The structure of CDK8/CycC implicates specificity in the CDK/cyclin family and reveals interaction with a deep pocket binder.

Cyclin-dependent kinase (CDK) 8 associates with cyclin C (CycC) and belongs to the CDK module of the Mediator of transcription, together with MED12 and MED13. CDK8 is involved in the regulation of mRNA transcription and was identified as a potent oncogene in colon cancerogenesis. We have solved the 2.2-Å crystal structure of CDK8/CycC in complex with sorafenib, an anti-cancer drug of clinical relevance. The CDK8 structure reveals a unique CycC recognition helix that explains the specificity of the CDK8/CycC pair and discrimination among the highly promiscuous binding in the CDK/cyclin family. In contrast to all CDKs, the CDK8 activation loop appears not to be phosphorylated. Based on the structure, we discuss an alternate mode of CDK8 activation to the general CDK activation by T-loop phosphorylation. Sorafenib binds to the catalytic cleft of CDK8. It displays a deep pocket binding mode and is the first small molecule to induce a DFG-out conformation in the CDK family, which is actually DMG-out in CDK8.

Flexibility and variability of TIMP binding: X-ray structure of the complex between collagenase-3/MMP-13 and TIMP-2.

The excessive activity of matrix metalloproteinases (MMPs) contributes to pathological processes such as arthritis, tumor growth and metastasis if not balanced by the tissue inhibitors of metalloproteinases (TIMPs). In arthritis, the destruction of fibrillar (type II) collagen is one of the hallmarks, with MMP-1 (collagenase-1) and MMP-13 (collagenase-3) being identified as key players in arthritic cartilage. MMP-13, furthermore, has been found in highly metastatic tumors. We have solved the 2.0 A crystal structure of the complex between the catalytic domain of human MMP-13 (cdMMP-13) and bovine TIMP-2. The overall structure resembles our previously determined MT1-MMP/TIMP-2 complex, in that the wedge-shaped TIMP-2 inserts with its edge into the entire MMP-13 active site cleft. However, the inhibitor is, according to a relative rotation of approximately 20 degrees, oriented differently relative to the proteinase. Upon TIMP binding, the catalytic zinc, the zinc-ligating side chains, the enclosing MMP loop and the S1' wall-forming segment move significantly and in concert relative to the rest of the cognate MMP, and the active site cleft constricts slightly, probably allowing a more favourable interaction between the Cys1(TIMP) alpha-amino group of the inhibitor and the catalytic zinc ion of the enzyme. Thus, this structure supports the view that the central N-terminal TIMP segment essentially defines the relative positioning of the TIMP, while the flanking edge loops determine the relative orientation, depending on the individual target MMP.

First steps towards effective methods in exploiting high-throughput technologies for the determination of human protein structures of high biomedical value.

The EC 'Structural Proteomics In Europe' contract is aimed specifically at the atomic resolution structure determination of human protein targets closely linked to health, with a focus on cancer (kinesins, kinases, proteins from the ubiquitin pathway), neurological development and neurodegenerative diseases and immune recognition. Despite the challenging nature of the analysis of such targets, approximately 170 structures have been determined to date. Here, the impact of high-throughput technologies, such as parallel expression of multiple constructs, the use of standardized refolding protocols and optimized crystallization screens or the use of mass spectrometry to assist sample preparation, on the structural biology of mammalian protein targets is illustrated through selected examples.

Crystal structure of the catalytic domain of MMP-16/MT3-MMP: characterization of MT-MMP specific features.

Membrane-type matrix metalloproteinases (MT-MMPs) have attracted strong attention, because four of them can activate a key player in the tumor scenario, proMMP-2/progelatinase A. In addition to this indirect effect on the cellular environment, these MT-MMPs degrade extracellular matrix proteins, and their overproduction is associated with tumor growth. We have solved the structure of the catalytic domain (cd) of MT3-MMP/MMP-16 in complex with the hydroxamic acid inhibitor batimastat. CdMT3-MMP exhibits a classical MMP-fold with similarity to MT1-MMP. Nevertheless, it also shows unique properties such as a modified MT-specific loop and a closed S1' specificity pocket, which might help to design specific inhibitors. Some MT-MMP-specific features, derived from the crystal structures of MT-1-MMP determined previously and MT3-MMP, and revealed in recent mutagenesis experiments, explain the impaired interaction of the MT-MMPs with TIMP-1. Docking experiments with proMMP-2 show some exposed loops including the MT-loop of cdMT3-MMP involved in the interaction with the proMMP-2 prodomain in the activation encounter complex. This model might help to understand the experimentally proven importance of the MT-loop for the activation of proMMP-2.

Substrate specificity determinants of human macrophage elastase (MMP-12) based on the 1.1 A crystal structure.

The macrophage elastase enzyme (MMP-12) expressed mainly in alveolar macrophages has been identified in the mouse lung as the main destructive agent associated with cigarette smoking, which gives rise to emphysema, both directly via elastin degradation and indirectly by disturbing the proteinase/antiproteinase balance via inactivation of the alpha1-proteinase inhibitor (alpha1-PI), the antagonist of the leukocyte elastase. The catalytic domain of human recombinant MMP-12 has been crystallized in complex with the broad-specificity inhibitor batimastat (BB-94). The crystal structure analysis of this complex, determined using X-ray data to 1.1 A and refined to an R-value of 0.165, reveals an overall fold similar to that of other MMPs. However, the S-shaped double loop connecting strands III and IV is fixed closer to the beta-sheet and projects its His172 side-chain further into the rather hydrophobic active-site cleft, defining the S3 and the S1-pockets and separating them from each other to a larger extent than is observed in other MMPs. The S2-site is planar, while the characteristic S1'-subsite is a continuous tube rather than a pocket, in which the MMP-12-specific Thr215 replaces a Val residue otherwise highly conserved in almost all other MMPs. This alteration might allow MMP-12 to accept P1' Arg residues, making it unique among MMPs. The active-site cleft of MMP-12 is well equipped to bind and efficiently cleave the AlaMetPhe-LeuGluAla sequence in the reactive-site loop of alpha1-PI, as occurs experimentally. Similarities in contouring and particularly a common surface hydrophobicity both inside and distant from the active-site cleft explain why MMP-12 shares many substrates with matrilysin (MMP-7). The MMP-12 structure is an excellent template for the structure-based design of specific inhibitors for emphysema therapy and for the construction of mutants to clarify the role of this MMP.

A novel endo-beta-galactosidase from Clostridium perfringens that liberates the disaccharide GlcNAcalpha 1-->Gal from glycans specifically expressed in the gastric gland mucous cell-type mucin.

We found that commercially available sialidases prepared from Clostridium perfringens ATCC10543 were contaminated with an endoglycosidase capable of releasing the disaccharide GlcNAcalpha1-->4Gal from glycans expressed in the gastric gland mucous cell-type mucin. We have isolated this enzyme in electrophoretically homogeneous form from the culture supernatant of this organism by ammonium sulfate precipitation followed by affinity chromatography using a Sephacryl S-200 HR column. The enzyme was specifically retained by and eluted from the column with methyl-alpha-Glc. By NMR spectroscopy, the structure of the disaccharide released from porcine gastric mucin by this enzyme was established to be GlcNAcalpha1-->4Gal. The specificity of this enzyme as an endo-beta-galactosidase was established by analyzing the liberation of GlcNAcalpha1-->4Gal from GlcNAcalpha1-->4Galbeta1-->4GlcNAcbeta1-->6(GlcNAcalpha1--> 4Galbeta1-->3)GalNAc-ol by mass spectrometry. Because this novel endo-beta-galactosidase specifically releases the GlcNAcalpha1-->4Gal moiety from porcine gastric mucin, we propose to call this enzyme a GlcNAcalpha1-->4Gal-releasing endo-beta-galactosidase (Endo-beta-Gal(GnGa)). Endo-beta-Gal(GnGa) was found to remove the GlcNAcalpha1-->4Gal epitope expressed in gastric adenocarcinoma AGS cells transfected with alpha1,4-N-acetylglucosaminyltransferase cDNA. Endo-beta-Gal(GnGa) should become useful for studying the structure and function of glycoconjugates containing the terminal GlcNAcalpha1-->4Gal epitope.

Insights into MMP-TIMP interactions.

The proteolytic activity of the matrix metalloproteinases (MMPs) involved in extracellular matrix degradation must be precisely regulated by their endogenous protein inhibitors, the tissue inhibitors of metalloproteinases (TIMPs). Disruption of this balance can result in serious diseases such as arthritis and tumor growth and metastasis. Knowledge of the tertiary structures of the proteins involved in such processes is crucial for understanding their functional properties and to interfere with associated dysfunctions. Within the last few years, several three-dimensional structures have been determined showing the domain organization, the polypeptide fold, and the main specificity determinants of the MMPs. Complexes of the catalytic MMP domains with various synthetic inhibitors enabled the structure-based design and improvement of high-affinity ligands, which might be elaborated into drugs. Very recently, structural information also became available for some TIMP structures and MMP-TIMP complexes, and these new data elucidated important structural features that govern the enzyme-inhibitor interaction.

Structural properties of matrix metalloproteinases.

Matrix metalloproteinases (MMPs) are involved in extracellular matrix degradation. Their proteolytic activity must be precisely regulated by their endogenous protein inhibitors, the tissue inhibitors of metalloproteinases (TIMPs). Disruption of this balance results in serious diseases such as arthritis, tumour growth and metastasis. Knowledge of the tertiary structures of the proteins involved is crucial for understanding their functional properties and interference with associated dysfunctions. Within the last few years, several three-dimensional MMP and MMP-TIMP structures became available, showing the domain organization, polypeptide fold and main specificity determinants. Complexes of the catalytic MMP domains with various synthetic inhibitors enabled the structure-based design and improvement of high-affinity ligands, which might be elaborated into drugs. A multitude of reviews surveying work done on all aspects of MMPs have appeared in recent years, but none of them has focused on the three-dimensional structures. This review was written to close the gap.

Endoproteinase-protein inhibitor interactions.

Nature uses protein inhibitors as important tools to regulate the proteolytic activity of their target proteinases. Most of these inhibitors for which 3D structures are available are directed towards serine proteinases, interacting with their active-sites in a substrate-like "canonical" manner via an exposed reactive-site loop of conserved conformation. More recently, some non-canonically binding serine proteinase inhibitors, two cysteine proteinase inhibitors, and three zinc endopeptidase inhibitors have been characterized in the free and complexed state, displaying novel mechanisms of inhibition with their target proteinases. These different interaction modes are briefly discussed, with particular emphasis on the interaction between matrix metalloproteinases (MMPs) and their endogenous tissue inhibitors of metalloproteinases (TIMPs).

Residue 2 of TIMP-1 is a major determinant of affinity and specificity for matrix metalloproteinases but effects of substitutions do not correlate with those of the corresponding P1' residue of substrate.

The unregulated activities of matrix metalloproteinases (MMPs) are implicated in disease processes including arthritis and tumor cell invasion and metastasis. MMP activities are controlled by four homologous endogenous protein inhibitors, tissue inhibitors of metalloproteinases (TIMPs), yet different TIMPs show little specificity for individual MMPs. The large interaction interface in the TIMP-1.MMP-3 complex includes a contiguous region of TIMP-1 around the disulfide bond between Cys1 and Cys70 that inserts into the active site of MMP-3. The effects of fifteen different substitutions for threonine 2 of this region reveal that this residue makes a large contribution to the stability of complexes with MMPs and has a dominant influence on the specificity for different MMPs. The size, charge, and hydrophobicity of residue 2 are key factors in the specificity of TIMP. Threonine 2 of TIMP-1 interacts with the S1' specificity pocket of MMP-3, which is a key to substrate specificity, but the structural requirements in TIMP-1 residue 2 for MMP binding differ greatly from those for the corresponding residue of a peptide substrate. These results demonstrate that TIMP variants with substitutions for Thr2 represent suitable starting points for generating more targeted TIMPs for investigation and for intervention in MMP-related diseases.

Crystal structure of the complex formed by the membrane type 1-matrix metalloproteinase with the tissue inhibitor of metalloproteinases-2, the soluble progelatinase A receptor.

The proteolytic activity of matrix metalloproteinases (MMPs) towards extracellular matrix components is held in check by the tissue inhibitors of metalloproteinases (TIMPs). The binary complex of TIMP-2 and membrane-type-1 MMP (MT1-MMP) forms a cell surface located 'receptor' involved in pro-MMP-2 activation. We have solved the 2.75 A crystal structure of the complex between the catalytic domain of human MT1-MMP (cdMT1-MMP) and bovine TIMP-2. In comparison with our previously determined MMP-3-TIMP-1 complex, both proteins are considerably tilted to one another and show new features. CdMT1-MMP, apart from exhibiting the classical MMP fold, displays two large insertions remote from the active-site cleft that might be important for interaction with macromolecular substrates. The TIMP-2 polypeptide chain, as in TIMP-1, folds into a continuous wedge; the A-B edge loop is much more elongated and tilted, however, wrapping around the S-loop and the beta-sheet rim of the MT1-MMP. In addition, both C-terminal edge loops make more interactions with the target enzyme. The C-terminal acidic tail of TIMP-2 is disordered but might adopt a defined structure upon binding to pro-MMP-2; the Ser2 side-chain of TIMP-2 extends into the voluminous S1' specificity pocket of cdMT1-MMP, with its Ogamma pointing towards the carboxylate of the catalytic Glu240. The lower affinity of TIMP-1 for MT1-MMP compared with TIMP-2 might be explained by a reduced number of favourable interactions.

Crystal structure of the catalytic domain of human tumor necrosis factor-alpha-converting enzyme.

Tumor necrosis factor-alpha (TNFalpha) is a cytokine that induces protective inflammatory reactions and kills tumor cells but also causes severe damage when produced in excess, as in rheumatoid arthritis and septic shock. Soluble TNFalpha is released from its membrane-bound precursor by a membrane-anchored proteinase, recently identified as a multidomain metalloproteinase called TNFalpha-converting enzyme or TACE. We have cocrystallized the catalytic domain of TACE with a hydroxamic acid inhibitor and have solved its 2.0 A crystal structure. This structure reveals a polypeptide fold and a catalytic zinc environment resembling that of the snake venom metalloproteinases, identifying TACE as a member of the adamalysin/ADAM family. However, a number of large insertion loops generate unique surface features. The pro-TNFalpha cleavage site fits to the active site of TACE but seems also to be determined by its position relative to the base of the compact trimeric TNFalpha cone. The active-site cleft of TACE shares properties with the matrix metalloproteinases but exhibits unique features such as a deep S3' pocket merging with the S1' specificity pocket below the surface. The structure thus opens a different approach toward the design of specific synthetic TACE inhibitors, which could act as effective therapeutic agents in vivo to modulate TNFalpha-induced pathophysiological effects, and might also help to control related shedding processes.

Temperature dependence of backbone dynamics in loops of human mitochondrial heat shock protein 10.

A highly flexible, yet conserved polypeptide loop of Hsp10 mediates binding to Hsp60 in the course of chaperonin-dependent protein folding. Previous transferred nuclear Overhauser effect (trNOE) studies with peptides based on the mobile loop of the Escherichiacoli and bacteriophage T4 Hsp10s suggested that the mobile loop adopts a characteristic hairpin turn upon binding to the E. coli Hsp60 GroEL. In this paper, we identify the sequence and characterize the nascent structure and dynamics of the 18-residue mobile loop in the 15N-enriched human Hsp10. We also identify four residues of another flexible loop, the roof beta hairpin. The mobile loop and/or roof beta hairpin of several subunits are absent from the X-ray crystal structure of human Hsp10. NMR data suggest that the mobile loop of Hsp10 preferentially samples a hairpin conformation despite the fact that the backbone motion resembles that of a disordered polypeptide. Analysis of backbone dynamics by measurement of 15N relaxation times, T1 and T2, and the 1H-15N nuclear Overhauser effect (1H-15N NOE) indicates that motion is greatest near the center of the loop. Inversion of the temperature dependence of the T1 near the center of the loop marks a transition to motion with a dominant time scale of less than 3 ns. Analysis of the relaxation data by spectral density mapping shows that subnanosecond motion increases uniformly along the loop at elevated temperatures, whereas nanosecond motion increases near the ends of the loop and decreases near the center of the mobile loop. The transition to dominance by fast motion in the center of the loop occurs at a distance from the well-structured part of Hsp10 that is equal to the persistence length of an unstructured polypeptide. Simulation of the spectral density function for the 15N resonance and its temperature dependence using the Lipari-Szabo formalism suggests that the dominant time scales of loop motion range from 0.6 to 18 ns. For comparison, the time scale for molecular rotation of the 70 kDa Hsp10 heptamer is estimated to be 37 ns. Complex behavior of the T2 relaxation time indicates that motion also occurs on longer time scales. All of the modes of loop motion are likely to have an impact on Hsp10/Hsp60 interaction and therefore affect Hsp10/Hsp60 function as a chaperonin.

Mechanism of inhibition of the human matrix metalloproteinase stromelysin-1 by TIMP-1.

Matrix metalloproteinases (MMPs) are zinc endopeptidases that are required for the degradation of extracellular matrix components during normal embryo development, morphogenesis and tissue remodelling. Their proteolytic activities are precisely regulated by endogenous tissue inhibitors of metalloproteinases (TIMPs). Disruption of this balance results in diseases such as arthritis, atherosclerosis, tumour growth and metastasis. Here we report the crystal structure of an MMP-TIMP complex formed between the catalytic domain of human stromelysin-1 (MMP-3) and human TIMP-1. TIMP-1, a 184-residue protein, has the shape of an elongated, contiguous wedge. With its long edge, consisting of five different chain regions, it occupies the entire length of the active-site cleft of MMP-3. The central disulphide-linked segments Cys 1-Thr 2-Cys 3-Val 4 and Ser 68-Val 69 bind to either side of the catalytic zinc. Cys 1 bidentally coordinates this zinc, and the Thr-2 side chain extends into the large specificity pocket of MMP-3. This unusual architecture of the interface between MMP-3 and TIMP-1 suggests new possibilities for designing TIMP variants and synthetic MMP inhibitors with potential therapeutic applications.

RBI, a one-domain alpha-amylase/trypsin inhibitor with completely independent binding sites.

The bifunctional inhibitor from Ragi (Eleusine coracana Gaertneri) (RBI) is the only member of the alpha-amylase/trypsin inhibitor family that inhibits both trypsin and alpha-amylase. Here, we show that both enzymes simultaneously and independently bind to RBI. The recently solved three-dimensional NMR structure of RBI has revealed that the inhibitor possesses a hitherto unknown fold for serine proteinase and alpha-amylase inhibitors. Despite its different fold, RBI obeys the standard mechanism observed for most protein inhibitors of serine proteinases and is a strong, competitive inhibitor of bovine trypsin (Ki = 1.2 +/- 0.2 nM). RBI is also a competitive inhibitor of porcine alpha-amylase (Ki = 11 +/- 2 nM) when a disaccharide is used as a substrate of alpha-amylase. However, the inhibition mode becomes complex when larger (> or = 7 saccharide units) alpha-amylase substrates are used. A second saccharide binding site on porcine alpha-amylase may enable larger oligosaccharides to displace RBI from its binding site in an intramolecular reaction.

Time-resolved fluorescence studies of tomaymycin bonding to synthetic DNAs.

Tomaymycin reacts covalently with guanine in the DNA minor groove, exhibiting considerable specificity for the flanking bases. The sequence dependence of tomaymycin bonding to DNA was investigated in synthetic DNA oligomers and polymers. The maximum extent of bonding to DNA is greater for homopurine and natural DNA sequences than for alternating purine-pyrimidine sequences. Saturation of DNA with tomaymycin has little effect on the melting temperature in the absence of unbound drug. Fluorescence lifetimes were measured for DNA adducts at seven of the ten unique trinucleotide bonding sites. Most of the adducts had two fluorescence lifetimes, representing two of the four possible binding modes. The lifetimes cluster around 2-3 ns and 5-7 ns; the longer lifetime is the major component for most bonding sites. The two lifetime classes were assigned to R and S diastereomeric adducts by comparison with previous NMR results for oligomer adducts. The lifetime difference between binding modes is interpreted in terms of an anomeric effect on the excited-state proton transfer reaction that quenches tomaymycin fluorescence. Bonding kinetics of polymer adducts were monitored by fluorescence lifetime measurements. Rates of adduct formation vary by two orders of magnitude with poly(dA-dG).poly(dC-dT), reacting the fastest at 4 x 10(-2) M-1 s-1. The sequence specificity of tomaymycin is discussed in light of these findings and other reports in the literature.

Determination of the three-dimensional structure of the bifunctional alpha-amylase/trypsin inhibitor from ragi seeds by NMR spectroscopy.

The three-dimensional structure of the bifunctional alpha-amylase/trypsin inhibitor (RBI) from seeds of ragi (Eleusine coracana Gaertneri) has been determined in solution using multidimensional 1H and 15N NMR spectroscopy. The inhibitor consists of 122 amino acids, with 5 disulfide bridges, and belongs to the plant alpha-amylase/trypsin inhibitor family for which no three-dimensional structures have yet been available. The structure of the inhibitor was determined on the basis of 1131 interresidue interproton distance constraints derived from nuclear Overhauser enhancement measurements and 52 phi angles, supplemented by 9 psi and 51 chi 1 angles. RBI consists of a globular four-helix motif with a simple "up-and-down" topology. The helices are between residues 18-29, 37-51, 58-65, and 87-94. A fragment from Val 67 to Ser 69 and Gln 73 to Glu 75 forms an antiparallel beta-sheet. The fold of RBI represents a new motif among the serine proteinase inhibitors. The trypsin binding loop of RBI adopts the "canonical", substrate-like conformation which is highly conserved among serine proteinase inhibitors. The binding loop is stabilized by the two adjacent alpha-helices 1 and 2. This motif is also novel and not found in known structures of serine proteinase inhibitors. The three-dimensional structure of RBI together with biochemical data suggests the location of the alpha-amylase binding site on the face of the molecule opposite to the site of the trypsin binding loop. The RBI fold should be general for all members of the RBI family because conserved residues among the members of the family from the core of the structure.

Efficient catalysis of disulfide formation during protein folding with a single active-site cysteine.

Protein disulfide isomerases (PDIs) catalyze disulfide bond formation during protein folding in vivo and are essential for viability in eukaryotic cells. They share the active-site sequence C-X-X-C that forms a catalytic disulfide. The recent finding that the EUG1 protein, a PDI-related yeast protein, with C-X-X-S sequence at its active sites can complement PDI-deficiency raised the general question of whether disulfide-isomerase activity is essential for cell viability or whether PDI variants with single active-site thiol groups can be catalytically active as disulfide isomerases. We investigated the function of the catalytic cysteine residues in DsbA, a PDI-related protein required for disulfide formation in the periplasmic space of Escherichia coli, by replacing C30 and C33 with alanine. While the mutant C30A and the double mutant CC30/33AA are inactive, C33A catalyzes disulfide-interchange reactions and oxidative renaturation of the reduced, unfolded thrombin inhibitor hirudin with close to wild-type efficiency. Thus, the single active-site thiol group of C30 is sufficient for disulfide-isomerase activity of the DsbA protein.

Time-resolved fluorescence studies of tomaymycin bonding to DNA.

Tomaymycin is an antibiotic that reacts at guanine N2 in the minor groove of the DNA helix. The number and type of tomaymycin-DNA adducts present on natural sequence DNA were identified using time-resolved fluorescence spectroscopy. At low bonding density, only two discrete species were observed with lifetimes of 4.3 and 7.1 ns and relative amplitudes of 40% and 60%. These two lifetime species are proposed to represent either R5' or S5' and S3' binding modes at the preferred bonding sequence 5'-AGA. R and S denote the configuration at C11 of tomaymycin, and 5' and 3' describe the orientation of the aromatic ring on the covalently modified strand. These two species were present over a range of solution conditions, including pH, nucleotide to drug ratio, DNA concentration, and DNA size. They have the same emission spectra, but slightly shifted absorption spectra. The weak temperature dependence of the fluorescence lifetimes presumably is due to the excited-state proton-transfer reaction that quenches tomaymycin fluorescence. The rate of formation of the longer lifetime species of DNA adduct is about twice as fast as that of the shorter lifetime species. Under saturating conditions, the fluorescence decay shows a bimodal lifetime distribution whether analyzed by least-squares assuming a Gaussian distribution model or by the maximum entropy method. The two groups of lifetimes are centered around 2-3 and 6-6.6 ns, reflecting multiple species on different bonding sequences.

NMR study of G.A and A.A pairing in (dGCGAATAAGCG)2.

One- and two-dimensional NMR, UV absorption experiments, and molecular mechanics calculations were conducted on an oligonucleotide duplex (dGCGAATAAGCG)2 which will be referred to as the T-11-mer. This oligonucleotide forms a duplex that is primarily B-form and contains two adjacent G.A and A.A base pairs and two 3' unpaired guanosines. The adjacent mismatch base pairs have an unusual structure which includes overwinding the helix and stacking with the base from the complementary strand (A4 with A8 and G3 with A7) instead of stacking with the base which is sequential on the strand. The exchangeable and nonexchangeable proton NMR spectra of the duplex have been characterized in H2O and D2O solution at neutral and acidic pH. The duplex is stabilized upon protonation; however, no additional hydrogen bonds are formed. We have observed the amino protons of adenosines A4 and A8 and guanosine G3 as a function of temperature and pH. These amino protons are involved in hydrogen bonds with the purine N3 or N7 acting as acceptors. Through the observation of a variety of NOE signals, the structure of the G.A and A.A mismatch base pairs has been defined.