https://orcid.org/0000-0001-7623-0935
Abstract
Aggrescan3D (A3D) standalone is a multiplatform Python package for structure-based prediction of protein aggregation properties and rational design of protein solubility. A3D allows the re-design of protein solubility by combining structural aggregation propensity and stability predictions, as demonstrated by a recent experimental study. It also enables predicting the impact of protein conformational fluctuations on the aggregation properties. The standalone A3D version is an upgrade of the original web server implementation—it introduces a number of customizable options, automated analysis of multiple mutations and offers a flexible computational framework for merging it with other computational tools.
A3D standalone is distributed under the MIT license, which is free for academic and non-profit users. It is implemented in Python. The A3D standalone source code, wiki with documentation and examples of use, and installation instructions for Linux, macOS and Windows are available in the A3D standalone repository at https://bitbucket.org/lcbio/aggrescan3d.
1 Introduction
Protein biopharmaceuticals are providing effective therapies for human diseases. Aggregation is a major limitation in the production and storage of these life-saving proteins (Hamrang et al., 2013). Therefore, there is a strong interest in tools that can predict aggregation and assist the design of soluble therapeutic proteins.
Aggregation is driven by permanent or transient exposure of aggregation prone regions (APRs), usually enriched in hydrophobic residues. A majority of algorithms identify and score APRs relying only on protein sequence. Those programs find difficulties predicting APRs of folded globular proteins, failing to detect APRs when residues are not contiguous in sequence or mistaking APRs for the buried hydrophobic core. Such limitations can be overcome by using information about protein structure, as implemented in Aggrescan3D (A3D) method for structure-based prediction of aggregation properties (Zambrano et al., 2015).
The A3D method has been made available in 2015 as a web server (Zambrano et al., 2015). Since its publication, A3D has reached several thousands of unique visits and has been applied in numerous aggregation prediction studies. Recently, an experimental study has shown that A3D predictions allow for designing mutations that improve proteins solubility without compromising their conformation or stability (Gil-Garcia et al., 2018). The solubility of unrelated polypeptides was easily tuned by A3D-designed non-destabilizing mutations at the proteins’ surfaces.
A3D was developed as an extension of the widely used, sequence-based, Aggrescan method (Conchillo-Sole et al., 2007; de Groot et al., 2012), the first algorithm using empirical in vivo data for calibrating an intrinsic amino acid aggregation scale. A3D adjusts Aggrescan’s scoring method taking into account the structural context, providing a unique structurally corrected aggregation value (A3D score) for each individual amino acid in the protein. Additionally, A3D method allows to introduce selected mutations and assess their impact on aggregation properties (using A3D score) and protein stability (using FoldX force field) (Buß et al., 2018; Schymkowitz et al., 2005). A3D method offers two modes of action: static mode; in which it analyses directly the input protein structure, and dynamic mode in which the analysis is performed on a set of protein models that reflect the flexibility of the input structure [predicted by CABS-flex method (Jamroz et al., 2013; Kurcinski et al., 2018; Kuriata et al., 2018)].
In this work, we present the A3D standalone, a fully functional implementation of A3D that is intended to work locally, therefore, addressing the important aspect of data privacy. A3D standalone allows users to incorporate the calculations into their own pipelines and control over every stage of the modeling process.
2 Features
A3D standalone is implemented as a Python 2.7 package, which can be installed via Python package managers (conda and pip) or as a docker image file and works on Linux, macOS and Windows operating systems. The program ships with a built-in local server which gives the user a graphical interface similar to that of a web server via their web browser and allows one to run simulations as well as to analyze their results easier with tools such as interactive plots or 3D molecule display.
The A3D’s pipeline is presented in Figure 1. The only required input is the protein structure data in the Protein Data Bank (PDB) file format. A3D incorporates other software packages into its pipeline: FoldX (Schymkowitz et al., 2005) for structure optimization, mutational design and scoring of structure stability and CABS-flex for fast simulations of protein flexibility (Kurcinski et al., 2018). Their use is not required, they can be replaced by any other software, or the A3D score calculations can be run without these external modules.
Aggrescan3D pipeline
A3D standalone offers extensive integration with the CABS-flex standalone package (Kurcinski et al., 2018) including interactive analysis and visualization of generated protein models and control over the dynamic mode simulation.
The program’s output depends on the chosen options. For example, using naccess algorithm (option) for calculations of solvent-accessible surface area instead of freeSasa (default) (Mitternacht, 2016) can result in small score differences. A3D returns a comma-separated values file with the scores for each residue and a score plot. Additional options might also yield a FoldX optimized PDB file, CABS-flex generated models in PDB format as well as score tables for the models, a file containing energy difference between the input and the mutant and more. Of special interest is the ‘automated mutations’ option, which automatically identifies the strongest APRs in the structure and suggests a series of point mutations that would increase the protein solubility without impacting its stability.
Instructions of use, description of options, installation instructions, along with working examples can be found in the A3D standalone repository available at https://bitbucket.org/lcbio/aggrescan3d.
3 Conclusions
The A3D standalone extends the capabilities of the original web server (Zambrano et al., 2015), and provides an alternative to other structure-based methods (Sormanni et al., 2015; Van Durme et al., 2016), for the prediction and rational re-design of protein aggregation propensity.
Thanks to its customizability, modularity and implementation in Python 2.7, A3D standalone can be easily incorporated into other pipelines for in silico design of antibodies (Sormanni et al., 2018) and other therapeutic proteins.
Funding
This work was supported from Ministry of Economy and Competitiveness, Spain [BIO2016 78310-R]; and the National Science Center (NCN, Poland) Grant [MAESTRO2014/14/A/ST6/00088].
Conflict of Interest: none declared.
References
