Sequencespecific Assignment Using Homonuclear 2d Spectra

Figure 12.21 shows the basic strategy for sequential assignment. Consider a sequential pair of amino acid residues: serine followed by valine (S-V). In the TOCSY spectrum, we see the characteristic patterns of crosspeaks on the F2 = HN vertical lines for the two residues. Because both patterns are unique, we know that the one on the left side is a valine residue and the one on the right side is a serine. In the DQF-COSY spectrum, we see only the Ha crosspeaks on the vertical lines corresponding to the HN chemical shifts in F2 because COSY mixing involves only a single "jump" of INEPT transfer via the Hn-H^ J coupling. This region (F2 = HN = 7-10 ppm, F1 = Ha = 3-5 ppm) is called the "fingerprint region" of the COSY spectrum and can be used to count up the number of crosspeaks, which should be equal to one fewer than the number of amino acid residues (not counting prolines). The first residue (N-terminal) will not show up because the amino-terminus is a protonated amine (H3N+) rather than an amide (HN-CO) and is exchanging with water far too rapidly to be observed. The COSY crosspeaks have a fine structure that is an antiphase doublet in the F2 dimension because the HN proton is only coupled to the single Ha proton (except for glycine residues where there are two Ha protons). Each crosspeak can be analyzed by curve fitting to extract the HN-Ha coupling constant from the antiphase F2 slice (Fig. 12.21, lower left). Usually the raw 2D data is reprocessed by zero filling and cutting out the relevant regions

Figure 12.21

of the spectrum (HN region in F2 and Ha region in Fi) to generate a 2D matrix including only the fingerprint region, with much greater digital resolution. Accurate values for the HN-Ha 3 J couplings can be used to determine the dihedral angle (related to the $ angle in biochemistry) defined by the path H-N-Ca-Ha. This angle, along with the N-Ca-CO-N or ^ dihedral angle, defines the conformation of the polypeptide backbone and is a crucial input for NMR structure calculations.

TheNOESY spectrum (Fig. 12.21, right) gives the sequential connectivity, the proof that the Ser residue on the right side is followed in the primary sequence by the Val residue on the left side. On the vertical HN = Val line we see the intraresidue crosspeaks to the Ha and Hg of valine (which also appear in the TOCSY spectrum), but there is a new crosspeak that connects the Ha of the Ser spin system (in F1) with the HN of the Val spin system (in F2). This crosspeak lines up with the intraresidue crosspeak on the serine HN line (HN = Ser in F2 and Ha = Ser in F1) that is in the exact position of a crosspeak in the DQF-COSY spectrum. These intraresidue HN-Ha crosspeaks may be weak or missing in the NOESY spectrum, so it may be necessary to mark the position of the COSY crosspeaks in the NOESY spectrum. With modern NMR software this is usually done by using correlated cursors (crosshairs) displayed in the separate COSY and NOESY spectra. In Figure 12.21 (right side), we also see sequential crosspeaks on the vertical HN = Val line corresponding to the j and j protons of the preceding (Ser) residue. These j,N NOE crosspeaks are very useful in confirming the sequential connection, especially if the Ha chemical shift in F1 falls in a crowded region, overlapped with the Ha shifts of other residues. Finally, the N,N region of the NOESY spectrum, near the diagonal, can be searched for sequential connectivities (Fig. 12.21, lower right). Unlike the a,N and j,N correlations, the HN ^ HN connections work in both directions, i to i + 1 and i to i - 1. These will be the primary connections for a-helical regions of the protein, but will be weak for the j-sheet regions. Because the crosspeaks in a NOESY spectrum are weak relative to the diagonal peaks (inefficient magnetization transfer), it may be difficult to see the N,N crosspeaks if they are close to the diagonal.

In the real world, we do not know beforehand which spin system preceeds the valine spin system shown in the figure. There may be a number of Val residues in the protein, so we do not even know which kind of spin system to look for. What we have to do is to identify the NOE crosspeaks on the valine HN vertical line in the Ha region that are not found in the TOCSY spectrum. These have to be interresidue NOE crosspeaks. Then we search the COSY spectrum to find a crosspeak with the exact F1 chemical shift of the interresidue NOE peak on the valine HN line. There may be more than one candidate due to overlap, and we can rule some of them out because we know the amino acid sequence. For example, if none of the valine residues are preceded by an AMX residue in the amino acid sequence, we can rule out any AMX residue even if it has the right Ha chemical shift. To find the spin system following the valine, locate the intraresidue Ha crosspeak on the valine HN line and search horizontally for an NOE crosspeak at the same level (same F1 shift) that is not found in the TOCSY spectrum. This sequential crosspeak will occur at the HN chemical shift (in F2) of the next residue in the sequence. The process will inevitably reach a dead end at some point—especially when you get to a proline residue in the sequence (no HN). But even before that you may get stuck because of overlap and ambiguity. There are many starting and stopping points in the sequence and eventually when enough spin systems are assigned you can assign others by process of elimination.

To illustrate the process with real data, we will assign three segments of Heregulin-a EGF domain using sequential a,N correlations. Figure 12.22 shows a portion of the 70-ms TOCSY spectrum of the protein in 90% H2O. The most downfield-shifted HN resonances (probably in a ^-sheet) can be easily classified: AMX, Lys, five-spin, Leu, Thr, AMX, Val, AMX, and so on, moving from left to right. Figure 12.23 shows the fingerprint region of the 150-ms jump-return NOESY spectrum. The positions of selected crosspeaks in the

H20 TOCSY 500 MHz

H20 TOCSY 500 MHz

1 1 1 l 1 1 1 1 l 1 1 1 1 l 1 1 1 1 ^o 9.5 9.0 8.5 8.0

1 1 1 l 1 1 1 1 l 1 1 1 1 l 1 1 1 1 ^o 9.5 9.0 8.5 8.0

Figure 12.23

—i—i—i—i—i—i—i—i—i—i—i—i—i—i—i—i—i—i— 9.5 9.0 8.5

Figure 12.23

DQF-COSY spectrum are marked as squares. Starting at the position of the COSY crosspeak for the leucine residue identified in the TOCSY (Fig. 12.22, HN = 9.44 ppm) we move to the right to a "fat" sequential crosspeak at HN = 8.84 ppm. This HN shift corresponds to an AMX spin system in the TOCSY spectrum. Moving down to the position of the COSY peak for this AMX system (square: Ha = 5.69), we start the search again horizontally for another sequential NOE crosspeak. This is found on the left side at HN = 9.48 ppm, corresponding to a lysine spin system in the TOCSY spectrum. Moving up to the Ha position of this system (square: Ha = 4.69 ppm), we set off again horizontally looking for another nice, fat, well-resolved sequential crosspeak. Here we run into some regions of overlap and the going gets rough. So far we have the sequence: Leu-AMX-Lys. Searching the amino acid sequence (Fig. 12.14), there are four sequences that start with leucine: LVK, LSN, LCK, and LY. Only LCK (Leu-Cys-Lys) fits the pattern Leu-AMX-Lys, so we can assign this to the region 33-34-35 in the sequence, which is followed by Cys-36 and Gln-37. An AMX system at HN = 9.04 has a "stretched" crosspeak corresponding to the sequential peak at Ha = 4.69 and the intraresidue peak (square) below it at Ha = 4.76 ("AMX"). Moving to the right from this square we come to an overlapped but strong sequential crosspeak at Hn = 8.68, corresponding to a five-spin system overlapped with an Ala. Because we are expecting Gln-37, we choose the five-spin system. This completes the assignments for the stretch LCKCQ from residue 33 to residue 37.

Figure 12.24 shows the same region of the NOESY spectrum with these assignments written in. Now we begin another "walk," starting with the well-resolved valine residue in the TOCSY at HN = 8.49 ppm (Fig. 12.22, top right) and working backward in the sequence. On this vertical line in the NOESY spectrum (Fig. 12.24) there is a lovely sequential peak all alone at Ha = 5.08 ppm, corresponding to the Ha chemical shift of a five-spin system with Hn = 9.45 ppm. At the position corresponding to the COSY crosspeak of this five-spin system (square, left side) there is a "stretched" crosspeak with the interresidue peak at Ha = 5.08 and the sequential peak at Ha = 5.07 ppm, corresponding to an AMX spin system with Hn = 9.68 ppm. Checking the amino acid sequence, there are five Val residues contained

Figure 12.24

Figure 12.24

in the tripeptides HLV, FCV, FMV, ENV, and MKV. Only the tripeptide FMV corresponds to the sequence AMX-five-spin-Val found in the NOESY, so we can assign this to Phe-21, Met-22, and Val-23. From the AMX (F21) at HN = 9.68, Ha = 5.07 we can move backward (down) along the HN = 9.68 line to a big sequential crosspeak at Ha = 5.20 ppm, another AMX system with HN = 8.43 ppm (moving to the square on the right side). The sequence tells us that this corresponds to Cys-20. Moving up along the HN = 8.43 line leads to a sequential crosspeak at Ha = 4.43, corresponding to a five-spin system (HN = 8.55) that we can assign to Glu-19.

A short, three-residue walk through the NOESY spectrum is shown in Figure 12.25. Although there is a significant overlap in this region, two prominent sequential NOE peaks connect a Leu to a Val (Ha (Leu) = 4.89 to HN (Val) = 8.88) and the same Val to a Lys

system (Ha (Val) = 4.58 to HN (Lys) = 8.64). The dipeptide LV occurs only once in the sequence, followed by lysine: LVK at position L3-V4-K5.

Figure 12.26 shows the TOCSY spectrum with all of the sequence-specific assignments indicated. There are some horrendously overlapped regions in the HN and Ha regions, and sorting all of this out requires some special tricks. One of these is to vary the temperature. The main effect of changing the sample temperature is to move the water peak because the H2O chemical shift is temperature-dependent due to the change in the extent of hydrogen bonding. The chemical shift of water changes with temperature according to the formula

This means that the water chemical shift moves downfield by 1/96.9 or 0.0103 ppm, which is 10.3 parts per billion (ppb) with every decrease in temperature of 1 °C. The HN chemical shift "contains" a certain contribution due to the H2O chemical shift because of exchange with water: fast exchange leads to a chemical shift that is the weighted average of the chemical shifts experienced by the spin over time. Some Hn protons in the protein exchange rapidly with water because they are "exposed" to solvent on the surface of the protein; others exchange slowly because they are "buried" in hydrophobic regions and tied up in stable hydrogen bonds to protein groups such as backbone carbonyls. If an HN proton is solvent exposed, it will exchange with water rapidly and its chemical shift will show a large down-field shift (6-8 ppb/°C) as temperature is decreased. A "protected" HN proton, however, will experience much lower temperature shifts, less than 4 ppb/°C. If an overlapped region has a mixture of exposed and protected HN resonances, changing the temperature will change the

Figure 12.27

Figure 12.27

relative HN chemical shifts in a way that may be very helpful in sorting out the assignments. Figure 12.27 shows a small portion of the TOCSY spectrum of Heregulin-a EGF domain at 20 0C (left) and 30 0C (right) with the assignments indicated. The HN of E47, for example, moves only slightly (it is part of a short ^-sheet) but the HN of M51 moves almost 0.10 ppm (it is in the disordered C-terminal region). With NOESY spectra at both temperatures many ambiguities in the sequential connectivities can be sorted out. The three HN resonances Q56, M51, and E60 are almost completely overlapped at 20 0C, but they are well separated at 30 0C. Keep in mind that a great deal of human effort and judgement is involved in data interpretation in protein NMR. A week of data acquisition can be followed by a year or two of tedious, eyesight-destroying analysis in front of a computer screen in a darkened room. After a while even raindrops on the car windshield begin to look like NOE crosspeaks!

0 0

Post a comment