WO2006117247A1 - Glycosyl hydrolase having both an alpha-l-arabinofuranosidase and a beta-d-xylosidase activity. - Google Patents

Abstract

The present invention relates to a glycosyl hydrolase having both an alpha-D-xylosidase and a beta-L-arabinofuranosidase activity, and to its uses.

Description

Glycosyl hydrolase having both an α-L-arabinofuranosidase and a β-D-xylosidase activity.

The present invention relates to a new glycosyl hydrolase which has both an α-L-arabinofuranosidase and a β-D-xylosidase activity, and to its uses.

5 Non-cellulosic polysaccharides of plant cell walls encompass pectins and hemicelluloses. After cellulose and lignin, they constitute major components of the plant biomass.

Pectin is the denomination for another group of chemically diverse polysaccharides, which are abundant in primary cell walls. Pectins comprise a backbone

0 containing a high proportion of (1— »4)-α-linked D-galacturonic acid residues that can be interspersed with (l—»2)-α-linked rhamnose residues. Long side chains consisting mainly of L-arabinose and D-galactose residues are often attached to these rhamnose residues.

Hemicellulose is a collective denomination for a group of non-cellulosic polysaccharides that represents the linking material between cellulose and lignin. There is a

5 variety of hemicelluloses, corresponding to different sugar components and different kinds of linkage between these sugars. Most of hemicelluloses are branched heteropolysaccharides, comprising a backbone chain with side chains linked through various bonds. The sugar components the more often found in hemicelluloses include hexoses such as glucose, galactose and mannose, and pentoses such as xylose and arabinose.

!0 Arabinose is one of the most abundant pentoses in plants and it is found in various non-cellulosic polysaccharides. For instance, in hemicellulosic arabinoxylans, the (l~>4)-β-D-xylose backbone is mainly branched with α-L-arabinofuranose residues linked in α(l— >3) position or, occasionally, in both α(l— >3) and α(l— >2) positions (PULS and SCHUSElL, Hemicellulose and hemicellulase, Coughlan and Hazlewood eds, 1-27, 1993); in

!5 pectins, arabinose is found in particular in arabinan side-chains which are composed of a backbone of (1— >5)-α-linked L-arabinofuranosyl residues branched with (1— >3)-α-linked and (l->2)-α-linked side chains of L-arabinose in the furanose conformation (BACIC et ai, The Biochemistry of Plants, Preiss Ed., 14: 297-371, 1988 ; RAHMAN et al, Carbohydr. Res., 338: 1469-1476, 2003).

>0 Plant biomass is used in a wide variety of applications, such as production of food, beverages, or food additives, production of fibres and cellulosic material, production of alcohol fuels and other chemicals. Many of them involve a partial or complete degradation of the hemicellulosic and/or pectic fraction. Enzymatic breakdown is currently a method of choice for performing this degradation. However, due to the heterogeneity and structural complexity of non- 15 cellulosic polysaccharides, their complete breakdown requires a number of enzymes with different specificities. For example, the complete hydrolysis of xylan requires not only endo-xylanase, that cleaves the backbone chains and β-xylosidase that hydrolyzes the resulting xylooligosaccharides to xylose, but also several accessory enzymes for cleaving the side-chains.

Among the various enzymes can hydrolyze either backbone or side chains of arabinose-containing polysaccharides, arabinofuranosidases (EC. 3.2.1.55) are particularly

5 important since they can cleave terminal non-reducing α-L-arabinofuranosyl groups from different arabinose-containing oligosaccharides and polysaccharides (SAHA, Biotechnol. Adv., 18: 403-423, 2000). They cooperate with endo-l ,5-α-L-arabinases (EC 3.2.1.99), which act by carrying out endohydrolysis of ( 1— »5)-α-L-arabinofuranosidic linkages in the arabinan backbone, resulting in their complete degradation.

0 Concerning α-L-arabinofuranosidase activities on natural polysaccharides,

BELDMAN et al. (Adv. Macromol. Carbohydr. Res., 1 : 1 -64, 1997) have classified these enzymes into three types depending on their mode of action and substrate specificity. α-L-arabinofuranosidases of Type-A preferentially degrade the (1— »5)-α-L-arabinofurano- oligosaccharide backbone arabinan. The second type of α-L-arabinofuranosidases, which are

5 called type-B, preferentially degrade L-arabinose residues from the side chains of arabinan or arabinoxylan. The third type of α-L-arabinofuranosidases, which are called type-C are specifically active on arabinosic linkages of arabinoxylans.

The use of L-arabinofuranosidases has been described in a number of applications, such as delignification in the pulp and paper industry, bioconversion of

!0 hemicellulosic biomass to alcohol fuel, improvement of the feed conversion efficiency of plant materials in animal feeds, clarification of juices, hydrolysis of grape monoterpenyl glycosides to increase wine aroma (for review cf. SAHA, 2000, aforementioned).

Generally L-arabinofuranosidases are used in preparations comprising other enzymes involved in the hydrolysis of non-cellulosic polysaccharides. These preparations usually

!5 consist of mixtures of purified enzymes with different substrate specificity. The use of preparations comprising naturally occurring or transformed microorganisms, or transgenic plants, producing the desired enzymes has also been described. For instance, US Patent 6,699,515 describes a process for the production of beer or whiskey, wherein a mixture of enzymes comprising an endo-β— (l,4)-xylanase, an arabinofuranosidase, an alpha-amylase, an endo-

0 protease and a β-(l,3;l,4)-glucanase is provided by seeds of transgenic plants expressing recombinant fungal enzymes.

Arabinofuranosidases have generally a narrow substrate specificity. Thus, there is a need to have at one's disposal a large variety of arabinofuranosidases differing in their substrate specificity, in order either to associate several of them to obtain a broad spectrum of

'5 activity, or conversely to select the more appropriate one for a selective hydrolysis of a given substrate. Most of the arabinofuranosidases that have been purified and characterized until now are of fungal or bacterial origin. Although numerous plant enzymes are involved in the structural modification and physical properties of cell wall polysaccharides, little information are available about them (FRY, New Phytologist, 161 : 641 -675, 2004).

5 All biochemically characterized plant α-L-arabinofuranosidases have been assigned to glycosyl hydrolase families 3 or 51 , according to the classification proposed by HENRlSSAT (Biochem. J., 280: 309-316, 1991), available on the CAZy (CArbohydrate-Active EnZymes) web site: http://afmb.cnrs-mrs.fr/CAZY/. Until now, two α-L-arabinofuranosidases from family 51 which remove α-L-arabinofuranosyl residues from the arabinoxylan polymer,

0 have been purified, sequenced and characterized from barley (Hordeum vulgare) seedlings. These two barley arabinoxylan hydrolases have been suggested to belong to type C of α-L-arabinofuranosidases (FERRE et al, Eur. J. Biochem., 267: 6633-6641 , 2000 ; LEE et al, Biochem. J., 356: 181 -189, 2001).

One α-L-arabinofuranosidase, from the same family 51 , has been purified from

5 stem tissues of Arabidopsis (MINIC et al, Plant Physiol., 135: 867-878, 2004). This α-L-arabinofuranosidase can hydrolyze various natural substrates such as xylan, arabinoxylan, arabinan and oligoxylan. This enzyme, ARAf, also belongs to type C of α-L-arabinofuranohydrolases because, although it possesses a relatively broad substrate specificity, it prefers arabinoxylan as a substrate.

!0 In addition, it was recently reported that some putative β-D-xylosidases, which belong to glycosyl hydrolase family 3, also possess α-L-arabinofuranosidase activity (LEE et al, J. Biol. Chem., 278: 5377-5387, 2003 ; MINIC et al, 2004, aforementioned). One of these, designated as ARA-I, has been purified, characterized and sequenced from barley seedlings (LEE et al, 2003, aforementioned). ARA-I can degrade (1— »5)-α-L-arabinofuranofexaose and

!5 (1 — >4)-β-D-xylopentaose (LEE et al, 2003, aforementioned). A second, XYLl, was purified and characterized from stem tissues of Arabidopsis (MINIC et al, 2004, aforementioned). XYLl can degrade (1 — »4)-β-D-xylobiose and (1 — >4)-β-D-xylotetraose and various natural substrates such as arabinoxylan, arabinan, oligoxylan and oligoarabinoxylan (MINIC et al, 2004, aforementioned). Therefore, enzymes showing α-L-arabinofuranosidase activity from both i0 families 3 and 51 of plant glycosyl hydrolases can hydrolyze various plant polysaccharides. However, their substrate specificity has yet to be determined.

As shown, these enzymes could be used to breakdown polysaccharides in oligosaccharides for industrial purposes (eg : delignification, bioconversion, clarification, etc .). Another use of these enzymes consists in modulate their in vivo expression to modify the

>5 biological characteristics of plants. For instance, BRICE and MORRISON (Carboh. Res., 101 : 93-100, 1982) demonstrated that the quality of arabinoxylanes plays a role in the digestibility of the ray-grass. The hemicellulosic compounds are involved in a lot of biological phenomenon in plant, e.g. cell growth, pathogen resistance, drought and mechanical resistance. Due to this, the modulation of their in vivo expression of enzymes involved in the hemicellulosic compounds catabolism or anabolism is one of the aims of plant amelioration and selection.

The inventors have now purified and characterized from Arabidopsis thaliana an enzyme that exhibits both an α-L-arabinofuranosidase and a β-D-xylosidase activity and which is specifically expressed in the seed albumen at the globular stage of the embryo (the higher expression is observed 4 days after pollination). Moreover, they surprisingly found that the expression of this enzyme has an influence on the seed size.

The enzyme, designated XYL3, has an apparent molecular mass of 80 kDa when purified to homogeneity. It was identified using MALDI-TOF as a putative β-D-xylosidase encoded by gene At5g09730, previously classified among family 3 of glycoside hydrolases (CAZY:GH 3). In particular, it comprises the consensus N-terminal and C-terminal domains of this family, respectively defined in CDD database (MARCHLER-BAUER et ai, Nucleic Acids Res., 31 : 383-387, 2003) as Pfam 00933 and Pfam 01915.

XYL3 hydrolyzes synthetic substrates such as /?-nitrophenyl-α - L-arabinofuranoside and /?-nitrophenyl-β-D-xyloside with similar catalytic efficiency. XYL3 releases L-arabinose from (1— >5)-α-L-arabinofuranobiose, arabinoxylan, sugar beet arabinan and debranched arabinan. The enzyme hydrolyzes both arabinosyl substituted side group residues, and terminal arabinofuranosyl residues (1— >5)-α-linked to the arabinan backbone. This indicates that XYL3 is able to degrade all terminal arabinosyl residues. However, it shows a higher substrate specificity for arabinan than for arabinoxylan. This substrate specificity was determined using natural arabinans consisting of (1— »5)-α-linked residues, which are substituted at α(l— »3) and α(l— >2) with single or multiple-unit side chains. These results show that XYL3 can hydrolyze (1— »3)-o>, (1— »2)-α-linked side group residues and non-reducing terminal L-arabinofuranose residues of debranched (1— >5)-α-L-arabinan backbone, thus functioning as an α-L-arabinofuranosidase of both types A and B.

XYL3, in addition to releasing L-arabinose, can also release D-xylose from various natural and synthetic substrates, in particular arabinoxylan and xylan. Until now, none of the glycosyl hydrolases of family 3 that have been identified and characterized from various sources has been shown to have substrate specificity comparable to that of XYL3.

The inventors further identified T-DNA null mutants for AtBXi. The mutant plants lacked the α -L-arabinofuranosidase and β-D-xylosidase activities corresponding to XYL3, due to an insertion in the coding sequence of XYL3. The growth of null mutant plants and the wild type is the same. However and surprisingly, siliques of the mutant lines are smaller. The mutant seeds have a slightly smaller size, but their weight is not changed when compared to WT seeds. Although germination was delayed, the percent of germination remained comparable to the wild type seeds.

The present invention provides an isolated glycosyl hydrolase having the following characteristics: - it belongs to family 3 of glycosyl hydrolases ;

- it has a β-D-xylosidase (EC 3.2.1.37) and an α-L-arabinofuranosidase (EC. 3.2.1.55) activity;

- it cleaves terminal arabinosyl residues from arabinan backbone and from the side chains of arabinan or arabinoxylan - it has a molecular weight of 80 ± 1 kDa;

- it is expressed in seed albumen in early stages of embryo formation. According to a preferred embodiment of the invention, said glycosyl hydrolase has at least 80%, preferably at least 85%, preferably at least 90% and still more preferably at least 95% identity, with the polypeptide SEQ ID NO: 1. According to another preferred embodiment, of the invention said glycosyl hydrolase has a pH optimum of about 4.7, and a temperature optimum of about 65 °C; for the β-D-xylosidase activity.

The sequence SEQ ID NO: 1 represents the mature form of the glycosyl hydrolase of the invention, obtainable from Arabidopsis thaliana. The sequence of its precursor, including the signal peptide, is available in databanks, for instance as EMBL:Q9LXD6 and is also represented in the enclosed sequence listing under SEQ ID NO: 3.

The DNA sequence encoding said precursor is represented in the enclosed sequence listing under SEQ ID NO: 2:

Unless otherwise specified, the sequence identity values provided herein are calculated using the BLAST programs (ALTSCHUL et al, Nucleic Acids Res., 25: 3389, 1997) under default parameters, on a comparison window including the whole sequence SEQ ID NO: 1.

According to a preferred embodiment of the present invention, said glycosyl hydrolase is obtainable from flowers or immature seeds of a plant naturally expressing said glycosyl hydrolase. Preferably, said plant is Arabidopsis thaliana. The glycosyl hydrolase of the invention is useful in all applications wherein it is desired to use at the same time a α-L-arabinofuranosidase and β-D-xylosidase activity, for instance for polysaccharides degradation, or for the control of polysaccharides synthesis in a plant in order to obtain desirable phenotypic characteristics.

In a preferred embodiment, the invention provides means for controlling the size and density of seeds through modulation of the expression of a glycosyl hydrolase of the invention. According to a first aspect the invention provides a method for producing a plant having seeds with a reduced size and/or an increased density wherein said method comprises reducing or eliminating the endogenous expression of a glycosyl hydrolase of the invention in said plant. According to a second aspect, the invention provides a method for producing a plant having seeds with an increased size and/or a reduced density wherein said method comprises over-expressing a glycosyl hydrolase of the invention in said plant.

Reduction or elimination of the endogenous expression of a glycosyl hydrolase of the invention in a plant can be obtained for instance by mutagenesis of the corresponding coding region or of its regulatory regions, in particular its promoter. For instance, a mutation within the coding region can induce, depending on the nature of the mutation, a complete lack of expression of the protein, or the expression of an inactive protein, or the expression of a protein with impaired activity; in the same way, a mutation within the regulatory sequences can induce a lack of expression of said glycosyl hydrolase, or a decrease thereof. Mutagenesis can be performed for instance by targeted deletion of the glycosyl hydrolase coding sequence or promoter, or of a portion thereof, or by targeted insertion of an exogenous sequence within said coding sequence or said promoter. Said mutations can be generated for instance by T-DNA insertional mutagenesis, or by targeted gene replacement technology (KEMPIN et al, Nature, 389: 802-3, 1997 ; PUCHTA, Plant MoI. Biol., 48: 173-82, 2002).

It can also be performed by random chemical or physical mutagenesis, followed by screening of the plants mutated within the gene of said glycosyl hydrolase. Methods for high throughput mutagenesis and screening are available in the art. By way of example, one can mention TILLING (Targeting Induced Local Lesions IN Genomes, described by McCALLUM et al, Plant Physiol., 123: 439-442, 2000)

The endogenous expression of a glycosyl hydrolase of the invention in a plant can also be eliminated or reduced by epigenetic inactivation, for instance by antisense inhibition or co-suppression, as described by way of example in US Patents 5,190,065 and 5,283,323, or by transcriptional gene silencing (METTE et al, EMBO J., 19: 5194-201, 2000) or post- transcriptional gene silencing, using RNAi constructs (NISHIKURA, Cell, 107: 415-418, 2001 ; TENLLADO et al, Virus Res., 102: 85-96, 2004 ; ZAMORE, Methods MoI. Biol., 252: 533-543, 2004). It is also possible to use ribozymes targeting the mRNA of said glycosyl hydrolase.

In another embodiment, the present invention provides a process for preparing a glycosyl hydrolase of the invention, as defined above, wherein said process comprises providing biological material comprising said glycosyl hydrolase and isolating said glycosyl hydrolase from said biological material. According to a particular embodiment of the invention, said biological material comprises flowers and/or seeds in early stages of embryo formation of a plant naturally expressing said glycosyl hydrolase.

According to another particular embodiment of the invention, said biological 5 material comprises host-cells genetically modified by a polynucleotide encoding a glycosyl hydrolase of the invention, and expressing said glycosyl hydrolase.

Said biological material may be for instance a culture of said host-cells. It can also consist of plant cells, organs, or tissues obtained from a transgenic plant containing a transgene encoding a glycosyl hydrolase of the invention.

0 The invention also comprises recombinant DNA constructs useful in the practice of the invention.

This encompasses recombinant DNA constructs for expressing a glycosyl hydrolase of the invention, as well as recombinant DNA constructs for reducing or eliminating the endogenous expression of a glycosyl hydrolase of the invention.

5 DNA constructs for expressing a glycosyl hydrolase of the invention include in particular recombinant expression cassettes and recombinant expression vectors comprising a polynucleotide encoding a glycosyl hydrolase of the invention, wherein said polynucleotide is under transcriptional control of a suitable promoter, i.e a promoter which is functional in the chosen host-cell. These recombinant DNA constructs classically also include transcription !θ terminator sequences. They may also include other regulatory sequences, such as transcription enhancer sequences.

The DNA constructs of the invention may also comprise a signal sequence allowing to direct the expressed glycosyl hydrolase for secretion or to a particular sub-cellular compartment of the host cell (for instance endoplasmic reticulum, chloroplast, vacuole, !5 membranes, cell walls).

The invention also comprises DNA constructs for reducing or eliminating the endogenous expression of a glycosyl hydrolase of the invention in a plant. Said DNA constructs include in particular DNA constructs comprising a polynucleotide encoding a glycosyl hydrolase of the invention of the invention in anti-sense-orientation, and DNA constructs expressing a >0 RNAi targeting the gene encoding a glycosyl hydrolase of the invention, under transcriptional control of a suitable promoter. Advantageously, said RNAi is a double stranded RNA (dsRNA) and in particular hairpin RNA (hpRNA) targeting the glycosyl hydrolase to be silenced (WESLEY et al, Plant J., 27: 581-590, 2001).

The choice of the more appropriate promoter and other regulatory sequences

'5 and, eventually, of the more appropriate signal sequence depends in particular on the chosen host, and on the targeted cellular compartment; in the case of a plant, it may also depend on the targeted organ or tissue. One can use the endogenous promoter and/or the endogeneous tenninator, and/or the endogeneous signal sequence of a glycosyl hydrolase of the invention. However, in many cases, one will prefer to replace at least one of these elements by an heterologous sequence, more adapted to the chosen host.

5 Non-limitative examples of constitutive promoters that are commonly used in plant cells are the cauliflower mosaic virus (CaMV) 35S promoter, the Nos promoter, the rubisco promoter, the rice actin promoter.

As a tissue specific promoter one can advantageously use an albumen specific promoter such as pBETLl (HUEROS et al., Plant Cell, 7: 747-757, 1995) or pBETL9 (DOAN et

O al., Plant MoI. Biol., 31 : 877-886, 1996) which are specific of the basal endospeπn transfer layer, or pESR2 which is specific of the embryo surrounding region (BONELLO et al., Gene, 246: 219-

227, 2000).

The present invention further provides a recombinant vector comprising a DNA construct of the invention. The choice of the vector depends on its intended use (propagation,

5 expression or both), on the intended host and on the intended method of transformation. The selection of suitable vectors and the methods for inserting DNA constructs therein are well known to persons of ordinary skill in the art.

Generally, vectors also include one or more marker genes, which allow for selection of transformed hosts.

¹O The DNA constructs and vectors of the invention can be obtained and introduced in a host cell or organism by the well-known techniques of recombinant DNA and genetic engineering.

The invention also comprises host cells or organisms genetically modified by a construct of the invention. The polynucleotide may be transiently expressed; it can also be 5 incorporated in a stable extrachromosomal replicon, or integrated in the chromosome.

In particular, the invention comprises transformed plant cells and transgenic plants genetically modified by a DNA construct of the invention, as well as mutant plants wherein the endogenous expression of a glycosyl hydrolase of the invention has been reduced or eliminated by said mutation.

0 The invention also encompasses isolated organs or tissues of said plants (such as seeds, leafs, flowers, roots).

Many methods for transformation of plants are available in the art for many plant species, dicotyledons or monocotyledons. By way of non-limitative examples, one can mention virus mediated transformation, transformation by microinjection, by electroporation, >5 microprojectile mediated transformation, Agrobacterium mediated transformation, and the like. Transgenic plants genetically modified by a DNA construct expressing a glycosyl hydrolase of the invention have seeds with an increased size compared to the wild-type plants. This is beneficial since seed size is an important component of seed yield.

Also, due to their greater α-L-arabinofuranosidase and β-D-xylosidase activities, they are useful in many applications wherein it is desired to control polysaccharides degradation, as further explained below.

Transgenic and mutant plants wherein the endogenous expression of a glycosyl hydrolase of the invention has been reduced or eliminated by mutation or by the expression of a DNA construct of the invention have seeds with a reduced size and/or an increased density compared to the wild-type plants.

Transgenic and mutant plants of the invention include dicotyledons as well as monocotyledons. Preferred monocotyledons include for instance maize, wheat, barley, and rice; preferred dicotyledons include for instance colza, pea, sunflower.

Besides plant cells, preferred host cells for expressing a DNA construct of the invention include fungi, especially yeast.

Transformation of many species of yeast, including for instance Saccharomyces cerevisiae, Schizosaccharomyces pombe, Pichia pastoris, Yarrowia lipolyticci, etc. is well known in the art.

Examples of production of glycosyl hydrolases in yeast are described for instance in US Patent 4,794,175 or US Patent 5,529,919. Examples of production of glycosyl hydrolases in plants cells or transgenic plants are disclosed for instance in US Patent 6,818,803, or in US Patent 6,699,515 cited above.

The invention further provides a method for producing L-arabinose and/or D-xylose from non-cellulosic polysaccharide material of plant walls, wherein said method comprises contacting said non-cellulosic material with a glycosyl hydrolase of the invention, and recovering the L-arabinose and/or the D-xylose produced.

For carrying out the method of the invention, one can use a purified preparation of glycosyl hydrolase of the invention, obtained, as indicated above, from biological material comprising said glycosyl hydrolase. One can also use directly said biological material. In particular, one can add transformed host cells, or transgenic plants of the invention or parts thereof to the L-arabinose and/or D-xylose containing material to hydrolyze.

For instance, a transgenic plant of the invention of parts thereof expressing an glycosyl hydrolase of the invention can be used directly as a food for livestock, or as a silage additive, since it is expected that its activity will ensure a better digestion of the cell-wall materials. It can also be used as a raw material or as an additive for food preparation, for instance in beer or wine making.

According to a preferred embodiment of the invention, said non-cellulosic material is contacted with a mixture of enzymes comprising at least glycosyl hydrolase of the invention and at least one other glycosyl hydrolase. Preferably, said other glycosyl hydrolase is selected among polygalacturonases, endoxylanases, and other β-D-xylosidases.

The present invention also provides a multienzymatic composition comprising at least an glycosyl hydrolase of the invention, and at least one other glycosyl hydrolase.

Said multienzymatic composition can consist of a mixture of purified enzymes. It can also consist of a mixture of transformed host-cells, each of them expressing one different glycosyl hydrolase. Advantageously, said multienzymatic composition comprises a transformed host cell or a transgenic plants of the invention or a part thereof, expressing a glycosyl hydrolase of the invention and further expressing another glycosyl hydrolase.

Foregoing and other objects and advantages of the invention will become more apparent from the following detailed description and accompanying drawings. It is to be understood however that this foregoing detailed description is exemplary only and is not restrictive of the invention.

LEGENDS OF THE DRAWINGS

Fig. 1. Elution profile of β-D-xylosidase activity in CM-Sepharose cation-exchange chromatography using protein extracts from various tissues of Arabidopsis.

A 2 ml dialyzed cell-free extract was eluted as described for purification step 1 (see Materiel and Methods). The first 10 fractions were not analyzed in enzymatic assays since these fractions contain pigments which showed high absorbance at 405 nm.

Fig. 2. SDS-PAGE of purified enzymes.

The purified enzyme containing about 5 μg of protein was analyzed by SDS- PAGE (10% polyacrylamide gel) and proteins were visualized with Coomassie Brilliant blue R 250. Lane M, marker proteins (molecular masses are indicated).

Fig. 3. Alignment of XYL3 and ARA-I amino acid sequences.

Identical residues in the three sequences are shown in bold. The XYL3 and ARA-I cDNAs encode polypeptides of 774 and 778 amino acids, respectively. The start of NH2- terminal residue, after removal of signal sequences, is indicated by arrowhead. A bold overline shows the peptide sequence confirmed by NH 2 -terminal Edman analysis. A vertical line is used to indicate the likely COOH terminus in ARA-I, and asterisks indicate potential N-glycosylation sites. Arrows indicate the putative catalytic nucleophiles (Asp-299 for XYL3), and putative catalytic acid/bases (Glu-503 for XYL3). Overlines indicate peptide sequencing with MALDI- TOF after proteolytic cleavage by trypsin for XYL3.

Fig. 4. Time course of hydrolysis of WAX, SBA and debranched arabinan by XYL3 and ARAf.

5 A reaction mixture containing 0.1 μg of enzyme, 2% substrate (w/v) in

25 mM Na-acetate buffer, pH 5.0, was incubated at 37°C for different time periods. Aliquots were taken at different intervals and boiled for 3 min to inactivate the enzyme followed by quantification of the reaction products with high-performance anion-exchange (HPAE) chromatography (Dionex X500) equipped with a CarboPac PA-I column combined with pulse

0 amperometric detection (PAD). (A) release of D-xylose from arabinoxylan; (B) release of L-arabinose from arabinoxylan; (C) release of L-arabinose from arabinan (D) release of L-arabinose from debranched arabinan.

Fig. 5. Expression profiles ofAtBX-3 gene.

Semi quantitative expression was determined using reverse-transcription

5 (RT)-PCR. (A) AtBX-I) detection in different organs of Arabidopsis (Fl, flowers ; Si, siliques ; Rl, rosette leaves; Cl, cauline leaves ; St, floral stem ; Ro, roots ; Pl, plantlets). (B) AtBX-Z detection in different stages of flower and silique formation (St 0, closed bud ; StI , flowers with stamens sticking out of flowers by 2 mm ; StI, flowers with stamen sticking out of flowers by 5 mm ; St3, fade flowers, St3, St4, developing siliques, stage 12 dried siliques ; (C) Si, silique envelopes

!0 4 daf; Se, isolated seeds 4 dap), a, hybridization with the AtBX-3 probe, b, hybridization with a the β-tubulin-4 gene considered as constitutively expressed.

Fig.6. In situ hybridization performed on developing seeds.

The probe corresponds to the antisense transcript of AtBX-3. The selected photos present the seed when the embryos is at the globular stage and correspond at the stage

!5 when this gene is expressed in the albumen (A). The control (B) corresponds to the same hybridization performed on seeds of a knockout mutant for this gene (see Exemple 4 for the obtaining of this mutant).

Fig. 7. Schematic representation of AtBX-3 gene and localization of the T-DNA insertions in each T-DNA line.

10 The number of nucleotides is indicated in the white rectangles representing each exon. The position of the T-DNA insertions for each mutant is indicated by flags. The position of the primers used to identify the mutant lines (PIf and PIr, P2f and P2r for Atbx3a and -3b respectively) and to verify the absence of mRNA expression in the mutant lines (P3f and P3r) are indicated by arrows. Fig. 8. Analysis of β-D-xylosidase and α-L-arabinofuranosidase activity in the AtBX-3 mutant using cation-exchange chromatography.

Protein extracts obtained from wild type and T-DNA mutant lines of Arabidopsis were analyzed by CM-Sepharose chromatography. The protein extracts were

5 dialyzed in 25 mM Na-acetate buffer (pH 5.0) in the presence of 5% glycerol (v/v). Two milliliters (about 1.4 mg of protein) extract were loaded on a CM-Sepharose (Sigma) column (1.5 x 4 cm) and eluted first with 20 mM Na-acetate buffer (pH 5.0) in the presence of 0.015% Triton X-100 (w/v), and then with the same buffer in presence of a 0.1-0.5 M NaCl discontinuous gradient. One milliliter fractions were collected and 100 μl assayed for

0 α-L-arabinofuranosidase and β-D-xylosidase activities at 37°C for 60 min.

Fig.9. Consequence of XYL3 absence on mutant lines.

Comparison between siliques of wildtype (WT) and a Atbx-3a mutant (Atbx-3α). The photos were taken using a binocular microscope (NIKON SMZ-IOA).

Fig.lO. Binary constructs used for AtXYL3 expression in maize.

5 (A) pActin-AtXYL3 comprises the AtXYL3 coding region under the control of the rice Actin promoter, (B) pESR2-AtXYL3 comprises the AtXYL3 coding region under the control of the maize ESR2 promoter, (C) pBETL9-AtXYL3 comprises the AtXYL3 coding region under the control of the BETL9 promoter.

EXAMPLES :

!0 Experimental Procedures

Plant Material:

The ecotype Wassilewskija (WS) was used in this work. Wild-type (WT) WS Arabidopsis plants were grown in the greenhouse at 20⁰C to 22°C with a 16-h photoperiod at 150 μE rrf^{2 s} '. Mutants and WT Arabidopsis were grown together in the same greenhouse to !5 ensure uniform environmental conditions. The Atbx-3 mutants were selected using a systematic border sequencing program (http://flagdb-genoplante-info.infobiofen.fr) of the Versailles collection of T-DNA lines. T3 homozygous mutant lines were selected on Estelle and Somerville (ESTELLE and SOMERVILLE, MoI. Gen Genet., 206: 200-206, 1987) medium containing kanamycin (100 mgL-¹).

50 Chemicals:

/?NP-Glycoside substrates, D-xyl, L-ara, />NP, and OSX (approx. 10% L-arabinose and 15% glucose residues), were purchased from Sigma (St. Louis). (1— »5)-α-L-arabinobiose, WAX, SBA and linear ( 1 — »5)-α-linked debranched arabinan (88:4:2:6 L-arabinose-galactose-rhamnose-galacturonic acid) were purchased from Megazyme International (Bray, Ireland).

Preparation of protein extracts from Arabidopsis tissues:

Tissue was harvested from Arabidopsis at the flowering stage. Approximately 2 g of tissue were suspended in 2 mL of ice-cold extraction buffer and blended for 5 min. The extraction buffer consisted of 25 raM BisTris, pH 7.0, 200 mM CaC12, 10% (v/v) glycerol,

4 μM Na-cacodylate, and 1/200 (v/v) protease inhibitor cocktail (P-9599; Sigma). The ground and suspended material was centrifuged twice at 4°C for 3 min at 10,000 g and the supernatant was additionally centrifuged for 15 min at 17,000 g. The resulting supernatant was used for chromatographic analyses.

Chromatographic purification of enzymes exhibiting β-D-xylosidase activity:

The protein extract obtained from flowers or siliques of WT Arabidopsis was used for purification of β-D-xylosidases. The protein purification was performed in four steps as described below. Step 1 : cation-exchange chromatography.

The pooled fractions were concentrated and then equilibrated in 25 mM Na- Acetate buffer (pH 5.0) in the presence 5% glycerol (v/v) and 0.015% Triton X-100 and loaded on a CM-Sepharose (Sigma) cation-exchange column (1.5 x 4 cm; Sigma). Proteins were eluted with the same buffer, first alone and then with a 0.0 - 0.5 M NaCl discontinuous gradient. One-milliliter fractions were collected and 50 μl assayed for β-D-xylosidase activity. Peak fractions showing β -D-xylosidase activity were pooled and used in the second step of purification. Step 2: lectin chromatography.

A 0.5 x 3 cm column was filled with 1 ml of ConA Sepharose (Sigma) and washed with 3 ml of 20 mM Tris-HCl,0.5 M Na Cl buffer (pH 7.4).The soluble protein extract was loaded and then column washed with 10 ml of the buffer. The column was eluted with 0.2 M methyL-α-glucopyranoside in the same buffer. The eluates were collected and 50 μl samples from each fraction were tested for β-D-xylosidase activity as described below. Step 3: gel-filtration chromatography. The pooled and concentrated fractions showing β-D-xylosidase activities were fractionated by FPLC (Pharmacia) on a Superdex 200 HR 10/30 column (Amercham Pharmacia Biotech) pre-calibrated with the following markers of known molecular mass: thyroglobulin (67O kDa), bovine gamma globulin (158 kDa), chicken ovalbumin (44 kDa), equine myoglobin (17 kDa) and vitamin B-12 (1.35 kDa). Equilibration and elution were performed at room temperature with 20 mM Na-acetate buffer (pH 5.0), containing 150 mM NaCl. Fractions of 0.4 ml were collected at a flow rate of 0.5 ml/min and 50 μl of each fraction was assayed for β-D-xylosidase activity. Fractions exhibiting β-D-xylosidase activities were pooled, concentrated (150 μl) and dialyzed against 25 mM Tris-HCl buffer (pH 7.4), in the presence of 5 5% glycerol (v/v).

Step 4: anion-exchange chromatography.

The pooled and dialyzed fractions were loaded on a DEAE Sephacel (Sigma) anion-exchange column (1.5 x 4 cm; Sigma). Proteins were eluted with 25 mM Tris-HCl buffer

(pH 7.4), first alone and then in a 0.0-0.4 M NaCl discontinuous gradient. One-milliliter fractions

0 were collected and 100 μl assayed for β-D-xylosidase activity. Peak fractions showing β-D-xylosidase activity were pooled, concentrated and used for biochemical analysis.

SDS-PAGE:

Protein-denaturing SDS-PAGE was carried out using 10% polyacrylamide gels (LAEMMLI, Nature, 227: 680-685, 1970). Standard markers (BenchMark; Invitrogen, Carlsbad, 5 CA) were used to determine the approximate molecular masses of purified proteins in gels stained with Coomassie Brilliant Blue R-250.

Identification of Proteins by Mass Spectroscopy:

The protein bands obtained after SDS-PAGE were excised and in-gel digested with trypsin according to conditions for loading and elution described by SANTONI et al.

!0 (Biochem. J., 373: 289-296, 2003). Tryptic peptides from each protein were analyzed by

MALDI-TOF mass spectroscopy on a REFLEX III instrument (Bruker Instruments, Billerica,

MA). Finally, proteins were identified using Mascot (http://www.matrixscience.com/). β-D-xylosidase and α-L-arabinofuranosidase activity assays:

The reaction mixture contained 2 mM /?-nitrophenyl β-D-xylopyranoside

!5 (pNPX)(Sigma) or 2 mM />-nitro phenyl α-L-arabinofuranoside (pNPAf) (Sigma),

0.1 MNa acetate buffer (pH 5 .0) and 50-100 μl of protein extract in a total volume of 0.5 ml

(GOUJON et al, Plant. J., 33: 677-690, 2003). The reaction was carried out at 37° C for 60 min and stopped by the addition of 0.5 ml of 0.4 M sodium bicarbonate to the assay mixture. Staining intensity was determined spectrophotometrically at 405 nm, and the amount of the resulting p-

10 nitrophenol was determined from a calibration curve. The activity toward other p- nitrophenylglycosides was determined as described above for β-D-xylosidase or α-L-arabinofuranosidase. Time course of hydrolysis of polysaccharides by the purified enzymes:

For the time course of hydrolysis of polysaccharides, the reaction mixture contained 0.1 μg purified enzyme in 0.5 ml, 25 mM sodium acetate buffer (pH 5.0) and 2% of polysaccharide. A control reaction was performed in the absence of protein extract. After

5 different times of incubations at 37°C, each reaction mixture was boiled 3 min to stop the reaction. The mixture was then centrifuged at 17000 g for 5 min and supernatant was used for analysis.

The cleavage products were fractionated on high-performance anion-exchange (HPAE) chromatographic system (Dionex X500) equipped with a CarboPac PA-I column, and

0 combined with pulse amperometric detection (PAD). Degradation products were quantified as described previously (MINIC et al, 2004, aforementioned). pH and Temperature Profiles :

The determination of temperature dependence was carried out at pH 5.0 as described above for the β-D-xylosidase and α-L-arabinofuranosidase assays, except that the 5 temperature ranged from 30⁰C to 70⁰C. For the determination of pH optimum, the temperature was 37⁰C and the pH varied from 3.0 to 8.0 in 100 mM acetate citrate buffer.

Kinetic Analyses:

Kinetic parameters of purified enzymes were determined for the substrate /?NPX in a concentration range of 0.05 to 4.0 mM and of 0.2 to 8.0 mM for pNPAf and !0 (1— >5)-α-L-arabinobiose substrates respectively. Assays were performed in 100 mM sodium acetate buffer, pH 5.0. SD values for assays were less than 5%. Kinetic data were processed using Kaleidagraph program (Synergy Software, Reading, PA) based on Michaelis-Menten enzyme kinetics (ATKINS and NIMMO, Anal. Biochem., 104: 1-9, 1980). mRNA expression

'.5 Total RNA was extracted from different Arabidopsis tissues using trizol

(Quiagen USA, Valencia, CA) and used for semi quantitative RT-PCR experiments. The following primers :

P3f 5' CAAGGCGGGTTTGGTAA 3' (SEQ ID NO: 4) and 10 P3r 5' GGCGAATGTAATCTCAAATC 3' (SEQ ID NO: 5) were used to amplify the AtBX-3 cDNA. A control RT-PCR was performed with the same cDNA with the β-tubulin gene At5g44340 using the following primers:

5' GTCCAGTGTCGTGATATTGCAC 3' (SEQ ID NO: 6) and 5' GCTTACGAATCCGAGGGTGCC 3' (SEQ ID NO: 7). In situ hybridization:

Antisense transcript for AtBXi cDNA was synthesised and labeled in vitro with digoxigenin 1-lUTP using T7 RNA polymerase (Promega Kit). In vitro transcription reaction

5 was performed following the manufacturer's protocol, except the ratio of labeled to unlabeled UTP was 1 :1.

The template for transcription of AtBXS antisense probes was derived from a PCR 626 bp fragment of the cDNA amplified by PCR with the following primers: 5' TGT AAT ACG ACT CAC TAT AGG GCG GCG AAT GTA ATG TCA 3' (SEQ ID NO: 8)

0 and

5' CAA GGC GGG TTT GGT AA 3' (SEQ ID NO: 4).

Tissues were fixed, dehydrated, and embedded according to the method of LINCOLD et al. (Plant Cell, 6: 1859-1876, 1994), except that hybridization with AtBX3 probe was performed at 45°C and the final wash was performed in 2 x SSC, 50% formamide at 45°C.

5 Immunological detection was performed as described by COEN et al. (Cell, 63: 131 1-1322, 1990). Molecular characterization of the mutants :

Genomic DNA was extracted from the leaves of WT and mutant plants as described by DOYLE and DOYLE (Focus, 12: 13-15, 1990). PCR was performed in standard ¹O conditions (SAMBROOK et al., Molecular cloning: A laboratory manual, Ed. 2. Cold Spring

Harbour Laboratory Press, 1989) using primers flanking the T-DNA insertion:

PIf 5 ' CCCAACACTCTCAAGCAA 3 ' (SEQ ID NO: 9) ;

PIr 5' CAAACCRGCCAATAGAGA 3' (SEQ ID NO: 10) ;

P2f 5' ATGTGTGAAACTGAGATTG 3' (SEQ ID NO: 11) ; '5 P2r 5 ' TGGTCGAATTT AGTGT AGG 3 ' (SEQ ID NO: 12).

Microscopic observation of siliques and seeds:

Young immature siliques of WT and Atbx-3 mutants were sampled and incubated in alcohol to eliminate chlorophyll before observation under a binocular microscope (Nikon SMZ-IOA).

>0 Seed size and weight:

The weight of 7 samples of 20 seeds of WT and mutants were determined with a precision balance (Sartorius M2P, about 1 μg precision). The size of WT and mutant seeds was determined using the NIH Image 1.56 software (http://rsb.info.nih.gov/nih-image). Bioinformatics Analyses:

Sequences were aligned using the FASTA program (http://fasta.bioch.virginia.edu/fasta/align.htm) (CORPET, Nucl. Acids Res., 16: 10881-10890, 1988). Cleavage and glycosylation sites of deduced proteins were studied using PSORT and 5 NetNGlyc software (NIELSEN et al., Int. J. Neural Syst., 8: 581-599, 1997; NAKAI and HORTON, Trends Biochem. Sci., 24: 34-36, 1999) at http://psort.nibb.ac.jp/form.html and http://www.cbs.dtu.dk/services/NetNGlyc/.

Protein Measurements:

Protein concentrations were determined as described by BRADFORD (Anal. O Biochem., 72: 248-254, 1976) using to the Coomassie Brilliant Blue method with bovine serum albumin as a standard.

EXAMPLE 1: IDENTIFICATION, PURIFICATION AND CHARACTERIZATION OF XYL3

EIution profile of β-D-xylosidase activities after cation exchange chromatography on CM- 5 Sepharose in various tissues of Arabidopsis

Seven putative xylosidase genes, which encode proteins that are phylogenetically related to β-D-xylosidases, have been identified in the genome of Arabidopsis (GOUJON et al., 2003, aforementioned). Enzymatic properties have been studied for two of these enzymes (XYLl and XYL4) (MINIC et al, 2004, aforementioned). These enzymes were purified

¹O and identified from Arabidopsis stem tissue (MINIC et al., 2004, aforementioned). In order to identify other β-D-xylosidases, this activity was examined in different Arabidopsis tissues. Protein extracts from various Arabidopsis tissues (stem, roots, leaves, flowers and siliques) were fractionated by cation-exchange chromatography on CM-Sepharose and pooled fractions were assayed for β-D-xylosidase activity. Three major peaks (I, II, III) of activity were present in

'5 different proportions in several tissues as shown in Fig. 1. Peaks I and III obtained from stem extracts corresponded to previously purified and characterized XYLl and XYL4 enzymes (MINIC et al., 2004, aforementioned). In addition, one major peak (II) of high β-D-xylosidase activity was observed in flower and silique tissues (Fig. 1 ). The elution position of this peak was different from the position of the peaks observed in stem tissues. This new peak, therefore, iO corresponded to an unknown β-D-xylosidase in Arabidopsis.

Purification of an enzyme with high β-D-xylosidase activity in Arabidopsis flower and silique tissues β-D-xylosidase activity was purified from a crude protein extract of Arabidopsis flowers. /?NPX was used as the substrate to monitor its enzymatic activity. Table I indicates the degree of purification and yield for each step. The purification protocol involved four steps of column chromatography: In the first step a major peak of activity was eluted from a CM-Sepharose column (Fig. 1). Pooled fractions showing β-D-xylosidase activity were loaded onto a Con -A Sepharose column. Resulting fractions with β-D-xylosidase activity were purified by gel filtration on Superdex 200. The pooled fractions with β-D-xylosidase activity from this step were further purified using anion exchange chromatography on DEAE Sephacel. The enzyme yields and purification factors are summarized in Table I.

TABLE I

Step of purification Yield Specific activity Recovery^* Purification

Protein Activity factor^* mg nmol/min nmol/min/mg % fold

Crude homogenate 2.500 15.00 5.8 100 1

CM-Sepharose 0.470 10.00 21.3 66.7 7.7

ConA Sepharose 0,087 4.28 48.4 28.5 8.2

Superdex 200 0.023 1.56 67.8 10.4 16.9

DEAE-Sephacel 0.009 0.73 81.1 4.8 12.1

* Recoveries are expressed as percentage of initial activity and purification factors are calculated on the basis of specific activities.

SDS-PAGE analysis of the isolated enzyme

The purity of the protein isolated after the DEAE Sepharose was examined by SDS-PAGE. Staining with Coomassie Brilliant Blue R-250 revealed only one band (Fig. 2). The apparent molecular mass of the protein was 80 kDa. In addition, Superdex 200 chromatography of the purified protein suggests that it is a native monomer since it was eluted at a position corresponding to a molecular mass similar to that determined on denaturing SDS-PAGE (data not shown).

Identification of the purified enzyme

In order to identify the purified enzyme, the protein band obtained after SDS- PAGE was analyzed by MALDI-TOF using "Mascof'algorithms. The enzyme was identified as a putative β-D-xylosidase, encoded in the Arabidopsis genome by the gene At5g09730 and named AtBXLi in GOUJON et al. (2003, aforementioned) (accession number T49925 in the PIR-PSD data bank).We designated this enzyme as XYL3. N-Terminal amino acid sequencing of the first 8 residues of the purified enzyme confirmed amino acid identity with XYL3 (Fig. 3). Analysis of the primary structure of the purified enzyme

A BLAST sequence similarity search revealed that XYL3 protein showed 76% amino acid sequence identity with XYL4 β-D-xylosidase from Arabidopsis (MfNIC et al., 2004, aforementioned), 62 % with ARA-I bifunctional α-L-arabinofuranosidase/β-D-xylosidase from barley (LEE et al., 2003, aforementioned) and 55% with XYLl bifunctional α-L-arabinofuranosidase/β-D-xylosidase from Arabidopsis (MINIC et al., 2004, aforementioned). The detailed molecular properties and position of putative catalytic sites has been reported for ARA-I (LEE et al., 2003, aforementioned). The amino acid sequence of XYL3 was aligned with the barley ARA-I amino acid sequence to identify molecular properties and the

5 position of putative catalytic residues (Fig. 3). The bioinformatic analysis using PSORT predicts the existence of a signal sequence containing 23 amino acids (Fig. 3). This predicted NH2- terminus residue of XYL3 was experimentally confirmed by Edman sequence analysis. This analysis showed that the NH2 terminal was the GIu residue of the EQSNNQSS sequence. The molecular mass calculated from the XYL3 sequence after removal of the NH2-terminal signal

0 sequences is 80.6 kDa (Fig. 3). This value corresponds well with that obtained by SDS/PAGE (Fig. 2). Based on the sequence alignment of XYL3 from Arabidopsis with ARA-I purified from barley seedlings, the putative catalytic nucleophile for XYL3 is predicted to be Asp-299. The alignment also suggests that the catalytic acid/base amino acid is at position Glu-503. These catalytic amino acid residues are highly conserved in family 3 of glycoside hydrolases (31 ;

5 afmb.cnrs-mrs.fr). In addition, some other common features of glycoside hydrolases from the family 3 were observed in XYL3 sequence, including the conserved WGR and KH motifs, beginning at residue Trp-173 and Lys-216 for XYL3. These motifs are thought to be involved in substrate binding (HRMOVA et al., Structure, 9: 1005-1016, 2001 ; HRMOVA et al., Plant Cell, 14: 1033-1052, 2002).

!0 Finally, XYL3 possesses 4 potential sites for 7V-glycosylation (Fig. 3).

Previous studies (LEE et al., J. Biol. Chem., 278: 5377-5387, 2003 ; MINIC et al., Plant Physiol., 135: 867-878, 2004) have shown that although plant β-D-xylosidases can have high amino acid homology their substrate specificities can be quite different. Therefore, to determine the substrate specificities of XYL3, its enzymatic activity was tested using various

!5 artificial pNP-glycosides and natural substrates.

EXAMPLE 2: FUNCTI ONN AL PROPERTIES OF XYL3

Determination of substrate specificities of the purified enzyme using pNP-2lvcosides

Aryl glycosides were used as substrates with purified proteins (0.1 mg) in standard assays at a final concentration of 4 mM, as described in Material and Methods. 10 Activities were expressed as the percent of activity compared to the maximal substrate activity obtained.

Relative activity of XYL3 with different substrates is shown in Table II. TABLE Il Substrate Relative activity (%) pNP-α-L-arabinofuranoside 100 pNP-β-D-xylopyranoside 61 oNP-β-D-xylopyranoside 26 pNP-α-L-arabinopyranoside 5 pNP-β-D-Galactopyranoside 2 pNP-α-D-Galactopyranoside 3

No activity was detected for /?NP-β-D-Fucopyranoside, />NP- β-D-glucopyranoside, /?NP-a-D-glucopyranoside, pNP-β-L-arabinopyranoside, Me- β-D-xylopyranoside, />NP-α-D-xylopyranoside, /?NP-a-D-Fucopyranoside, />NP- β-D-Mannopyranoside, j?NP-α-D-Mannopyranoside, and />NP-β-D-glucuronide.

On the basis of the results reported in Table II, the purified enzyme showed activity for />-nitrophenyL-α-L-arabinofuranoside (/7NPAf) and for p- nitrophenyL-β-D-xylopyranoside (pNPX). Similar findings have been observed for α-L-arabinofuranosidase from Arabidopsis and for bifunctional α-L-arabinofuranosidase/β-D-xylosidases isolated from Arabidopsis (MINIC et al., 2004, aforementioned) and from barley (LEE et al., 2003, aforementioned). In addition to />NPX and />NPAf, the purified XYL3 could hydrolyze other substrates such as /λNPβGal, />NPαGal and />NPA but less efficiently (Table II).

Kinetic properties and enzymatic characterization obtained using NP-glucosides

To confirm the bifunctional specificity of XYL3, we tested its catalytic efficiency towards pNPAf and pNPX substrates. The results are reported in Table HIA.

TABLE IMA

Substrate Km (37°C) Vmax Vmax/Km

(mM) (nmol/min) (nmol/min/mM)

NPX 0.26 ± 0.02 1.02 ± 0.02 3.92 ± 0.01

NPAf 3.52 ± 0.25 18.96 ± 1.51 5.38 ± 0.22

The Vmax/Km ratios obtained were 5.4 nmol.min-l .mM-1 and 3.9 nmol.min '.mM-l for pNPAf and pNPX, respectively. The small difference in Vmax/Km ratios suggests that XYL3 acts as bifunctional α-L-arabinofuranosidase/β-D-xylosidase.

The influence of pH on the activity of the enzyme was investigated between pH 3 to 8. The pH optima determined for XYL3 was 4.7. The temperature dependence of XYL3 activity was tested between 20°C to 75°C, and the apparent optimum temperature was 65 ⁰C. Enzyme activities of XYL3 towards the synthetic />-nitrophenol glycosides were similar to those observed for α-L-arabinofuranosidase (ARAf), which was recently purified from Arabidopsis stem tissue (MINIC et al., 2004, aforementioned). Hydrolysis of xylopolysaceharides, arabinan, and arabinobiose by XYL3

In order to deteπnine eventual functional differences between XYL3 (family 3 of glycosyl hydrolases) and ARAf (family 51 of glycosyl hydrolases), we tested their hydrolysis efficiency towards natural plant polysaccharides that contain L-arabinose and D-xylose.

5 Incubation of the purified enzyme with xylopolysaccharides and arabinan was carried out to determine if it could degrade native plant polysaccharides. Neoformed products were analyzed by HPAE and compared with those released by ARAf. Time course hydrolysis of these polysaccharides are shown in Fig. 4. The results obtained indicate that both XYL3 and ARAf release D-xylose in similar ratio from wheat arabinoxylan (WAX) (Fig 4A) and oat spelt xylan

0 (OSX; data not shown). The release of L-arabinose from WAX by ARAf is more efficient than the release by XYL3 (Fig 4B). The highest production of L-arabinose by XYL3 was observed using sugar beet arabinan (SBA) as the substrate (Fig 4C) whereas ARAf gives low yields of L-arabinose from this substrate. Thus, the degradation activities of the two enzymes on wheat arabinoxylan were the opposite of their activities on sugar beet arabinan. Moreover, both

5 enzymes degraded debranched α(l— »5)-L-arabinan backbones (Fig. 4D).

To further confirm the functional specificity of XYL3 and ARAf, their catalytic efficiency was tested with (1 — >5)-α-L-arabinobiose as a substrate; the results of this experiment are shown in Table IHB. TABLE IHB

Km (37°C) Vmax Vmax/Km

(mM) (nmol/min) (nmol/min/mM)

XYL3 5.5 ± 0.9 25.1 ± 2.0 4.6 ± 0.7

ARAf 5.2 + 0.9 11.4 ± 1.0 2.2 ± 0.5

!0 The Vmax/Km ratios obtained were 2.2 ±0.5 nmol.min^" ImM-I and

4.6 ±0.7 nmol.min- LmM- 1 for ARAf and XYL3, respectively. Thus, the Vmax/Km ratio of XYL3 towards arabinobiose is higher than that of ARAf.

These results indicate that XYL3 and ARAf have similar specificities towards synthetic substrates such as NPX and NPAf, but different specificities towards the natural !5 polysaccharides tested. In contrast to ARAf which prefers arabinoxylan as substrate, XYL3 has a preference for the arabinan substrates tested (SBA, debranched arabinane and arabinobiose).

EXAMPLE 3: EXPRESSION PROFILE OF ATBX-3

Expression profile of AtBX-3

To determine the precise transcriptional expression profile of AtBX-3, the gene

10 encoding XYL3, total RNA was extracted from different organs of Arabidopsis (plantlets, floral stems, rosette leaves, cauline leaves, flowers and siliques) and used for RT-PCR experiments. The transcript was observed in flowers and siliques (Fig. 5A). In order to deteπnine the precise stage of expression in flower and silique formation, RNA of 12 different flower/silique developmental stages (st) were used in RT-PCR. AtBX-i signal was observed in the 3 first stages which correspond to closed buds, and flowers with stigma protruding from the flowers from two to five millimeters respectively. Highest expression levels were observed at the 2 mm stage (Fig. 5B for st 0 to 4; not shown for st 5 to 12). AtBXi signal was high in seeds 4 days after pollinisation (dap), very low in seeds 9 dap and not observed in silique envelopes at 4 daf (Fig. 5C and not shown). AtBXi expression was not detected in mature embryos, testa and in silique envelopes (green mature stage; data not shown). In conclusion, AtBX-i mRNA was specifically expressed in the early stage of seed formation and not at seed maturation.

In situ hybridization

In order to localize precisely the expression in the seeds, in situ hydridization experiment was performed on developing seeds of Arabidopsis at different stages of development using a AtBX-i probe. A high hydridization signal was exclusively observed in the albumen of very young seeds when the embryo was at the globular stage. No signal was observed in older stages of seed development. The specificity of the observed hybridization signal was demonstrated by the lack of hybridization in seeds of a knockout mutant for the gene encoding AtBX-3 at the same developing stage (Fig.6)

EXAMPLE 4: ISOLATION AND CHARACTERIZATION OF ATBX-3 MUTANTS

To further define the role of XYL3 we identified two null mutant lines for the AtBX-3 gene in the Versailles T-DNA insertion collection (BOUCHE and BOUCHEZ, Curr. Opin. Plant Biol., 4: 1 11-1 17, 2001). These lines (named Atbx-ia and Atbx-ib) were identified by systematic border sequencing (http://flagdb-genoplante-info.infobiogen.fr). The segregation of progeny of these lines, germinated on selective medium containing kanamycin, allowed us to infer that only one npt\\ insertion locus was present in each line. The sites of the insertions were localized in the fourth and sixth exons respectively (Fig. 7).

Homozygous lines were obtained and the impact of the T-DNA insertion on mRNA expression was determined by RT-PCR on total RNA from flowers. The absence of AtBX-i mRNA was confirmed in Atbx-ia and Atbx-ib (data not shown).

The homozygous mutant lines and wild-type Arabidopsis were cultivated at the same time in greenhouse conditions. β-D-xylosidase and α-L-arabinofuranosidase activities were determined in flowers of the mutant lines and the peak corresponding to XYL3 was absent from mutant extracts (Fig. 8 ioτ Atbx-ia, non shown for Atbx-ib). No obvious phenotype was observed for growth parameters on the two mutant lines when compared to the WT. However, siliques of the mutant lines were smaller than those of WT and inside, the seeds were smaller in size than WT seeds (Fig. 9).

The weight and size of mature dried seeds of WT and Atbx-3a was determined. The results are shown in Table IV.

TABLE IV

Line WT Abxb-3a

Weight of 20 seeds (μg) 329 ± 15 322 ± 12

Area of a seed (pixels) 805 ± 14 750 ± 18

The size of mature dried seeds of Atbx-3a was also slightly reduced but their weight was not changed when compared to WT seeds.

The rate of germination of the mutant seeds was reduced: 91.5% of wild type seeds were germinated five days after sawing whereas only 70.5% were germinated for the two 0 mutant lines. However all the mutant seeds were able to germinate later and no delay in plantlet development was noticed.

EXAMPLE 5: OVEREXPRESSION OF ATXL3 IN MAIZE FROM A CONSTITUTIVE PROMOTER..

Since reduction of XYL3 expression in Arαbidosis leads to a reduction in seed 5 size overexpression of XYL3 can be performed in order to increase seed size. Such overexpression can be achieved by expressing XYL3 from a constitutive promoter such as the rice Actin promoter (McELROY et αi, MoI. Gen. Genet., 231 : 150-160, 1991). A full length

Arαbidopsis XYL3 clone was obtained by PCR from a cDNA library made from mixed floral stages. A first round of PCR used the primers : !0 Ol 1080 XYL-F : 5 ' ATACAATGGCGAGCCGAAAC 3 ' (SEQ ID NO : 13) and

011065 XYL-R : 5' CG AGCCGTCGAATC AAAC ACTA 3' (SEQ ID NO: 14).

The second round of PCR introduced an Ncol site around the initiating ATG of

AtXYL3 and used the primers: !5 Ol 1079 XYL-F Long Ncol : 5 ' ATACCATGGCG AGCCGAAACAGAG 3 ' (SEQ ID NO: 15) and

011065 XYL-R : 5' CG AGCCGTCGAATC AAAC ACTA 3' (SEQ ID NO: 14).

The PCR product from the second PCR reaction was digested with Ncol and cloned into between the Ncol and EcoRV sites of the vector pENTR4 (Invitrogen) forming >0 pENTR4-XYL-NcoI. The AtXYL3 coding region was then placed under the control of the rice

Actin promoter by performing an LR recombination reaction with the GATEWAY destination binary vector pBIOS1 124 forming pActin-AtXYL3. The vector pBIOS1 124 is a derivative of pSB12 (KOMARI et αl., Plant J., 10: 165-174, 1996) containing a pActin-Bar gene for selection of maize transformants, a pCsVMV-GFP gene to follow the presence of the transgene in plants and seeds and a rice Actin promoter linked to an actin intron (McELROY et al., 1991, aforementioned) followed by a GATEWAY cassette and a polyadenylation sequence derived from the Arabidopsis Sac66 gene (JENKINS et al., Plant Cell Environ., 22: 159-167, 1999). pActin-AtXYL3 (Fig. 10A) was transferred into agrobacteria LBA4404 (pSBl) according to KOMARI et al. (1996, aforementioned) and the Maize cultivar A 188 was transformed with this agrobacterial strain essentially as described by ISHIDA et al. (Nature Biotechnol., 14: 745-750, 1996).

The transformed plants overexpressing AtXYL3 possess a normal vegetative phenotype however seed inheriting the transgene have a larger size compared to seed on the same cob that lack the transgene.

EXAMPLE 6: OVEREXPRESSION OF ATXL3 IN MAIZE FROM A PROMOTER EXPRESSED IN THE EMBRYO SURROUNDING REGION.

Since reduction of XYL3 expression in Arabidosis leads to a reduction in seed size overexpression of XYL3 can be performed in order to increase seed size. Such overexpression can be achieved by expressing XYL3 from an endosperm-specific promoter that is expressed early in endosperm development. Such a promoter is pESR2 (BONELLO et al., 2000, aforementioned) that is specifically expressed in the maize endosperm surrounding region (ESR). As described in Example 5, a full length Arabidopsis XYL3 clone was obtained by PCR from a cDNA library made from mixed floral stages and cloned into the vector pENTR4 (Invitrogen) forming pENTR4-AtXYL-Ncol.

The AtXYL3 coding region was then placed under the control of the maize ESR2 promoter by performing an LR recombination reaction with the GATEWAY destination binary vector pBIOS-ESR2 forming pESR2-AtXYL3. The vector pBIOS-ESR2 is a derivative of pSB12 (KOMARI et al., 1996, aforementioned) containing a pActin-Bar gene for selection of maize transformants, a pCsVMV- GFP gene to follow the presence of the transgene in plants and seeds and an ESR2 promoter followed by a GATEWAY cassette and a polyadenylation sequence derived from the Arabidopsis Sac66 gene (JENKINS et al., 1999, aforementioned). pESR2-AtXYL3 (Fig. 10B) is transferred into agrobacteria LBA4404 (pSBl) according to KOMARI et al. (1996, aforementioned) and the Maize cultivar Al 88 is transformed with this agrobacterial strain essentially as described by ISHIDA et al. (1996, aforementioned).

The transformed plants overexpressing AtXYL3 possess a normal vegetative phenotype however seed inheriting the transgene have a larger size compared to seed on the same cob that lack the transgene. EXAMPLE 7: OVEREXPRESSION OF ATXYL3 IN MAIZE FROM A PROMOTER EXPRESSED IN THE BASAL ENDOSPERM TRANSFERLAYER.

Since reduction of XYL3 expression in Arabidosis leads to a reduction in seed size overexpression of XYL3 can be performed in order to increase seed size. Such

5 overexpression can be achieved by expressing XYL3 from an endosperm-specific promoter that is expressed early in endosperm development. Such a promoter is BETL9 that is specifically expressed in the maize endosperm basal transfer layer (BETL). The BETL9 gene is homologous to the barley ENDl gene (DOAN et al., 1996, aforementioned) and has an expression pattern similar to the BETLl gene (HUEROS et al., 1995, aforementioned). As described in Example 5 a

IO full length Arabidopsis XYL3 clone was obtained by PCR from a cDNA library made from mixed floral stages and cloned into the vector pENTR4 (Invitrogen) forming pENTR4-AtXYL- Ncol.

The AtXYL3 coding region was then placed under the control of the maize BETL9 promoter by performing an LR recombination reaction with the GATEWAY destination

15 binary vector pBIOS 960 forming pBETL9-AtXYL3. The vector pBIOS 960 is a derivative of pSB12 (KOMARl et al., 1996, aforementioned) containing a pActin-Bar gene for selection of maize transformants, a pCsVMV-GFP gene to follow the presence of the transgene in plants and seeds and a BETL9 promoter followed by a GATEWAY cassette and a polyadenylation sequence derived from the Arabidopsis Sac66 gene (JENKINS et al., 1999, aforementioned). The 1941 bp

!0 maize BETL9 promoter was PCRed from genomic DNA of the inbred line F2 using the primers: pBETL9fw 5' CGATGGTACTTACTCATGATGGTCATCTAGG 3' (SEQ ID NO: 16) and pBETL9rw 5' CCATGGT AT AACTTCAACTGTTGACGG 3' (SEQ ID NO: 17). pBETL9-AtXYL3 (Fig. 10C) was transferred into agrobacteria LBA4404

!5 (pSBl) according to KOMARI et al. (1996, aforementioned) and the Maize cultivar Al 88 was transformed with this agrobacterial strain essentially as described by ISHIDA et al. (1996, aforementioned).

The transformed plants overexpressing AtXYL3 possess a normal vegetative phenotype however seed inheriting the transgene have a larger size compared to seed on the same

SO cob that lack the transgene.

Claims

1) An isolated glycosyl hydrolase having the following characteristics:

- it belongs to family 3 of glycosyl hydrolases ;

- it has a β-D-xylosidase and an α-L-arabinofuranosidase activity; - it cleaves terminal arabinosyl residues from arabinan backbone and from the side chains of arabinan or arabinoxylan

- it has a molecular weight of 80 ± 1 kDa;

- it is expressed in seed albumen in early stages of embryo formation.

2) An isolated glycosyl hydrolase of claim 1 , characterized in that it has at least 80% identity with the polypeptide SEQ ID NO: 1.

3) A method for producing a plant having seeds with a reduced size and an increased density wherein said method comprises reducing or eliminating the endogenous expression of a glycosyl hydrolase of claim 1 in said plant.

4) A method for producing a plant having seeds with an increased size and a reduced density wherein said method comprises over-expressing a glycosyl hydrolase of any of claims 1 or 2 in said plant

5) A recombinant DNA construct comprising a polynucleotide encoding a glycosyl hydrolase of any of claims 1 or 2, under transcriptional control of a promoter.

6) A recombinant DNA construct expressing a RNAi targeting the gene encoding a glycosyl hydrolase of any of claims 1 or 2, under transcriptional control of a promoter.

7) A recombinant vector comprising a DNA construct of any of claims 5 or 6.

8) An host cell genetically modified by a DNA construct of any of claims 5 or 6

9) A transgenic plant containing a transgene comprising a DNA construct of any of claims 5 or 6.

10) Use of a polynucleotide encoding a glycosyl hydrolase of any of claims 1 or 2 for increasing the size of seeds in a plant.

1 1) Use of a RNAi targeting the gene encoding a glycosyl hydrolase of any of claims 1 or 2, for decreasing the size of seeds in a plant.