Gene mining of metagenomes for novel enzymes and therapeutics

The expression in E.coli cannot be guaranteed in all cases and therefore an alternative could be the use of Rhizobium leguminosarum as a host, which has a good secretion system. Two metagenomic libraries, an activated sludge library J251(GL11, around 70,000 colonies) and an anaerobic digester library J251 (GL7-10, around 100,000) were provided by Partner 6 (Marge Wexler). Both non-cult DNA were partially digested with Sau3A and cloned into BamHI digested pLAFR3, a cosmid vector (Staskawicz B, Dahlbeck D, Keen N, Napoli C.(1987) J Bacteriol. 169(12):5789-94). 15.000 cfu of each library were screened on TY-AZCl-substrates for the presence of amylase, galactomannanase, xyloglucanase, and cellulose. A lipase assay by diffusion plate using mineral oil was also included. The agar plates were incubated at 30 °C for two days or more. No positive hits were detected so far.

Micro-Mar (http://egg.umh.es/micromar/) is a novel database storing mainly publicly available marine prokaryotes sequences along with their geographical and ecological information. Each entry represents an individual marine prokaryote with one or more DNA sequences coming from a particular sampling location and depth. It aims to become a virtual and interactive book of marine microbiology. It would be useful for the growing community of marine microbiologists and help in integrating all the information that is often published without being properly incorporated into the mainstream of marine microbiology. Micro-Mar currently has 7743 entries consisting of 11430 sequences including 5648 16S rDNA, 1778 ITS, 170 23S rDNA, 3448 CDS and one 5S rDNA. Micro-Mar sequences cover 188 different sampling sites widespread on the world oceanic map representing almost all the oceans. Inshore and offshore representatives are also present demonstrating a wide range of depth, going from surface water (0.5m) to the deepest of the Mariana Trench (10,898m). The entries fall in two super-kingdoms of Archea (959) and Bacteria (5907). There is also a group of 877 unclassified entries. Micro-Mar is updated every week to include the most recent marine prokaryotic sequences.

Microbial proteases are among the most important hydrolytic enzymes. Proteolytic enzymes are ubiquitous in occurrence, being found in all living organisms, and are essential for cell growth and differentiation. The extra-cellular proteases are of commercial value and find multiple applications in various industrial sectors such as detergent industry. A very interesting result with respect to an application was the detection of novel proteases with unusual domain structures during functional screening of metagenomic libraries. The screening was based on the ability of positive protease-producing clones to hydrolyse skim milk. The sequence analysis of the sequences from the recovered metagenomic plasmids led to the identification of two protease-encoding genes. These were designated proA and proB. The deduced proA and proB gene products revealed some similarity to family M4 metalloproteases. Both proteases exhibit unusual domain structures. The proteins harboured a signal sequence, N-terminal domain, catalytic domain, and alpha-helical domain, two C-terminal propeptides, and a regulatory P-domain. The latter domain is unknown to be present in proteases and in bacteria. Usually, this domain is only found in secreted proteins from eukaryotes. The function of this domain is unknown. Analysis of the produced proteases revealed that both are secreted in the culture supernatant and processed. Experiments with derivatives of both proteases showed that the presence of signal sequence, N-terminal domain, catalytic domain, and alpha-helical domain is required for the production and secretion of active proteases. The proteases showed high activity and stability at alkaline pH values and at high temperatures. Thus, the screening of the metagenomic libraries resulted in the identification of proteases with a novel domain structure and excellent activity and stability at alkaline pH values and high temperatures. Therefore, both enzymes have a high potential for an application.

Functional screening assay by using -D-methylumbelliferone-substrates in microtitre plates was performed as described by Cottrell et al. (AEM, 65:2553-2557 (1999)) where an excised copy of the library was used for screening. The assay was performed in microtitre plates with phagemids adsorbed to XLOLR (E.coli) at a multiplicity of infection of 2 x 10-5 (150 infected cells). The screening assay was performed with MUF-beta-D-cellobioside. Screening 100.000 cfu from a sandy soil metagenomic library (SBL0588), 6 putative false positive hits were obtained last year. The hits were analysed again and one hit showed 25% identity with a secreted endoxylanase (from protein database). The functional assay by using fluorescence has the disadvantage that it detects false positive hits due probably to redox reactions.

It was noted that one of the metagenomic clones caused E. coli to become pigmented, the colonies and the surrounding medium becoming reddish brown after three day's incubation. We do not know if this compound has any biological activity, but initial spectroscopic analyses by Partner 3 indicate that, chemically, it is related to melanin. This was confirmed by cloning and sequencing the genes that were responsible for this phenotype. Their deduced products had very high similarity to known tyrosinases, which, in other bacteria, are responsible for the synthesis of melanin. We also constructed a fosmid library of 400,000 clones using DNA from an activated sludge laboratory-scale reactor. From this library we identified a gene that encodes a phenol hydroxylase, which conferred the ability to make the blue pigment indigo to some bacterial strains. This easily observable trait provided a useful generalised marker for the identification of a wide range of oxidases that can act on indole to make the blue pigment. In this particular case, we believe that the "natural; "substrate of this metagenomic phenol oxidase is the phytohormone indole acetic acid, and, indeed, strains of E. coli containing this gene were able to partially catabolise IAA.

DeLisa et al. (2004) described that pspA overexpression resulted in a 3-fold increase of Tat-dependent secretion in E. coli. In S. lividans, a pspA homologue was found, but no homologues of pspB and pspC genes, thought to be needed for a functional phage shock-protein (psp) response (Darwin et al., 2005). To study the impact of the S. lividans pspA gene on secretion in general and on Tat-dependent protein translocation in particular, we isolated the gene from the chromosome by means of PCR amplification and placed it under control of the vsi promoter. Next, the gene was cloned in pIJ486 and introduced in S. lividans TK24. When grown in NM medium, the wild-type S. lividans and the pspA-over-expressing strain showed similar growth curves. Subsequent measurement of the protein content of the spent medium revealed a significant increase of 32% in the total amount of secreted proteins, an observation confirmed by visualization of supernatant proteins of both cultures separated on a SDS-PAA gel by Coomassie staining. To assess the specific effect of pspA over-expression on Tat translocation in S. lividans, Tat-directed XylC and EGFP production were compared under wild-type and pspA-over-expressing conditions. The xylanase activity in the culture supernatant of S. lividans [pSSV-VXylC] amounted 4.8 U/ mg dry weight, while the xylanase activity increased up to 18.5 U/mg dry weight in case pspA was concomitantly over-expressed. This means an almost 4-fold increase in xylanase activity. A similar effect was observed with co-expression of pspA in the egfp-recombinant strains. The EGFP yield increased from 6.9 to 20 mg EGFP/l. On the other hand, the effect of pspA over-expression on Sec-routed protein secretion was studied by comparing the activity of the efficiently, Sec-dependently secreted Sti-1 in the wild type versus the pspA-over-expressing strain. A significant increase of 20% in Sti-1 activity was observed when pspA was over-expressed. Also the amount of Sec-dependently translocated hTNFa encoded on a pSSV05 derivative could be slightly enhanced through concomitant expression of pspA present on the pIJ486 plasmid, but this increase was not statistically significant. Thus, pspA might have a crucial role in strain optimisation aiming to increase recombinant protein yield in the S. lividans host.

Novozymes has obtained a plasmid containing a family M4 protease (thermolysin like) from GEMINI partner 7 (R. Daniel, GEORG-AUGUST-UNIVERSITY GOETTINGEN). The gene was identified by them via screening of a activated sludge metagenomic library on casein plates. The enzyme was difficult to express in E. coli and hence our task was to see if it could be expressed in B. subtilis. The protease was cloned into B. subtilis without the Cterminal domains since Daniel showed previously, that the domains were not necessary for enzyme activity and do not have orthologs in other thermolysin sequences. No elevated protease activity was detected in Bacillus expression clone and no protein band was visible on SDS-PAGE. The enzyme might not be expressed in Bacillus, which may not be surprising since the gene had a high G+C mol% of 67%. The work on this protein was stopped.

A 16S rRNA gene-containing genomic fragment of a Group II mesopelagic euryarchaeota was completely sequenced. The clone was retrieved from a genomic library generated from 500m deep picoplankton of the Antarctic Polar Front. Very close relatives were detected at 3000 m depth in this location and in other mesopelagic regions, so that this euryarchaeote appears a bona fide representative of deep-sea planktonic euryarchaeota. Our analysis revealed a number of important genes in the immediate surroundings of the 16S rRNA gene, which is separated of other rRNA genes in the genome, as is the case for Thermoplasma spp. These included a canonical ribosomal protein spc operon, and genes involved in the maintenance of protein and RNA conformation (Hsp60 and one RNA-helicase), but also genes that are most likely essential for the energy metabolism of this archaeon. We identified a putative operon composed of two genes that encode membrane proteins probably involved in succinate oxidation and/or fumarate reduction. One of these proteins has a unique arrangement of motifs in the same polypeptide (ferredoxin, flavodoxin and succinate dehydrogenase/fumarate reductase Fe-S subunit). The finding of these genes may constitute a clue to identify one of the metabolic abilities of Group II euryarchaeota.

Method for the direct isolation of environmental DNA (metagenome DNA) from various habitats has been developed. Subsequently, metagenomic DNA libraries have been constructed and several screening procedures have been developed. The libraries are rich sources for the isolation of novel enzymes and genes for biotechnological applications. The screening procedures are all directed to the isolation of novel enzymes and metabolite genes. This technology has led to the isolation of a significant number of novel biocatalysts with unusual and novel properties. The list of the identified novel biocatalyst genes comprises esterases, lipases, proteases, alcohol dehydrogenases, pectinases, cellulases, chitinases, amylases and other enzymes with relevance to industry.

The assay is based on the increasing number of reducing ends when carbohydrases degrade their specific substrates (Voragen et al). The reducing ends are detected by using bicinchoninic acid (BCA) on the screening assay: the first step is a Biuret reaction, which reduces Cu+2 to Cu+1 performed by sugar reducing ends. In the second step BCA forms a complex with Cu+1, which it purple coloured and is detectable at 550nm. The detection limit ranges between 0.2-50µg. High background absorbances to be considered: The medium in which the clones are grown. Media commonly used for the growth of fungi or bacteria usually contain interfering substances such as yeast extracts, tryptone, peptone or glucose. The enzymes themselves: BCA can detect concentrations of proteins as low as 2,5ug/ml. The high protein background can easily be avoided by diluting the samples ten times before measuring. One can also correct for the background by measuring t=0 blanks. The substrate: this problem can be avoided by using a final substrate concentration of 1mg/ml. The tenfold dilution necessary to avoid a high protein background was enough to ensure that the substrate background was not too high. So far the screening assay has been tested on positive controls, two wild alpha-amylase bacillae and one alpha-amylase recombinant clone in E.coli, including an E.coli strain (DH10B) alone. Screening assay of a Bacillus sp. genomic library (10.000 cfu) is in progress.

The overall background to this study was to identify genes that were present in the huge, but little studied populations of bacteria in Waste Water Treatment Plants (WWTPs). The functions and effective performance of these sites is dependent on bacterial action and we know only a very little of the processes involved. Furthermore, because of the unusual sets of substrates available to the bacteria in WWTPs, there is great potential for the identification of genes of biotechnological importance. To a large extent our ignorance of the details of WWTP microbiology is because the great majority of these bacteria cannot be cultured. Therefore, our approach taken was to construct metagenomic libraries directly from these bacteria and to screen the libraries for functions of interest. The unusual aspect of this work is that, in contrast to other studies on metagenomics, we used a wide host-range cloning vector so that the functions of the cloned genes could be screened in many different bacterial taxa and not, as is usually the case, in the model organism, Escherichia coli. Metagenomic libraries were constructed in the broad host range cosmid pLAFR3 using DNA isolated from different stages of a waste- water treatment plant. Approximately 200,000 cosmid clones have been isolated, each with an average insert of ~25kb of DNA. Cosmids that conferred interesting phenotypes to other bacteria were obtained. These included: (a) resistance to various antibacterial agents including the aminoglycoside antibiotic kanamycin and to mercury. In these cases, the genes resembled those that had been found in other bacteria that grow in free-living culture. (b) a novel gene that specified an alcohol dehydrogenase (see Result Summary below) (c) several genes that conferred the ability to make tryptophan to various tryptophan auxotrophic mutants of some species of bacteria Although not of biotechnological interest in their own right, these "trp" genes were excellent markers for the quality control of the libraries. More detailed analyses of several of these clones showed that some species of bacteria were more effective in being able to express metagenomic trp genes than others and illustrated the value of using more than one bacterial linage to screen any metagenomic library (d) genes that conferred to E. coli the ability to make an unusual pigment (see Result Summary below). In addition, metagenomic libraries were constructed in a narrow host-range fosmid, pCC1FOS, to obtain gene libraries with larger insert sizes, using DNA from a laboratory scale sequencing batch reactor inoculated with activated sludge. Approximately 400,000 fosmid clones have been isolated, each with an average insert of ~40 kb of DNA. This included: (e) a fosmid that enabled E. coli to synthesise indigo and degrade the plant auxin indole-3-acetic acid.

The metagenomically identified csx007/008 genes encoding an intracellular alpha-amylase and an intracellular 1,4-alpha-glucan branching enzyme, respectively, were amplified from the plasmids pRS241 and pRS245 (Georg-August University, Gottingen). Both genes were fused in frame with the vsi signal sequence and placed under control of the vsi promoter. The expression/secretion cassettes vsi-csx007 and vsi-csx008 were next transferred to pIJ486 and pSSV05. In case of vsi-csx007, recombinant S. lividans [pSSV-vsicsx007] strains could be obtained. By plating these clones on starch-containing media and subsequently flooding the plates with lugol, the production of Csx007 was shown. However, it was not possible to visualize the Csx007 protein via Coomassie staining upon electrophoretic separation of spent medium proteins, implicating a lower yield. In order to optimise enzyme production of this strain, recombinant protein yield will be further evaluated under fermentation conditions. With respect to vsi-csx008, no positive Streptomyces recombinants could initially be selected using the pIJ486 or pSSV cloning vectors. Nevertheless, applying pIJ2587 as cloning vector, recombinant clones with starch-degrading activity could be isolated. By means of the DNS assay up to 700 U of starch-degrading enzyme per mg dry weight was measured in spent medium of S. lividans [pIJ2587-vsicsx008]. Again, Csx008 was secreted in rather low amounts, as the protein could not be visualized in Coomassie-stained SDS-PAA gels on which supernatant proteins were electrophoretically separated. To simplify further purification of the Csx008 from the spent medium, a new pIJ2587 construct was developed carrying the vsi-csx008-6His cassette in which a 6His tag was included at the 3- end of the fusion gene. The large-scale fermentation of the relevant S. lividans clones and subsequent purification of the 6His-tagged Csx008 enzyme are planned.

The ability to screen metagenomic libraries containing large inserts for potentially valuable enzymes and other microbial gene products has been examined using an available BAC (bacterial artificial chromosome) library of DNA inserts from the rumen contents of cattle. Average insert sizes are estimated at around 250kb based on analysis by pulsed field electrophoresis, and the library consists of around 80,000 clones (equivalent to approximately 4000 bacterial genomes). 50,000 clones from this library, provided by Dr. Ron Teather (AgCanada), have now been arrayed into microtitre plates using robotic equipment at the Rowett Institute (RRI). Functional screening at RRI of a subset of 19,200 clones has so far yielded: 17 clones expressing acetyl esterase activity (0.088 %); 10 clones expressing carboxymethyl cellulase activity (0.052 %); >100 clones expressing xylanase (using Remazol brilliant blue xylan) (approximately 4.4%). Secondary screening of positive clones using fluorogenic substrates (4-methylumbelliferyl-?-D-cellobioside, 4-methylumbelliferyl, L-arabinofurunoside, 4-methylumbelliferyl; and ? mannoside) has revealed some clones carrying multiple enzymatic activities. Screening with additional polysaccharides (mannan, lichenan, crystalline cellulose and acid swollen cellulose) is in progress. The main aim is to obtain novel enzymes, or assemblages of co-operative enzymes encoded by gene clusters, that have biotechnological potential. In addition the library is also being screened for certain other traits including resistance and antimicrobial activity; clones resistant to the antibiotic tetracycline were very rare (2 in 19,200 clones).

Chitin, a homopolymer of ?-1,4-linked N-acetyl-D-glucosamine residues, is the most abundant renewable resource after cellulose. It is widely distributed in nature as a structural component of crustacea, fungi, protozoa, and insects. Chitinases (EC 3.2.1.14) are glycosyl hydrolases, which catalyse the degradation of chitin. These enzymes are present in a wide range of organisms such as bacteria, fungi, insects, plants, and animals. Some chitinases of chitinolytic bacteria are potential agents for the biological control of plant diseases caused by various phytopathogenic fungi. These enzymes inhibit fungal growth by hydrolysing the chitin present in the fungal cell wall. Antifungal proteins such as chitinases are of great biotechnological interest because of their potential use as food and seed preservative agents and for engineering plants for resistance to phytopathogenic fungi. In order to identify metagenomic clones and isolates producing chitinases with antifungal properties a two-step procedure has been employed. The first step was the identification of metagenomic clones and isolates exhibiting chitinase activity on chitin-containing indicator agar. The thereby identified chitinase-producing clones and isolates were then in the second step tested for their antifungal activity against phytopathogenic fungi using a hyphal extension-inhibition assay. The screening of metagenomic libraries resulted in 6 chitinolytic clones, which als conferstrong antifungal activities. The genes encoding the chitinases from the metagenomic clones were sequenced and analysed. Chitinases known so far belong to either family 18 or family 19 of glycosyl hydrolase. The enzymes of the two families do not share similarities of the amino acid sequences or the three-dimensional structures. In addition, members of family 19 chitinases are known for strong antifungal activity. The encountered sequences of the chitinases from the metagenomic clones s fall into three categories: i. The chitinase from the one clone showed strong similarity to enzymes belonging to family 18. In accordance with this finding, the enzymee show only weak antifungal activity. ii. Sequence analysis of the chitinase genes from three- metagenomic clones revealed significant homology to chitinases that are a member of family 19 of the glycosyl hydrolases. iii. The chitinase genes from 2 metagenomic clones showed no significant similarities to chitinases, chitin-binding, and catalytic domains of bacterial chitinases belonging to family 18 or family 19 of the glycosyl hydrolases. In addition, no significant similarities to any other characterized proteins of the deduced gene products were found. Thus, the chitinases from the 2 metagenomic clones, which revealed the strongest antifungal activities of all enzymes, represent probably an entire novel class of chitinases. Therefore, these enzymes have a high potential for industrial applications.

PCR cloning approach is currently being employed to target conserved domains of protein families such as Glycosyl hydrolases family 44 (GH44). After alignment of in- and ex-house GH44, the Consensus degenerate-hybrid oligonucleotide primer program or Codehop (http://bioinformatics.weizmann.ac.il/blocks/codehop.html) was used for optimal primer design. As template a sheep rumen gut DNA provided by Partner 2 (Marco Rincon) was used for PCR screening. 10 independently PCR clones showed 53% identity (amino acid level) to a endoglucanase from Ruminococcus flavefaciens (SWALL:Q52743) and 57% to a cellulase (fragment) from an uncultured bacterium (SWALL:Q6IZE3). The N- and C-terminal part of the PCR fragment (1200 bp) was isolated by using the DNA SpeedUp Walking kit (Seegene). It was possible to clone a 1910 bp endoglucanase gene fragment, where 100 bp at the N-terminal part is still missing. Nevertheless, a Bacillus subtilis fusion clone is under construction using an optimised signal peptide including two amino acids for optimal cleavage.

The aim of this work is to analyse the biodiversity and thereby quality of metagenomic libraries before the set-up of functional- and PCR-based screening assays. For this purpose, functional complementation of an E. coli purine synthesis mutant (delta purD) was set up. Genes encoding the purine biosynthetic enzymes have been identified from many sources ranging from bacteria to eukaryotes. A multiple alignment of purD genes from different sources shows that organisms belonging to the same species are clustered (not shown), and hence, this gene can be used for phylogenetic tree construction and phylogenetic analysis. Six environmental libraries were transformed in the E. coli mutant, plated on minimal media and colonies were sequenced with library vector primers. The diversity of the thereby identified purD genes should have given information about the total gene diversity in the metagenome. However, we observed difficulties during the project with the identification of the diverse purD genes from E. coli recombinants. Therefore it was assumed that the amount of obtained recombinants from each library does reflect the metagenome diversity in that library.

Primers and protocols for the amplification of different classes of P450s from actinomycetes have been developed. PCR products have been sequenced and analysed. Genome cosmid libraries have been constructed for selected actinomycetes and the full-length genes have been sequenced and subcloned into expression vectors. Subcloned genes will be used to manipulate the biosynthesis pathways for natural products in order to generate novel derivatives. The genome libraries are rich sources for the isolation of novel genes and gene clusters coding for natural product biosynthesis pathways. Identified genes and cosmids have been used in a combinatorial biosynthesis approach.

Metagenomic libraries from human gut microbiota have the potential to benefit research on human gut health by providing information on uncultured organisms, as well as yielding potentially exploitable genes. A successful method for preparing DNA of moderate size (25-50kb) from human faecal samples was based on the use of a liquid nitrogen chilled freezer mill, followed by CsCl density gradient ultra-centrifugation, dialysis, and washing of the resulting DNA while embedded in agarose. The resulting DNA can be cleaved by restriction enzymes and ligated into standard vectors. DNA extracted from human faecal material was used at RRI to construct libraries in the bacteriophage vector lambda ZAPII. So far the best library has a titre greater of 1.2 x 106pfu/ml, without amplification, with an estimated insert size between 5 and 10kb. A pool-based screening approach demonstrated the ability to efficiently detect inserts expressing the enzymes -galactosidase and glucosidase, using the appropriate fluorogenic substrates. Screening with further chromogenic and fluorogenic substrates can now be used, among other things, to identify clones carrying activities against sugars found on host receptors and against various diet-derived substrates.

A cosmid that enables R. leguminosarum to utilise ethanol as a sole carbon source was identified. A gene associated with this phenotype is a novel alcohol/aldehyde dehydrogenase. Although it is similar in sequence to the AdhE alcohol/aldehyde dehydrogenase of Escherichia coli and has an even closer match to a related enzyme in Clostridium it is unusual in that, instead of being involved in the biosynthesis of ethanol it appears to be catabolic in its activity. We are currently investigating whether this unusual enzyme is of commercial benefit. Our results demonstrate that the screening of metagenomic libraries in different bacterial hosts results in the identification of novel genes and gene products that may otherwise be overlooked if only E. coli is used as a host.

Discovery of biological activities in the metagenome: functional and PCR-based screening. Main focus during the first year was the construction of metagenomic libraries from different environmental samples. The in situ DNA extraction method used was based on the mechanic disruption of cells with the use of silica and glass beads. The average DNA amount reached per 500 mg soil sample was approximately 10 - 20ug, which was directly used for partial digestion, size fractionation (3 - 10 kb DNA inserts), and construction of metagenomic libraries in lambdaZAP vector arms. The phage library was converted into a plasmid library in the presence of two E.coli strains and a M13 helper phage. The titre of the metagenomic libraries varied between 1,000,000 and 50,000 cfu (colony forming units). A quality check of the libraries is under discussion and necessary in order to assure coverage of a high biodiversity. The libraries were then screened by using functional and PCR-based assays. Approximately 100.000 cfu of each metagenomic library was screened on azurine-coupled (AZCl)-substrates contained in agar plates. This screening method has not been successful, at least proven not to be sensitive enough, so this assay was dropped. Functional screening assay by using -D-methylumbelliferone-substrates in microtitre plates was performed as described by Cottrell et al. (AEM, 65:2553-2557 (1999)) where an excised copy of the library was used for screening. The assay was performed in microtitre plates with phagemids adsorbed to XLOLR (E.coli) at a multiplicity of infection of 2 x 10-5 (150 infected cells). The screening assay was performed with MUF-D-cellobioside. Screening 100.000 cfu from a sandy soil metagenomic library (SBL0588), 6 putative false positive hits were obtained last year. The hits were analysed again and one hit showed 25% identity with a secreted endoxylanase (from protein database). The functional assay by using fluorescence has the disadvantage that it detects false positive hits due probably to redox reactions. The expression in E.coli cannot be guaranteed in all cases and therefore an alternative could be the use of Rhizobium leguminosarum as a host, which has a good secretion system. Two metagenomic libraries, an activated sludge library J251(GL11, around 70,000 colonies) and an anaerobic digester library J251 (GL7-10, around 100,000) were provided by Partner 6 (Marge Wexler). Both non-cult DNA were partially digested with Sau3A and cloned into BamHI digested pLAFR3 (Staskawicz B, Dahlbeck D, Keen N, Napoli C.(1987) J Bacteriol. 169(12):5789-94). 15.000 cfu of each library were screened on TY-AZCl-amylose, -galactomannan, -xyloglucan and -cellulose agar plates at 30 °C for two days. No positive hits were detected so far. Implementation of a carbohydrase-screening assay on microtitre plates (Voragen et al). The assay is based on the increasing number of reducing ends when carbohydrases degrade their substrates. The reducing ends are detected by using bicinchoninic acid (BCA) on the screening assay: the first step is a Biuret reaction, which reduces Cu+2 to Cu+1 performed by sugar reducing ends. In the second step BCA forms a complex with Cu+1, which it purple coloured and is detectable at 550 nm. The detection limit ranges between 0.2-50µg. High background absorbances to be considered: The medium in which the clones are grown. Media commonly used for the growth of fungi or bacteria usually contain interfering substances such as yeast extracts, tryptone, peptone or glucose. The enzymes themselves: BCA can detect concentrations of proteins as low as 2,5ug/ml. The high protein background can easily be avoided by diluting the samples ten times before measuring. One can also correct for the background by measuring t=0 blanks. The substrate: this problem can be avoided by using a final substrate concentration of 1mg/ml. The tenfold dilution necessary to avoid a high protein background was enough to ensure that the substrate background was not too high. So far the screening assay has been tested on positive controls, two wild alpha-amylase bacillae and one alpha-amylase recombinant clone in E.coli, and on a Bacillus sp. genomic library. WP 2.2 PCR-based screening Employed to target conserved domains of protein families such as Glycosyl hydrolases (GH) family 44.

One of the major tasks of Novozymes in the GEMINI project was to identify novel cellolytic enzymes from metagenomic source. One approach to get hold of the immense diversity is by functional screening genomic libraries on indicator substrates. The initial screening of the metagenomic libraries was made by using the following indicator substrates: Azurine-crosslinked polysaccharides (AZCL, Megazyme) and 4-methylumbelliferyl-beta-D-cellobioside (MUF-cellobioside, Sigma). Another screening assay was performed by using MUF-cellobioside as a substrate in microtitre plates. The functional screening host was E. coli. In total, sixteen metagenomic libraries were screened (approx. 100.000 clones each). Several novel genes were identified by functional screening of the metagenomic E. coli libraries. Novozymes focussed on enzymes, where a potential industrial value was assumed. We have identified a cellobiosidase (Glycosyl Hydrolase family 6, CBH II), a xylanase (GH10), a heat stable amylase (GH13) by screening soil metagenomes. The sequences of three xyloglucanases (GH44) that were found by a PCR screening technique remained partial. Other GEMINI partners have identified proteases (M4), glucanases (GH5) and mannanases (GH26) by functional screening of sheep rumen and activated sludge metagenomes. However, we were disappointed by the very low hit rate of positive clones from screening techniques on metagenomes. During the project many false positive hits had been analysed and many metagenomic libraries had been screened with no positive result at all. We have screened libraries with relatively low titer and small insert sizes, which may explain the low hit rate. Future high-throughput screening techniques like single cell microcapsules (droplets) combined with cytometry may overcome the vast amount of clones to be screened in metagenomic screening projects. In the GEMINI project we had to limit ourselves to smaller screening programmes, but nevertheless successfully identified several novel enzymes.

In first instance, the use of the newly developed pSSV05 in order to obtain efficient heterologous protein production by the streptomycete host, was evaluated by inserting existing expression/secretion cassettes. These cassettes consist of the promoter and signal sequence of the S. venezuelae subtilisin inhibitor gene (vsi) fused in-frame to cDNA of human or mouse tumour necrosis factor (TNF). The Vsi protein is secreted via the Sec pathway. Upon conjugative transfer of the recombinant plasmids from E. coli to S. lividans and culturing the selected exconjugants, the TNF yield was determined. The results indicated that the same high-level TNF production could be obtained in exconjugants containing the pSSV05 derivatives as in transformants carrying the pIJ486 derivatives with the respective expression/secretion cassettes. This proved the suitability of the pSSV05 conjugative vector. In a next step, the production of Jonesia xyloglucanase (ca. 100 kDa) of which the gene was selected from a metagenomic library, was tested in S. lividans since production in E. coli and Bacillus failed. To this purpose, either the vsi promoter or the vsi promoter followed by the vsi signal sequence was cloned in the MCS of pSSV05. Subsequently, the xyloglucanase gene was cloned downstream the vsi promoter or the mature part of the gene was fused in-frame with the vsi signal sequence which was preceded by the vsi promoter. The S. lividans exconjugants carrying one of both constructs, were evaluated for xyloglucanase production. In both conditions, the xyloglucanase enzyme was detected in the extracellular medium via enzymatic activity tests. The positive results obtained with the production of Jonesia xyloglucanase in S. lividans, proves that Streptomyces is indeed a valuable alternative host organism for the validation of metagenomically isolated genes.

For phylogenetic and metagenomic analyses samples from different environments including extreme habitats were collected. The samples were derived from hot springs, geysers and volcanic areas. The temperatures ranged from 58°C to 78°C and the pH values ranged from 2.5 to 3.5 and from 9 to 9.3. The isolation of high-quality DNA that is suitable for cloning and that covers the microbial diversity present in the original sample from thermophilic and other extreme environments is still a technological challenge. Reasons for this include the reluctance of many microorganisms present in these samples to lysis by the protocols that have been mainly developed for DNA extraction from mesophilic samples and the release of very stable nucleases upon cell lysis. These technological challenges have been addressed and methods allowing the isolation of high-quality DNA and complex metagenomic libraries from these samples have been developed. For the thermophilic samples, total DNA was extracted by using a modified direct lysis method of Henne et al. (1999). To facilitate cell lysis of the extremophilic organisms' presentsamples were suspended in DNA extraction buffer and treated in a bead-beater. After incubation at 60°C for 2 h and centrifugation of the supernatant DNA was extracted with chloroform/isoamylalcohol. The DNA was precipitated with isopropanol. The precipitated DNA was dissolved in water. The isolated amount of DNA was sufficient to construct complex large-insert and small-insert metagenomic libaries. In this way, several libraries harbouring more than 7 Gbp of environmental DNA derived from these habitats have been generated. These libraries are an excellent source for the discovery of novel thermostable biocatalysts.

Two plasmid encoding putative isoamylase genes were obtained from partner Combinature. The respective genes were cloned in B. subtilis and over-expressed at low levels. Characterization showed, that the proteins showed no isoamylase activity and hence the work on these enzymes was stopped.

A conjugative E. coli/Streptomyces vector was developed using the ori/rep functions of E. coli plasmid pUC18 and Streptomyces plasmid pSVH1, and the oriT locus of the broad host range conjugative vector RK2. Additionally, for selection reasons the apramycin resistance gene functional both in E. coli and Streptomyces, was inserted in the vector. By minimising the vector size and inserting a 72-bp multiple cloning site, the vector was further optimised. It was proven that the resulting 5.4-kb pSSV05 maintained the capacity to transfer DNA from E. coli to Streptomyces via conjugation. Importantly, pSSV05 could be stably replicated in Streptomyces lividans overcoming the problem of the observed instability of pIJ101-based conjugative vectors. The presence of the MCS allows the easy insertion and/or replacement of several genetic elements such as promoters, signal sequences, heterologous genes of interest and sequences encoding a tag. Consequently, pSSV05 forms the base of an enormous set of versatile expression/secretion vectors to be used in Streptomyces. This newly developed conjugative vector allows all DNA manipulations to be done in E. coli, which means a gain in time. By a simple conjugation protocol, the expression/ secretion vectors can be brought in the Gram-positive Streptomyces to finally analyse the production of the protein of interest in this host organism.

The isolation of high-quality DNA that is suitable for cloning and that covers the microbial diversity present in the original sample from "cold" environments such as snow and glacier ice is a technological challenge. Reasons for this include the low biomass in the samples. This technological challenge has been addressed. Snow samples were collected from a pristine meadow and ditch in Gottingen, Germany, and glacier ice was collected from the Schneeferner glacier at the Zugspitze (47° 25N, 10° 59O), Germany, to a depth of 0.5m. Subsequently, glacier ice and snow were melted at 4°C, and filtered through a 0.2µm filter. Filter membranes were stored until further processing at 70°C. Total DNA from snow and glacier samples was extracted directly from filters using the "Nucleo Spin Tissue kit" (Macherey-Nagel, Duren, Germany) according to the manufacturer's directions with the following modifications: for the initial pre-lysis step 2 filters were cut in small pieces and incubated overnight at 56°C with 2.5ml pre-lysis buffer T1 and 1.8mg/ml proteinase K. Due to the low biomass in the samples from and the resulting low amounts of purified metagenomic DNA the isolated DNA was amplified by multiple displacement amplification. In this way, a sufficient amount of DNA for the construction of complex large-insert and small-insert metagenomic libraries from glacier ice and snow has been recovered. These libraries harbour approximately 3 Gbp of environmental DNA from these habitats. These libraries can serve as starting materials to explore and to exploit the genetic content of these unique environments.

Our main focus during the first year of the project was the construction of metagenomic libraries from different environmental samples. The in situ DNA extraction method used was based on the mechanic disruption of cells with the use of silica and glass beads. The average DNA amount reached per 500 mg soil sample was approximately 10 - 20 ug, which was directly used for partial digestion, size fractionation (3 - 10kb DNA inserts), and construction of metagenomic libraries in lambdaZAP vector arms (ZAP Express pre-digested vector kit and ZAP Express pre-digested Gigapack cloning kit, Stratagene). The phage library was converted into a plasmid library in the presence of two E.coli strains (XL1-Blue MRF/ and XLOLR strain) and M13 helper phage. Approximately 100.000 pfu (plaque forming units) were mass-excised into plasmid libraries and used for functional and PCR-based screening assays. A quality check of the metagenomic libraries is in progress and necessary in order to assure coverage of a high biodiversity.

It has recently been shown (De Keersmaeker et al., 2006) that TatABC overproduction in S. lividans improves Tat-dependent secretion capacity as concluded from the increased yield of the homologous, exclusive Tat-dependent xylanase C. On the other hand, tatABC over-expression negatively influenced the yield of the homologous Sec-dependent xylanase B and trypsin inhibitor Sti-1. In this project, the effect of TatABC overproduction on the yield of enhanced green fluorescent protein (EGFP) and hTNFa was tested. Therefore, egfp was amplified from plasmid pIJ8668 (Sun et al., 1999), placed under control of the vsi promoter and the xylC signal sequence and finally cloned into pIJ2587 or pIJ2587tatABC containing the 3 S. lividans tat genes (De Keersmaeker et al., 2005). EGFP production in S. lividans transformants containing either pIJ2587VXEgfp or pIJ2587VXEgfpTatABC was evaluated by means of fluorescence measurements in 1 ml spent medium. TatABC overproduction increased the amount of active EGFP in the supernatant, but this increase was not significant. Similarly, the hTNFa cDNA placed under control of the vsi promoter and xylC signal sequence was cloned in pIJ2587 and pIJ2587TatABC resulting in pIJ2587VXhTNFa and pIJ2587VXhTNFaTatABC. S. lividans transformants carrying the respective plasmids were grown and hTNFa ELISA measurements were carried out on supernatant samples. While no hTNFa could be detected in S. lividans [pIJ2587VXhTNFa] cultures, up to 0.64 ± 0.04 ng hTNFa/ml medium was measured in spent medium of S. lividans [pIJ2587VXhTNFaTatABC] cultures. This means that for both heterologous proteins TatABC overproduction improved their Tat-dependent secretion in S. lividans.

The rumen as an organ specialized in the microbial fermentation of complex polysaccharides is still relatively unknown and represents a source to find novel polysaccharidases or associated enzymes with potential use in many biotechnological applications. This study demonstrated the efficacy of phage libraries for functional-based screening in E. coli, taking advantage of the promiscuity in promoter recognition and efficient release of cloned enzyme accompanying cell lysis. Phages allowed cloning of relatively small insert DNA easily recovered at high efficiency by in vivo excision and at the same time permitting activity-based screening in high throughput format without resorting to archiving and organization of the library by robotic clone selection. The genes recovered included two potentially novel hydrolases with very little relationship to known families. In addition, many others represented new branches on the phylogenetic trees of the relevant enzyme families. This may reflect the use of new screening substrates, as in the case of xyloglucan, where a group of family 5 enzymes that show this activity appeared to cluster together. In other cases, it shows that the most abundant enzymes detected from the metagenome (eg. for CMCase) do not correspond to those so far recovered from cultured organisms. The functional properties of these novel enzymes therefore deserve further study. In contrast to the majority of enzymes reported from cellulolytic rumen bacteria, very few of the metagenome sequences coded for multi-domain enzymes, and there was no evidence of substrate-binding modules or dockerin sequences involved in protein: protein interactions. The most likely reasons for this are: 1) that primary cellulolytic bacteria make up a small proportion of the total rumen population; 2) that they are difficult to recover because of tight association with the substrate, and are also relatively difficult to lyse in the case of the Gram-positive ruminococci; 3) that multi-domain enzymes may be less readily expressed as active products in E. coli because of their size, a tendency to proteolysis and possibly toxicity of certain domains (eg. dockerins) for the host.

A method was developed for the heterologous expression/production/purification of a metagenomic xyloglucanase of more than 900aa residues. The gene of the xyloglucanase was transferred to the Gram-positive bacterium Streptomyces lividans and the protein was expressed as a secreted polypeptide in this host. A purification protocol was developed that allowed high level and purity production of the heterologous xyloglucanase. In a highly active form this method has the potential to be applied for other polypeptides of interest for industrial or biopharmaceutical use.

A metagenomic fosmid library was used to investigate genes conferring antibiotic resistance in a faecal sample from a human volunteer who had received long-term tetracycline therapy. A mosaic tet(O/32/O) gene was found to be the most abundant gene present that conferred tetracycline resistance in an E. coli host. Further sequence analysis identified a novel efflux-type resistance gene, that encodes a predicted membrane-associated protein (the closest related protein has 43% amino acid identity), immediately downstream from tet(O/32/O). This arrangement was also found in a Clostridium related TcR bacterial isolate from the same donor, and the two genes were co-transferred into matings with a sensitive recipient. By itself, the new efflux gene, which is designated tet(40), conferred resistance to revealed a MIC of 60 ug tetracycline/ml in E. coli. The discovery of a new gene that confers high levels of resistance in bacteria, that may possibly extend to other antimicrobial agents, is potentially important for researchers and policy makers working on the control of pathogens in human and veterinary medicine. This also applies to the finding that mosaic genes, which appear to confer higher levels of resistance than their "parent" genes, can arise by homologous recombination and become more common than non-mosaics under conditions of intense selection.

Livrables

Partager cette page

Télécharger