Triple-resonance protein sequence assignment

If the program is not already open, start CcpNmr Analysis on the command line by typing:

-> analysis

We will now look at the sequentially assigning protein backbone spin systems using triple-resonance experiments. There are two basic, although linked, parts to the process. The first is the linking of sequential spin systems (collections of resonances that relate to one residue) on the basis of matching peak positions. The second is the matching of runs of unassigned spin systems to residues within a sequence.

Open an existing project

We will now leave the previous CCPN project behind and load a new one; this differs from the old one in that it has two more spectra; HNcoCaCB and HNCACB, and that all the peaks have been picked and linked to the amide 'root'; resonances as described above (i.e. using the Assignment:Pick & Assign from Roots option). In the Analysis menu bar select M:Project:Open Project. Select to close the exiting project, but there is no need to save. Navigate to find and select the CcpnCourse2a project, then click [Open].

Spectrum Setup

First make sure that window2 and window4 are visible and arranged such that window2 is tall and narrow and on the left of a tall and wide window4 (both windows should be as tall as possible). Both window2 and window4 have HCN axis and can display the triple resonance spectra. Note that to make a new window you could use either use  M:Window:New window or clone an existing window via R:Window:Clone. Using the "Spectra" tab in the windows ensure that the HNcoCA and HNcoCACB are the only spectra turned on in the narrow window and that all four triple resonance spectra are turned on in the wide window.

Select M:Assignment:Protein Sequence Assignment. You should start out in the 'Window & Spectra' tab. Select window2 as the 13C Window in the Query section and window4 as 13C Window in the Match section. Make sure that the "Use" column is set to "Yes" for the query HNcoCA and HNcoCACB spectra in the top table by double-clicking. For the bottom table set the "Use" column to "Yes" for only the HNCA and HNCACB spectra (i.e. not the through-carbonyl experiments). This setup means that we are going to compare specified 13C peak positions in the HNcoCA and HNcoCACB experiments with potentially matching peak positions in the HNCA and HNCACB experiments.

The rationale here is that the through-carbonyl experiment's peaks have the carbon shift of the preceding alpha and beta carbons along the polypeptide chain at a given amide location, where the HNCA & HNCACB have both the intra-residue and preceding alpha & beta carbon peaks for each amide. Thus we can potentially use both spectra to say two amide spin systems are sequentially connected; by saying that an inter-residue peak of the HNcoCA or HNcoCACB derives from the same resonance as an intra-residue peak of the HNCA or HNCACB.

Note that this system can readily use other backbone experiments like HNcaCO, HNCO, HAcacoNH, HAcaNH  etc. with the same approach and that this tutorial only uses the alpha & beta carbon experiments for simplicity.

Linking Sequential Spin Systems

Now go to the {Spin System Table} tab and in the to table scroll down and click on the row that corresponds to spin system 8. Enlarging the poup window to the full heigh of the screen will help view the tables more clearly. You will see that the two triple resonance windows move to new locations. The location of window2 is at spin system {8} and split into two regions one for the CA and the other for the CB peak. The other, window4, has moved to the position of any peaks that match the carbon frequencies of the query peaks. In this case there are three potential matches, but the first strip that corresponds to spin system {71} is the only one that matches both the CA and CB positions well. Note that if the "Filter 13C By Inter/Intra Type" option in the {Options} tab is set to off then spin system {8} matches even more strips. Because we already have a good match we don't really need this option on, but it is useful if the previous-residue and same-residue peaks overlap significantly in the match strips. In the case of spin system {8} matching {71} the CA position does not intersect the purple HNcoCA position, it clearly matches the separate orange HNCA peak (and similarly for the CB position).

In the {Spin System Table} tab, go to the middle "Match Peak Positions" table a click to highlight the row corresponding to spin system {71} (rank 1), and then click [Set Seq Link]. You will now see that the tables of the popup update to show that {71} is set as "i-1" of {8}. Also note that in the spectrum window the peak annotations of the aligned CA & CB peaks have changed to illustrate that they are both assigned to the same 13C resonances. Now click [Goto i-1], which will repeat the carbon shift matching, but this time from a sequence position one earlier in the sequence.  Repeat the above procedure for spin system {71}: select the best match and link. For this exercise go on to sequentially link six spin systems in total. If all goes according to plan, the order of spin systems (going i-1each time) will be {8} -  {71} -  {42} -  {37} -  {48} - {41}.

Assigning Segments to the Sequence

You will see that as you select the various spin systems in the main table, in the lower right Residue Types table there is a display of the probable types of amino acid residue. This prediction is based upon how well the shifts within the spin system match the chemical shifts in the RefDB database. As spin systems are connected sequentially, amino acid type predictions are made for the whole sequentially connected section. In the lower left hand table the connected spin systems, given their probable amino acid types, are matched to the protein sequence. Here the highest scoring positions of residue type match for various five residue sections are listed. You will see that the unassigned residue positions are coloured grey, and the one assigned regions become blue.

Click in the upper table on the row for spin system {37}, which is in the middle of our linked region, and you will see that the residue type predictions are strong at and either side of this position. The highest scoring option in the Sequence Locations table (hopefully with a score of about 66.7) should correspond to the region from residue 17 Gly to 21 Leu, with the other sequence locations having lesser scores. Simply select the row for this highest-scoring location and click [Assign Selected] and then [OK] to confirm the assignment. You will see that all the residues in the section (by virtue of their links for the most part) become assigned to the selected section, and that the colors in the 'Sequence Locations' table change.

Select the option M:Molecule: Atom Browser.  Make sure that the elements [N] and [H] are displayed (click the button to get the green hydrogen assignment options) and look at the amide atom for 19 Thr. You will see that not only is 19 Thr assigned (i.e. the atom option goes dark green), but the atoms in the residues which we just connected sequentially are also assigned. Go through the spectra and the Protein Sequence Assignment (with 'Goto i+1')  to verify that all the connected spin systems have been assigned. Note that in this instance it was possible to assign the resonances to unique atoms, as well as assign the backbone spin systems to the sequence, because the resonances had their atom type set previously.

Automatic Protein Backbone Assignment

The semi-automated sequence assignment mechanism described above is supplemented in CcpNmr Analysis by automatic assignment routines. At the moment only one called "Nexus" is publicly available, but MARS, AutoAssign and other routines will be incorporated in the future. The automated assignment routines are great for saving time, but they won't always work for all parts of proteins, especially where peaks are missing or severely overlapped. In such instances you can run the automation to assign the easy parts and then fill in the rest more carefully, where possible, using the more manual routines.

Fortunately the data in the example CCPN project being used here is pretty good, and the peaks positions have been checked and have undergone a degree of curation. Accordingly, we can assign most of pure protein sequences automatically with a couple of clicks. To do this open M:Assignment:Automated Seq. Assignment and in the resulting popup you will see that the four triple-resonance spectra are already selected; changing the "Use" column entry to "No" would mean the spectrum was not used for the assignment. All of these default settings are correct, so move on to the {Spin Systems} tab. In the spin systems table you will be presented with all of the resonances, available in the chosen spectra, that are linked to a backbone amide as either same-residue (Intra Residue) or previous-residue (Inter Residue). Initially the table is quite bare, with only a few resonances (and their chemical shift values) displayed for the few residues that we have already assigned. To fill in the rest of the table click [Find Resonances From Peaks], selecting [OK] at the confirmation dialogue box. This function will then automatically extract and label the CA and CB chemical shifts from the same- and previous-residue peaks. This is done by considering the overlap of the through-carbonyl spectra with other spectra, together with the peak intensities and relative chemical shifts that are expected to be visible (in an experiment of that type).

With the resonances populated in the spin systems table, move on to select the {Automation} tab. Here, set "Use existing assignments?" to on and accept the other default settings and click [Run Nexus]. Once the procedure has finished all of its iterations you will be shown a graph of the assignment scores for each residue; where blue means good, yellow dubious and red bad. Moving to the {Predictions} tab you will hopefully see large regions of good (blue) prediction. Before a spin system (and thus all of its contained resonances) can be assigned to the predicted residue, the prediction must be confirmed. To confirm the prediction either select a series of good rows (use <Crtl> + click in the table) and click [Confirm Selected] or set all blue regions en masse by clicking [Conform Above Score Threshold]. With your choices conformed click [Commit Assignments] and confirm [OK] to actually do the full assignment; which will affect the peak labels.

Look at the peaks in window1 to see the result of the assignment, and admire your handywork.

Finding Side Chain Resonances Using TOCSY Data

The next kind of resonance assignment that we will look as is for the assignment of amino acid side chain atoms. As an example we will choose a relatively simple residue, with relatively clear peaks, naturally some residues with obscures or weak peaks will be more difficult to assign than this one, but the general principles remain the same. Firstly we will load a new project with 3D TOCSY spectra loaded. Select to close the exiting project, but do not save, then navigate to find and select the CcpnCourse2b project and lastly click [Open]. This project contains an assigned HSQC spectrum, some tripe resonance backbone (HNC) experiments and three different kinds of three-dimensional TOCSY spectra.

In window2 a three-dimensional CCcoNH TOCSY spectrum is shown. Like the HNcoCA and HNcoCACB spectra, this shows peaks belonging to amide resonances which correlate with the carbon resonances of the previous residue. For example, below  we will look for the 20 Ala H,N (amide) position but the side chain peaks on the Y-axis will be for residue 19 Thr.

In window3 a three-dimensional 15N HSQC TOCSY is shown. This spectrum shows side chain hydrogen resonances on the Y-axis and amide (HSQC) resonances on the other axes. Here the amide resonances are from the same residue as the amides.

In window5 a three-dimensional HCcH TOCSY is shown. This spectrum has two hydrogen axes and one carbon axis. The hydrogen dimension on the X-axis represents resonances that are directly bound to the carbon resonances on the Z-axis (depth axis). This contracts with the other side chain spectra where the Z axis is for amide nitrogen. Thus here we are looking at CH pairs not NH pairs. The Y-axis of this spectrum represents hydrogen resonances that are in the same residue as the bound CH pair on the other axes.

In this example we will assign the side chain peaks for the 19 Thr residue, thus we expect to find threecarbon resonances; CA, CB, & CG2  and three hydrogen resonances; HA, HB, HG2* (methyl). To find the hydrogen and carbon resonances in the amide-linked spectra we will navigate from the HSQC peak positions. Accordingly, open M:Peak:Peak Lists, select the {Peak Table} tab and set the peak list pulldown menu to "HSQC:115:1". Now sort the HSQC peaks by sequence by clicking on one of the "Assign" column headings in the table. Scroll down so that peaks for 19Thr and 20Ala are visible. Select the 20Ala row, then above the table set pulldown menu next to "Go To Position" to "1H - 15N in window2" and click [Go To Position]. This will cause the display of window2 to move to the amide of 20Ala, but because the CCcoNH spectrum shows side chain resonances for the previous residue, the peaks will derive from 19Thr in their middle dimension. Go back to the peak list table and select the row for 19Thr, change the pulldown menu next to "Go To Position" to "1H - 15N in window3" and click [Go To Position]. This will show the amide and side chain peaks for 19Thr in the HSQC TOCSY spectrum.

With window2 and window3 located at the carbon and hydrogen side chain positions for 19Thr we next assign the non-amide (middle) dimension of all the relevant peaks. For the CCcoNH spectrum in window2 place the mouse over the peak labelled as "20Ala H,-,N" near 69 ppm on the Y-axis (carbon) and press <a> to open the Assignment Panel. This is the CB peak for 19Thr, so when the Assignment Panel opens, you will see "19ThrCb" as an assignment option in the middle table. Click on the "19ThrCb" row to assign this to the middle dimension of the peak. Note that this CB resonance exists and has a known chemical shift because if the assignments in the HNcoCACB and HNCACB backbone experiments. Repeat this procedure for the peak at 66 ppm ( press <a> over it) to give it a "19ThrCa" assignment in its middle dimension.

Assigning New Side Chain Resonances

For the last carbon position at 20 ppm we will make an assignment to the CG2 atom, however this resonance is new to the project, and hence there is no assignment suggestion that appears automatically for the middle dimension. Instead we will add a new resonance that we will link to the right atoms. To do this click on the middle [<New>] button (F2 dimension). You will see a new resonance number appear of the form [number] in the middle dimension. Click on this new resonance in the left most column, then click [Assign [number]] at the left of the blue buttons. In the Atom Browser that appears, switch on the hydrogen and carbon atoms using the[C] and [H] buttons. Then scroll down to the 19 Thr residue and click on the right most "Cg2" (the "g" will be a Greek gamma). Returning to the Assignment Panel you will now see that the resonance assigned to the middle dimension is labelled as "19ThrCg2".

Next move to window3 and assign the side chain peaks of the HSQC TOCSY spectrum in a similar way. In this case the peaks you need to consider will be initially labelled as "19Thr H,-,N" because the middle dimension will carry the hydrogen assignments for the same residue as the amide. There will be three peaks at 4.26, 3.82 and 1.21 ppm, these correspond to the HB. HA and HG2 resonances respectively. Note that the HB peak is below the HA peak because this is a Thr residue. For each of these peaks you will need to make a new resonance with [<New>] and use [Assign [number]] to link it to the right atoms.

Assigning the HCcH TOCSY

For the  HCcH TOCSY spectrum we will use the location of the side chain resonances we have just assigned in the other TOCSY spectra to locate peak positions. Note that this spectrum does not have any peaks picked in it. The idea is that the CCcoNH spectrum will provide the ppm positions for the carbon resonances (on the Z-axis) and the HSQC TOCSY will provide the ppm positions for the hydrogen resonances (X- and Y-axes). The lines of intersection between the carbon and hydrogen resonances will hopefully show where the peaks (with HHC dimensions) can be found within the HCcH spectrum.

Open window5 to view the HCcH TOSCY spectrum. You can close the other windows to save space. Go to the main Analysis menu and select M:Resonance:Spin Systems. In the Spin Systems table that opens, click on the "Residue" column to sort the spin systems (groups of resonances in a asingle residue) into sequential order. Scroll down to the row for 19Thr, select the row, check that window5 is selected in the top right pulldown menu,  and click [Display Strips]. In window5 you will hopefully see three strips; for the CA-HA, CB-HB & CG2-HG2* positions. Each strip will have a vertical line at its centre, and there will be three horizontal lines, one for each side chain hydrogen. Where the horizontal lines intersect the vertical ones you will see intensity maxima in the HCcH spectrum.

With 19Thr and window5 still selected in the Spin Systems table click [Display Cells]. You will now see that the display in window 5 is now also split vertically to give separate HG2*, HA, HB and HN regions: these are for all hydrogens in the spin system although the HN resonance is not visible in the HCcH TOCSY, however it would be in a 13C HSQC NOSEY. For each of the nine intersection points with obvious maxima pick new peaks (<Shift> + <Ctrl> + left click). Assign each peak by pressing <a> with the mouse over the peak cross. The relevant resonances should be automatically presented for you, so all you have to do is click on the options in the right hand side oe the Assignment Panel. Note that you can assign several the peaks that align vertically or horizontally to the same resonances (in the matching dimensions) by selecting the peaks (left <Shift> + click) and spreading the assignments with R:Assign:Propagate assignments. Note that the 19Thr Ha, Hb, Ca peak at 3.82, 4.26, 66.0 ppm is slightly distorted along the Z-axis, so you may have to manually place the peak in the CA plane before it can be assigned to 19ThrCa; either pick it with <Ctrl> + click or select it and press <p> wile in the exact CA plane.

*Half-way point *

Copying Assignments, Multiple Shift Lists & Synthetic Peaks

If the program is not already open, start CcpNmr Analysis on the command line by typing:

-> analysis

Setting up and linking a new shift list

So far during this tutorial, all of the chemical shift values we have recorded, by picking peaks and linking resonances & atoms, have gone into one single list. However, there are several circumstances when it is helpful to use more than one list of chemical shifts, so that you may for example separate sets of shifts based upon the experimental conditions they relate to; you may use different shift lists for different temperatures or when doing titration experiments. Commonly in such circumstances the underlying resonances (and hence atoms) that we work with are the same, but the positions of the peaks move significantly and we would not want to average chemical shift values over such a variation. In the Analysis menu bar select M:Project:Open Project. Select to close the exiting project, but there is no need to save. Navigate to find and select the CcpnCourse2c project, then click [Open].

We are going to set up a new shift list for the second HSQC experiment present in the demonstration project. This second HSQC experiment relates to exactly the same sample (molecule) as the first HSQC but was recorded at a different temperature, which has caused the peak positions to move. Given that the second HSQC experiment is already loaded, all we need to do is go to the M:Experiment:Experiments {Experiments} and in the Shift List column double click on the cell corresponding to HSQC_2. In the pulldown menu that appears simply select <New>. You will see that a new shift list is now listed. From now on whenever an assignment is made in HSQC_2 the shift remains separate from the other spectra. Indeed, any resonances that were already connected to HSQC_2 before its shift list was changed will have their shifts recalculated in two separate groups. You could move the experiment back to the original shift list at any time and the chemical shift values will be appropriately recalculated. However, if you disconnect experiments from a shift list (or delete a peak etc) you might have chemical shift values that are not defined by any peaks. Under such circumstances the chemical shifts persist, but are described as "orphans".

Assignment of HSQC_2, with its separately curated shifts proceeds pretty much as assignment normally does, with the only major difference being that for a new shift list (i.e. without any assignments under it) there will be no resonance possibilities that can appear when assigning a peak. - The only known shifts are distinctly different, representing different conditions.

To assign HSQC_2, even though we don't have shifts set, we can say that certain peaks in the two HSQC spectra are equivalent if we can see how they have moved under the different situations. Accordingly we can select a peak in the first HSQC and the corresponding peak in HSQC_2 and propagate assignments. That is to say that the peak dimensions link to the same resonances (and hence atoms), even though things have moved. Go to spectrum window1 and try this for any pair of close, isolated peaks; one peak in each of the HSQC spectra. Use the mouse with a left click to ensure that these peaks (and only these peaks) are selected. Then in the right mouse menu select R:Assign:Propagate assignments. Note that this function usually checks to ensure resonance positions are within tolerances, but here with the two shift lists it cannot so the connection is made anyway, which is exactly what we want.

You may like to repeat the above procedure for several HSQC peaks and their counterparts. With a few resonances represented in the new shift list (should be "ShiftList 2") have a look at the resonance table M:Resonance:Resonances table to see evidence of these links. In the resonance table note that you can choose between the different shift lists in a pulldown menu at the top, i.e set the ShiftList: pulldown to "ShiftList 2". Note that when the shift list changes the resonance positions shift and the number of peaks linked also changes. Apart from the Resonance Table, Analysis also provides direct access to the shift measurements. Looking in the menu at M:Data Analysis:Measurement Lists, you will see that there are tables of the shift values corresponding to each list. Note that other types of value like T1 rates, or Hetero-NOE values will also appear as measurements in these tables if they are calculated.

Copying peaks between related spectra

Propagating assignments between peaks by manually selecting them, although sometimes necessary, can be tedious if you have large number of peaks that are equivalent and have closely matching positions. In this instance Analysis provides a utility to compare all of the peak positions in two related spectra and transfer the resonance assignments efficiently, whilst still leaving sufficient scope for human intervention. 

In this exercise we will transfer the amide assignments from the peaks in the HSQC experiment to the HSQC_2 experiment. In the main menu we select M:Assignment:Copy Assignments. In the Copy Assignments popup ensure that the first "HQSC:115:1" experiment is set as the Source Peak List and that the second "HQSC_2:48:1" is set as the Target Peak List.  In the top table you will see a list of the source peaks, together with an indication of how many destination peaks they match, within the given tolerances. Note that when the Follow Peaks? option is set clicking on a source row will cause peak to be highlighted in a relevant window (window1). Also, selecting the source will display the potentially matching destination peaks in the lower table. Clicking on a target row will, as before, highlight the relevant peak.

This utility can be used in one of two ways. One can manually choose the correct destination for each source peak and [Assign Selected Target], or if you trust the completeness and accuracy of the destination one can automatically assign the closest destination peak to each source one with [Assign Singly Matched]. When taking the more automated approach it is common to start with strict tolerances to assign the more certain matches and then relax the tolerances to fill in the remainder.

Synthetic Peak Lists

Relaxation Rate Analyses

Relaxation Analysis