Aggregation properties of a disordered protein are tunable by pH and depend on its net charge per residue

Highlights

• Intrinsically disordered proteins lose solubility at isoelectric point (pI).

• The extent of solubility loss depends on net charge per residue (NCPR).

• Chimeric proteins with high- and low-NCPR moieties lose solubility at their average pI.

• In chimeric proteins, high-NCPR moiety drives the loss of solubility.

Abstract

Intrinsically disordered proteins (IDPs) possess a peculiar amino acid composition that makes them very soluble. Nevertheless, they can encounter aggregation in physiological and pathological contexts. In this work, we addressed the issue of how electrostatic charges can influence aggregation propensity by using the N-terminus moiety of the measles virus phosphoprotein, PNT, as a model IDP. Taking advantage of the high sequence designability of IDPs, we have produced an array of PNT variants sharing the same hydrophobicity, but differing in net charges per residue and isoelectric points (pI). The solubility and conformational properties of these proteins were analysed through biochemical and biophysical techniques in a wide range of pH values and compared with those of the green fluorescence protein (GFP), a globular protein with lower net charge per residue, but similar hydrophobicity. Tested proteins showed a solubility minimum close to their pI, as expected, but the pH-dependent decrease of solubility was not uniform and driven by the net charge per residue of each variant. A parallel behaviour was observed also in fusion proteins between PNT variants and GFP, which minimally contributes to the solubility of chimeras. Our data suggest that the overall solubility of a protein can be dictated by protein regions endowed with higher net charge per residue and, hence, prompter to respond to pH changes. This finding could be exploited for biotechnical purposes, such as the design of solubility/aggregation tags, and in studies aimed to clarify the pathological and physiological behaviour of IDPs.

Introduction

Protein aggregation is involved in a number of physiological and pathological events. Moreover, it is a major hurdle in the production and storage of recombinant proteins, included drugs. Hence, understanding the physical and chemical bases of protein aggregation could help not only to figure out how physio-pathological processes occur, but also to exploit this phenomenon for biotechnical purposes, for instance to increase in-vitro solubility of proteins [1], to design biomaterials with tunable aggregation properties [2], or even to design tags exploitable in the production of recombinant proteins [3].

How to recognize or predict protein solubility? Different definitions and criteria have been proposed, based on experimental observations, databases of soluble and insoluble proteins, or on the employment of machine-learning algorithms [4], [5], [6]. It emerges that besides sequence and structural features, the electrostatic properties of proteins, i.e. their net charge, can play a key role. It is well recognized that proteins behave as amphoteric molecules, showing significantly reduced solubility and even precipitation at their isoelectric points (pIs) [7]. On the other hand, charges can produce opposite effects. Indeed, “supercharging” of proteins, especially with negative charges, may enhance solubility [8], [9], whereas positively-charged surface patches correlate with insolubility of proteins expressed in a cell-free Escherichia coli system [10]. Systematic studies on protein solubility find obvious limitations in the disastrous structural effects induced by extensive replacement of charged residues on globular proteins. In this context, intrinsically disordered proteins (IDPs) provide a very versatile tool to extend the “host-guest approach” [11] from peptides to larger molecules, minimizing structural effects. IDPs are usually well soluble proteins lacking strict spatial constraints and compositional complexity [12], [13], [14], [15]. Due to their high designability [16], starting from a “prototypical” sequence, it is possible to generate an ideally infinite series of ad-hoc proteins sharing some properties (e.g., hydrophobicity, length, “depth” of structural disorder, etc) and differing in others (e.g., net charge, charge density and distribution etc.). IDPs have proved to be less prone to β-aggregation [17] and more stable to heat and pH than their folded counterparts [18], [19]. These features arise from the peculiar amino acid composition of IDPs and are consistent with the abundance of highly soluble residues (proline, charged and polar residues) and the paucity of aromatic and hydrophobic residues [20], [21], [22]. These properties have suggested the use of IDPs as solubility enhancers, and the hypothesis they can act as “entropic bristles” sweeping the space around the fusion protein and preventing large molecules to participate in aggregation [23]. Nevertheless, the high solubility of IDPs does not imply a lower propensity to collective interactions, such those giving rise to aggregates or coacervates. Indeed, besides hydrophobicity, also entropic factors, hydrogen bonding and electrostatic interactions can cause aggregation [17]. Moreover, the water solubility of IDPs might depend on conformational compactness that, in its turn, is influenced by the water exposure of solubility-promoting amino acids [24]. An attempt to rationalize the relationships between electrostatic charges and conformation of IDPs is represented by charge-hydropathy plots [20] and, more recently, by diagrams of states. Through these latter empirical diagrams, conformational states of IDPs have been related to fraction of charged residue (FCR) and net charge per residue (NCPR) [25], [26], [27].

In this complex scenario, we aimed to shed light on the role of charged amino acids on IDPs solubility at pI, using as a model the N-terminus moiety of measles virus phosphoprotein (PNT) [28]. We compared at various pHs the aggregation propensity of wild-type PNT (wt PNT) and synthetic variants of PNT with higher net charge and markedly more acidic or basic pI. We included in our study the green fluorescent protein (GFP), a globular protein very similar to wt PNT in terms of net charge and pI, but differing in NCPR, which is the worthiest parameter to compare the net charge of proteins of different length. Furthermore, we explored the ability of all PNT variants and GFP to reciprocally influence their solubility in chimeric constructs.

Our study shows that overall PNTs are more pH-responsive than GFP, which has lower NCPR. Among PNT variants, the loss of solubility occurs to varying degree, depending on the protein net charge. PNT variants endowed with highest NCPR promptly undergo aggregation at or near their pI, whereas low-NCPR proteins mildly react to pH, remaining mostly soluble. We further report that PNT variants “transmit” their solubility profile to chimeric constructs with GFP. This information would greatly help in the de-novo design of synthetic, disordered solubility/aggregation tags and hopefully in understanding in-vivo processes of IDP condensation and aggregation.