(C) PLOS One
This story was originally published by PLOS One and is unaltered.
. . . . . . . . . .



Engineering of a chitin deacetylase to generate tailor-made chitosan polymers [1]

['Martin Bonin', 'Institute For Biology', 'Biotechnology Of Plants', 'University Of Münster', 'Münster', 'Laboratory Of Biochemistry', 'Institut Químic De Sarrià', 'University Ramon Llull', 'Barcelona', 'Antonia L. Irion']

Date: 2024-01

Chitin deacetylases (CDAs) emerge as a valuable tool to produce chitosans with a nonrandom distribution of N-acetylglucosamine (GlcNAc) and glucosamine (GlcN) units. We hypothesized before that CDAs tend to bind certain sequences within the substrate matching their subsite preferences for either GlcNAc or GlcN units. Thus, they deacetylate or N-acetylate their substrates at nonrandom positions. To understand the molecular basis of these preferences, we analyzed the binding site of a CDA from Pestalotiopsis sp. (PesCDA) using a detailed activity screening of a site-saturation mutagenesis library. In addition, molecular dynamics simulations were conducted to get an in-depth view of crucial interactions along the binding site. Besides elucidating the function of several amino acids, we were able to show that only 3 residues are responsible for the highly specific binding of PesCDA to oligomeric substrates. The preference to bind a GlcNAc unit at subsite −2 and −1 can mainly be attributed to N75 and H199, respectively. Whereas an exchange of N75 at subsite −2 eliminates enzyme activity, H199 can be substituted with tyrosine to increase the GlcN acceptance at subsite −1. This change in substrate preference not only increases enzyme activity on certain substrates and changes composition of oligomeric products but also significantly changes the pattern of acetylation (PA) when N-acetylating polyglucosamine. Consequently, we could clearly show how subsite preferences influence the PA of chitosans produced with CDAs.

Funding: This work was part of the EU programme “NanoBioEngineering of BioInspired BioPolymers (Nano3Bio)”, which was financed by the European Union’s Seventh Framework Programme under grant agreement no. 613931 (to M.B., S.P., S.CL., A.P., B.M.). A.P. was further funded from Ministry of Science and Innovation (MICINN), Spain under grant no. PID2019-104350RB-I00. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

In this study, we aimed to combine these in silico and in vitro approaches to investigate the substrate binding site of a previously described CDA from the plant-endophytic fungus Pestalotiopsis sp. [ 4 ]. We created a site-saturation mutagenesis (SSM) library of 27 residues along the binding site, which we screened according to Pascual and Planas [ 27 ]. In silico analyses then helped to narrow down the most interesting residues that were screened and analyzed in more detail. Finally, muteins, which showed different subsite preferences on oligomeric substrates, were used to N-acetylate polyglucosamine to evaluate the impact of these mutations on the PA generated on polymers.

In addition to determining the 3D structure of the enzyme–substrate complex, knowledge-based enzyme mutagenesis has been used in previous studies to investigate the function of specific residues along the substrate binding site [ 5 , 22 , 26 ]. Although proven effective, this approach is rather time consuming and, thus, limited to a small number of residues. As an alternative to the knowledge-based approach, a random mutagenesis library was successfully screened by Pascual and Planas to identify the VcCDA mutein K275E/H127R that has a significantly increased activity on chitin tetramers [ 27 ]. Besides these in vitro experiments, in silico studies are increasingly performed to analyze the binding site in more detail. These include homology modeling with substrate docking [ 5 , 28 , 29 ] and molecular dynamics (MD) simulations [ 26 , 30 – 32 ]. Despite great advancements in these fields, in silico predictions still need to be backed up by laboratory experiments, or vice versa, but they can help to understand experimental results.

So far, several key residues have been described, mainly located in the 5 conserved motifs (MT1 to MT5) defined for carbohydrate esterase family 4 (CE4) enzymes [ 22 ], to which CDAs belong. These motifs include the metal binding triad consisting of an aspartate (MT1) and 2 histidines (MT2), and the catalytic aspartate (MT1) and histidine (MT5) with their supporting residues asparagine (MT3) and aspartate (MT4), respectively. Furthermore, a hydrophobic pocket, formed by 2 residues (MT3 and MT5 or MT4 and MT5), has been described, accommodating the acetate methyl group upon hydrolysis [ 22 , 23 ]. All of these residues are located at subsite 0, i.e., the region of the binding site that receives the GlcNAc unit to be deacetylated. Residues further along the binding site at negative subsites towards the nonreducing end as well as at positive subsites towards the reducing end of the substrate have been investigated less frequently. We assume, however, that the preferences for GlcNAc or GlcN units at these nonzero subsites deeply influence the PA of the products, as suggested earlier [ 14 , 24 , 25 ]. Therefore, amino acid residues that interact with the substrate at these subsites should be studied in more detail.

Theoretically, oligomers up to DP5 with all possible PAs can be produced using the CDAs known and characterized to date [ 20 ]. In order to produce defined PAs for larger oligomers and, even further, to control the PA of polymers, new CDAs are needed, or available CDAs need to be characterized in more detail. In addition to exploiting the natural diversity of CDAs found in different organisms, an alternative approach would be the targeted engineering of known CDAs, as it has been done for other chitin-modifying enzymes [ 21 ], to alter the PA of their products. Both approaches require a detailed understanding of the structure–function relationships in these enzymes, to identify crucial molecular differences between CDAs that are responsible for the different PAs they produce.

Chitosans also have a considerable potential to be used as multifunctional biopolymers in diverse application fields. In contrast to chitin, which is insoluble in all common solvents, chitosans are soluble in slightly acidic aqueous solutions due to their higher percentage of GlcN units [ 10 ]. Thus, they can be used more easily, e.g., as plant strengthening and antimicrobial agents in agriculture or antiseptical material for wound dressings in medicine [ 11 – 13 ]. Depending on the application, the degree of polymerization (DP), the F A , and even the pattern of acetylation (PA) of the chitosan used need to be adjusted for optimum performance [ 14 , 15 ]. Both DP and F A can be controlled during the chemical production route of commercially available chitosans, but the PA remains close to a random distribution [ 16 – 18 ]. In contrast, chitosans produced using CDAs or via N-acetylation of polyglucosamines using CDAs in reverse have different, nonrandom PAs impacting the biodegradability and even bioactivity of these polymers [ 19 ]. However, this enzymatic approach to control the PA is limited by the number of CDAs characterized in detail so far, as the PA produced strongly depends on the enzyme used.

Chitin and its derivative chitosan are polymers consisting of β-1,4-linked N-acetylglucosamine (GlcNAc, A) and glucosamine (GlcN, D) units in differing ratios. Chitin, with a high percentage of GlcNAc units up to 100%, forms strong crystalline fibers, and due to its strength, it is widely found in nature as a structural building block in the exoskeletons of arthropods and insects or the cell walls of fungi and algae [ 1 , 2 ]. Due to its wide abundance in nature, it is targeted by a variety of enzymes such as chitinases found in most organisms, e.g., to cleave the chitin of invading pathogenic fungi, weakening the pathogens and at the same time generating immunogenic chitin oligomers [ 3 ]. To disguise their chitin, some fungi are thought to have evolved chitin deacetylases (CDAs), hydrolyzing the acetyl groups, thus lowering the fraction of acetylation (F A ) and yielding chitosans [ 4 ]. Chitin-to-chitosan conversion was shown to reduce immune defense reactions of host organisms [ 5 ], whereas the knock-out of CDAs was shown to increase survival of infected host organisms [ 6 ]. Beyond pathogenic fungi, CDAs are also found in chitin-degrading bacteria [ 7 , 8 ], and they are involved in chitin modulation in insects [ 9 ].

The weight average A- and D-block sizes block(A) w and block(D) w are calculated as follows: where DP is the degree of polymerization, I the relative oligomer intensity, N(A) the number of GlcNAc units, and N(D) the number of GlcN units for each oligomer i obtained by chitinosanase digestions.

After enzyme production, chitosan 134 (1 mg/ml, MW 190 kDa, F A 0.01) was incubated with 25 μg/ml of each enzyme in 1.5 M sodium acetate buffer at pH 7 at 37°C. Samples were taken after 0.33 h, 0.66 h, 1 h, 2 h, 3 h, 4 h, 6 h, 8 h, 10 h, 14 h, and 24 h, and the reaction was stopped by polymer precipitation with 1 vol. of acetone and 0.066 vol. of 1% ammonia. All samples were centrifuged for 40 min, 16,000g at 4°C, and the polymer pellet was vacuum dried. The pellet was resuspended in 0.3 vol. water and vacuum dried again 2 times before it was finally resuspended in 1 vol. 200 mM sodium acetate buffer at pH 4.2. The chemical controls were produced with chitosan 134 as the starting material as described previously [ 42 ]. The F A of all samples as well as the chemical controls (15 μg polymer per sample) was determined as described by Wattjes and colleagues [ 43 ]. For each chitosan (produced chemically or enzymatically), samples were chosen, which are close to F A 0.1, 0.2, 0.3, 0.4, and 0.5. These samples were split into 3 reactions to determine the PA via enzymatic mass spectrometric fingerprinting with chitinosanase [ 44 ] in triplicates. Samples were measured by HILIC-ESI-MS as described previously [ 45 ] and via SEC-RI-ESI-MS. For the latter method, 1.5 μg of the polymer digests were separated at 40°C with a ACQUITY UPLC Protein BEH SEC column (125 Å, 1.7 μm, 4.6 mm × 300 mm, 1 K to 80 K, Waters Corporation) using 0.4 ml/min of 150 mM ammonium acetate and 200 mM acetic acid in water as the eluent. After separation, the flow was split equally going into an ERC RefractoMax 520 (Thermo Fisher Scientific) and an ESI-MS (amaZon Speed, Bruker). The RI signal was acquired with a rate of 10 Hz, a recorder range of 512 μRIU and an integrator range of 125 μRIU/V at 40°C, while the ESI-MS was operated as described before [ 38 ].

Polyglucosamine was N-acetylated with PesCDA nm , PesCDA H199Y, and PesCDA H199K, which showed an increased and decrease GlcNAc preference at subsite −1 during our SSM library screening, respectively. These were first expressed in 500 ml LB Amp100 autoinduction medium [ 37 ] for 48 h at 26°C, 120 rpm. The cells were harvested by centrifugation at 4,000g for 20 min at 4°C, and the pellet was resuspended in 30 ml FPLC buffer (20 mM TEA, 400 mM NaCl at pH 8) and stored at −20°C. The cell mixtures were thawed at room temperature before adding 3 μl benzonase (Merck KGaA, 25 U/μl) in 250 μl 2 M MgCl 2 . After 15-min incubation at room temperature, 2 ml high salt buffer (1 M TEA, 1 M NaCl at pH 8) were added before the cells were lyzed by sonication with a Branson Digital Sonifier model 250-D (Emerson) using five 15-s pulses at 40% amplitude. After cell lysis, insoluble debris was pelleted for 60 min at 40,000g, 4°C. The enzymes were purified from the supernatant by affinity chromatography using the Strep-TactinXT purification system (IBA). Then, the enzymes were concentrated with Amicon Ultra-15 centrifugal filters (Merck KGaA) and rebuffered into FPLC buffer. Enzyme concentrations were determined using the Bradford method [ 41 ].

The muteins for 7 positions (N75, W76, Y140, F141, D164, Y168, and H199) were screened by hand using DAAA, ADAA, AADA, and AAAD as substrates as described in the previous section Detailed screening of 7 positions with A4. The HILIC-ESI-MS detection of the paCOS products was performed with a Waters Synapt XS HDMS 4k mass spectrometer coupled to a Waters ACQUITY Premier UPLC System. The oligomers were separated with the same columns and elution profile as described in the previous section Detailed screening of 7 positions with A4 only with the 0.4 min equilibration phase shifted to the beginning of each run. Solvent B and solvent A were used as the wash and purge solvent, respectively. MS 1 measurements were conducted in positive resolution mode in a mass range from m/z 650 to 900 with a scan time of 0.1 s. Mass accuracy was ensured using the proton adduct of leucine enkephalin (m/z 555.2692) as the lock mass in 10-s intervals for 1 s. The capillary voltage of the source was set to 1 kV, source and desolvation temperature to 150°C and 250°C, respectively. Cone gas and desolvation gas flow were set to 0 l/h and 750 l/h, respectively, resulting in a nebulizer pressure of 6.5 bar. MS 2 spectra were measured using the same conditions in TOF MS/MS mode with target masses set for A3D1 (m/z 789.325), A2D2 (m/z 747.314), and A1D3 (m/z 705.304) with respective collision energies of 22 V, 24 V, and 24 V at the expected elution time. The scan range was set to m/z 100 to 1,000 with an LM resolution of 4.7 for the quadrupole. The sample data were processed as described for A4, and the relative acetate release for the A3D1 substrates (raA1D3) was calculated based on the relative amounts of A2D2, A1D3, and D4 (raA2D2, raA1D3, and raD4) as follows:

Samples taken after 48 h were additionally measured via MS 2 . They were separated as described for the MS 1 measurement, and time windows were set for the corresponding target masses (A3D1: m/z 789; A2D2: m/z 747), which were fragmented both with an amplitude of 1.2 and an isolation width of m/z 3 using the MRM mode. Fragments were observed in a scan range from m/z 200 to 900. The data were converted to mzML format and analyzed with an in-house python script using pymzML v2.0 [ 39 ]. The PA was determined based on the method introduced by Cord-Landwehr and colleagues [ 40 ], assuming that mainly b- and y-ions were present.

The HILIC-ESI-MS 1 detection of chitooligosaccharide (COS) and partially acetylated COS (paCOS) products was conducted as described previously [ 38 ] with the following adaptations to reduce the time for each run. A shorter ACQUITY UPLC BEH Amide column (1.7 μm, 2.1 × 50 mm; Waters Corporation) was used with a flow rate of 0.8 ml/min, and the column oven temperature was increased to 70°C. A volume of 2 μl of each sample was injected and separated with the following elution profile: 0 to 0.8 min, 15% to 100% B; 0.8 to 1.1 min, 100% to 15% B; 1.1 to 1.5 min, 15% B. Mass spectra were acquired over a smaller scan range from m/z 650 to 900. The data were first inspected by hand using Data Analysis 4.1 (Bruker) before they were converted to mzML format and analyzed with an in-house python script using pymzML v2.0 [ 39 ]. The relative acetate release for the A4 (rarA4) was calculated based on the relative amounts of A3D1, A2D2, and A1D3 (raA3D1, raA2D2, and raA1D3, respectively) as follows:

The muteins for 7 positions (N75, W76, Y140, F141, D164, Y168, and H199) were screened again by hand using A4 as a substrate, which was produced as described previously [ 32 ]. First, 1.5 ml LB Amp100 autoinduction medium [ 37 ] in 96-deep well plates were inoculated with 10 μl from the glycerol stocks. The plates were sealed with breathable foil and incubated at 26°C, 240 rpm. After 48 h, the plates were centrifuged at 4°C, 2,500 rpm for 20 min. The supernatant was discarded and the pellets were stored at −20°C. Then, 250 μl lysis buffer (50 mM Tris-HCl, 1% Triton X-100 at pH 8) were added before shaking the plates at room temperature, 1,400 rpm for 60 min. Cell debris was removed via centrifugation at 4°C, 2,500 rpm for 20 min before transferring 175 μl of the supernatant to a new plate. This plate was centrifuged again at 4°C, 2,500 rpm for 5 min before 75 μl of the supernatant was added to 10 μl CBMs, which have been washed before as described above. After an incubation at 4°C, 600 rpm for 60 min, the CMBs were washed twice with 145 μl 50 mM TEA buffer at pH 8.5 on a magnetic plate. Finally, the reaction was started by adding prewarmed 50 μl 50 mM TEA buffer at pH 8.5 with 0.5 mM substrate to the washed CBMs. The reaction plate was sealed with sealing mats and incubated at 37°C and 600 rpm. After 2 h, 6 h, 18 h, and 48 h, the reaction plate was centrifuged for 2 to 3 min at 2,000 rpm to remove water from the sealing mat, placed on a magnetic plate, and 5 μl of each sample were mixed with 5 μl 1% formic acid. These samples were stored at 4°C or −20°C before measuring the products via HILIC-ESI-MS 1 .

All data were evaluated with an in-house script, which executed the following steps. First, a regression line is calculated for the glucosamine standard to calculate the amount of free primary amines in each well. The average activity of 4 PesCDA nm per plate was set to one, and the muteins’ activities were adjusted accordingly. Finally, the mean and standard deviation from 4 independent measurements for each mutein were calculated. As no activity or very high activity was measured in some wells, these outliers were automatically removed with the following method: If the standard deviation is higher than 0.2, the value furthest away from the mean is eliminated. This elimination is withdrawn if it does not decrease the standard deviation by more than 20%. A detailed list of all values with and without the outlier elimination can be found in S1 Data .

The screening assay is based on the method developed by Pascual and Planas [ 27 ] and was conducted on the same Automated Liquid Handling Bravo Platform from Agilent Technologies. Before running the screening with the SSM library muteins, it was first conducted using the PesCDA nm in comparison to an empty vector control using different dilutions of CMBs (see S1 Fig ). Based on these results, we decided to use a 1:10 dilution of CMBs assuming that they are saturated with enzymes to have an upper limit for enzyme concentration. Besides this adjustment, only minor changes have been made for this library screening. Instead of 25 μl, 75 μl of the supernatant after cell lysis were transferred to a new plate containing 10 μl 1:10 diluted CMBs, which were previously washed twice with water and 3 times with 50 mM TEA buffer at pH 8.5 and finally resuspended in 50 mM TEA buffer at pH 8.5. Similarly, after incubating the CMBs to allow protein binding, the beads were washed twice with 50 mM TEA buffer at pH 8.5. Finally, the reaction was performed with a final concentration of 0.5 mM A4 in 50 mM TEA buffer at pH 8.5 in a total volume of 50 μl. To quantify the amount of free primary amines, a fluorescamine-based assay was used. As a standard, 4 glucosamine concentrations (0 μM, 75 μM, 150 μM, and 250 μM) were used in triplicates. For the quantification, 70 μl of 50 mM TEA buffer at pH 8.5 were mixed with 30 μl of the reaction supernatant or the glucosamine standard in a black 96-well plate before adding 20 μl of 2 mg/ml fluorescamine in dimethylformamide (DMF). The mixture was incubated for 10 min at room temperature before 150 μl 1:1 DMF:H 2 O were added. Using the FLx800 (BioTek Instruments), the samples were mixed for 15 s before the fluorescence was measured with 360 ± 20 nm excitation and 460 ± 20 nm emission.

The PesCDA together with the VcCDA CBD’s section of the construct was codon optimized by GeneArt (Thermo Fisher Scientific). This section (PesCDA-VcCBD1+2-opt) was reintroduced into the plasmid, and the resulting fusion construct is referred to as the nonmutated PesCDA, termed “PesCDA nm ” in this article. For the library, the codons of 27 amino acid positions (sec. SSM library preparation and generation) were individually mutated by GeneArt to generate a pool of SSM PCR products containing all 19 non-wild-type amino acid substitutions for each position. Each pool of PCR products was cloned into the original plasmid, which was then transformed into E. coli BL21 (DE3) cells. For each position, 64 colonies were stored as a glycerol stock in a 96-well plate format and used for a PCR to amplify the region of interest. For the PCR, the KAPA2G Fast ReadyMix (no dye) (Roche) was used in a total volume of 5 μl. Excess dNTPs and primers were removed by adding 0.65 μl Antarctic Phosphatase Buffer, 0.1 μl Antarctic Phosphatase, and 0.03 μl Exonuclease I (NEB) and subsequent incubation at 37°C and 80°C both for 15 min. The PCR products were sequenced by GeneArt and E. coli clones carrying the plasmids with the desired mutations were replated as new glycerol stocks in 96-well plates. Each 96-well plate contained up to 76 library and 4 PesCDA nm colonies. In total, 43 muteins could not be identified during sequencing resulting in 470 muteins in the final PesCDA SSM library.

The PesCDA construct created previously [ 4 ], containing both an N- and C-terminal Strep-Tag II and the maltose binding protein (MBP) cloned into the pET22b plasmid (pET22b::NSt-MBP-CDA-CSt), was used as a template for cloning in this work. Here, the PesCDA CBD was replaced by the 2 CBDs from VcCDA (UniProt ID: Q9KSH6), for which the screening method was originally designed [ 27 ]. The resulting construct pET22b::NSt-MBP-PesCDA-VcCBD1+2-CSt (see S12 Fig for the complete sequence excluding the plasmid backbone) was cloned into E. coli Rosetta 2 (DE3) cells to express the protein. To verify the usability of this constructs, the protein was purified both via Strep-tag II affinity chromatography as well as chitin magnetic beads (CMBs) purification, before successfully testing its activity on A5.

A PesCDA (UniProt ID: A0A1L3THR9) homology model was generated with SWISS-MODEL [ 33 ] using ClCDA (PDB ID: 2IW0 [ 34 ]) as a template, including its zinc ion in the active site. As these were not part of the template, the signal peptide and the chitin binding domain (CBD) were removed, resulting in a homology model covering residues 26 to 238. The model was evaluated based on literature data for important CDA residues such as the catalytic and metal binding residues. It shows a good similarity to the AlphaFold2 model available in the AlphaFold Protein Structure Database with an RMSD = 1.341 Å on all atoms and an RMSD = 0.782 Å on the 27 SSM library amino acids (see S2 Fig ) [ 35 , 36 ]. Ligand generation and docking was performed as described previously [ 32 ]. To find 3 conformations of each ligand in binding mode [−3,0] (for nomenclature explanation, see sec. Results ), the grid box size and position were adjusted to exclude subsites +1 and +2. MD simulations with 3 conformations for each ligand and binding mode and subsequent root mean square fluctuation (RMSF), hydrogen bond, and molecular mechanics generalized Born surface area (MMGBSA) analysis were carried out as described previously [ 32 ].

Results

For a better understanding of the following results, we use several notations to indicate the substrate, its pattern of acetylation, and, if applicable, we visualize how the substrate was placed in the enzymes binding site. The substrate composition is indicated by the number of A (GlcNAc) and D (GlcN) units. Accordingly, chitotetraose consisting of 4 A units will be denoted as A4, while the first deacetylation product will be denoted as A3D1. To indicate which of the GlcNAc units was deacetylated, all units will be written out, starting at the nonreducing and proceeding to the reducing end, such as AADA. In this example, the A4 substrate had been bound to the enzyme from subsite −2 to subsite +1 (with subsite 0 being the one where deacetylation occurs), which will be written as binding mode [−2,+1]. If we refer to a specific substrate in a certain binding mode, this will be indicated with a lower case letter for the sugar unit bound at subsite 0. As an example, A4 in binding mode [−2,+1] will be written as AAaA.

It should be noted, that the term “subsite” does not necessarily imply a significant interaction between the corresponding region in the binding site and the closest sugar unit. Instead, it refers to the region closest to the indicated sugar unit relative to subsite 0 regardless of whether or not any significant interactions occur.

SSM library preparation and generation PesCDA was previously shown to deacetylate A4 at the third position from the nonreducing end in binding mode [−2,+1], producing AADA [4]. Thus, to identify interesting residues, potentially interacting with the substrate, a chitin tetramer (AAaA) was docked in silico into a PesCDA homology model in binding mode [−2,+1] (see Fig 1). Based on their close proximity to the substrate, 27 residues were chosen for mutational studies. These are either part of the conserved motifs (MTs) or 6 loops (L1 to L6) surrounding the binding site defined by the subsite capping model [30]. Both catalytic residues D46 from MT1 and H196 from MT5 were included as negative controls. Therefore, the metal binding residues from MT1 and MT2 were not included as their muteins were expected to be mainly inactive as well. From MT2, only S102 was chosen due to its close proximity to the substrate. Further residues are from L1 (Q74 to S78), MT3 and part of L3 (R137 to S142), MT4 and part of L4 (I163 to H170) as well as MT5 and the beginning of L6 (L194 to H199). For some of these residues, such as the catalytic aspartate (D46) and histidine (H196) and their supporting residues arginine (R137) and aspartate (D167), their function has been well described for other CDAs such as ClCDA [34]. These also include S102 from MT2, the hydrophobic pocked forming leucine from MT5 (L194) and the main chain of Y140 from MT3. However, for the latter, the role of the residue’s side chain was only briefly described for SpPgdA [22], even though it is highly conserved in many CDAs. To our knowledge, the structural or functional role of the remaining residues has not been investigated in detail yet, even though some seem to be conserved in many CDAs. PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 1. Amino acids in close proximity to AAaA in binding mode [−2,+1], which were chosen for the SSM library. In the structure, the substrate is shown with green carbon atoms and the sugar units are labeled according to their subsite. Loops (L1, L3, L4, and L6 according to Andres and colleagues [30]) and motifs (MT3, MT4, and MT5) are color-coded and the corresponding residues are labeled accordingly. Hidden residues are indicated by a small arrow. Single residues from MT2 (D46) and MT2 (S102) are highlighted separately in blue. In the amino acid sequence on the right (UniProt ID: A0A1L3THR9), the chosen residues are colored accordingly. In addition, the 5 CDA motifs (MT1–MT5), the catalytic CE4 domain, the signal peptide, and the chitin binding domain are indicated. https://doi.org/10.1371/journal.pbio.3002459.g001 The plasmid used for library generation was based on an available construct [4], replacing the CBDs of PesCDA by the CBD 1 and 2 of VcCDA, as the screening method used was originally designed for VcCDA [27]. We aimed to mutate all 27 chosen positions to all non-wild-type amino acids, resulting in a library size of 513 muteins of which 470 were successfully verified by sequencing.

First screening of all 27 positions This SSM library was screened according to Pascual and Planas [27] with some modifications (see sec. Complete library screening with A4 for more details). In short, the E. coli cells expressing the different muteins or the nonmutated construct termed “PesCDAnm” were grown, harvested, and lyzed in 96-well plates. The target enzymes were purified using small amounts of chitin-coated magnetic beads, based on preparatory test (see S1 Fig), aiming at saturating the beads to ensure similar enzyme concentrations in all wells. Then, the substrate was added and the plates were incubated at 37°C for 2 h before the amount of free primary amines was detected using a fluorescamine-based method. The muteins’ activities were normalized to PesCDAnm present in each plate (Figs 2 and S3). PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 2. Activity of PesCDA muteins on A4 normalized to PesCDAnm estimated by fluorescence-based quantification of free primary amines. Each column shows the activity of all available muteins at a given position, as indicated below the matrix. Missing muteins are grayed out. For orientation, PesCDAnm is included in each column at the corresponding position of the wild-type (WT) amino acid. The muteins are grouped by the properties—nonpolar, polar, charged (+/−), or aromatic—of the residue by which the WT amino acid was exchanged. Corresponding standard deviations can be found in a separate heatmap in S3 Fig (n = 3–4). All values can be found in S1 Data. https://doi.org/10.1371/journal.pbio.3002459.g002 It should be pointed out that we evaluated the results carefully, avoiding to rely on the activity levels of a single mutein without further rationalization. Due to the large number of muteins and because the library generation was mainly done manually, the results are likely to include some false positive or false negative results. Furthermore, despite the small amount of chitin-coated beads used to ensure saturation and, therefore, equal amounts of enzymes in all wells, it is still possible that some muteins were misfolded or expressed with very low yields. Consequently, the screening is not capable of distinguishing between these cases; however, if a mutation likely results in misfolding and, thus, a reduced activity, this will be further elaborated below. First, we looked at the overall activity pattern for each position. Based on these patterns and to allow better understanding of the results, the positions tested are sorted into the following 4 groups: Group 1 (D46, N75, R137, D167, H196): All muteins show no or strongly reduced activity. Group 2 (W76, Y140, Y168): Only muteins with aromatic residues show good or even PesCDAnm activities. Group 3 (P139, F141, I163, D164, T165, L194, D197, V198, H199): A group of active or inactive muteins share a common characteristic of the muteins’ residues. Group 4 (Q74, G77, S78, S102, P138, S142, L166, E169, H170, M195): Muteins’ residues of muteins with reduced or no activity do not share a common characteristic and most muteins show PesCDAnm activity levels. As expected, both catalytic residues (D46 and H196) can be found in group 1 with all muteins being inactive. This also applies to R137 stabilizing the catalytic D46, and D167, which increases the pK a value of the catalytic histidine. The seemingly high activity level of R137V is possibly a false positive result, as valine is not expected to compensate for the missing arginine. Group 1 also contains N75 as part of an elongated L1, which can be found in several fungal CDAs, all containing an asparagine in this position [5,34,46,47]. Another residue of L1, namely W76, is present in group 2, where only aromatic residues can substitute the wild-type (WT) amino acid. A similar pattern can be found for Y140 and Y168, suggesting that the aromatic character of these group 2 residues plays an important role for substrate binding. The activity patterns for both group 1 and group 2 suggest that these residues play an important role for enzyme activity, either by participating in the catalytic mechanism or by substrate binding. A more detailed analysis of their role will be elaborated in the next section In silico studies of PesCDA. Residues that belong to group 3 can be replaced with several other residues, still resulting in active enzymes; however, some or even most substitutions strongly reduce their activity. This applies to P139, which is conserved among most CDAs and can only be replaced by very small amino acids, indicating spatial constraints at this position. F141 can be replaced by larger hydrophobic residues and, to a certain extent, by other aromatic residues. This could match the role of the aligning leucine in ClCDA, which forms a hydrophobic pocket for the acetyl methyl group at subsite 0 [34]. The opposite site of this hydrophobic pocket is formed by L194, which cannot be replaced without losing activity in this screening. The 3 consecutive positions I163, D164, and T165 as part of MT4 also show distinct activity patterns. D164 can only be substituted with the other negatively charged amino acid without strongly losing activity, indicating that the negative charge is needed for enzyme activity, likely for enzyme substrate interaction as further discussed in the following section. In contrast to D164, the side chains of the 2 neighboring residues I163 and T165 point towards the inside of the enzyme (Fig 1), probably resulting in misfolding if the substituted amino acids are too big or charged. The last 3 residues from group 3 are part of L6 following MT5. D197 cannot be replaced by any amino acid without losing activity and most muteins are even inactive. It seems that only a glutamate can somewhat replace the WT residue, while the other substitutions that still result in activity are very small amino acids, likely to not interfere with protein folding at this position. The activity pattern at position V198 seems surprising as some muteins show activity levels above 0.5 (i.e., half of the PesCDAnm’s activity), whereas this is not the case for similar residues such as isoleucine or leucine. The high activity of V198P hints at a structural function where the proline is the only other amino acid that facilitates the tight turn at this position, whereas other amino acids likely disturb the folding to different extents. H199 is another interesting position with only a few nonpolar amino acid substitutions strongly reducing the activity at this position, while especially polar and charged residues retain PesCDAnm activity levels. As this position shows additional interesting activity patterns in further analyses, the role of this position will be discussed in more detail below. All positions that belong to group 4 show PesCDAnm activity levels for most of their muteins. The 3 positions Q74, G77, and S78 form the beginning and end of L1 and are all surface exposed residues, seemingly not crucial for the correct folding of this elongated loop. The same applies to P138 in MT3, S142 in L3 as well as E169 and H170 at the beginning of L4. Even though the remaining 3 residues S102, L166, and M195 are all enclosed by highly conserved residues of MT2, MT4, and MT5, respectively, they seem to have a negligible impact on protein activity. The activity patterns observed for all positions in this group indicate that the WT amino acids do not have a distinct feature necessary for enzyme activity. Only a few substitutions reduce enzyme activity, suggesting a negative effect of these substitutions rather than a missing positive effect caused by the WT amino acids. Using this screening method, which gave valuable insight into the muteins’ activities on the fully acetylated chitin tetramer A4, we screened a subset of the SSM library using the mono-deacetylated chitosan tetramer ADAA as well. However, due to the high background fluorescence of the substrate and the comparatively low fluorescence increase upon enzyme activity, the results showed large standard deviations making them very difficult to evaluate. Furthermore, PesCDAnm has a reduced activity on ADAA compared to A4, and, thus, testing different reaction times would be needed to find activity patterns that could be compared properly between the 2 substrates. Thus, we decided to stop the reaction at different times and analyze the products via high pressure liquid chromatography–mass spectrometry (HPLC-MS). As this approach requires dramatically increased analysis time, we first evaluated the experimental results obtained so far in more detail using in silico tools, to further narrow down the most interesting positions to be eventually screened with A4 as well as several mono-deacetylated A3D1 substrates.

[END]
---
[1] Url: https://journals.plos.org/plosbiology/article?id=10.1371/journal.pbio.3002459

Published and (C) by PLOS One
Content appears here under this condition or license: Creative Commons - Attribution BY 4.0.

via Magical.Fish Gopher News Feeds:
gopher://magical.fish/1/feeds/news/plosone/

You are viewing proxied material from magical.fish. The copyright of proxied material belongs to its original authors. Any comments or complaints in relation to proxied material should be directed to the original authors of the content concerned. Please see the disclaimer for more details.