Liability Analysis for Biologics

Tutorial Created with Software Release: 2026-1
Topics: Antibody Design, Biologics Drug Discovery
Products Used: BioLuminate

Tutorial files

0.2 MB
This tutorial is written for use with a 3-button mouse with a scroll wheel.
Words found in the Glossary of Terms are shown like this: Workspacethe 3D display area in the center of the main window, where molecular structures are displayed

 

Tip: You can hover over a glossary term to display its definition. You can click on an image to expand it in the page.
Abstract:

 

In this tutorial, you will learn how the Protein Surface Analysis tool can be used for liability detection and mitigation in biologics.

 

Tutorial Content

  1. Introduction to Aggregation Propensity Prediction

  1. Creating Projects and Importing Structures

  1. Running Protein Surface Analysis

  1. Comparing the Protein Surfaces

  1. Analyzing the Correlation Between AggScore and HP-SEC Retention Time

  1. Conclusion and References

  1. Glossary of Terms

1. Introduction to Aggregation Propensity Prediction

Developability assessment is a key stage of the lead optimization stage of the protein-based therapeutic development process. One property of particular interest during developability assessment is aggregation propensity as it is a key concern during both the production and storage of protein-based therapeutics. As the propensity for aggregation has historically been expensive to predict experimentally, there is a clear need for computational tools that could help predict the relative aggregation propensity of a set of protein variants.

While most existing predictors for aggregation propensity such as Zyggregator and AGGRESCAN are sequence-based, here we will be focusing on a structure-based descriptor, called AggScore, which can be used to prioritize protein variants based on predicted aggregation propensity. See the AggScore: Prediction of aggregation-prone regions in proteins based on the distribution of surface patches for more information.

While hydrophobic patches have traditionally been viewed as a key driver of aggregation, AggScore considered whether those patches are surrounded (in 3-dimensional space) by neighboring polar patches. Residue-specific contributions to aggregation (as determined by positive, negative, and hydrophobic patches) are calculated on the interaction surface of the protein and smoothed over a window of five residues. This smoothing allows for the polar patches to limit or possibly negate entirely the predicted impact a hydrophobic patch could have on aggregation propensity.

Consider the two 5-residue peptides below. While both have polar phenylalanine in the center, the phenylalanine in the peptide on the left is neighbored by two polar serines, while the phenylalanine in the peptide on the right is surrounded by alanines.

Figure 1. AggScore results for two 5-residue peptides

As the phenylalanine in the peptide on the left is flanked by two polar residues which counteract the impact the hydrophobic patch could have on aggregation propensity, the AggScore for all of the residues is zero. For the peptide on the right, however, we can see that the AggScore associated with the phenylalanine is 3.2, since it is surrounded by non-polar alanine residues which do not negate the impact of the hydrophobic patch. It is the interplay between the polar and hydrophobic patches in a 3-dimensional environment that afford the AggScore a predictive power beyond that of the commonly used sequence-based predictors.

The AggScore function provides propensity values for each amino acid position in a protein, thus allowing for the prediction of aggregation hotspots within the protein. It also provides a single value for an entire protein, which can be used for rank-ordering several variants by their relative aggregation propensity

It is important to note that AggScore is not dependent on the presence of natural amino acids and is not trained on beta-amyloid aggregation data.

In this tutorial you will compare the AggScore values for a series of monoclonal antibodies that were developed to bind to nerve growth factor described in Dobson et al., 2016. You will also see how AggScore correlates with HP-SEC retention time.

2. Creating Projects and Importing Structures

  1. Open BioLuminate.
  2. Download the tutorial zip file including input files and reference outputs here: https://www.schrodinger.com/sites/default/files/s3/release/current/Tutorials/zip/liability_analysis_biologics.zip
  3. After downloading the zip file, unzip the contents in your Working Directorythe location that files are saved for ease of access throughout the tutorial.

Figure 2-1. Opening the project.

  1. Go to File > Open Project.
  2. Choose MEDI1912_aggregation.prjzip.
  3. Click Open.
    • Structures are added to the Entriesa simplified view of the Project Table that allows you to perform basic operations such as selection and inclusion, with the top entry includedthe entry is represented in the Workspace, the circle in the In column is blue in the Workspacethe 3D display area in the center of the main window, where molecular structures are displayed.

 

Figure 2-2. Saving the project.

  1. In the Save scratch project dialog box, click OK.
  2. Go to  File > Save Project As.
  3. Change the File name to aggregation_tutorial and click Save.
    • The project is now named aggregation_tutorial.prj.

3. Running Protein Surface Analysis

Before loading anything into the Protein Surface Analysis panel, the protein structure must already be prepared using the Protein Preparation Workflow. Please see the Introduction to Structure Preparation and Visualization tutorial for more details on using the Protein Preparation Workflow. Also see the Best Practices for Protein Preparation for more information. For more information about the Protein Surface Analysis, see the documentation page for the panel. While we show how this can be run for a single protein through the Graphical User Interface, you can also run this analysis in batch through the command line using protein_patch_calculation.py.

Figure 3-1. Opening the Protein Surface Analyzer panel.

While we have pre-run Protein Surface Analysis for the 6 mutants of MEDI-1912 (as you can see by the symbol next to the entry title), you will now run it for MEDI-1912.

 

  1. Find and open Protein Surface Analyzer panel from Tasks.

Figure 3-2. Protein Surface Analyzer panel.

  1. In the Protein Surface Analyzer panel, click Analyze.
    • This step takes around ~ 1 minute.
    • Once the job is completed, a surface is generated surrounding the protein in the Workspacethe 3D display area in the center of the main window, where molecular structures are displayed.
    • The Protein Surface Analyzer panel now displays patch and aggregation information for the protein.

 

Note: Previous analyses can be imported from file to the Protein Surface Analyzer panel. Additionally, completed analyses can be saved for later reference.

Figure 3-3. The Aggregation tab in the Protein Surface Analyzer panel.

While we will be focusing on the AggScore across the whole protein, it is good to understand which residues are contributing more/less to the AggScore value (and therefore predicted to potentially drive aggregation).

 

  1. Switch to the Aggregation tab.
  2. Double click the AggScore column header.
    • The table is sorted by descending AggScore.

If you’d like, you can adjust Color by to AggScore such that the surface more represents the per-residue AggScore contribution instead of just the charged and hydrophobic patches. This is generally easier for visual inspection. This can make it a lot more apparent where the potentially ‘bad’ patches are located on the protein.

 

Figure 2. MEDI-1912 with a surface colored by per-residue AggScore

 

We can see that for MEDI-1912 F31 and W30 have the highest per-residue AggScore. This aligns well with Dobson et al., 2016 where they identified W30, F31, and L56 to be the drivers of the self-association. In fact, the letters next to the names of the entries of the mutants in the BioLuminate project represent the amino acids at those three positions respectively for each mutant.

 

If you want to generate new mutants based on this data, you can select Export to Residue Scanning, to send the selected residues to the Residue Scanning panel. Please consult the Improving Antibody Stability/Affinity Using MM-GBSA Residue tutorial for additional guidance.

 

You can also click Profiles to view the predicted Aggregation profiles from AggScore, Aggrescan, and Zyggregator.

Figure 3-4. The Reactive Residues tab in the Protein Surface Analyzer panel.

In addition to the AggScore prediction, the Protein Surface Analysis calculation will also predict which residues might be susceptible to post-translational modifications based on side chain accessibility and pattern matching. There is also a separate Reactive Residues panel that you can use for similar analysis (though the pattern matching is more customizable there).

 

  1. Switch to the Reactive Residues tab.

W30 is identified as a potential oxidation site. There is a chance that when we mutate the site to improve the aggregation profile we will also address this potential liability.

Figure 3-5. The Properties tab in the Protein Surface Analyzer.

The Properties tab provides some high-level, structure-wide properties that might be of interest. Most notably, it includes the charge, which could be helpful when thinking about the developability of certain biologics, as well as the Sum of AggScore.

 

  1. Switch to the Properties tab.

 

 

It is important to note that the “Sum of AggScore” property is not useful in the absolute sense - it should only be used to compare the predicted relative aggregation propensity of similar protein structures. There is therefore no score that would always be viewed as ‘bad’ or ‘good’ - it all depends on your baselines.

4. Comparing the Protein Surfaces

While we will explore more quantitative comparisons of the AggScore-predicted aggregation propensity later in the tutorial, we will now try a more qualitative approach of comparing the surfaces and looking for the hydrophobic patches that could be driving the aggregation.

Figure 4-1. Including the entries in the Entries table.

  1. Use Shift+Click to include MEDI-1912 and all of the mutants in the Entriesa simplified view of the Project Table that allows you to perform basic operations such as selection and inclusion.
    • 7 structures are included in the Workspacethe 3D display area in the center of the main window, where molecular structures are displayed but they are all overlapped.  

Figure 4-2. Tiling the Workspace structures.

  1. Click the icon and select Tile.
    • The Workspacethe 3D display area in the center of the main window, where molecular structures are displayed is tiled such that each included entry is in its own tile.

Figure 4-3. Zooming in to W30.

  1. In the Hierarchy, click the search toggle and type 30.
    • The Hierarchy is filtered to show the residues that are labeled 30 (or have 30 in their name).
  2. Select TRP 30 in the top entry of the searched results.
  3. Click the Fit view to selected atoms button.
    • Each tile is now zoomed to the 30 position and you can now see the impact the mutations at that position might have had on the AggScore.

5. Analyzing the Correlation Between AggScore and HP-SEC Retention

To understand how AggScore can be used to rank-order protein therapeutics based on aggregation propensity, we will plot the AggScore predicted through the Protein Surface Analyzer versus experimental HP-SEC retention times obtained from Dobson et al., 2016. This retrospective analysis is essential before using AggScore prospectively on a project.

Figure 5-1. Opening the Project Table

  1. Click the Table icon to open the Project Tabledisplays the contents of a project and is also an interface for performing operations on selected entries, viewing properties, and organizing structures and data.  

Figure 5-2. Loading patch Sum AggScore into the Project Table.

We are now going to add the patch Sum AggScore property which we calculated with the Protein Surface Analysis panel to the Project Tabledisplays the contents of a project and is also an interface for performing operations on selected entries, viewing properties, and organizing structures and data.

 

  1. In the Property Tree, search AggScore.
  2. Check the box next to patch Sum AggScore.
    • Patch Sum AggScore now appears in the Project Tabledisplays the contents of a project and is also an interface for performing operations on selected entries, viewing properties, and organizing structures and data.

Figure 5-3. Open the Chart Manager

We are now going to create a scatter plot with patch Sum AggScore and HP-SEC retention time as the axes.

 

  1. Shift+Click to select(1) the atoms are chosen in the Workspace. These atoms are referred to as "the selection" or "the atom selection". Workspace operations are performed on the selected atoms. (2) The entry is chosen in the Entries (and Project Table) and the row for the entry is highlighted. Project operations are performed on all selected entries all of the Entries in the Project Tabledisplays the contents of a project and is also an interface for performing operations on selected entries, viewing properties, and organizing structures and data.
  2. Click the Gadgets menu at the top right corner and choose Chart Manager.
    • The Manage Charts dialog box opens.

Figure 5-4. The Manage Charts dialog box.

  1. Click Create and choose Scatterplot.

 

Figure 5-5. Generating the Scatter plot.

  1. For the X-axis, choose patch Sum AggScore.
  2. For the Y-axis, choose HP-SEC.
  3. Check Best fit.
  4. Check Title.

From the scatter plot we can easily see that there is a very nice correlation between the predicted AggScore and the HP-SEC retention time. Mutants 1-3 are all single mutants of MEDI-1912 and showed only moderate improvement in both AggScore and HP-SEC retention time, the double mutants (Mutants 4-5) were even better according to both prediction and experiment, and the triple mutant (Mutant 6) was found to both have the lowest retention time and the lowest predicted AggScore.

6. Conclusion and References

In this tutorial we performed Protein Surface Analysis on the MEDI-1912 structure described in  Dobson et al., 2016. We then went over some of the properties predicted as part of that analysis, visually compared the different surfaces for the 7 protein structures, and finally plotted the predicted AggScore versus the experimental HP-SEC retention time. While we showed a few tools that could be helpful for addressing potential liabilities for biologics in this tutorial, additional workflows described in the further learning section below could be useful as well.

7. Glossary of Terms

Entries - a simplified view of the Project Table that allows you to perform basic operations such as selection and inclusion

included - the entry is represented in the Workspace, the circle in the In column is blue

incorporated - once a job is finished, output files from the Working Directory are added to the project and shown in the Entry List and Project Table

Project Table - displays the contents of a project and is also an interface for performing operations on selected entries, viewing properties, and organizing structures and data

Scratch Project - a temporary project in which work is not saved, closing a scratch project removes all current work and begins a new scratch project

selected - (1) the atoms are chosen in the Workspace. These atoms are referred to as "the selection" or "the atom selection". Workspace operations are performed on the selected atoms. (2) The entry is chosen in the Entries (and Project Table) and the row for the entry is highlighted. Project operations are performed on all selected entries

Working Directory - the location that files are saved

Workspace - the 3D display area in the center of the main window, where molecular structures are displayed