M. tuberculosis), ID 1Y11 (Rv1264 from M. tuberculosis), ID 1FX2 (receptor-type AC GRESAG4.1 from T. brucei), ID 3MR7 (a/b-fold family AC from Silicibacter pomeroyi), ID 3UVJ (human sGC-Cb), ID 2WZ1 (human sGC-Ca), and ID 3ET6 (sGCb from Chlamydomonas reinhardtii).

Acknowledgements We thank the Protein Sample Production Facility at HZI Braunschweig for help with protein expression, the Berlin Electron Storage Ring Society for Synchrotron Radiation MX staff for technical support, and Deutsche Forschungsgemeinschaft for financial support (grant STE1701/11 to C. Steegborn).

Activity measurements Activity assays were performed with 100 ng of purified, His-tagged sAC-cat in 50 mM Tris/HCl (pH 8.0), 50 mM NaCl and 5 mM ATP at 37 °C. As divalent cations, either 10 mM MgCl2 and 10 mM CaCl2, or either 20 mM CaCl2 or 20 mM MgCl2, were used. The reactions were stopped by flash-freezing, and analyzed by RP-HPLC on a C18 column (3.5 lm, 4.6 9 50-mm XBridge columns; Waters, Milford, USA) in 100 mM ammonium acetate (pH 8.8) and 10% (v/ v) acetonitrile, with cAMP eluting after 6.6 min at a flow rate of 0.5 mLmin1. Signal areas were integrated in LABSOLUTIONS software (Shimadzu, Kyoto, Japan). Measurements were performed in duplicate, and the data shown are representative of at least two repetitions.

Structural similarity searches Searches for similar structures were performed with aposAC (PDB ID 4CLF [12]) and the sAC–ApCpp complex (4CLK [12]) with DALILIGHT 3 (http://ekhidna.biocenter.helsinki.fi/dali_server/) [21] and 3D-BLAST (http:// 3d-blast.life.nctu.edu.tw/) [24]. For 3D-BLAST searches, the PDB was used as the search set, with the maximum number of hits set to 100 and the E-value threshold to 1010.

Sequence alignments and phylogenetic tree calculation For structure-based sequence alignments and phylogenetic tree calculations, the program STRAP [42] (superimposed native combinatorial extension method [43]) was used.

4162

Author contributions S. Kleinboelting performed experiments; J. van den Heuvel expressed protein; S. Kleinboelting and C. Steegborn planned experiments, analyzed data, and wrote the paper.

References 1 Hall RA, De Sordi L, Maccallum DM, Topal H, Eaton R, Bloor JW, Robinson GK, Levin LR, Buck J, Wang Y et al. (2010) CO(2) acts as a signalling molecule in populations of the fungal pathogen Candida albicans. PLoS Pathog 6, e1001193. 2 Hanoune J & Defer N (2001) Regulation and role of adenylyl cyclase isoforms. Annu Rev Pharmacol Toxicol 41, 145–174. 3 Tresguerres M, Parks SK, Salazar E, Levin LR, Goss GG & Buck J (2010) Bicarbonate-sensing soluble adenylyl cyclase is an essential sensor for acid/base homeostasis. Proc Natl Acad Sci USA 107, 442–447. 4 Danchin A (1993) Phylogeny of adenylyl cyclases. Adv Second Messenger Phosphoprot Res 27, 109–162. 5 Kamenetsky M, Middelhaufe S, Bank EM, Levin LR, Buck J & Steegborn C (2006) Molecular details of cAMP generation in mammalian cells: a tale of two systems. J Mol Biol 362, 623–639. 6 Sunahara RK & Taussig R (2002) Isoforms of mammalian adenylyl cyclase: multiplicities of signaling. Mol Interv 2, 168–184. 7 Zippin JH, Farrell J, Huron D, Kamenetsky M, Hess KC, Fischman DA, Levin LR & Buck J (2004)

FEBS Journal 281 (2014) 4151–4164 ª 2014 FEBS

€lting et al. S. Kleinbo

8

9

10

11

12

13

14

15

16

17

18

19

20

Bicarbonate-responsive ‘soluble’ adenylyl cyclase defines a nuclear cAMP microdomain. J Cell Biol 164, 527–534. Litvin TN, Kamenetsky M, Zarifyan A, Buck J & Levin LR (2003) Kinetic properties of ‘soluble’ adenylyl cyclase. Synergism between calcium and bicarbonate. J Biol Chem 278, 15922–15926. Zippin JH, Chen Y, Straub SG, Hess KC, Diaz A, Lee D, Tso P, Holz GG, Sharp GW, Levin LR et al. (2013) CO2/HCO3- and calcium regulated soluble adenylyl cyclase as a physiological ATP sensor. J Biol Chem 288, 33283–33291. Chen Y, Cann MJ, Litvin TN, Iourgenko V, Sinclair ML, Levin LR & Buck J (2000) Soluble adenylyl cyclase as an evolutionarily conserved bicarbonate sensor. Science 289, 625–628. Sinha SC & Sprang SR (2006) Structures, mechanism, regulation and evolution of class III nucleotidyl cyclases. Rev Physiol Biochem Pharmacol 157, 105–140. Kleinboelting S, Diaz A, Moniot S, van den Heuvel J, Weyand M, Levin LR, Buck J & Steegborn C (2014) Crystal structures of human soluble adenylyl cyclase reveal mechanisms of catalysis and of its activation through bicarbonate. Proc Natl Acad Sci USA 111, 3727–3732. Tesmer JJ & Sprang SR (1998) The structure, catalytic mechanism and regulation of adenylyl cyclase. Curr Opin Struct Biol 8, 713–719. Steegborn C, Litvin TN, Levin LR, Buck J & Wu H (2005) Bicarbonate activation of adenylyl cyclase via promotion of catalytic active site closure and metal recruitment. Nat Struct Mol Biol 12, 32–37. Tesmer JJ, Sunahara RK, Gilman AG & Sprang SR (1997) Crystal structure of the catalytic domains of adenylyl cyclase in a complex with Gsalpha.GTPgammaS. Science 278, 1907–1916. Rauch A, Leipelt M, Russwurm M & Steegborn C (2008) Crystal structure of the guanylyl cyclase Cya2. Proc Natl Acad Sci USA 105, 15720–15725. Mou TC, Masada N, Cooper DM & Sprang SR (2009) Structural basis for inhibition of mammalian adenylyl cyclase by calcium. Biochemistry 48, 3387–3397. Hess KC, Jones BH, Marquez B, Chen Y, Ord TS, Kamenetsky M, Miyamoto C, Zippin JH, Kopf GS, Suarez SS et al. (2005) The soluble adenylyl cyclase in sperm mediates multiple signaling events required for fertilization. Dev Cell 9, 249–259. Steegborn C, Litvin TN, Hess KC, Capper AB, Taussig R, Buck J, Levin LR & Wu H (2005) A novel mechanism for adenylyl cyclase inhibition from the crystal structure of its complex with catechol estrogen. J Biol Chem 280, 31754–31759. Schlicker C, Rauch A, Hess KC, Kachholz B, Levin LR, Buck J & Steegborn C (2008) Structure-based

FEBS Journal 281 (2014) 4151–4164 ª 2014 FEBS

Crystal structures of sAC nucleotide complexes

21

22

23

24

25

26

27

28

29

30

31

32

33

34

development of novel adenylyl cyclase inhibitors. J Med Chem 51, 4456–4464. Holm L & Rosenstrom P (2010) Dali server: conservation mapping in 3D. Nucleic Acids Res 38, W545–W549. Z€ahringer F, Lacanna E, Jenal U, Schirmer T & Boehm A (2013) Structure and signaling mechanism of a zinc-sensory diguanylate cyclase. Structure 21, 1149–1157. Zhou B-L, Pata JD & Steitz TA (2001) Crystal structure of a DinB lesion bypass DNA polymerase catalytic fragment reveals a classic polymerase catalytic domain. Mol Cell 8, 427–437. Tung C-H, Huang J-W & Yang J-M (2007) Kappaalpha plot derived structural alphabet and BLOSUM-like substitution matrix for rapid search of protein structure database. Genome Biol 8, R31.1-16. Yang J-M (2006) Protein structure database search and evolutionary classification. Nucleic Acids Res 34, 3646–3659. Sinha SC, Wetterer M, Sprang SR, Schultz JE & Linder JU (2005) Origin of asymmetry in adenylyl cyclases: structures of Mycobacterium tuberculosis Rv1900c. EMBO J 24, 663–673. Topal H, Fulcher NB, Bitterman J, Salazar E, Buck J, Levin LR, Cann MJ, Wolfgang MC & Steegborn C (2012) Crystal structure and regulation mechanisms of the CyaB adenylyl cyclase from the human pathogen Pseudomonas aeruginosa. J Mol Biol 416, 271–286. Cann MJ, Hammer A, Zhou J & Kanacher T (2003) A defined subset of adenylyl cyclases is regulated by bicarbonate ion. J Biol Chem 278, 35033–35038. Kobayashi M, Buck J & Levin LR (2004) Conservation of functional domain structure in bicarbonate-regulated ‘soluble’ adenylyl cyclases in bacteria and eukaryotes. Dev Genes Evol 214, 503–509. Kleinboelting S, Van den Heuvel J, Kambach C, Weyand M, Leipelt M & Steegborn C (2014) Expression, purification, crystallization, and preliminary X-ray diffraction analysis of a mammalian type 10 adenylyl cyclase. Acta Crystallogr F 70, 467–469. Litvin TN (2004) A novel mechanism of adenylyl cyclase activation conserved from cyanobacteria to man. PhD thesis, Cornell Medical College, NY, USA. Zheng H, Chruszcz M, Lasota P, Lebioda L & Minor W (2008) Data mining of metal ion environments present in protein structures. J Inorg Biochem 102, 1765–1776. Linder JU & Schultz JE (2003) The class III adenylyl cyclases: multi-purpose signalling modules. Cell Signal 15, 1081–1089. Ochoa de Alda JAG, Ajlani G & Houmard J (2000) Synechocystis strain PCC 6803 cya2, a prokaryotic

4163

€lting et al. S. Kleinbo

Crystal structures of sAC nucleotide complexes

35

36

37 38

39

gene that encodes a guanylyl cyclase. J Bacteriol 182, 3839–3842. Wasilko DJ, Lee SE, Stutzman-Engwall KJ, Reitz BA, Emmons TL, Mathis KJ, Bienkowski MJ, Tomasselli AG & Fischer HD (2009) The titerless infected-cells preservation and scale-up (TIPS) method for large-scale production of NO-sensitive human soluble guanylate cyclase (sGC) from insect cells infected with recombinant baculovirus. Protein Expr Purif 65, 122–132. Mueller U, Darowski N, Fuchs MR, Forster R, Hellmig M, Paithankar KS, Puhringer S, Steffien M, Zocher G & Weiss MS (2012) Facilities for macromolecular crystallography at the Helmholtz-Zentrum Berlin. J Synchrotron Radiat 19, 442–449. Kabsch W (2010) Xds. Acta Crystallogr D 66, 125–132. Diederichs K & Karplus PA (2013) Better models by discarding data? Acta Crystallogr D 69, 1215–1222. Vagin AA & Isupov MN (2001) Spherically averaged phased translation function and its application to the search for molecules and fragments in electron-density maps. Acta Crystallogr D 57, 1451–1456.

4164

40 Emsley P, Lohkamp B, Scott WG & Cowtan K (2010) Features and development of Coot. Acta Crystallogr D 66, 486–501. 41 Murshudov GN, Vagin AA & Dodson EJ (1997) Refinement of macromolecular structures by the maximum-likelihood method. Acta Crystallogr D 53, 240–255. 42 Gille C, Birgit W & Gille A (2014) Sequence alignment visualization in HTML5 without Java. Bioinformatics 30, 121–122. 43 Shindyalov IN & Bourne PE (1998) Protein structure alignment by incremental combinatorial extension (CE) of the optimal path. Protein Eng 11, 739–747.

Supporting information Additional supporting information may be found in the online version of this article at the publisher’s web site: Table S1. Structural comparison of sAC structures against the PDB by the use of DALI and 3D-BLAST (top 100 hits). Fig. S1. Alternative refinements for an sAC–ATP complex.

FEBS Journal 281 (2014) 4151–4164 ª 2014 FEBS

Structural analysis of human soluble adenylyl cyclase and crystal structures of its nucleotide complexes-implications for cyclase catalysis and evolution.

The ubiquitous second messenger cAMP regulates a wide array of functions, from bacterial transcription to mammalian memory. It is synthesized by six e...
1MB Sizes 0 Downloads 3 Views