Group I lytic polysaccharide monooxygenase (LPMO1) is required for e�cient chitinous cuticle turnover during insect molting

22 Microbial lytic polysaccharide monooxygenases (LPMOs) catalyze the oxidative 23 cleavage of crystalline polysaccharides including chitin and cellulose. The discovery 24 of a large assortment of LPMO-like proteins widely distributed in insect genomes 25 suggests that they could be involved in assisting chitin degradation in the exoskeleton, 26 tracheae and peritrophic matrix during development. However, the physiological 27 functions of insect LPMO-like proteins are still undetermined. To investigate the 28 functions of insect LPMO subgroup I-like proteins, which contain an AA15 LPMO 29 catalytic domain and a conserved C-terminal cysteine-rich motif, two evolutionarily 30 distant species, Tribolium castaneum and Locusta migratoria , were chosen for study. 31 RNAi for the T. castaneum protein, TcLPMO1, caused molting arrest at all 32 developmental stages, whereas RNAi of the L. migratoria protein, LmLPMO1, 33 prevented only adult eclosion. In both species, LPMO1-deficient animals were unable 34 to shed their exuviae and died. TEM analysis revealed failure of turnover of chitinous 35 cuticle, which is critical for completion of molting. Purified recombinant LPMO1-like 36 protein from Ostrinia furnacalis (rOfLPMO1) exhibited oxidative cleavage activity 37 and substrate preference for chitin. These results reveal for the first time the 38 physiological importance of catalytically active LPMO1-like proteins from distant 39 insect species and provide new insight into the enzymatic mechanism of chitin 40 turnover during molting. 41


Introduction
Insect cuticle consists of morphologically distinct tissue layers including the outermost envelope, epicuticle and innermost procuticle (exocuticle and endocuticle), the latter consisting of a large number of horizontally oriented chitin-protein rich laminae [1][2][3] .To accommodate growth, insects periodically replace their cuticle in a process involving ecdysis and molting.The turnover of chitin in the cuticle by enzymes presents several challenges because of the crystallinity of the anti-parallel chitin chains and potential covalent cross-linking to cuticular proteins.The roles of chitinolytic enzymes are relatively well understood [4][5][6] .However, the functions of lytic polysaccharide monooxygenases (LPMOs) that have the potential to digest insect chitin are less well studied and thus, is the focus of this research.
Bacterial LPMOs initially were identified as non-catalytic carbohydrate-binding proteins (CBP21 and CBM33), which enhanced polysaccharide degradation catalyzed by glycosyl hydrolases 19 .However, Vaaje-Kolstad et al. 20 demonstrated subsequently that the proteins CBP21 and CBM33A from Serratia marcescens and Enterococcus faecalis, respectively, are able to cleave glycosidic bonds of crystalline chitin in an oxidative fashion, creating new access points to the substrate for chitinolytic enzymes.After these initial findings, other enzymatic properties such as substrate specificity and modes of action of LPMOs from microorganisms and viruses were examined extensively.Those studies demonstrated that LPMOs could cleave recalcitrant polysaccharides including chitin, cellulose, starch, xylan, pectin and various other hemicelluloses as well as soluble cello-oligosaccharides [13][14][15][21][22][23][24][25][26][27] . Only afew studies on the enzymatic properties of LPMO-like proteins from insects have been reported so far.For example, Sabbadin et al. 8 identified 23 genes encoding LPMO-like proteins (TdLPMOs) in the transcriptome of the firebrat, Thermobia domestica (Td), which is an ancient insect species capable of digesting crystalline cellulose by using its own enzymes without the assistance of enzymes from microbial symbionts, with 21 of them (including TdAA15A and TdAA15B) present in the gut proteome.The recombinantly expressed TdAA15A protein, which is one of the most abundant LPMO-like proteins from T. domestica, exhibited synergistic activity with enzymes of glycosylhydrolase families GH6 (cellobiohydrolase) and GH18 (endochitinase) on the breakdown of cellulose and chitin, respectively.In contrast, another recombinant enzyme, TdAA15B, showed activity on crystalline chitin but not on cellulose.
The primary role of bacterial LPMOs appears to be the degradation of polysaccharides as a nutrient carbon source.In addition, some studies of entomopathogenic bacteria, oomycetous fungi and viruses have demonstrated their functional importance in pathogenicity.For instance, Paneibacillus larvae is a bacterial pathogen that causes a serious disease of honeybees, American Foulbrood (AFB).PlCBP49 protein containing an LPMO domain (AA10 family) of P. larvae is critical for the degradation of the chitin-rich peritrophic matrix (PM) in the lining of the midgut of bee larvae, which is a vital step in the invasion of P. larvae during infection 29 .Phytophthora infestans is a damaging crop pathogenic oomycete that infects both potato and tomato crops.A new family of AA17 LPMO from P. infestans has been reported to oxidatively cleave the backbone of pectin, playing an important role as virulence factors 18 .Similarly, spindles produced by entomopoxviruses (EVs) are cellular crystals of the LPMO-domain-containing protein, fusolin (AA10 family).This protein disrupts the host's PM, leading to the greatly enhanced infectivity of EVs 30 .
Because chitin is a major structural component of insect cuticle, tracheae and the PM 6 , LPMO-like proteins could be involved in the degradation of chitin in those tissues during development.However, the exact physiological functions of insect LPMO genes are still not determined.In this study, we report on the physiological function of LPMO-like proteins comprising one of the subgroups of insect LPMO (denoted as group I LPMO) whose members have an AA15 LPMO catalytic domain and a conserved long C-terminal stretch of ~120 amino acids containing a cysteine-rich motif.Three economically important agricultural pests from three orders of insects were used as model species to investigate the function of insect LPMO1s.
The holometabolous red flour beetle, Tribolium castaneum (Coleoptera), and hemimetabolous migratory locust, Locusta migratoria (Orthoptera), were utilized to study the biological function(s) of LPMO1s by RNA interference (RNAi), while an LPMO1 from the holometabolous Asian corn borer, Ostrinia furnacalis (Lepidoptera), was used to study the enzymatic properties of this class of enzymes.We provide experimental evidence for a functional importance of insect LPMO1s in the catabolic oxidative breakdown of cuticular chitin, which is also degraded by a mixture of molting fluid chitinases and β-N-acetylhexosaminidases [4][5][6] .This work provides new insight into the molecular mechanism underlying the vital process of chitin degradation in insect molting.

Phylogenetic analysis of insect LPMOs
A gene family encoding LPMO-like proteins was found recently in an ancient insect species, T. domestica, and several other species including D. melanogaster and C. gestroi 8,31 .This family is classified as members of auxiliary activity family 15 (AA15) in the Carbohydrate Active enZymes (CAZy) database 17 .By our initial search using the TdAA15A of T. domestica as a query, we identified nine genes in the genome of T. castaneum that encode LPMO (AA15 family)-like proteins.Using these predicted LPMO protein sequences as queries, we searched other insect genomes that have been fully sequenced and annotated.A phylogenetic analysis of these sequences indicated that insect LPMO-like proteins can be divided into at least four major clusters, which we denoted as groups I (LPMO1s), II (LPMO2s), III (LPMO3s) and IV (lepidopteran-specific LPMOs) (Fig. 1a).Interestingly, 21 TdLPMOs identified in the gut proteome of T. domestica 8 comprise a separate clade.The groups LPMO1 and LPMO2 present in all orders of Insecta appear to have a single representative in all of the species identified, whereas the number of representatives in the LPMO3 group ranged from one to seven in different species.Note that group III LPMOs do not include any representatives from lepidopteran species characterized so far.Instead, there is a separate group with representatives from lepidopterans only, often with more than one member.All LPMO-like proteins identified consist of a putative signal peptide and an AA15 catalytic domain.In addition, LPMO1s have a C-terminal stretch containing a cysteine-rich motif, whereas LPMO2s from group 2 have a predicted transmembrane span.Phylogenetic trees were constructed by MEGA 7.0 software using the Neighbor-Joining method.See Supplementary Table 2 for the accession numbers of LPMO1 proteins used in this study.

Sequencing of TcLPMO1, LmLPMO1 and OfLPMO1 cDNAs
For this study we focused only on members of the LPMO1 subgroup because their single-copy orthologous genes encoding not only a AA15 catalytic domain but also a C-terminal cysteine-rich motif are present in all insect orders so far examined, indicating that they are likely to have essential physiological functions.In addition to the insect genomes that have been fully sequenced and annotated, we also performed BLAST searches of the L. migratoria and O. furnacalis transcriptomes using the TcLPMO1 and/or TdAA15A protein as queries.We identified LPMO1 orthologs from L. migratoria (LmLPMO1) and O. furnacalis (OfLPMO1) as well as from other species of other insect orders including the Coleoptera, Hymenoptera, Diptera, Hemiptera, Orthoptera and Isoptera (Fig. 1b).Our searches confirmed that there is single member of the LPMO1 subgroup in species of those insect orders and that TcLPMO2 members form an outgroup.The LPMO1 subgroup is of an ancient origin and must have diverged from a common progenitor of the LPMO2 subgroup.Using primers flanking the predicted start and stop codon regions, we were able to amplify TcLPMO1, LmLPMO1 and OfLPMO1 cDNAs including the entire open reading frames that encode proteins with 337, 339 and 343 amino acid residues and theoretical molecular masses for the mature proteins of 35.8, 34.6 and 35.3 kDa, respectively (Supplementary Fig. 1).Each protein, as seen in the members of group I LPMOs, has a putative signal peptide and a conserved 192 or 193 amino acids-long AA15 catalytic domain followed by a ~120 amino acids-long C-terminal stretch containing a 74 amino acids-cysteine-rich motif that contains two 6-cysteine-containing internal repeats (C-X 15 -C-X 3 -C-X 6-9 -C-X 4 -C-X 1 -C) (Supplementary Fig. 2).Sabbadin et al. 8 reported about several amino acid residues that are critical for LPMO activity as revealed by the crystal structure of TdAA15A in the 190 amino acids-long AA15 catalytic domain (see Supplementary Fig. 2); they include two histidines (His1 and His91), which directly coordinate a copper ion with a T-shaped geometry in the catalytic site, known as the histidine brace; alanine (Ala89) and tyrosine (Tyr184), which are also involved in forming the copper-containing active center as non-coordinating amino acid residues occupying the apical site and axial position of the copper ion, respectively; and two other tyrosines (Tyr24 and Tyr166) located at the boundaries of the flat surface surrounding the active center, which could be involved in substrate binding.All insect LPMO1-like proteins in our analysis (Supplementary Fig. 2) have the 192 amino acids-long AA15 catalytic domain except for some orthopteran LPMO1s, which are composed of 193 amino acids.All members of this group share a high degree of amino acid sequence identity/similarity (54-99 and 78-100%, respectively).The catalytically critical amino acid residues (H1, Y24, A/S89, H91, W168/169 and F186/187) occupy the same positions in the sequence alignment.All of them differ from the TdAA15A enzyme in the replacement of Y166 by W168 in the boundaries of the flat surface surrounding the active center and Y184 by F186 in the axial position of the copper-binding site (Supplementary Fig. 2).

Developmental and tissue-specific expression of LPMO1 genes in T. castaneum and L. migratoria
To determine the role(s) of LPMO1s in insect development, the expression profiles of TcLPMO1 and LmLPMO1 during various growth stages were conducted using real-time PCR.With T. castaneum, transcripts of TcLPMO1 were detected in all developmental stages analyzed with lowest expression in embryos and mature adults and highest expression in the pharate pupal stage (Fig. 2a).During late stages of development from pharate pupae to day 7 adults, high transcript levels of TcLPMO1 were detected at early pharate pupal (PP1) and early pupal (P0-P2) stages and declined thereafter (Fig. 2b).To assess the tissue specificity of expression of TcLPMO1, we dissected late stage larvae to obtain midgut and carcass (whole body minus midgut) tissue preparations.The transcript level of TcLPMO1 in the carcass was substantially higher than that in the midgut (Fig. 2c).
In L. migratoria, the temporal expression pattern of LmLPMO1 was analyzed during the later stages of development from the 4 th instar nymph to the adult stage.Transcripts of LmLPMO1 were observed at all stages with higher levels detected in the 4 th instar day 5 and 5 th instar days 1-3 during the nymph-nymph molt (Fig. 2d).The tissue-specific expression analysis showed that LmLPMO1 is expressed at a higher level in the epidermis than in other tissues analyzed such as the midgut, testis and ovary (Fig. 2e).These results suggest a role of both TcLPMO1 and LmLPMO1 in the turnover of chitin in the cuticle.

Biochemical properties of OfLPMO1
To further illustrate that the LPMO1s participate in the degradation of chitin during molting, we attempted to express the full-length proteins, TcLPMO1s, LmLPMO1 and OfLPMO1 from T. castaneum, L. migratoria and O. furnacalis, respectively, in yeast cells.Only the recombinant expression of OfLPMO1, which shares an amino acid sequence identity of 71% and 63% with those of TcLPMO1 and LmLPMO1, respectively, was successful (Fig. 3a).The recombinant OfLPMO1 was found to have a peroxidase activity when using 2,6-dimethoxyphenol (2,6-DMP) as a chromogenic substrate and H 2 O 2 as a co-substrate at pH 6.0 and 30°C (Fig. 3b).The reaction was linear over a 10-minute reaction time and the specific activity of the purified enzyme was estimated to be about 16.0 U/g under the experimental conditions, comparable to the values obtained for two enzymes purified from C. gestroi (6.2 and 7.6 U/g) 31 .
rOfLPMO1 also oxidatively hydrolyzed -chitin and produced a series of chitin oligosaccharides in aldonic acid or lactone forms.The mass-to-charge ratio (m/z) profiles of the digestion products in the mass spectral analysis consisted of oxidized oligosaccharides with different degrees of polymerization (DP).The products corresponding to even-numbered oligosaccharides with DP6 and DP8 had higher intensities than the odd-numbered products with DP5 and DP7 (left panel in Fig. 3c).
No oxidized product was detected either in the enzyme-omitted negative controls or in a reaction where micro-cellulose was used as the substrate (middle and right panels in Fig. 3c), suggesting LPMO1s are not involved in cellulose degradation.A potential synergistic effect between rOfLPMO1 and recombinant chitinase h (rOfChi-h) was examined using either αor β-chitin as the substrate.As shown in Fig. 3c and Table 1, a mixture of rOfLPMO1 and rOfChi-h generated more reducing sugar from both substrates than the sum generated from reactions catalyzed by individual enzymes, indicating that there is a synergistic effect in chitin degradation between LPMO1 and chitinase h.(Supplementary Fig. 3a) when the targeted gene is maximally expressed (see Fig. 2b).
Similarly, injection of dsLmLPMO1 into the 5 th instar day 1 nymphs caused a substantial decrease in the level of LmLPMO1 transcripts (Supplementary Fig. 3b).
In T. castaneum, injection of dsTcLPMO1 into early instar larvae had no effect on the subsequent molt and the resulting penultimate instar larvae developed normally (Fig. 4a).However, all of the larvae failed to complete the molt to the last larval instar (phenotype 1 in Fig. 4a).Slippage of the old larval cuticle was observed, but the penultimate instar larvae were trapped inside the old cuticle and died.When penultimate instar larvae were treated with dsTcLPMO1, the insects molted normally to the last instar larvae, but they subsequently failed to complete the larval-pupal molt and died as pharate pupae entrapped in the old larval cuticle (Phenotype 2 in Fig. 4a).When dsTcLPMO1 was injected into last instar larvae, developmental arrest occurred during the pupal-adult molt when the pharate adults became entrapped in their pupal cuticle.
Pupal cuticle slippage was evident, but the insects were unable to shed the pupal exuvium and died (Phenotype 3 in Fig. 4a).Similarly, injection of dsTcLPMO1 into day 0 pupae also caused an incomplete pupal-adult molting in all of the insects (Phenotype 3 in Fig. 4a).These results indicate that TcLPMO1 is required for all types of molting (larval-larval, larval-pupal and pupal-adult) in T. castaneum.
In contrast, injection of dsLmLPMO1 into the 5 th instar day 1 nymphs resulted in terminal developmental arrest.Approximately 10% of the nymphs died without dorsal splitting (Phenotype 1 in Fig. 4b), while ~70% of the nymphs initiated a molting process exhibiting both a dorsal split and new adult cuticle.However, the insects failed to shed the antecedent cuticle and died (Phenotype 2 in Fig. 4b).All of these results suggest that insect LPMO1 appears to play an essential role during the molting process, presumably in turnover of cuticle chitin, which is critical for completion of the molt.To evaluate this hypothesis, we further analyzed by TEM the ultrastructure of cuticles from TcLPMO1-and LmLPMO1-deficient insects.or exhibiting a dorsal splitting during the molt but entrapped in their nymphal cuticle (Phenotype 2, ~70%).

Ultrastructure of cuticles from LPMO1-deficient insects
Because loss of function of LPMO1 by RNAi caused molting arrests in both T. castaneum and L. migratoria, we analyzed by TEM the morphology and ultrastructure of the old and newly forming cuticles during molting periods in both species.In T.
castaneum, there were no obvious differences in morphology of the newly forming body wall cuticles of dsTcLPMO1-treated insects isolated at the penultimate larval instar, pharate pupal and pharate adult stages when compared with the dsTcVer-treated controls.New cuticles from both control and TcLPMO1-depleted insects exhibited well-organized, horizontal alternating electron-dense and electron-lucent chitinous laminae, as well as vertically oriented helicoidal pore canals in the larval and pupal cuticles or wide vertical structures with a central chitin fiber core of pore canal fibers (PCFs) in the adult cuticle (bottom panels in Fig. 5a-c).In the old cuticles at each molting stage analyzed, the endocuticles of dsTcVer-treated controls had been degraded (top left panels in Fig. 5a-c).In contrast, RNAi for TcLPMO1 resulted in failure of turnover of the endocuticular layer in which the horizontal chitinous laminae and vertical pore canals remained essentially intact (top right panels in Fig. 5a-c).To investigate the functional importance of LmLPMO1 in turnover/morphology of the old and/or new cuticles during adult eclosion in L. migratoria, those tissues were dissected from the third abdominal segment of pharate adults (5 th instar day 9 nymphs) that had been injected with dsGFP (control) or dsLmLPMO1 on 5 th instar day 1 and again on day 3. Paraffin sections of tissues stained with hematoxylin and eosin showed that the old cuticle of dsLmLPMO1-treated insects was much thicker than that of dsGFP-treated controls (Fig. 6a), suggesting that LmLPMO1-deficient insects might have failed to digest the old cuticle.To confirm that the turnover of the old cuticle was affected, we further performed TEM analysis of old and new cuticles.The endocuticle portion of the old cuticle of dsGFP-control insects was nearly completely digested, whereas that of dsLmLPMO1-treated insects appeared to be intact, retaining numerous horizontal chitin laminae and vertical pore canals (left panels in Fig. 6b).As observed with cuticles of TcLPMO1-depleted T. castaneum, RNAi for LmLPMO1 yielded no obvious differences in morphology of either the horizontal laminae or vertical pore canals in the newly formed adult cuticle compared with that of the dsGFP-control (right panels in Fig. 6a).All of these results indicate that LPMO1 plays a role during molting in digestion of chitin in the old cuticle, which is critical for the completion of the molt, but not in the formation of new cuticle or its morphology including the chitinous laminar organization and pore canal/PCF structure in both T. castaneum and L. migratoria.

Discussion
The degradation of extracellular matrix polysaccharides in insects and plants presents special challenges because of the crystallinity and possible cross-linking of the substrates.A commonly held view envisaged that a combination of families of endoand exo-acting hydrolytic enzymes could accomplish the digestion of recalcitrant polysaccharides including chitin and cellulose 32,33 .However, the relatively recent discovery that proteins collectively known as LPMOs, which bind to these substrates, promote utilization of insoluble substrates and have oxidase activities of their own, has provided new insight about the mechanism of turnover of naturally occurring polysaccharides by microbes as well as arthropods 8,[12][13][14][15][16][25][26][27] . LPMOs rom microbes break the C-H bond of either C1 or C4 carbons of polysaccharides using a peroxidase activity with H 2 O 2 as the co-substrate resulting in aldonic acid or lactone products.This process results in the generation of internal entry sites in the polysaccharide for the hydrolytic enzymes such as chitinases and cellulases in addition to facilitating the decrystallization of chitin or cellulose chains from crystalline bundles 34 .
Besides microbes, arthropods also contain a variety of LPMO-like proteins.
Recently, analysis of the gut proteome and transcriptome of the ancient insect species, T. domestica from the order Zygentoma, has confirmed the expression in the gut of a large family of LPMOs belonging to the AA15 family of enzymes.These proteins promote digestion of dietary crystalline cellulose without the assistance of gut microbes, unlike termite species that require such microbial assistance for cellulose digestion.
Individual TdLPMOs differed in their ability to act on either chitin or cellulose or both.
On the other hand, two LPMOs from the lower termite, C. gestroi, could act only on chitin and not on cellulose, indicating their role in structural matrix remodeling rather than in digestion.So far, the only evidence for the idea that LPMOs might have a role in insect cuticle turnover comes from genome-wide RNAi of D. melanogaster and the iBeetle project that identified pupal lethality and other defects in T. castaneum (http://ibeetle-base.uni-goettingen.de/).However, there have been no direct studies on chitin turnover in the cuticle or on the enzymatic activity/specificity of LPMO-like proteins in these species.Our study addresses these specific issues and points to an essential role for these enzymes in chitin turnover, which is not met alone by the assortment of endo-and exochitinases.
The nine TcLPMO homologs and other LPMO-like proteins deduced from 12 insect genomes obtained with TdLPMO as query are distributed into at least four distinct branches and display unusual properties.There is only one representative from each insect species in LPMO groups I and II, which have distinctly different protein domain compositions.All group I proteins have two copies of a 6-cysteine repeat often associated with chitin-binding domains.However, these cysteine repeats are not related to the CBM14 domains present in insect chitinase-like proteins 35 .Group II LPMOs lack this domain but have a predicted C-terminal transmembrane domain.Group III proteins in several insect genomes contain redundant enzymes that have only the AA15 catalytic domain.Group IV proteins have an interesting distribution among insects, being prevalent only in lepidopteran species.The significance of their unique distributions remains unknown.The presence of these distinct groups among insect species suggests functional differences among them.This inference is further supported by the limited data that we currently have on tissue specificity of expression of these LPMO groups.Groups I and II enzymes from both T. castaneum and L. migratoria are expressed predominantly in the epidermis but not in the midgut, suggesting their involvement in chitin remodeling in the cuticle rather than a digestive function.They are also expressed at comparable periods in the molt cycle and not in the adult stage.In contrast, all seven group III enzymes from T. castaneum are expressed almost exclusively in the gut and not in the carcass (Supplementary Fig. 4), suggesting their involvement in digestion of dietary chitin and/or peritrophic matrix-associated chitin, but not in cuticle chitin turnover.
We were able to express recombinantly the full-length OfLPMO1 protein from O.
furnacalis, which was also identified in our BlastP search.OfLPMO1 shares 71% identity with the TcLPMO1 from T. castaneum and 63% identity with the LmLPMO1 from L. migratoria and also has the 12 cysteine-containing C-terminal motif.Besides exhibiting peroxidase activity with 2,6-DMP as the substrate, the highly purified enzyme oxidized crystalline αand β-chitins, but not cellulose, yielding chitooligosaccharide products consistent with a role in chitin turnover, but not in the digestion of dietary cellulose, similar to TdAA15B of T. domestica.It should be emphasized that besides sharing a high degree of amino acid sequence identity, the three LPMO1 proteins from T. castaneum, L. migratoria and O. furnacalis have the same four essential amino acids (H1, H91, A89 and F186) involved in the T-brace structure containing the catalytic Cu atom first identified in the TdAA15B enzyme, which is specific for chitin 8 .Our studies provide clear cut evidence that, compared to the control insects showing dissolution of the old cuticle, following RNAi of just one of several LPMO genes present in insect genomes, the laminar architecture of the old cuticle remains mostly intact, even though the assortment of chitinolytic enzymes is produced normally.
Note that the LPMO-depleted insects do complete one molt cycle presumably utilizing the chitinolytic enzymes.This report is the first demonstration of the essential role of an LPMO in promoting the turnover of the chitinous cuticle.The failure to molt after depletion of a single oxidative enzyme supports the hypothesis that LPMOs are as equally important as the chitinolytic enzymes in the digestion of cuticle chitin at each molt cycle.For applications in economic and medical entomology for the development of novel insecticides, LPMOs may be potential targets identified via comparative and functional insect genomics.

Insects
The GA-1 strain of T. castaneum 36 was used for this study.Beetles were reared at 30°C and 50% relative humidity in whole wheat flour containing 5% brewer's yeast as described previously 37 .L. migratoria was kindly provided by the Institute of Zoology, Chinese Academy of Sciences (CAS).Nymphs were reared on fresh wheat sprouts in the laboratory at 28°C under a 14 h light/10 h dark diurnal cycle.α-factor at the N-terminus in frame.The recombinant plasmid was linearized using Sac I and then transformed into Pichia pastoris strain GS115.Positive clones were selected and the recombinant OfLPMO1 protein (rOfLPMO1) was obtained by induction with 1% methanol for 120 h.rOfLPMO1 was purified by 75% saturation of ammonium sulfate precipitation, followed by affinity chromatography using β-chitin beads as described previously 40 .The purity of affinity-purified rOfLPMO1 was analyzed by 15% SDS-PAGE.

Enzyme assay
Enzymatic activity of the purified rOfLPMO1 was measured by utilizing the peroxidase activity associated with this class of enzymes using 2,6-dimethoxyphenol (2,6-DMP) and H 2 O 2 as co-substrates 41 with modifications.The reaction solution containing 5 mM 2,6-DMP and 100 µM H 2 O 2 in 200 µl of 100 mM sodium phosphate (pH 6.0) was preincubated at 30°C for 10 min, and then the absorbance at 469 nm was measured every 30 s shortly after adding rOfLPMO1 (final concentration of 5 µ M ).In the control group, the reaction solution contained the same components except that no enzyme was added.For determination of activity of rOfLPMO1 toward chitin and cellulose, rOfLPMO1 (5 µ M ) was incubated with 2 mg/ml β-chitin or microcrystalline cellulose (Sigma) in 300 µl of 20 mM sodium phosphate buffer (pH 6.0) containing 1 mM ascorbic acid at 30°C for 24 h with rotation.The reaction mixture was centrifuged at 17,000 x g for 10 min and 0.5 µL of the reaction product in the supernatant was analyzed by MALDI-TOF/TOF mass spectrometry (Waters).
Both α-chitin and β-chitin were used to detect the synergistic effect (if any) between rOfLPMO1 and recombinant OfChi-h (rOfChi-h).For the single enzyme hydrolysis condition, a final concentration of 1 µM rOfChi-h or 5 µM rOfLPMO1 was added in a total volume of 1.0 mL containing 100 mM sodium phosphate buffer (pH 6.0) and 2 mg/mL chitin in a 2 mL Eppendorf tube.For the two-enzyme combination condition, 1 μM rOfChi-h and 5 μM rOfLPMO1 were added to the reaction system together.In addition, 1 mM ascorbic acid was added to the reaction mixture whenever LPMO was included in the assay.These tubes were incubated horizontally in an incubator at 200 rpm for 48 h at 30°C.A 0.06 mL sample was withdrawn from well-mixed digestion mixtures at selected time-points during digestions.An aliquot of 0.18 mL potassium ferriferrocyanide (2 mg/mL) was then added and the mixture was boiled for 15 min.The amount of reducing sugar generated corresponded to the potassium ferriferrocyanide consumption, which was quantified by measuring the absorbance at 420 nm.All the assays were performed in triplicate.

RNA interference (RNAi)
Templates for synthesis of double-stranded RNA (dsRNA) for TcLPMO1 (dsTcLPMO1) and LmLPMO1 (dsLmLPMO1) were amplified by PCR using the gene-specific primers containing T7 RNA promoter sequences at the 5′-ends.dsRNAs were synthesized and purified according to the protocol described previously 42,43 .In T.
castaneum, dsTcLPMO1 (200 ng per insect) was injected into early instar larvae, penultimate instar larvae, last instar larvae or day 0 pupae (n = 20-40 for each of the three independent experiments).To analyze knockdown levels of TcLPMO1 transcripts, total RNA was isolated from whole day 1 pupae (n = 3) that had been injected with dsRNA at last instar larval stage.In L. migratoria, dsLmLPMO1 (10 µg per insect) was injected into the 5 th instar day 1 nymphs (1st injection), and the same amount of dsRNA was injected a second time 3 d after the first injection to increase RNAi efficiency (n = 10 for each of the three independent experiments).To analyze knockdown levels of LmLPMO1 transcripts, total RNA was isolated from the epidermis of 5 th instar day 5 nymphs (2 d after 2nd dsRNA injection).dsRNAs for T. castaneum Vermilion (dsTcVer) and green fluorescent protein (dsGFP) were synthesized as described previously 43,44 and injected to serve as a negative control for T. castaneum and L. migratoria, respectively.The primer sequences used and lengths of the dsRNAs are listed in Supplementary Table 1.

Histochemistry
L. migratoria pharate adults (5 th instar day 9 nymphs) that had been injected with dsLmLPMO1 or dsGFP into 5 th instar day 1 nymphs were collected, and their third abdominal segments were dissected.Samples were fixed in 4% paraformaldehyde for 24 h, dehydrated in an ethanol gradient of 30, 50, 70, 90 and 100% for 30 min each, and then embedded in paraffin.Paraffin sections (5 μm) were stained with hematoxylin and eosin (Beyotime Biotechnology) and then observed using the Olympus IX-83 inverted microscope.

Transmission electron microscopy (TEM)
Penultimate instar larvae, pharate pupae (day 2 prepupae) and pharate adults (day 5 pupae) of T. castaneum that had been treated previously with dsTcLPMO1 or dsTcVer were collected and fixed in a mixture of 4% paraformaldehyde and 0.1% glutaraldehyde in 0.1 M sodium cacodylate buffer (pH 7.4) for 24 h at room temperature.Ultrastructure of cuticles from the dsRNA-treated animals was analyzed by TEM as described previously 45 .In the case of samples of L. migratoria, pharate adults (5 th instar day 9 nymphs) that had been injected with dsLmLPMO1 or dsGFP at the 5 th instar day 1 nymph stage were collected, and the epidermis of their third abdominal segments were dissected and fixed in a mixture of 4% paraformaldehyde and 0.1% glutaraldehyde in 0.1 M phosphate buffer (pH7.4) for 24 h at room temperature.Samples were washed with 0.1 M phosphate buffer, and then fixed with 1% osmium tetroxide for 3 h.The tissues were rinsed with phosphate buffer, dehydrated in acetone, and then embedded in Epon 812 (Sigma) for 2 h at room temperature and baked in a 62°C oven for 48 h, followed by ultrathin sectioning.Ultrathin sections

Fig. 1
Fig. 1 Phylogenetic analysis of insect LPMOs.a A phylogenetic tree of putative

Fig. 2
Fig. 2 Expression profiles of TcLPMO1 and LmLPMO1.a Transcript levels of

"
rOfLPMO1 + rOfChi-h" indicates the calculated sum of the reducing sugar generated by the individual enzymes."rOfLPMO1 & rOfChi-h" indicates the reducing sugar produced by combining these two enzymes in the same reaction.Synergism coefficient = activity (rOfLPMO1 & rOfChi-h)/(activity rOfLPMO1 + activity rOfChi-h) Effect of RNAi for TcLPMO1 and LmLPMO1 on insect molting and survival Double-stranded RNA (dsRNA)-mediated transcript down-regulation (RNAi) was performed to determine the role of TcLPMO1 and LmLPMO1 in development and molting of T. castaneum and L. migratoria, respectively.To analyze the transcript abundance for each LPMO1 gene after RNAi, real-time PCR experiments were carried out.Injection of dsRNA for TcLPMO1 (dsTcLPMO1) into the last instar larvae led to a substantial depletion of TcLPMO1 transcripts at the young pupal stage (day 1 pupae)

Fig. 4
Fig. 4 Lethal phenotypes produced by RNAi for LPMO1 in T. castaneum and L.

Finally, we demonstrated
by RNAi that the LPMO1 enzymes in two different insect orders are critical for insect molting as well as cuticle digestion.Administration of dsRNA for the single copy of the gene encoding this enzyme resulted in molting failure in T. castaneum at all developmental stages.In cases other than the pupal stage, the insects managed to molt to the next developmental stage (larva, prepupa, or pupa) but died without completing the next molt cycle due to a high level of transcript depletion of the targeted gene.While we do not have a precise explanation for the delay in the development of the phenotype until the next molt cycle, we hypothesize a delay in protein depletion due to a long half-life of the LPMO protein past the point of transcript depletion.
TcLPMO1, LmLPMO1 and OfLPMO1 cDNAs T. castaneum, L. migratoria and O. furnacalis homologs of TdAA15A from T. domestica (accession number: GASN01405718.1)were identified by performing a BLAST search of the T. castaneum genome and the L. migratoria and O. furnacalis transcriptomes.To clone TcLPMO1 cDNA, total RNA was isolated from a pool of six whole T. castaneum pupae (mixture of day 0-5 pupae) by using the RNeasy Mini Kit (Qiagen).First strand cDNAs were synthesized with the Super Script III First-Strand Synthesis System (Invitrogen) using an oligo-(dT) 18 primer.To clone LmLPMO1 and OfLPMO1 cDNAs, total RNA was isolated from L. migratoria 5 th instar day 1 nymphs and O. furnacalis pharate pupae, respectively, using Trizol reagent (Invitrogen), and then treated with DNase I (Takara).First strand cDNAs were synthesized using the Reverse Transcriptase M-MLV (Takara) and used as template for cDNA amplification.cDNAs containing the predicted full-length coding sequence of TcLPMO1, LmLPMO1 and OfLPMO1 were amplified by PCR using the gene-specific primers shown in AGG GGT ATC TCT CGA GAA AAG ACA TGG AAG ATT GAT GGA CCC-3´ and 5´-GAA TTA ATT CGC GGC CGC TTA GTA ACA CCT ACA CCT GT-3´.The forward and reverse primers contain Xho I and Not I recognition sites (underlined), respectively, to facilitate directional cloning into the pPIC9 vector (Invitrogen).The PCR product was digested with Xho I and Not I and subcloned into the same sites of the pPIC9 plasmid DNA behind the signal cleavage site with

Table 1 .
The synergistic effect between rOfLPMO1 and rOfChi-h.
*Activity was calculated at 2 h after incubation.