Reprints
PubMed
CAVEAT LECTOR
Mauno Vihinen1, Pekka T. Mattsson2,3 and C. I. Edvard Smith2
1Department of Biosciences, Division of Biochemistry, P. O. Box 56, FIN-00014 University of Helsinki, Finland
2Center for BioTechnology, Department of Bioscience at Novum, Karolinska Institute, S-141 57 Huddinge and Department of Immunology, Microbiology, Pathology and Infectious Diseases (IMPI), Karolinska Institute, Huddinge University Hospital, S-141 86 Huddinge, Sweden
3Department of Biochemistry and Food Chemistry, University of Turku, Vatselankatu 2, Arcanum, FIN-20014 Turku, Finland
Received 11/24/96; Accepted 12/16/96; On-line 01/01/97
6. STRUCTURAL CONSEQUENCES OF BTK MUTATIONS
The structure of four of the five Btk domains has been modeled to study structure-function and genotype-phenotype interactions (69-72). The gene defect leading to XLA has been characteized in a large number of patients. The mutations have been collected into a database called BTKbase (28-31). Recent analysis of the registry indicated that in the 368 XLA patients, in 318 unrelated families, mutations are scattered throughout the entire length of BTK gene (31). The proportion of unique mutations is 72% (228 cases), and the distribution of the mutations in the five structural domains corresponds to the length of the domains. Exonic mutations are distributed as follows: 123 families had missense mutations, 66 had nonsense mutations, 24 showed insertions, and 57 had deletions. In addition, there are 49 intron mutations affecting splice sites. Three double mutations and a single triple mutation have been detected. The gene defect of nine gross deletions have not been characterized in detail. As expected, the missense mutations appear mainly in the first two positions within the codon. Altogether, there were in the missense and nonsense mutations 135 transitions and 54 transversions corresponding to 71 and 29% of the single amino acid substitutions, respectively. Eight of 18 CpG containing arginine residues were affected, whereas none of the residual 15 CpG sites encoding non-arginine residues were mutated. CpG dinucleotides are involved in all the cases where at least five families have the same mutation except for the initiation site. The larger deletions encompass whole exons.
The models of the domains have been used to give putative structural description for each of the XLA mutations. The BTKbase is available at World Wide Web at:
[http://www.helsinki.fi/science/signal/btkbase.html].
Highly divergent pleckstrin homology (PH) domains of about 120 residues have been found in a number of signaling and cytoskeletal proteins including protein kinases and their substrates, phospholipase C, GTPase activating proteins, guanine nucleotide releasing factors, and adaptor proteins (73-79). The Tec family kinases are the only PTKs which contain a PH domain. The 3D structure has been determined for several PH domains. Although these proteins share very limited sequence identity, they have the same fold consisting of a ß-barrel formed of two ß-sheets and a C-terminal alpha-helix that caps one end of the ß-barrel. The Btk PH domain was modeled (72) based on the dynamin structure (80, 81).
The N-terminal half of at least certain PH domains bind phosphoinositides and the binding residues have been localized in the pleckstrin and spectrin PH domains (57, 59, 63). Site-directed mutagenesis of the three conserved lysines in a charged patch of the pleckstrin PH domain significantly decreased the binding (61). The binding is specific to PIP2, having a Kd of 41 µM (57).
In view of clustering of XLA mutations, the corresponding region in Btk PH domain initially was thought to be involved in binding (72). It was later shown that mutations of Btk residues close to the sites corresponding to the conserved lysines led to the disease (72). According to biosensor assays, Btk PH domain specifically binds to unilamellar liposomes containing PIP3 in a R28 dependent manner with a Kd of 1.23 µM (82). A point mutation in the PH domain has been shown to cause X-linked immunodeficiency (Xid) in mice (R28C) (83, 84) and XLA in man (31). The modeled Btk structure indicated presence of a putative binding site that could consist of two parts; a highly charged patch and a cleft formed by hydrophobic and aromatic residues (72). The PH domain has been suggested to replace the function of myristylation in membrane targeting of at least some cytoplasmic proteins. Recently, substitution of Btk PH domain residue, E41, by lysine was shown to increase phosphorylation of tyrosine residues and membrane targeting (51). Thus, the Btk phosphorylation might be linked to membrane interaction. Most of the Btk PH domain mutations are concentrated in the binding site region where they could disturb interactions (Fig. 3). The other mutations usually distort the folding of the domain.
Figure 3. The modeled structure of Btk PH domain with bound PIP3 (red) in the same site as in pleckstrin and spectrin. The PH domain is displayed as surface presentations. Residues affected by XLA-causing mutations are shown in yellow.
Many PH domains, including Btk, have been shown to bind to ßgamma subunits of heterotrimeric G proteins (85, 86). Only the C-terminal half of the PH domain and some 30 residues from the following TH domain are required for this interaction (85). Btk and Itk kinase activity is stimulated by Gßgamma subunits and some unidentified membrane factor(s) (50).
The Tec family members contain a unique region between the PH and SH3 domains which is tentatively designated the TH (Tec homology) domain (87, 88). Conserved N-terminal Btk motif is followed by a proline rich region (PRR). Although the whole TH domain can be found only in the Tec family members, the PH domain followed by the Btk motif is present also in several forms of Ras GTPase-activating protein 1 (Ras-GAP1) and in a putative interferon-gamma-binding protein (88). The Btk motif contains invariant histidine and cysteines residues which in many cases are involved in metal binding.
Two 10 amino acid motifs in the PRR of Btk have been shown to interact with the SH3 domains of Fyn, Lyn and Hck (41-43), but no data are yet available for association of full length Src family kinases. Itk PRR is also bound by the same proteins with the same specificity (41). The corresponding region of Tec binds to Lyn (89). The Src family kinases are activated early in the B cell activation. Erythropoietin and IL3 stimulation induces the specific binding of Vav to Tec through the TH domain (90). An unidentified, 72 kDa protein, binds to residues 186-192 in the TH domain of Btk suggesting this domain to mediate stable protein-protein interactions (43).
Of the TH domain binding Src family SH3 domains (41, 43), the 3D structure has been determined for Fyn (91-93). The first of the Btk proline rich repeats was modeled based on the high-affinity peptide binding to c-Src (94) and docked into the SH3 domains (Fig. 4), because the Src family kinases have been shown to preferentially bind this region (43). The binding of the Btk PRR peptide is similar to the other known high affinity interactions. The TH domain PRRs have RLP type sequences (94).
Figure 4. The first proline rich repeat of Btk TH domain docked into Fyn SH3 domain (90). The binding residues in the Fyn are color coded as follows: Y91 orange, Y93 blue, D100 green, W119 white, and Y137 turquoise. In the Btk PRR peptide (cyan), proline residues are red.
Compared to Src for which high-affinity peptide complex structure is available (94), the polyproline binding residues in Fyn (Y91, Y93, D100, W119, and Y137) are identical. The positions of these residues are also similar, but still the binding specificities are different (42, 95, 96). Subtle changes are known to alter affinity and specificity of SH3 domains (97-99).
Site-directed mutations of the polyproline II (PPII) helix forming proline residues in the PRRs of Btk abolish binding to SH3 domain (41, 43). Mutations, P189A and P192A (41, 43), are likely to alter the conformation such that the polyproline stretch can no longer be recognized. Also, mutation in the conserved polyproline binding region of Fyn (W119L) abolished the binding (43). On the other hand, mutation in another PPII binding site amino acid, D100N, had little or no effect on binding (43).
The Btk PRR does not bind to SH3 domains from Abl, Blk, Btk or Crk (43). The structure of the Btk SH3 domain has been modeled (69). Although Btk SH3 domain has not been shown to bind to the TH domain (41, 43) the modeled PRR peptide was also docked into the SH3 domain and the binding was compared to the Src family binding. The binding site is very similar to that of Fyn and Src. All the major binding residues are conserved (Y223, Y225, D232, W251, and Y268) and have corresponding positions. Although, binding by these residues may occur, other interactions outside this region are likely to be crucial. It is known that Hck affinity is more than 300 fold higher for a full length Nef protein compared to synthesized peptide motif (99).
The SH3 domains are modules which bind polyproline stretches containing polypeptides and proteins. The Btk SH3 domain was modeled based on the Fyn structure (69). The Btk SH3 domain has been shown to interact with the proline-rich c-cbl protooncogene (54).
There are several nonsense mutations in the Btk SH3 domain, but no known missense mutations have been found (30, 31). Aberrant splicing and skipping of exon 9 leads to an in-frame deletion of 21 residues containing the 14 C-terminal residues of the SH3 domain in two unrelated families (69, 100). Even though this protein is expressed in a stable form in cells and has full kinase activity in vitro, the patients have classical XLA (69). Deletion of the last three ß-strands seems to distort the structure. According to molecular dynamics simulation, the mutant protein has a stable structure. The spacing between the termini in the mutant protein corresponds to the normal Btk SH3 domain thus facilitating connection to the rest of the Btk without major changes in the overall scaffolding (69).
SH2 domains bind phosphotyrosine (pY) containing peptides and proteins. The specificity is gained by recognizing residues from C-terminal to pY. The Btk SH2 was modeled using v-Src as a template (71). The two ß-sheets and the terminal alpha-helices are conserved. Most of the XLA-causing Btk SH2 domain mutations disrupt the pY peptide binding sites (28-31).
In many kinases, SH3 and SH2 domains are in close proximity to each other. In Src and Tec families, the domains have only a few intervening residues and only have few intramolecular contacts in the Lck SH2-SH3 dimer (101).
Binding specificities of several SH2 domains have been determined by using phoshotyrosine peptide libraries (102, 103). We have predicted the Btk SH2 domain sequence to be pYEXL/I. These peptides were docked into the binding site (71). The sites for phosphotyrosine and residue +1 are formed predominantly by charged residues, whereas the site +3 is mainly hydrophobic. When searching databases, the isoleucine peptide was found in Hck and both YEXI and YEXL motifs of ßARK-1 and ßARK-2 (71). The modeled Btk SH2 domain with pYEAI peptide is seen in Fig. 5. While both Btk and ßARK contain PH domain which can interact with Gßgamma it is not known if these proteins interact.
Figure 5. The Btk SH2 domain model with the putative binding peptide, pYEAI (71). The binding residues are indicated with different colors.
The kinase domain of about 250 residues is the only catalytic part in most kinases including the Tec family PTKs. The 3D structure has been solved for several protein kinases. The first 3D structure was for cAPK, which was subsequently refined and crystallized with cofactors and inhibitors (104-108). Although overall sequence similarities are generally low, all the known protein kinases share several conserved residues (109). The known 3D structures have the same scaffolding consisting of two lobes, where ATP is bound in a cleft between the two lobes and substrate interacts mainly with the lower lobe.
Protein kinases are generally regulated by phosphorylation in the activation loop. In cAPK, the phosphothreonine is highly coordinated by residues from the activation and catalytic loops (104, 110, 111). When the enzyme is activated, the upper lobe rotates to lock the ATP molecule between the two domains (112). The ATP binding residues are the most conserved sites in all protein kinases suggesting that both PSKs and PTKs have the same direct in-line reaction mechanism (111, 113).
Btk kinase domain was originally modeled based on the cAPK structure (70) and subsequently based on the IRK and FGF. The models have been used to study the functional implications of XLA causing mutations (28-31, 70, 114, 115). The kinase domain model was also used to design a novel mutation, that altered the enzyme activity in a predictable fashion (116). Residue, W563, is rather conserved and according to the available models, it is sandwiched between residues R562 and A582 (70). Although W563 and the two surrounding amino acids are not directly involved in catalysis, mutations in the lining residues cause XLA (70), presumably by affecting the orientation of the W563 side chain. Conservative mutation W563F inactivated the enzyme in a predicted manner (116). The expressed protein had no kinase activity, but it presumably folded correctly.
Almost half of the XLA-causing mutations are in the kinase domain which forms more than 40% of Btk. The mutations are almost generally distributed along the Btk sequence, except for the upper lobe, which forms about one third of the domain's length incorporating only 16% of the kinase domain mutations (31). Putative structural description has been given for each XLA mutation (28-30). There are several different types of missense mutations affecting structural, functional and interacting residues. The Btk kinase domain models in Fig. 6 indicate the distribution of the mutations along the polypeptide chain. The severe XLA mutations are mainly in the ATP-binding cleft, the putative substrate binding region or in other functionally or structurally crucial sites (29, 31). Milder XLA causing mutations can be further away from the functional regions, with some exceptions (29, 31). The very same mutation causes sometimes classical XLA in one patient and only a mild one in another.
Figure 6. Btk kinase domain model (70). On the left side of the stereo pair the XLA causing missense mutations are shown with color coding for a number of unrelated families. To the right, the other types of mutations are color coded in the kinase domain. The number of affected families is indicated as follows: one family, yellow; two families, cyan; three families, green; four families, white; five families, orange and six or more families in red.