- Email: [email protected]
Structural and Functional Analysis of the cAMP Binding Domain from the Regulatory Subunit of Mucor rouxii Protein Kinase A
Archives of Biochemistry and Biophysics Vol. 382, No. 2, October 15, pp. 173–181, 2000 doi:10.1006/abbi.2000.2018, available online at http://www.idealibrary.com on
Structural and Functional Analysis of the cAMP Binding Domain from the Regulatory Subunit of Mucor rouxii Protein Kinase A 1,2 Marı´a Rosa Sorol,* Ricardo L. Pastori,† Andre´s Muro,‡ Silvia Moreno,* and Silvia Rossi* ,3 *Departamento de Quı´mica Biolo´gica, Facultad de Ciencias Exactas y Naturales, Ciudad Universitaria, 1428 Buenos Aires, Argentina; †Diabetes Research Institute, University of Miami, Miami, Florida 33021; and ‡International Center for Genetic Engineering and Biotechnology, Trieste, Italy I-34012
Received April 24, 2000, and in revised form July 14, 2000
The cAMP binding domain of the regulatory subunit (R) of Mucor rouxii protein kinase A was cloned. The deduced amino acid sequence was highly homologous in sequence and in size to the corresponding region in fungal and higher eukaryotic regulatory subunits (47– 54%), but particularly homologous (62%) to Blastocladiella emersonii, a fungus classified in a different phylum. Amino acids reported to be important for interaction with cAMP, for cooperativity between the two cAMP binding domains, in the general folding of the domain, and for interaction with the catalytic subunit were conserved in all the fungal sequences. Based on either sequence or functional behavior, the M. rouxii R subunit cannot be classified as being more similar to RI or RII of mammalian systems. The M. rouxii protein sequence was modeled using as template the coordinates of the crystallized bovine regulatory subunit type I ␣. The quality of the model is good. The two backbones could be perfectly overlapped, except for two loop regions of high divergence. The ␣ helix C of domain A, proposed to have a strong interaction with the catalytic subunit, contains a leucine replacing a basic residue (arginine or lysine) commonly found in RI or RII. The domains A and B of the M. rouxii regulatory subunit were overexpressed as fusion proteins with GST. GST domain B protein was inactive. GST domain A was active; the kinetic parameters of affinity 1 This work was supported by grants from the University of Buenos Aires, Agencia Nacional de Promocio´n Cientı´fica y Te´cnolo´gica and Consejo Nacional de Investigaciones Cientı´ficas y Te´cnicas. M.R.S. had a doctoral fellowship and S.R. a postdoctoral short fellowship from FOMEC. 2 Sequence data are deposited with the DDBJ/EMBL/GenBank database under Accession No. AF240461. 3 To whom correspondence should be addressed. Fax: 54 114 576 3342. E-mail: [email protected].
0003-9861/00 $35.00 Copyright © 2000 by Academic Press All rights of reproduction in any form reserved.
toward cAMP analogs, site selectivity, and dissociation kinetics of bound cAMP were analogous to the properties of the domain in the whole regulatory subunit. © 2000 Academic Press Key Words: Mucor rouxii; protein sequence; cAMP binding.
cAMP-dependent protein kinases (PKA) 4 are tetrameric holoenzymes composed of a dimer of regulatory (R) subunits interacting and inhibiting with high affinity two catalytic (C) subunits. Cooperative binding of cAMP to the R subunit decreases the affinity between both subunits by several orders, promoting the activation of the catalytic moiety. The regulatory subunits are modular proteins composed of several distinct, well-defined domains. At the amino terminus there is a dimerization/docking domain; the hinge region that follows contains the consensus site sequence that binds to the substrate site in C in the absence of cAMP, and at the carboxy terminus there are two cAMP binding domains A and B. The three-dimensional structure of the cAMP binding domains, bound to cAMP, has been solved (1); each domain is composed of an A helix, an eight-stranded -barrel, and a B and a C helix. The hinge region, containing a small consensus site peptide, is not sufficient to justify the high affinity binding of R to C. Additional interaction sites are required. Genetic and biochemical evidence suggests that the A domain contains these additional sites required for high affinity binding to the C subunit, while 4 Abbreviations used: PKA, protein kinase A; R, regulatory subunit; C, catalytic subunit.
173
174
SOROL ET AL.
the B domain regulates access of cAMP to domain A (2). The precise molecular interactions that justify the high affinity between both subunits and explain the activation mechanism of protein kinases A await the crystallization of the holoenzyme as well as the regulatory subunit in its cAMP-free form. The protein kinase A from the fungus Mucor rouxii is present as a unique isoform and has been characterized and proved to be a tetrameric enzyme like those of higher eukaryotes (3, 4), with two cAMP binding sites per R subunit monomer exhibiting site-analog selectivity (5). The M. rouxii R subunit seems particularly large, since the sedimentation coefficient of its dimer is 6.2 S (the holoenzyme has a sedimentation coefficient of 8.8 S) and the MW estimated through SDS–PAGE analysis of the photoaffinity-labeled subunit (4) and from Western blots (6) is in the range of 70 –75 kDa. The fungal holoenzyme has the unusual property of not being dissociated by cAMP alone (3); however, the dissociation is complete by the addition of NaCl, histone or protamine together with cAMP (3, 7, 8). These properties make M. rouxii PKA an interesting model to study the mechanism of activation and dissociation by cAMP. All the residues in the R subunit proposed until now to be responsible for the high-affinity interaction with C, besides those of the highly conserved hinge region, reside in the cAMP binding domain. As a first step in the study of the structural basis for the high-affinity interaction between M. rouxii R and C subunits, we undertook the cloning and expression of these domains. MATERIALS AND METHODS Materials, bacterial strains, and plasmids. Restriction endonucleases were from Promega and GIBCO BRL, Taq DNA polymerase and the RACE-PCR kit were from GIBCO BRL, [ 3H]cAMP was purchased from New England Nuclear, cyclic nucleotide analogs were obtained from Sigma Chemical Co., and Rp- and Sp-cAMP analogs were a generous gift of H. Genieser (BIOLOG Life Science Institute, Bremen). Oligonucleotides were synthesized by Bio-Synthesis Inc., and GST–Sepharose affinity resin and the Sequenase version 2.0 DNA sequencing kit were from Amersham Pharmacia Biotech. Bacterial Escherichia coli strain DH5␣ was used for preparation of plasmids and cloning. The plasmid pGEX-4T-3 (Amersham Pharmacia Biotech) was used for the construction of the expression system. Fungal material. Aerobic mycelium of M. rouxii (NRRL 1894) was grown in complex rich medium YPG (yeast extract, peptone, glucose) until the early exponential phase (6 –7 h of growth). Genomic DNA and total RNA were prepared from the fungal material according to the protocols described in Refs. 9 and 10, respectively. Cloning and sequencing. Two degenerate primers (SM1 and SM2) were designed based on two conserved regions among protein kinase A regulatory subunits from a wide range of organisms: one (SM1) immediately before the beginning of cAMP binding domain A and the other (SM2) inside domain B. The primers were synthesized taking into account the Saccharomyces cerevisiae R subunit amino
acid sequence. Primer sequences: SM1, 5⬘-GGATCCCGAAYAAYTTYYTRTT-3⬘ (sense); SM2, 5⬘-GAATTCGYTCACCRAART-3⬘ (antisense). The template used for the polymerase chain reaction (PCR) in the first stage was M. rouxii genomic DNA. The PCR conditions for this stage began with four cycles of 1 min at 95°C, 1 min at 37°C, and 1 min at 72°C, which were then followed by more rigorous conditions in the annealing step. The product of the reaction, visualized as a clean band of the expected size of approximately 600 bp, was cloned and sequenced. The sequence corresponded to the complete site A and half of the site B cAMP domains as expected. Two new specific primers, SMA and SMB, were designed over the 3⬘-terminal sequence in order to make a nested RACE-PCR to obtain the Cterminal sequence of the R subunit. Primer sequences: SMA, 5⬘CAAGGTGACATAGGCGACCAG-3⬘ (sense); SMB, 5⬘-CCGACTCTAACGGAGATCAAC-3⬘ (sense). The conditions for this RACE-PCR were those indicated in the commercial kit. With a 17-mer oligo-dT as a primer, single-stranded cDNA was synthesized by reverse transcription from total RNA isolated from a M. rouxii culture grown 6 –7 h and subsequently amplified by PCR using oligo-dT and SMA and SMB primers. A 500 bp fragment could be identified corresponding to the C-terminus of the protein. Four new specific primers were designed based on the cloned sequences: SM4 and SM6, limiting the site A, and SM20 and SM21, limiting domain B. Primers sequences: SM4, 5⬘-GCGAATTCCATGGACGAGGAACACT-3⬘ (sense); SM6, 5⬘-CCGCTCGAGATAAGGTTCTAATGA-3⬘ (antisense); SM20, 5⬘-CCGGCATCCGCCTTGTTAATATCA-3⬘ (sense); SM21, 5⬘-CCGCCCGGGAATAGCTTGATAGTT-3⬘ (antisense). With genomic DNA of M. rouxii as a template, the primer pairs SM4 –SM6 and SM20 –SM21 were used to amplify by PCR the sequences codifying cAMP binding domains A and B, respectively. The fragments obtained by PCR were cloned into pGEX-4T-3 (all these primers have restriction sites in their 5⬘ termini to allow the cloning) for the expression of the domains fused to GST. The nucleotide sequence of all the cloned DNA fragments, derived from at least two independent polymerase chain reactions, was determined on both strands by the Sanger dideoxy method, using the Sequenase 2.0 kit. Expression and purification of the isolated cAMP binding domains. pGEX-A and pGEX-B are plasmids derived from pGEX-4T-3 carrying the DNA fragments corresponding to site A and to site B. They were expressed in E. coli DH5␣. Bacterial cultures (40 ml) from the logarithmic stage of growth were induced with isopropyl ␣-D-thiogalactopyranoside (0.1 mM) for 4 h at 37°C with shaking. Induced cells were pelleted at 6000g max for 15 min at 4°C, resuspended in ice-cold PBS (4 ml), repelleted, and stored at ⫺20°C until homogenization or used immediately. Except where indicated, all subsequent steps were performed at 4°C. Bacterial pellets were disrupted with glass beads in PBS buffer (1 ml). The extracts were clarified by centrifugation at 12,000g max for 10 min. These crude extracts were used directly for cAMP binding assays or loaded onto glutathione–Sepharose 4B resin for protein purification. A suspension of 40 l of resin per milliliter of crude extract was incubated overnight at 4°C with agitation. The suspension was centrifuged, the supernatant removed, and the glutathione–Sepharose pellet washed three times with 10 bed volumes of PBS. The protein was eluted with three washes of two bed volumes each of glutathione elution buffer (10 mM reduced glutathione in 50 mM Tris–HCl, pH 8) and used immediately. cAMP binding assays. cAMP binding was measured by incubation of the preparations with 1 M [ 3H]cAMP, plus the additions indicated in each case, in 20 mM Tris–HCl buffer, pH 7.4, 0.5 M NaCl at 4°C for 16 h, followed by filtering through nitrocellulose membranes (3). Sequence alignment and translation. The multiple sequence alignment and the translation of M. rouxii sequence were performed with DNAMAN software, version 4.1.
Mucor rouxii PROTEIN KINASE A cAMP BINDING DOMAIN
175
these domains in R subunits for other lower eukaryotic organisms. Homology with Other Regulatory Subunit cAMP Binding Domains
FIG. 1. cDNA and deduced amino acid sequence of M. rouxii protein kinase A regulatory subunit cAMP binding domains. cAMP binding domains A and B are underlined with thin and thick lines, respectively. DNA and protein sequence numbers are relative to the first position, arbitrarily chosen as 1. The DNA and deduced amino acid sequences are deposited with DDBJ/EMBL/GenBank under Accession No. AF240461.
Prediction of tertiary structure. For 3D structure prediction with RI␣ bovine as template (PDB entry 1RGS), the sequence of the cloned M. rouxii cAMP binding domains was submitted to SwissModel, “An Automated Comparative Protein Modelling Server” (http://expasy.hcuge.ch/swissmod/SWISS-MODEL.html) (11), which runs on the software ProMod II (12). The 3D structure was visualized with the software Swiss-Pdb Viewer 2.1 (13).
RESULTS AND DISCUSSION
Cloning and Sequence of cAMP Binding Domains A and B The strategy used for cloning M. rouxii regulatory subunit cAMP binding domains can be summarized as follows: in a first step, a specific product was obtained by PCR amplification from genomic DNA using degenerate primers derived from conserved sequences in all the R subunits. The 3⬘-terminus was obtained by RACE-PCR from a cDNA preparation using oligo-dT as a primer for reverse transcription. The complete cAMP binding domain was finally amplified from genomic DNA. The nucleotide and deduced amino acid sequences are presented in Fig. 1. From the comparison of the genomic DNA and cDNA sequences, it could be immediately observed that this region of the sequence is intronless, as occurs with the sequence codifying
The sequence of M. rouxii cAMP binding domains was compared and aligned with the homologous sequences from seven fungi and with the cAMP binding domains of bovine RII ␣ and RI ␣ (Fig. 2). The first conclusion derived from this alignment is that the general conservation of these domains throughout the whole phylogenetic scale of eukaryotic organisms is highly conserved. The degree of identity of the M. rouxii sequence with almost all of the fungal sequences ranges between 50 and 54%, while the identity with the proteins from a highly unrelated species (bovine) is only slightly lower (47%). It is interesting that the identity between the M. rouxii and B. emersonii sequences is much higher (62%). According to classical taxonomical parameters, these two fungi, M. rouxii and B. emersonii, belong to two different phyla: Zygomycota and Chytridiomycota. However, construction of a neighbor-joining tree of the fungi based on comparison of their 18-S rDNA sequence (TreeBASE M536) (14) has drawn into question whether these two phyla are monophyletic. In this tree, the homology between the 18-S rDNA sequences from M. racemosus and B. emersonii is higher than between the rDNA of other members of their own phylum. The results of the degree of identity of the cAMP binding domain of the regulatory subunits from these two fungi confirm that the comparison of protein sequences is a useful tool to complement rDNA data in the molecular phylogenetic studies used to reinvestigate the systematic position of fungi (15). A second conclusion derived from these comparisons is that the M. rouxii R subunit cannot be classified as a typical type I or type II regulatory subunit. The concentration of NaCl at which it elutes from a DEAE– cellulose column (3) and the capacity of being autophosphorylated (6) make it resemble type II regulatory subunits. However, comparing the sequences of each of the two cAMP binding domains A and B to the corresponding sequences of bovine RI ␣ and RII ␣, one can observe that M. rouxii cAMP binding domain A more closely resembles a type II domain A than a type I domain A (52.0 vs 46.3% of identity) and, on the contrary, M. rouxii domain B is more similar to type I domain B than type II domain B (45.6 vs 37.5%). It is interesting to note that lower eukaryotes have only one gene codifying for the R subunit and that, therefore, only one type of holoenzyme (regarding the R subunit) might be responsible for participating in processes described to be regulated differentially by PKAI and PKAII holoenzymes in higher eukaryotic organisms
176
SOROL ET AL.
FIG. 2. Multialignment of M. rouxii R subunit cAMP binding domains (1–248) with the corresponding sequences of the R subunits from Blastocladiella emersonii (160 – 403), Ustylago maydis (217–536), Emericella nidulans (154 – 402), Magnaporthe grisea (134 –390), Neurospora crassa (133–385), Saccharomyces cerevisiae (187– 416), Schizosaccharomyces pombe (147– 412), bovine RII ␣ (139 – 401), and bovine RI ␣ (139 –379). Only the sequence of bovine RI ␣ is numbered in the alignment. The amino acid sequences were aligned first using the DNAMAN software and further manually to obtain maximal identity, by inserting blanks. The arrow marks the limit between domains A and B. The asterisks correspond to the two loops where the backbone of M. rouxii R structure diverges from the backbone of bovine RI ␣, shown as two orange loops in Fig. 3. The box encloses the ␣ helix C of domain A, described in Fig. 4.
(16, 17). Interesting conclusions can be drawn from the comparison of the fungal cAMP binding domains with those of bovine RI ␣ and RII ␣, shown in Table I. Fungal
cAMP binding domains A are much more identical to bovine domains A than fungal domains B are to the corresponding bovine domain. Fungal cAMP binding
TABLE I
Homology of Fungal cAMP Binding Domains A and B with Bovine RI ␣ and RII ␣ Domains A and B a Domain A
M. rouxii B. emersonii U. maydis E. nidulans Ma. grisea N. crassa S. cerevisiae Sc. pombe a
Domain B
IA
IIA
IB
IIB
IA
IIA
IB
IIB
46.3 51.2 46.3 44.6 47.1 45.5 44.3 46.7
52.0 55.2 44.8 51.2 48.8 50.4 50.4 42.4
26.1 31.4 28.6 32.8 29.4 27.7 31.0 31.1
27.4 29.3 21.9 25.4 23.8 23.1 31.0 24.6
32.5 31.5 23.9 36.4 31.6 31.3 31.4 37.9
28.0 27.9 26.5 29.6 25.8 26.7 27.4 29.4
45.6 44.0 28.9 39.6 31.6 36.0 36.8 32.7
37.5 46.6 23.2 38.1 33.1 35.9 36.4 28.2
The results are expressed in percentage of identity obtained through the alignment program of the DNAMAN software, without manual correction. Italic characters: homology between nonhomologous sites. Bold characters: values representing a distinct greater homology.
Mucor rouxii PROTEIN KINASE A cAMP BINDING DOMAIN
177
Analysis of Residues Important for the High-Order Structure of the cAMP Binding Domains
FIG. 3. Backbone model of the three-dimensional structure of the cAMP binding domains of the M. rouxii regulatory subunit. The cloned sequence of cAMP binding domain was modeled through the Swiss-Model Modelling Server, using as template the coordinates of the bovine RI ␣ subunit (1RGS). In orange are the loops that are not superimposed between both backbones, corresponding to M. rouxii amino acids 48 –55 in domain A and 173–180 in domain B (the amino acids corresponding to the orange loops are marked with an asterisk in the alignment of Fig. 2).
domains B seem to diverge more from the mammalian counterparts, and, in fact in most cases, there is not a clear preference of homology with either bovine domain A or B, either of isoforms I or II. In fact, the degree of identity is, in most cases, not greater than the homology existing between any domain A and B of the same species. It seems as if domain A, important in the interaction with the catalytic subunit, would have been more conserved than domain B, important in the cooperativity between both cAMP binding sites.
Conserved residues involved in the interaction with cAMP, in the cooperativity between the cAMP binding domains, and in the general folding of the R subunit have been identified from data obtained from the crystal structure of cAMP-bound RI ␣ and from genetic studies (1). The residues described to be important in the interaction with the phosphate and ribose of cAMP through several hydrogen bonds and electrostatic contacts are conserved in M. rouxii as well as in all other fungal sequences, namely, Glu200 and Arg209 in domain A and Glu324 and Arg333 in domain B. Both sites are described to have an aromatic residue that, through its side chain, stacks with the adenine ring of cAMP: Trp260 in domain A and Tyr371 in domain B for bovine RI ␣. In M. rouxii and in the rest of the fungi, except U. maydis, Trp260 is replaced by another aromatic residue, Tyr, which can also have stacking interactions with the adenine ring. It is interesting to note that RII ␣ lacks an aromatic residue at this position. In the corresponding position of domain B, the alignment is less clear, due to the nonconserved sequences of the carboxy termini of all the R subunits; however, a Tyr can be aligned for all the sequences except S. cerevisiae. In addition to the electrostatic interactions, the tight binding of cAMP involves a number of strong hydrophobic interactions (1). The residues involved are Val182, Val184, Ala189, Ala210, and Ala211 in site A and Val300, Leu316, Val313, Ile325, and Ala335 in site B. Almost all these residues are identical in M. rouxii or replaced by nonpolar aliphatic amino acids, the only exception being the replacement of Val182 by a Cys.
FIG. 4. (Left) Alignment of the amino acid sequence of the fungal regulatory subunits and RII ␣ as compared to the C helix of domain A of bovine RI ␣ (amino acids 235–244). In bold are the residues aligned with Arg239 of RI ␣. This C helix is enclosed in a box in the multialignment of Fig. 2. (Right) Modeling of M. rouxii amino acids 103–112, corresponding to the C helix of cAMP binding domain A (amino acids 235–244 from bovine RI ␣). The side chains of Leu107, Arg109, Arg110, and Tyr112 are shown. These residues are numbered according to the M. rouxii sequence.
178
SOROL ET AL.
FIG. 5. Expression of domains A and B and purification of isolated domain A: 10% SDS–PAGE gels, stained with Coomassie blue. (A) Twenty micrograms of total proteins from E. coli crude extracts from an induced culture of the bacterial strain transformed with the empty pGEX vector (lane 2) and from uninduced and induced cultures (lanes 3 and 4) from the bacterial strain transformed with GST-A plasmid. Lane 1: MW markers. The dash indicates the 44-kDa overexpressed GST-A product. (B) Ten micrograms of total proteins from E. coli extract transformed with GST-B plasmid, without (lane 1) and with (lane 2) IPTG induction. The dash indicates the 38-kDa overexpressed GST-B product. (C) Purification of GST-A protein; 8 ml of an extract containing 16 mg of protein was purified as described under Materials and Methods. Ten-microliter aliquots of each fraction were loaded on the gel: lane 1, total extract; lane 2, flowthrough; lanes 3 and 4, second and third washes; lanes 5 and 6, first and second elutions.
The three invariant glycine residues in each domain, described to be essential for maintenance of the structural integrity of the -barrels and even conserved in E. coli CRP protein (18), are also conserved in the fungal regulatory subunits. The molecular basis for cooperativity between cAMP domains A and B was described from the crystallization model derived for RI ␣ (1) and it consists of an extended hydrophobic surface. Specifically, the hydrophobic surface in domain B formed by Val265, Leu269, Ile292, Leu294, Tyr321, Val346, Val356, Leu357, Cys369, Ile363, and Leu364 is covered by Tyr244, Phe247, Leu248, Val251, Ile253, Leu254, and Leu257 from domain A. All these residues are identical or have a conservative replacement in M. rouxii cAMP binding domains, except Leu294, which is aligned with a glutamic acid in the fungal proteins and in bovine RII ␣. Molecular Modeling of M. rouxii Structure The 3D structure of M. rouxii cAMP binding domains was obtained by molecular modeling through the Swiss-Model Protein Modelling Server using as template the coordinates of the crystallized bovine RI ␣ subunit (PDB entry 1RGS). The Z values of the root mean square of data from bond lengths, bond angles, and torsion angles were not very far from 1 and were within the range of values expected for a naturally folded protein. The Ramachandran score, expressing how well the backbone conformations of all residues are corresponding to allowed areas in the Ramachandran plot, is within the expected ranges for well-refined structures. The superposition of the backbones for M. rouxii cAMP binding domains and those of its template, shown in Fig. 3, indicates an almost complete
overlap, except for two loop regions (shown in orange) comprising M. rouxii amino acids 48 –55 in domain A and the corresponding region in domain B (amino acids 173–180). These two loops (indicated with asterisks in Fig. 2) correspond to the two regions where the major divergence in sequence arises: in the loop corresponding to domain A, the bovine protein has a smaller number of amino acids, while in the loop corresponding to domain B, there is nonconservative replacement of amino acids throughout the whole loop. Even though the understanding of the details of the interaction between the R and C subunits awaits the crystallization of an R 䡠 C complex, Zhao et al. (19) have modeled the quaternary structure of the holoenzyme, taking into account their own results of neutron contrast variation and results taken from the literature derived from genetic approaches. There are results characterizing the residues in C implicated in the R–C interaction (20 –23), as well as those likely not to be involved (20, 21), and results identifying residues in R that were proposed to be candidate binding sites for the C subunit (2). In this model, the interacting surface of the R subunit includes the following important residues from the cAMP binding domains implicated in R–C interaction, namely Asp140, Glu143, Asp170, Lys242, and Asp258 (corresponding to bovine RI ␣). All these residues are conserved in all the regulatory subunits except Glu143; however, the fungal enzymes and RII ␣ have an acidic residue one position more amino than Glu143 in the RI subunit. Recently, using a yeast two-hybrid screen with the C subunit as bait, Huang and Taylor identified a critical helix in the cAMP binding domain A (helix C) that modulates interactions of this domain with cAMP, with C, and with cAMP bind-
Mucor rouxii PROTEIN KINASE A cAMP BINDING DOMAIN
179
ing domain B (24). This helix corresponds to the region comprising amino acids 234 –244 in bovine RI ␣ and protrudes from the surface defined in R by Zhao et al. (19), to interact with C. The residue in this region important for cooperative interaction between the two cAMP sites, Arg241, is conserved (Arg109 for the Mucor sequence; Fig. 4). However, the opposite surface of this helix, facing toward the C binding site, is not so conserved. Particularly, an important residue in this surface, Arg239 in RI ␣, is replaced by a Leu107 in M. rouxii (Fig. 4), implying a different kind of interaction between this helix in the R subunit of both organisms and the complementary surface in the respective C subunit. This may define a critical difference in the interaction between R and C subunits in the different holoenzymes. However, the interaction through this domain does not seem to explain the low affinity known to exist between the R and C subunits in S. cerevisiae (25), since the R subunit from this yeast possesses a Lys residue in this position, just like bovine RII ␣. Expression and Characterization of M. rouxii Isolated cAMP Binding Domains M. rouxii cAMP binding domains A and B were expressed as fusion proteins with GST in order to use the proteins, if functional, in future studies of interaction with catalytic subunit. Figures 5A and 5B show that the two constructions were overexpressed when inducing the E. coli culture with IPTG. The apparent MW for GST-domain A (GST-A) corresponded to the expected MW (44 kDa), while the MW of GST-domain B (GST-B) was smaller (38 kDa). Extracts overexpressing GST-B protein had a very low cAMP binding activity, which did not correspond to the amount of overexpressed protein; besides, this activity, although severalfold higher than the background activity of extracts transformed with the vector without insert, was not saturated by cAMP up to 100 M, suggesting a poorly folded, unstable protein, with very low affinity for cAMP. However, the overexpression of a functional GST-domain A protein was relatively successful, since the cAMP binding activity of crude extracts was 50- to 100-fold higher than in control extracts (no induction). This activity could be saturated by 1.5 M [ 3H]cAMP. It was relatively unstable, since cAMP binding activity decreased upon incubation of the protein at 30°C; however, the amount and size of the stained band corresponding to the fusion protein did not change with this incubation, suggesting that the decrease in binding activity corresponded to an unfolding of the protein and not to proteolytic degradation (not shown). Therefore, cAMP binding assays were performed by overnight incubations at 4°C. GST-A protein was purified through glutathione–Sepharose columns (Fig. 5C) in order to study some of its kinetic properties. The stoi-
FIG. 6. cAMP-analog selectivity of isolated domain A. Two-microgram aliquots from the purified preparation of GST-A protein were incubated with 1 M [ 3H]cAMP in the presence of 1, 10, or 100 M of each cAMP analog or unlabeled cAMP overnight at 4°C. Bound [ 3H]cAMP was measured by the nitrocellulose filter assay. Binding of [ 3H]cAMP in the absence of added analogs was taken as 100% (7 pmol of cAMP bound). (Top panel) The dotted line indicates the competition curve of the partially purified regulatory subunit with 8-bromo-cAMP (from Ref. 5). N6 ben, N 6 -benzoyl-cAMP; N6 mono, N 6 -monobutyryl-cAMP; 8 Br, 8-bromo-cAMP; 8 ben, 8-thiobenzylcAMP. (Bottom panel) The dotted line indicates the behavior of the partially purified regulatory subunit with Sp-8-bromo-cAMP. Sp, Sp-cAMP; Rp, Rp-cAMP; Sp8, Sp-8-bromo-cAMP; Rp8, Rp-8-bromocAMP.
chiometry of cAMP binding of the purified protein indicated that not more than 10 –15% of the purified protein was active. Site-selective cAMP analogs were used as competitors of [ 3H]cAMP binding in order to analyze whether the isolated domain maintained the characteristics of a cAMP binding site A, which should be selective for N 6 - cAMP analogs. The top panel of Fig. 6 shows that the two site A selective cAMP analogs, N 6 -monobutyryl-cAMP and N 6 -benzoyl-cAMP, were very good competitors of [ 3H]cAMP binding, while the two site B selective analogs, 8-Br-cAMP and 8-thiobenzyl-cAMP, were very poor competitors. C8-cAMP analogs were always found to be poor competitors of [ 3H]cAMP binding of the whole subunit (5); but the behavior of the isolated A domain was clearly different from the behavior of the whole subunit (dotted line in top panel of Fig. 6). The bottom panel of Fig. 6 shows the behavior of the Rp- and Sp-cAMP analogs and of its C8 derivatives as competitors of [ 3H]cAMP binding.
180
SOROL ET AL.
FIG. 7. Dissociation of [ 3H]cAMP bound to isolated domain A. One-microgram aliquots from the purified preparation of GST-A protein were incubated with 1 M [ 3H]cAMP overnight at 4°C. Dissociation kinetics was measured after addition of 100 M nonradioactive cAMP. Samples were filtered at the indicated times. The dotted line represents the dissociation kinetics of cAMP bound to site A of the whole R subunit, partially purified from M. rouxii extracts; the data were recalculated from those of Ref. 3 to express them as dissociation from site A; during the 180 min of the experiment at 4°C, the dissociation of [ 3H]cAMP bound to site B was negligible. A value of 100% binding represents 3 pmol of bound cAMP.
Rp-cAMP analogs did not compete, as occurs with the whole subunit (C. Mizyrycki and S. Moreno, unpublished results); however, Sp-8Br-cAMP, which competed with the binding of [ 3H]cAMP to the whole subunit (dotted line), did not compete with the binding to the isolated A domain. The dissociation kinetics at 4°C of the [ 3H]cAMP bound to GST-A protein is shown in Fig. 7. The dissociation curve is monophasic, as expected for only one site, suggesting there is one population of molecules. The estimated half-life for the bound cAMP is around 45 min, in the same range of the estimate we had for the dissociation kinetics from this site in the whole subunit at 4°C (3), drawn as a dotted line in the same figure for comparison. The results from this section show that even though only a small percentage of the protein is active, the kinetic properties of the isolated recombinant cAMP domain A from the M. rouxii regulatory subunit are analogous to the properties of the domain in the whole subunit. This suggests that domain B does not contribute significantly to the affinity, dissociation kinetics, selectivity, and folding of domain A. This result contrasts with those reported for a deletion mutant of RI ␣ (26), in which upon removal of domain B from the whole protein, the dissociation kinetics from [ 3H]cAMP bound to domain A was greatly increased; in this case, the amino terminus of the protein is still conserved. The isolation of cAMP-binding fragments with binding activity is not new. However, the identity of those fragments as one or the other of the cAMP binding domains has not been determined. Small fragments, presumably derived from proteolysis of the native reg-
ulatory subunits, have been reported in bovine spermatozoa (27), rat liver (28), and yeast (29). Trypsin digestion products of purified mammalian type I and II R subunits have been reported to have binding activity (30, 31). The fact that it was not possible to obtain an active cAMP binding domain B from the recombinant protein from M. rouxii, while an active domain A could be obtained, suggested us, as a first analysis, that in the case of Mucor, domain A might be necessary for the adequate folding of domain B. However, a recombinant cAMP binding domain B from a mammalian system has been obtained in an active form and reported to function independently of any other domain structure (32). In this report, we establish that the physical limits for a functional cAMP domain A in M. rouxii can be defined to the limits of the functional domain in the whole subunit. REFERENCES 1. Su, Y., Dostmann, W. R. G., Herberg, F. W., Durick, K., Xuong, N., Ten Eyck, L. F., Taylor, S. S., and Varughese, K. I. (1995) Science 269, 807– 813. 2. Gibson, R. M., Ji-Buechler, Y., and Taylor, S. S. (1997) J. Biol. Chem. 272, 16343–16350. 3. Moreno, S., and Passeron, S. (1980) Arch. Biochem. Biophys. 199, 321–330. 4. Pastori, R., Moreno, S., and Passeron, S. (1985) Mol. Cell. Biochem. 69, 55– 66. 5. Paveto, C., Passeron, S., Corbin, J. D., and Moreno, S. (1989) Eur. J. Biochem. 179, 429 – 434. 6. Rossi, S., Guthmann, M., and Moreno, S. (1992) Cell. Signal. 4, 443– 451. 7. Pastori, R., Kerner, N., Moreno, S., and Passeron, S. (1981) Biochem. Biophys. Res. Commun. 101, 663– 671. 8. Guthmann, M., Pastori, R., and Moreno, S. (1990) Cell. Signal. 2, 395– 402. 9. Hoffman, C. S., and Winston, D. (1987) Gene 57, 267–272. 10. Ausubel, F. M., Brent, R., Kingston, R., Moore, D., Seidman, J. G., Smith, J. A., and Struhl, K. (1993) Short Protocols in Molecular Biology, 2nd ed., Greene Publishing Associates/Wiley, New York. 11. Peitsch, M. C. (1995) Biotechnology 13, 658 – 660. 12. Peitsch, M. C. (1996) Biochem. Soc. Trans. 24, 274 –279. 13. Guex, N., and Peitsch, M. C. (1977) Electrophoresis 18, 2714 – 2723. 14. Tanabe, Y., Nagahama, T., Saikawa, M., and Sugiyama, J. (1999) Mycologia 91, 830 – 835. 15. Thon, M. R., and Royse, D. J. (1999) Mycologia 91, 468 – 474. 16. Cho-Chung, Y. S., and Clair, T. (1993) Pharmacol. Ther. 60, 265–288. 17. Tasken, T., Skalhegg, B. S., Tasken, K. A., Solberg, R., Knutsen, H. K., Levy, F. O., Sandberg, M., Orstavik, S., Larse, T., Johansen, A. K., Vang, T., Schrader, H. P., Reinton, N. T., Torgersen, K. M., Hansson, V., and Jahnsen, T. (1997) Adv. Second Messenger Phosphoprotein Res. 31, 191–204. 18. Shabb, J. B., and Corbin, J. D. (1992) J. Biol. Chem. 267, 5723– 5726. 19. Zhao, J., Hoye, E., Boylan, S., Walsh, D. A., and Trewhella, J. (1998) J. Biol. Chem. 273, 30448 –30459.
Mucor rouxii PROTEIN KINASE A cAMP BINDING DOMAIN 20. Gibbs, C. S., and Zoller, M. J. (1991) J. Biol. Chem. 266, 8923–8931. 21. Gibbs, C. S., Knighton, D. R., Sowadski, J. M., Taylor, S. S., and Zoller, M. J. (1992) J. Biol. Chem. 267, 4806 – 4814. 22. Orellana, S. A., Amieux, P. S., Zhao, X., and McKnight, G. S. (1993) J. Biol. Chem. 268, 6843– 6846. 23. Gibson, R. M., Buechler, Y. J., and Taylor, S. S. (1997) Protein Sci. 6, 1825–1834. 24. Huang, L. J., and Taylor, S. S. (1998) J. Biol. Chem. 273, 26739 – 26746. 25. Kuret, J., Johnson, K. E., Nicolette, C., and Zoller, M. J. (1988) J. Biol. Chem. 263, 9149 –9154. 26. Woodford, T. A., Correll, L. A., McKnight, G. S., and Corbin, J. D. (1989) J. Biol. Chem. 264, 13321–13328.
181
27. Garbers, D. L., First, N. L., and Lardy, H. A. (1973) J. Biol. Chem. 248, 875– 879. 28. Srivastava, A. K., and Stellwagen, R. H. (1978) J. Biol. Chem. 253, 1752–1755. 29. Takai, Y., Yamamura, H., and Nishizuka, Y. (1974) J. Biol. Chem. 249, 530 –535. 30. Rannels, S. R., and Corbin, J. D. (1979) J. Biol. Chem. 254, 8605– 8610. 31. Rannels, S. R., and Corbin, J. D. (1980) J. Cycl. Nucleotide Res. 8, 201–215. 32. Shabb, J. B., Poteet, C. E., Kapphahn, M. A., Muhonen, W. M., Baker, N. E., and Corbin, J. D. (1995) Protein Sci. 4, 2100 – 2106.
Structural and Functional Analysis of the cAMP Binding Domain from the Regulatory Subunit of Mucor rouxii Protein Kinase A 1,2 Marı´a Rosa Sorol,* Ricardo L. Pastori,† Andre´s Muro,‡ Silvia Moreno,* and Silvia Rossi* ,3 *Departamento de Quı´mica Biolo´gica, Facultad de Ciencias Exactas y Naturales, Ciudad Universitaria, 1428 Buenos Aires, Argentina; †Diabetes Research Institute, University of Miami, Miami, Florida 33021; and ‡International Center for Genetic Engineering and Biotechnology, Trieste, Italy I-34012
Received April 24, 2000, and in revised form July 14, 2000
The cAMP binding domain of the regulatory subunit (R) of Mucor rouxii protein kinase A was cloned. The deduced amino acid sequence was highly homologous in sequence and in size to the corresponding region in fungal and higher eukaryotic regulatory subunits (47– 54%), but particularly homologous (62%) to Blastocladiella emersonii, a fungus classified in a different phylum. Amino acids reported to be important for interaction with cAMP, for cooperativity between the two cAMP binding domains, in the general folding of the domain, and for interaction with the catalytic subunit were conserved in all the fungal sequences. Based on either sequence or functional behavior, the M. rouxii R subunit cannot be classified as being more similar to RI or RII of mammalian systems. The M. rouxii protein sequence was modeled using as template the coordinates of the crystallized bovine regulatory subunit type I ␣. The quality of the model is good. The two backbones could be perfectly overlapped, except for two loop regions of high divergence. The ␣ helix C of domain A, proposed to have a strong interaction with the catalytic subunit, contains a leucine replacing a basic residue (arginine or lysine) commonly found in RI or RII. The domains A and B of the M. rouxii regulatory subunit were overexpressed as fusion proteins with GST. GST domain B protein was inactive. GST domain A was active; the kinetic parameters of affinity 1 This work was supported by grants from the University of Buenos Aires, Agencia Nacional de Promocio´n Cientı´fica y Te´cnolo´gica and Consejo Nacional de Investigaciones Cientı´ficas y Te´cnicas. M.R.S. had a doctoral fellowship and S.R. a postdoctoral short fellowship from FOMEC. 2 Sequence data are deposited with the DDBJ/EMBL/GenBank database under Accession No. AF240461. 3 To whom correspondence should be addressed. Fax: 54 114 576 3342. E-mail: [email protected].
0003-9861/00 $35.00 Copyright © 2000 by Academic Press All rights of reproduction in any form reserved.
toward cAMP analogs, site selectivity, and dissociation kinetics of bound cAMP were analogous to the properties of the domain in the whole regulatory subunit. © 2000 Academic Press Key Words: Mucor rouxii; protein sequence; cAMP binding.
cAMP-dependent protein kinases (PKA) 4 are tetrameric holoenzymes composed of a dimer of regulatory (R) subunits interacting and inhibiting with high affinity two catalytic (C) subunits. Cooperative binding of cAMP to the R subunit decreases the affinity between both subunits by several orders, promoting the activation of the catalytic moiety. The regulatory subunits are modular proteins composed of several distinct, well-defined domains. At the amino terminus there is a dimerization/docking domain; the hinge region that follows contains the consensus site sequence that binds to the substrate site in C in the absence of cAMP, and at the carboxy terminus there are two cAMP binding domains A and B. The three-dimensional structure of the cAMP binding domains, bound to cAMP, has been solved (1); each domain is composed of an A helix, an eight-stranded -barrel, and a B and a C helix. The hinge region, containing a small consensus site peptide, is not sufficient to justify the high affinity binding of R to C. Additional interaction sites are required. Genetic and biochemical evidence suggests that the A domain contains these additional sites required for high affinity binding to the C subunit, while 4 Abbreviations used: PKA, protein kinase A; R, regulatory subunit; C, catalytic subunit.
173
174
SOROL ET AL.
the B domain regulates access of cAMP to domain A (2). The precise molecular interactions that justify the high affinity between both subunits and explain the activation mechanism of protein kinases A await the crystallization of the holoenzyme as well as the regulatory subunit in its cAMP-free form. The protein kinase A from the fungus Mucor rouxii is present as a unique isoform and has been characterized and proved to be a tetrameric enzyme like those of higher eukaryotes (3, 4), with two cAMP binding sites per R subunit monomer exhibiting site-analog selectivity (5). The M. rouxii R subunit seems particularly large, since the sedimentation coefficient of its dimer is 6.2 S (the holoenzyme has a sedimentation coefficient of 8.8 S) and the MW estimated through SDS–PAGE analysis of the photoaffinity-labeled subunit (4) and from Western blots (6) is in the range of 70 –75 kDa. The fungal holoenzyme has the unusual property of not being dissociated by cAMP alone (3); however, the dissociation is complete by the addition of NaCl, histone or protamine together with cAMP (3, 7, 8). These properties make M. rouxii PKA an interesting model to study the mechanism of activation and dissociation by cAMP. All the residues in the R subunit proposed until now to be responsible for the high-affinity interaction with C, besides those of the highly conserved hinge region, reside in the cAMP binding domain. As a first step in the study of the structural basis for the high-affinity interaction between M. rouxii R and C subunits, we undertook the cloning and expression of these domains. MATERIALS AND METHODS Materials, bacterial strains, and plasmids. Restriction endonucleases were from Promega and GIBCO BRL, Taq DNA polymerase and the RACE-PCR kit were from GIBCO BRL, [ 3H]cAMP was purchased from New England Nuclear, cyclic nucleotide analogs were obtained from Sigma Chemical Co., and Rp- and Sp-cAMP analogs were a generous gift of H. Genieser (BIOLOG Life Science Institute, Bremen). Oligonucleotides were synthesized by Bio-Synthesis Inc., and GST–Sepharose affinity resin and the Sequenase version 2.0 DNA sequencing kit were from Amersham Pharmacia Biotech. Bacterial Escherichia coli strain DH5␣ was used for preparation of plasmids and cloning. The plasmid pGEX-4T-3 (Amersham Pharmacia Biotech) was used for the construction of the expression system. Fungal material. Aerobic mycelium of M. rouxii (NRRL 1894) was grown in complex rich medium YPG (yeast extract, peptone, glucose) until the early exponential phase (6 –7 h of growth). Genomic DNA and total RNA were prepared from the fungal material according to the protocols described in Refs. 9 and 10, respectively. Cloning and sequencing. Two degenerate primers (SM1 and SM2) were designed based on two conserved regions among protein kinase A regulatory subunits from a wide range of organisms: one (SM1) immediately before the beginning of cAMP binding domain A and the other (SM2) inside domain B. The primers were synthesized taking into account the Saccharomyces cerevisiae R subunit amino
acid sequence. Primer sequences: SM1, 5⬘-GGATCCCGAAYAAYTTYYTRTT-3⬘ (sense); SM2, 5⬘-GAATTCGYTCACCRAART-3⬘ (antisense). The template used for the polymerase chain reaction (PCR) in the first stage was M. rouxii genomic DNA. The PCR conditions for this stage began with four cycles of 1 min at 95°C, 1 min at 37°C, and 1 min at 72°C, which were then followed by more rigorous conditions in the annealing step. The product of the reaction, visualized as a clean band of the expected size of approximately 600 bp, was cloned and sequenced. The sequence corresponded to the complete site A and half of the site B cAMP domains as expected. Two new specific primers, SMA and SMB, were designed over the 3⬘-terminal sequence in order to make a nested RACE-PCR to obtain the Cterminal sequence of the R subunit. Primer sequences: SMA, 5⬘CAAGGTGACATAGGCGACCAG-3⬘ (sense); SMB, 5⬘-CCGACTCTAACGGAGATCAAC-3⬘ (sense). The conditions for this RACE-PCR were those indicated in the commercial kit. With a 17-mer oligo-dT as a primer, single-stranded cDNA was synthesized by reverse transcription from total RNA isolated from a M. rouxii culture grown 6 –7 h and subsequently amplified by PCR using oligo-dT and SMA and SMB primers. A 500 bp fragment could be identified corresponding to the C-terminus of the protein. Four new specific primers were designed based on the cloned sequences: SM4 and SM6, limiting the site A, and SM20 and SM21, limiting domain B. Primers sequences: SM4, 5⬘-GCGAATTCCATGGACGAGGAACACT-3⬘ (sense); SM6, 5⬘-CCGCTCGAGATAAGGTTCTAATGA-3⬘ (antisense); SM20, 5⬘-CCGGCATCCGCCTTGTTAATATCA-3⬘ (sense); SM21, 5⬘-CCGCCCGGGAATAGCTTGATAGTT-3⬘ (antisense). With genomic DNA of M. rouxii as a template, the primer pairs SM4 –SM6 and SM20 –SM21 were used to amplify by PCR the sequences codifying cAMP binding domains A and B, respectively. The fragments obtained by PCR were cloned into pGEX-4T-3 (all these primers have restriction sites in their 5⬘ termini to allow the cloning) for the expression of the domains fused to GST. The nucleotide sequence of all the cloned DNA fragments, derived from at least two independent polymerase chain reactions, was determined on both strands by the Sanger dideoxy method, using the Sequenase 2.0 kit. Expression and purification of the isolated cAMP binding domains. pGEX-A and pGEX-B are plasmids derived from pGEX-4T-3 carrying the DNA fragments corresponding to site A and to site B. They were expressed in E. coli DH5␣. Bacterial cultures (40 ml) from the logarithmic stage of growth were induced with isopropyl ␣-D-thiogalactopyranoside (0.1 mM) for 4 h at 37°C with shaking. Induced cells were pelleted at 6000g max for 15 min at 4°C, resuspended in ice-cold PBS (4 ml), repelleted, and stored at ⫺20°C until homogenization or used immediately. Except where indicated, all subsequent steps were performed at 4°C. Bacterial pellets were disrupted with glass beads in PBS buffer (1 ml). The extracts were clarified by centrifugation at 12,000g max for 10 min. These crude extracts were used directly for cAMP binding assays or loaded onto glutathione–Sepharose 4B resin for protein purification. A suspension of 40 l of resin per milliliter of crude extract was incubated overnight at 4°C with agitation. The suspension was centrifuged, the supernatant removed, and the glutathione–Sepharose pellet washed three times with 10 bed volumes of PBS. The protein was eluted with three washes of two bed volumes each of glutathione elution buffer (10 mM reduced glutathione in 50 mM Tris–HCl, pH 8) and used immediately. cAMP binding assays. cAMP binding was measured by incubation of the preparations with 1 M [ 3H]cAMP, plus the additions indicated in each case, in 20 mM Tris–HCl buffer, pH 7.4, 0.5 M NaCl at 4°C for 16 h, followed by filtering through nitrocellulose membranes (3). Sequence alignment and translation. The multiple sequence alignment and the translation of M. rouxii sequence were performed with DNAMAN software, version 4.1.
Mucor rouxii PROTEIN KINASE A cAMP BINDING DOMAIN
175
these domains in R subunits for other lower eukaryotic organisms. Homology with Other Regulatory Subunit cAMP Binding Domains
FIG. 1. cDNA and deduced amino acid sequence of M. rouxii protein kinase A regulatory subunit cAMP binding domains. cAMP binding domains A and B are underlined with thin and thick lines, respectively. DNA and protein sequence numbers are relative to the first position, arbitrarily chosen as 1. The DNA and deduced amino acid sequences are deposited with DDBJ/EMBL/GenBank under Accession No. AF240461.
Prediction of tertiary structure. For 3D structure prediction with RI␣ bovine as template (PDB entry 1RGS), the sequence of the cloned M. rouxii cAMP binding domains was submitted to SwissModel, “An Automated Comparative Protein Modelling Server” (http://expasy.hcuge.ch/swissmod/SWISS-MODEL.html) (11), which runs on the software ProMod II (12). The 3D structure was visualized with the software Swiss-Pdb Viewer 2.1 (13).
RESULTS AND DISCUSSION
Cloning and Sequence of cAMP Binding Domains A and B The strategy used for cloning M. rouxii regulatory subunit cAMP binding domains can be summarized as follows: in a first step, a specific product was obtained by PCR amplification from genomic DNA using degenerate primers derived from conserved sequences in all the R subunits. The 3⬘-terminus was obtained by RACE-PCR from a cDNA preparation using oligo-dT as a primer for reverse transcription. The complete cAMP binding domain was finally amplified from genomic DNA. The nucleotide and deduced amino acid sequences are presented in Fig. 1. From the comparison of the genomic DNA and cDNA sequences, it could be immediately observed that this region of the sequence is intronless, as occurs with the sequence codifying
The sequence of M. rouxii cAMP binding domains was compared and aligned with the homologous sequences from seven fungi and with the cAMP binding domains of bovine RII ␣ and RI ␣ (Fig. 2). The first conclusion derived from this alignment is that the general conservation of these domains throughout the whole phylogenetic scale of eukaryotic organisms is highly conserved. The degree of identity of the M. rouxii sequence with almost all of the fungal sequences ranges between 50 and 54%, while the identity with the proteins from a highly unrelated species (bovine) is only slightly lower (47%). It is interesting that the identity between the M. rouxii and B. emersonii sequences is much higher (62%). According to classical taxonomical parameters, these two fungi, M. rouxii and B. emersonii, belong to two different phyla: Zygomycota and Chytridiomycota. However, construction of a neighbor-joining tree of the fungi based on comparison of their 18-S rDNA sequence (TreeBASE M536) (14) has drawn into question whether these two phyla are monophyletic. In this tree, the homology between the 18-S rDNA sequences from M. racemosus and B. emersonii is higher than between the rDNA of other members of their own phylum. The results of the degree of identity of the cAMP binding domain of the regulatory subunits from these two fungi confirm that the comparison of protein sequences is a useful tool to complement rDNA data in the molecular phylogenetic studies used to reinvestigate the systematic position of fungi (15). A second conclusion derived from these comparisons is that the M. rouxii R subunit cannot be classified as a typical type I or type II regulatory subunit. The concentration of NaCl at which it elutes from a DEAE– cellulose column (3) and the capacity of being autophosphorylated (6) make it resemble type II regulatory subunits. However, comparing the sequences of each of the two cAMP binding domains A and B to the corresponding sequences of bovine RI ␣ and RII ␣, one can observe that M. rouxii cAMP binding domain A more closely resembles a type II domain A than a type I domain A (52.0 vs 46.3% of identity) and, on the contrary, M. rouxii domain B is more similar to type I domain B than type II domain B (45.6 vs 37.5%). It is interesting to note that lower eukaryotes have only one gene codifying for the R subunit and that, therefore, only one type of holoenzyme (regarding the R subunit) might be responsible for participating in processes described to be regulated differentially by PKAI and PKAII holoenzymes in higher eukaryotic organisms
176
SOROL ET AL.
FIG. 2. Multialignment of M. rouxii R subunit cAMP binding domains (1–248) with the corresponding sequences of the R subunits from Blastocladiella emersonii (160 – 403), Ustylago maydis (217–536), Emericella nidulans (154 – 402), Magnaporthe grisea (134 –390), Neurospora crassa (133–385), Saccharomyces cerevisiae (187– 416), Schizosaccharomyces pombe (147– 412), bovine RII ␣ (139 – 401), and bovine RI ␣ (139 –379). Only the sequence of bovine RI ␣ is numbered in the alignment. The amino acid sequences were aligned first using the DNAMAN software and further manually to obtain maximal identity, by inserting blanks. The arrow marks the limit between domains A and B. The asterisks correspond to the two loops where the backbone of M. rouxii R structure diverges from the backbone of bovine RI ␣, shown as two orange loops in Fig. 3. The box encloses the ␣ helix C of domain A, described in Fig. 4.
(16, 17). Interesting conclusions can be drawn from the comparison of the fungal cAMP binding domains with those of bovine RI ␣ and RII ␣, shown in Table I. Fungal
cAMP binding domains A are much more identical to bovine domains A than fungal domains B are to the corresponding bovine domain. Fungal cAMP binding
TABLE I
Homology of Fungal cAMP Binding Domains A and B with Bovine RI ␣ and RII ␣ Domains A and B a Domain A
M. rouxii B. emersonii U. maydis E. nidulans Ma. grisea N. crassa S. cerevisiae Sc. pombe a
Domain B
IA
IIA
IB
IIB
IA
IIA
IB
IIB
46.3 51.2 46.3 44.6 47.1 45.5 44.3 46.7
52.0 55.2 44.8 51.2 48.8 50.4 50.4 42.4
26.1 31.4 28.6 32.8 29.4 27.7 31.0 31.1
27.4 29.3 21.9 25.4 23.8 23.1 31.0 24.6
32.5 31.5 23.9 36.4 31.6 31.3 31.4 37.9
28.0 27.9 26.5 29.6 25.8 26.7 27.4 29.4
45.6 44.0 28.9 39.6 31.6 36.0 36.8 32.7
37.5 46.6 23.2 38.1 33.1 35.9 36.4 28.2
The results are expressed in percentage of identity obtained through the alignment program of the DNAMAN software, without manual correction. Italic characters: homology between nonhomologous sites. Bold characters: values representing a distinct greater homology.
Mucor rouxii PROTEIN KINASE A cAMP BINDING DOMAIN
177
Analysis of Residues Important for the High-Order Structure of the cAMP Binding Domains
FIG. 3. Backbone model of the three-dimensional structure of the cAMP binding domains of the M. rouxii regulatory subunit. The cloned sequence of cAMP binding domain was modeled through the Swiss-Model Modelling Server, using as template the coordinates of the bovine RI ␣ subunit (1RGS). In orange are the loops that are not superimposed between both backbones, corresponding to M. rouxii amino acids 48 –55 in domain A and 173–180 in domain B (the amino acids corresponding to the orange loops are marked with an asterisk in the alignment of Fig. 2).
domains B seem to diverge more from the mammalian counterparts, and, in fact in most cases, there is not a clear preference of homology with either bovine domain A or B, either of isoforms I or II. In fact, the degree of identity is, in most cases, not greater than the homology existing between any domain A and B of the same species. It seems as if domain A, important in the interaction with the catalytic subunit, would have been more conserved than domain B, important in the cooperativity between both cAMP binding sites.
Conserved residues involved in the interaction with cAMP, in the cooperativity between the cAMP binding domains, and in the general folding of the R subunit have been identified from data obtained from the crystal structure of cAMP-bound RI ␣ and from genetic studies (1). The residues described to be important in the interaction with the phosphate and ribose of cAMP through several hydrogen bonds and electrostatic contacts are conserved in M. rouxii as well as in all other fungal sequences, namely, Glu200 and Arg209 in domain A and Glu324 and Arg333 in domain B. Both sites are described to have an aromatic residue that, through its side chain, stacks with the adenine ring of cAMP: Trp260 in domain A and Tyr371 in domain B for bovine RI ␣. In M. rouxii and in the rest of the fungi, except U. maydis, Trp260 is replaced by another aromatic residue, Tyr, which can also have stacking interactions with the adenine ring. It is interesting to note that RII ␣ lacks an aromatic residue at this position. In the corresponding position of domain B, the alignment is less clear, due to the nonconserved sequences of the carboxy termini of all the R subunits; however, a Tyr can be aligned for all the sequences except S. cerevisiae. In addition to the electrostatic interactions, the tight binding of cAMP involves a number of strong hydrophobic interactions (1). The residues involved are Val182, Val184, Ala189, Ala210, and Ala211 in site A and Val300, Leu316, Val313, Ile325, and Ala335 in site B. Almost all these residues are identical in M. rouxii or replaced by nonpolar aliphatic amino acids, the only exception being the replacement of Val182 by a Cys.
FIG. 4. (Left) Alignment of the amino acid sequence of the fungal regulatory subunits and RII ␣ as compared to the C helix of domain A of bovine RI ␣ (amino acids 235–244). In bold are the residues aligned with Arg239 of RI ␣. This C helix is enclosed in a box in the multialignment of Fig. 2. (Right) Modeling of M. rouxii amino acids 103–112, corresponding to the C helix of cAMP binding domain A (amino acids 235–244 from bovine RI ␣). The side chains of Leu107, Arg109, Arg110, and Tyr112 are shown. These residues are numbered according to the M. rouxii sequence.
178
SOROL ET AL.
FIG. 5. Expression of domains A and B and purification of isolated domain A: 10% SDS–PAGE gels, stained with Coomassie blue. (A) Twenty micrograms of total proteins from E. coli crude extracts from an induced culture of the bacterial strain transformed with the empty pGEX vector (lane 2) and from uninduced and induced cultures (lanes 3 and 4) from the bacterial strain transformed with GST-A plasmid. Lane 1: MW markers. The dash indicates the 44-kDa overexpressed GST-A product. (B) Ten micrograms of total proteins from E. coli extract transformed with GST-B plasmid, without (lane 1) and with (lane 2) IPTG induction. The dash indicates the 38-kDa overexpressed GST-B product. (C) Purification of GST-A protein; 8 ml of an extract containing 16 mg of protein was purified as described under Materials and Methods. Ten-microliter aliquots of each fraction were loaded on the gel: lane 1, total extract; lane 2, flowthrough; lanes 3 and 4, second and third washes; lanes 5 and 6, first and second elutions.
The three invariant glycine residues in each domain, described to be essential for maintenance of the structural integrity of the -barrels and even conserved in E. coli CRP protein (18), are also conserved in the fungal regulatory subunits. The molecular basis for cooperativity between cAMP domains A and B was described from the crystallization model derived for RI ␣ (1) and it consists of an extended hydrophobic surface. Specifically, the hydrophobic surface in domain B formed by Val265, Leu269, Ile292, Leu294, Tyr321, Val346, Val356, Leu357, Cys369, Ile363, and Leu364 is covered by Tyr244, Phe247, Leu248, Val251, Ile253, Leu254, and Leu257 from domain A. All these residues are identical or have a conservative replacement in M. rouxii cAMP binding domains, except Leu294, which is aligned with a glutamic acid in the fungal proteins and in bovine RII ␣. Molecular Modeling of M. rouxii Structure The 3D structure of M. rouxii cAMP binding domains was obtained by molecular modeling through the Swiss-Model Protein Modelling Server using as template the coordinates of the crystallized bovine RI ␣ subunit (PDB entry 1RGS). The Z values of the root mean square of data from bond lengths, bond angles, and torsion angles were not very far from 1 and were within the range of values expected for a naturally folded protein. The Ramachandran score, expressing how well the backbone conformations of all residues are corresponding to allowed areas in the Ramachandran plot, is within the expected ranges for well-refined structures. The superposition of the backbones for M. rouxii cAMP binding domains and those of its template, shown in Fig. 3, indicates an almost complete
overlap, except for two loop regions (shown in orange) comprising M. rouxii amino acids 48 –55 in domain A and the corresponding region in domain B (amino acids 173–180). These two loops (indicated with asterisks in Fig. 2) correspond to the two regions where the major divergence in sequence arises: in the loop corresponding to domain A, the bovine protein has a smaller number of amino acids, while in the loop corresponding to domain B, there is nonconservative replacement of amino acids throughout the whole loop. Even though the understanding of the details of the interaction between the R and C subunits awaits the crystallization of an R 䡠 C complex, Zhao et al. (19) have modeled the quaternary structure of the holoenzyme, taking into account their own results of neutron contrast variation and results taken from the literature derived from genetic approaches. There are results characterizing the residues in C implicated in the R–C interaction (20 –23), as well as those likely not to be involved (20, 21), and results identifying residues in R that were proposed to be candidate binding sites for the C subunit (2). In this model, the interacting surface of the R subunit includes the following important residues from the cAMP binding domains implicated in R–C interaction, namely Asp140, Glu143, Asp170, Lys242, and Asp258 (corresponding to bovine RI ␣). All these residues are conserved in all the regulatory subunits except Glu143; however, the fungal enzymes and RII ␣ have an acidic residue one position more amino than Glu143 in the RI subunit. Recently, using a yeast two-hybrid screen with the C subunit as bait, Huang and Taylor identified a critical helix in the cAMP binding domain A (helix C) that modulates interactions of this domain with cAMP, with C, and with cAMP bind-
Mucor rouxii PROTEIN KINASE A cAMP BINDING DOMAIN
179
ing domain B (24). This helix corresponds to the region comprising amino acids 234 –244 in bovine RI ␣ and protrudes from the surface defined in R by Zhao et al. (19), to interact with C. The residue in this region important for cooperative interaction between the two cAMP sites, Arg241, is conserved (Arg109 for the Mucor sequence; Fig. 4). However, the opposite surface of this helix, facing toward the C binding site, is not so conserved. Particularly, an important residue in this surface, Arg239 in RI ␣, is replaced by a Leu107 in M. rouxii (Fig. 4), implying a different kind of interaction between this helix in the R subunit of both organisms and the complementary surface in the respective C subunit. This may define a critical difference in the interaction between R and C subunits in the different holoenzymes. However, the interaction through this domain does not seem to explain the low affinity known to exist between the R and C subunits in S. cerevisiae (25), since the R subunit from this yeast possesses a Lys residue in this position, just like bovine RII ␣. Expression and Characterization of M. rouxii Isolated cAMP Binding Domains M. rouxii cAMP binding domains A and B were expressed as fusion proteins with GST in order to use the proteins, if functional, in future studies of interaction with catalytic subunit. Figures 5A and 5B show that the two constructions were overexpressed when inducing the E. coli culture with IPTG. The apparent MW for GST-domain A (GST-A) corresponded to the expected MW (44 kDa), while the MW of GST-domain B (GST-B) was smaller (38 kDa). Extracts overexpressing GST-B protein had a very low cAMP binding activity, which did not correspond to the amount of overexpressed protein; besides, this activity, although severalfold higher than the background activity of extracts transformed with the vector without insert, was not saturated by cAMP up to 100 M, suggesting a poorly folded, unstable protein, with very low affinity for cAMP. However, the overexpression of a functional GST-domain A protein was relatively successful, since the cAMP binding activity of crude extracts was 50- to 100-fold higher than in control extracts (no induction). This activity could be saturated by 1.5 M [ 3H]cAMP. It was relatively unstable, since cAMP binding activity decreased upon incubation of the protein at 30°C; however, the amount and size of the stained band corresponding to the fusion protein did not change with this incubation, suggesting that the decrease in binding activity corresponded to an unfolding of the protein and not to proteolytic degradation (not shown). Therefore, cAMP binding assays were performed by overnight incubations at 4°C. GST-A protein was purified through glutathione–Sepharose columns (Fig. 5C) in order to study some of its kinetic properties. The stoi-
FIG. 6. cAMP-analog selectivity of isolated domain A. Two-microgram aliquots from the purified preparation of GST-A protein were incubated with 1 M [ 3H]cAMP in the presence of 1, 10, or 100 M of each cAMP analog or unlabeled cAMP overnight at 4°C. Bound [ 3H]cAMP was measured by the nitrocellulose filter assay. Binding of [ 3H]cAMP in the absence of added analogs was taken as 100% (7 pmol of cAMP bound). (Top panel) The dotted line indicates the competition curve of the partially purified regulatory subunit with 8-bromo-cAMP (from Ref. 5). N6 ben, N 6 -benzoyl-cAMP; N6 mono, N 6 -monobutyryl-cAMP; 8 Br, 8-bromo-cAMP; 8 ben, 8-thiobenzylcAMP. (Bottom panel) The dotted line indicates the behavior of the partially purified regulatory subunit with Sp-8-bromo-cAMP. Sp, Sp-cAMP; Rp, Rp-cAMP; Sp8, Sp-8-bromo-cAMP; Rp8, Rp-8-bromocAMP.
chiometry of cAMP binding of the purified protein indicated that not more than 10 –15% of the purified protein was active. Site-selective cAMP analogs were used as competitors of [ 3H]cAMP binding in order to analyze whether the isolated domain maintained the characteristics of a cAMP binding site A, which should be selective for N 6 - cAMP analogs. The top panel of Fig. 6 shows that the two site A selective cAMP analogs, N 6 -monobutyryl-cAMP and N 6 -benzoyl-cAMP, were very good competitors of [ 3H]cAMP binding, while the two site B selective analogs, 8-Br-cAMP and 8-thiobenzyl-cAMP, were very poor competitors. C8-cAMP analogs were always found to be poor competitors of [ 3H]cAMP binding of the whole subunit (5); but the behavior of the isolated A domain was clearly different from the behavior of the whole subunit (dotted line in top panel of Fig. 6). The bottom panel of Fig. 6 shows the behavior of the Rp- and Sp-cAMP analogs and of its C8 derivatives as competitors of [ 3H]cAMP binding.
180
SOROL ET AL.
FIG. 7. Dissociation of [ 3H]cAMP bound to isolated domain A. One-microgram aliquots from the purified preparation of GST-A protein were incubated with 1 M [ 3H]cAMP overnight at 4°C. Dissociation kinetics was measured after addition of 100 M nonradioactive cAMP. Samples were filtered at the indicated times. The dotted line represents the dissociation kinetics of cAMP bound to site A of the whole R subunit, partially purified from M. rouxii extracts; the data were recalculated from those of Ref. 3 to express them as dissociation from site A; during the 180 min of the experiment at 4°C, the dissociation of [ 3H]cAMP bound to site B was negligible. A value of 100% binding represents 3 pmol of bound cAMP.
Rp-cAMP analogs did not compete, as occurs with the whole subunit (C. Mizyrycki and S. Moreno, unpublished results); however, Sp-8Br-cAMP, which competed with the binding of [ 3H]cAMP to the whole subunit (dotted line), did not compete with the binding to the isolated A domain. The dissociation kinetics at 4°C of the [ 3H]cAMP bound to GST-A protein is shown in Fig. 7. The dissociation curve is monophasic, as expected for only one site, suggesting there is one population of molecules. The estimated half-life for the bound cAMP is around 45 min, in the same range of the estimate we had for the dissociation kinetics from this site in the whole subunit at 4°C (3), drawn as a dotted line in the same figure for comparison. The results from this section show that even though only a small percentage of the protein is active, the kinetic properties of the isolated recombinant cAMP domain A from the M. rouxii regulatory subunit are analogous to the properties of the domain in the whole subunit. This suggests that domain B does not contribute significantly to the affinity, dissociation kinetics, selectivity, and folding of domain A. This result contrasts with those reported for a deletion mutant of RI ␣ (26), in which upon removal of domain B from the whole protein, the dissociation kinetics from [ 3H]cAMP bound to domain A was greatly increased; in this case, the amino terminus of the protein is still conserved. The isolation of cAMP-binding fragments with binding activity is not new. However, the identity of those fragments as one or the other of the cAMP binding domains has not been determined. Small fragments, presumably derived from proteolysis of the native reg-
ulatory subunits, have been reported in bovine spermatozoa (27), rat liver (28), and yeast (29). Trypsin digestion products of purified mammalian type I and II R subunits have been reported to have binding activity (30, 31). The fact that it was not possible to obtain an active cAMP binding domain B from the recombinant protein from M. rouxii, while an active domain A could be obtained, suggested us, as a first analysis, that in the case of Mucor, domain A might be necessary for the adequate folding of domain B. However, a recombinant cAMP binding domain B from a mammalian system has been obtained in an active form and reported to function independently of any other domain structure (32). In this report, we establish that the physical limits for a functional cAMP domain A in M. rouxii can be defined to the limits of the functional domain in the whole subunit. REFERENCES 1. Su, Y., Dostmann, W. R. G., Herberg, F. W., Durick, K., Xuong, N., Ten Eyck, L. F., Taylor, S. S., and Varughese, K. I. (1995) Science 269, 807– 813. 2. Gibson, R. M., Ji-Buechler, Y., and Taylor, S. S. (1997) J. Biol. Chem. 272, 16343–16350. 3. Moreno, S., and Passeron, S. (1980) Arch. Biochem. Biophys. 199, 321–330. 4. Pastori, R., Moreno, S., and Passeron, S. (1985) Mol. Cell. Biochem. 69, 55– 66. 5. Paveto, C., Passeron, S., Corbin, J. D., and Moreno, S. (1989) Eur. J. Biochem. 179, 429 – 434. 6. Rossi, S., Guthmann, M., and Moreno, S. (1992) Cell. Signal. 4, 443– 451. 7. Pastori, R., Kerner, N., Moreno, S., and Passeron, S. (1981) Biochem. Biophys. Res. Commun. 101, 663– 671. 8. Guthmann, M., Pastori, R., and Moreno, S. (1990) Cell. Signal. 2, 395– 402. 9. Hoffman, C. S., and Winston, D. (1987) Gene 57, 267–272. 10. Ausubel, F. M., Brent, R., Kingston, R., Moore, D., Seidman, J. G., Smith, J. A., and Struhl, K. (1993) Short Protocols in Molecular Biology, 2nd ed., Greene Publishing Associates/Wiley, New York. 11. Peitsch, M. C. (1995) Biotechnology 13, 658 – 660. 12. Peitsch, M. C. (1996) Biochem. Soc. Trans. 24, 274 –279. 13. Guex, N., and Peitsch, M. C. (1977) Electrophoresis 18, 2714 – 2723. 14. Tanabe, Y., Nagahama, T., Saikawa, M., and Sugiyama, J. (1999) Mycologia 91, 830 – 835. 15. Thon, M. R., and Royse, D. J. (1999) Mycologia 91, 468 – 474. 16. Cho-Chung, Y. S., and Clair, T. (1993) Pharmacol. Ther. 60, 265–288. 17. Tasken, T., Skalhegg, B. S., Tasken, K. A., Solberg, R., Knutsen, H. K., Levy, F. O., Sandberg, M., Orstavik, S., Larse, T., Johansen, A. K., Vang, T., Schrader, H. P., Reinton, N. T., Torgersen, K. M., Hansson, V., and Jahnsen, T. (1997) Adv. Second Messenger Phosphoprotein Res. 31, 191–204. 18. Shabb, J. B., and Corbin, J. D. (1992) J. Biol. Chem. 267, 5723– 5726. 19. Zhao, J., Hoye, E., Boylan, S., Walsh, D. A., and Trewhella, J. (1998) J. Biol. Chem. 273, 30448 –30459.
Mucor rouxii PROTEIN KINASE A cAMP BINDING DOMAIN 20. Gibbs, C. S., and Zoller, M. J. (1991) J. Biol. Chem. 266, 8923–8931. 21. Gibbs, C. S., Knighton, D. R., Sowadski, J. M., Taylor, S. S., and Zoller, M. J. (1992) J. Biol. Chem. 267, 4806 – 4814. 22. Orellana, S. A., Amieux, P. S., Zhao, X., and McKnight, G. S. (1993) J. Biol. Chem. 268, 6843– 6846. 23. Gibson, R. M., Buechler, Y. J., and Taylor, S. S. (1997) Protein Sci. 6, 1825–1834. 24. Huang, L. J., and Taylor, S. S. (1998) J. Biol. Chem. 273, 26739 – 26746. 25. Kuret, J., Johnson, K. E., Nicolette, C., and Zoller, M. J. (1988) J. Biol. Chem. 263, 9149 –9154. 26. Woodford, T. A., Correll, L. A., McKnight, G. S., and Corbin, J. D. (1989) J. Biol. Chem. 264, 13321–13328.
181
27. Garbers, D. L., First, N. L., and Lardy, H. A. (1973) J. Biol. Chem. 248, 875– 879. 28. Srivastava, A. K., and Stellwagen, R. H. (1978) J. Biol. Chem. 253, 1752–1755. 29. Takai, Y., Yamamura, H., and Nishizuka, Y. (1974) J. Biol. Chem. 249, 530 –535. 30. Rannels, S. R., and Corbin, J. D. (1979) J. Biol. Chem. 254, 8605– 8610. 31. Rannels, S. R., and Corbin, J. D. (1980) J. Cycl. Nucleotide Res. 8, 201–215. 32. Shabb, J. B., Poteet, C. E., Kapphahn, M. A., Muhonen, W. M., Baker, N. E., and Corbin, J. D. (1995) Protein Sci. 4, 2100 – 2106.