Synopsis:

SCAssign (Side-Chain Assignment) is a Sparky extension written in Python to assist the assignment of aliphatic side-chain resonances of uniformly ¹³C,¹⁵N-labled large proteins. It is based on a general strategy recently developed in our lab, which makes use of 4D ¹³C,¹⁵N-edited NOESY, MQ-(H)CC_mH_m-TOCSY, and the prior backbone assignments. SCAssign runs on virtually all operating systems on which Sparky is available, and is easy to install, setup and use. The above screenshots show how it looks in Linux.

Features:

SCAssign offers plenty of useful features that can greatly accelerate the assignment process. Many tasks that used to take weeks or even months can now be done in just a few days. These features include:

• Import and isotope correction of chemical shifts from the prior backbone assignments
• Friendly GUI for defining (H^N, N, C) spin triplets and examining the potential matching NOE peaks
• Real time peak picking and automatic peak match to the defined spin triplets
• Simultaneous view of both the current and the referential C-H planes for a more reliable assignment
• One-click to assign/unassign NOE peaks and auto-aliasing of the assigned peaks
• Strip plot in CCH-TOCSY of selected NOE peaks to confirm assignment or resolve ambiguity
• Ability to identify and assign many weak NOE peaks using CCH-TOCSY

Availability:

The source code of SCAssign is provided as SCAssign (download) at this website. The download page also contains a step-by-step installation guide. Once you have it installed, carefully read the instructions on how to setup and use SCAssign, and then try out with your own data (infeasible for us to provide download of sample data set due to the huge file size of 3D/4D spectra). Some known issues with SCAssign are also listed out for your information, together with the workaround. Feel free to contact us if you need further support.

After extraction of the zip archive (~18KB), you should see four files with .py extension in the SCAssign folder, namely spectra_setup.py, import_shifts.py, sidechain_assign.py, and sparky_init.py. A brief description of each file is listed below.

spectra_setup.py	Setup spectra and preferences
import_shifts.py	Import chemical shifts and correct for deuterium labeling
sidechain_assign.py	Main program for side-chain assignment
sparky_init.py	Load SCAssign extension upon Sparky startup

 

Installation:

For more information on Sparky extensions, you can read the Extensions in Python section of the Sparky manual.

Normally all user-customized Sparky extensions are stored in the Python folder under a user’s home Sparky directory. If you have never installed any customized Sparky extension before, simply create a Python folder under your home Sparky directory, and copy all files in the SCAssign folder to this folder. After you have done, launch Sparky as usual (no compilation is required since Sparky will do that in real time) and you should find in the “Extensions” menu a new menu entry named “Sidechain assign” with an accelerator “sa”.

Please note that the above operation may overwrite any existing sparky_init.py file in the Python folder. So If you already have a few of your own Sparky extensions installed, here are the steps to add SCAssign to your extension list:

1. Copy only spectra_setup.py, import_shifts.py, and sidechain_assign.py to the Python folder.
2. Locate sparky_init.py in the same folder and open it with your favorite text editor.
3. Insert the following lines into the initialize_session() function and save the file. def sa_command(s = session):
   

   import sidechain_assign
sidechain_assign.show_sidechain_assign_dialog(s)

   session.add_command("sa", "Sidechain assign", sa_command) You need to replace session with the parameter you used when defining the initialize_session() function.

How to use:

Format of Shifts File:

In order for SCAssign to use the chemical shifts from prior backbone assignments, first you need to prepare a file in plain text that specifies those shifts. The format of this shifts file is illustrated below.

1	LYS	CA	56.304
1	LYS	CB	33.336
2	THR	CA	62.145
2	THR	CB	69.736
2	THR	H	8.236
2	THR	N	116.334
3	GLU	CA	56.124
3	GLU	CB	31.485
3	GLU	H	8.576
.	...	.	...

As seen from the above example, the file usually contains multiple lines, each consisting of four data fields (i.e., residue number, residue name, atom name, and chemical shift) separated by spaces or tabs. The following table summarizes the required data format for each field. Please do comply with it to avoid error during import.

Residue number:	Sequence number of the residue, counting from 1 and increasing monotonically
Residue name:	Amino acid name of the residue, in standard 3-letter or 1-letter code
Atom name:	In BMRB nomenclature, H for amide hydrogen, N for amide nitrogen, CA for α-carbon, ...
Chemical shift:	Value of the chemical shift, floating point number, can be positive or negative

 

Error Checking:

SCAssign checks the shifts file for any format error before import. If an error was found, a message would be displayed indicating the location and type of the error (as shown below). In this case, edit the file to make corrections accordingly, and then try again.

Assign Peaks:

Once you have identified which peak is the real match, you can assign it by Shift-clicking (press and hold the “Shift” key while clicking) on its entry in SCAssign’s peak list instead of manually filling in the group and atom names for each axis in the “Assignment” dialog window (accelerator “at”). SCAssign will generate the assignment label according to the selected spin triplet, and display it in the spectrum view next to the peak. At the same time, an asterisk mark will appear at the end of the entry to indicate that the peak has been assigned. You may increase the size of the labels in the “Ornament Sizes” dialog window (shown below, accelerator “oz”) If it is too small to be seen. To unassign a peak, simply Shift-click on its entry again. Both the assignment label and the asterisk mark will then be cleared.

When a peak is assigned, SCAssign will check its frequency in each of the four dimensions and alias it if necessary to bring the value closest to the expected frequency. This step would be ignored if the peak was aliased by the user prior to the assignment. For H^N, N, CA and CB, the expected frequency is taken as the respective chemical shift obtained from backbone assignments; for CG, CD, etc. and all side-chain protons, it is taken as the respective mean chemical shift in the empirical range.

Many amino acid residues contain side-chain carbon atoms that carry more than one hydrogen atoms (e.g., CA in Gly has HA2 and HA3). They are reflected on the 4D NOESY spectrum as distinct peaks with the same aliphatic carbon shift but slightly different aliphatic proton shifts. Since in this case SCAssign is unable to differentiate one hydrogen atom from another, it will simply assign those peaks with the same label. To append a suffix number to a hydrogen atom, you need to edit the label in the “Assignment” dialog window.

Edit Peaks:

Most of the time the auto-alias performed by SCAssign gives good results and hence you save the hassle of doing it manually. In case of wrong alias, you can correct using Sparky command “a1”, “a2”, … or “A1”, “A2”, … (refer to Aliased peaks in the Sparky manual). Please be reminded that all aliases made to a peak will be cleared when you unassign the peak by Shift-click.

As you may have noticed, in fact all Sparky commands is callable from within SCAssign. Besides zooming in and out of a spectrum view or aliasing a peak, you can type “lt” to call up a peak list for the currently selected spectrum, or “js” to save the project, and so on. In addition, you may still use the function keys (F1 to F12, some keys may not work on certain machines) in SCAssign to change the pointer mode.

SCAssign always highlights the entry of the selected peak in its peak list. This is true even when you select multiple peaks in the spectrum view. If a peak is dragged, SCAssign will immediately update to show the new position and data height, and re-sort its peak list accordingly. In consistent with way that Sparky works, you can delete unwanted peaks in SCAssign simply by pressing the “Delete” key.

Plot Strips:

Before plotting strips, please make sure that you have properly configured the X, Y and Z dimensions of the CCH-TOCSY spectrum during setup. SCAssign constructs a strip plot by first locating its position on the X-Z plane using the above mentioned C-H frequencies, and then displaying the spectrum view along the Y dimension. You may notice that the “Strip plot” extension itself provides a “Show peak strips” dialog window (accelerator “ss”, refer to How to Show Strips in Sparky manual for a full description and screenshot), where you can choose the spectra for which strip plot will be performed and define how their axes correspond to the X, Y and Z dimensions. Do not confuse this with SCAssign’s own spectra setup. In fact all settings made in this window will be totally ignored by SCAssign in strip plot. They are neither sufficient nor necessary.

Select and right click on any peak in SCAssign’s peak list will show the corresponding strip in CCH-TOCSY. A “Strip plot” window will be created if no one exists. Otherwise, it will be restored (if previously minimized) and raised on top of all other windows. The newly drawn strip is highlighted in a color frame. SCAssign labels each strip at the bottom with the respective C-H frequencies, the identity of the residue containing the C-H group, and the position of the C-H group (α, β, γ, etc.). It also aliases the C frequency to get the expected position of that C-H carbon atom on the Y axis of the spectrum, and marks the position in the strip with a bright line for easy comparison.

The auto-alias for calculating the C-H frequencies of a strip is performed only on unaliased peaks and affects only the strip. The peak itself will not be auto-aliased until it is assigned by the user. Once a peak is aliased, SCAssign will instead use the C-H frequencies of that peak for strip plot. In addition, SCAssign synchronizes all strips with the respective peaks, so at any time you can drag a peak to adjust its position, or manually alias it if you are not satisfied with the result of the strip plot based on auto-alias. The strip will be redrawn and updated immediately to reflect such changes. Sometimes the “closest to the expected value” principle fails and SCAssign may mistakenly alias to get the C-H frequencies for strip plot while they are in fact identical to the on-spectrum frequencies of the peak. If this happens, just turn off the auto-alias feature by deselecting the checkbotton located on the upper right corner of the “Strip plot” window as shown below, and plot the strip again.

Analyze Strips:

The “Strip plot” window is interlinked with SCAssign’s peak list and the spectrum view to allow convenient analysis and cross-check. For example, when you click on an peak entry in the list, if it has a strip but the strip is not currently displayed in the plotting area, SCAssign will automatically scroll to that strip and highlight it. Similarly when you double-click on a strip, SCAssign will show the corresponding peak in the spectrum view and select it, pinpoint with a crosshair the same C-H position on the reference plane, and highlight the peak entry in the list if present. A Flash demo is available below. Please note that the reference plane is not shown due to space constraint.

For any strip in the strip plot, once its corresponding NOE peak is deleted, the strip will be deleted as well. If you just want to delete the strip without affecting the peak, click to select the strip first (which will be highlighted upon selection) and then type the command “sd”. Typing “sD” will delete all strips. You can zoom in or out on a selected strip using the command “si” or “so”. You can also type “sw” to bring up a dialog window for adjusting the strip width and the gap between the strips. All these commands are accessible under the “Show” menu as highlighted below. For more details on the functions provided by the “Strip plot” extension, you may read the Strip Plot section of the Sparky manual.

Based on the general strategy that will be introduced later, the strips can be compared and analyzed to confirm assignments or resolve ambiguities that could not be resolved with the 4D NOESY spectrum alone. The same as with entries in SCAssign’s peak list, Shift-click on a strip will assign the corresponding NOE peak and auto-alias it closest to the expected frequencies.

Identify Weak NOE

Not only does SCAssign speed up the process of assigning aliphatic side-chain resonances in large proteins, but also it produces a more complete set of assignments, since with the help of SCAssign you can now assign some side-chain resonances that could not be manually assigned before. In this section, we will present an example in which weak NOEs between an amide proton and aliphatic protons at the distal end of a side-chain can be identified by SCAssign and used for resonance assignment.

Our general strategy to assign aliphatic side-chain resonances was developed based on the statistics of interatomic distances which indicate that nearly all HAs and HBs, many HGs, and some HDs will give rise to both intraresidue and sequential NH-CH NOEs. However, a number of such NOEs, especially those involving HDs and HEs, may appear too weak due to their usually longer distances to amide protons, and hence may not be observed in the contour plot with threshold set for manual analysis. The spectrum view below on the left shows a C-H plane from 4D NOESY defined by the NH of T2 in a maltose binding protein (MBP, 370 residues, 42 kDa). The contour plot has a threshold of 2.4×10⁶ for both positive and negative peaks, a setting used most of the time during manual assignment to maximally eliminate background noises. With SCAssign, we can quickly assign HA and HB of K1 by searching for peaks whose chemical shifts match the triplet (H^N₂, N₂, CA₁) and (H^N₂, N₂, CB₁). Possible peaks for CG and HG of K1 can be similarly found by SCAssign using the empirical range of CG’s chemical shift, and the ambiguities are resolved with additional information obtained from the strip plot. However, when we are trying to assign CD, HD, CE and HE of K1, the program returns no possible peaks at the current threshold.

In an attempt to assign these resonances, we first lower the threshold of the contour plot to 1.2×10⁶ for the 4D NOESY spectrum. As a result, more peaks emerge and meanwhile we start to see some noises (as shown above on the right). We then search for possible peaks again in SCAssign, and this time 10 are found for (H^N₂, N₂, CD₁) and 8 for (H^N₂, N₂, CE₁). Right click on each entry in the peak list to display the strip plot, the peak whose corresponding strip contains no meaningful pattern is most likely the noise and hence can be deleted straight away. The remaining strips are compared with reference to the strips of K1A, K1B and K1G to resolve ambiguities on the basis of pattern matching. In the following screenshot (due to space constraint only representative strips of K1D and K1E are shown here), it is not difficult to realize that the peak pattern in the 3rd strip of K1D and the 2nd strip of K1E (counting from the left) aligns most well with that in the reference strips. Therefore, the C-H frequencies of these two strips can be respectively assigned as the chemical shifts of CD, HD, CE and HE.

General Strategy:

Please refer to the paper below if you want a comprehensive description.

Xu, Yingqi; Lin, Zhi; Chien, Ho; Yang, Daiwen. A General Strategy for the Assignment of Aliphatic Side-Chain Resonances of Uniformly 13C,15N-Labeled Large Proteins. J. Am. Chem. Soc. 2005, 127(34):11920-11921

Traditionally, structure determination by NMR is suitable for small proteins (<25 kDa) of which backbone and side-chain assignments can be obtained using uniformly ¹³C,¹⁵N-labeled samples with triple resonance experiments. Such approach often fails when applied to proteins larger than 30 kDa due to increased transverse relaxation rates. With the introduction of deuteration and TROSY techniques, it is possible to assign backbone and ¹³C^β resonances for proteins up to 100 kDa. However, deuteration also significantly reduces distance constraints derived from NOEs among aliphatic and aromatic protons, leading to low resolution structure. Although more long-range NOEs can be observed by selectively reintroducing methyl protons into otherwise deuterated samples since methyl groups are often involved in hydrophobic cores, preparation of deuterated and methyl-protonated samples is always costly and time-consuming and may not be suitable for every protein. To further improve structure resolution, it is necessary to constrain side chains of all or most residues using NOEs among protons located at side chains. This implies that complete or partial protonation at most side chains is inevitable.

For fully protonated large proteins, our group has developed a novel experiment MQ-(H)CC_mH_m-TOCSY and a strategy to assign side-chain resonances of methyl-containing residues. The method has been successfully applied to a 42 kDa AcpS trimer and a 65 kDa chain-selectively labeled hemoglobin. So far, no method can be applied to assign side-chain resonances of residues that contain no methyl groups.

The new strategy introduced here makes use of 4D ¹³C,¹⁵N-edited NOESY and prior backbone assignments to assign side-chain resonances of all residues in uniformly ¹³C,¹⁵N-labeled large proteins. Although most triple resonance experiments involving both ¹³C and ¹⁵N spins have very poor sensitivity for protonated large proteins, NOESY experiments are still sensitive enough to provide through-space correlations between spins separated by 4.5 Å or less (5.5 Å for methyl groups). Statistics on interatomic distances indicate that nearly all intraresidue H^N_i-H^α_i, H^N_i-H^β_i and sequential H^N_i-H^α_i-1, H^N_i-H^β_i-1 NOEs can be observed. In 4D NOESY, each amide correlates with a number of CH_n groups at positions [ω(H^N_i), ω(N_i), ω(C^k_j), ω(H^k_j)], where ω is the chemical shift; k is the kth carbon/hydrogen of residue j. H^α and H^β can be assigned from intraresidue or sequential NOEs, provided that these NOEs are distinct from other interresidue NOEs on the basis of prior assignment of H^N, N, C^α, and C^β spins. Otherwise, ambiguities in assignment can be resolved using both intraresidue and sequential NH-CH NOE correlations. If the ambiguity cannot be resolved due to a lack of sequential or intraresidue NOEs, an MQ-(H)CC_mH_m-TOCSY experiment can be applied to confirm the assignment. Assignments of H^γ, H^δ, and H^ε are much more challenging since the chemical shifts of respective carbon spins are unknown. According to the statistics on H-H distances, many H^γs and some H^δs give rise to both intraresidue and sequential NH-CH NOEs and thus can be assigned in 4D NOESY. The remaining spins can be assigned by combining TOCSY and NOESY experiments.

This strategy was tested on a cell adhesion protein (DdCAD-1, 214 residues) and β-chain of human normal adult hemoglobin in the carbonmonoxy form (rHbCO A, ~65 kDa with two identical α-chains and two identical β-chains). By comparing the results with those obtained previously, it was shown that most aliphatic side-chain resonances of these large proteins can be reliably assigned.

 

The procedure for assigning H^α and H^β is as follows.

Step 1:
Identify peaks whose chemical shifts match the shifts [ω(H^N_i), ω(N_i), ω(C^α_i)] on the C-H plane in 4D NOESY defined by N_i-H_i. If only one peak matches, the aliphatic proton shift of this peak is presumably assigned as the chemical shift of H^α_i.

Step 2:
Similarly by substituting ω(C^α_i-1), ω(C^β_i) and ω(C^β_i-1) for ω(C^α_i) above, H^α_i-1, H^β_i and H^β_i-1 may be presumably assigned, provided that unique matches also exist (for CH₂ groups, two peaks with identical carbon shifts are also regarded as a unique match).

Step 3:
In step 1 and 2, if an assignment obtained from intraresidue NOE is consistent with that obtained from sequential NOE (Figure a, b on the right), the assignment is confirmed.

Step 4:
For an unconfirmed assignment on the C-H plane defined by N_i-H_i, if its C-H shifts match those of any peak on the plane defined by N_i+1-H_i+1 or N_i-1-H_i-1 (Figure c, d), the assignment is also confirmed.

Step 5:
When no assignment can be made in step 1 or 2 due to ambiguities, directly compare the planes defined by N_i-H_i and N_i+1-H_i+1 (or N_i-1-H_i-1) to resolve the ambiguities.

Since degeneracy of (H^N, N, C) spin triplets occurs in a much lower chance than that of (H^N, N) spin pairs, most H^αs and H^βs could be presumably assigned with only intraresidue or sequential NOEs (result table, columns A and B). For DdCAD-1, we found that all assignments obtained with both intraresidue and sequential NOEs (result table, column C) were correct, while three of the unconfirmed assignments (result table, column D) were incorrect. In rare cases where the above methods fail to unambiguously assign H^α or H^β, a 3D CCH-TOCSY spectrum can be used to resolve the ambiguities, as will be described in the next section.

The procedure for assigning H^α and H^β can be similarly applied to assign other side-chain resonances. Although exact chemical shifts of C^γ and C^δ are unknown, their empirical ranges may serve as a guide for locating possible peaks (Figure a, b on the right). However, due to the obvious problem of chemical shift degeneracy as well as usually longer distances to amide protons, less number of C^γH^γ_n and C^δH^δ_n groups can be assigned with 4D NOESY alone (result table, columns A and B). In this case, a 3D CCH-TOCSY spectrum has to be used in conjunction with 4D NOESY to assign any unconfirmed and remaining resonances (Figure c-g on the right). Details are given below.

Once a peak at position [ω(H^N_i), ω(N_i), ω(C^k_j), ω(H^k_j)] in 4D NOESY is unambiguously assigned (most often would be H^α and H^β peaks), draw its corresponding strip plot at [ω(C^k_j), ω(H^k_j)] in CCH-TOCSY and mark the C^k peak on the strip. Such strips will be our “reference strips”. Later on if ambiguities arise when assigning other resonances (e.g. C^γH^γ_n ) in residue j, similar strip plots can be drawn and compared for the number of matching peaks (as those joined by lines in Figure c-e and f, g). The more the matches a strip shares with the “reference strips”, the more likely that the NOE peak to which it corresponds is the correct one to receive the assignment.

The above method can also be used to assign H^α or H^β that could not be assigned previously with 4D NOESY alone, provided that there are other confirmed assignments from the same residue for comparison, or the strip plot itself provides enough information.

After all these steps, the aliphatic side-chain assignment can reach a completeness of up to about 96% in DdCAD-1 and 80% in rHbCO A (the ratio of the assigned to total aliphatic CH_n groups, result table, column E). Hence using our strategy, much more distance constraints may be obtained for accurate structure determination of large proteins.