African, fungus-growing termites Pseudacanthotermes militaris were procured from a laboratory-based colony that had been maintained for several years in the University of Dijon, France . The colony was initially established in the Republic of Congo in 1992, and was thereafter maintained in Dijon in vivariums made from Altuglass, containing clayish soil and held at 28±1°C, 80% relative humidity and subjected to a daily cycle of 12 h light and 12 h dark. Decayed wood from the Burgundy region and distilled water were supplied regularly.
Construction of metagenomic libraries
Metagenomic libraries were constructed by Libragen S.A (Toulouse, France). Basically, termites were first sorted to essentially isolate the workers, which were then dissected in two stages. First, working under a binocular microscope, the abdominal parts were separated from the thorax and head. Then, the entire digestive tract was recovered and transferred to a microcentrifuge tube containing physiological solution (0.8% w/v NaCl). Digestive tracts were crushed on ice using a micropestle and bacterial cells were separated out by high-speed centrifugation on a Nycodenz density gradient (Nycomed Pharma AS, Oslo, Norway) as described by Courtois et al. . The bacterial cells were then suspended in 10 mM Tris-500 mM EDTA (pH 8.0), incorporated into low melting point agarose and subjected to enzymatic lysis as previously described . High molecular weight DNA were separated using pulsed-field gel electrophoresis according to a previously described method  and was then cloned into the fosmid pCC1FOS and packaged into the lambda phage particles according to the suppliers recommendations for library construction (Epicentre Technologies, USA). After transduction of E. coli EPI100 cells by the recombinant fosmid library and growth at 37°C on solid LB medium containing 12.5 μg/mL chloramphenicol, individual colonies were transferred to 384-well microtiter plates containing freezing medium (Luria-Bertani, 8% glycerol complemented with 12.5 g/mL chloramphenicol), using an automated colony picker (QPixII, Genetix, UK). After 22 h of growth at 37°C without any agitation, the plates were stored at -80°C.
Chromogenic glycosides and polysaccharides
Most chromogenic compounds used for screening were purchased from either Megazyme (Ireland) or, in the case of 5-bromo-4-chloro-3-indolyl-α-l-arabinofuranoside (BCI-Araf), from Carbosynth (Berkshire, UK). However, 5-Bromo-3-indolyl β-d-xylopyranoside (BI-Xylp) was synthesized in-house using a two-step protocol. First, N-acetyl-5-bromo-3-indolyl 2,3,4-tri-O-acetyl-β-d-xylopyranoside (1) was prepared from 1-acetyl-5-bromo-indoxyl-3-ol (0.333 g, 1.31 mmol, 1.05 eq.) , which was dissolved under nitrogen in anhydrous (10 mL) in a two-neck flask equipped with a pressure equalising dropping funnel. The reaction was then cooled to 0°C and boron trifluoride diethyl etherate (77 μL, 0.62 mmol, 0.5 eq.) was added, before slowly (over 5 min using the dropping funnel) transferring dry (dried on activated 4Å molecular sieves) 2,3,4-tri-O-acetyl-d-xylopyranosyl trichloroacetimidate (0.525 g, 1.25 mmol, 1 eq.) , in anhydrous dichloromethane (5 mL), into the reaction mixture. The funnel was rinsed with 5 mL of anhydrous dichloromethane, which were further added to the reaction. After stirring for 2 h at 0°C, the mixture was raised to room temperature and then quenched by adding triethylamine. Dilution with ethyl acetate was followed by washing with saturated aqueous sodium hydrogen carbonate and then brine, before drying over anhydrous sodium sulfate, filtering, concentrating under reduced pressure, and purifying compound 1 by flash chromatography (ethyl acetate/petroleum ether, 8:2 to 3:2, v/v), which was obtained as an amorphous, white solid in 49% yield (0.313 g, 0.61 mmol). At all stages of the preparation process, reactions were monitored by analytical thin-layer chromatography, using silica gel 60 F254 precoated plates (E. Merck).
To obtain BI-Xyl (2), a suspension of 1 (0.206 mg, 0.40 mmol, 1 eq.) in dry methanol (10 mL), was cooled in an ice-water bath and treated with sodium methoxide (1M in methanol, 200 μL, 0.20 mmol, 0.5 eq.) for 2.5 h. The solution was neutralized with Amberlite IRN-120 (H+), filtered, concentrated under reduced pressure, dissolved in deionized water and freeze-dried to yield compound 2 in 93% yield (0.128 mg, 0.37 mmol) as a slightly dark blue foam. Analysis using NMR spectroscopy and HRMS provided full verification of the successful synthesis of both 1 and 2.
For NMR experiments, a Bruker Avance II 500 spectrometer was used. Chemical shifts (δ) are reported in ppm downfield with internal reference of residual solvents . Coupling constants (J) are reported in Hertz (Hz) with singlet (s), doublet (d), doublet of doublet (dd), triplet (t), multiplet (m), broad (br). Analysis and assignments were made using correlated spectroscopy (COSY), J-modulated spin-echo (Jmod) and Heteronuclear Single Quantum Coherence (HSQC) NMR experiments.
High-resolution mass spectra (HRMS) analyses were performed at the CRMPO (Rennes University, France) in positive ionisation mode (ES+) on either a Waters Q-Tof 2.
1H NMR (500 MHz, CDCl3, 298 K): δ 8.29 (1H, br s, H-indolyl), 7.63 (1H, d, J 0.4 and 2.0, H-indolyl), 7.47 (1H, dd, J 2.0 and 8.9, H-indolyl), 7.10 (1H, br s, H-indolyl), 5.24-5.18 (3H, m, H-1, H-2, and H-3), 5.00-4.97 (1H, m, H-4), 4.28 (1H, dd, J 4.0 and 12.5, H-5a), 3.63 (1H, dd, J 6.0 and 12.5, H-5b), 2.56 (3H, s, N-Ac), 2.15, 2.13, 2.11 (9H, 3s, O-Ac); 13C NMR (125 MHz, CDCl3, 298 K): δ 169.8, 169.7, 169.3 (C=O, O-Ac), 168.2 (C=O, N-Ac), 140.0, 132.3 (Cq-indolyl), 129.2 (CH-indolyl), 125.6 (Cq-indolyl), 120.4, 118.2 (CH-indolyl), 117.0 (Cq-indolyl), 109.5 (CH-indolyl), 99.5 (C-1), 69.3, 69.0 (C-2 and C-3), 68.0 (C-4), 61.4 (C-5), 23.8 (CH3, N-Ac), 20.8, 20.8, 20.7 (CH3, O-Ac); HRMS calcd for [M+Na]+ C21H22NO9BrNa+ 534.0376; found 534.0372 (1 ppm).
1H NMR (500 MHz, CD3OD): δ 7.80 (1H, br d, J 1.7, H-indolyl), 7.21-7.15 (2H, m, H-indolyl), 7.03 (1H, s, H-indolyl), 4.62 (1H, d, J 7.5, H-1), 3.95 (1H, dd, J 5.8 and 11.5, H-5a), 3.61-3.56 (1H, m, H-4), 3.47 (1H, dd, J 7.5 and 9.1, H-2), 3.40 (1H, t, J 9.0, H-3), 3.26 (1H, dd, J 10.3 and 11.5, H-5b; 13C NMR (125 MHz, CD3OD): δ 138.1, 133.8 (Cq-indolyl), 125.6 (CH-indolyl), 123.2 (Cq-indolyl), 121.2, 114.1, 114.0 (CH-indolyl), 112.7 (Cq-indolyl), 106.7 (C-1), 77.7 (C-3), 74.9 (C-2), 71.1 (C-4), 67.0 (C-5); HRMS calcd for [M+Na]+ C13H14NO5BrNa+ 365.9953; found 365.9957 (1 ppm).
Primary high-throughput screening of metagenomic libraires
Functional screening of metagenomic libraries was performed using a core facility comprised of a QPixII colony picker (Genetix, UK), a Biomek 2000 liquid handling station (Beckman, USA) and a Genesis RSP-200 configured for enzyme assay miniaturization (TECAN, Switzerland).
The initial screening of 40,000 fosmid clones was performed on 22 × 22 cm Q-tray bioassay plates (2304 clones per plate) containing solid PLA medium and chloramphenicol supplemented with chromogenic substrates: 5-bromo-3-indolyl-β-d-xyloside and 5-bromo-4-chloro-3-indolyl-α-l-arabinofuranoside (60μg/mL each), or AZCL-HE-Cellulose (0.2% w/v), or AZCL-Xylan (0.2% w/v), or AZCL-β-(1,3)-β-(1,4)-Glucan (0.2% w/v) (Megazyme, Ireland). Plates were incubated for up to 2 weeks at 30°C, and the appearance of colony coloration or haloes was monitored on a daily basis.
Secondary screening of library hits in microtiter plates
For secondary screening of metagenomic clones that had been positively identified in the primary screen, pre-cultures were prepared in sterile 96-well microtiter plates containing 200 μL of LB medium and chloramphenicol (12.5 μg/mL) and grown at 30°C for 16 h with shaking (700 rpm). After, 100 μL of pre-culture was transferred to 1 mL of LB medium and chloramphenicol (12.5 μg/mL) contained within wells of deep-well microtiter plates, which were then incubated at 30°C for 16 h with shaking (700 rpm). Bacterial cells were lysed by adding 100 μL of a solution containing 5 g/L lysozyme and 5 mg/L deoxyribonuclease I (Euromedex, France), followed by incubation at 37°C for 1 h with shaking (200 rpm) and then a freeze-thaw cycle (-80° C/37°C). Clarified cell extracts were obtained by transferring the lysates to FiltrEX™ 96 well microtiter plates (Corning, USA) equipped with glass fiber filters (0.25 mm) followed by centrifugation (1,000 × g, 7 min at 10°C). The clarified extracts were then used to perform enzyme assays, using p NP-α-l-arabinofuranoside (p NP-Araf), p NP-β-d-xylopyranoside (p NP-Xylp) or Azo-functionalized arabinoxylans (Megazyme, Ireland) as substrates. To vary the pH conditions the following buffer were employed: 50 mM citrate buffer, pH 4 and 50 mM sodium/potassium phosphate, pH 6 and pH 8 and 50 mM Glycine-NaOH, pH 10. Generally, reactions were performed in wells of thermoresistant polypropylene 96-well microtiter plates containing 40 μL cell extract, 50 μL 0.1 M buffer and 10 μL p NP-Araf or p NP-Xylp (10 mM) and sealed using Easy Pierce film (Thermo Scientific, USA) and an ALPS 50V thermosealer (ABgene). Sealed plates were incubated at different temperatures (30, 40, 50, 60 or 70°C) for 2 h and reactions were stopped by adding 100 μL of sodium carbonate (2.5 M) and placing plates on ice. To measure absorbance (405 nm), reactions mixtures (150 μL) were transferred to 96-well polystyrene microtiter plates (Greiner, Bio-One, Austria and Germany) and analysed using a microtiter plate absorbance reader (Sunrise™, Tecan, Switzerland). Then, the absorbance was converted to mM of released pNP, using the Beer-Lambert formula. For each reaction condition, relative activity was calculated as the ratio of the clone activity in this condition and the clone highest activity during the test. For reactions involving polysaccharides, Azo-xylans from different botanical sources were used: birchwood glucuronoxylan, BGAX (arabinose/xylose or A/X=0.015; uronic acid/xylose or U/X = 0.15), oat spelt xylan, OAX (A/X=0.12; A/U = 0.054) and wheat arabinoxylan, WAX (A/X=0.61; A/U < 0.054). Reactions were performed in sealed deep-well microtiter plates containing 112 μL cell extract, 140 μL buffer (0.1 M) and 28 μL of Azo-linked xylan (4% w/v). After incubation for 2 h at 30, 40, 50, 60 or 70°C, reactions were stopped by adding 700 μL of ethanol (95% v/v) to each well and the precipitated polymers were eliminated by centrifugation (10 min at 1,000 × g). The supernatants (150 μL) were transferred into 96-well polystyrene microtiter plates (Greiner Bio-One, Austria and Germany) and analysed at 590 nm using a microtiter plate absorbance reader (Sunrise™, Tecan, Switzerland).
Fosmid quality control and sequencing
Fosmids were extracted from positively identified library hits using a NucleoBond® DNA miniprep kit (Macherey Nagel, France), following the manufacturer’s instructions for the isolation of low copy number vectors. Prior to sequencing, the quality and the potential redundancy of the extracted fosmids was assessed using restriction fragment length polymorphism (RFLP) analysis. Each fosmid was digested (2 hours at 37°C) using Bam HI and Pst I restriction enzymes (New England Biolabs® Inc.) and then analysed on a 0.8% w/v agarose gel, prepared using Pulsed Field Certified™ agarose (BioRad, France), immersed in TBE buffer (45 mM Tris, 45 mM Borate, 1mM EDTA) and running on a CHEF-DRIII Pulse Field Gel Electrophoresis system (switch time 2 to 6 s, 4.5 V, angle of 120°, for 11 h at 14°C) coupled to a pump and a cooling module (BioRad, France). After migration, gels were stained with ethidium bromide (0.5 μg/mL) and visualized under UV light.
Once the quality of the fosmids was ascertained, sequences were determined using Roche 454 GS FLX Titanium technology, according to the manufacturer’s protocols (Roche Applied Science, Indianapolis). 500 ng of fosmid DNA were used and up to 12 fosmids were assembled in the sequencing mix, using MID adapters to differentiate them. The assembly of sequence reads was achieved using CAP3  and vector sequences were removed from contigs using Crossmatch (http://www.phrap.org/phredphrapconsed.html#block_phrap). Only contigs presenting lengths > 1,000 bp and a sequencing coverage > 8 fold were considered for further analyses. Open reading frames (ORF) were detected using Metagene (http://weizhong-lab.ucsd.edu/metagenomic-analysis/server/metagene/, ) and ORFs and contigs were analysed using both blastx (http://blast.ncbi.nlm.nih.gov/), searching against non-redundant NCBI and Swissprot databases, and by performing another search using the CAZy database (http://www.cazy.org/). Annotated sequences were deposited in the European Nucleotide Archive (http://www.ebi.ac.uk/ena/) as 68 accessions, numbered HF548269 through to HF548336.
For the taxonomic assignment of metagenomic fragments, two methods were used. The first one simply relied on the results obtained from the blast search. Basically, among hits displaying an e-value lower than 10-8 and a percentage of identity higher than 90%, if more than 50% of the ORFs of one contig were assigned to the same species, then the contig was assigned to this species. The second method employed the Megan software (http://ab.inf.uni-tuebingen.de/software/megan/, [21, 22]. For COGs assignment, a RPS-BLAST search was performed using the COG database [46, 47] and results were filtered, selecting only hits displaying e-values > 10-8.
Subcloning, expression and enzyme purification
Five ORFs encoding family GH51 or GH43 enzymes, and one encoding a hybrid protein containing both GH51 and GH43 domains, were cloned into the T7 promoter-based expression vector pET28a (Merck KGaA, Germany). To achieve this, appropriate primers were designed to simultaneously PCR amplify the target sequences and introduce Nhe I and Xho I restriction sites at the 5′ and 3′ extremities of the amplicons respectively (Additional file 1: Table S2). Amplification was achieved using Phusion™ high-fidelity DNA polymerase (Finnzymes) and the appropriate fosmid DNA as the template. After PCR, amplicons were purified using the GenElute™ Extraction Kit (Sigma, France), digested with Nhe I and Xho I and ligated to pET28a DNA. The resultant plasmids were ultimately used to transform to E. coli BL21(DE3) (EMD Millipore, Germany). For protein expression, a standard protocol for T7-driven expression was employed. Briefly, E. coli BL21(DE3) cells bearing one of the recombinant plasmids were cultured in LB broth containing 50 μg/ml kanamycin. Overnight cultures were diluted in fresh medium and grown at 37°C until an OD (600 nm) value of 0.5-0.6 was reached. Isopropyl-β-d-thiogalactopyranoside (IPTG) was added to a final concentration of 0.5 mM, cultures were further grown overnight at 16°C. Cells were harvested by centrifugation (15 min, 6,000 × g, 4°C), resuspended in 20 mM Tris-HCl, 300 mM NaCl, pH 8 and lysed by sonication (over 2 min using 0.5 s pulses). The proteins were purified using immobilized metal ion affinity chromatography (IMAC) and Talon® Metal Affinity Resin (Clontech, USA). Proteins were eluted from the column using Talon buffer containing 100 mM imidazole. Fractions containing the purified protein were pooled and dialysed in 20 mM Tris-HCl pH 7.
Protein concentrations were determined spectrophotometrically, by measuring absorbance at 280 nm and employing theoretical molecular extinction coefficients, determined using the ProtParam Tool (http://web.expasy.org/protparam/). Specific activities of arabinofuranosidases and xylosidases present in cell lysates or obtained in purified recombinant form (e.g. GH43 enzymes from clones A3 and G12 respectively ) were determined by measuring the release of paranitrophenol (p NP) release from p NP-α-l-Araf or p NP-β-d-Xylp. To achieve this, reactions performed in 50 mM phosphate buffer pH 7 (for cell lysates), containing BSA (1 mg/mL) and a p NP-glycoside (5 mM), were incubated at 30°C. Aliquots (100 μL) were removed at regular intervals and added to 500 μL NaCO3. After mixing, the absorbance at 405 nm was recorded using a Cary 100 spectrophotometer (Agilent, USA). The amount of p NP released was quantified using a standard curve and one unit (U) of activity was defined as the amount of enzyme releasing one μmol of p NP per minute. To determine the optimal pH for the activities of GH43 enzymes from clones A3 and G12 respectively, activities were measured in a similar way, using different buffers (citrate, pH 3-6; phosphate, pH 6-8; bicine, pH 8-9; glycine, pH 9-10) at a concentration of 50 mM and working at 40°C. Arabinanase activities were assayed at 30°C in 50 mM phosphate buffer (pH 7), containing BSA 1mg/mL and 10 mg/mL of debranched arabinan or sugar beet arabinan (Megazyme, Ireland), by monitoring the solubilization of reducing sugars. To achieve this, aliquots were removed from the reaction mixture at regular intervals and added to an aliquot of DNS (3,5-dinitrosalicylic acid) reagent. After mixing and incubation in a water bath at 95°C for 20 min, absorbance at 540 nm was recorded using a Cary 100 spectrophotometer (Agilent, USA) and compared to a standard calibration curve prepared in 50 mM phosphate buffer and 10 mg/mL arabinan using l-arabinose. One unit (U) of activity was defined as the amount of enzyme releasing one μmol.mL-1 of free l-arabinose per minute.