Efficient identification of amino acid types for fast protein backbone assignments

We describe a procedure that allows for very efficient identification of amino acid types in proteins by selective ¹⁵N-labeling. The usefulness of selective incorporation of ¹⁵N-labeled amino acids into proteins for the backbone assignment has been recognized for several years. However, widespread use of this method has been hindered by the need to purify each selectively labeled sample and by the relatively high cost of labeling with ¹⁵N-labeled amino acids. Here we demonstrate that purification of the selectively ¹⁵N-labeled samples is not necessary and that background-free HSQC spectra containing only the peaks of the overexpressed heterologous protein can be obtained in crude lysates from as little as 100 ml cultures, thus saving time and money. This method can be used for fast and automated backbone assignment of proteins.     

New amino acid residue type identification experiments valid for protonated and deuterated proteins

Two experiments are presented that yield amino acid type identification of individual residues in a protein by editing the ¹H?C¹⁵N correlations into four different 2D subspectra, each corresponding to a different amino acid type class, and that can be applied to deuterated proteins. One experiment provides information on the amino acid type of the residue preceding the detected amide ¹H?C¹⁵N correlation, while the other gives information on the type of its own residue. Versions for protonated proteins are also presented, and in this case it is possible to classify the residues into six different classes. Both sequential and intraresidue experiments provide highly complementary information, greatly facilitating the assignment of protein resonances. The experiments will also assist in transferring the assignment of a protein to the spectra obtained under different experimental conditions (e.g. temperature, pH, presence of ligands, cofactors, etc.).     

4D APSY-HBCB(CG)CDHD experiment for automated assignment of aromatic amino acid side chains in proteins

A four-dimensional (4D) APSY (automated projection spectroscopy)-HBCB(CG)CDHD experiment is presented. This 4D experiment correlates aromatic with aliphatic carbon and proton resonances from the same amino acid side chain of proteins in aqueous solution. It thus allows unambiguous sequence-specific assignment of aromatic amino acid ring signals based on backbone assignments. Compared to conventional 2D approaches, the inclusion of evolution periods on ¹H^β and ¹³C^δ efficiently removes overlaps, and provides two additional frequencies for consequent automated or manual matching. The experiment was successfully applied to three proteins with molecular weights from 6 to 13 kDa. For the complementation of the assignment of the aromatic resonances, TOCSY- or COSY-based versions of a 4D APSY-HCCH^aro sequence are proposed.     

Sequential resonance assignments in protein 1H nuclear magnetic resonance spectra: Glucagon bound to perdeuterated dodecylphosphocholine micelles

The assignment of the ¹H nuclear magnetic resonance spectrum of glucagon bound to perdeuterated dodecylphosphocholine micelles with the use of two-dimensional ¹H nuclear magnetic resonance techniques at 360 MHz is described. Sequential resonance assignments were obtained for all backbone and C^β protons except the N-terminal amino group and the amide proton of Ser2. The assignments of the non-labile amino acid side-chain protons are complete except for the γ-methylene protons of Gln20 and Gln24. These assignments provide a basis for the determination of the three-dimensional structure of lipid-bound glucagon.     

Amino acid selective unlabeling for sequence specific resonance assignments in proteins

Sequence specific resonance assignment constitutes an important step towards high-resolution structure determination of proteins by NMR and is aided by selective identification and assignment of amino acid types. The traditional approach to selective labeling yields only the chemical shifts of the particular amino acid being selected and does not help in establishing a link between adjacent residues along the polypeptide chain, which is important for sequential assignments. An alternative approach is the method of amino acid selective ‘unlabeling’ or reverse labeling, which involves selective unlabeling of specific amino acid types against a uniformly ¹³C/¹⁵N labeled background. Based on this method, we present a novel approach for sequential assignments in proteins. The method involves a new NMR experiment named, {¹²CO _i –¹⁵N _i+1}-filtered HSQC, which aids in linking the ¹H^N/¹⁵N resonances of the selectively unlabeled residue, i, and its C-terminal neighbor, i + 1, in HN-detected double and triple resonance spectra. This leads to the assignment of a tri-peptide segment from the knowledge of the amino acid types of residues: i − 1, i and i + 1, thereby speeding up the sequential assignment process. The method has the advantage of being relatively inexpensive, applicable to ²H labeled protein and can be coupled with cell-free synthesis and/or automated assignment approaches. A detailed survey involving unlabeling of different amino acid types individually or in pairs reveals that the proposed approach is also robust to misincorporation of ¹⁴N at undesired sites. Taken together, this study represents the first application of selective unlabeling for sequence specific resonance assignments and opens up new avenues to using this methodology in protein structural studies.     

Bridging the gap: A set of selective 1H-15N-correlations to link sequential neighbors of prolines

Triple-resonance experiments are standard in the assignment of protein spectra. Conventional assignment strategies use ¹H-¹⁵N-correlations as a starting point and therefore have problems when proline appears in the amino acid sequence, which lacks a signal in these correlations. Here we present a set of amino acid selective pulse sequences which provide the information to link the amino acid on either side of proline residues and thus complete the sequential assignment. The experiments yield amino acid type selective ¹H-¹⁵N-correlations which contain signals from the amino protons of the residues either preceding or following proline in the amino acid sequence. These protons are correlated with their own nitrogen or with that of the proline. The new experiments are recorded as two-dimensional experiments and their performance is demonstrated by application to a 115-residue protein domain.     

Sequential resonance assignments in protein 1H nuclear magnetic resonance spectra. Basic pancreatic trypsin inhibitor

The assignment of the ¹H nuclear magnetic resonance spectrum of the basic pancreatic trypsin inhibitor with the use of two-dimensional ¹H nuclear magnetic resonance techniques at 500 MHz is described. The assignments are based entirely on the known amino acid sequence and the nuclear magnetic resonance data. Individual resonance assignments were obtained for all backbone and C^β protons, with the exception of those of Arg1, Pro2, Pro13 and the amide proton of Gly37. The side-chain resonance assignments are complete, with the exception of Pro2 and Pro13, the N^δ protons of Asn44 and the peripheral protons of the lysine residues and all but two of the arginine residues.     

Extension of the HA-detection based approach: (HCA)CON(CA)H and (HCA)NCO(CA)H experiments for the main-chain assignment of intrinsically disordered proteins

Extensive resonance overlap exacerbates assignment of intrinsically disordered proteins (IDPs). This issue can be circumvented by utilizing ¹⁵N, ¹³C′ and ¹H^N spins, where the chemical shift dispersion is mainly dictated by the characteristics of consecutive amino acid residues. Especially ¹⁵N and ¹³C′ spins offer superior chemical shift dispersion in comparison to ¹³C^α and ¹³C^β spins. However, HN-detected experiments suffer from exchange broadening of amide proton signals on IDPs especially under alkali conditions. To that end, we propose here two novel HA-detected experiments, (HCA)CON(CA)H and (HCA)NCO(CA)H and a new assignment protocol based on panoply of unidirectional HA-detected experiments that enable robust backbone assignment of IDPs also at high pH. The new approach was tested at pH 6.5 and pH 8.5 on cancer/testis antigen CT16, a 110-residue IDP, and virtually complete backbone assignment of CT16 was obtained by employing the novel HA-detected experiments together with the previously introduced iH(CA)NCO scheme. Remarkably, also those 10 N-terminal residues that remained unassigned in our earlier HN-detection based assignment approach even at pH 6.5 were now readily assigned. Moreover, theoretical calculations and experimental results suggest that overall sensitivity of the new experiments is also applicable to small or medium sized globular proteins that require alkaline conditions.     

Prospects for resonance assignments in multidimensional solid-state NMR spectra of uniformly labeled proteins

Summary The feasibility of assigning the backbone ¹⁵N and ¹³C NMR chemical shifts in multidimensional magic angle spinning NMR spectra of uniformly isotopically labeled proteins and peptides in unoriented solid samples is assessed by means of numerical simulations. The goal of these simulations is to examine how the upper limit on the size of a peptide for which unique assignments can be made depends on the spectral resolution, i.e., the NMR line widths. Sets of simulated three-dimensional chemical shift correlation spectra for artificial peptides of varying length are constructed from published liquid-state NMR chemical shift data for ubiquitin, a well-characterized soluble protein. Resonance assignments consistent with these spectra to within the assumed spectral resolution are found by a numerical search algorithm. The dependence of the number of consistent assignments on the assumed spectral resolution and on the length of the peptide is reported. If only three-dimensional chemical shift correlation data for backbone ¹⁵N and ¹³C nuclei are used, and no residue-specific chemical shift information, information from amino acid side-chain signals, and proton chemical shift information are available, a spectral resolution of 1 ppm or less is generally required for a unique assignment of backbone chemical shifts for a peptide of 30 amino acid residues.     

Specific assignment of resonances in the 1H-NMR spectrum of the polypeptide toxin I from Anemonia sulcata

The assignment of a large number of resonances in the 300-MHz ¹H-nmr spectrum of the polypeptide neurotoxin Anemonia sulcata toxin I is described. The initial identification of spin systems is made using both one- and two-dimensional nmr spectra. The subsequent assignment of these spin systems to specific residues in the molecule is based largely on the observation in two-dimensional spectra of through-space connectivities between H^α and NH resonances from adjacent residues in the amino acid sequence. Using these techniques, the full spin systems of 22 residues are specifically assigned, together with partial assignments for a further 8. Many of the spin systems from the remaining 16 residues have been defined, although not yet specifically assigned. From the pattern of through-space connectivities between protons from adjacent residues in the sequence, some inferences may be drawn concerning the secondary structure of this polypeptide in aqueous solution.     

A general method for assigning NMR spectra of denatured proteins using 3D HC(CO)NH-TOCSY triple resonance experiments

Summary A general approach for assigning the resonances of uniformly ¹⁵N- and ¹³C-labeled proteins in their unfolded state is presented. The assignment approach takes advantage of the spectral dispersion of the amide nitrogen chemical shifts in denatured proteins by correlating side chain and backbone carbon and proton frequencies with the amide resonances of the same and adiacent residues. The ¹H resonances of the individual amino acid spin systems are correlated with their intraresidue amide in a 3D ¹⁵N-edited ¹H, ¹H-TOCSY-HSQC experiment, which allows the spin systems to be assigned to amino acid type. The spin systems are then linked to the adjacent i-1 spin system using the 3D H(C)(CO)NH-TOCSY experiment. Complete ¹³C assignments are obtained from the 3D (H)C(CO)NH-TOCSY experiment. Unlike other methods for assigning denatured proteins, this approach does not require previous knowledge of the native state assignments or specific interconversion rates between the native and denatured forms. The strategy is demonstrated by assigning the ¹H, ¹³C, and ¹⁵N resonances of the FK506 binding protein denatured in 6.3 M urea.     

Acceleration of protein backbone NMR assignment by combinatorial labeling: Application to a small molecule binding study

Selective labeling with stable isotopes has long been recognized as a valuable tool in protein NMR to alleviate signal overlap and sensitivity limitations. In this study, combinatorial ¹⁵N‐, ¹³C^α‐, and ¹³C'‐selective labeling has been used during the backbone assignment of human cyclophilin D to explore binding of an inhibitor molecule. Using a cell‐free expression system, a scheme that involves ¹⁵N, 1‐¹³C, 2‐¹³C, fully ¹⁵N/¹³C, and unlabeled amino acids was optimized to gain a maximum of assignment information from three samples. This scheme was combined with time‐shared triple‐resonance NMR experiments, which allows a fast and efficient backbone assignment by giving the unambiguous assignment of unique amino acid pairs in the protein, the identity of ambiguous pairs and information about all 19 non‐proline amino acid types. It is therefore well suited for binding studies where de novo assignments of amide ¹H and ¹⁵N resonances need to be obtained, even in cases where sensitivity is the limiting factor.     

Multidimensional1H and15N NMR investigation of glutamine-binding protein ofEscherichia coli

Summary Specific and uniform¹⁵N labelings along with site-directed mutagenesis of glutamine-binding protein have been utilized to obtain assignments of the His¹⁵⁶, Trp³² and Trp.²²⁰ residues. These assignments have been made not only to further study the importance of these 3 amino acid residues in protein-ligand and protein-protein interactions associated with the active transport ofl-glutamine across the cytoplasmic membrane ofEscherichia coli, but also to serve as the starting points in the sequence-specific backbone assignment. The assignment of H₂ of His¹⁵⁶ refines the earlier, model where this particular proton formas an intermolecular hydrogen bond to the -carbonyl ofl-glutamine, while assignments of both Trp³² and Trp²²⁰ show the variation in local structures which ensure the specificity in ligand binding and protein-protein interaction. Using 3D NOESY-HMQC NMR, amide connectivities can be traced along 8–9 amino acid residues at a time. This paper illustrates the usefulness of combining¹⁵N isotopic labeling and multinuclear, multidimensional NMR techniques for a structural investigation of a protein with a molecular weight of 25 000.     

Optimization of amino acid type-specific <Superscript>13</Superscript>C and <Superscript>15</Superscript>N labeling for the backbone assignment of membrane proteins by solution- and solid-state NMR with the UPLABEL algorithm

We present a computational method for finding optimal labeling patterns for the backbone assignment of membrane proteins and other large proteins that cannot be assigned by conventional strategies. Following the approach of Kainosho and Tsuji (Biochemistry 21:6273–6279 (1982)), types of amino acids are labeled with ¹³C or/and ¹⁵N such that cross peaks between ¹³CO(i – 1) and ¹⁵NH(i) result only for pairs of sequentially adjacent amino acids of which the first is labeled with ¹³C and the second with ¹⁵N. In this way, unambiguous sequence-specific assignments can be obtained for unique pairs of amino acids that occur exactly once in the sequence of the protein. To be practical, it is crucial to limit the number of differently labeled protein samples that have to be prepared while obtaining an optimal extent of labeled unique amino acid pairs. Our computer algorithm UPLABEL for optimal unique pair labeling, implemented in the program CYANA and in a standalone program, and also available through a web portal, uses combinatorial optimization to find for a given amino acid sequence labeling patterns that maximize the number of unique pair assignments with a minimal number of differently labeled protein samples. Various auxiliary conditions, including labeled amino acid availability and price, previously known partial assignments, and sequence regions of particular interest can be taken into account when determining optimal amino acid type-specific labeling patterns. The method is illustrated for the assignment of the human G-protein coupled receptor bradykinin B2 (B₂R) and applied as a starting point for the backbone assignment of the membrane protein proteorhodopsin.     

Towards structural investigations on isotope labelled native bacteriorhodopsin in detergent micelles by solution-state NMR spectroscopy

¹H NMR signals of the retinal moiety in detergent-solubilizedbacteriorhodopsin are assigned, enabling the interpretation of NOEs within thechromophore. To achieve this, a number of differently labelled samples wereprepared to test the applicability of the various assignment and distancemeasurement strategies. In measurements with and without light,¹H and ¹³C chemical shifts of the retinal in thenative protein were partially assigned for both the dark- and thelight-adapted states. Additionally, samples with residue-specific¹H amino acids and/or retinal in an otherwise deuterated proteinwere prepared to measure the distances between either two kinds of amino acidsor between individual amino acids and the retinal moiety. With the observationof NOE within the bound retinal and between retinal and its neighbouring aminoacids, an important step towards the elucidation of distance constraints inthe binding pocket of the proton pump is made.     

RESCUE: An artificial neural network tool for the NMR spectral assignment of proteins

The assignment of the ¹H spectrum of a protein or a polypeptide is the prerequisite for advanced NMR studies. We present here an assignment tool based on the artificial neural network technology, which determines the type of the amino acid from the chemical shift values observed in the ¹ H spectrum. Two artificial neural networks have been trained and extensively tested against a non-redundant subset of the BMRB chemical shift data bank [Seavey, B.R. et al. (1991) J. Biomol. NMR, 1, 217–236]. The most promising of the two accomplishes the analysis in two steps, grouping related amino acids together. It presents a mean rate of success above 80% on the test set. The second network tested separates down to the single amino acid; it presents a mean rate of success of 63%. This tool has been used to assist the manual assignment of peptides and proteins and can also be used as a block in an automated approach to assignment. The program has been called RESCUE and is made publicly available at the following URL: http://www.infobiosud.univ-montp1.fr/rescue.     

Analysis of the 1H-NMR spectra of ferrichrome peptides. I. The non-amide protons

The thus far unexplored aliphatic region of the proton magnetic resonance spectra of ferrichrome peptides was investigated at 360 MHz. Six isomorphic diamagnetic analogues of the ferric cyclohexapeptide differing in the coordinated cation (AL³⁺ or Ga³⁺) and the amino acid composition were studied in d6 -DMSO solution. By use of a novel resolution enhancement technique which applies a sinusoidal half-wave window to the free induction decay combined with multiplication by an increasing exponential, the proton chemical shifts and spin-spin couplings were accurately measured. Homonuclear decoupling combined with Fourier difference spectroscopy was used to selectively extract resonances out of crowded spectral regions. The spectra revealed unique features of fine structure in the proton resonance lines. Thus, the conformation-dependent geminal coupling constants of glycyl α-protons were found to be constant throughout the suite of analogue peptides. A similar invariance was observed for the vicinal coupling constants between α-, β-, γ-, and δ-protons in the ornithyl side chains. Comparison of the proton spin–spin coupling constants with the crstallographic dihedral angles led to a unique stereochemical assignment of the side-chain resonances. The combined data sets of x-ray atomic coordinates and ¹H-nmr spin-spin coupling parameters have been used to calibrate the coefficients for a Karplus curve related to the torsional x angles in amino acid side chains: Structurial information was also obtained for the seryl residues, where the multiplet structures of the OH resonances indicate preferred spatial arrangements of the side chains.     

Application of neural networks to automated assignment of NMR spectra of proteins

Summary Simulated neural networks are described which aid the assignment of protein NMR spectra. A network trained to recognize amino acid type from TOCSY data was trained on 148 assigned spin systems from E. coli acyl carrier proteins (ACPs) and tested on spin systems from spinach ACP, which has a 37% sequence homology with E. coli ACP and a similar secondary structure. The output unit corresponding to the correct amino acid is one of the four most activated units in 83% of the spin systems tested. The utility of this information is illustrated by a second network which uses a constraint satisfaction algorithm to find the best fit of the spin systems to the amino acid sequence. Application to a stretch of 20 amino acids in spinach ACP results in 75% correct sequential assignment. Since the output of the amino acid type identification network can be coupled with a variety of sequential assignment strategies, the approach offers substantial potential for expediting assignment of protein NMR spectra.     

Assignment of amide proton signals by combined evaluation of HN,NN and HNCA MAS-NMR correlation spectra

In this paper, we present a strategy for the ¹H^N resonance assignment in solid-state magic-angle spinning (MAS) NMR, using the -spectrin SH3 domain as an example. A novel 3D triple resonance experiment is presented that yields intraresidue H^N-N-C correlations, which was essential for the proton assignment. For the observable residues, 52 out of the 54 amide proton resonances were assigned from 2D (¹H-¹⁵N) and 3D (¹H-¹⁵N-¹³C) heteronuclear correlation spectra. It is demonstrated that proton-driven spin diffusion (PDSD) experiments recorded with long mixing times (4 s) are helpful for confirming the assignment of the protein backbone ¹⁵N resonances and as an aid in the amide proton assignment.     

An automated assignment-free Bayesian approach for accurately identifying proton contacts from NOESY data

The identification of proton contacts from NOE spectra remains the major bottleneck in NMR protein structure calculations. We describe an automated assignment-free system for deriving proton contact probabilities from NOESY peak lists that can be viewed as a quantitative extension of manual assignment techniques. Rather than assigning contacts to NOESY crosspeaks, a rigorous Bayesian methodology is used to transform initial proton contact probabilities derived from a set of 2992 protein structures into posterior probabilities using the observed crosspeaks as evidence. Given a target protein, the Bayesian approach is used to derive probabilities for all possible proton contacts. We evaluated the accuracy of this approach at predicting proton contacts on 60 ¹⁵N separated NOESY and ¹³C separated NOESY datasets simulated from experimentally determined NMR structures and compared it to CYANA, an established method for proton constraint assignment. On average, at the highest confidence level, our method accurately identifies 3.16/3.17 long range contacts per residue and 12.11/12.18 interresidue proton contacts per residue. These accuracies represent a significant increase over the performance of CYANA on the same data set. On a difficult real dataset that is publicly available, the coverage is lower but our method retains its advantage in accuracy over CANDID/CYANA. The algorithm is publicly available via the Protinfo NMR webserver .