Structural Characterization of Two Short Unspecific Peroxygenases: Two Different Dimeric Arrangements

Unspecific peroxygenases (UPOs) are extracellular fungal enzymes of biotechnological interest as self-sufficient (and more stable) counterparts of cytochrome P450 monooxygenases, the latter being present in most living cells. Expression hosts and structural information are crucial for exploiting UPO diversity (over eight thousand UPO-type genes were identified in sequenced genomes) in target reactions of industrial interest. However, while many thousands of entries in the Protein Data Bank include molecular coordinates of P450 enzymes, only 19 entries correspond to UPO enzymes, and UPO structures from only two species (Agrocybe aegerita and Hypoxylon sp.) have been published to date. In the present study, two UPOs from the basidiomycete Marasmius rotula (rMroUPO) and the ascomycete Collariella virescens (rCviUPO) were crystallized after sequence optimization and Escherichia coli expression as active soluble enzymes. Crystals of rMroUPO and rCviUPO were obtained at sufficiently high resolution (1.45 and 1.95 Å, respectively) and the corresponding structures were solved by molecular replacement. The crystal structures of the two enzymes (and two mutated variants) showed dimeric proteins. Complementary biophysical and molecular biology studies unveiled the diverse structural bases of the dimeric nature of the two enzymes. Intermolecular disulfide bridge and parallel association between two α-helices, among other interactions, were identified at the dimer interfaces. Interestingly, one of the rCviUPO variants incorporated the ability to produce fatty acid diepoxides—reactive compounds with valuable cross-linking capabilities—due to removal of the enzyme C-terminal tail located near the entrance of the heme access channel. In conclusion, different dimeric arrangements could be described in (short) UPO crystal structures.


Introduction
Unspecific peroxygenases (UPOs) are enzymes catalyzing the oxyfunctionalization of both aromatic and aliphatic molecules [1][2][3], including specific products that are difficult to be obtained only by chemical means. Therefore, they are of high biotechnological interest [4][5][6][7] for the regio-and/or stereoselective synthesis of biobased compounds and other fine chemicals and pharmaceuticals. These enzymes form a group of fungal hemethiolate peroxidases (HTPs). Hence, they share a proximal cysteine residue acting as the fifth ligand of the heme iron with the phylogenetically distant cytochrome P450 monooxygenases (P450s) [8], present in all living organisms. This results in similar reaction chemistry but UPOs, as self-sufficient secreted peroxygenases, catalyze selective oxygenation reactions with advantages over P450s [9]. The latter often require an auxiliary flavin-containing nucleotide sequences using the Optimizer software [39], cloned into the pET23a plasmid, and transformed into E. coli BL21 cells, which were grown in an auto-induction medium [40] at 16 • C for 4 days to promote soluble protein production. After cell lysis and debris removal, the recombinant enzymes were purified by a combination of ion-exchange and size-exclusion chromatography (SEC) using an Äkta (GE Healthcare, Chicago IL, USA) fast liquid chromatography system. UPO fractions were identified by their absorbance around 420 nm, and activity was followed by 2,6-dimethoxyphenol (DMP) oxidation in the presence of H 2 O 2 , under the conditions described below.
In the case of rMroUPO, the first purification step was on a 6 mL Resource™ Q cartridge using 10 mM phosphate (pH 7), supplemented with 1 M NaCl for elution. The appropriate fractions were pooled, concentrated, desalted, and loaded into a Mono Q high-resolution 5/5 column under the same conditions. Finally, the appropriate fractions were concentrated and, using 10 mM citrate (pH 4), dialyzed and loaded into a Mono S high-resolution 5/5 column and eluted under a NaCl gradient.
In the case of rCviUPO, the first step was cation exchange chromatography with a HiTrap SPFF column (GE Healthcare, Chicago, IL, USA) in 10 mM Tris (pH 7.4). The proteins, eluted with a gradient of the same buffer supplemented with 1 M NaCl (single peak with absorbance at 420 nm), were concentrated in an Amicon 3K device (Sigma-Aldrich, Saint Louis, MO, USA). The second step (to ensure purity) was SEC with a Superdex 75 column (10/300 GL; GE Healthcare, Chicago, IL, USA) in 10 mM Tris (pH 7.4) containing 0.15 M NaCl.
Enzyme purity was confirmed under denaturing conditions by 12% polyacrylamide gel electrophoresis (PAGE) in the presence of 0.1% sodium dodecyl sulfate (SDS) and 1% mercaptoethanol (5% in the loading buffer) [41]. Detection of the dimeric forms of the pure enzymes was performed also in 12% PAGE without any sample heating and using a loading buffer without any denaturing agent. Proper folding and binding of the cofactor were verified by analyzing the UV-visible spectrum of the enzymes (at the resting state) in 10 mM Tris (pH 7.4) using a Cary 60 spectrophotometer (Agilent, Santa Clara, CA, USA) ( Figure S1, black profiles). Additionally, formation of the characteristic complex between reduced (ferrous) heme-thiolate enzymes (such as P450s and UPOs) and carbon monoxide (CO) was assessed in 0.2 M phosphate (pH 8) after addition of Na 2 S 2 O 4 and CO flushing ( Figure S1, red profiles). Concentrations were calculated using the rMroUPO and rCviUPO molar extinction coefficients of

Protein Expression and Purification
The production and purification of rMroUPO and rCviUPO as active proteins in E. coli were reported by Carro et al. [34] and Linde et al. [33], respectively. In short, the amino-acid sequences of MroUPO [32] and CviUPO [38] were converted into optimized nucleotide sequences using the Optimizer software [39], cloned into the pET23a plasmid, and transformed into E. coli BL21 cells, which were grown in an auto-induction medium [40] at 16 °C for 4 days to promote soluble protein production. After cell lysis and debris removal, the recombinant enzymes were purified by a combination of ion-exchange and size-exclusion chromatography (SEC) using an Äkta (GE Healthcare, Chicago IL, USA) fast liquid chromatography system. UPO fractions were identified by their absorbance around 420 nm, and activity was followed by 2,6-dimethoxyphenol (DMP) oxidation in the presence of H2O2, under the conditions described below.
In the case of rMroUPO, the first purification step was on a 6 mL Resource™ Q cartridge using 10 mM phosphate (pH 7), supplemented with 1 M NaCl for elution. The appropriate fractions were pooled, concentrated, desalted, and loaded into a Mono Q high-resolution 5/5 column under the same conditions. Finally, the appropriate fractions were concentrated and, using 10 mM citrate (pH 4), dialyzed and loaded into a Mono S high-resolution 5/5 column and eluted under a NaCl gradient.
In the case of rCviUPO, the first step was cation exchange chromatography with a HiTrap SPFF column (GE Healthcare, Chicago, IL, USA) in 10 mM Tris (pH 7.4). The proteins, eluted with a gradient of the same buffer supplemented with 1 M NaCl (single peak with absorbance at 420 nm), were concentrated in an Amicon 3K device (Sigma-Aldrich, Saint Louis, MO, USA). The second step (to ensure purity) was SEC with a Superdex 75 column (10/300 GL; GE Healthcare, Chicago, IL, USA) in 10 mM Tris (pH 7.4) containing 0.15 M NaCl.
Enzyme purity was confirmed under denaturing conditions by 12% polyacrylamide gel electrophoresis (PAGE) in the presence of 0.1% sodium dodecyl sulfate (SDS) and 1% mercaptoethanol (5% in the loading buffer) [41]. Detection of the dimeric forms of the pure enzymes was performed also in 12% PAGE without any sample heating and using a loading buffer without any denaturing agent. Proper folding and binding of the cofactor were verified by analyzing the UV-visible spectrum of the enzymes (at the resting state) in 10 mM Tris (pH 7.4) using a Cary 60 spectrophotometer (Agilent, Santa Clara, CA, USA) ( Figure S1, black profiles). Additionally, formation of the characteristic complex between reduced (ferrous) heme-thiolate enzymes (such as P450s and UPOs) and carbon monoxide (CO) was assessed in 0.2 M phosphate (pH 8) after addition of Na2S2O4 and CO flushing ( Figure S1, red profiles). Concentrations were calculated using the rMroUPO and rCviUPO molar extinction coefficients of ℇ420 115.0 and 114.2 mM −1 ·cm −1 , respectively.

Site-Directed Mutagenesis
To investigate MroUPO dimerization through the intermolecular disulfide bridge visible in the crystal structure, the C227A mutation was introduced at its C-terminal tail using the Expand Long Template PCR kit from Roche (Basel, Switzerland) and the following primer (direct sequence with the mutated triplets underlined): 5'-CCA ACT GGC GAT AAC GCC GGC GCT ATC GTT CTC.
The above tail region was not solved in the rCviUPO crystal structure. However, to clarify its dimerization mechanism, (i) the single substitution C235A variant (changing the residue equivalent to MroUPO Cys227 after sequence alignment) and (ii,iii) the tail deletion K228stop and P201stop variants (including the TAA stop codon) were generated using the above PCR kit and the following primers (direct sequences with the mutated triplets underlined): C235A, 5'-GGC AAA CAG GCA CCG CAA GCC C; K228stop, 5'-GCG TTA GTG ACC CCG GAA AAA CTG ATC GAC TAA; and P201stop, 5'-GGT TGG CGT CCG TAA AAA GCT GAA CTG. 420 115.0 and 114.2 mM −1 ·cm −1 , respectively.

Site-Directed Mutagenesis
To investigate MroUPO dimerization through the intermolecular disulfide bridge visible in the crystal structure, the C227A mutation was introduced at its C-terminal tail using the Expand Long Template PCR kit from Roche (Basel, Switzerland) and the following primer (direct sequence with the mutated triplets underlined): 5 -CCA ACT GGC GAT AAC GCC GGC GCT ATC GTT CTC.
The above tail region was not solved in the rCviUPO crystal structure. However, to clarify its dimerization mechanism, (i) the single substitution C235A variant (changing the residue equivalent to MroUPO Cys227 after sequence alignment) and (ii,iii) the tail deletion K228stop and P201stop variants (including the TAA stop codon) were generated using the above PCR kit and the following primers (direct sequences with the mutated triplets underlined): C235A, 5 -GGC AAA CAG GCA CCG CAA GCC C; K228stop, 5 -GCG TTA GTG ACC CCG GAA AAA CTG ATC GAC TAA; and P201stop, 5 -GGT TGG CGT CCG TAA AAA GCT GAA CTG.
The PCR products were digested with DpnI and transformed into E. coli DH5α for propagation. Different rMroUPO and rCviUPO variants were produced in E. coli as active cytosolic proteins and purified as described for the corresponding native enzymes.

Crystallization, Data Collection, and Processing
For the screening of crystallization conditions, the purified recombinant enzymes were extensively dialyzed at 4 • C against 10  Diffraction data were collected at the BL13-XALOC beamline of the ALBA Synchrotron (Cerdanyola del Vallés, Barcelona, Spain). Crystallographic data were processed using XDS [42] and merged and scaled with AIMLESS [43]. Details are shown in Table 1. Statistics for the highest-resolution shell are shown in parenthesis. a R meas = Σ hkl .(n/n−1) 1/2 Σ i |I i (hkl)-<|>(hkl)/Σ hkl Σ i I i (hkl), where I i (hkl) is the intensity measured for the ith reflection and <I>(hkl)) is the average intensity of all reflections with indices hkl. b CC 1/2 is the correlation coefficient between two random half datasets.

Structure Determination and Refinement
The structures were solved by molecular replacement. The rMroUPO structure was solved with the program PHASER [44] implemented in the CCP4 package, using the crystal structure (2YOR) of AaeUPO [14] as the search model. Structural refinement of the initial model was carried out using Refmac [45] alternating manual building, addition of water molecules, and positioning of the heme group and ligands using Coot [46]. Once the structure of rMroUPO was solved, the structural model was used to determine the structure of rCviUPO and its C235A and K228stop variants. Structural refinement of the latter enzyme and variants used the same workflow as the applied for rMroUPO. Details of the model refinements are given in Table 2.
The coordinates and structure factors have been deposited in the Protein Data Bank (as entries 7ZBP and 7ZCL). Figures for structural representations were drawn with the Pymol [47] and Swiss PdbViewer [48] programs. Protein-protein interactions were analyzed with PISA [49] available at EMBL-EI (PDBePISA; http://www.ebi.ac.uk/msd-srv/prot_ int/pistart.html, accessed on 1 March 2022). The final molecular models were analyzed and validated with MolProbity [50]. The nonresolved C-terminal tail was tentatively

Size-Exclusion Chromatography and Analytical Ultracentrifugation
The UPO molecular masses at the native as well as at the unfolded and disulfidefree states were first estimated as described at the end of the purification section by using (i) Superdex-75 SEC and (ii) SDS-PAGE under denaturing and nondenaturing conditions (with and without heating at 100 • C with SDS and mercaptoethanol, respectively). Molecular-mass standards were used in both cases: (i) conalbumin (75 kDa), ovalbumin (43 kDa), carbonic anhydrase (29 kDa), ribonuclease A (13.7 kDa), and aprotinin (6.5 kDa) for SEC; and (ii) the Precision Plus protein standards (10-250 kDa) from Bio-Rad (Hercules, CA, USA) for SDS-PAGE.
Moreover, analytical ultracentrifugation was used in sedimentation velocity experiments with native rCviUPO and two mutated variants. These experiments were performed at the Molecular Interactions facility of CIB using a Beckman-Coulter (Brea, CA, USA) analytical ultracentrifuge Optima XLI. The sedimentation coefficient distributions were calculated by least-square boundary modeling of sedimentation velocity data using the continuous distribution c(s) Lamm equation mode [53], as implemented in SEDFIT 16 1c, with a confidence level of 0.68.
In parallel to the above sedimentation velocity experiments, dynamic light scattering (DLS) experiments were performed in a Protein Solutions (Piscataway, NJ, USA) DynaPro MS/X instrument, using a 90 • light scattering cuvette. Data were collected and analyzed with the Dynamics V6 software yielding the corresponding diffusion coefficients (D) that, together with the sedimentation coefficients (s), were used to establish the molecular mass (M) of the enzymes using the Svedberg equation, Equation (1), where T, R, First, the optimal pH alcohol (10 mM), 2,2'-azi , and ρ are the absolute temperature, the universal gas constant, the partial specific volume of the protein, and the density of the solution, respectively.
using the continuous distribution c(s) Lamm equat SEDFIT 16 1c, with a confidence level of 0.68.
In parallel to the above sedimentation velocity ex (DLS) experiments were performed in a Protein DynaPro MS/X instrument, using a 90° light scatterin analyzed with the Dynamics V6 software yielding t cients (D) that, together with the sedimentation coeffi molecular mass (M) of the enzymes using the Svedbe R, ῡ, and are the absolute temperature, the univer volume of the protein, and the density of the solution = 1 − ῡῡ
Kinetic curves for the enzyme-oxidizing H2O2 su mM H2O2 in 0.1 M acetate (pH 4) containing 7.5 mM tion was monitored for activity estimation, and the H that two moles of reducing substrate are oxidized by Curve fitting and data analysis for kinetic consta tions) were carried out with Sigma Plot 11.0. Michae number (catalytic constant, kcat), catalytic efficiency were generally obtained by nonlinear fitting the ko lis-Menten model).

= S + S
However, Equation (3) was used for rCviUPO k enzyme inhibition was observed (with the ki inhibitio producing half of the maximal inhibition).
Then, kinetic curves for the above enzyme-reducing substrates were obtained from the initial (10-30 s) increase in absorbance due to product formation, using a Thermo (Waltham, MA, USA) Spectronic spectrophotometer at the optimal pH values. Reactions included oxidation of 0.09-10 mM veratryl alcohol in 0.1 M acetate, pH 4.6 for rMroUPO and pH 3 or 5 for rCviUPO (veratraldehyde were concentrated and, using 10 mM citrate (pH 4), dialyzed and loaded into a Mono S high-resolution 5/5 column and eluted under a NaCl gradient.
In the case of rCviUPO, the first step was cation exchange chromatography with a HiTrap SPFF column (GE Healthcare, Chicago, IL, USA) in 10 mM Tris (pH 7.4). The proteins, eluted with a gradient of the same buffer supplemented with 1 M NaCl (single peak with absorbance at 420 nm), were concentrated in an Amicon 3K device (Sigma-Aldrich, Saint Louis, MO, USA). The second step (to ensure purity) was SEC with a Superdex 75 column (10/300 GL; GE Healthcare, Chicago, IL, USA) in 10 mM Tris (pH 7.4) containing 0.15 M NaCl.
Enzyme purity was confirmed under denaturing conditions by 12% polyacrylamide gel electrophoresis (PAGE) in the presence of 0.1% sodium dodecyl sulfate (SDS) and 1% mercaptoethanol (5% in the loading buffer) [41]. Detection of the dimeric forms of the pure enzymes was performed also in 12% PAGE without any sample heating and using a loading buffer without any denaturing agent. Proper folding and binding of the cofactor were verified by analyzing the UV-visible spectrum of the enzymes (at the resting state) in 10 mM Tris (pH 7.4) using a Cary 60 spectrophotometer (Agilent, Santa Clara, CA, USA) ( Figure S1, black profiles). Additionally, formation of the characteristic complex between reduced (ferrous) heme-thiolate enzymes (such as P450s and UPOs) and carbon monoxide (CO) was assessed in 0.2 M phosphate (pH 8) after addition of Na2S2O4 and CO flushing ( Figure S1, red profiles). Concentrations were calculated using the rMroUPO and rCviUPO molar extinction coefficients of ℇ420 115.0 and 114.2 mM −1 ·cm −1 , respectively.

Site-Directed Mutagenesis
To investigate MroUPO dimerization through the intermolecular disulfide bridge visible in the crystal structure, the C227A mutation was introduced at its C-terminal tail using the Expand Long Template PCR kit from Roche (Basel, Switzerland) and the following primer (direct sequence with the mutated triplets underlined): 5'-CCA ACT GGC GAT AAC GCC GGC GCT ATC GTT CTC.
The above tail region was not solved in the rCviUPO crystal structure. However, to clarify its dimerization mechanism, (i) the single substitution C235A variant (changing the residue equivalent to MroUPO Cys227 after sequence alignment) and (ii,iii) the tail deletion K228stop and P201stop variants (including the TAA stop codon) were generated using the above PCR kit and the following primers high-resolution 5/5 column under the same conditions. Finally, the appropriat were concentrated and, using 10 mM citrate (pH 4), dialyzed and loaded into high-resolution 5/5 column and eluted under a NaCl gradient.
In the case of rCviUPO, the first step was cation exchange chromatograp HiTrap SPFF column (GE Healthcare, Chicago, IL, USA) in 10 mM Tris (pH proteins, eluted with a gradient of the same buffer supplemented with 1 M Na peak with absorbance at 420 nm), were concentrated in an Amicon 3K de ma-Aldrich, Saint Louis, MO, USA). The second step (to ensure purity) was S Superdex 75 column (10/300 GL; GE Healthcare, Chicago, IL, USA) in 10 mM 7.4) containing 0.15 M NaCl.
Enzyme purity was confirmed under denaturing conditions by 12% polya gel electrophoresis (PAGE) in the presence of 0.1% sodium dodecyl sulfate (SD mercaptoethanol (5% in the loading buffer) [41]. Detection of the dimeric for pure enzymes was performed also in 12% PAGE without any sample heating a loading buffer without any denaturing agent. Proper folding and binding of th were verified by analyzing the UV-visible spectrum of the enzymes (at the res in 10 mM Tris (pH 7.4) using a Cary 60 spectrophotometer (Agilent, Santa C USA) ( Figure S1, black profiles). Additionally, formation of the characteristi between reduced (ferrous) heme-thiolate enzymes (such as P450s and UPOs) a monoxide (CO) was assessed in 0.2 M phosphate (pH 8) after addition of Na2S2O flushing ( Figure S1, red profiles). Concentrations were calculated using the and rCviUPO molar extinction coefficients of ℇ420 115.0 and 114.2 mM −1 ·cm −1 , res

Site-Directed Mutagenesis
To investigate MroUPO dimerization through the intermolecular disulfi visible in the crystal structure, the C227A mutation was introduced at its C-te using the Expand Long Template PCR kit from Roche (Basel, Switzerland) an lowing primer (direct sequence with the mutated triplets underlined): 5'-CCA GAT AAC GCC GGC GCT ATC GTT CTC.
The above tail region was not solved in the rCviUPO crystal structure. Ho clarify its dimerization mechanism, (i) the single substitution C235A variant the residue equivalent to MroUPO Cys227 after sequence alignment) and (ii,i deletion K228stop and P201stop variants (including the TAA stop codon) were using the above PCR kit and the following primers (direct sequences with th triplets underlined): C235A, 5'-GGC AAA CAG GCA CCG CAA GCC C; K2 GCG TTA GTG ACC CCG GAA AAA CTG ATC GAC TAA; and P201stop, 5'-CGT CCG TAA AAA GCT GAA CTG. 436 = 29,300 M −1 ·cm −1 ), and 0.03-30 mM DMP in 0.1 M tartrate, pH 4 for rMroUPO and pH 5 for rCviUPO ( sing 10 mM phosphate (pH 7), supplemented with 1 M NaCl for elution. The riate fractions were pooled, concentrated, desalted, and loaded into a Mono Q solution 5/5 column under the same conditions. Finally, the appropriate fractions ncentrated and, using 10 mM citrate (pH 4), dialyzed and loaded into a Mono S solution 5/5 column and eluted under a NaCl gradient. the case of rCviUPO, the first step was cation exchange chromatography with a SPFF column (GE Healthcare, Chicago, IL, USA) in 10 mM Tris (pH 7.4). The s, eluted with a gradient of the same buffer supplemented with 1 M NaCl (single ith absorbance at 420 nm), were concentrated in an Amicon 3K device (Sigrich, Saint Louis, MO, USA). The second step (to ensure purity) was SEC with a ex 75 column (10/300 GL; GE Healthcare, Chicago, IL, USA) in 10 mM Tris (pH taining 0.15 M NaCl. zyme purity was confirmed under denaturing conditions by 12% polyacrylamide trophoresis (PAGE) in the presence of 0.1% sodium dodecyl sulfate (SDS) and 1% toethanol (5% in the loading buffer) [41]. Detection of the dimeric forms of the zymes was performed also in 12% PAGE without any sample heating and using a buffer without any denaturing agent. Proper folding and binding of the cofactor rified by analyzing the UV-visible spectrum of the enzymes (at the resting state) M Tris (pH 7.4) using a Cary 60 spectrophotometer (Agilent, Santa Clara, CA, Figure S1, black profiles). Additionally, formation of the characteristic complex n reduced (ferrous) heme-thiolate enzymes (such as P450s and UPOs) and carbon de (CO) was assessed in 0.2 M phosphate (pH 8) after addition of Na2S2O4 and CO g ( Figure S1, red profiles). Concentrations were calculated using the rMroUPO iUPO molar extinction coefficients of ℇ420 115.0 and 114.2 mM −1 ·cm −1 , respectively.
-Directed Mutagenesis investigate MroUPO dimerization through the intermolecular disulfide bridge in the crystal structure, the C227A mutation was introduced at its C-terminal tail he Expand Long Template PCR kit from Roche (Basel, Switzerland) and the folprimer (direct sequence with the mutated triplets underlined): 5'-CCA ACT GGC AC GCC GGC GCT ATC GTT CTC. e above tail region was not solved in the rCviUPO crystal structure. However, to its dimerization mechanism, (i) the single substitution C235A variant (changing due equivalent to MroUPO Cys227 after sequence alignment) and (ii,iii) the tail Kinetic curves for the enzyme-oxidizing H 2 O 2 substrate were obtained with 0.5 to 30 mM H 2 O 2 in 0.1 M acetate (pH 4) containing 7.5 mM DMP, whose (one electron) oxidation was monitored for activity estimation, and the H 2 O 2 constants were obtained (note that two moles of reducing substrate are oxidized by each mole of peroxide).
Curve fitting and data analysis for kinetic constant estimation (from triplicate reactions) were carried out with Sigma Plot 11.0. Michaelis-Menten constant (K m ), turnover number (catalytic constant, k cat ), catalytic efficiency (k cat /K m ), and their standard errors were generally obtained by nonlinear fitting the k obs values to Equation (2) (Michaelis-Menten model).
However, Equation (3) was used for rCviUPO kinetics with ABTS and H 2 O 2 , where enzyme inhibition was observed (with the k i inhibition constant being the concentration producing half of the maximal inhibition).
Products (and remaining substrates) were liquid-liquid extracted with methyl tertbutyl ether (Sigma-Aldrich, Saint Louis, MO, USA), which was evaporated under a N 2 stream. N,O-Bis(trimethylsilyl)trifluoroacetamide (Supelco, Bellefonte, PA, USA) was used to prepare trimethylsilyl derivatives. Analyses were performed with an Agilent (Santa Clara, CA, USA) gas chromatography-mass spectrometry (GC-MS) QP2010 Ultra equipment using a fused-silica DB-5HT 30 m capillary column from J&W Scientific (Folsom, CA, USA). The oven was heated from 120 • C (1 min) to 300 • C (15 min) at 5 • C min −1 . The injector and transfer line were kept at 300 • C. Compounds were identified by mass fragmentography and comparison of their mass spectra with authentic standards. Quantifications were obtained from total-ion peak areas (partially overlapping peaks were deconvoluted when required) using external standard curves and molar response factors of the same or similar compounds.

Production of Native UPOs and Site-Directed Variants
The nonmutated recombinant (hereinafter native) rMroUPO and rCviUPO were obtained as soluble and active proteins in E. coli, purified without any tags, and used for crystallization, analytical ultracentrifugation, and catalytic studies. These studies were often favored by the higher expression levels and easier purification of the rCviUPO enzyme.
Moreover, to confirm some of the results obtained, the rMroUPO C227A and rCviUPO C235A, K228stop, and P201stop variants were generated by site-directed mutagenesis, isolated by ion-exchange chromatography and SEC, and characterized by biophysical and crystallographic methods, as done with the native enzymes, except for the P201stop variant whose extremely low expression levels prevented isolation with significant yields.
The results obtained in the above studies are described in the next five sections. For consistency with residue numbering in wild enzymes (mature proteins isolated from fungal cultures), the signal sequence and introduced initial methionine were not considered for residue numbering in recombinant UPOs (rMroUPO and rCviUPO included) except for HspUPO [17].

Overall Crystallographic Structures
Crystals of recombinant rMroUPO and rCviUPO were obtained at sufficiently high resolution (1.45 and 1.95 Å, respectively), and the corresponding structures solved by molecular replacement. The rCviUPO C235A and K228stop variants were also crystallized, but the molecular structures obtained (not shown) were fully superimposable with that of the native (nonmutated) enzyme except for the residue(s) substituted or removed.
The asymmetric unit of the rMroUPO crystal structure ( Figure 1A) includes four polypeptide chains that associate to form a tetramer composed of two homo-dimers in a head-to-tail arrangement, each of them stabilized by an intermolecular disulfide bridge (between the Cys227 residue of both molecules). Each of the rMroUPO monomers ( Figure 1B) is compact and spherical in shape and consists of residues from Ser1 to Glu235, which displayed clearly defined electron density (only lacking the C-terminal Leu236), together with one heme molecule and one Mg 2+ ion. The overall fold is mainly helical with eleven α-helices and two short β-strands (V120-N122 and E174-P176). Moreover, a clear electron density could be detected at the active site ( Figure 2A) that could be unequivocally assigned to an acetate ion from the crystallization medium. Both active sites in the dimer are open and accessible from the solvent (see below).
The rCviUPO asymmetric unit reveals a compact dimer through a parallel association of helix α12 from each monomer, providing a closely packed helix interface ( Figure 1C). The rCviUPO structure includes all amino acid residues, except for the initial Glu1 and the thirty-two C-terminal residues-from Lys228 to Lys259, whose lack is due to their flexible conformation-one heme molecule and one Mg 2+ ion. The monomer adopts an α fold ( Figure 1D) formed by twelve helices and two short β strands (E124-V126 and K177-P179). The α-helices include two short α5 and α7 helices not found in rMroUPO, whereas the short α9 helix present in rMroUPO is absent from this structure. In spite of the analogies, a noticeable feature was the different dimer association, in which no intermolecular disulfide bridge was observed. Instead, the abovementioned parallel association between two α-helices at the dimer interface is proposed to play a key role. The monomeric rCviUPO structure contains 12 α-helices (yellow) and two short antiparallel β-strands (blue). As in (B), a 90°rotation is shown in (D). One heme cofactor (CPK sticks with Fe 3+ as a red sphere) and one Mg 2+ ion (green sphere) are also present in each UPO monomer.
The rCviUPO asymmetric unit reveals a compact dimer through a parallel association of helix α12 from each monomer, providing a closely packed helix interface ( Figure  1C). The rCviUPO structure includes all amino acid residues, except for the initial Glu1 and the thirty-two C-terminal residues-from Lys228 to Lys259, whose lack is due to their flexible conformation-one heme molecule and one Mg 2+ ion. The monomer adopts an α fold ( Figure 1D) formed by twelve helices and two short β strands (E124-V126 and K177-P179). The α-helices include two short α5 and α7 helices not found in rMroUPO, whereas the short α9 helix present in rMroUPO is absent from this structure. In spite of the analogies, a noticeable feature was the different dimer association, in which no intermolecular disulfide bridge was observed. Instead, the abovementioned parallel association between two α-helices at the dimer interface is proposed to play a key role. The monomeric rCviUPO structure contains 12 α-helices (yellow) and two short antiparallel β-strands (blue). As in (B), a 90 • rotation is shown in (D). One heme cofactor (CPK sticks with Fe 3+ as a red sphere) and one Mg 2+ ion (green sphere) are also present in each UPO monomer.

Heme Pocket
The stereochemical quality of the above crystal models is illustrated with the unambiguous electron density map at 1.45 Å resolution, shown in Figure 2A. The map corresponds to the heme cofactor, located at a central position in the crystal structure of rMroUPO ( Figure 1B), its surrounding residues, and the acetate and Mg 2+ ions.
The heme cofactor sits in an internal pocket with its iron ion coordinated by the sulfur atom of rMroUPO Cys17 at a distance of 2.4 Å in the lower (also known as proximal because of the cysteine ligand) side of the heme cofactor. The heme pocket includes residues lining the upper (also known as distal) side of the cavity, such as His86, Ile84, Phe160, Glu157, Ser156, Ile55, and Ala59, among others (Figure 2A). One of the acetate oxygens is near the sixth coordination position of the heme iron. The Mg 2+ ion appears coordinated (at 2.07-2.19 Å distances) by the Ser89 hydroxyl, the His86 (backbone) carbonyl, two water molecules (wat1 and wat2), and the carboxylates of Glu85 and heme ring-D propionate (neighbor Asp87 is not involved in Mg 2+ coordination). In this way, the Mg 2+ cation would contribute to anchoring the heme cofactor inside the UPO central pocket.
x FOR PEER REVIEW 9 of 22

Heme Pocket
The stereochemical quality of the above crystal models is illustrated with the unambiguous electron density map at 1.45 Å resolution, shown in Figure 2A. The map corresponds to the heme cofactor, located at a central position in the crystal structure of rMroUPO ( Figure 1B), its surrounding residues, and the acetate and Mg 2+ ions.
The heme cofactor sits in an internal pocket with its iron ion coordinated by the sulfur atom of rMroUPO Cys17 at a distance of 2.4 Å in the lower (also known as proximal because of the cysteine ligand) side of the heme cofactor. The heme pocket includes residues lining the upper (also known as distal) side of the cavity, such as His86, Ile84, Phe160, Glu157, Ser156, Ile55, and Ala59, among others (Figure 2A). One of the acetate oxygens is near the sixth coordination position of the heme iron. The Mg 2+ ion appears coordinated (at 2.07-2.19 Å distances) by the Ser89 hydroxyl, the His86 (backbone) carbonyl, two water molecules (wat1 and wat2), and the carboxylates of Glu85 and heme ring-D propionate (neighbor Asp87 is not involved in Mg 2+ coordination). In this way, the Mg 2+ cation would contribute to anchoring the heme cofactor inside the UPO central pocket.
x FOR PEER REVIEW 9 of 22

Heme Pocket
The stereochemical quality of the above crystal models is illustrated with the unambiguous electron density map at 1.45 Å resolution, shown in Figure 2A. The map corresponds to the heme cofactor, located at a central position in the crystal structure of rMroUPO ( Figure 1B), its surrounding residues, and the acetate and Mg 2+ ions.
The heme cofactor sits in an internal pocket with its iron ion coordinated by the sulfur atom of rMroUPO Cys17 at a distance of 2.4 Å in the lower (also known as proximal because of the cysteine ligand) side of the heme cofactor. The heme pocket includes residues lining the upper (also known as distal) side of the cavity, such as His86, Ile84, Phe160, Glu157, Ser156, Ile55, and Ala59, among others (Figure 2A). One of the acetate oxygens is near the sixth coordination position of the heme iron. The Mg 2+ ion appears coordinated (at 2.07-2.19 Å distances) by the Ser89 hydroxyl, the His86 (backbone) carbonyl, two water molecules (wat1 and wat2), and the carboxylates of Glu85 and heme ring-D propionate (neighbor Asp87 is not involved in Mg 2+ coordination). In this way, the Mg 2+ cation would contribute to anchoring the heme cofactor inside the UPO central pocket. , and Phe160 at the upper side and the characteristic proximal Cys17 at the lower side (neighbor I55, A59, and S156 residues and acetate ion are also shown). The 1.4 Å resolution 2Fo-Fc electron density map (blue) contoured at 1.0 σ is included. (B) Active site differences between rMroUPO (gray) and rCviUPO (yellow). Only noncommon residues in rCviUPO, including Leu57, Ile61, Phe88, Gly161, and Lys165, are labeled. (C,D) Semitransparent surfaces show the access channel to the heme in rMroUPO and rCviUPO, respectively. The equivalent positions of Ala59, Ile84, and Phe160 in rMroUPO change to Ile61, Phe88, and Lys165 in rCviUPO, reducing the entrance to the channel. These residues and acetate ion are shown as CPK sticks, while the Fe 3+ and Mg 2+ ions are shown as red and green spheres, respectively.
In rCviUPO, the neighbor residues located at the upper side of the cofactor that outline the cavity shell include Leu57, Ile61, His90, Phe88, Gly161, Glu162, and Lys165, with the heme iron coordinated by the thiol side chain of Cys19 at the lower side ( Figure 2B). The Mg 2+ -coordinating residues in rCviUPO (at 2.01-2.14 Å) are the same found in rMroUPO, but no acetate was found at the heme pocket. This is in agreement with its absence from the rCviUPO crystallization medium, and resulted in the presence of one water molecule (wat255) near the sixth coordination position of the heme iron.
Three of the above residues are invariable in the two enzymes (Figure 2A,B): (i) Cys17/Cys19, acting as the fifth ligand of the heme iron (together with the four N atoms of the heme macrocycle); and (ii) Glu157/Glu162 and His86/His90, putatively contributing to the reaction with H 2 O 2 for heme activation (the conserved histidine also contributes to the coordination of the Mg 2+ ion through its backbone carbonyl). Concerning the other five residues (Figure 2A,B), it is interesting to mention that a phenylalanine was present in both rMroUPO and rCviUPO structures, although it occupies opposite positions (Phe160 and Phe88, respectively). Phe160 of rMroUPO was replaced by Lys165 in rCviUPO and, as a result of this and other amino acid substitutions, the access to the substrate binding site in rCviUPO was narrowed ( Figure 2C,D).

Dimeric Arrangements
As explained above, the rMroUPO and rCviUPO crystal structures revealed dimeric proteins in both cases, but suggested the existence of two different dimeric arrangements. As already depicted in Figure 1A,C, an intermolecular disulfide bridge and the association between parallel α-helices appeared as main dimerization motifs in the rMroUPO and rCviUPO crystal structures, respectively. However, other interactions are also produced at the dimer interfaces of the two enzymes ( Figure 3A,B), most probably contributing to the whole dimeric arrangement.
tions (Phe160 and Phe88, respectively). Phe160 of rMroUPO was replaced by Lys165 rCviUPO and, as a result of this and other amino acid substitutions, the access to substrate binding site in rCviUPO was narrowed ( Figure 2C,D).

Dimeric Arrangements
As explained above, the rMroUPO and rCviUPO crystal structures revealed dim proteins in both cases, but suggested the existence of two different dimeric arrangeme As already depicted in Figure 1A,C, an intermolecular disulfide bridge and the asso tion between parallel α-helices appeared as main dimerization motifs in the rMroU and rCviUPO crystal structures, respectively. However, other interactions are also p duced at the dimer interfaces of the two enzymes ( Figure 3A,B), most probably cont uting to the whole dimeric arrangement. Figure 3. Schematic representation of the dimer interfaces in rMroUPO (A) and rCviUPO Subunits of the dimers are shown in gray and green for rMroUPO, and gray and yellow rCviUPO. Hydrogen bonds and salt bridges together with hydrophobic interactions contribut the stability of the dimers, together with the C227-C227′ disulfide bridge and the parallel asso tion of the two α12 helices (also see Figure 1).
In rMroUPO, the dimer interface buries 825 Å 2 of each monomer and is formed fr the interactions between residues of the C-terminal loop and residues from the helix ( Figure 3A). From the C-terminal region, an intermolecular disulfide bond between sulfhydryl side chains of Cys227 and the hydrophobic interactions established by Ile and Leu232 contribute to the clustering of the dimer. In addition, residues Ile62-Gl from the helix α4 increase the dimer stability by polar interactions of Lys61 with Se and Glu69, and by a patch of hydrophobic residues at the interface (Ile62, Leu65).  rCviUPO (B). Subunits of the dimers are shown in gray and green for rMroUPO, and gray and yellow for rCviUPO. Hydrogen bonds and salt bridges together with hydrophobic interactions contribute to the stability of the dimers, together with the C227-C227 disulfide bridge and the parallel association of the two α12 helices (also see Figure 1).
In rMroUPO, the dimer interface buries 825 Å 2 of each monomer and is formed from the interactions between residues of the C-terminal loop and residues from the helix α4 ( Figure 3A). From the C-terminal region, an intermolecular disulfide bond between the sulfhydryl side chains of Cys227 and the hydrophobic interactions established by Ile230 and Leu232 contribute to the clustering of the dimer. In addition, residues Ile62-Glu69 from the helix α4 increase the dimer stability by polar interactions of Lys61 with Ser67 and Glu69, and by a patch of hydrophobic residues at the interface (Ile62, Leu65).
In the rCviUPO crystal structure (Figure 3B), the buried interface area of the dimer, including the two α12 helices, was 1304 Å 2 . Dominant at the upper side of the interface are Arg156 and Lys224, which establish polar contacts on both sides with residues Glu216 and Glu223, respectively. At the corner of the dimer, Ser54 and Glu56 also bridge the interface by hydrogen bond interactions.
PDBePISA analyses were employed to unveil the interface interactions between monomers of the native and mutated enzymes. The dimeric model built with AlphaFold for the whole C227A variant of rMroUPO revealed a contribution of hydrophobic interactions to the homo-dimer formation, with an estimated free energy of −18.8 kcal/mol. This value was similar to that produced in the crystallographic dimer of native rMroUPO (−20.1 kcal/mol). In fact, even lower differences were found when the C227A model was compared with an AlphaFold model for the whole rMroUPO, with −18.3 kcal/mol free energy. This supports that, in addition to the disulfide bond contribution to dimeric arrangement, hydrophobic interactions also play an important role in the formation of a rMroUPO stable dimer (even in the absence of Cys227).
In the case of rCviUPO, the PDBePISA analyses of the AlphaFold models of C236A, K228stop, and P201stop showed lower numbers of H bonds (four, two, and one, respectively) compared to the whole (AlphaFold model) native rCviUPO (with nine H bonds). Moreover, the hydrophobic interactions in the C236A and K228stop interface (between −24.0 and −24.3 kcal/mol free energies) were similar to those found in the whole native rCviUPO. However, this interaction strongly decreased in the P201stop model (with a free energy of −16.9 kcal/mol), confirming the central contribution of such parallel association between the α12 helices of the two monomers, being absent from the P201stop variant.

Biophysical Properties: Molecular-Mass and Oligomerization State
Firstly, the molecular masses of rMroUPO and rCviUPO (under native conditions) were estimated by SEC after enzyme purification or during the last purification step (when purification concluded with SEC and the expression level was too low, as for the P201stop variant of rCviUPO). These results were compared with the molecular masses estimated after protein unfolding (also resulting in cofactor release) and breakdown of disulfide bridges, by SDS-PAGE in the presence of mercaptoethanol. As shown in Figure 4 (black profiles) and insets, the SEC molecular masses (41-45 kDa) of the native UPOs were nearly two-fold the masses from SDS-PAGE (26-27 kDa), which in turn were similar to the theoretical masses estimated from the protein sequences (23-30 kDa). Monomeric and dimeric forms of both enzymes could also be observed by SDS-PAGE with and without heating at 100 • C with SDS and mercaptoethanol, respectively ( Figure S2). These results indicate that both rMroUPO and rCviUPO are dimeric proteins under physiological conditions. Molecular masses (from SEC) similar to or even higher than those found for the native enzyme were obtained for the C227A variant (~59 kDa) of rMroUPO ( Figure 4A, blue profile) and the first two rCviUPO variants analyzed-K228stop (~41 kDa) and C235A (~52 kDa) that present 32-residues shorter C-terminal tail and the Cys→Ala mutation, respectively ( Figure 4B, green and blue profiles, respectively)-suggesting that their oligomerization state was not modified. However, the SEC analysis of the P201stop variant, with a 27-residue shorter C-terminal tail than K228stop (and 59-residue shorter than native rCviUPO), provided a mass value of only~17 kDa (indicative of a monomeric protein), revealing that these additional 27 residues removed were strongly involved in dimerization.
To confirm the above SEC analyses of rCviUPO and variants, analytical ultracentrifugation was performed and the sedimentation velocity profiles ( Figure S3 left) and c(s) distributions ( Figure S3 right) were obtained for the native enzyme and its C235A and K228stop variants (not enough P201stop enzyme was available, as explained above). In the sedimentation velocity assays, sedimentation coefficients of the major native rCviUPO, C235A, and K228stop species (representing 95.1, 94.6, and 98.7%, respectively) were determined to be 4.00, 3.78, and 3.93 S ( Figure 5).
With the above S values and the diffusion coefficients from parallel DLS analyses, molecular masses of 60 (rCviUPO), 66 (C235A), and 57 (K228stop) kDa were estimated using the Svedberg equation. The slightly lower S values and higher masses estimated for the two variants lacking (i) only Cys235 (substituted by an alanine residue) and (ii) Cys235 and other 31 residues in a shortened C tail (K228stop) could be due to a modification of the hydrodynamic (and sedimentation) properties of the enzyme by the breakdown of the intramolecular disulfide bridge in the mobile C tail region predicted by AlphaFold ( Figure S4), which would be similar to that stabilizing the C-terminal tail of AaeUPO [14]. In any case, the results support the SEC analysis confirming that rCviUPO remains a dimer in the absence of Cys235. To confirm the above SEC analyses of rCviUPO and variants, analytical ultra trifugation was performed and the sedimentation velocity profiles (Figure S3 left) c(s) distributions ( Figure S3 right) were obtained for the native enzyme and its C2 and K228stop variants (not enough P201stop enzyme was available, as explained ab In the sedimentation velocity assays, sedimentation coefficients of the major n   Figure S3). The estimated molecular masses we 57-66 kDa, revealing dimeric proteins in all the cases.
With the above S values and the diffusion coefficients from paralle molecular masses of 60 (rCviUPO), 66 (C235A), and 57 (K228stop) kDa using the Svedberg equation. The slightly lower S values and higher mass the two variants lacking (i) only Cys235 (substituted by an alanine re Cys235 and other 31 residues in a shortened C tail (K228stop) could be d cation of the hydrodynamic (and sedimentation) properties of the e breakdown of the intramolecular disulfide bridge in the mobile C tail regi AlphaFold ( Figure S4), which would be similar to that stabilizing the C-AaeUPO [14]. In any case, the results support the SEC analysis confirmin remains a dimer in the absence of Cys235.

Comparison of Catalytic Properties
To investigate if the dimeric arrangements would affect the catalyt the two short UPOs under investigation, the kinetic constants of the nativ some of the mutated variants on four peroxygenase/peroxidase classical estimated (Table 3). First, the constants for H2O2 were determined usin zyme-reducing substrate. It was observed that rCviUPO was inhibited concentrations (with ki of 1.1 mM). Therefore, the rCviUPO and rMroUP stants (towards veratryl alcohol, ABTS and DMP) were determined at 1 H2O2 concentrations, respectively.
In the case of rMroUPO, a comparison with the activities reported f zyme isolated from M. rotula cultures [22], confirmed that the recombi able to oxidize the same substrates and to reduce H2O2 with catalytic e similar order of magnitude. However, removal of the intermolecular d observed in the crystal structure by replacement of Cys227 by alanine c loss of the enzyme activity, in spite of the fact that the enzyme maintains ture, as explained above.  Figure S3). The estimated molecular masses were in the range of 57-66 kDa, revealing dimeric proteins in all the cases.

Comparison of Catalytic Properties
To investigate if the dimeric arrangements would affect the catalytic properties of the two short UPOs under investigation, the kinetic constants of the native enzymes and some of the mutated variants on four peroxygenase/peroxidase classical substrates were estimated (Table 3). First, the constants for H 2 O 2 were determined using DMP as enzyme-reducing substrate. It was observed that rCviUPO was inhibited by high H 2 O 2 concentrations (with k i of 1.1 mM). Therefore, the rCviUPO and rMroUPO catalytic constants (towards veratryl alcohol, ABTS and DMP) were determined at 1 mM and 5 mM H 2 O 2 concentrations, respectively. In the case of rMroUPO, a comparison with the activities reported for the wild enzyme isolated from M. rotula cultures [22], confirmed that the recombinant enzyme is able to oxidize the same substrates and to reduce H 2 O 2 with catalytic efficiencies in a similar order of magnitude. However, removal of the intermolecular disulfide bridge observed in the crystal structure by replacement of Cys227 by alanine caused complete loss of the enzyme activity, in spite of the fact that the enzyme maintains its dimeric nature, as explained above.
Comparison with the wild CviUPO could not be performed, since this UPO is only known as a recombinant enzyme. Concerning the two recombinant enzymes, rCviUPO was over 60-fold less efficient oxidizing veratryl alcohol than rMroUPO (due to lower substrate affinity) but over 10-fold more efficient oxidizing ABTS (due to higher turnover) revealing marked substrate preferences. Moreover, in contrast to activity loss by the C227A mutation in rMroUPO, no decrease in the catalytic efficiency was produced by the C235A mutation in rCviUPO. Interestingly, strong changes in both veratryl alcohol activity (25-fold higher turnover) and affinity (19-fold higher K m ) were caused by the K118stop mutation (although the resulting efficiency was only slightly increased).
Finally, the oxygenation patterns by native rCviUPO and its C-terminal tail variants were studied on 18-C unsaturated fatty acids. In all cases, the conversion yields lay in the 50-98% range and (different) epoxides were found as the main products. The most interesting chromatographic profile was observed for the linoleic acid reaction with the above K228stop variant ( Figure S5). As shown by the quantitative results provided in Table 4, the variant and native enzyme produced similarly high (97-98%) conversions, but over 50% of the products formed by K228stop were diepoxides, which are of interest as cross-linking molecules, while they were fully absent from the rCviUPO reactions, where monoepoxides predominate. This higher diepoxide content resulted in up to 68% epoxidation yield (defined as the percentage of total double bonds converted into epoxides) by the variant with shortened tail. Table 4. Fatty acid conversion, percentages of main products, and epoxidation yield in the reactions of linoleic acid with native rCviUPO and the short-tail K228stop variant (see Figure S5 for chromatographic profiles). The products from 30 min reactions of 0.4 µM enzyme in 50 mM phosphate (pH 7) with 20% acetone were extracted with methyl tert-butyl ether and analyzed by GC-MS as trimethylsilyl derivatives.

Discussion
Novel structural-functional information on short UPO enzymes is provided by a combination of crystallographic, biophysical, and molecular biology techniques, with the focus on the dimerization mechanisms. For these studies, we took advantage of the heterologous production of rMroUPO and rCviUPO in E. coli as soluble enzymes [33,34], which strongly facilitate crystallization by reducing the heterogeneity due to the glycosidic moiety, as shown for in vitro deglycosylated yeast-expressed recombinant UPOs (with 30-50% carbohydrate content) [14,15,17]. Moreover, it has been shown that the activity and stability properties of the rMroUPO obtained are fairly similar to those of wild MroUPO isolated from fungal cultures [54].

Short and Long UPO Families
Phylogenetic analyses revealed two major UPO families, characterized by their shorter or longer protein chains, each family including several thousand sequences in different subfamilies [12]. The long AaeUPO is the best known UPO in terms of crystal structure [14], directed evolution for yeast expression [27], and reactions of biotechnological interest [4,55], among other studies. While long UPOs are only present in higher fungi, short UPOs have a wider distribution and more ancient origin inside the fungal kingdom [12].
Structural and sequence differences are illustrated in Figure 6, where MroUPO and CviUPO were compared with the model AaeUPO. The two main structural characteristics of the latter long UPO with respect to the two short UPOs ( Figure 6A left) concern its (i) 18-residues long internal loop, and (ii) longer C-terminal tail with respect to short UPOs (the AaeUPO C-terminal tail is >50-residues longer than the rMroUPO tail and >30-residues longer than the rCviUPO tail). AaeUPO also has a near 17-19 residues longer N-terminal tail with respect to the two short UPOs (Figure 6A right). Significant similarities and differences between the two UPO families are also found at the active-site level, as discussed below. The former include the conserved proximal cysteine and distal glutamic acid, indicated with a red background in Figure 6B, together with other conserved residues that, interestingly, are especially frequent in the N-terminal moiety of these proteins (which harbors 74% of the conserved residues).
Concerning their active sites, all UPOs from basidiomycetes-such as AaeUPO, CciUPO, and MroUPO-and ascomycetes-such as HspUPO, DcaUPO, CglUPO, HinUPO, and CviUPO-whose molecular structure has been modeled [14,17,33,[56][57][58][59] (Figure S6), show two invariable cysteines (Cys19 in CviUPO) and glutamic acid (Glu162 in CviUPO) residues at the lower and upper sides of the heme pocket. These conserved residues would (i) act as the fifth ligand of the heme iron, and (ii) contribute to the reaction with H 2 O 2 , respectively. A water oxygen is often near the sixth coordination position of the heme iron in the resting enzyme (as shown in Figure S6H for rCviUPO) that is substituted by the oxygen of acetate, when present in the crystallization medium (as shown in Figure S6G for rMroUPO), or by O or N atoms of other buffers or ligands as described for HspUPO [17] and evolved AaeUPO [15]. A second residue would contribute to the activation of UPOs by H 2 O 2 , being an arginine (Arg189) in the long AaeUPO and CciUPO ( Figure S6A,B), and a histidine (His90 in CviUPO) in the other (short) UPOs mentioned above ( Figure S6C-H). These two residues act as an acid-base pair in compound I formation through a mechanism [60] shared by peroxidases and peroxygenases.
Several aromatic residues are also present at the UPO heme pocket and most probably modulate the access of different substrates to the reactive iron-oxo group of the peroxideactivated heme, as suggested for rAaeUPO [15]. These residues include: (i) AaeUPO/CciUPO Phe121 near the heme edge, conserved in CviUPO (as Phe88) but substituted by isoleucine (Ile84 in MroUPO) or leucine (in the other four UPOs mentioned above); and (ii) AaeUPO/CciUPO Phe199, with one tyrosine (Tyr166 in CviUPO), one phenylalanine (Phe160 in MroUPO), or both tyrosine and phenylalanine residues (in the other four UPOs) occupying neighbor (but not homologous) positions at the upper side of heme. AaeUPO Phe199 and HspUPO Phe176 have been reported to interact with aromatic substrates [14,15,17]. As a result of differences in the heme pocket and neighboring residues, the size of the access channel varies in the different UPO crystal structures, with the internal diameter being in the order rMroUPO > rHspUPO > rCviUPO > AaeUPO ( Figure S7A-C, respectively). Antioxidants 2022, 11, x FOR PEER REVIEW 16 of 22

Two Different Dimeric Arrangements
Long UPOs are only known as monomeric enzymes, while dimerization has been reported in short UPOs, such as MroUPO, MweUPO [23], and CviUPO [33], although not for short HspUPO [17]. The first experimental evidence for dimer formation by rMroUPO and rCviUPO came from double molecular-mass estimation by techniques potentially preserving or destroying (by the action of SDS and mercaptoethanol) the tertiary and quaternary structure of the native UPOs. The results obtained, a molecular mass reduction from 40-45 to 26-27 kDa, clearly showed that both rMroUPO and rCviUPO are dimeric enzymes.
Analytical ultracentrifugation is a powerful technique to study protein associations under physiological (ecological) conditions [61] and, in combination with dynamic light scattering, enabled more precise estimation of the molecular masses (57-66 kDa) of the two native UPOs and their mutated variants, although the molecular mass of the P201stop variant could only be estimated by SEC, due to its very low expression level as an active soluble enzyme in E. coli.
Unexpectedly, considering the proposed dimerization of MroUPO by an intermolecular disulfide bridge [1,57] based on the unpublished 5FUJ/5FUK coordinates of the wild enzyme available at PDB (which are superimposable with those of the recombinant enzyme, 7ZBP, solved here), the C227A variant still appeared as a dimer. The same result was found, using both SEC and analytical centrifugation, by the C235A mutation in rCviUPO. The solved crystal structures explained this situation, and agree with a dimeric form in the solution state. First, both asymmetric units showed dimeric structures (associated in a tetramer in the case of rMroUPO crystals) confirming the biophysical analyses. Moreover, a strong parallel association between two α-helices was observed at the interface of the rCviUPO dimer. Its removal in the shortened P201stop variant strongly reduced the molecular mass, demonstrating a main role in the dimerization of this enzyme. The extremely low expression and difficult purification of this variant, mentioned above, could be related to lower stability of the monomeric form, as claimed for some dimeric hemeproteins [62]. No such α-helix association was observed in rMroUPO, but the described hydrophobic and polar interactions at the dimer interface could contribute to the dimerization mechanism in this enzyme, maintaining the dimeric structure of the C227A variant (as shown by SEC and analytical centrifugation). A comparison of the two dimeric arrangements found in rMroUPO and rCviUPO is illustrated in Figure 7.
Antioxidants 2022, 11, x FOR PEER REVIEW reduced the molecular mass, demonstrating a main role in the dimerization of zyme. The extremely low expression and difficult purification of this variant, me above, could be related to lower stability of the monomeric form, as claimed fo dimeric hemeproteins [62]. No such α-helix association was observed in rMroU the described hydrophobic and polar interactions at the dimer interface could con to the dimerization mechanism in this enzyme, maintaining the dimeric structur C227A variant (as shown by SEC and analytical centrifugation). A comparison of dimeric arrangements found in rMroUPO and rCviUPO is illustrated in Figure 7.

Some (Bio)Technological Implications
Owing to oxirane ring reactivity, fatty acid epoxides are of interest for the in Figure 7. Two dimeric arrangements in short UPOs, as shown by the superimposition of the rMroUPO (purple) and rCviUPO (cyan) crystal structures. The interface of the dimeric ribbon models shows parallel association of two α-helices in rCviUPO and an intermolecular disulfide bridge in rMroUPO. Heme cofactors are shown as CPK-colored sticks and Mg 2+ ions as spheres.

Some (Bio)Technological Implications
Owing to oxirane ring reactivity, fatty acid epoxides are of interest for the industrial production of biobased chemicals and intermediates [63,64] including binder ingredients as cross-linkable materials. UPOs catalyze the epoxidation of (poly)unsaturated fatty acids [65] without the drawbacks of chemical [66] and chemoenzymatic [67] epoxidation via peracids. Although rMroUPO and rCviUPO engineering has been addressed to improve the epoxidation selectivity, monoepoxides are generally the main products obtained [34][35][36].
In this context, the strong improvement in diepoxide production (representing >50% of products after 98% conversion of linoleic acid by the K228stop variant of rCviUPO) is a result of industrial relevance, since diepoxides will confer the desired cross-linking properties to biobased epoxy binders. The C-terminal region removed by the K228stop mutation could not be solved in the rCviUPO crystal structure due to high mobility, as found for other UPOs [17]. However, its AlphaFold [51] modeling revealed that it extends directly over the heme access in the crystal structure ( Figure S4). In this way, C-terminal tail removal in rCviUPO would facilitate the access of linoleic acid to the active site providing the structural basis for its improved diepoxidation activity.
Moreover, improved catalytic efficiency of the rCviUPO variants was observed for other substrates, including the bulky ABTS molecule. It had been suggested that dimerization would not affect the access to the heme cofactor in MroUPO [57]. However, the C227A variant generated here, in spite of remaining a dimer as discussed above, completely lost its catalytic activity. This would be explained by a different disposition of the mutated C-terminal tail (lacking the stabilizing disulfide bond) at the entrance of the heme access channel.

Concluding Remarks
In conclusion, although additional interactions at the dimer interface seem to contribute to the observed UPO dimerization, as suggested by directed mutagenesis combined with SEC experiments, the solved crystal structures show the existence of the intermonomer disulfide bridge previously proposed in MroUPO, and the parallel association between helix-12 of both rCviUPO monomers, as main elements of two different dimeric arrangements existing in short UPOs.