This section describes how the assignment principles described under Assignment - Theory can be but put into practice using the CCCPNmr Analysis software. There are several ways in which triple resonance backbone assignment, in particular, can be approached in CCPNmr Analysis using more or less automated methods. Initially a more manual method will be described, as this makes it easier to understand the process of assignment for those who are new to protein NMR assignment. This is followed by an outline of the slightly more automated method using the Link Sequential Spin Systems function within CCPNmr Analysis.
This is an old webpage which describes assignment using Analysis 1.0. A new description using Analysis 2.1 can be found here.
This section assumes that you have created a new project, provided details of the Molecular System and read in your spectra. See CCPNmrAnalysis Basics for details on how to do this, as well as how to change the appearance of your spectra and windows, use the mouse and navigate around your spectra.
By initialising the HSQC you select all its peaks and then create new resonances for each peak. In addition, resonances from backbone peaks are grouped into spin systems and their atom types are also assigned.
Begin by peakpicking your HSQC. The easiest way to do this is to zoom out at a fairly high contour level and then drag the mouse over all peaks while holding down Shift and Ctrl. Avoid picking too many artifacts by making sure that only positive peaks can be picked (this option can be selected under Peak Find Parameters in the Crosspeaks menu). You will probably still find that some small positive artifact peaks have been picked which you do not want - delete these by selecting them and pressing Delete. There will also be peaks which have not been picked correctly because of overlap - simply set a peak at the place where you want it by right-clicking the mouse and selecting Peak and then Add New Peak.
Once you are happy that you have picked all the peaks you want, go to Initialise Root Spectra in the Assignment menu. You will get the Initialise Peak Lists pop-up in which you should make sure that the correct peak list is selected (top left of the pop-up). When you click Initialise Peak List Analysis will assign two resonances to each peak - one in each dimension (resonance numbers are always surrounded by square brackets []). Because Analysis knows that this is an HSQC, it will also create a spin system for each peak, and both resonances from the peak will be added to it (spin system numbers are always surrounded by curly brackets {}). Furthermore, the hydrogen resonances will all be assigned the atom type HN and the nitrogen resonances the atom type N. While this is the correct way to treat the backbone peaks in an HSQC, this is not the correct treatment for side-chain peaks, in particular the Asn and Gln side-chain peaks. Here the atom types are different (Hδ, Hε, Nδ and Nε instead of HN and N) and because we are dealing with NH2 groups, there will be two peaks which belong to the same spin-system and indeed will have the same resonance in the nitrogen dimension. Analysis therefore helps you identify the Asn and Gln side-chain peaks, so as initialise them seperately, and only assign resonances to them, not spin systems and atom types.
The Amide Side Chain Peaks table in the Initialise Peak List pop-up lists all the pairs of peaks that Analysis can find which have a nitrogen chemical shift within a certain tolerence of one another. The size of this tolerance can be set in the top right hand box of the Initialise Peak List pop-up. When you click on one of the rows in the Amide Side Chain Peaks table, Anaylsis will automatically navigate to that pair of peaks and mark them. The window used for this can be selected from the drop-down menu in the top part of the pop-up.
If a given row has (in your opinion) correctly found an amide side-chain peak pair, then either click Confirm Selected or double-click the cell in the Confirmed? column. This will now switch from No to Yes. Check all the rows. If this has not identified all the amide side-chain peaks, then increase the tolerance (top right hand corner of the pop-up) and click Refresh Table to update it until all pairs are picked up. Once all amide side-chain peaks are confirmed, the initialisation can be done by clicking Initilise Peak List!.
All the peaks in a 3D spectrum should in theory be arranged in columns above the peaks of the 2D root spectrum it derives from (see Visualising 3D spectra for an explanation). Therefore the fastest way of peak-picking a 3D (and also the best way of avoiding picking too much noise) is to pick peaks only in the columns above the peaks you have in your 2D root spectrum. So if, for instance, you are wanting to peak-pick an HNCO or a CBCACONNH spectrum, you would base it on the peaks in your HSQC spectrum. Analysis is then able not only to do the peak picking, but will also assign the nitrogen and hydrogen resonances and atom types from the root HSQC peaks to CBCACONNH peaks.
Before you peakpick each 3D spectrum it is usually wise to save your project. The first time you pick a spectrum you may not know exactly what tolerances and contour heights to use. So if you then end up picking too few or too many peaks, the easiest way to start again, is to close and reopen your project without saving. Set your peak picking parameters in the Crosspeaks / Peak Find Parameters pop-up. Make sure that you have selected positive and negative or positive only depending on the peaks that you want to pick in your 3D. Scale relative to contour levels is probably set to 1.0, so you should now set your contour level such, that you only see (and therefore pick) real peaks and not noise. Later you may want to lower the contour level again, but for the main peak picking operation it is usually useful to have the contour level set relatively high.
To peakpick your 3D CBCACONNH spectrum go to the Macro menu and select pickAssignSpecFromRoot. A pop-up will appear in which you need to select your root spectrum, i.e. in this case your HSQC. Following this, a couple of pop-ups will appear asking you to specify the 1H and 15N tolerances - this specifies is the tolerance around your HSQC peak for the column in which the program will look for peaks in the 3D. You can simply go with the suggested tolerances to start with and then alter these later, if they don't seem appropriate. Finally you will be asked to select which 3D spectrum in which you would like to pick the peaks. Analysis will then pick each column of peaks and display a message in the terminal window to this effect. When the peak picking is finished, a pop-up tells you how many peaks have been picked. You should have a reasonlby good idea of many peaks should have been picked. If you have 100 peaks in your HSQC, then the HNCO should also contain 100 peaks (1 peak in each strip corresponding to the CO of the previous residue). The CBCACONNH should contain around 200 (2 peaks in each strip corresponding to the Cα and Cβ of the previous residue) and the CBCANNH should contain about 400 (4 peaks in each strip - the Cα and Cβ of the own residue plus the Cα and Cβ of the previous residue). If the number of peaks picked differs significantly from what your expect, your tolerances or contour level are probably too high or low. Simply close your project without saving and have another go with different parameters.
When the peak picking is done you will find that the atom type information as well as the resonance and spin system numbers have been transferred from the HSQC onto the 3D. This means you are only left with having to assign the carbon dimension of the 3D peaks which will speed things up considerably.
You will almost certainly find that some strips will be stronger than others. Even when you have set all the peak picking parameters as sensibly as you can, you are likely to end up with some strips that contain a large degree of noise and where additional peaks have been picked. Don't worry about this - simply delete such peaks when you come across them.
There may be cases where two peaks are very close to one another in the HSQC and within the tolerances you set for the 3D peak picking. In this case you will probably find that that all 3D peaks have been assigned the same resonance and spin system information. If you come across such strips in your 3D, you will need to compare the spectrum carefully with your HSQC and work out which 3D peak belongs to which root peak in the HSQC. Then the assignments on the 3D peak will need to be changed.
In order to do this, you will need to call up the assignment panel. Simply place the mouse over the required peak and press a. This way the assignment panel will immediately show the assignment for that peak.
To change the assignment, in this case from resonances {72}H[143] and {72}N[144] to {155}H[313] and {155}N[314], select the {72}H[143] resonance on the left and click Clear contrib to remove it from that peak. The new resonance {155}H[313] can be assigned to the peak by clicking on it on the right hand side of the panel. The same process will then have to be repeated for the nitrogen dimension.
Once you are happy with the peak picking for each 3D spectrum, save the project and peak pick the next spectrum.
The description here assumes that the backbone assignment will be carried out using CBCACONNH and CBCANNH spectra. Many of the steps are the same if using HNCA, HN(CO)CA, HNCO and HN(CA)CO spectra, and differences are highlighted in a section below.
Now comes the actual assignment part which follows the process outlined in the theory section. This section shows how to do this manually and in so doing goes through each step one at a time. Once you feel comfortable with the general process you can speed things up by using the semi-automated Link Sequential Spin Systems function. This simply condences several steps into one, thus speeding the process up. The flow diagram below shows the different steps that need to be taken.
1. Find a peak in the HSQC to be used as a starting point
Begin by selecting a peak in the HSQC which looks nice and is not overlapped.
2. Navigate to this NH position in the CBCANH
Navigate from here to your CBCANNH by right clicking the mouse and selecting Navigate and 1H - 15N in windowX (where windowX is the window in which you have the CBCANNH). The strip should contain two Cα peaks and two Cβ peaks.
3. Identify the own and previous Cα/Cβ pairs by comparison with the CBCA(CO)NH
If you overlay the CBCACONNH, then one of the Cα and one of the Cβ peaks (generally the weaker ones) should be present in the CBCACONNH. These are the Cα and Cβ of the previous residue, the other two are the Cα and Cβ of the same residue as the NH group. The two carbons belonging to the same residue as the NH group can be assigned.
4. Assign the own Cα/Cβ
The assignment is done using the assignment panel (press a when the mouse is over the peak to be assigned). Select New for the carbon dimension. This will create a new carbon resonance and assign it to this peak. Since it belongs to the same spin system as the N and H resonances, click on Set Same Spin System, so as to add the new carbon resonance the NH spin system. Now assign the atom type (Cα or Cβ) by clicking Set Atom Type. This will bring up the Browse Atoms panel which will initially not show any atoms at all. Click on C to toggle the carbon atoms on and select any Ca (or Cb) to set the atom type.
5. Select the own Cα/Cβ and search for matching peak pairs in the CBCA(CO)NH
Now select your assigned Cα and Cβ peaks (drag the mouse over them together or individually while holding down Shift) and to make things easier, place a mark through each peak (hold the mouse of the peak and press m). Then hold the mouse over one of the two peaks, right click and select Peak, Match Peaks, In CBCACONNH and F3 (make sure this is the carbon dimension!). Analysis will now look for strips in the CBCACONNH spectrum where there are peaks that match these two carbon dimensions. In this way you will find the NH which follows your current one in the sequence, since the CBCACONNH peaks are of the type Hi-Ni-Cαi-1/Cβi-1. Be aware that at the peak matching feature will only offer those spectra which are visible in your query window as possible options for matching! So if the spectrum in which you want to search for matches is not given as an option, make sure that it is 'switched on' and visible in the current window from where you are doing your matching.
Analysis will then bring up a new panel in which the Options are presented. The case below two matches were found which are ranked according to a scoring function. The top match has matches to both peaks, the lower one only to one peak. Thus the top match looks as though it is probably the correct one, though this should always be checked visually in the spectra as well (a peak found by analysis may for instance be noise and not a real signal, or if two peaks are overlapped the peak may not be placed correctly).
6. Display matching peaks in strips and select best match
To visually inspect your results, you can select as many of the possible matches as you like. Diplay Groups in Strips will then bring up these strips in whichever target window you have selected.
Once you have brought up your possible matches for visual inspection you may be lucky and find that there is quite definitely one match which is much better than the others.
7a. Good match: Assign the CBCA(CO)NH Cα/Cβ and set the NH as the sequential spin system
In this case you can now go ahead and assign the carbon dimensions in the matching strip of the CBCA(CO)NNH. Simply bring up the Assignment Panel for each peak (press m while the mouse is on the peak) and select the relevant carbon resonance.
To set the sequential spin system, select either peak, right click the mouse and go to Assign, Set Sequential Spin Systems, F1 0, F2 0, F3 -1. Analysis now knows that the spin system assigned to the F3 dimension of this peak belongs to the residue i-1 relative to the spin system assigned to the F1 and F2 dimensions.
Sometimes you will find tha there are two very good matches and it is not possible to tell which is correct. In this case simply make a note of the spin system involved and come back to this later when you have an idea of the position in the sequence and the amino acid types that are involved.
7b. No good match
If you are unlucky and the spectra are not very good, you may find that none of the possible matches found by Analysis match very well. Alternatively, the following amino acid may be a proline - because it has no H attached to its N, proline does not give any signals in CBCA(CO)NH spectra and there is no possible match to find. If cannot find a good match, simply make a note of the spin system you have investigated and try your luck with another by starting at the beginning again.
To use the Link Sequential Spin Systems function, select this from the Assignment menu and the Link Sequential Spin Systems panel will come up. First of all you need to select the window and spectrum from which you do your search, i.e. the CBCANNH. Then select the spectrum in which you want to find your matches, i.e. the CBCA(CO)NNH and the window in which you want the possible matches to be displayed.
When you select a spin system in the Spin System box, Analysis will automatically navigate to this position in your query spectrum (CBCANNH) in the window specified. It will also immediately look for matches to the peaks in that strip. These are then listed in the Matched Peaks box of the panel and are displayed as strips in the other window you selected above.
First of all go to the CBCANNH and assign the Cα and Cβ resonances which belong to that strip's own spin system (i.e. the strong peaks which do not overlap with the CBCA(CO)NNH). Simply press a when holding the mouse over each peak, add a New carbon resonance, then Set Same Spin System and Assign Atom Type.
Now you can check whether any of the matches found in the CBCA(CO)NNH fit to this resonance. In the example case, spin system {54} is the only one which matches and must therefore be the one which follows {88} in the sequence. In order to assign the carbon dimension of the {54} CBCA(CO)NNH peaks and set it as the spin system following {88}, simply select the {54} row in the Matched Peaks box and click Set Seq Link.
Now you can move on to the next spin system. Simply click Goto i+1 and Analysis will now set spin system {54} as the starting spin system from which to look for matches. It immediately navigates to strip {54} in the CBCANNH and shows the CBCA(CO)NNH matches. All you have to do is assign the {54} Cα and Cβ, select the match and move on. Simple! :)
If you cannot find a good match, or there are several that match equally well, just make a note of this and move on to another spin system (e.g. with Goto Next). Note that if you like, at the top of the Matched Peaks box, you can experiment with setting your Tolerance to be larger or smaller or with the number of strips to be displayed in the CBCA(CO)NNH window.
If you find that you have made a mistake at some point, simply select the spin system in question in the Spin Systems box and click Clear i-1 links, Clear i+1 links or Clear all links and then you can set new links again.
By linking sequential spin systems you will have ended up with a number of long strings of spin systems in their correct sequential order. Now you need to work out which part of the sequence they match up to. In order to do this you need to identify the amino acid type of some of your spin systems. For some amino acid types this is fairly straight forward. Glycine for instance, has a very characteristic Cα and no Cβ chemical shift. Alanine, Serine and Threonine all have very characteristc Cα and Cβ chemical shifts. You can see the distributions of chemical shifts by amino acid type as found in the BMRB within Analysis if you go to Molecules and Ref Chemical Shifts. The following pdf file has a visually very appealing way of displaying the average carbon chemical shifts. Using this you can quickly see that Isoleucine, Proline and Valine have lower Cα chemical shifts than all the other amino acids, for example.
Once you have identified (or excluded!) the amino acid type of a few of your spin systems in your linked stretch, you can start to compare this to your protein sequence. You may for instance have a stretch which is S-X-A-X-X-G, where X is, for example, anything other than A, S, G, T or P. If this motif only appears once in your sequence, then you can make a sequence specific assignment. If you like, you can immediately enter this into Analysis. But I tend to wait until I have assigned most of the sequence and am feeling fairly confident with my assignment before doing this, as it makes undoing mistakes easier. To enter your sequence specific assignment in Analysis, simply bring up the Edit Assignment panel for one of the peaks in one of the spin systems you want to assign. Select a resonance and click Assign {SpinSystem}ATOM[Resonance]. The Browse Atoms panel will appear and you now need to select the exact atom which this resonance corresponds to.
This atom will then be shaded in a slightly darker colour in the Browse Atoms panel to indicate that it is assigned. But you will find that there are lots of other atoms which will become shaded at the same time, because Analysis is able to assign them based on the assignment you have just made. For instance, if you assign the Cβ in one spin system, then Analysis will automatically be able to sequence-specifically assign the rest of that spin system, too. And if there is another spin system which has been set to be i-1 to this spin system, that is automatically assigned, too. So with one click, you can set the sequence specific assignment of a whole stretch of residues.
If you are using HNCA, HN(CO)CA, HN(CA)CO and HNCO spectra for your assignment the easiest thing is to use the Link Sequential Spin Systems function in a mannner very similar to that described above. When using the CBCANNH/CBCA(CO)NNH spectra the Cα and Cβ resonances are matched. When using the HNCA and HNCO based spectra, it is the Cα and CO resonances which are matched. The Cα and Cβ resonances occur at similar ppm values (about 75-15 ppm). However, the CO resonances have chemical shifts between about 170 and 180 ppm. For this reason, the best way to visualise the spectra is by adding a horizontal separator to your windows. This means that you can look at the HNCA and HNCO spectra in the same window rather than having to juggle additional windows. To add a horizontal separator, right click the mouse and go to Strip and Add horizontal separator. The spectrum will now be split in half and you can adjust the top half to coincide with the Cα chemical shift range and the bottom half with the CO chemical shift range.
Within the Link Sequential Spin Systems panel you will now have to set the HNCA and HN(CA)CO as your query spectra. In these spectra you will need to assign the Cα and CO resonances belonging to the query NH group. The matched spectra in which you look for your matches will be HN(CO)CA and HNCO spectra.
When selecting a spin system in the Link Sequential Spin Systems panel, you can match them in the normal way, but don't forget to use a horizontal separator in the window displaying the matched strips.
Assigning proteins which are only 15N-labelled is harder than assigning 15N,13C-labelled proteins. But for proteins of up to about 130 residues in length this is not an unreasonable task, especially, if the secondary structure and topology are known from a crystal structure or a homologous protein. The general principle of assignment is outlined in the theory section. There is no set way of doing this kind of assignment in Analysis, but several useful moves and strategies are outlined below.
In your 15N-NOESY-HSQC spectrum you should be able to see NOEs between neighbouring NH groups. There are several ways of identifying neighbouring residues. To begin with, select a peak in your HSQC with which you want to start and navigate to that position in the 15N-NOESY-HSQC (right-click the mouse and go to Navigate and windowX where windowX is the window in which you have your 15N-NOESY-HSQC).
With any luck there should be two or more NOEs in this strip to other NH groups. Two should be to the neighbouring residues (i.e. NOEs to NHi ± 1) and additional ones will be medium or long-range NOES. In α-helical secondary structure you may well see NOEs to NHi ± 2 and NHi ± 3. In β-sheet secondary structure you may see long-range NOEs to the neighbouring strand which can be considerably stronger than the NHi ± 1 NOEs.
In principle you want to find the correct spin system that each NOE originates from. In some cases it may be possible to identify the spin system unambiguously because the hydrogen chemical shift is unique to one spin system. However, in many cases there will be several options. One way to know that you have found the correct spin system, is by looking for the symmetrical NOE peak. If spin system i has an NOE to spin system j, then j should also have one to i. When you place the strips alongside one another this makes a nice rectangular pattern.
Note that in the region between about 6.5 and 7.5 ppm you can not only observe NOEs from NH groups, but also from certain side chains such as Phe, Tyr, Trp, Asn or Gln. If your protein contains Trp residues, you may also be able to observe NOEs to the Trp side-chain Hε 1 nuclei between about 9.0 and 11.0 ppm. The Asn, Gln and Trp Hε 1 resonances are also visible in your HSQC, however the Phe, Tyr and other Trp resonances are not.
It can be difficult working out which NOEs are from the neighbouring residues and which are from others. And of course when you have found out which are the neighbouring residues, you then still need to determine which is on the N-terminal side and which on the C-terminal side of your central residue.
On the whole the NOEs to neighbours will be the stronger ones, and they should be of comparable strength to one another. The main exception to this, is the cross-strand NOE in an antiparallel β-sheet which is stronger than the Ni± 1 NOEs. NOE intensity scales with the distance between the hydrogen atoms (as r-6). Below are some typical distances found in seconary structure (taken from Wüthrich (1986) NMR of Proteins and Nucleic Acids, WileyBlackwell).
| i, i ± 1 | i, i ± 2 | i, j | |
| α-helix | 2.8 Å | 4.2 Å | |
| β-sheet (parallel) | 4.3 Å | 4.8 Å | |
| β-sheet (antiparallel) | 4.2 Å | 3.3 Å |
You now have several options of how to identify which are your neighbouring strips. One option is to use the Match Peaks function. An alternative would be to select the assignment panel for each NOE peak (hold the mouse over the peak and press a) and then make a note of each of the spin systems that is an option for that NOE. Then you can line their strips up using the precedure outlined below and look at the matches.
The intensity of peaks varies with their mobility. Sections of the protein in secondary structure and will be fairly rigid. They should all show similar, reasonably strong, peak intensities. Slightly mobile regions, such as loop regions tend to show reduced peak intensities and in the worst case scenario are simply not visible at all. Very highly flexible regions, such as the C-terminus show very high peak intensities. Peak intensities will tend to go in waves along the backbone with neighbouring residues tending to have similar peak intensities.
The C-terminus can usually be identified fairly easily due to its strong peak in a very characteristic part of the HSQC spectrum (see here for a figure).
Serine peaks are often rather weak.
