The functional and structural characterization of Trichoderma reesei dehydrogenase belonging to the PQQ dependent family of Carbohydrate-Active Enzymes Family AA12

21 AEM Accepted Manuscript Posted Online 11 October 2019 Appl. Environ. Microbiol. doi:10.1128/AEM.00964-19 Copyright © 2019 Turbe-Doan et al. This is an open-access article distributed under the terms of the Creative Commons Attribution 4.0 International license. on M arch 5, 2020 by gest ht://aem .sm .rg/ D ow nladed fom

2 Pyrrolo-quinoline quinone (PQQ) is an ortho-quinone cofactor of several prokaryotic 22 oxidases. Widely available in diet and necessary for the correct growth of mice, PQQ has 23 been suspected to be a vitamin for eukaryotes. However, no PQQ-dependent eukaryotic 24 enzyme had been identified to use the PQQ until 2014, when a basidiomycete enzyme 25 catalyzing saccharide dehydrogenation using PQQ as a cofactor was characterized and served 26 to define Auxiliary Activity family 12 (AA12). Here we report the biochemical 27 characterization of the AA12 enzyme encoded by the ascomycete Trichoderma reesei 28 (TrAA12). Surprisingly, only weak activity against uncommon carbohydrates like L-fucose or 29 D-arabinose was measured. The three-dimensional structure of TrAA12 reveals important 30 similarities with bacterial glucose dehydrogenases (s-GDH). The enzymatic characterization 31 and the structure solved in the presence of calcium confirms the importance of this ion in 32 catalysis, as observed for s-GDH. The structural characterization of TrAA12 was completed 33 by modeling PQQ and L-fucose in the enzyme active site. Based on these results, family 34 8 crystals were obtained by one to five minutes soak in reservoir solution added with 1M of 164 potassium iodide followed by a back soak and then freezing. Several soaks were done using 165 calcium, PQQ and L-fucose to obtain a holoenzyme structure. The final concentrations in the 166 drop during the soaking were approximately 1mM CaCl 2 , 5μM PQQ and 500mM L-fucose. 167 Crystals were collected at the French national synchrotron facility, Soleil, on the Proxima 2 168 beamline for the native dataset, and on the Proxima 1 beamline for the iodide crystal. The 169 crystals containing calcium were collected at ESRF on the ID30B beamline. 170

Structure determination and refinement 171
The crystals belong to P4 3 2 1 2 space group with cell parameters a=b=83 Å, c=143 Å. One 172 protein is present in the asymmetric unit. Single-wavelength Anomalous Dispersion was used 173 to obtain the phase. Data were collected at a wavelength of 1.653 Å. The initial phases were 174 calculated with Phaser suite program (29,30). Phase improvement was done by Parrot in the 175 CCP4 program suite (30). Only one iodide ion was found in the structure. The first model was 176 built by buccaneer and refined with Refmac5 in the CCP4 program suite (30-32). Coot was 177 used to finalize the model building (33). The other structures were solved by molecular 178 replacement using Refmac5 and coot programs (30,32,33). The assignment of the two 179 calcium ions was verified by using the web server CheckMyMetal (34). 180

Molecular docking studies 181
To estimate the binding mode of the cofactor with the substrate, molecular docking studies 182 were performed using the Autodock Vina Program (35). The cofactor (PQQ) and substrate 183 (L-fucose: α-L-fucopyranose) were built and minimized with Gasteiger charges in UCSF 184 Chimera ( (17). Among the analyzed sequences, this modularity with three domains is rarely observed 226 and seems to be mainly occurring in the Agaricales order of the Basidiomycota. On the other 227 hand, the bimodular structure containing the AA8 domain and the AA12 cannot be linked to a 228 particular taxonomic clade and is more frequently observed than the AA8-AA12-CBM1 229 Various monosaccharides were tested to evaluate the enzyme activity and specificity. Several 238 monosaccharides were tested in standard condition requiring calcium and PQQ as cofactor. 239 No detectable activity against D-glucose or D-fucose was found. TrAA12 oxidizes 240 preferentially L-fucose, and a weak activity was measured against D-arabinose (11.6% 241 relative activity), D-galactose (5.7%), L-arabinose (2.8%) and D-lyxose (2.2%) (Fig2A). The 242 activity against D-glucosone was also tested but the activity of TrAA12 was 10 times lower 243 than that measured with L-fucose. Unfortunately, the reductive activity of D-glucosone on 244 cytochrome c was not stable enough to yield a reliable activity. The kinetic constants 245 determined in steady-state for L-fucose in standard conditions revealed a Km of 99.9 (± 9.7) 246 mM and a k cat of 0.012 (±0.0003) s -1 , resulting in a low catalytic efficiency k cat /Km of 0.119 s -1 247 M -1 . These kinetic data assume the full-occupancy of the cofactor, which under these 248 experimental conditions remains uncertain. The temperature profile was determined up to 249 80°C, and TrAA12 activity increased gradually with the temperature between 30 and 80°C 250 (Fig2B). The thermostability of the purified TrAA12 was tested in the range 0 -75 °C. 251 TrAA12 was stable up to 50°C and its activity decreased by 20% after heating at 60°C for 30 252 min. No activity was found after 30 min incubation at 70°C (Fig2C). The optimal pH was 4.5. 253 When the pH was below 3.5 and above 7, only 50% and 10% of the maximum activity were 254 attained respectively (Fig2D). For stability in relation to pH, TrAA12 was stable from pH 3.5 255 to pH 8. Below pH 3.5, only 50% of the activity could be measured (Fig2E). 256

Three-dimensional structures of TrAA12 257
Three structures of TrAA12 were solved, namely that of the iodide adduct for phasing, of the 258 native protein and of the catalytic calcium-containing protein (PDB codes 6I1Q, 6H7T and 259 on March 5, 2020 by guest http://aem.asm.org/ Downloaded from 6I1T respectively). The three structures superimposed very well and no major differences in 260 the main chains were observed. The polypeptide chain observable in the crystal structure of 261 TrAA12 covers amino acids Gln24 to Asn435. This domain folds as a 6-bladed β-propeller 262 (Fig3A). This fold is frequently found in various enzymes families where the number of 263 blades varies from 4 to 12 (42). According to the common terminology, each blade of 264 TrAA12 is numbered from 1 to 6 and contains four antiparallel β strands named from A to D 265

Molecular docking of PQQ and L-fucose on TrAA12 336
Because no density was observed for L-fucose and only a partial density was obtained for 337 PQQ in our crystals, molecular docking studies were initiated to confirm the binding position 338 of PQQ and to estimate the putative position of the substrate. The docking simulation was 339 done in two steps. Firstly, PQQ was docked and the best complex protein-cofactor was kept.

Downloaded from
In the s-GDH family, the oxidation mechanism involves a direct hydride transfer between 384 substrate C1 and the C5 of the PQQ followed by a general based-catalyzed proton abstraction 385 (37). Based on the quaternary structure of A. calcoaceticus s-GDH, His144 is the best 386 candidate catalytic base, functionally assisted by calcium and Arg228 to enhance the 387 reactivity of the C5-O5 bond of PQQ (12, 47). The orientation of the L-fucose, found by 388 docking simulation on TrAA12, is the same as that observed for the D-glucose in the structure 389 of A. calcoaceticus s-GDH (PDB code 1CQ1) (37). The proton at C1 position of the L-fucose 390 is well oriented and points down towards C5 of PQQ, which would be coherent with a 391 catalytic mechanism similar to that of s-GDH. Moreover, the equivalent of His144 in A. 392 calcoaceticus s-GDH, His153, is close enough to the anomeric carbon of L-fucose (3 Å) to 393 act as the general base. In addition to the catalytic histidine, the two amino acids surrounding 394 the PQQ ketones (Arg228, Asn229 in A. calcoaceticus s-GDH) and potentially important for 395 the catalytic activity are also conserved in TrAA12 (Arg220, Asp221). If TrAA12 operates 396 with the same catalytic mechanism as s-GDH, then His153 would be the catalytic base. 397 However, the phylogenetic analysis shows that this histidine is not strictly conserved in 398 family AA12 (Fig7). In more than 25% of the sequences, this histidine is not conserved and is 399 frequently replaced by a leucine. Interestingly, the conservation rate of His153 is directly 400 linked to the subgroup classification. More than 93% of the sequences of the mixed subgroup 401 harbor the conserved putative catalytic histidine whereas the rate drops to 78% for the 402 sequences belonging to the Ascomycota and to only 30% for the Basidiomycota sequences. In 403 comparison, the couple Arg/Asx (Arg220, Asp221 in TrAA12) which completes the PQQ 404 interaction with the reactive ketones is more conserved (around 94% for each) (Fig7). The 405 conservation of Arg/Asx dyad in family AA12 suggests an implication of these amino acids in 406 catalysis. The structural similarities observed between TrAA12 and s-GDH suggest that the 407 two families have a common evolutionary ancestor. However, the high sequence divergence 408 on March 5, 2020 by guest http://aem.asm.org/ Downloaded from  Various temperatures ranging from 30°C to 80 °C and pH values (3.5 to 7) were tested under 642 standard conditions. The effects of temperature (C) and pH (E) on the stability of the purified 643 protein after 1 hour in 100mM tartrate buffer (pH 2-3), 100mM sodium acetate buffer (pH 644 3.5-6), 100mM HEPES (pH 7-8). All assays were performed with L-fucose as substrate.