Structural basis of the intrasteric regulation of myosin light chain kinases

Intrasteric regulation of myosin light chain kinase

Ca2+/calmodulin activates myosin light chain kinase by reversal of an autoinhibited state. The effects of substitution mutations on calmodulin activation properties implicate 4 of the 8 basic residues between the catalytic core and the calmodulin-binding domain in maintaining autoinhibition. These residues are further amino-terminal to the basic residues comprising the previously proposed pseudosubstrate sequence and suggest involvement of the connecting region in intrasteric autoinhibition. The pseudosubstrate model for autoinhibition proposes that basic residues within the autoinhibitory region mimic basic residues in the substrate and bind to defined acidic residues within the catalytic core. Charge reversal mutations of these specific acidic residues, however, had little or no effect on the Km value for regulatory light chain. From a total of 20 acidic residues on the surface of the substrate binding lobe of the catalytic core, 7 are implicated in binding directly or indirectly to the autoinhibitory domain but not to the light chain. Only 2 acidic residues near the catalytic site may bind to the autoinhibitory domain and the arginine at P-3 in the light chain. Exposure of these 2 residues upon calmodulin binding may be necessary and sufficient for light chain phosphorylation.

Mutational analysis of Ca(2+)-independent autophosphorylation of calcium/calmodulin-dependent protein kinase II

Previous studies with synthetic peptides indicate that residues 290-309, corresponding to the calmodulin (CaM)-binding domain of Ca2+/CaM-dependent protein kinase II interact with the catalytic core of the enzyme as a pseudosubstrate (Colbran, R. J., Smith, M. K., Schworer, C. M., Fong, Y. L., and Soderling, T. R. (1989) J. Biol. Chem. 264, 4800-4804). In the present study, we attempted to locate the pseudosubstrate motif by generation or removal of potential substrate recognition sequences (RXXS/T) at selected positions using site-directed mutagenesis. Based on previous results, Arg297, Thr305/306, and Ser314 were selected as key residues. Single mutations such as N294S, K300S, A302R, A309R, and R311A were expressed, purified, and characterized. Several of the mutants exhibited decreased binding of and activation by CaM, not surprising since the mutations were within the CaM-binding domain. None of the mutants exhibited enhanced Ca(2+)-independent kinase activity toward exogenous substrate, but the K300S and N294S mutants showed a significant enhancement in the rate and stoichiometry of 32P incorporation during Ca(2+)-independent autophosphorylation. Using two-dimensional peptide mapping and phosphoamino acid analyses, enhanced phosphorylation of the introduced Ser residue was demonstrated in the K300S mutant but not in the N294S mutant. This specific Ca(2+)-independent autophosphorylation of Ser300 is consistent with the hypothesis that Arg297 may occupy the P (-3) position in a pseudosubstrate autoinhibitory interaction with the catalytic core in the nonactivated state of the kinase.

Covalent control of 6-phosphofructo-2-kinase/fructose-2,6-bisphosphatase: insights into autoregulation of a bifunctional enzyme

The hepatic bifunctional enzyme, 6-phosphofructo-2-kinase/fructose-2,6-bisphosphatase (6PF-2-K/Fru-2,6-P2ase), E.C. 2.7-1-105/E.C. 3-1-3-46, is one member of a family of unique bifunctional proteins that catalyze the synthesis and degradation of the regulatory metabolite fructose-2,6-bisphosphate (Fru-2,6-P2). Fru-2,6-P2 is a potent activator of the glycolytic enzyme 6-phosphofructo-1-kinase and an inhibitor of the gluconeogenic enzyme fructose-1,6-bisphosphatase, and provides a switching mechanism between these two opposing pathways of hepatic carbohydrate metabolism. The activities of the hepatic 6PF-2-K/Fru-2,6-P2ase isoform are reciprocally regulated by a cyclic AMP-dependent protein kinase (cAPK)-catalyzed phosphorylation at a single NH2-terminal residue, Ser-32. Phosphorylation at Ser-32 inhibits the kinase and activates the bisphosphatase, in part through an electrostatic mechanism. Substitution of Asp for Ser-32 mimics the effects of cAPK-catalyzed phosphorylation. In the dephosphorylated homodimer, the NH2- and COOH-terminal tail regions also have an interaction with their respective active sites on the same subunit to produce an autoregulatory inhibition of the bisphosphatase and activation of the kinase. In support of this hypothesis, deletion of either the NH2- or COOH-terminal tail region, or both regions, leads to a disruption of these interactions with a maximal activation of the bisphosphatase. Inhibition of the kinase is observed with the NH2-truncated forms, in which there is also a diminution of cAPK phosphorylation to decrease the Km for Fru-6-P. Phosphorylation of the bifunctional enzyme by cAPK disrupts these autoregulatory interactions, resulting in inhibition of the kinase and activation of the bisphosphatase. Therefore, effects of cyclic AMP-dependent phosphorylation are mediated by a combination of electrostatic and autoregulatory control mechanisms.

Two structures of the catalytic domain of phosphorylase kinase: an active protein kinase complexed with substrate analogue and product

BACKGROUND

Control of intracellular events by protein phosphorylation is promoted by specific protein kinases. All the known protein kinase possess a common structure that defines a catalytically competent entity termed the 'kinase catalytic core'. Within this common structural framework each kinase displays its own unique substrate specificity, and a regulatory mechanism that may be modulated by association with other proteins. Structural studies of phosphorylase kinase (Phk), the major substrate of which is glycogen phosphorylase, may be expected to shed light on its regulation.

RESULTS

We report two crystal structures of the catalytic core (residues 1-298; Phk gamma trnc) of the gamma-subunit of rabbit muscle phosphorylase kinase: the binary complex with Mn2+/beta-gamma-imidoadenosine 5'-triphosphate (AMPPNP) to a resolution of 2.6 A and the binary complex with Mg2+/ADP to a resolution of 3.0 A. The structures were solved by molecular replacement using the cAMP-dependent protein kinase (cAPK) as a model.

CONCLUSIONS

The overall structure of Phk gamma trnc is similar to that of the catalytic core of other protein kinases. It consists of two domians joined on one edge by a 'hinge', with the catalytic site located in the cleft between the domains. Phk gamma trnc is constitutively active, and lacks the need for an activatory phosphorylation event that is essential for many kinases. The structure exhibits an essentially 'closed' conformation of the domains which is similar to that of cAPK complexed with substrates. The phosphorylated residue that is located at the domain interface in many protein kinases and that is believed to stabilize an active conformation is substituted by a glutamate in Phk gamma trnc. The glutamate, in a similar manner to the phosphorylated residue in other protein kinases, interacts with an arginine adjacent to the catalytic aspartate but does not participate in interdomain contacts. The interactions between the enzyme and the nucleotide product of its activity, Mg2+/ADP, explain the inhibitory properties of the nucleotides that are observed in kinetic studies.

Photoaffinity labeling of a peptide substrate to myosin light chain kinase

The substrate binding properties of skeletal muscle myosin light chain kinase were investigated with a synthetic peptide containing the photoreactive amino acid p-benzoylphenylalanine (Bpa) incorporated amino-terminal of the phosphoacceptor serine (BpaKKRAARATSNVFA). When photolyzed at 350 nm, the peptide was cross-linked stoichiometrically to myosin light chain kinase in a Ca2+/calmodulin-dependent manner. Peptide incorporation into kinase inhibited light chain phosphorylation, and the loss of kinase activity was proportional to the extent of peptide incorporated. After peptide I was incorporated into myosin light chain kinase, it was partially phosphorylated in the absence of Ca2+/calmodulin. The extent of phosphorylation increased in the presence of Ca2+/calmodulin. The cross-linked photoadduct was digested, labeled peptides were purified by high performance liquid chromatography, and sites of covalent modification were determined by amino acid sequencing and analysis. The covalent modification in the catalytic core occurred on Ile-373 (66%) and in a peptide containing residues Asn-422 to Met-437 (14%), respectively. Lys-572 in the autoinhibitory region accounted for 20% of the incorporated label. The coincident covalent modification of the autoinhibitory domain suggests that it is located near the catalytic site. Based upon a model of the catalytic core, the substrate peptide is predicted to bind in the cleft between the two lobes of the kinase. The orientation of the substrate peptide on myosin light chain kinase is similar to the orientation of the substrate recognition fragment, but not the high affinity binding fragment, of inhibitor peptide of cAMP-dependent protein kinase in the catalytic subunit of the cAMP-dependent protein kinase.

Identification of the substrate and pseudosubstrate binding sites of phosphorylase kinase gamma-subunit

Using site-directed mutagenesis, we proposed that an autoinhibitory domain(s) is located at the C-terminal region (301-386) of the phosphorylase kinase gamma-subunit (Huang, C.-Y.F., Yuan C.-J., Livanova, N.B., and Graves, D.J. (1993) Mol. Cell. Biochem. 127/128, 7-18). Removal of the putative inhibitory domain(s) by truncation results in the generation of a constitutively active and calmodulin-independent form, gamma 1-300. To probe the structural basis of autoinhibition of gamma-subunit activity, two synthetic peptides, PhK13 (gamma 303-327) and PhK5 (gamma 343-367), corresponding to the two calmodulin-binding regions, were assayed for their ability to inhibit gamma 1-300. Competitive inhibition of gamma 1-300 by PhK13 was found versus phosphorylase b (Ki = 1.8 microM) and noncompetitive inhibition versus ATP. PhK5 showed noncompetitive inhibition with respect to both phosphorylase b and ATP. Calmodulin released the inhibition caused by both peptides. These results indicate that there are two distinct auto-inhibitory domains within the C terminus of the gamma-subunit and that these two domains overlap with the calmodulin-binding regions. Two mutant forms of gamma 1-300, E111K and E154R, were used to probe the enzyme-substrate-binding region using peptide substrate analogs corresponding to residues 9-18 of phosphorylase b (KRK11Q12ISVRGL). The data suggest that Glu111 interacts with the P-3 position of the substrate (Lys11) and Glu154 interacts with the P-2 site (Gln12). Both E111K and E154R were competitively inhibited with respect to phosphorylase b by PhK13, with 14- and 8-fold higher Ki values, respectively, than that observed with the wild-type enzyme. These data are consistent with a model for the regulation of the gamma-subunit of phosphorylase kinase in which PhK13 acts as a competitive pseudosubstrate that directly binds the substrate binding site of the gamma-subunit (Glu111 and Glu154).

Stabilization of calmodulin-dependent protein kinase II through the autoinhibitory domain

The active 30-kDa chymotryptic fragment of calmodulin-dependent protein kinase II (CaM kinase II), devoid of the autoinhibitory domain, and the enzyme, autothiophosphorylated at Thr286/Thr287, were much more labile than was the original native enzyme. They were markedly stabilized by synthetic peptides, designed after the sequence around the autophosphorylation site in the autoinhibitory domain, such as autocamtide-2 and CaMK-(281-309), but such marked stabilizations were not observed with the ordinary exogenous substrates, such as syntide-2. These results suggest that the autoinhibitory domain of CaM kinase II plays a crucial role in stabilizing the enzyme. A nonphosphorylatable analog of autocamtide-2, AIP, strongly inhibited the activity of the 30-kDa fragment. Kinetic analysis revealed that the inhibition by AIP was competitive with respect to autocamtide-2 and CaMK-(281-289) and noncompetitive with respect to syntide-2 and ATP/Mg2+, suggesting that CaM kinase II possesses at least two distinct substrate-binding sites; one for ordinary exogenous substrates such as syntide-2 and the other for an endogenous substrate, the autophosphorylation site (Thr286/Thr287) in the autoinhibitory domain. Fluorescence analysis of the binding of 7-nitrobenz-2-oxa-1,3-diazole-4-yl labeled AIP to the 30-kDa fragment also supported this contention. Thus, the autoinhibitory domain appears to play a crucial role in keeping the enzyme stable by binding to the substrate-binding site for the autophosphorylation site.

Targeted disruption of Ca(2+)-calmodulin signaling in Drosophila growth cones leads to stalls in axon extension and errors in axon guidance

Ca(2+)-calmodulin (CaM) function was selectively disrupted in a specific subset of growth cones in transgenic Drosophila embryos in which a specific enhancer element drives the expression of the kinesin motor domain fused to a CaM antagonist peptide (kinesin-antagonist or KA, which blocks CaM binding to target proteins) or CaM itself (kinesin-CaM or KC, which acts as a Ca(2+)-binding protein). In both KA and KC mutant embryos, specific growth cones exhibit dosage-dependent stalls in axon extension and errors in axon guidance, including both defects in fasciculation and abnormal crossings of the midline. These results demonstrate an in vivo function for Ca(2+)-CaM signaling in growth cone extension and guidance and suggest that Ca(2+)-CaM may in part regulate specific growth cone decisions, including when to defasciculate and whether or not to cross the midline.

Calmodulin-binding domains: just two faced or multi-faceted?

The Ca(2+)-binding protein calmodulin binds to and activates several cellular enzymes in response to a rise in Ca2+ concentration. It binds certain basic amphiphilic helices within these enzymes, which also act as autoinhibitory domains. The modulation of the binding equilibrium of these helices between intramolecular (inhibition) and intermolecular (activation) sites forms a focal point for crosstalk between various signalling pathways.

Structural basis for chromosome X-linked agammaglobulinemia: a tyrosine kinase disease

X-linked agammaglobulinemia (XLA) is a hereditary defect of B-cell differentiation in man caused by deficiency of Bruton tyrosine kinase (BTK). A three-dimensional model for the BTK kinase domain, based on the core structure of cAMP-dependent protein kinase, was used to interpret the structural basis for disease in eight independent point mutations in patients with XLA. As Arg-525 of BTK has been thought to functionally substitute for a critical lysine residue in protein-serine kinases, the mutation Arg-525-->Gln was studied and found to abrogate the tyrosine kinase activity of BTK. All of the eight mutations (Lys-430-->Glu, Arg-520-->Glu, Arg-525-->Gln, Arg-562-->Pro, Ala-582-->Val, Glu-589-->Gly, Gly-594-->Glu, and Gly-613-->Asp) were located on one face of the BTK kinase domain, indicating structural clustering of functionally important residues.

Substrate recognition by casein kinase-II: the role of histidine-160

Casein kinase-II (CK-II) belongs to the protein kinases recognizing serine/threonine in proximity to acidic residues in protein substrates. Crystallography and mutagenesis studies on the cAMP-dependent protein kinase (PKA) disclosed that glutamic acid-170 (E170), is important for interaction of substrates with the enzyme. At a position corresponding to E170 in PKA most Ser/Thr kinases have an aspartic or glutamic acid, while CK-II has a histidine residue (H160). In order to examine the relevance of this substitution for CK-II substrate specificity, a mutant of the catalytic alpha subunit (H160D), in which H160 was changed to aspartic acid, was made. Our results show that H160 is not primarily involved in canonical substrate recognition, but does interact with an acidic residue located at position -2 with respect to the target Ser/Thr.

Protein kinases

The structures of four serine/threonine protein kinases have been determined recently. By comparing these structures with that of the cAMP-dependent protein kinase (cAPK), it is now possible to see how the activity of these regulatory enzymes is controlled. Low activity is maintained through the conformation of the phosphorylation lip, domain rotations, and binding of substrate analog inhibitors and autoinhibitory domains.

The biochemical basis of the regulation of smooth-muscle contraction

The primary signal for smooth-muscle contraction is an increase in sarcoplasmic free Ca2+ concentration ([Ca2+]i). This triggers activation of calmodulin-dependent myosin light-chain kinase, which catalyses myosin phosphorylation, thereby activating crossbridge cycling and the development of force or contraction of the muscle cell. Restoration of resting [Ca2+]i deactivates the kinase; myosin is dephosphorylated by myosin light-chain phosphatase and the muscle relaxes. Recent evidence suggests that other signal-transduction pathways can modulate the contractile state of a smooth-muscle cell by affecting specific steps in the myosin phosphorylation-dephosphorylation mechanism.

Calmodulin and the regulation of smooth muscle contraction

Calmodulin, the ubiquitous and multifunctional Ca(2+)-binding protein, mediates many of the regulatory effects of Ca2+, including the contractile state of smooth muscle. The principal function of calmodulin in smooth muscle is to activate crossbridge cycling and the development of force in response to a [Ca2+]i transient via the activation of myosin light-chain kinase and phosphorylation of myosin. A distinct calmodulin-dependent kinase, Ca2+/calmodulin-dependent protein kinase II, has been implicated in modulation of smooth-muscle contraction. This kinase phosphorylates myosin light-chain kinase, resulting in an increase in the calmodulin concentration required for half-maximal activation of myosin light-chain kinase, and may account for desensitization of the contractile response to Ca2+. In addition, the thin filament-associated proteins, caldesmon and calponin, which inhibit the actin-activated MgATPase activity of smooth-muscle myosin (the cross-bridge cycling rate), appear to be regulated by calmodulin, either by the direct binding of Ca2+/calmodulin or indirectly by phosphorylation catalysed by Ca2+/calmodulin-dependent protein kinase II. Another level at which calmodulin can regulate smooth-muscle contraction involves proteins which control the movement of Ca2+ across the sarcolemmal and sarcoplasmic reticulum membranes and which are regulated by Ca2+/calmodulin, e.g. the sarcolemmal Ca2+ pump and the ryanodine receptor/Ca2+ release channel, and other proteins which indirectly regulate [Ca2+]i via cyclic nucleotide synthesis and breakdown, e.g. NO synthase and cyclic nucleotide phosphodiesterase. The interplay of such regulatory mechanisms provides the flexibility and adaptability required for the normal functioning of smooth-muscle tissues.

Three protein kinase structures define a common motif

Structural comparisons between cAMP-dependent protein kinase, cyclin-dependent kinase 2 and mitogen-activated protein kinase reveal which features are common to the protein kinase family and which are enzyme-specific.

Crystallization and preliminary X-ray analysis of the auto-inhibited twitchin kinase

An auto-inhibited fragment of twitchin kinase (residues 5890 to 6262) has been crystallized by vapor diffusion techniques using polyethylene glycol 4000 as the precipitant at pH 7.25 to 7.5 at 4 degrees C. We have found that MgSO4 and glycerol were essential for large crystal growth. The crystals belong to the orthorhombic space group P2(1)2(1)2, with unit cell dimensions of a = 144.1 A, b = 168.3 A and c = 60.6 A. They are suitable for X-ray analysis and diffract to a resolution of at least 2.8 A.

The essential mitotic target of calmodulin is the 110-kilodalton component of the spindle pole body in Saccharomyces cerevisiae

Two independent methods identified the spindle pole body component Nuf1p/Spc110p as the essential mitotic target of calmodulin. Extragenic suppressors of cmd1-1 were isolated and found to define three loci, XCM1, XCM2, and XCM3 (extragenic suppressor of cmd1-1). The gene encoding a dominant suppressor allele of XCM1 was cloned. On the basis of DNA sequence analysis, genetic cosegregation, and mutational analysis, XCM1 was identified as NUF1/SPC110. Independently, a C-terminal portion of Nuf1p/Spc110p, amino acid residues 828 to 944, was isolated as a calmodulin-binding protein by the two-hybrid system. As assayed by the two-hybrid system, Nuf1p/Spc110p interacts with wild-type calmodulin and triple-mutant calmodulins defective in binding Ca2+ but not with two mutant calmodulins that confer a temperature-sensitive phenotype. Deletion analysis by the two-hybrid system mapped the calmodulin-binding site of Nuf1p/Spc110p to amino acid residues 900 to 927. Direct binding between calmodulin and Nuf1p/Spc110p was demonstrated by a modified gel overlay assay. Furthermore, indirect immunofluorescence with fixation procedures known to aid visualization of spindle pole body components localized calmodulin to the spindle pole body. Sequence analysis of five suppressor alleles of NUF1/SPC110 indicated that suppression of cmd1-1 occurs by C-terminal truncation of Nuf1p/Spc110p at amino acid residues 856, 863, or 881, thereby removing the calmodulin-binding site.

The essential mitotic target of calmodulin is the 110-kilodalton component of the spindle pole body in Saccharomyces cerevisiae

Two independent methods identified the spindle pole body component Nuf1p/Spc110p as the essential mitotic target of calmodulin. Extragenic suppressors of cmd1-1 were isolated and found to define three loci, XCM1, XCM2, and XCM3 (extragenic suppressor of cmd1-1). The gene encoding a dominant suppressor allele of XCM1 was cloned. On the basis of DNA sequence analysis, genetic cosegregation, and mutational analysis, XCM1 was identified as NUF1/SPC110. Independently, a C-terminal portion of Nuf1p/Spc110p, amino acid residues 828 to 944, was isolated as a calmodulin-binding protein by the two-hybrid system. As assayed by the two-hybrid system, Nuf1p/Spc110p interacts with wild-type calmodulin and triple-mutant calmodulins defective in binding Ca2+ but not with two mutant calmodulins that confer a temperature-sensitive phenotype. Deletion analysis by the two-hybrid system mapped the calmodulin-binding site of Nuf1p/Spc110p to amino acid residues 900 to 927. Direct binding between calmodulin and Nuf1p/Spc110p was demonstrated by a modified gel overlay assay. Furthermore, indirect immunofluorescence with fixation procedures known to aid visualization of spindle pole body components localized calmodulin to the spindle pole body. Sequence analysis of five suppressor alleles of NUF1/SPC110 indicated that suppression of cmd1-1 occurs by C-terminal truncation of Nuf1p/Spc110p at amino acid residues 856, 863, or 881, thereby removing the calmodulin-binding site.

A conserved helix motif complements the protein kinase core

Residues 40-300 of the mammalian catalytic (C) subunit of cAMP-dependent protein kinase define a conserved bilobal catalytic core shared by all eukaryotic protein kinases. Contiguous to the core is an extended amphipathic alpha-helix (A helix). Trp30, a prominent feature of this helix, fills a deep hydrophobic pocket between the two lobes on the surface opposite to the active site. The C subunit in Dictyostelium discoideum shows sequence conservation of residues 40-350 with the mouse enzyme but contains an N-terminal extension of 332 residues. A sequence corresponding to the A helix contiguous to the core is absent. However, we have now identified a remote A-helix motif (residues 77-98). When the core of the Dictyostelium C subunit was modeled, based on the mouse C subunit, complementarity between this putative A helix and the surface of the core was found to be conserved. Analysis of other protein kinases reveals that the A-helix motif is not restricted to cAMP-dependent protein kinase. In the Src-related family of protein kinases, for example, an A helix is very likely contiguous to the core, thus serving as a linker between the conserved catalytic core and the Src homology 2 domain. We predict that an A-helix motif complementary to the core will be a conserved feature of most eukaryotic protein kinases.

Affinity labelling of smooth-muscle myosin light-chain kinase with 5'-[p-(fluorosulphonyl)benzoyl]adenosine

5'-(p-(Fluorosulphonyl)[14C]benzoyl)adenosine (FSBA) was synthesized and used as a probe to study the ATP-binding site of smooth-muscle myosin light-chain kinase (MLCK). FSBA modified both free MLCK and calmodulin/MLCK complex, resulting in inactivation of the kinase activity. Nearly complete protection of the calmodulin/MLCK complex against FSBA modification was obtained by addition of excess ATP whereas MLCK activity alone was lost in a dose-dependent manner even in the presence of excess ATP. These results suggest that FSBA modified ATP-binding sites and ATP-independent sites, and the latter sites are protected by calmodulin binding. The results also suggest that the ATP-binding site is accessible to the nucleotide substrate regardless of calmodulin binding. The FSBA-labelled MLCK was completely proteolysed by alpha-chymotrypsin, and the 14C-labelled peptides were isolated and sequenced. The sequence of the labelled peptide was Ala-Gly-X-Phe, where X is the labelled residue. The sequence was compared with the known MLCK sequence, and the labelled residue was identified as lysine-548, which is located downstream of the GXGXXG motif conserved among ATP-utilizing enzymes.

A peptide analog of the calmodulin-binding domain of myosin light chain kinase adopts an alpha-helical structure in aqueous trifluoroethanol

A 22-residue synthetic peptide encompassing the calmodulin (CaM)-binding domain of skeletal muscle myosin light chain kinase was studied by two-dimensional NMR and CD spectroscopy. In water the peptide does not form any regular structure; however, addition of the helix-inducing solvent trifluoroethanol (TFE) causes it to form an alpha-helical structure. The proton NMR spectra of this peptide in 25% and 40% TFE were assigned by double quantum-filtered J-correlated spectroscopy, total correlation spectroscopy, and nuclear Overhauser effect correlated spectroscopy spectra. In addition, the alpha-carbon chemical shifts were obtained from (1H,13C)-heteronuclear multiple quantum coherence spectra. The presence of numerous dNN(i, i + 1), d alpha N(i, i + 3), and d alpha beta(i, i + 3) NOE crosspeaks indicates that an alpha-helix can be formed from residues 3 to 20; this is further supported by the CD data. Upfield alpha-proton and downfield alpha-carbon shifts in this region of the peptide provide further support for the formation of an alpha-helix. The helix induced by TFE appears to be similar to that formed upon binding of the peptide to CaM.

A transmembrane protein with a cdc2+/CDC28-related kinase activity is required for signaling from the ER to the nucleus

In eukaryotic cells, the accumulation of unfolded proteins in the endoplasmic reticulum (ER) triggers a signaling pathway from the ER to the nucleus. Several yeast mutants defective in this pathway map to the ERN1 gene, which protects cells from lethal consequences of stress by signaling for increased expression of BiP and other ER proteins. ERN1 encodes a 1115 amino acid transmembrane protein (Ern1p) whose glycosylated N-terminal portion is located inside microsomes and whose cytoplasmic C-terminal portion carries an essential protein kinase activity. We postulate that Ern1p is the proximal sensor of events in the ER and that binding of ligand causes transduction of information across the ER membrane, leading to activation of a specific set of transcription factors.

A three-dimensional model of the Cdc2 protein kinase: localization of cyclin- and Suc1-binding regions and phosphorylation sites

The Cdc2 protein kinase requires cyclin binding for activity and also binds to a small protein, Suc1. Charged-to-alanine scanning mutagenesis of Cdc2 was used previously to localize cyclin A- and B- and Suc1-binding sites (B. Ducommun, P. Brambilla, and G. Draetta, Mol. Cell. Biol. 11:6177-6184, 1991). Those sites were mapped by building a Cdc2 model based on the crystallographic coordinates of the catalytic subunit of cyclic AMP-dependent protein kinase (cAPK) (D. R. Knighton, J. Zheng, L. F. Ten Eyck, V. A. Ashford, N.-H. Xuong, S. S. Taylor, and J. M. Sowadski, Science 253:407-414, 1991). On the basis of this model, additional mutations were made and tested for cyclin A and Suc1 binding and for kinase activity. Mutations that interfere with cyclin A binding are localized primarily on the small lobe near its interface with the cleft and include an acidic patch on the B helix and R-50 in the highly conserved PSTAIRE sequence. Two residues in the large lobe, R-151 and T-161, influence cyclin binding, and both are at the surface of the cleft near its interface with the PSTAIRE motif. Cyclin-dependent phosphorylation of T-161 in Cdc2 is essential for activation, and the model provides insights into the importance of this site. T-161 is equivalent to T-197, a stable phosphorylation site in cAPK. On the basis of the model, cyclin binding very likely alters the surface surrounding T-161 to allow for T-161 phosphorylation. The two major ligands to T-197 in cAPK are conserved as R-127 and R-151 in Cdc2. The equivalent of the third ligand, H-87, is T-47 in the PSTAIRE sequence motif. Once phosphorylated, T-161 is predicted to play a major structural role in Cdc2, comparable to that of T-197 in cAPK, by assembling the active conformation required for peptide recognition. The inhibitory phosphorylation at Y-15 also comes close to the cleft interface and on the basis of this model would disrupt the cleft interface and the adjacent peptide recognition site rather than prevent ATP binding. In contrast to cyclin A, both lobes influence Suc1 binding; however, the Suc1-binding sites are far from the active site. Several mutants map to the surface in cAPK, which is masked in part by the N-terminal 40 residues that lie outside the conserved catalytic core. The other Suc1-binding site maps to the large lobe near a 25-residue insert and includes R-215.

A three-dimensional model of the Cdc2 protein kinase: localization of cyclin- and Suc1-binding regions and phosphorylation sites

The Cdc2 protein kinase requires cyclin binding for activity and also binds to a small protein, Suc1. Charged-to-alanine scanning mutagenesis of Cdc2 was used previously to localize cyclin A- and B- and Suc1-binding sites (B. Ducommun, P. Brambilla, and G. Draetta, Mol. Cell. Biol. 11:6177-6184, 1991). Those sites were mapped by building a Cdc2 model based on the crystallographic coordinates of the catalytic subunit of cyclic AMP-dependent protein kinase (cAPK) (D. R. Knighton, J. Zheng, L. F. Ten Eyck, V. A. Ashford, N.-H. Xuong, S. S. Taylor, and J. M. Sowadski, Science 253:407-414, 1991). On the basis of this model, additional mutations were made and tested for cyclin A and Suc1 binding and for kinase activity. Mutations that interfere with cyclin A binding are localized primarily on the small lobe near its interface with the cleft and include an acidic patch on the B helix and R-50 in the highly conserved PSTAIRE sequence. Two residues in the large lobe, R-151 and T-161, influence cyclin binding, and both are at the surface of the cleft near its interface with the PSTAIRE motif. Cyclin-dependent phosphorylation of T-161 in Cdc2 is essential for activation, and the model provides insights into the importance of this site. T-161 is equivalent to T-197, a stable phosphorylation site in cAPK. On the basis of the model, cyclin binding very likely alters the surface surrounding T-161 to allow for T-161 phosphorylation. The two major ligands to T-197 in cAPK are conserved as R-127 and R-151 in Cdc2. The equivalent of the third ligand, H-87, is T-47 in the PSTAIRE sequence motif. Once phosphorylated, T-161 is predicted to play a major structural role in Cdc2, comparable to that of T-197 in cAPK, by assembling the active conformation required for peptide recognition. The inhibitory phosphorylation at Y-15 also comes close to the cleft interface and on the basis of this model would disrupt the cleft interface and the adjacent peptide recognition site rather than prevent ATP binding. In contrast to cyclin A, both lobes influence Suc1 binding; however, the Suc1-binding sites are far from the active site. Several mutants map to the surface in cAPK, which is masked in part by the N-terminal 40 residues that lie outside the conserved catalytic core. The other Suc1-binding site maps to the large lobe near a 25-residue insert and includes R-215.

The ligand-binding domain in metabotropic glutamate receptors is related to bacterial periplasmic binding proteins

Receptors for the major excitatory neurotransmitter glutamate include metabotropic (G protein-coupled) and ionotropic (glutamate-gated ion channel) types. These receptors have large, presumably extracellular, amino-terminal domains. Sensitive sequence analysis techniques indicate that the metabotropic receptor extracellular domain is similar to bacterial periplasmic amino acid binding proteins. A structural model built using the observed similarity predicts a ligand-binding site, and mutants with conservative amino acid substitutions at this site are shown to have reduced ligand affinity. The metabotropic receptor extracellular domain is a member of a family of structural domains linked to a variety of receptor types, including ionotropic glutamate receptors.

Structural features that specify tyrosine kinase activity deduced from homology modeling of the epidermal growth factor receptor

To identify structural features that distinguish protein-tyrosine kinases from protein-serine kinases, a molecular model of the kinase domain of epidermal growth factor receptor was constructed by substituting its amino acid sequence for the amino acid sequence of the catalytic subunit of cAMP-dependent protein kinase in a 2.7-A refined crystallographic model. General folding was conserved as was the configuration of invariant residues at the active site. Two sequence motifs that distinguish the two families correspond to loops that converge at the active site of the enzyme. A conserved arginine in the catalytic loop is proposed to interact with the gamma phosphate of ATP. The second loop provides a binding surface that positions the tyrosine of the substrate. A positively charged surface provides additional sites for substrate recognition.

Crystal structure of the catalytic subunit of cAMP-dependent protein kinase complexed with MgATP and peptide inhibitor

The structure of a ternary complex of the catalytic subunit of cAMP-dependent protein kinase, MgATP, and a 20-residue inhibitor peptide was determined at a resolution of 2.7 A using the difference Fourier technique starting from the model of the binary complex (Knighton et al., 1991a). The model of the ternary complex was refined using both X-PLOR and TNT to an R factor of 0.212 and 0.224, respectively. The orientation of the nucleotide and the interactions of MgATP with numerous conserved residues at the active site of the enzyme are clearly defined. The unique protein kinase nucleotide binding site consists of a five-stranded antiparallel beta-sheet with the base buried in a hydrophobic site along beta-strands 1 and 2 and fixed by hydrogen bonds to the N6 amino and N7 nitrogens. The small lobe secures the nucleotide via a glycine-rich loop and by ion pairing with Lys72 and Glu91. While the small lobe fixes the nontransferable alpha- and beta-phosphates in this inhibitor complex, the gamma-phosphate is secured by two Mg2+ ions and interacts both directly and indirectly with several residues in the large lobe--Asp184, Asn171, Lys168. Asp166 is positioned to serve as a catalytic base. The structure is correlated with previous chemical evidence, and the features that distinguish this nucleotide binding motif from other nucleotide binding proteins are delineated.

A template for the protein kinase family

The crystal structure of the catalytic subunit of cAMP-dependent protein kinase, complexed with ATP and a 20-residue inhibitor peptide, is reviewed and correlated with chemical and genetic data. The striking convergence of the structure with the biochemistry and genetics provides for the first time a molecular basis for understanding how this enzyme functions, as well as an explanation for the highly conserved residues that are scattered throughout the molecule. Because these residues probably serve a common role in all eukaryotic protein kinases, this first protein kinase structure serves as a general template for the entire family of enzymes.

Regulation of intrasteric inhibition of the multifunctional calcium/calmodulin-dependent protein kinase

A regulatory region involved in both autoinhibition and calmodulin (CaM) binding has previously been identified in the multifunctional Ca2+/CaM-dependent protein kinase (CaM kinase II). We have tested the role of various segments of the regulatory region in autoinhibition by the analysis of a series of truncation, substitution, and deletion mutants of the CaM kinase II alpha subunit (CaM kinase II alpha). Unexpectedly, the sequence Lys-Lys-Phe-Asn at positions 291-294, adjacent to the CaM binding domain, was found to be sufficient to maintain an inhibited state in a truncated form of the kinase. However, these residues are not essential in the context of the full-length protein, indicating the importance of additional residues from the overlapping CaM binding domain. We propose here a molecular model for CaM kinase II alpha based on the three-dimensional structure of the cAPK-PKI-(5-24) (protein kinase inhibitor fragment) complex. It is predicted from this model that autoinhibition is of the pseudosubstrate variety and that autophosphorylation of Thr-286 could occur by an intersubunit reaction in the holoenzyme complex.