A C4-oxidizing Lytic Polysaccharide Monooxygenase Cleaving Both Cellulose and Cello-oligosaccharides*

Background: Lytic polysaccharide monooxygenases (LPMOs) are recently discovered enzymes that cleave polysaccharides. Results: We describe a novel LPMO and use a range of analytical methods to characterize its activity. Conclusion: Cellulose and cello-oligosaccharides are cleaved by oxidizing the sugar at the nonreducing end in the C4 position. Significance: This study provides unequivocal evidence for C4 oxidation of the nonreducing end sugar and demonstrates a novel LPMO substrate specificity. Lignocellulosic biomass is a renewable resource that significantly can substitute fossil resources for the production of fuels, chemicals, and materials. Efficient saccharification of this biomass to fermentable sugars will be a key technology in future biorefineries. Traditionally, saccharification was thought to be accomplished by mixtures of hydrolytic enzymes. However, recently it has been shown that lytic polysaccharide monooxygenases (LPMOs) contribute to this process by catalyzing oxidative cleavage of insoluble polysaccharides utilizing a mechanism involving molecular oxygen and an electron donor. These enzymes thus represent novel tools for the saccharification of plant biomass. Most characterized LPMOs, including all reported bacterial LPMOs, form aldonic acids, i.e., products oxidized in the C1 position of the terminal sugar. Oxidation at other positions has been observed, and there has been some debate concerning the nature of this position (C4 or C6). In this study, we have characterized an LPMO from Neurospora crassa (NcLPMO9C; also known as NCU02916 and NcGH61–3). Remarkably, and in contrast to all previously characterized LPMOs, which are active only on polysaccharides, NcLPMO9C is able to cleave soluble cello-oligosaccharides as short as a tetramer, a property that allowed detailed product analysis. Using mass spectrometry and NMR, we show that the cello-oligosaccharide products released by this enzyme contain a C4 gemdiol/keto group at the nonreducing end.

In the emerging bio-economy, plant biomass will gradually substitute fossil resources for the production of fuels, chemicals, and materials. One of the main bottlenecks in such biorefining processes is the depolymerization of cellulose, a major constituent of the plant cell wall, to fermentable sugars. In nature this process is catalyzed by cellulases and the recently discovered lytic polysaccharide monooxygenases (LPMOs) 2 (1). Enzymes and binding domains interacting with polysaccharides are categorized in the CAZy database, which comprises families of structurally related carbohydrate-active enzymes, such as glycoside hydrolases (GH), and carbohydrate-binding modules (CBMs) (2). LPMOs were originally classified as CBM33 (family 33 carbohydrate-binding module) or GH61 (family 61 glycoside hydrolase). However, CAZy has recently been revised, and GH61 and CBM33 are now named LPMOs and classified under the heading "auxiliary activities" (AA) as families AA9 and AA10, respectively (3).
The enzyme activities of LPMOs were first discovered in 2010 for an AA10 protein (CBP21) acting on chitin (4). Following this study, cellulose active LPMOs were found in both the AA10 (5) and AA9 (6 -8) families. These copper-dependent enzymes carry out oxidative cleavage of the ␤-1,4-glycosidic bonds in polysaccharides, using molecular oxygen and an electron donor (1). Electrons may be supplied by small molecule reductants such as ascorbic acid and gallic acid (4,6) or by enzymes, such as cellobiose dehydrogenase (CDH), that are co- expressed with LPMOs (8 -10). The products of the reaction are oxidized oligosaccharides and native oligosaccharides containing reducing ends originally present in the polymeric substrate (1, 7) (Fig. 1). Although cleavage of polysaccharides by C1 oxidizing LPMOs, which yields aldonic acids, has been thoroughly demonstrated and analyzed (4,5,7,8,11,12), oxidation at the nonreducing end is more difficult to analyze. Based on mass spectrometry, both oxidation at C4 and C6 have been suggested (6,8,13). For products generated by NcLPMO9D (also known as NCU01050 or NcGH61-4), oxidation at C4 rather than at C6 has been shown indirectly by detection of the C4 epimer of glucose, galactose, upon reduction of reaction products and by the absence of glucuronic acid upon hypoiodite oxidation of reaction products (14). However, direct evidence for the identity of the nonreducing end oxidized species is lacking. The filamentous ascomycete Neurospora crassa is an efficient degrader of plant cell walls and produces a wide range of LPMOs and hydrolytic enzymes. The genome of N. crassa is predicted to contain 14 AA9 family LPMOs, six of which are attached to a CBM1 carbohydrate-binding module (15). These CBM1 modules contain ϳ40 amino acids, typically bind cellulose, and are almost exclusively found in fungi. The activities of three of these LPMOs on cellulose have been qualitatively characterized by HPLC and MS analyses of released products. NcLPMO9E (NCU08760, NcGH61-5; attached to a CBM1) oxidizes C1, NcLPMO9D exclusively oxidizes the nonreducing end, and NcLPMO9M (NCU07898, NcGH61-13) seems to be capable of oxidizing both C1 and the nonreducing end (8,16). In another study, using CDH as an electron donor, it was shown that three additional N. crassa AA9s, NcLPMO9C (NCU02916, NcGH61-3; attached to a CBM1), NcLPMO9F (NCU03328, NcGH61-6), and NcLPMO9J (NCU01867, NcGH61-10; attached to a CBM1), degrade cellulose, but no attempts were made to unravel details of the reaction products of these enzymes (17).
In this study, we have characterized the activity of NcLPMO9C using NMR, mass spectrometry, HPLC, and a previously described activity assay (17). Interestingly, NcLPMO9C turned out to be active on soluble substrates, which is an activity not previously described for LPMOs. Exploiting this unique property, we used NMR analysis to identify the products generated by NcLPMO9C.

EXPERIMENTAL PROCEDURES
Production and Purification of Enzymes-The AA9 encoding N. crassa gene NCU02916 was codon-optimized, cloned with its native signal sequence under control of the methanol inducible AOX1 promotor, and recombinantly produced in Pichia pastoris X-33 following a published protocol (10). The protein was purified from 0.4 liter of culture supernatant by three subsequent chromatographic steps following a published method (17). In total, 28 mg of purified NcLPMO9C was obtained, and the homogeneity was verified by SDS-PAGE. Cellobiose dehydrogenase from Myriococcum thermophilum carrying a C-terminal CBM1 (MtCDH, Uniprot accession number A9XK88) (18) was recombinantly expressed in P. pastoris using methanol for induction (19). The enzyme was purified from 1 liter of culture supernatant by two chromatographic steps according to the procedure described by Harreither et al. (20), and 180 mg of homogeneous MtCDH was obtained.
HPLC Analysis-From the standard reactions, samples were taken at different time points, and the reaction was stopped by adding NaOH to a final concentration of 0.05 M. After removing insoluble substrates by centrifugation, the supernatant was centrifuged and analyzed by high performance anion exchange chromatography (HPAEC) using an ICS3000 system (Dionex, Sunnivale, CA) as described previously (12). In brief, a 2-l sample was injected on a CarboPac PA1 2 ϫ 250 mm analytical column (Dionex) coupled to a CarboPac PA1 2 ϫ 50 mm guard column kept at 30°C. Cello-oligosaccharides were eluted at 0.25 ml/min using a stepwise linear gradient from 100% eluent A (0.1 M NaOH) toward 10% eluent B (1 M NaOAc in 0.1 M NaOH) 10 min after injection and 30% eluent B 25 min after injection, followed by a 5 min exponential gradient to 100% B. The column was reconditioned between each run by running initial conditions for 9 min.
Analysis by Mass Spectrometry-For time resolved product analysis, electrospray ionization mass spectrometry (ESI-MS) was used with a linear ion trap LTQ Velos Pro (Thermo Scien-tific, San Jose, CA USA) coupled to an UltiMate 3000 RS UHPLC from Dionex (Sunnyvale, CA USA) which delivered a constant flow and performed injection. No chromatographic separation was employed. The UHPLC delivered a flow of 0.2 ml/min of 30/70 (v/v) H 2 O and acetonitrile via the auto-sampler. Standard reaction mixtures were incubated in the thermostatted auto-sampler of the UHPLC at 15°C during the entire reaction time and samples of 2 l were injected at given time points. The electrospray was operated in positive mode at 4 kV spray current, with a sheath gas flow of 30 (arbitrary units), an auxiliary gas flow of 5 (arbitrary units) and a capillary temperature of 250°C. The acquisition time was set to 0.2 min with a data collection time of 10 ms per acquisition. Full scans were performed in the m/z 100 -1000 mass range and fragmentation was done using higher energy collisional dissociation (HCD) with N 2 as the collision gas and normalized energy levels of 65, to enable observation of lower mass fragments. During fragmentation, data were collected in the m/z 100 -400 mass range. The data were further processed using Xcalibur 2.2 SP1.48 (Thermo Scientific).
All homo-and heteronuclear NMR experiments were recorded on a Bruker Avance 600 MHz NMR spectrometer (Bruker BioSpin AG, Fällanden, Switzerland) equipped with a 5-mm cryogenic CP-TCI z-gradient probe at 25°C. For chemical shift assignment, the following spectra were recorded: one-dimensional proton, two-dimensional double quantum filter correlated spectroscopy (DQF-COSY), two-dimensional inphase correlation spectroscopy (IP-COSY) (23), two-dimensional total correlation spectroscopy (TOCSY) with 70 ms of mixing time, two-dimensional 13 C heteronuclear single quantum coherence (HSQC) with multiplicity editing, two-dimensional 13 C HSQC-[ 1 H, 1 H]TOCSY with 70 ms of mixing time on protons, and two-dimensional heteronuclear multibond correlation (HMBC) with BRID filter to suppress first order correlation. The NMR data were processed and analyzed with TopSpin 2.1 and TopSpin 3.0 software (Bruker BioSpin).
H 2 O 2 Analysis-A fluorimetric assay based on Amplex Red and horseradish peroxidase (17) was used to measure the extent of H 2 O 2 generation, which is a futile side reaction catalyzed by the reduced LPMO copper center. The peroxidase catalyzed conversion of Amplex Red to resorufin is proportional to H 2 O 2 production (stoichiometry ϭ 1). The increase of fluorescence was measured with an Enspire Multimode plate reader (PerkinElmer Life Sciences) using an excitation wavelength of 569 nm and an emission wavelength of 585 nm. The well plate assay (total volume of 200 l, 30°C, 6 min) was performed in 100 mM potassium phosphate buffer, pH 6.0, containing 50 M Amplex Red, 7.1 units ml Ϫ1 horseradish peroxidase, 0.87 M LPMO, and 30 M ascorbate as reductant in 100 mM sodium phosphate buffer, pH 6.0. Cello-oligosaccharides (DP2-6; Sigma-Aldrich) were added to a final concentration of 5 mM.
Sequence Alignment and Modeling-A structure-guided sequence alignment to compare NcLPMO9C with two structurally characterized AA9-type LPMOs, NcLPMO9D (NCU01050; Protein Data Bank code 4EIR), and PcLPMO9D (PcGH61D; Protein Data Bank code 4B5Q), was constructed using the Expresso mode of the T-Coffee multiple sequence alignment server (24). A homology model of NcLPMO9C was made based on a structure prediction by HHpred (25) and by using Modeler (26) with the crystal structure of NcLPMO9D as a template. All structural comparisons were carried out using PyMOL (PyMOL Molecular Graphics System, version 1.5.0.4; Schrödinger, LLC). The coordinates for cellopentaose were derived from Protein Data Bank entry 2EEX (27).

RESULTS AND DISCUSSION
Enzyme Activity on Cellulose and Cello-oligosaccharides-Initial activity screening of the enzyme was done with a well established HPAEC method for the detection of C1-oxidized cello-oligosaccharide products (12). Incubation of NcLPMO9C with polymeric substrates (PASC, Avicel, steam-exploded spruce; Fig. 2A) showed release of soluble cello-oligosaccharides (DP2 and DP3), but C1 oxidized species were not detected. Instead, two dominant later eluting peaks (between 26 and 29 min) were observed that could represent a different kind of oxidized product. Similar late eluting peaks have also been observed for PASC degradation by NcLPMO9D (8). MS analysis (described in detail below) indicated that the later eluting of these two peaks represented a trimeric product, whereas the earlier peak represented a dimeric product.
Cellulose-active LPMOs characterized so far typically produce oligosaccharides with a DP up to 6 or 7 (5)(6)(7)14). The production of relatively short oligosaccharides by NcLPMO9C could be the result of a glucanase background activity in the enzyme sample or by NcLPMO9C having activity on soluble oligosaccharides. Fig. 2B shows that NcLPMO9C indeed is able to oxidatively degrade cello-oligosaccharides. The enzyme readily degraded Glc 5 and Glc 6 , while showing lower activity on Glc 4 and no or minute activity on Glc 3 . There is abundant evidence for these conversions being caused by NcLPMO9C and not an impurity in the enzyme preparation: 1) the reaction generates oxidized species (see below for detailed characterization), 2) product formation was not observed in the absence of a reducing agent (results not shown), and 3) product formation was not observed in the presence of reducing agent only (results not shown). The oxidized products released were mainly DP2 from Glc 4 and Glc 5 and a mix of DP2 and DP3 from Glc 6 (Fig.  2B).
The degradation patterns shown in Fig. 2 (A and B) were independent of the reductant used (we tested hydroquinone, ascorbic acid, and catechin in concentrations varying from 1.5 to 10 mM). Fig. 2B (CDH trace) further shows that use of MtCDH as electron donor for degradation of Glc 5 led to a clear change in the product profile. As expected, native oligosaccharides were no longer observed because they were oxidized to aldonic acids. The oxidized species observed between 26 and 29 min for the reactions with a reducing agent present were not seen, but a new peak appeared at ϳ40 min that may represent double oxidized species, an interpretation that is supported by mass spectrometry and NMR analysis of the samples (see below). This shift clearly shows that the products eluting between 25 and 30 min cannot be aldonic acids, because the CDH oxidizes reducing ends.
MS analysis confirmed the presence of only two major products upon incubating NcLPMO9C with Glc 5 , namely an oxidized dimer and a nonoxidized trimer (see below for details). Nevertheless, the chromatograms (Fig. 2) show additional peaks eluting in between the native and the main oxidized products. We propose that this is due to tautomerization, because keto groups on ring carbons are prone to this process at the elevated pH values used during the HPAEC runs. Migration of the 4-keto group to C3 and maybe C2 will change the elution behavior of the oligosaccharides.
NcLPMO9C is the first LPMO unequivocally shown to be active on soluble cello-oligosaccharides. The fact that the pentamer is degraded faster than the tetramer and the clear preference for releasing an oxidized dimer from Glc 5 indicates the presence of at least five subsites on the enzyme running from Ϫ3 to ϩ2 (subsites numbered according to the nomenclature used for glycoside hydrolases) (28). Degradation of Glc 6 yielded both DP2 and DP3 oxidized species, indicating binding to Ϫ4 to ϩ2 and Ϫ3 to ϩ3. Notably, the location of these subsites and the orientation of the productively bound substrate relative to the catalytic center are currently unknown.
To test whether the cello-oligosaccharide oxidizing activity of NcLPMO9C is specific or an unspecific side reaction, the enzyme was incubated with other oligosaccharides (Man 6 , Xyl 5 , Xyl 6 , chitopentaose, and maltodextrin) and both crystalline ␣-chitin and nanofibrillar ␣and ␤-chitin. No activity was observed on any of these substrates. Thus, NcLPMO9C seems to be specific for ␤-1-4-linked glucose units.
Product Identification by MS-MS analysis of product mixtures showed that the main products from degradation of Glc 5 were two species with m/z values of 527 and 381, corresponding to the sodium adducts of Glc 3 and a cellobiose with a gemdiol (i.e., a hydrated keto group) at the nonreducing end, respectively. Minor amounts of cellobiose and oxidized cellotriose were also detected. To investigate this further, the time course of the enzyme reaction was studied in H 2 18 O. This resulted in the same signal of m/z 527, whereas a new peak at m/z 383 appeared (Fig. 3A). This shows that the oxygen of the glycosidic linkage remains in the Glc 3 product as the OH group at C1, whereas the oxidized product acquires one oxygen atom from water and one from oxygen as schematically shown in Fig. 3B. As indicated in Fig. 3A, a m/z 385 peak appears after some time, which is due to the lactone-gemdiol equilibrium leading to exchange with water and, thus, to the eventual incorporation of two 18 O atoms.
Because the glycosidic bond in cello-oligosaccharides (and cellulose) links the C1 of one glucose to the C4 of the adjacent glucose, it is logical to suggest that the oxidation carried out by NcLPMO9C takes place at the C4 position of the nonreducing end moiety. The resulting keto sugar will be in equilibrium with the C4 gemdiol in water solution (a feature that is common to keto saccharides (29)). Generally, it is not straightforward to prove the position of LPMO generated oxidations using mass spectrometry because the masses of various possible products are identical and because the mass difference between sodium and potassium adducts equals the mass of an oxygen atom. As shown in Fig. 1, the aldonic acid and gemdiol forms have identical masses, as do the corresponding lactone and keto forms. Thus, MS analysis alone cannot determine the type of LPMO activity. Exploiting the low complexity product mixtures obtainable thanks to the activity of NcLPMO9C on soluble substrates, we have addressed this issue by carrying out MS/MS analyses of products after lithium doping, which facilitates reducing end cross-ring fragmentation (30), as well as by NMR analyses. Fig. 4 shows the result of MS/MS analysis of the m/z 365 [MϩLi] ϩ products (oxidized dimers) generated by NcLPMO9C and C1 oxidizing PcLPMO9D (7). The spectra show differences including the occurrence of fragment ions that are diagnostic for C1 or C4 oxidations. Cellobionic acid (fragments indicated in blue in Fig. 4) readily loses mass corresponding to one carboxyl group (m/z 319). The C4 oxidized dimer (fragments indicated in black in Fig. 4) readily loses masses corresponding to both one and two water molecules (m/z 347 and 329) but does not generate ions corresponding to the loss of a carboxyl group. In addition, the most prevalent Y 1 and Z 1 ions (31) from glycosidic bond cleavage will be different for the two oxidized compounds. The Y 1 and Z 1 ions for glycosidic bond cleavage of cellobionic acid have m/z values of 203 and 185, respectively, species that include the carboxylic acid. The Y 1 and Z 1 ions for the C4 oxidized dimer have m/z values of 187 and 169, respectively, meaning no oxidation in the reducing end. The corresponding B 1 and C 1 ions (for the

MS analysis of native and oxidized cellobiose and cellotriose compounds
The table shows the m/z values of sodium and lithium adducts of the different fragments. n.a., not analyzed. # represents [Mϩ2Metal-H] ϩ . Double ox represents a double oxidized oligosaccharide with a gemdiol in the nonreducing end and an aldonic acid in the downstream end. Fragmentation nomenclature according to Domon and Costello (31).
tively high m/z 187 signal is also typical for Glc4gemGlc. In the case CDH is used as the electron donor for a C4-oxidizing LPMO, double oxidized products will emerge (Fig. 2B, CDH trace) with yet another characteristic MS/MS pattern, as illustrated in Fig. 5. Lithium and sodium adducts of MS/MS fragments obtained for the different LPMO products are summarized in Table 1.
Product Identification by NMR-To obtain proof for the identity of the oxidized reaction products, NMR spectroscopy was used to analyze products generated from cellopentaose by NcLPMO9C in the presence of either CDH or hydroquinone (Fig. 6). The individual monosaccharide residues were assigned by starting at the anomeric signal and/or at the primary alcohol group at C6 and then following the proton-proton connectivity using TOCSY, DQF-COSY/IP-COSY, and 13 C HSQC-[ 1 H, 1 H]TOCSY, whereas connectivity between the individual monosaccharide residues was obtained from the HMBC spectrum (see Table 2 for assignment of chemical shifts). These experiments showed the presence of dimeric and trimeric products in both reactions, as expected from the MS analyses described above. An overlay of the 13 C HSQC spectra of product mixtures obtained in the presence of hydroquinone or CDH (Fig. 6A) readily shows different occurrence of oxidations. C4 oxidation occurs in both reactions, whereas signals reflecting C1 oxidation are only observed in the reaction with CDH (red signals in Fig. 6A). There were no indications of oxidation of C6 because this part of the 13 C HSQC spectrum was essentially identical to that observed for nontreated cellopen-taose and because no novel signals possibly reflecting additional oxidations were observed. 13 C HMBC spectra provided further insight into the nature of the products. This is illustrated by Fig. 6B, showing an overlay of the 13 C HSQC and the 13 C HMBC spectra of products obtained in the reaction with NcLPMO9C and CDH. The overlay shows correlations from H/C-5 and H/C-3 to peaks with a carbon chemical shift at C4 of 95.9 and 175.2 ppm, corresponding well to the chemical shift of a gemdiol (29) and keto group, respectively. The gemdiol and keto groups account for ϳ80 and 20% of the signal intensity, respectively. Although documentation on the keto:gemdiol ratio in literature is limited, the gemdiol would be expected to dominate at pH 6.0 because the keto group is easily hydrated (29) in aqueous medium. The overlay also shows a correlation from the H/C-2 to a peak with a carbon chemical shift of 181.1 ppm that corresponds well to the shift expected when a carboxylate group is present at position C1 (12). Even though there are two different Glc1A groups in the product mixture (one in a trimeric and one in a dimeric product), their chemical shifts are very similar, meaning that the peaks appear nearly at the same position in the spectra, looking like one broad peak (see Table 2 for more details). Thus, for the first time, NMR has been used to prove that the products generated by a nonreducing end active LPMO are oxidized in the C4 position and that the products primarily exist as a gemdiol.
Suppression of H 2 O 2 Production-In the absence of a cellulosic substrate, activated LPMOs produce H 2 O 2 (17). We employed this ability to demonstrate binding of cello-oligo- FIGURE 6. NMR analysis of reaction products. A, overlay of 13 C HSQC spectra for reaction products generated by treating 0.9 mg/ml cellopentaose with 2.9 M NcLPMO9C in the presence of 10 mM hydroquinone (black signals) or 0.9 M MtCDH (red signals). The samples were both in 99.996% D 2 O with 5 mM sodium acetate, pH 6.0, and spectra were recorded at 25°C. Peaks in the proton/carbon signals of the C4 oxidized monosaccharide residue are marked by H/C#, where # refers to the ring carbon number (overlapping red and black signals). Peaks in the proton/carbon signals of the C1 oxidized monosaccharide residue are marked by H/C#* (only red signals). For the sake of simplicity, peaks related to nonoxidized monosaccharide residues are not marked (a full assignment of chemical shifts is provided in Table 2). B, two details of an overlay of a 13 C HSQC spectrum and a 13 C HMBC spectrum recorded for products obtained in a reaction with both NcLPMO9C and CDH. The left panel shows a correlation (indicated by a vertical line) from the H/C2* peak in HSQC (red) to a peak with a carbon chemical shift of 181.1 ppm in HMBC (blue), corresponding well to the (expected) presence of a carboxylate group at position C1. The right panel shows correlation from the H/C3 and H/C5 peaks in HSQC (red) to carbon peaks with a carbon chemical shift at C4 of 95.9 and 175.2 ppm in HMBC (blue), corresponding to the presence of a gemdiol and a keto group at C4, respectively. meric substrates to NcLPMO9C. Fig. 7 shows that production of H 2 O 2 was diminished in the presence of cello-oligosaccharides with a DP Ͼ4. Although only minor inhibition was observed in the presence of Glc 4 , almost complete inhibition of H 2 O 2 formation was observed in the presence of Glc 6 . Thus, the data in Fig. 7 show that cello-oligomers with a minimal length of five sugars form complexes with the enzyme that are sufficiently strong to suppress the futile H 2 O 2 generating side reaction. This corresponds very well with the HPLC data, which showed high activity for Glc 5 and Glc 6 and much lower activity for Glc 4 (Fig. 2B). As mentioned above, NcLPMO9C is not active on Xyl 5 or Man 6 , and Fig. 7 shows that these substrates do not suppress the H 2 O 2 production.
Sequence and Structural Model of NcLPMO9C-For comparison, the sequence of NcLPMO9C was aligned to a C1 oxidizing and a nonreducing end oxidizing LPMO. The AA9 domain of NcLPMO9C shares 36.6% sequence identity with C1-oxidizing PcLPMO9D (7) and 47.5% with nonreducing end oxidizing NcLPMO9D (8,16) (Fig. 8A). The metal coordinating residues (His-1 and His-83 in NcLPMO9C) as well as four residues surrounding the copper site (Asn-26, His-155, Gln-164, and Tyr-166 in NcLPMO9C) are conserved in all three AA9s. Looking at the LPMO structures available, AA9s tend to contain several surface-exposed aromatic residues that seem to be aligned to interact with a cellulose chain that would then traverse the catalytic center (32). Indeed, these aromatic residues have roughly the same spatial orientation as residues that, by experiment, have been shown to interact with chitin in an AA10 LPMO (33). This adds confidence to the notion that these aromatic residues interact with the substrate, possibly analogous to what is been suggested for PcLPMO9D (32).
Based on the crystal structure of NcLPMO9D a model was built for NcLPMO9C. Comparison of the surface-exposed residues potentially involved in substrate binding (Fig. 8B) shows that there are more aromatic residues on the surfaces of NcLPMO9D and PcLPMO9D compared with NcLPMO9C. Because of this difference, the length of the substrate-binding surface seems shorter in NcLPMO9C. Even if one considers the contribution of protruding polar residues on the surface, the binding surface of NcLPMO9C seems less extended (Fig. 8B) and possibly more adapted to binding shorter substrates compared with PcLPMO9D and NcLPMO9D. Interestingly, the structural model of NcLPMO9C indicates that this enzyme has a cluster of three asparagine residues (Asn-25, Asn-26, and Asn-27) in a location that could potentially be a ϩ2 subsite. One could speculate that these asparagines interact with the reducing end sugar. Indeed, as illustrated in Fig. 8B, a cellopentaose would span the putative binding surface of NcLPMO9C, when its reducing end is positioned near these asparagines. Such an orientation would be in accordance with the observation that cellopentaose primarily is cleaved into a cellotriose and a Glc4gemGlc.

TABLE 2 Assignment of chemical shifts
The individual monosaccharide residues were assigned by starting at the anomeric signal and/or at the primary alcohol group at C6 and then following the proton-proton connectivity using TOCSY, DQF-COSY/IP-COSY, and 13 C HSQC-TOCSY spectra. 13 C HSQC was used for assigning the carbon chemical shifts. The 13 C HMBC spectrum provided long range bond correlations, allowing identification of the connectivity between the sugar units and allowing identification of the carboxyl group at C1, as well as the gem-diol group and keto group at C4. Chemical shifts were assigned for the major products resulting from treatment of 0.9 mg/ml cellopentaose with 2.9 M NcLPMO9C in the presence of 0.9 M MtCDH or 10 mM hydroquinone in 99.996% D 2 O with 5 mM sodium acetate pD 6.0, for spectra recorded at 25°C. Numbers 1-6 represent ring carbon numbers to which the chemical shift values ( 1 H, 13 C or 1 H, 1 H, 13 C for C6) are assigned for a nonreducing end glucose (NR), a second after NR glucose (SNR), the ␣ anomer of glucose (␣), the ␤ anomer of glucose (␤), the aldonate glucose (C1), and the keto/gemdiol glucose (C4). The chemical shifts reported are with accuracy of 0.01 ppm for 1 H and 0.1 ppm for 13   Conclusions-In this article, we have shown that NcLPMO9C cleaves both crystalline cellulose as well as cello-oligosaccharides yielding products oxidized in the nonreducing end. By applying NMR, we unambiguously showed that the nonreducing end sugar was oxidized at the C4 position and that this sugar primarily exists as a gemdiol in solution. We have also shown that MS/MS fragmentation analysis of oxidized products can be applied to differentiate between C1 and C4 oxidizing LMPOs. A range of different substrates were tested, but NcLPMO9C was only active on ␤-1,4-linked glucose units, and the enzyme FIGURE 8. Sequence and structural comparisons of PcLPMO9D, NcLPMO9D and NcLPMO9C. A, structure-guided sequence alignment of PcLPMO9D, NcLPMO9D, and NcLPMO9C with fully conserved residues shaded in gray. Residues involved in coordination of the copper are marked with asterisks above the sequence (His-1, His-83 and Tyr-166 in NcLPMO9C), whereas aromatic surface residues and protruding polar surface residues potentially involved in substrate binding are colored in red (all highlighted in B). Cysteines forming disulfide bonds are indicated in boxes with numbers above showing which cysteines are connected. B, side chains on the substrate binding surface of PcLPMO9D (Protein Data Bank code 4B5Q) and NcLPMO9D (Protein Data Bank code 4EIR) compared with the modeled structure of NcLPMO9C. Copper coordinating residues are colored dark gray, aromatic and protruding polar residues putatively involved in substrate binding are colored yellow and violet, respectively, and the copper is shown as an orange sphere. For illustration purposes only, a cellopentaose (coordinates derived from Protein Data Bank code 2EEX; shown in green with oxygens in red) is placed above the surface of the modeled structure of NcLPMO9C. The view in the right panels is rotated 90°relative to the view in the left panels, looking down at the flat surface containing the copper binding site. seemed to require a minimum stretch of at least four glucose units. It has not escaped our attention that such an activity on short cello-oligosaccharides could imply that the enzyme is active on hemicellulose structures containing ␤-1,4-linked glucose units.