Engineering of a critical membrane-anchored enzyme for high solubility and catalytic activity

Muhammad S. Hussain a, Qinzhe Wang a,1, Ronald E. Viola a

Abstract

Membrane-associated proteins carry out a wide range of essential cellular functions but the structural characterization needed to understand these functions is dramatically underrepresented in the Protein Data Bank. Producing a soluble, stable and active form of a membrane-associated protein presents formidable challenges, as evidenced by the variety of approaches that have been attempted with a multitude of different membrane proteins to achieve this goal. Aspartate N-acetyltransferase (ANAT) is a membrane-anchored enzyme that performs a critical function, the synthesis of N-acetyl-L-aspartate (NAA), the second most abundant amino acid in the brain. This amino acid is a precursor for a neurotransmitter, and alterations in brain NAA levels have been implicated as a causative effect in Canavan disease and has been suggested to be involved in other neurological disorders. Numerous prior attempts have failed to produce a soluble form of ANAT that is amenable for functional and structural investigations. Through the application of a range of different approaches, including fusion partner constructs, linker modifications, membrane-anchor modifications, and domain truncations, a highly soluble, stable and fully active form of ANAT has now been obtained. Producing this modified enzyme form will accelerate studies aimed at structural characterization and structure-guided inhibitor development.

Keywords:
Membrane-Associated enzyme
Fusion partners
Aspartate N-Acetyltransferase
Protein engineering

1. Introduction

The expression and purification of mammalian membrane proteins extensive investigation [14]. In spite of these numerous approaches to remains challenging, particularly when needed for structural charac- aid in their characterization this class of proteins still constitutes less terization, where expression in higher quantities and purification to a than 5% of the protein structures deposited in Protein Data Bank (PDB), higher degree are required. This task is complicated when using an despite an estimation that membrane proteins constitute more than 60% E. coli or other bacterial expression system for enhanced expression, of the current drug targets [15]. However, through the expanded use because bacteria lack the machinery that is frequently required for the and optimization of different solubilization strategies, and concerted modification, maturation and targeting of mammalian membrane pro- studies on particularly significant targets, the number of membrane teins. However, through the use of strong promoters to improve protein structures in the PDB continues to grow. expression, modified E. coli cell lines to enhance membrane incorpora- Our target enzyme, aspartate N-acetyltransferase (ANAT), is pretion, and different tags inserted on either end of membrane proteins to dicted to be a membrane-anchored protein that catalyzes the synthesis improve solubility, bacterial strains can offer enhanced flexibility in the of N-acetyl aspartate (NAA) in neurons (Fig. 1). N-acetyl aspartate expression of eukaryotic membrane proteins [1–5]. The vast majority of (NAA) is an important metabolite in the brain, and elevation in brain the extraction and solubilization approaches for membrane proteins NAA levels is the biomarker used to confirm the presence of Canavan involves the use of detergents [6–9]. However, the selection of the disease in pediatric patients [16]. The enzyme aspartoacylase subseappropriate detergent for a particular membrane protein of interest re- quently degrades NAA into L-aspartate and acetate, the building block mains a trial-and-error approach since detergents can affect both the required for the formation of lipids in the brain. Canavan disease paoligomeric states and enzymatic activity of membrane-associated tients are deficient in a fully active form of this NAA metabolizing enzyme, resulting in a build-up in brain NAA levels [17]. However, surprisingly, normal levels of myelin lipid synthesis are observed in double knockout mice in which both the NAA synthesizing enzyme (aspartate N-acetyltransferase, ANAT) and the NAA degrading enzyme (aspartoacylase) are eliminated. These results suggest that NAA is not absolutely essential as a precursor for myelin lipid synthesis [18], and that it is the elevated levels of NAA itself [19] and not the failure to produce acetate in the brain that is the underlying cause of Canavan disease. Suppressing NAA synthesis by inhibiting the biosynthetic enzyme ANAT will reduce the elevated brain NAA levels and could subsequently prevent the loss of neurons observed in Canavan leukodystrophy. However, the complete elimination of this enzyme activity can lead to enhanced risks for patients, as seen by the premature death of homozygous double knockout mice [20]. Controlling rather than completely shutting down NAA synthesis can provide a more flexible treatment of Canavan disease, since rebalancing of the metabolic flux between NAA synthesis and degradation is less likely to have adverse consequences.
Human ANAT has been expressed and purified in a soluble form with maltose binding protein (MBP) as a fusion partner [21]. Some inhibitors of ANAT with low micromolar affinity have already been identified from the development of fragment library lead compounds [22], and also from a set of high scoring compounds selected by ATOMNET, a virtual screening platform [23]. Discovering more potent inhibitors that will selectively target ANAT would be aided by acquiring more detailed structural information about its active site. To support the recent successful stories in homology modeling of ANAT [23,24], determining the experimental structure of this NAA biosynthetic enzyme is an important goal to guide the design and optimization of the ANAT inhibitors. The importance of different protein regions and the roles of selected amino acid residues for the activity of ANAT has been suggested by mutation studies [25], yet the structural details of the functional domains of ANAT that comprise the active site are still unknown. Here we report the use of an extensive series of protein engineering approaches for the production of a stable, soluble and active form of human ANAT that will help to achieve this structural characterization goal.

2. Materials and methods

2.1. Materials

E. coli BL21 (DE3)-RIL and Rosetta (DE3) cell lines were purchased from Fisher Scientific and Novagen. Luria broth media was purchased from Research Products International (RPI). 5,5′-Dithiobis (2-nitrobenzoic acid) and all buffers were purchased from Sigma. The human nat8l gene was a generous gift from Dr. Namboodiri (Uniformed Services University, Bethesda, MD). The Pet28a plasmid was purchased from Novagen, and a modified version of the Pet28a plasmid used to express the fusion constructs was a gift from Dr. Ronning (University of Toledo). PreScission protease, purified in lab [26] and tested for activity, was used to cleave the MBP tagged proteins.

2.2. Molecular biology

The domain truncations, replacements and alternations of the membrane-anchored region and all of the other mutations were carried out by using the Q5 site-directed mutagenesis kit (New England Biolabs). The truncation constructs without an MBP tag were made using restriction cloning between the NcoI and XhoI restriction sites in the pET28a vector, whereas the truncation constructs with an N-terminal MBP tag were made by restriction cloning between the BamHI and XhoI sites in the modified pET28a vector. A figure showing the annotated sequences of all of the ANAT constructs is included in the supplementary data (supplementary Fig. 1).

2.3. Bioinformatics tools

The disordered regions in the full-length protein sequence were predicted using the PrDOS protein disorder prediction tool [27] which calculates the expected disorder in the secondary structure of proteins. The prediction was done for the full-length protein with a 5% false positive rate on a scale of probability from 0 to 1. The regions of the protein furthest above the threshold (0.5) are predicted to be increasingly disordered. The location of the membrane-anchored region for ANAT was predicted using the TMHMM server [28] that predicts the membrane region or helices in transmembrane proteins. The solubility of the different domains of ANAT were predicted from the sequence using Protein-sol, an online sequence-based protein solubility prediction tool [29].

2.4. General expression and purification of ANAT constructs

Ten ml starter cultures of each construct were grown overnight at 37 ◦C. Each starter culture was diluted into 1 L LB media and grown until the OD600 reached 0.6, followed by induction with isopropyl thiogalactopyranoside (IPTG) (0.1 mM) and continued grown at 16 ◦C for 16–20 h. Cells were then harvested by centrifugation at 6000 rpm for 10 min and resuspended in 20 mM potassium phosphate, pH 7.4, 300 mM NaCl, 10% glycerol and 20 mM imidazole. The cells were lysed by adding lysozyme (100 μg/ml) and DNA was cleaved with DNase (10 μg/ ml). The cell extracts were then clarified by centrifugation at 11000 rpm for 30 min. The supernatant was loaded onto an amylose column using an AKTA FPLC system and, after washing with buffer, the fusion protein was eluted by using a 0–10 mM maltose gradient. Fractions from the amylose column were pooled and loaded onto a Ni-NTA column for further purification and were eluted using a 20–300 mM imidazole gradient.

2.5. Expression and purification of soluble functional domains of ANAT

Two soluble and functional domains of MBP-ANAT, containing ANAT (68–302) and ANAT (78–302) were expressed in E. coli BL21 (DE3)-RIL or Rosetta (DE3) competent cells. A 10 ml starter culture of each construct was grown overnight at 37 ◦C in the presence of 50 μg/ml kanamycin. These cultures were then diluted into 1 L LB media and grown until the OD600 reached 0.6. IPTG was then added (from 0.1 to 0.5 mM) and the cultures were grown for 16–20 h at 16 ◦C. The cells were harvested by centrifugation for 15 min at 5000 rpm and resuspended in buffer A consisting of 20 mM potassium phosphate, pH 7.5, 500 mM NaCl, 10% glycerol, and 10 mM imidazole. The cells were lysed by ultrasonication for 6–10 min using lysozyme (100 μg/ml) and DNA was cleaved with DNase (10 μg/ml). Clarification of the cells was carried out by centrifugation for 30 min at 11000 rpm, followed by filtering through a 0.45 μm syringe filter. A two-step purification scheme was used for each construct. The MBP-ANAT (68–302) protein was first loaded onto a 20 ml IMAC column pre-charged with nickel chloride and then eluted with a linear gradient of Buffer B (buffer A containing 400 mM imidazole). The active fractions from the IMAC column were pooled and dialyzed against Source 30Q buffer A (50 mM HEPES, pH 7.4, 5% glycerol, 0.5 mM EDTA) overnight at 4 ◦C. The clarified protein solution was loaded onto a 15 ml Source 30Q column and ANAT was eluted by using a NaCl gradient from 0 to 500 mM. The active fractions were pooled and dialyzed overnight into the final buffer (50 mM HEPES, pH 7.4, 5% glycerol, 100 mM NaCl) and concentrated to 1 mg/ml.
A similar, but slightly different approach was used for the purification of ANAT (78–302). The MBP-ANAT (78–302) protein was first purified with an amylose column, eluted with a 0–10 mM maltose gradient, followed by cleavage of the MBP tag using PreScission protease. After digestion the protein sample was loaded onto a pre-equilibrated Talon-Sepharose superflow column, and washed with 20 mM potassium phosphate, pH 7.4, containing 300 mM NaCl, 10% glycerol and 10 mM imidazole. The ANAT (78–302) enzyme was eluted by using a 10–100 mM imidazole gradient in this phosphate buffer. Active fractions eluted from Talon-Sepharose column were pooled and dialyzed overnight into the final HEPES buffer listed above. SDS-PAGE gel was run to confirm the molecular weight and purity of each construct, and the activity of each construct was assayed by a thiol-exchange reaction using DTNB as described below.

2.6. Enzyme activity assay

Aspartate N-acetyltransferase activity was measured by monitoring the production of the thiol group [30] of coenzyme A after acetyl transfer using a SpectraMax 190 spectrophotometer. The assay contains 20 mM HEPES, pH 7.5, 150 mM NaCl, 5% glycerol, 40 μM DTNB (A412 = 14,150 M− 1 cm− 1), 40 μM acetyl-CoA, 2 mM L-aspartate, and 10 μg of enzyme in a total volume of 200 μl. The reaction is initiated by adding the substrate L-aspartate into the 96-well plate and quickly transferring the plate into the spectrophotometer. The increasing absorption at 412 nm was monitored for up to 20 min during the reaction time course. This assay is quite sensitive, and is capable of measuring rates as low as 0.01 μmol/min above background.

3. Results

3.1. Properties of MBP-ANAT fusion protein

Previous results have shown that the MBP tag that had been used to enhance the solubility of different proteins [31] is a useful fusion partner for solubilization of ANAT [24]. However, the purified MBP-ANAT fusion protein still has a propensity to aggregate at higher concentrations, and ANAT loses solubility once the MBP tag is removed. While the MBP-ANAT construct possesses good enzymatic activity, the low solubility and aggregation of the MBP-ANAT fusion protein makes it unlikely that successful crystallization of this fusion protein can be achieved. To achieve the goal of finding a suitable construct for crystallization and structural characterization extensive engineering of the enzyme has now been carried out to optimize the expression of a highly soluble and fully functional form of ANAT.

3.2. Solubilization of ANAT with new fusion partners

The tendency for aggregation of the MBP-ANAT construct suggests the need to consider using smaller solubility tags that could offer some advantages over the larger fusion partners previously investigated [32]. Thioredoxin (Trx) has previously been used successfully as a solubilizing partner [33], however ANAT showed only low expression with Trx as the solubility tag and the resulting Trx-ANAT construct was found to be highly heterogeneous. To extend the screening of potential solubilizing partners for ANAT the small ubiquitin-like modifier protein (SUMO, ~11 KDa) [34], a small metal binding protein (SmbP, ~10 KDa) [35] from Nitrosomonas europaea, and a novel peptide sequence NT11 (~1.1 KDa) from carbonic anhydrase [36] were each examined as possible N-terminal fusion tags for expressing the full-length ANAT. Unfortunately, no significant soluble expression was seen for the fusion proteins expressed with either N-terminal SUMO or NT11 tags. The SmbP-ANAT fusion protein does yield some soluble expression, however the expressed protein has only about 20% of the solubility and less than 5% of the specific activity of the MBP-ANAT enzyme form. Overall, of the fusion partners that have now been investigated (Table 1), MBP was the only solubility tag that supported expression of the full-length ANAT in a soluble and active form.

3.3. Alterations in the fusion linker region to enhance solubility

Although MBP-ANAT shows higher solubility and activity relative to the other fusion constructs, this form of ANAT is still quite heterogeneous in solution when examined by dynamic light scattering. In addition to varying the identity of the fusion partner, changing the nature of the linker used to couple MBP and ANAT can also have an effect on the overall properties of the construct. Altering the linker properties (length, flexibility, hydrophobic vs. hydrophilic) has the potential to improve the overall properties of the fusion construct [37] by allowing more intimate contacts between the protein partners. The original MBP-ANAT construct has a long (21 amino acid) and fairly flexible linker between MBP and ANAT that contains a PreScission protease recognition sequence for subsequent tag removal. To examine the effect of altering these linker characteristics on the properties of the fusion protein the linker was shortened to 4 aa (Ile-Ile-Ile-Ser, hydrophobic), 3 aa (Ala-Ala-Ala, hydrophobic), 2 aa (Gly-Ser, hydrophilic), and finally to 0 aa (no linker, nl). The purified MBP-nl-ANAT and MBP-GS-ANAT constructs each yield about 20% more soluble expression compared to the full-length MBP-ANAT. Unfortunately, these new fusion proteins were not obtained in pure form after the two-stage purification protocol. The MBP-ANAT fusion proteins with the GS or IIIS linkers, as well as the construct without any linker, each have about 70–80% specific activity when compared to the full length MBP-ANAT with the long linker, while the MBP-AAA-ANAT has slightly higher specific activity as measured by the DTNB-based assay when compared to the original construct.

3.4. Engineering of the putative membrane-anchored region to enhance the solubility of native ANAT

Since changes in the fusion protein partner and in the nature of the linker between the protein partners have failed to lead to a construct with significantly improved properties, the putative membrane- anchored region became the next focus to determine if alterations in this region would lead to a soluble enzyme form. Based on a structural model of ANAT [25] the membrane-anchored region is predicted to form a helix-turn-helix encompassing amino acids 118 to 144 as it inserts into the membrane. Each of the membrane-associated mammalian ANATs contain this central domain (Fig. 2A), while this domain is absent in related soluble N-acetyltransferases. A series of studies were carried out to change the properties of this region with the goal of reducing its binding affinity to the membrane and facilitating the folding and solubilization of the protein outside of the membrane environment.
The initial modifications in this region were designed to make this region somewhat less hydrophobic. Two Leu to Asn mutations were created by site-directed mutagenesis to convert the “SRSLLL” sequence proposed to be located near the turn in the anchor region into “SRSNLN”. However, the resulting mutated protein was only found in the insoluble pellet by Western blot analysis. Next, one turn was eliminated from each putative helix sequence through the removal of 4 amino acids on each side to shorten the length of the membrane anchor (Fig. 2B). This was followed by the removal of two helix turns (7 aa) from each side and finally the complete elimination of each proposed helix by the removal of 14 amino acids from each side to leave only an “SRS” sequence connecting the N- and C-terminal domains (Fig. 2B). Unfortunately, while good protein expression was observed, in each case the resulting ANAT enzymes were only found in the insoluble pellet. In an attempt to continue with the purification of these altered enzyme forms the insoluble cell pellets were resuspended in buffer containing from 2 to 6 M urea or guanidine-HCl. However, the refolding of ANAT into an active form after removal of these denaturants could not be achieved.
Since these modifications did not lead to a fully soluble protein, the entire membrane-anchored region was next replaced by a portion of the location and the extent of these modifications [25]. Also, the existence peptide sequence that is found in the corresponding position in a soluble of possible regions in the sequence of ANAT that could be inherently bacterial N-acetyltransferase (Fig. 2A). Three different constructs were disordered in an aqueous environment would limit soluble expression by made with either a four, eight or twelve amino acids peptide from the preventing the proper folding of this protein outside of the confines of bacterial enzyme inserted into this sequence as a replacement for the the membrane. Determination of the boundaries between the structural putative anchor region (Fig. 2B). Finally, the membrane-anchored re- and functional domains of ANAT and the identification of disordered gion was completely eliminated by constructing a circularly permutated regions in the protein sequence will aid in improving both the solubility sequence in which the C-terminus is linked to the N-terminus. Based on and the stability issues inherent with this enzyme. The amino acid the ANAT model structure [25] the two termini are predicted to be fairly sequence of human ANAT was examined through an online tool which close to each other in the folded protein, and a short and flexible “GSSG” predicts regions of potential disorder in a protein based on its amino acid linker was inserted to join these domains (Fig. 2B). These constructs sequence [27]. This algorithm suggests that a significant portion of the designed either by truncating or replacing different portions of the N-terminal domain will likely be disordered, as well as the final amino membrane region were then transformed into E. coli BL21 (DE3) cell line acids at the C-terminus (Fig. 3). for expression. Unfortunately, no significant improvement in soluble In a further attempt to improve the solubility of ANAT a series of expression was seen for any of these modified constructs, suggesting an truncation constructs were designed through the selective removal of important role for the membrane-anchored region in support of the different regions of the protein that were predicted to be inherently folding of ANAT into a stable and active structure. disordered in the N-terminal domain and at the end of the C-terminal domain (Fig. 3). Because the precise domain boundaries remain unde-

3.5. Formation of chimeric C-terminal domain

The sequence similarity and conserved residues of the ANAT C-ter- properties of each truncation construct. The truncated constructs that minal domain aligns well with polyamine N-acetyltransferase from Ba- were expressed and purified showed a range of soluble expression in cillus subtilis, and this information had previously been used to suggest E. coli cell lines when linked to an N-terminal MBP tag. A comparison of the location of the putative substrate binding site in ANAT [38]. How- five different truncation constructs (with the beginning and ending ever, unlike the mammalian ANATs, this polyamine N-acetyltransferase amino acids listed) with the full length MBP-ANAT found that the lowest is a soluble protein. To take advantage of the soluble nature of this expression level was seen for MBP-ANAT (12–302), while the highest related N-acetyltransferase the C-terminal domain of ANAT was fused level of expression was seen for MBP-ANAT (153–302) (Fig. 4). The N- with a 30-amino acid peptide from the N-terminal domain in polyamine terminal truncations starting at positions 68 and 78 were designed to N-acetyltransferase that was selected based on the sequence alignment eliminate predicted regions of disorder (Table 2), while the truncations and the presence of a stable secondary structure at the beginning of paiA (supplementary Fig. 2). Incorporation of this peptide region from paiA in place of the membrane-anchored region could have both a stabilizing and a solubilizing effect on the C-terminal domain of ANAT. Two new constructs were designed, one by appending the solubilizing peptide to the 153–302 region of the C-terminal domain of ANAT, and a second by appending this peptide to the 153–289 region of the ANAT C-terminal. The truncation at the end of the C-terminal domain was made to remove a predicted disordered region from positions 290 to 302 in the ANAT amino acid sequence (Fig. 3). While some protein expression was seen in the insoluble pellet, neither chimeric construct led to any significant soluble expression in different E. coli cell lines, despite several rounds of optimization of the expression conditions.

3.6. Solubilization of ANAT by domain truncations

Fig. 4. SDS-PAGE gel showing different expression levels of the soluble fraction of the MBP-ANAT truncated constructs. Lane 1: full length MBP-ANAT, Lane 2: an active and soluble form of ANAT can likely be attributed to reliance on only a low resolution structural model of this region to guide the The failure of membrane-anchored region modifications to produce MBP-ANAT (89= molecular weight markers. –302), Lane 5(12–302): MBP-ANAT , Lane 3:(116 MBP-ANAT –302), Lane 6(78–:302) MBP-ANAT , Lane 4(153: MBP-ANAT –302), M starting at positions 89 and 116 probed the effects of removing more substantial portions of the N-terminal domain. In general, the highest expression of soluble protein was obtained for the truncated fusion constructs comprised of only the C-terminal domain of ANAT fused with the MBP tag (supplementary Fig. 3).

3.7. Properties of truncated MBP-ANAT constructs

The stabilized constructs which showed the highest level of solubility and homogeneity were subsequently purified using the two-step affinity chromatography approach described in Methods. The constructs containing only the C-terminal domain of ANAT and an N-terminal MBP tag achieved high purity (supplementary Fig. 3), however the resulting purified proteins did not show any detectable catalytic activity (Table 2). Trimming several amino acids that were predicted to be disordered from the C-terminal end (Fig. 3) did not lead to further improvements in the properties of these N-terminal truncation constructs. The constructs comprising the amino acid sequences from 68 to 302 and 78–302 each possess good solubility in the presence of the MBP fusion partner and also have measureable catalytic activity. In particular, the specific activity of ANAT (78–302) is essentially unchanged from that of the full-length MBP-ANAT and the Km value of the substrate L-aspartic acid is also unaffected (Table 3), demonstrating that the removal of the first 77 amino acids from ANAT does not adversely affect the catalytic activity, but does result in improved stability and solubility.

3.8. Solubilization of ANAT in the absence of a fusion partner

In each of the previous cases the purified fusion protein containing only the C-terminal domain of ANAT were highly soluble, but became insoluble once the MBP tag was cleaved using PreScission protease. In an attempt to overcome this solubility problem a set of additives, stabilizing agents, and chaotropic agents were screened to identify conditions that could keep the C-terminal domain constructs soluble after cleaving the MBP tag [39,40]. Of the 17 different additives that were tested (supplementary Table 1) only five additives were identified that significantly enhanced the solubility of the C-terminal domain after cleavage of the MBP tag. In particular, a quite soluble ANAT (153–302) domain was obtained in the presence of each of these additives after cleavage and removal of the MBP fusion partner (Fig. 5).
Now that constructs of the MBP-C-terminal domain fusion protein have been produced that remained soluble without substantial aggregation, additional constructs were examined with the goal of obtaining a soluble form of ANAT that also possesses significant catalytic activity, and that remains active after removal of the solubilizing protein partner. MBP-ANAT (78–302), the most active truncated fusion construct, was digested by using PreScission protease and loaded onto a Talon Sepharose superflow column for purification and recovery of ANAT (78–302) without using any solubilizing additives. The MBP that is released does not contain a his-tag and was eliminated by washing the column with the low (10 mM) imidazole-containing phosphate buffer. ANAT (78–302) was recovered by elution with a 10–100 mM imidazole gradient, while the his-tagged PreScission protease remained bound to the column under these conditions. An SDS-PAGE gel was run to follow the PreScission protease digestion and confirm the molecular weight and purity of the ANAT (78–302) (Fig. 6). This truncated enzyme form is estimated to be >85% pure, with a small amount of undigested MBP- ANAT (78–302) still detected. Most significantly, this newly produced enzyme form is found to retain ~50% of the catalytic activity relative to the full length MBP fusion construct.

4. Discussion

4.1. General approaches to solubilizing membrane proteins

Each protein uses a unique set of internal interactions to achieve their final folded and stable conformation. Similarly, the interactions between a protein and membrane also utilize a set of unique binding interactions that are specific to each membrane-associated protein. So, perhaps it is not surprising that no single approach has been successful in extracting membrane-associated proteins into an aqueous environment in a stable and active form. Detergents have been the most commonly used extract tool for solubilizing membrane proteins, and many new classes of detergents have been produced with improved properties for the solubilization of membrane proteins while minimizing their denaturation [41]. The examination of a wide range of these improved detergents led to the identification of conditions for the extraction and solubilization of human ANAT [21]. Unfortunately, the presence of these detergents interfered with enzyme purification, and the partially purified and solubilized enzyme was not particularly stable and had a tendency to aggregate.
When detergent and lipid extraction methods fail to produce sufficient levels of important membrane proteins it has frequently become necessary to modify the protein structure through the application of different protein engineering methods ranging from specific mutations in the membrane anchor region [42] to non-specific approaches such as directed evolution [43]. Solving these solubility issues for ANAT required the investigation of a broad scope of different protein engineering approaches to ultimately achieve the goal of producing a soluble, stable and active enzyme form.

4.2. The use of solubilizing fusion partners

The full-length human ANAT shows no detectable soluble expression in E. coli in the absence of a solubilizing fusion partner, a result which is not unexpected for a mammalian membrane-anchored enzyme. Extensive expression and growth optimizations were carried out, including the use of combinations of different plasmids and cell lines for expression, including cells specifically modified for the enhanced expression of membrane proteins [5], as well as variations in both growth and induction conditions. Unfortunately, examining this broad range of conditions still failed to yield any soluble expression for ANAT. Previous studies had shown that the addition of MBP as an N-terminal soluble protein partner resulted in good expression and allowed the subsequent purification of soluble ANAT fusion constructs [21]. Unfortunately, even this improved form has a propensity to aggregate upon concentration and has remained refractory to crystallization despite an extensive screening effort. Testing some small proteins that had been successfully utilized to serve as solubilizing partners for other membrane proteins did not yield any improvements in the solubility and stability of ANAT solubilization or stabilization.

4.3. Modifications of the fusion linker region

The nature of the linker between fusion partners has been shown to influence the folding and stability of numerous proteins [37], but the improvements in protein properties that have been achieved are not universal and seem to depend on the particular protein partners involved [44]. No general guidance has emerged to be followed in optimizing the linker design within these chimeric constructs. For the MBP-ANAT fusion constructs successive shortening of the original long and flexible linker, including altering the hydrophobic/hydrophilic properties and the eventual complete elimination of the linker, had no adverse effects on protein expression or enzyme activity, but did not lead to any significant improvement in either the solubility or the stability of these constructs.

4.4. Modifications in the membrane-anchored region

Changes in the properties of the protein domain that is in direct contact with the membrane should have the most dramatic effect on the solubilization of ANAT. With this as a guiding principle an extensive series of modifications were conducted in this putative membrane- anchored region, with the aim of decreasing the affinity of ANAT for its membrane environment and shifting towards a more soluble enzyme form. However, the truncations, modifications and deletions that have been carried out in the membrane-anchored region did not lead to any significant level of soluble protein expression. These results suggest that this region is involved not only in membrane association but might also play a role in proper protein folding.

4.5. Solubilization of ANAT through domain truncations

A more aggressive modification approach was then pursued in which entire regions of this protein were either truncated or eliminated to determine their effect on the properties of this protein. The shorter constructs starting at amino acids from 89 to 153 were soluble, but had very little catalytic activity, supporting an important role for this region of the N-terminal domain in stabilizing the enzyme in its active conformation. Complete elimination of the N-terminal domain and putative membrane-anchored region (constructs starting at amino acids 143 to 153) produced protein forms that were generally quite soluble and were even able to remain soluble after cleavage of the MBP tag in the presence of certain additives. These constructs contain about half of the ANAT protein sequence and, despite their lack of catalytic activity, will be useful for the structural characterization of a major conserved domain of this enzyme.

4.6. Role of protein domains in catalytic activity

Because of the intense interest in the structure of this membrane- anchored enzyme various models of the structure of ANAT have been generated. Two earlier models [25,45] aimed to identify the issues that were preventing successful solubilization, while later models [22,23] were attempts to provide more detailed structural information. Each model predicts somewhat different sets of amino acids that are suggested to be involved in substrate binding and catalysis, but there have been only limited experimental support of the proposed roles of these amino acids. What is found in common between these different models is that none of the suggested active site amino acids are located in the N-terminal domain. Experimentally, only constructs that also contained the membrane-anchored region were found to be catalytically active. The purified ANAT (78–302) construct showed the highest activity relative to the full length MBP-ANAT enzyme, confirming that the removal of the domain from amino acids 1–77 has no significant effect on catalytic activity. Determination of the structure of ANAT (78–302) will provide insights into the active site structure and the catalytic mechanism of this enzyme.

4.7. Elimination of the fusion partner

Now that some optimized constructs of ANAT have been produced with the inclusion of a solubilizing partner, the remaining question to be addressed is whether any of these truncated constructs will remain both soluble and active if the MBP partner is removed. Cleavage at a PreScission protease site incorporated within the longer flexible linker, followed by purification of the his-tagged ANAT (78–302) construct yielded a purified and active enzyme form containing a portion of the N- terminal domain, the complete C-terminal domain, and the putative membrane-anchored region. This construct retains high catalytic activity relative to that of the initial MBP-fusion construct and has substantially improved solubility and stability.

5. Conclusions

The combination of fusion constructs with a solubilizing protein partner and extensive protein engineering of human ANAT has produced a modified form of this membrane-associated brain enzyme that retains both its solubility and its catalytic activity after removal of the solubilizing partner. From among the extensive set of truncation constructs that were produced and characterized the ANAT construct that is missing amino acids 1 to 77, but still contains the putative membrane- anchor region and the C-terminal domain, shows the best combination of improved solubility while retaining high catalytic activity. Screening of this modified ANAT has begun for the identification of selective inhibitors and for the production of diffraction-quality crystals for structural characterization of this essential metabolic enzyme.

References

[1] G. Hannig, S.C. Makrides, Strategies for optimizing heterologous protein expression in Escherichia coli, Trends Biotechnol. 16 (1998) 54–60.
[2] S. Wagner, M.M. Klepsch, S. Schlegel, A. Appel, R. Draheim, M. Tarry, M. Hogbom,¨ K.J. van Wijk, D.J. Slotboom, J.O. Persson, J.-W. de Gier, Tuning Escherichia coli for membrane protein overexpression, Proc. Natl. Acad. Sci. U.S.A. 105 (2008) 14371–14376.
[3] C.L. Young, Z.T. Britton, A.S. Robinson, Recombinant protein expression and purification: a comprehensive review of affinity tags and microbial applications, Biotechnol. J. 7 (2012) 620–634.
[4] C.R. Midgett, D.R. Madden, Breaking the bottleneck: eukaryotic membrane protein expression for high-resolution structural studies, J. Struct. Biol. 160 (2007) 265–274.
[5] B. Miroux, J.E. Walker, Over-production of proteins in Escherichia coli: mutant hosts that allow synthesis of some membrane proteins and globular proteins at high levels, J. Mol. Biol. 260 (1996) 289–298.
[6] M.I. Schimerlik, Overview of membrane protein solubilization, Curr. Protocol Neurosci. 2 (1998) 5.9.1–5.9.5.
[7] S.H. Lin, G. Guidotti, Purification of membrane proteins, Methods Enzymol. 463 (2009) 619–629.
[8] J.M. Vergis, M.D. Purdy, M.C. Wiener, A high-throughput differential filtration assay to screen and select detergents for membrane proteins, Anal. Biochem. 407 (2010) 1–11.
[9] P. Loll, Membrane proteins, detergents and crystals: what is the state of the art? Acta Crystallogr. F70 (2014) 1576–1583.
[10] B. Neumann, C.C.Y. Chang, T.Y. Chang, Triton x-100 or octyl glucoside inactivates acyl-CoA:Cholesterol acyltransferase 1 by dissociating it from a two-fold dimer to a two-fold monomer, Arch. Biochem. Biophys. 671 (2019) 103–110.
[11] B. Neumann, K. Chao, C.C.Y. Chang, T.-Y. Chang, Nanodisc scaffold Glumetinib peptide (NSPR) replaces detergent by reconstituting acyl-CoA:Cholesterol acyltransferase 1 into peptidiscs, Arch. Biochem. Biophys. 691 (2020) 108518.
[12] J.F. Bada Juarez, A.J. Harper, P.J. Judge, S.R. Tonge, A. Watts, From polymer chemistry to structural biology: the development of SMA and related amphipathic polymers for membrane protein extraction and solubilisation, Chem. Phys. Lipids 221 (2019) 167–175.
[13] P.I. Semenyuk, E.V. Moiseeva, Y.Y. Stroylova, M. Lotti, V.A. Izumrudov, V. I. Muronetz, Sulfated and sulfonated polymers are able to solubilize efficiently the protein aggregates of different nature, Arch. Biochem. Biophys. 567 (2015) 22–29. [14] T.K. Ritchie, Y.V. Grinkova, T.H. Bayburt, I.G. Denisov, J.K. Zolnerciks, W. M. Atkins, S.G. Sligar, Reconstitution of membrane proteins in phospholipid bilayer nanodiscs, in: N. Düzgünes (Ed.), Methods Enzymol, Academic Press, 2009, pp. 211–231.
[15] H. Yin, A.D. Flynn, Drugging membrane protein interactions, Annu. Rev. Biomed. Eng. 18 (2016) 51–76.
[16] W. Grodd, I. Krageloh-Mann, D. Petersen, F.K. Trefz, K. Harzer, In vivo assessment of N-acetylaspartate in brain in spongy degeneration (Canavan’s disease) by proton spectroscopy, Lancet 336 (1990) 437–438.
[17] R. Matalon, K. Michals-Matalon, M. Sebesta, M. Deanching, P. Gashkoff, J. Casanova, Aspartoacylase deficiency and N-acetylaspartic aciduria in patients with Canavan disease, Am. J. Med. Genet. 29 (1988) 463–471.
[18] J. Sohn, P. Bannerman, F. Guo, T. Burns, L. Miers, C. Croteau, N.K. Singhal, J. A.McDonough, D. Pleasure, Suppressing N-acetyl-aspartate synthesis prevents loss of neurons in a murine model of Canavan leukodystrophy, J. Neurosci. 37 (2017) 413–421.
[19] M.H. Baslow, N-acetylaspartate in the vertebrate brain: metabolism and function, Neurochem. Res. 28 (2003) 941–953.
[20] H. Maier, L. Wang-Eckhardt, D. Hartmann, V. Gieselmann, M. Eckhardt, N- acetylaspartate synthase deficiency corrects the myelin phenotype in a canavan disease mouse model but does not affect survival time, J. Neurosci. 35 (2015) 14501–14516.
[21] Q. Wang, M. Zhao, G.G. Parungao, R.E. Viola, Purification and characterization of aspartate N-acetyltransferase: a critical enzyme in brain metabolism, Protein Expr. Purif. 119 (2016) 11–18.
[22] B. Thangavelu, V. Mutthamsetty, Q. Wang, R.E. Viola, Design and optimization of aspartate N-acetyltransferase inhibitors for the potential treatment of Canavan disease, Bioorg. Med. Chem. 25 (2017) 870–885.
[23] A. Stecula, M.S. Hussain, R.E. Viola, Discovery of novel inhibitors of a critical brain enzyme using a homology model and a deep convolutional neural network, J. Med. Chem. 63 (2020) 8867–8875.
[24] I.V. Polyakov, A.E. Kniga, B.L. Grigorenko, A.V. Nemukhin, Structure of the brain N-acetylaspartate biosynthetic enzyme NAT8l revealed by computer modeling, ACS Chem. Neurosci. 11 (2020) 2296–2302.
[25] G. Tahay, E. Wiame, D. Tyteca, P.J. Courtoy, E. Van Schaftingen, Determinants of the enzymatic activity and the subcellular localization of aspartate N- acetyltransferase, Biochem. J. 441 (2012) 105–112.
[26] R. Ullah, M.A. Shah, S. Tufail, F. Ismat, M. Imran, M. Iqbal, O. Mirza, M. Rhaman, Activity of the human rhinovirus 3c protease studied in various buffers, additives and detergents solutions for recombinant protein production, PloS One 11 (2016) e0153436.
[27] T. Ishida, K. Kinoshita, Prdos: prediction of disordered protein regions from amino acid sequence, Nucleic Acids Res. 35 (2007) W460–W464.
[28] A. Krogh, B. Larsson, G. Von Heijne, E.L. Sonnhammer, Predicting transmembrane protein topology with a hidden markov model: application to complete genomes, J. Mol. Biol. 305 (2001) 567–580.
[29] M. Hebditch, M.A. Carballo-Amador, S. Charonis, R. Curtis, J. Warwicker, Protein–Sol: a web tool for predicting protein solubility from sequence, Bioinformatics 33 (2017) 3098–3100.
[30] S.K. Wright, R.E. Viola, Evaluation of methods for the quantitation of cysteines in proteins, Anal. Biochem. 265 (1998) 8–14.
[31] P. Sun, J.E. Tropea, D.S. Waugh, Enhancing the solubility of recombinant proteins in Escherichia coli by using hexahistidine-tagged maltose-binding protein as a fusion partner, in: J.T.C. Evans, M.-Q. Xu (Eds.), Heterologous Gene Expression in E. coli: Methods and Protocols, Humana Press, Totowa, NJ, 2011, pp. 259–274.
[32] J.J. Lichty, J.L. Malecki, H.D. Agnew, D.J. Michelson-Horowitz, S. Tan, Comparison of affinity tags for protein purification, Protein Expr. Purif. 41 (2005) 98–105.
[33] E.R. LaVallie, E.A. DiBlasio, S. Kovacic, K.L. Grant, P.F. Schendel, J.M. McCoy, A thioredoxin gene fusion expression system that circumvents inclusion body formation in the E. coli cytoplasm, Bio Technol. 11 (1993) 187–193.
[34] T.R. Butt, S.C. Edavettal, J.P. Hall, M.R. Mattern, SUMO fusion technology for difficult-to-express proteins, Protein Expr. Purif. 43 (2005) 1–9.
[35] B.M. Barney, R. LoBrutto, W.A. Francisco, Characterization of a small metal binding protein from Nitrosomonas europaea, Biochemistry 43 (2004) 11206–11213.
[36] T.K.M. Nguyen, M.R. Ki, R.G. Son, S.P. Pack, The NT11, a novel fusion tag for enhancing protein expression in Escherichia coli, Appl. Microbiol. Biotechnol. 103 (2019) 2205–2216.
[37] X. Chen, J. Zaro, W.-C. Shen, Fusion protein linkers: property, design and functionality, Adv. Drug Deliv. Rev. 65 (2013) 1357–1369.
[38] F. Forouhar, I.-S. Lee, J. Vujcic, S. Vujcic, J. Shen, S.M. Vorobiev, R. Xiao, T. B.Acton, G.T. Montelione, C.W. Porter, L. Tong, Structural and functional evidence for Bacillus subtilis paiA as a novel N-spermidine/spermine acetyltransferase, J. Biol. Chem. 280 (2005) 40328–40336.
[39] D.J. Leibly, T.N. Nguyen, L.T. Kao, S.N. Hewitt, L.K. Barrett, W.C. Van Voorhis, Stabilizing additives added during cell lysis aid in the solubilization of recombinant proteins, PloS One 7 (2012), e52482.
[40] S.E. Bondos, A. Bicknell, Detection and prevention of protein aggregation before, during, and after purification, Anal. Biochem. 316 (2003) 223–231.
[41] B.T. Arachea, Z. Sun, N. Potente, R. Malik, D. Isailovic, R.E. Viola, Detergent selection for enhanced extraction of membrane proteins, Protein Expr. Purif. (2012) 12–20.
[42] D.J. Scott, L. Kummer, D. Tremmel, A. Plückthun, Stabilizing membrane proteins through protein engineering, Curr. Opin. Chem. Biol. 17 (2013) 427–435.
[43] M.W. Lluis, J.I. Godfroy III, H. Yin, Protein engineering methods applied to membrane protein targets, Protein Eng. Des. Sel. 26 (2012) 91–100.
[44] T. Jin, W. Chuenchor, J. Jiang, J. Cheng, Y. Li, K. Fang, M. Huang, P. Smith, T. S. Xiao, Design of an expression system to enhance MBP-mediated crystallization, Sci. Rep. 7 (2017) 40991.
[45] P.S. Ariyannur, J.R. Moffett, P. Manickam, N. Pattabiraman, P. Arun, A. Nitta, T. Nabeshima, C.N. Madhavarao, A.M.A. Namboodiri, Methamphetamine-induced neuronal protein NAT8L is the NAA biosynthetic enzyme: implications for specialized acetyl coenzyme a metabolism in the CNS, Brain Res. 1335 (2010) 1–13.