Users' Guide

Table of Contents
Introduction
Getting Started
Loading PDB Files
Running Analysis
Evaluating Analysis Results
Building Mutant Proteins
Saving Results & Models
Tips & Tricks
Technical Details
Release Notes
References
Papers Using DbD

---

Disulfide by Design 2 (DbD2) is a redesigned and enhanced version of our original DbD application (Dombkowski, 2003) for the rational design of disulfide bonds in proteins. For a given protein structural model, all residue pairs are rapidly assessed for proximity and geometry consistent with disulfide formation, assuming the residues were mutated to cysteines. The output displays residue pairs meeting the appropriate criteria. The input model will typically be a Protein Data Bank (PDB) structure for the protein of interest; however, structures developed through homology modeling may also be used. Engineered disulfides have proven useful for increasing the stability of proteins and to assist the investigation of protein dynamics and interactions. This software was written by Dr. Alan Dombkowski and Douglas Craig, and is based on algorithms created for disulfide identification in protein fold recognition methods (Dombkowski & Crippen, 2000). The Disulfide by Design algorithm has been successfully used for disulfide engineering in a wide variety of applications (see Papers Using DbD below).

Disulfide by Design 2 has been tested with the following browsers:

• Chrome (v42.0)
• FireFox (v38.0)
• Safari   (v6.0 iOS, v5.0 Win)
• Opera   (v27.0)   NOT RECOMMENDED due to performance issues with 3D model viewing
• IE          (v10)   NOT RECOMMENDED due to compatibility and performance issues

Contact us with any compatibility issues.

---

Try loading the sample PDB file by clicking on  Load Example PDB 
(or load your own PDB from a local machine, or directly from www.PDB.org).

Click on the  Run  button at the bottom of the Options Pane.

Inspect the analysis results. Find candidate residue pairs based on bond energy, B-factor, associated secondary structure, or 3-D considerations.

Select one or more residue pairs using the checkboxes under the ANALYSIS tab.
Click on  Create/View Mutant  to build the new disulfide bonds.

Inspect the resulting model under the 3-D VIEWS tab.
Click on  Save Mutant PDB  to save the new PDB file.

---

All DbD analysis begins by loading an existing PDB file. PDB files can be loaded three different ways.

Figure 1. Loading a protein (PDB) file.

	PDB files may be loaded from a local or networked drive. Click on the  Choose...  or  Browse...  button (depending on browser) and locate the PDB file to load. Then click the  Upload  button to load the PDB file into memory. NOTE: there is a 2MB size limit for local files.
	PDB files may be loaded directly from the PDB.org website. Type in the PDB identifier (e.g., 7RSA), then click the  Get from PDB.org  button to load the PDB into memory.
	If you are just getting started you can load a sample PDB to experiment with. Click on  Load Example PDB .

Once the PDB file is successfully loaded into memory, the FILE INFO tab will display summary information about the loaded protein. This information is gathered from PDB file fields. The actual content of the file will also be displayed in the scrollable/resizable Content pane.

Figure 2. FILE INFO Tab.

---

The Options Pane (to the left of the FILE INFO tab) is used to set all DbD analysis parameters.

Figure 3. Options Pane.

	By default, DbD checks all potential inter and intra-chain disulfides. To limit analysis to only those potential bonds between different chains, uncheck Intra-chain (leaving Inter-chain checked).
	To limit analysis to only those potential bonds within a chain, uncheck Inter-chain (leaving Intra-chain checked).
	The DbD algorithm requires coordinates for Cβ atoms to determine the potential for disulfide formation. Since glycine residues do not include a Cβ atom they cannot be used in the analysis unless one is created. The Build Cβ for Gly checkbox enables the construction of Cβ atoms. The Cβ location is determined by using the coordinates of the residue backbone atoms.
	The χ3 torsion angle is formed by the Cβ-Sγ-Sγ-Cβ bonds, with rotation about the Sγ-Sγ bond (see Figure 14 below). The distribution of χ3 angles observed in disulfides of known protein structures is bimodal with sharp peaks at +97° and -87° (see Figure 15 below). It may be desirable to restrict candidate disulfides to those having an estimated χ3 value that falls within the region of observed χ3 values. The χ3 angle tolerance can be increased or decreased using the drop-down list of values. The default setting is +97° ±30° and -87° ±30°. If numerous putative disulfides are identified using the default setting, it may be useful to decrease the tolerance resulting in a shorter list with preferential characteristics.
	The distribution of Cα−Cβ−Sγ angles observed in known disulfides has a peak near 115° and covers a range from approximately 105° to 125° (Petersen et al., 1999). The analysis tolerance for this bond angle can be adjusted using the drop-down list of values. The default setting is 114.6° ±10°.
	Click  Run  to start the analysis. Each possible pair of residues will be assessed for potential disulfide formation, assuming the residues were mutated to cysteines. For large proteins progress will be displayed at the bottom of the Options Pane as well as under the ANALYSIS tab. You may switch tabs while analysis is running without affecting the results. To abandon analysis click the  Abort  button.
	To reset all options to their default values click the  Reset  button.
	To load a new PDB file click the  New File  button. Any current analysis or mutation results will be lost unless previously saved.

Figure 4. Multiple model selection.

If the loaded PDB has more than one model (e.g., from NMR) you may select which model to analyze from the drop-down list located at the top of the Options Pane. By default, the first model is selected from multiple model files. Note, this dropdown is not visible for single model files.

---

After running DbD analysis, three categories of information are available from which to assess results:

• Residue pair information, including energy, angle and B-factor
• Detailed secondary-structure
• Three-dimensional visualization

Each is available under a separate tab, and is described in detail below.

This tab displays a list of all residue pairs which analysis indicates may potentially form a disulfide bond. All information (except for energy and angle) is drawn directly from the PDB file.

Figure 5. ANALYSIS Tab.

Residue	Detailed information for each of the residues that makes up a potential disulfide.
Chain	Chain identifier for the residue.
Seq #	Residue sequence number.
AA	Residue name.
Struct	Gives any name and structure information from the PDB associated with the residue's secondary structure. Possible structure types are: Sheet [name]: strand [m] of [n],[sense] Helix [name]: [type]     where type is one of the following: α-helix, RH ω-helix, RH π-helix γ-helix, RH 310 helix α-helix, LH ω-helix, LH γ-helix, LH 27 ribbon poly-Proline
Bond	Calculated information about the potential disulfide bond.
χ3	Estimated χ3 torsion angle (see Technical Details).
kcal/mol	Calculated bond energy in kcal/mol (see Technical Details).
Σ B-factor	Sum of the temperature factors (from the PDB) associated with the two residues, using backbone and Cβ atoms.
Create/View Mutant	See Building Mutant Proteins.
Save Results	See Saving Results & Models.

This tab displays detailed secondary structure for the protein, including where potential disulfide bonds are located based on DbD analysis. All information is drawn directly from the PDB file.

Figure 6. 2° STRUCTURE Tab.

	Graphical representation of secondary structure: sheet, helix, none, or missing (dotted gray line). Move mouse over to see residue details.
	Residue one-letter abbreviation.
	Color coded temperature (B) factor. The color scale is normalized so that blue represents the lowest value among residues in the protein and red the highest value. Move mouse over to see actual numeric value from the PDB.
	Chain name.
	Residue sequence number (only every tenth number is displayed).
	The residue background is highlighted red when selected/checked under the ANALYSIS tab.
	The residue background is highlighted gold for all residues listed under the ANALYSIS tab.
	Moving the mouse over the secondary structure graphic pops up a list of details for each residue, including greater detail about sheet and helix types, as well as temperature (B) factor values.

This tab displays an interactive 3D representation of the protein using the open-source Jmol Java plugin, as well as the mutated protein if one has been generated (see Building Mutant Proteins).

Figure 7. 3-D VIEW Tab.

	Click directly in the display region to manipulate the protein model: Mouse over to identify residue names and numbers. Left-click and drag to rotate the model. Scroll-button/wheel to zoom in or out. Right-click to view Jmol options menu.
	Click to reload the model in Jmol.
	Click the radio button to display using cartoon rendering.
	Click the radio button to display using wireframe rendering.
	Toggles the display of disulfide bonds (yellow bars).
	COPY ORIENTATION  can be used to replicate orientation and zoom/magnification between the Original and Mutant views.

---

Once DbD analysis has been run you will have a list of residue pairs under the ANALYSIS tab which could potentially form a disulfide bond if mutated into cysteines.

Figure 8. Select residue pairs and create mutant.

	Check all residue pairs you wish to mutate into cysteines.
	Click on  Create/View Mutant  to generate a new PDB in memory with the mutated cysteines and new disulfide bonds. The 3-D VIEWS tab will automatically open with the mutated protein displayed side-by-side with the original protein.

Figure 9. 3-D VIEW Tab with mutant.

---

The results of DbD analysis can be saved to a local text file in comma-separated value (CSV) format.

Figure 10. Save analysis results.

Click on  Save Results  under the ANALYSIS tab, then select a local file to save analysis results to.

The output can then be opened in any spreadsheet application (e.g., Excel), and includes the following information:

Figure 11. Analysis results CSV file contents.

Any mutated protein (i.e., one with new disulfide bonds) can be saved to a new PDB file.

Figure 12. Save mutated PDB file.

Click on  Save Mutant PDB  under the 3-D VIEWS tab, then select a local file to save the new PDB to.

Note that for multiple model PDB files, only the one modified model is saved to the new PDB.
In addition to new   SSBOND   records, a   REMARK   record is added to the new PDB above all residues mutated to cysteines, as follows:

Figure 13. Mutant PDB file disulfide.

---

When analyzing a PDB with multiple models it is possible to generate a mutant for one model of the PDB and then load a second model (using the Options Pane drop-down). This allows a side-by-side visual comparison of the two model variants under the 3-D VIEW tab.

---

The Figure 14 below represents a cysteine pair coupled by a disulfide bond. The approach uses a disulfide model with fixed C_β-S_γ and S_γ-S_γ bond lengths, 1.81 and 2.04 Ĺ respectively, and fixed C_β-S_γ-S_γ bond angles of 104.15°. These bond lengths and angles are consistent with values observed in a survey of protein disulfide bonds (Petersen et al., 1999). The χ₃ torsion angle, formed by rotation of the C_β atoms about the S_γ-S_γ bond, is allowed to vary in the model and can be described as a function of the distance between C_β atoms (Dombkowski & Crippen, 2000).

Figure 14. Anatomy of a disulfide.

To "fit" the disulfide model between a pair of residues, the algorithm simply rotates the χ₃ angle to obtain a C_β-C_β distance that matches the C_β-C_β distance measured between the residues. Numerous S_γ locations are possible, so all possible S_γ orientations are examined and the atomic coordinates providing the lowest energy (E_ij) are selected, based on the χ₁ and χ₃ torsion angles and the two C_α-C_β-S_γ angles. The χ₁ torsion angle is defined by the N-C_α-C_β-S_γ atoms. E_ij is calculated per equations (1-4), where i and j are residue indices, θ is the C_α-C_β-S_γ angle, and θ₀ is set to 114.6°. Energy units are kcal/mol.

(1)	Eij = E(χ1,i) + E(χ1,j) + E(χ3) + E(θi) + E(θj)
(2)	E(χ1) = 1.4 [ 1 + cos(3χ1) ]
(3)	E(χ3) = 4.0 [ 1 - cos(1.957[χ3 + 87.0]) ]
(4)	E(θ) = 55.0 [ θ - θ0 ]2

The energy calculation provides minima at χ₃ values of +97° and -87° (Figure 15) and χ₁ values of ±60° and ±180°. These values are derived from a sample of 331 proteins containing 1418 known disulfide bonds. Since the disulfide model uses fixed bond lengths, no term is included for bond stretching.

Figure 15. χ₃ distribution of 1418 known disulfide bonds.

The distribution of energy values for these 1418 known disulfide bonds (Figure 16) reveals a mean energy value, calculated per equations (1 -4 above), of 0.89 kcal/mol and a maximum of 8.35 kcal/mol.

Figure 16. Energy distribution of 1418 known disulfide bonds.

The energy calculation also provides a means to compare potential disulfides during disulfide design. The following Figure 17 compares the calculated energy of actual disulfide bonds to their corresponding designed disulfide bonds.

Figure 17. Energy for 1418 designed disulfides versus energy of the actual disulfides.

A list of PDB identifiers used to validate DbD2 can be found here.

---

• Some PDB files do not load. This is most often due to "non-standard" residue numbering. Although the PDB file format does not specify that residues should be numbered starting with 1, DbD does expect this. Workaround: it may be possible to generate a new PDB with corrected residues using available PDB file utilities.

• Converted 3D Viewer to JSmol to eliminate dependency on the Java browser plug-in.

• Removed the Analysis composite-rank calculation which was shown to be misleading.
• Added Temperature Factor (B-factor) range and mean to the File Info tab.

• Added Analysis composite-rank calculation.
• Added mouse-over text for 2D structure to show associated SS bond residues.
• Corrected Temperature Factor (B-factor) nomenclature.
• Fixed a problem with column sorting on the Analysis tab.

• Corrected Microsoft Internet Explorer incompatibilities.
• Improved Jmol 3D Viewer performance.

• Windows version converted to a completely web-based application.
• Integrated file loading directly from PDB.org
• Automatic link to PDB.org based on file identifier (DBREF).
• Support for multiple-model PDB files.
• Support for very large files using incremental analysis.
• Graphical display of 2° structure and disulfide bonds.
• Detailed 2° structure for potential disulfide bonds.
• Inclusion of constituent residue Temperature Factor (B-factor) values for bonds.
• Interactive 3D view of original and mutant models using integral Jmol viewer.
• Enhanced PDB file information display, including file source.
• Save Results now uses standard CSV format.
• Sortable columns for Analysis results.

• Corrected bug that caused text to go beyond the right-hand border of the dialog box with some terminal settings.

• Energy units are now in kcal/mol.
• For residues with multiple conformers only the first set of coordinates encountered are used.
• For NMR structures with multiple models only the first model found in the PDB file is used.
• Fixed bug in torsion angle calculation that caused crashes for a small number of PDB structures.

---

---