Author for correspondence:Stephen C. Fry Tel: +44 131 650 5320 Fax: +44 131 650 5392 Email: S.Fry@ed.ac.uk
I. Primary cell walls: composition, deposition and roles000
II. Reactions that have been proposed to occur in primary cell walls000
III. Tracking the careers of wall components in vivo: evidence for action of enzymes in the walls of living plant cells000
IV. Evidence for the occurrence of nonenzymic polymer scission in vivo?000
Numerous examples have been presented of enzyme activities, assayed in vitro, that appear relevant to the synthesis of structural polysaccharides, and to their assembly and subsequent degradation in the primary cell walls (PCWs) of higher plants. The accumulation of the corresponding mRNAs, and of the (immunologically recognized) proteins, has often also (or instead) been reported. However, the presence of these mRNAs, antigens and enzymic activities has rarely been shown to correspond to enzyme action in the living plant cell. In some cases, apparent enzymic action is observed in vivo for which no enzyme activity can be detected in in-vitro assays; the converse also occurs. Methods are reviewed by which reactions involving structural wall polysaccharides can be tracked in vivo. Special attention is given to xyloglucan endotransglucosylase (XET), one of the two enzymic activities exhibited in vitro by xyloglucan endotransglucosylase/hydrolase (XTH) proteins, because of its probable importance in the construction and restructuring of the PCW's major hemicellulose. Attention is also given to the possibility that some reactions observed in the PCW in vivo are not directly enzymic, possibly involving the action of hydroxyl radicals. It is concluded that some proposed wall enzymes, for example XTHs, do act in vivo, but that for other enzymes this is not proven.
I. Primary cell walls: composition, deposition and roles
1. Biological roles
This article discusses the dynamics of the primary cell wall (PCW), the first-formed wall layer, whose cellulosic microfibrillar skeleton is deposited while the cell is (at least potentially) still expanding. The PCW is a unique fabric that is strong but usually thin, flexible, and capable of both plastic and elastic extension. Its shape and size dictate most of the externally recognizable features of plant development (Brett & Waldron, 1996; Fry, 2000). The PCW governs the rate and direction of cell expansion, and thus the ultimate size and shape of the cell. For example, sudden changes in the properties of the PCW trigger the explosive increase in cell expansion rate at the onset of germination (e.g. when an imbibed seed that had been held in ‘suspended animation’ by abscisic acid (ABA) is washed free of ABA) (Schopfer & Plachy, 1985). However, the nature of these sudden changes is still a mystery. Another role for wall polysaccharides is as the precursors of ‘oligosaccharins’; these are oligosaccharides that exert hormone-like effects (Darvill et al., 1992; Aldington & Fry, 1993; Dumville & Fry, 2000).
Modification of wall components in vivo can thus make a major contribution to the mechanics and control of plant growth and morphogenesis. The main themes of the present article are the metabolic reactions that modify PCW polymers, the existence of enzymes that could feasibly catalyse such reactions and a critical evaluation of the limited evidence that ‘feasible’ reactions actually happen in vivo.
2. Primary wall composition and cross-links
Primary cell walls are composed of polysaccharides, smaller proportions of glycoproteins and, in some specialized cell-types, various noncarbohydrate substances such as lignin, suberin, cutin, cutan or silica. Wall polysaccharides fall into three categories: pectins, hemicelluloses and cellulose. Pectins and hemicelluloses are components of the wall ‘matrix’, within which are embedded the skeletal, cellulosic microfibrils. The chemistry of these wall polymers is described in detail elsewhere (McNeil et al., 1984; Carpita & McCann, 2000; Fry, 2000). Briefly, pectins are galacturonate (GalA)-rich, acidic polysaccharides (partly methylesterified galacturonan, linked to the rhamnogalacturonans RG-I and RG-II, which are rich in arabinose (Ara), galactose (Gal), rhamnose (Rha). Hemicelluloses are GalA-free, neutral or slightly acidic polysaccharides that are extractable from the wall with aqueous NaOH (optimally 6 m at 37°C; Edelmann & Fry, 1992a) and can be adsorbed from neutralized aqueous solutions by hydrogen-bonding to cellulose (e.g. filter paper). Major hemicelluloses are the xylans (including arabinoxylans, glucuronoarabinoxylans, etc.), xyloglucans (composed mainly of glucose (Glc), xylose (Xyl), galactose (Gal), fucose (Fuc) residues), and mixed-linkage β-(1→3),(1→4)-d-glucans (MLGs; only found in the Gramineae and a few related families, such as the Restionaceae; Smith & Harris, 1999). Cellulose is composed of neutral, β-(1→4)-d-glucan chains, hydrogen-bonded together to form microfibrils. The ratio and exact composition of the various wall polysaccharides differs phylogenetically between plant taxa (Popper & Fry, 2003), spatially between the tissues of a given plant (Selvendran & O’Neill, 1987; Lynch & Staehelin, 1995), and temporally during the development of a given cell (Brett & Waldron, 1996). Major glycoproteins of the PCW include the extensins, composed of a basic polypeptide backbone rich in hydroxyprdine (Hyp), Ser, Lys and Tyr, the Hyp residues bearing mono- to tetra-saccharide side-chains of Ara (Josè & Puigdomènech, 1993).
Each polysaccharide and glycoprotein of the PCW matrix is, in isolation, water-soluble. However, in the intact wall, these molecules are crosslinked to form a ‘fabric’, which is essential if a structurally competent material is to be formed that holds together in an aqueous environment (Fry, 1986; Iiyama et al., 1994). Some of the crosslinks involved are noncovalent (individually weak, but numerous enough to confer strength): for example, the hemicelluloses hydrogen-bond to cellulose; galacturonans are crosslinked to each other by divalent Ca2+ bridges, and basic glycoproteins are ionically crosslinked to acidic polysaccharides. Other crosslinks are covalent (few but individually strong): for example, pairs of RG-II molecules are bonded together through borate diester cross-links (O’Neill et al., 1996, 2001); some polysaccharide chains may be crosslinked to each other by oxidatively coupled phenolic (e.g. ferulic acid) side-chains (Markwalder & Neukom, 1976; Fry, 1979; Ishii, 1991; Wende et al., 1999; Fry et al., 2000); some glycoproteins are crosslinked through oxidatively coupled tyrosine residues (Fry, 1982; Brady et al., 1996), and lignin may be covalently attached to certain polysaccharides through benzyl ether bridges to form lignin–carbohydrate complexes (Watanabe et al., 1989; Wallace et al., 1995).
In addition to true (lateral) crosslinks of the types mentioned, through which infinite macromolecules can, in principle, be created, there are also interpolymeric glycosidic bonds, which can link an individual polysaccharide, via its single reducing terminus, to another polysaccharide or protein molecule. For example, pectin probably consists of galacturonan, RG-I and RG-II covalently linked to each other end-to-end via glycosidic bonds (Mort, 2002). With branched polysaccharides, glycosidic bonds can potentially form complicated, tree-like structures: for example, some xyloglucans appear to be glycosidically linked to the side-chains of RG-I (Thompson & Fry, 2000). However, since polysaccharide–polysaccharide glycosidic linkages must involve a reducing terminus, of which there is only one per polysaccharide molecule, such linkages are not comparable to true cross-links (i.e. those where more than one can form with each partner). Hence, the statement that polysaccharides can be glycosidically linked together as long chains or as tree-like structures (the base of the tree trunk representing the reducing terminus), but separate trees cannot be cross-linked to each other through glycosidic bonds between their branches.
Our understanding of PCW cross-links and other interpolymer bonds, and thus of wall architecture, is far from complete. However, the formation of these cross-links is thought, in many cases, to be enzyme-catalysed — a major proposed role for wall-located enzymes.
3. Biosynthesis and deposition of the primary cell wall
Subcellular sites of polysaccharide synthase action The PCW polysaccharides are not synthesized within the wall itself; instead, they are produced by various membranes of protoplast. Cellulose and callose are synthesized on the apoplastic side of the plasma membrane using a cytosolic donor substrate, probably UDP-glucose (Köhle et al., 1985). (The apoplast is the aqueous solution that permeates the cell wall and is separated from the symplast by the plasma membrane.) Cellulose is produced by ‘terminal complexes’ (visible by electron microscopy) in the plasma membrane (Doblin et al., 2002). Each terminal complex churns out simultaneously several dozen cellulose chains, which almost instantly hydrogen-bond to each other to form a microfibril. That the hydrogen bonding is not quite instantaneous is suggested by the fact that adding Congo red, which intercalates with β-(1→4)-polysaccharides, can prevent newly synthesized cellulose from organizing into microfibrils, without blocking the synthesis of individual β-(1→4)-glucan chains (Robinson, 1981). Extruded lengths of microfibril quickly become fixed within the PCW; therefore, as the microfibril elongates, it must push its own terminal complex around within the plasma membrane. Part of the reason why extruded lengths of microfibril become fixed in the PCW is the fact that they become embedded in the newly secreted matrix polymers, some of which hydrogen-bond to and may tether the microfibrils (Fry, 1989; Hayashi, 1989), a process that can be modelled in vitro by allowing the cellulose-synthesizing bacterium Acetobacter to extrude its new cellulose chains into a solution of hemicelluloses (Iwata et al., 1998).
Matrix polysaccharides other than callose are synthesized at the inner face of Golgi membranes. In elegant experiments where 3H-labelled sugars were fed to living cells, the first-formed [3H]polysaccharides were detected (by electron microscopy (EM)–autoradiography) in the vicinity of Golgi cisternae, then a little later in the Golgi-derived vesicles, and finally in the PCW (Northcote & Pickett-Heaps, 1966). The cellular route for matrix polysaccharides appears to be cis Golgi cisternae → medial Golgi cisternae → trans Golgi cisternae → trans Golgi network → Golgi-derived vesicles → plasma membrane → apoplast (including primary wall), although not all polysaccharides begin life in cis-Golgi cisternae (some may start to form mainly in medial or trans cisternae). Later work has used immunocytochemical techniques to localize specific polysaccharides in specific Golgi cisternae (Moore et al., 1991; Zhang & Staehelin, 1992). These observations have great spatial resolution and may provide more rigorous information on the identity of the polysaccharides that was available to Northcote & Pickett-Heaps (1966); however, they lack the temporal information which can be provided by radiolabelling experiments. The abundance of a specific polysaccharide in the cis-Golgi cisternae, for example, does not prove that they are its major site of synthesis. This location could be a cellular ‘backwater’ where, for example, only a small amount of the polymer is synthesized but those molecules that are synthesized rarely proceed to the next stage of the vesicular traffic system and therefore accumulate to readily detectable levels.
Visualization of polysaccharide deposition within the wall by EM–autoradiography shows that new (i.e. recently radiolabelled) polysaccharides are principally laid down on the inner surface of the wall, adjacent to the plasma membrane — i.e. they have limited mobility within the wall — (Ray, 1967). By contrast, cultured cells permit a proportion of their ‘matrix’ polysaccharides to pass through the wall and to be sloughed into the culture medium (Becker et al., 1964; Aspinall et al., 1969; Kerr & Fry, 2003). Thus, polysaccharides are able to permeate the wall, perhaps at particular sites, at least in the living cell (whose wall may be stretched owing to turgor). In addition, it appears that the glycoprotein extensin can pass through the whole thickness of the wall and become deposited at various distances from the plasma membrane. However, extensins are unable to pass through the middle lamella from one cell to its neighbour, presumably because of ionic entrapment in the acidic polysaccharides of the middle lamella (Stafstrom & Staehelin, 1988).
Substrates of polysaccharide synthases Isolated membrane preparations can frequently generate polysaccharides in vitro using NDP-sugars as glycosyl donors (Brett & Waldron, 1996); these are therefore generally assumed to be the natural sugar donors in-vivo. However, this assumption is supported by surprisingly little in vivo evidence. For example, evidence that NDP-sugars act as precursors of cellulose in developing cotton hairs was obtained from experiments in which 14C from exogenous sucrose was incorporated into the UDP-glucose pool soon enough after administration of the precursor, and with adequate flux, for the UDP-glucose pool to be deemed ‘adequate’ for cellulose synthesis (Carpita & Delmer, 1981). Further evidence was provided by pulse-labelling experiments in which [3H]arabinose was fed to living cells and showed a progressive transfer of radiolabel, with increasing time, from Ara to Ara-1-P to UDP-Ara to UDP-Xyl and finally to the pentose residues of polysaccharides. No NDP-sugars other than UDP-pentoses became appreciably radioactive. The UDP-Ara and UDP-Xyl became radiolabelled shortly before the polysaccharides, supporting a precursor → product relationship (Fry & Northcote, 1983).
Protein ‘anchors’ for polysaccharide biosynthesis It has been suggested that matrix polysaccharides are synthesized (or at least initiated) as carbohydrate side-chains of membrane proteins (Crossthwaite et al., 1994). Such proteins await detailed characterization and it is unclear quite when, where and how the polysaccharides are cut free of the putative protein anchors.
How long does it take to synthesize a polysaccharide? Little work has been done to find out how long it takes the cell to complete the synthesis of a wall-destined polysaccharide chain in vivo. The total time interval between the incorporation of 3H-sugar residues into nascent polysaccharides and the emergence of 3H-polysaccharides external to the plasma membrane is typically in the order of 20 min (Myton & Fry, 1994). However, it is unclear whether this represents 1 s of synthesis followed by 20 min of transport, vice versa, or something between these extremes.
Sequential action of cooperating polysaccharide synthases Major polysaccharides synthesized in the Golgi bodies include pectins, xyloglucans, xylans and mannans. It appears usually to be polysaccharide synthase activities (rather than the supply of NDP-sugars) that are the dominant factors controlling wall polysaccharide biosynthesis during plant development. For example, pectin synthase and xylan synthase decrease and increase, respectively, during the transition from primary to secondary wall production during the development of wood (Bolwell & Northcote, 1981). Also, legume species differ in the Gal : Man ratio of their seed storage galactomannans, and this is dictated by the differing abilities of the polysaccharide synthase systems to add further α-Gal residues to the β-(1→4)-mannan backbone in the vicinity of β-Man residues that are already galactosylated (Edwards et al., 1999).
Polysaccharide labelling experiments in vivo enable an investigation of the action of enzymes, as distinct from their activity (assayed under optimized conditions in vitro). Studies of action can yield information about the relative timing, within the vesicular traffic system, of the biosynthesis of different hemicelluloses. After the administration of [3H]arabinose to cultured maize cells, intraprotoplasmic [Xyl-3H]xyloglucans and [Xyl-3H]xylans both started to be formed at about the same time (after a very short lag period due to the time taken to incorporate 3H into UDP-Xyl and for this to enter the Golgi lumen). (The two 3H-polysaccharides can readily be distinguished by digesting them with Driselase: the α-[3H]Xyl residues of xyloglucan end up in a disaccharide, isoprimeverose, since Driselase lacks α-xylosidase activity, whereas the β-[3H]Xyl residues of xylan end up in free xylose and xylobiose, separable from isoprimeverose by paper chromatography.) However, [Xyl-3H]xylans took a little longer than [Xyl-3H]xyloglucans to begin being secreted (i.e. appearing in the apoplast) (Kerr & Fry, 2003). This indicates that the α-xylosyltransferase involved in incorporating Xyl residues into xyloglucan acts further downstream in the Golgi system than the β-xylosyltransferase involved in incorporating Xyl residues into xylan.
This in-vivo conclusion agrees with the in-vitro finding that xylan β-xylosyltransferase activity occurs predominantly in low- and medium-density Golgi membrane fractions (interpreted as cis and medial cisternae, respectively) while xyloglucan α-xylosyltransferase activity is detectable principally in the high-density fractions (probably trans Golgi cisternae) (Baydoun & Brett, 1997). Surprisingly, xyloglucan fucosyltransferase activity (which can act only after xyloglucan α-xylosyltransferase has acted) was found to be greater in the low-density than the high-density membranes (Baydoun et al., 2001), an observation that emphasizes the point that enzyme activity does not always correctly predict enzyme action. As an explanation for this apparent anomaly, it is possible that the fucosyltransferase activity is found in all Golgi cisternae (though somewhat more abundantly in the cis cisternae) but that, in vivo, it acts only in the trans cisternae because its acceptor substrate (nonfucosylated xyloglucan) is only found there. Indeed, Brummell et al. (1990) reported the presence of xyloglucan α-xylosyltransferase activity (assayed in vitro) throughout the range of Golgi membrane densities; again, this does not prove the distribution of enzyme action in vivo.
Immunocytochemistry provides an alternative approach to study polysaccharide biosynthesis. Zhang & Staehelin (1992) showed that the product of xyloglucan α-xylosyltransferase (i.e. the Xyl/Glc-rich backbone of xyloglucan) is concentrated in the trans Golgi cisternae. However, this type of observation indicates the site of accumulation of an enzymic product and does not necessarily indicate where the enzyme acts: since c. 6% of the total xyloglucan backbone was observed in the cis and medial cisternae, it is possible, from Zhang & Staehelin's (1992) data that the xyloglucan backbones are synthesized in the cis and medial cisternae but then quickly carried into the trans cisternae, where they encounter a bottleneck and therefore build up to a high concentration. In conclusion, it is emphasized that in-vivo radiolabelling methods can provide access to information on the distribution of enzyme action, unlike studies on the distribution of enzyme activity or of reaction products.
Polysaccharide size changes before secretion In-vivo radiolabelling can also monitor qualitative changes, for example in Mr, of polysaccharides as they progress from their site of synthesis towards the plasma membrane and ultimate secretion. For example, in cultured maize cells, newly synthesized xyloglucans had Mr ≈ 300 000, whereas about 15 min later (but while still intraprotoplasmic), they had Mr ≈ 2000 000 (Kerr & Fry, 2003). It is to be expected that the earliest-labelled polysaccharides will have a lower mean Mr than the corresponding finished chains simply because the former were nascent – still elongating – when labelled. However, it can be shown that the lowest possible average Mr that can be expected on this basis is 50% of the mature size (Kerr & Fry, 2003). Since the [3H]xyloglucan actually increased in Mr by c. sevenfold before secretion, the data indicate that some form of molecular grafting occurs within the protoplast: the challenge now exists to explain what may be the chemical nature of this grafting process.
Addition of nonsugar moieties to nascent polysaccharides Other qualitative changes that often befall polysaccharides intraprotoplasmically include the addition of ester and ether groups. The PCW polysaccharides contain not only sugar residues but also certain noncarbohydrate ‘decorations’, especially methyl, acetyl and feruloyl esters (Fig. 1), and methyl ethers (Brett & Waldron, 1996; Fry, 2000; Perrone et al., 2002). These decorations are added to matrix polysaccharides during, or very soon after, the biosynthesis of the latter within the Golgi bodies. Feruloyl ester groups are of particular interest because they can undergo oxidative phenolic coupling and may thus crosslink the polysaccharide chains to which they are attached (see Primary wall composition and cross-links earlier in this section).
Acyltransferase activities have been demonstrated that can transfer acetyl groups from acetyl-CoA to polysaccharides (Pauly & Scheller, 2000). By contrast, feruloyl-CoA:polysaccharide feruloyltransferase activity has not yet been conclusively demonstrated in vitro (Meyer et al., 1991), although recent work demonstrates an acyltransferase activity that can transfer feruloyl groups from feruloyl-CoA to a trisaccharide model of arabinoxylan (Yoshida-Shimokawa et al., 2001). Methyl esters of pectic GalA residues are not produced by acyltransferases but by methyltransferases, using S-adenosyl methionine (SAM) as donor substrate (Kauss et al., 1967). These observations relate to enzyme activities rather than enzyme action. In-vivo labelling studies would potentially test the action of these enzymes in vivo (e.g. if the acetyl or methyl group were carefully traced in a time-course).
Conclusion It seems clear that our knowledge of wall polysaccharide biosynthesis is fragmentary and lacking in detail in many respects. Thus, although apparently relevant enzyme activities can be detected in vitro, and progress is now being made towards the identification and characterization of the genes specifying polysaccharide synthases, much work remains to be done on the action of the products of these genes — i.e. on the timing, location and molecular intricacies of wall polysaccharide biosynthesis.
II. Reactions that have been proposed to occur in primary cell walls
1. Enzyme-catalysed reactions
Types of enzyme involved Numerous enzyme activities occur in the PCW. The main examples are listed in Table 1. The present article does not discuss in detail the evidence that these enzymes are located in the PCW, but this point is well established in many cases and has been reviewed (Fry, 1995). Also, the present article focuses on enzymes that appear likely to act on structural wall components rather than on low-Mr solutes (e.g. sucrose, raffinose or putrescine) present in the apoplast. Some of the enzymes listed (e.g. a β-glucosidase that is able to digest β-(1→3)-d-glucan; Kim et al., 2000), are located on the outer surface of the plasma membrane rather than in the wall itself, and may thus have preferential access to polysaccharides on the inner face of the wall.
Table 1. Information on primary cell wall enzymes, with special reference to evidence for their action in vivo
PCW, primary cell wall; MLG, mixed-linkage β-(1→3),(1→4)-d-glucan; XET, xyloglucan endotransglucosylase; XTH, xyloglucan endotransglucosylase/hydrolase. Note that most published reports deal with the accumulation of their mRNAs, the assay of the enzyme activities (under optimized conditions in vitro), and the detection of the enzymes’ immonologically recognizable epitopes. *For hydrolases, substrate in which a bond is cleaved; for transferases, donor substrate; for oxidoreductases, electron donor. ¶For transferases, acceptor substrate; for oxidoreductases, electron acceptor.
Most PCW enzymes are hydrolases, which cleave the primary substrate and require only H2O as the second substrate (Table 1). In addition, there are oxidoreductases, most of them oxidases or peroxidases that use O2 or H2O2, respectively, as electron acceptor. The O2 required by oxidases is certainly plentiful in the apoplast in vivo; H2O2 is often also present (Frahry & Schopfer, 1998; Ros Barcelo, 2000; Lin & Kao, 2001; Kärkönen et al., 2002; Lin & Kao, 2002). One lyase, pectate lyase, appears to be present in some walls, although the evidence for this rests largely on the observation that putative pectate lyase genes are transcribed in plant cells; there are few reports of pectate lyase activity in plants (Domingo et al., 1998; Marin-Rodriguez et al., 2003; Payasi & Sanwal, 2003). Recently, a transferase activity (xyloglucan endotransglucosylase; XET) that breaks and remakes glycosidic bonds in the backbone of xyloglucan has been discovered in PCWs. Since XET activity potentially plays a very significant role in modifying the PCW in vivo, akin to the role of the penicillin-sensitive transpeptidase in bacterial walls, it will be considered in detail. Other transferases (acting on different glycosidic bonds, or on ester or amide bonds) possibly await discovery.
Observations relating to the occurrence and functioning of PCW enzymes have been made by various experimental approaches, which are summarized in the remainder of this section, more or less in order of increasingly strong evidence that the enzyme is present in the plant and can act.
Gene sequences The genome of Arabidopsis includes 730 open reading frames that encode putative glycosyltransferases and glycosylhydrolases, 170 of which are putative pectin-hydrolysing enzymes (Henrissat et al., 2001). Although this suggests a great importance for such enzymes, it remains to be shown which of them are transcribed and translated, and which of the translation products possess the enzymic activities claimed. A putative α-l-fucosidase gene from pea was shown to encode a protein that lacked fucosidase activity (Tarrago et al., 2003), emphasising the need for caution in ascribing enzymic activities to proteins known only from their gene sequences.
mRNA accumulation It has become relatively simple to demonstrate that a given gene, identified in silico as being likely to encode an enzyme of interest, is transcribed in a particular cell and at a particular time. However, without further data, such information is not satisfactory since it remains possible that the mRNA detected is not translated to produce the corresponding protein. If a protein is produced, it might not undergo any necessary post-translational modifications and if it does undergo these, it might not possess the enzymic activity claimed for it on the basis of its sequence; if it does possess the claimed activity, the protein might not be transported into the wall, and if it is transported there, it might not act in vivo if it fails to come into contact with its substrate(s) or if inhibitors are present and/or activators absent.
An interesting example of post-transcriptional regulation is provided by a ripening-related cellulase in Capsicum annuum: the gene is transcribed to give two distinct populations of mRNA, sized 1.7 kb and 2.1 kb, respectively. Blocking the formation of the 1.7-kb transcript alone abolished formation of the enzyme, implying that the 2.1-kb transcript is not translated (Harpster et al., 2002a).
Protein sequences detected Some workers have exploited the ‘proteomics’ approach of screening for the proteins present in the cell wall (or secreted into the medium) and identifying these by partial (N-terminal) sequencing. This approach bypasses the uncertainty of whether the mRNA is transcribed, but the protein's enzyme activity and in-vivo action remain to be established. A protein secreted by clover root cells in response to phosphate starvation was supposed from its amino acid sequence to be an α-fucosidase (Hay et al., 1998), although this activity now appears questionable (Tarrago et al., 2003).
Immunologically detected proteins The immunocytochemical approach (Table 1) overcomes some of these objections. A labelled antibody can be used to detect the enzyme within the PCW (Nematollah & Roux, 1999). It may even be possible to obtain an antibody that demonstrates that the appropriate post-translational modifications have occurred (although this has rarely been confirmed). However, other gaps remain in the evidence obtained through immunocytochemistry: the protein bearing the epitope which the antibody recognizes may not possess the claimed enzymic activity; and if it does, it may not act in vivo to its full potential (exemplified by the lack of correlation between β-glucanase protein levels and β-glucan hydrolysis in auxin-treated maize coleoptiles (Inouhe & Nevins, 1998)). In addition, a new gap arises: the antibody may not be completely specific for the protein of interest, especially if the antibody recognizes a carbohydrate side-chain of the enzyme. A monoclonal antibody against a maize coleoptile β-glucosidase also, unexpectedly, recognized a peroxidase (Nematollah & Roux, 1999).
Enzyme activities assayed after extraction Evidence for the presence of active enzymes in the PCW can be obtained by assay of the enzyme activities in isolated PCWs or in apoplastic fluid. Sometimes, the enzyme can be solubilized from the isolated walls, but in case this is unsuccessful it may be more satisfactory to assay the enzyme activity in situ in the isolated wall fragments. An alternative strategy for demonstrating the existence of wall enzymes is to isolate protoplasts (i.e. remove the wall) and attempt to show that the enzymic activity of interest is thereby lost. However, both methods have pitfalls: it is impossible to isolate absolutely pure walls or to prevent all postmortem wall-binding of protoplasmic enzymes, and the enzymes used for protoplast isolation may be taken up into the protoplasts, giving inflated values of, for example, cellulase activity (J. E. Thompson & S. C. Fry, unpubl. data).
No in-vitro assay of enzyme activity can show that the enzyme really acts within the wall in vivo. Nevertheless, this approach completely overcomes the objection that the protein under investigation might lack the enzyme activity of interest. Quantitative studies of enzyme activity in PCWs, sampled, for example, at various stages of development, can suggest possible changes in enzyme action. In addition, such in-vitro studies can help to define the mechanism and substrate-specificity of the reaction(s) catalysed by the enzyme. The approach can also provide valuable information on factors likely to control the action of the enzyme in vivo: for example, its pH optimum and the influence of various inorganic ions and organic metabolites.
Enzymological activity does not necessarily equate with enzyme action in vivo— for example:
• the enzyme may be spatially separated from its ‘primary’ substrate (e.g. enzyme in secondary wall, substrate in primary wall; Bourquin et al., 2002).
• the enzyme, although possessing the ability to attack a particular chemical moiety (e.g. to hydrolyse an α-xylosyl group off a chosen substrate tested in vitro) might be unable to recognize the ‘same’ moiety within a PCW polysaccharide structure. Thus, a PCW α-xylosidase might attack the repeat heptasaccharide of xyloglucan (XXXG) but have no discernible effect on high-Mr xyloglucan (Koyama et al., 1983; Edwards et al., 1985). Several other examples have been documented of PCW enzymes that appear to be designed for digesting chemical moieties of PCW-located polymers but fail to attack them in vitro (Dopico et al., 1991; Konno et al., 1994, 1996, 2002). Isoenzymes can differ in this respect: for example, of two β-galactosidases isolated from avocado fruit, isozyme III was able to release galactose from wall polysaccharides in in-vitro assays, whereas isozyme I was not (Tateishi et al., 2001b). Also, the putative substrate can change in its susceptibility to one enzyme owing to the action of another: for example, endopolygalacturonase (Wakabayashi et al., 2003) and pectin acetylesterase (Bordenave et al., 1995) require pectic substrates with a sufficiently low degree of methylesterification; thus, it is possible that the actions of these two enzymes in the PCW in vivo are dependent on prior action in the cell wall of a third enzyme, pectin methylesterase.
• the local pH or ionic strength may not be optimal. The activity of enzymes is normally assayed in vitro using optimized conditions (e.g. of pH, ionic strength, etc.). However, in vivo, such conditions might not always prevail. For example, tomato endopolygalacturonase activity, assayed on isolated PCWs in vitro, is strongly dependent on K+ and pH and their ratio (Chun & Huber, 1998), and this could explain why the enzyme activity assayed in vitro (or pectin depolymerization occurring during the manufacture of tomato paste; Brummell & Labavitch, 1997) exceeds its action as observed in vivo (Chun & Huber, 1998). This suggests that in the walls of living cells, endopolygalacturonase action is restricted by factors present in (or absent from) the apoplast. Similarly, Ca2+ activates tomato seed exopolygalacturonase (Sitrit et al., 1999) and banana fruit pectate lyase (Payasi & Sanwal, 2003). Pectin methylesterase activity is strongly regulated by the ionic composition of the medium, probably mainly because of effects of the ions on the pectic substrate (Goldberg et al., 1992a). Xyloglucan endotransglucosylase activity assayed in vitro is critically dependent on a minimum ionic strength (T. Takeda & S. C. Fry, unpubl. data), which might not always be reached in the apoplast in vivo.
• specific inhibitors or promoters may be present. An enzyme's action may be dependent on the presence of specific promoters such as the 230-kDa protein that activates Cuscuta cellulase (Chatterjee & Sanwal, 1999) or the low-Mr peptide that activates Cuscuta endopolygalacturonase (Bar Nun et al., 1999).
There are cases where an enzyme action, observed in vivo, cannot be ascribed to any enzyme activity detectable in vitro. For example, sucrose treatment, which retards the senescence-related loss of galactose residues from the PCW polysaccharides in petals of cut flowers, had no effect on the petals’ content of extractable β-galactosidase activity (O’Donoghue et al., 2002). Similarly, ethylene-treatment of watermelon fruit causes extensive depolymerization of pectic polysaccharides, while having little consistent effect on the content of extractable endo-polygalacturonase activity, as assayed in vitro (Karakurt & Huber, 2002). As a further example, the use of a truncated sense-gene suppression method to block the production of a ripening-related cellulase in Capsicum annuum was able to prevent ‘completely’ the production of the enzyme but did not prevent the ripening-related depolymerization of hemicelluloses in the living fruit (Harpster et al., 2002a). It must therefore be concluded that this particular enzyme activity was not responsible for the biologically relevant action – hemicellulose cleavage –in vivo. Furthermore, expression of this Capsicum gene in tomato, resulting in a 20-fold increase in cellulase activity, had little or no effect on xyloglucan cleavage in vivo or on fruit softening (Harpster et al., 2002b). The reasons for these apparent mismatches between enzyme action and enzyme activity remain to be determined. The action observed in vivo could be nonenzymic or, in some cases, an essential cofactor might be absent from the in-vitro assay mixtures.
Studies on XET activity in vitro Xyloglucan endotransglucosylase is taken here as an example of a significant wall enzymic activity that can usefully be studied in vitro. Xyloglucan endotransglucosylase activity is one of the two enzymic activities exhibited by a group of proteins known as xyloglucan endotransglucosylase/hydrolases (XTHs). Most XTHs exhibit the XET activity, some also exhibit the corresponding hydrolase activity (XEH), and a few exhibit only XEH activity (Rose et al., 2002). Although XET activity was first detected in vivo (Baydoun & Fry, 1989; Smith & Fry, 1991), it was first rigorously characterized in terms of the reaction catalysed and its substrate-specificity during studies of the extracted enzyme. Crude extracts from essentially any growing plant tissue were found to catalyse transglycosylation when mixed in vitro with xyloglucan (polysaccharide) plus a suitable tritium-labelled, xyloglucan-derived oligosaccharide (3H-XGO): segments of various lengths of the polysaccharide became glycosidically linked to the 4-position of the Glc residue at the nonreducing end of the 3H-XGO (Fig. 2) (Fry et al., 1992; Nishitani & Tominaga, 1992). This reaction is easy to detect because it results in some of the 3H, initially 100% present in the XGO (Mr ≈ 103), being incorporated into a much larger reaction-product (Mr ≈ 104−105), which is easily distinguished from the XGO by dialysis, ethanol precipitation, gel-permeation chromatography or, most simply, by the ability of the high-Mr product to remain bound to filter paper when washed in running water. In a similar vein, it was demonstrated by gel-permeation chromatography that the enzyme (originally misidentified as a ‘cellulase’ acting in transglycosylation rather than hydrolytic mode) could transfer a segment from a moderate-Mr xyloglucan to a high-Mr xyloglucan (McDougall & Fry, 1990) (Fig. 3). More challenging to demonstrate was the ability of the enzyme to transfer a segment from one high-Mr xyloglucan chain to another essentially identical one. This, however, is possible by use of density labelling: one population of substrate molecules was labelled with ‘heavy’ isotopes (e.g. 13C and/or 2H, at high isotopic abundances) and with a radio-isotope (3H) whereas the second population (although potentially identical chemically) was unlabelled; transglycosylation resulted in some of the 3H being transferred from the ‘heavy’ population to a new, intermediate-density population, as shown by a shift in the buoyant density estimated by isopycnic centrifugation in a density gradient of caesium trifluoroacetate (Thompson et al., 1997).
The new bond formed during XET action does not involve the reducing terminal [3H]glucose moiety of the XGO, as shown by the observation that this group retained its characteristic properties (e.g. it could still be converted to [3H]glucitol upon treatment with NaBH4) after the enzymic reaction (Smith & Fry, 1991; Fry et al., 1992). The radiolabelled products were, on average, smaller than the substrate polysaccharide used, indicating that bond cleavage was occurring (Nishitani & Tominaga, 1992; Steele et al., 2001). Bond cleavage associated with the formation of a new (alkali-stable) bond is diagnostic of a transglycosylation reaction: a glycosidic bond was broken, and a new one formed. Nuclear magnetic resonance analysis suggested that the bond formed was not chemically different from the bond cleaved (Nishitani & Tominaga, 1992), a conclusion supported by the observation that the new bond formed by XET could be broken by treatment with endo-β-(1→4)-glucanase (Baydoun & Fry, 1989; Smith & Fry, 1991). Thus, this intriguing enzyme activity is able to catalyse a group-transfer reaction in which the products may, in theory, be identical with the substrates (Fig. 4b).
One proposed role for XET is the ‘grafting’ of newly secreted xyloglucan chains into the existing wall matrix. For this to occur with minimum wastage of material, the leaving group (Fig. 4a) should be as short as possible (i.e. the cleavage site should be as near as possible to the reducing terminus of the donor substrate). The position of the cleavage site can be investigated by in-vitro enzyme assays: the length of the polysaccharide segment that becomes attached to the 3H-XGO (Fig. 2) indicates the distance along the polysaccharide chain (measured from its nonreducing terminus) that it was cleaved by the enzyme. This distance was found to be essentially random (Steele et al., 2001), an observation which argues against the hypothesis that XET chooses cleavage sites a specific distance from the end of its donor substrate chain.
In-vitro assays have also indicated that XETs form a remarkably long-lived complex with a portion of the (cleaved) donor substrate (Sulováet al., 1998) (Fig. 4a). Chemical evidence suggested that the polysaccharide–enzyme linkage was a glucosyl ester bond, probably involving the –COOH group of an Asp residue at the enzyme's active site. This ability of the enzyme to remain covalently attached to its substrate for some time (minutes to hours) formed the basis of generally applicable purification methods for proteins with XET activity (Steele & Fry, 1999; Sulová & Farkas, 1999). It also provides a perhaps surprising vision of XET cleaving a donor substrate in the cell wall and then spending some considerable time ‘seeking’ a suitable acceptor substrate elsewhere in the wall. The longevity of the substrate–enzyme complex may be important in the enzyme's functioning because the acceptor substrate being sought is a scarce commodity — i.e. the single nonreducing terminus of a long polysaccharide chain.
The assay using 3H-XGOs is very convenient for defining the donor and acceptor substrate specificities of XETs. In studies of the donor substrate specificity, a standard acceptor substrate is employed (e.g. [3H]XLLGol) and the nature of the polysaccharide offered as donor substrate is varied: xyloglucans were the only polysaccharides to act as donor substrates, and they worked almost equally whether fucosylated or not (Fry et al., 1992; Purugganan et al., 1997). For studies of the acceptor substrate specificity, tamarind xyloglucan (a readily available standard) can be employed as donor and a low concentration of a standard 3H-XGO (e.g. [3H]XLLGol) as acceptor; other XGOs (not radiolabelled) can then be added at a higher concentration, and those that act as acceptor substrates will compete with the 3H-XGO. This approach has shown that the minimum structural requirement for a good acceptor substrate is Xyl2·Glc3 (Fig. 5) (Lorences & Fry, 1993). These, and similar in-vitro assays have identified differences in pH optima, temperature optima, substrate specificity and Km, of various XET isoforms (Campbell & Braam, 1999; Steele & Fry, 2000).
Similar assays can be used to search for novel transglycosylases using substrates other than xyloglucan. Such work provided evidence against the existence of transglycosylases that act on pectic substrates such as galacturonans (García-Romera & Fry, 1994). There is evidence, however, for an exotransglucosylase, which can transfer nonreducing terminal glucose residues from one oligosaccharide chain to another (Crombie et al., 1998), and the possibility of pectin methylesterase acting as an acyltransferase has been raised (Gelineo-Albersheim et al., 2001).
Cytochemically detected enzyme activities The spatial distribution of active (though not necessarily acting, on endogenous substrates) enzymes can be mapped by cytochemical enzyme assays in which the tissue is infiltrated with artificial substrate(s) that when acted on by the enzyme of interest generate a product that is detectable by microscopy. The product may be coloured, fluorescent or electron-opaque; preferably it should be water-insoluble so that it does not diffuse far from its site of formation. For example, peroxidases can be localized by infiltration of tissues with H2O2 plus a chromogenic electron donor such as tetramethylbenzidine or syringaldazine, which react to generate water-insoluble coloured products. Many hydrolases can be detected by suitable chromogenic substrates which after hydrolysis release a coloured, immobile product: for example, the colourless substrate ‘X-Gal’ (5-bromo-4-chloro-3-indoxyl-β-d-galactopyranoside) when hydrolysed by β-galactosidase yields an immobile blue product (Biswas et al., 2003). Such assays can be made semiquantitative if there is a suitable method of measuring the yield of coloured product. It should be borne in mind, however, that the cell wall might well contain endogenous, nonchromogenic, substrates for the enzyme of interest, and that such substrates will compete with the exogenous chromogenic substrate, compromising any attempt at quantification of the enzyme activity.
As with enzyme assays (see the previous section), the cytochemical approach can give clear evidence that the gene has been transcribed, the mRNA has been translated, and the protein has undergone any necessary post-translational modifications (such as are required by XETs (Campbell & Braam, 1998) and endo-xylanase (Caspers et al., 2001) and are observed in the case of pectin methylesterase (Gaffe et al., 1997)), possesses the enzymic activity of interest and has been secreted into the cell wall. The cytochemical approach has the advantage over conventional enzyme assays of providing good spatial resolution of the cellular and even subcellular distribution of the enzyme; however, cytochemistry gives poorer quantification of the enzyme activity and little information on the enzyme's specificity for natural (nonchromogenic) substrates.
2. Potential nonenzymic reactions in the cell wall
Active oxygen species Not all polysaccharide cleavage reactions need be protein-mediated. It has been proposed recently that wall polysaccharides may be subject in vivo to nonenzymic scission mediated by hydroxyl radicals (•OH). Wall polysaccharides are readily cleaved by •OH radicals in vitro (Fry, 1998), and •OH can solubilize pectic polysaccharides from tomato fruit cell walls in vitro (Dumville & Fry, 2003). •OH can be generated by Fenton reactions, which require H2O2 and a reduced transition metal ion (Cu+ being 60 times more effective than Fe2+),
Cu+ + H2O2 → Cu2+ + •OH + OH−
both of which are thought to occur in the cell wall. Although •OH is exceedingly reactive and can cause mutation, membrane damage and protein denaturation, its production could be controlled (e.g. by the correct siting of Cu atoms relative to wall polysaccharide molecules and by control of the formation of reductants necessary to convert Cu2+ to Cu+). Once produced at a particular site within the cell wall, an •OH radical would rarely diffuse more than 1 nm before reacting with some organic substance (cf. thickness of a typical PCW ≈ 100 nm). Thus, if produced at the right time and right place within the wall, •OH could be very precisely targeted to cause polysaccharide scission (Schopfer, 2001; Fry et al., 2002).
An interesting and possibly related hypothesis is that apoplastic active oxygen species can attack cell wall proteins, modifying them in such a way that they are earmarked for preferential hydrolysis by apoplastic proteinases (Gomez et al., 1995).
Ca2+ chelation Calcium ions help to cross-link the nonesterified GalA residues of pectin chains in the wall and middle lamella (Jarvis, 1982). Chelation of this Ca2+, by naturally occurring chelators such as citrate or oxalate, could contribute to pectin solubilization (Brady, 1987). The citrate content of tomato fruit (30 mmol kg−1 f. wt) exceeds the total Ca2+ content (approx. 3.5 mmol kg−1 f. wt) (Stevens & Scott, 1988); and some oxalate is also present (MacDougall et al., 1995; Islam et al., 1996). Nonesterified pectic GalA residues, potentially bound to Ca2+, are present in unripe tomato pericarp at approx. 3 mmol kg−1 f. wt (MacDougall et al., 1996). Thus, citrate is present in quantities potentially able to displace all pectin-bound Ca2+. Citrate and malate concentrations increase during ripening (Dalal et al., 1966; Islam et al., 1996), compatible with a role in softening. Such a role is supported by the observation that pectin solubilization in disks of unripe tomato pericarp is inhibited by added Ca2+ (Mingani et al., 1995). Throughout normal ripening, the concentration of free Ca2+ in the apoplast is a steady 4–5 mm, which is enough to restrict pectin solubilization (Almeida & Huber, 1999). This suggests that natural chelators have the potential to promote the solubilization of a proportion of the pectin, if these chelators are apoplastic. It has been reported that a proportion of them are indeed apoplastic (MacDougall et al., 1995; Ruan et al., 1996). The possible role of Ca2+ chelators in modifying cell wall properties deserves careful investigation.
Borate cross-linking Borate, an essential element for land plants, at least above the bryophytes, has been shown to form diester bonds with specific apiose residues in the pectic polysaccharide, rhamnogalacturonan-II (O’Neill et al., 1996). Such bonds can be formed in vitro but, for their efficient formation, certain heavy metals such as Pb2+ must be added. It seems certain that Pb2+ does not have this role in vivo, and it is currently unknown how borate diester linkages are formed in the living plant cell. It is also unclear whether such linkages are formed before or after secretion of the polysaccharide.
III. Tracking the careers of wall components in vivo: evidence for action of enzymes in the walls of living plant cells
Much evidence shows that cell walls contain enzymes, and their specificities suggest that they could act on apoplastic substrates in vivo. But do they? It is important to emphasize the distinction between enzyme activity (usually measured in katals, in vitro, under optimized conditions) and enzyme action (observed, in vivo, in the ‘rough and tumble’ of a living cell).
Detecting enzyme action in vivo may be technically difficult, especially with enzymes (such as XET) that catalyse freely reversible reactions. However, it is an important goal if we are more interested in the biological function of an enzyme than in its mere presence or its physicochemical properties. Evidence that an enzyme actually acts in the walls of living plant cells, and that the wall is therefore not just a ‘dump’ for redundant enzymes, can be sought from several sources, some of which are reviewed in this section.
Certain cell types exhibit visible wall lysis during the normal course of development — e.g. the formation of perforation plates in xylem vessel elements and of sieve pores in sieve tube members and the fusion of cells e.g. in the development of laticifers. In addition, most cell types exhibit localized wall lysis during cell division (Jeffree et al., 1986). Also, extensive cleavage of hemicelluloses in vivo during rapid growth can be monitored in the outer epidermal PCWs of pea stems by interference microscopy (Bret-Harte & Talbott, 1993). Such lysis seems likely to entail the chemical cleavage of structural wall components, although the enzymes or other agents responsible are not easy to determine without more detailed chemical analysis.
3. In-muro synthesis of immobile products
Lignin biosynthesis One source of evidence for the in-vivo action of wall-located enzymes is the finding, in the wall, of immobile products of enzyme action. Lignin, a polymer of monolignols (coniferyl, p-coumaryl and sinapyl alcohols), is a good example. Lignin is a product of the action of enzymes (probably mainly peroxidases; possibly also laccases),
so the occurrence of lignin indicates that the enzymes responsible are active in vivo. Furthermore, lignin (unlike its precursors) is water-insoluble and thus immobile; therefore, lignin must be made where it is found (in the wall). In fact, lignin is initially formed in the middle lamella and PCW of cells such as xylem vessel elements and phloem fibres, which also have thick secondary walls (Wardrop, 1971). The initial site of lignification may thus be approx. 10 m distant (the thickness of the secondary wall) from the nearest living protoplasm. This therefore provides a clear example of a wall-localized enzyme acting in vivo.
4. Changes in the chemistry of previously deposited wall components
Another line of evidence for the in-vivo action of wall-located enzymes is the detection of an enzyme-catalysed chemical change, with time, in some previously deposited PCW component.
Turnover of noncellulosic, wall-bound glucans An excellent demonstration of the in-vivo modification of a previously deposited wall component is provided by Franz's (1972) evidence for turnover (simultaneous synthesis and degradation of essentially identical polymer molecules) of a noncellulosic PCW β-glucan, probably callose. Franz ink-marked a standardized zone in the hypocotyls of several identical mung bean seedlings, and then injected [14C]glucose into the marked zones and at intervals analysed the wall-bound 14C-polysaccharides in the marked zone. Rapid accumulation of [14C]glucan occurred on the day of injection, after which time little free [14C]glucose remained. During the following 2 wk, however, a proportion of the [14C]glucan disappeared from the marked zone. If the (intact) [14C]glucan is assumed to be immobile within the tissue (a reasonable assumption for a wall polysaccharide), the only tenable explanation for the disappearance of previously wall-bound [14C]glucan molecules is their degradation to soluble sugars. This methodology may underestimate the glucan degradation if some of the breakdown product ([14C]glucose) is immediately reincorporated into new 14C-polysaccharides that are indistinguishable from the old. The biological role of the degradation is unknown; possibilities include removal of surplus callose initially deposited as a wound-response. The β-glucanase presumably involved in the [14C]glucan degradation in mung bean seedlings has not been characterized, and theoretically the possibility remains that the degradation was nonenzymic (e.g. through hydroxyl radical attack). Evidence for hydrolysis rather than •OH-mediated oxidative scission would be provided by identification of the newly formed reducing termini (characteristic only of hydrolysis), although this has apparently not been checked. Nevertheless, Franz's (1972) data clearly demonstrate turnover of a wall polysaccharide in vivo.
Several other wall polysaccharides have been shown by similar methods to exhibit turnover in vivo. These include xyloglucan, especially in auxin- or acid-treated dicot tissues (Labavitch & Ray, 1974; Nishitani & Masuda, 1982), and MLG, especially in auxin-treated cereal coleoptiles (Nevins, 1975). Among the clearest evidence for this statement is that which comes from the demonstration that, in auxin-treated pea stem segments, previously radiolabelled, wall-bound xyloglucans decrease in amount or in Mr (Labavitch & Ray, 1974) and increase in extractability from the wall (Terry et al., 1981). Trimming and solubilization of previously wall-bound [3H]xyloglucan also occurs in rapidly growing rose cell suspension cultures (Thompson & Fry, 1997).
The role usually proposed for turnover of xyloglucan and MLG is wall loosening and hence cell expansion. In support of this hypothesis, in dicots, both auxin-induced elongation and auxin-induced xyloglucan cleavage are inhibited by lectins and antibodies that bind to xyloglucans and may thereby shield them from enzymic attack (Hoson & Masuda, 1991; Hoson et al., 1991). Such experiments do not indicate whether the xyloglucan degradation is catalysed by cellulases or by XTHs (XET or XEH activity, or both), or is caused by •OH attack, any of which could, theoretically, be blocked by a foreign protein that coats the xyloglucan. In gramineous monocots (maize coleoptiles), application of antibodies against exoglucanases and endoglucanases can inhibit auxin-induced elongation, supporting a biological role for these wall enzymes (Inouhe & Nevins, 1991). However, evidence not easily explicable in terms of a causal role for MLG degradation in auxin-induced elongation has also been presented (Miyamoto & Schopfer, 1997).
Net degradation of wall polysaccharides Radiolabelling methods can enable the demonstration of turnover (i.e. the degradation of polysaccharides simultaneously with (possibly greater rates of) polysaccharide synthesis). In some situations, polysaccharide degradation exceeds continued polysaccharide synthesis (if any) and radiolabelling is therefore not necessary for demonstrating degradation. It is sufficient to analyse equivalent samples of the tissue, collected at different ages, and to demonstrate a loss of a wall component. For example, in maturing Phaseolus vulgaris pods, levels of pre-existing galactose-rich pectins decreased with time (Stolle-Smits et al., 1999), indicating a degradation, possibly enzymic, of these polysaccharides. A similar approach appeared to demonstrate breakdown of some of the cellulose during grape berry softening, demonstrating cellulose attack, possibly by the action of cellulase; and simultaneous partial cleavage (decrease in Mr) of xyloglucan (Yakushiji et al., 2001), although the enzyme, if any was involved in this case, cannot be specified (XTH or cellulase).
This approach will underestimate polysaccharide degradation if there is concurrent de-novo synthesis.
Net loss of a specific side-chain of wall polysaccharides The repeat units of xyloglucan may contain Gal residue(s). As pea seedling tissues mature, with little further de-novo synthesis of xyloglucan, the proportion of the repeat units, in the preformed xyloglucan, that possess a Gal residue has been observed to decrease (Pauly et al., 2001). This provides evidence for the in-vivo action of a wall-located β-galactosidase.
Pectin methylesterase action Another possible example of evidence for wall-localized metabolism of a polysaccharide in vivo is provided by studies on pectin methylesterase action. In young tissue (sampled at time t1 after germination or fruit initiation, etc.), pectin is heavily methyl-esterified (carries –COOCH3 groups) and thus relatively neutral; in the same cells at a later stage of development (t2), a higher proportion is in the ionizable free acid form (–COO−) (Goldberg et al., 1992a; Barnavon et al., 2001). This compositional change is physiologically important because it increases the susceptibility of pectin to being crosslinked via Ca2+-bridges (Yamaoka et al., 1983) and may thus restrict cell expansion. It also increases the susceptibility of the pectin to hydrolysis by other enzymes (Bordenave et al., 1995; Wakabayashi et al., 2003). The observed decrease in the degree of esterification between t1 and t2 suggests the action of a wall-bound pectin methylesterase. However, this evidence rests on the assumption that there was no appreciable de-novo synthesis of low-methyl pectin between t1 and t2 (i.e. that the equivalent population of pectin molecules was being analysed at both times). This assumption is difficult to establish, especially if the possibility is considered that a constant pectin content per cell is achieved by balanced synthesis and degradation (turnover). Therefore, this proposed evidence for changes in the chemistry of previously deposited wall components is equivocal. A pulse–chase radiolabelling approach, in which methyl-esterified [14C]GalA residues are distinguished from nonesterified ones (Kim & Carpita, 1992), would potentially provide stronger evidence for in-vivo de-esterification of pectic polysaccharides.
Addition of new feruloyl groups to existing wall-bound xylans? Another example of the need for great caution in interpreting changes in wall composition as evidence for the occurrence of chemical reactions within the wall is provided by data on the ferulate:xylan ratio in gramineous cell walls. The cell walls of 2-d-old maize coleoptiles have c. 20 feruloyl groups per 1000 Xyl residues, whereas those of 10-d-old coleoptiles have > 100 per 1000 Xyl residues (Nishitani & Nevins, 1990). These observations might superficially suggest that pre-existing wall-bound xylans get new feruloyl groups attached to them; however, the increasing ferulate:xylan ratio could equally represent the de-novo synthesis, in older coleoptiles, of new xylan molecules with a higher ferulate:xylan ratio, thus diluting the previously deposited, low-feruloyl xylans. In-vivo radiolabelling of the sugar residues in cultured spinach cells did not provide any evidence for the feruloylation of Ara residues once these had become part of a wall-bound polysaccharide chain (Fry, 1987).
5. Xyloglucan endotransglucosylase/hydrolase action observed in vivo
Methods to observe XTH action in vivo In view of the proposed ‘tethering’ role of xyloglucan chains in wall architecture (Fry, 1989; Hayashi, 1989) and the ability of XET activity, at least in vitro, to cut and rejoin xyloglucan chains, this enzyme activity could, in theory, play a role in reversibly loosening the cell wall, thus permitting controlled cell expansion. In favour of this hypothesis, extractable XET activity is often correlated with growth rate (Fry et al., 1992; Hetherington & Fry, 1993; Pritchard et al., 1993; Potter & Fry, 1994; Xu et al., 1995; Palmer & Davies, 1996; Antosiewicz et al., 1997; Cataláet al., 1997). It is therefore of special interest to trace qualitatively and quantitatively the reactions catalysed by XTHs in the walls of living plant cells.
Xyloglucan endotransglucosylases that are in the process of acting in vivo can be caught ‘red-handed’ by a simple but effective technique in which cell walls are treated with XGOs (Sulováet al., 2001). The polysaccharide–enzyme covalent complex, formed within the cell wall after an XET attacks a xyloglucan (donor substrate) but before it completes net transglycosylation (Fig. 4a), can readily be broken upon addition of a high concentration of a suitable low-Mr acceptor substrate (XGO). The latter will act as a readily available acceptor substrate, thwarting the enzyme finding a high-Mr acceptor substrate within the wall architecture, and will thereby cause the release the enzyme from the wall into solution, where it can be collected, freed of XGOs and assayed in vitro (Sulováet al., 2001). Any XET solubilized by this method is considered to have been ‘acting’in vivo.
An alternative approach is to detect XET action by ‘labelling’ the participating substrates in vivo. This, however, is a particular challenge because the substrates of XET are chemically indistinguishable from the products; therefore, standard pulse–chase radiolabelling approaches, such as reported by Franz (1972) for β-glucan turnover, are not suitable for detecting the action of XET in vivo. An alternative approach that has been successfully devised uses dual labelling of xyloglucan: label 1 need merely be detectable (e.g. a radiolabel, although in principle it could alternatively be a fluorescent group), whereas label 2 must form the basis of a separation method whereby the labelled population of xyloglucan molecules can be physically separated from the unlabelled population. The transfer, in either direction, of label 1 between the population with label 2 and the population without label 2 would provide unambiguous evidence for the cutting–rejoining action of XET. This dual-labelling approach is an in-vivo adaptation of the approach described in Section II.1. It should be emphasized that dual-labelling with two different radio-isotopes (3H and 14C), or with 3H and a fluorescent label, is not helpful because there is no practical way of telling whether the two labels are present in a single xyloglucan chain or in two identical chains. The labelling experiments critically depend on creating two physically separable populations of xyloglucan in vivo (with and without label 2).
In recent studies from this laboratory, using cultured ‘Paul's Scarlet’ rose (Rosa sp.) cells, label 1 was a radioisotope, 3H, fed in vivo as [1-3H]arabinose; and label 2 was one or more stable isotope(s), 13C or 13C + 2H together, supplied as [13C6,2H7]glucose. The 3H was used at a low isotopic abundance: it is detectable with great sensitivity (by scintillation counting) even if only a small percentage of the H atoms in the polysaccharide are 3H. By contrast, the 13C and 2H were used at high isotopic abundances (c. 99 atom%13C and c. 50 atom%2H; i.e. the glucose normally used as the culture's sole carbon source was entirely replaced by [13C6,2H7]glucose), so that their presence caused a clear-cut increase in the buoyant density of the polysaccharides, as observed during density-gradient ultracentrifugation in caesium trifluoroacetate (CsTFA) solutions (an approach inspired by that of Meselson & Stahl (1958) for studies of DNA replication). The reason for using 13C and 2H together instead of just one stable isotope was to maximize the density difference between the ‘heavy’ and ‘light’ polysaccharides. Caesium trifluoroacetate was preferred over CsCl because the former can produce solutions of higher density; CsCl solutions, although suitable for banding DNA, are generally not dense enough for polysaccharides or RNA. Isotopic labelling has the advantage over fluorescent labelling that in the former all substrates are endogenous and chemically unchanged by the label.
After 28 d of growth in [13C6,2H7]glucose, the cells had isoprimeverose units (Xyl–Glc, the unique disaccharide building-block of xyloglucan) in which the most common isotopic composition was for all 11 C atoms to be 13C, and for four of the 13 stable (carbon-linked) H atoms to be 2H, as shown by mass spectrometry (Thompson & Fry, 2001b). Thus, the isoprimeverose obtained had a mean Mr of c. 327 instead of 312. Isopycnic centrifugation in CsTFA gradients showed that the ‘heavy’ xyloglucan had a buoyant density of 1.6432 ± 0.0010 g ml−1, as opposed to 1.5846 ± 0.0006 g ml−1 for ‘light’ xyloglucan (Thompson & Fry, 2001a).
In various 13C/2H/3H pulse–chase experiments, XET action could be detected by physical changes (in buoyant density): radiolabel, incorporated into the xyloglucan during a defined, narrow, time interval was passed between high- and low-density chains, known to have been synthesized at different times depending on the design of the experiment. Such experiments have provided evidence for the in-vivo involvement of XET in two distinct, previously proposed, biological roles: (1) the integration of newly synthesized xyloglucan into the wall fabric, and (2) wall restructuring (transglycosylation between pairs of xyloglucan molecules, both of which were already wall-bound). These two roles, and the evidence for them, are described next.
Xyloglucan endotransglucosylase action in vivo: role in wall assembly The action of XET could, in theory, favour the integration of newly synthesized xyloglucans into the PCW, i.e. have a role in wall assembly (Xu et al., 1996; Nishitani, 1997). Such integration is necessary for continued cell expansion if excessive wall weakening is to be avoided.
Edelmann & Fry (1992b) showed that suspension-cultured rose cells quickly secrete newly synthesized [3H]xyloglucan into the wall, and then very rapidly (within approx. 1 min) strongly bind it there. This integration process was not blocked by up to 1 d of treatment of the cells with 2,6-dichlorobenzonitrile (DCB), an inhibitor of cellulose synthesis (Edelmann & Fry, 1992c). Therefore, hydrogen-bonding to newly generated cellulosic microfibrils cannot be the sole mechanism of xyloglucan's wall-binding, although more prolonged treatment with DCB did abolish the ability of cells to bind newly secreted xyloglucan (Shedletzky et al., 1990).
The discovery of XET activity suggested an interesting new means (molecular cutting–rejoining) by which newly secreted xyloglucans may become covalently integrated into the wall: the newly secreted molecule may undergo a transglycosylation reaction in which it is bonded to a pre-existing wall-bound chain (Fig. 4). Since XET activity cannot achieve net bond synthesis (a high-energy reactant, such as UDP-glucose or ATP would be necessary for that), either a fragment of the newly secreted xyloglucan would fail to become wall-bound, or a fragment of the pre-existing xyloglucan could lose its foothold in the wall during the process. However, such a fragment could be a small proportion of the total.
Xyloglucan endotransglucosylase could serve this role in xyloglucan integration: but does it?
The 13C/3H pulse–chase experiments have shown that transglycosylation accompanies, and may therefore cause, the wall-binding of newly synthesized xyloglucans in vivo (Thompson et al., 1997; Thompson & Fry, 2001a). In rose cell cultures maintained constantly on either heavy or light medium (i.e. with [13C]glucose or [12C]glucose as sole carbon source, respectively), and then given a brief pulse of [3H]arabinose, the buoyant density of the pulse-radiolabelled [3H]xyloglucans remained constant as they ‘aged’ in the cell wall (e.g. 0–7 d after radiolabelling). This shows that ageing xyloglucan does not undergo any inevitable density changes that might be attributable, for example, to defucosylation (Thompson & Fry, 2001a). With the foregoing as an important control experiment, [13C]glucose-grown Rosa cells were given a 2-h pulse of l-[1-3H]arabinose, resulting in the synthesis of a cohort of ‘hot, heavy’ (3H,13C)-xyloglucan molecules. At 6 h after the pulse of radiolabel, the cells were transferred into [12C]glucose medium so that the chains synthesized thereafter would be ‘cold, light’ (1H,12C)-xyloglucans (Thompson et al., 1997). The mean buoyant density of the wall-bound [3H]xyloglucan decreased during the 7 d after the 13C → 12C shift. This indicates that during or after the wall-binding of newly synthesized [12C,1H]xyloglucan, it became covalently attached to lengths of previously wall-bound [13C,3H]xyloglucan, thus reducing the mean buoyant density of the 3H-labelled material.
In a variation on this theme, [12C]glycerol-grown rose cells were transferred into [13C]glucose medium, 20 min before a 2-h pulse of [3H]arabinose (Thompson et al., 1997). Twenty minutes in glucose medium is sufficient to saturate a glycerol-grown cell's pools of the relevant intermediary metabolites (UDP-Glc, UDP-Xyl, etc.) with the carbon isotope (13C) present in the new carbon source (glucose). Thus, by the time of 3H-labelling, all the polysaccharides being synthesized were c. 100%‘heavy’, and the newly synthesized chains would have been [3H,13C]xyloglucans. It was therefore interesting to discover that the earliest wall-bound [3H]xyloglucan detected had a density that indicated a 12C : 13C ratio of approx. 1 : 1. This shows that, during wall-binding, (segments of) the newly synthesized [13C,3H]xyloglucan became covalently attached to (segments of) previously synthesized [12C]xyloglucan. The only known mechanism by which such ‘grafting’ could occur in the cell wall is by XET-catalysed transglycosylation. The observations therefore indicate that XET action accompanies, and may cause, the integration of newly secreted xyloglucan chains into the existing cell wall architecture.
This would account for the very rapid, and strong, wall-binding of newly secreted xyloglucans (Edelmann & Fry, 1992b). This type of XET action is referred to as ‘integrational transglycosylation’. In the dual-labelling system described (Thompson et al., 1997) both the donor and the acceptor substrate were endogenous, natural polysaccharides; therefore the ‘grafting’ process detected can confidently be stated to occur in the living cell.
Other 3H pulse–chase experiments, without any density-labelling, have shown that, especially in the Gramineae, xyloglucans greatly increase in Mr when they become integrated into the PCW (Kerr & Fry, 2003). This observation also supports the involvement of some form of molecular ‘grafting’, most likely polymer-to-polymer transglycosylation, in wall assembly.
Xyloglucan endotransglucosylase action in vivo: role in wall restructuring Once radiolabelled xyloglucan is integrated into the walls of cultured rose cells, a proportion of it is subsequently (over many hours) sloughed into the medium (Edelmann & Fry, 1992b; Thompson & Fry, 1997). The weight-average relative molecular mass (Mw) of [3H]xyloglucan freshly deposited in the cell wall was initially 160 000 or 240 000 (in fast- and slow-growing cells, respectively); the wall-bound [3H]xyloglucan of both cultures underwent a decrease in Mw of approx. 40 000 during the first 2 d after the pulse-labelling (Thompson & Fry, 1997). At the same time, 20–30% of the initially deposited [3H]xyloglucan disappeared from the cell wall, and a similar amount appeared in solution in the culture medium. Its failure to remain bound to the cell wall and its low Mw (c. 39 000) indicated that this soluble extracellular [3H]xyloglucan was formed by partial degradation of segments of wall-bound xyloglucan that were not directly hydrogen-bonded to microfibrils (‘loose ends’ and ‘tethers’). It was not established in this work (Thompson & Fry, 1997) whether the ‘trimming’ was achieved by hydrolysis, transglycosylation or nonenzymic cleavage. Nevertheless, the data support the view that xyloglucan is subject to an active metabolism in the walls of cultured rose cells, as previously reported for auxin-treated pea stem segments (Labavitch & Ray, 1974) and Vigna hypocotyl segments (Nishitani & Masuda, 1982).
In the experiments to test the contribution of XET action to cell wall loosening, and specifically to look for transglycosylation in vivo, ‘heavy’ rose cells were suddenly shifted into medium with only a ‘light’ carbon source, then given a brief (2-h) pulse of [3H]arabinose followed by a chase with nonradioactive arabinose (Fig. 6). This protocol would result in light, radioactive xyloglucans being secreted into heavy, nonradioactive walls and then being chased by new, light, nonradioactive xyloglucans. As soon as any radioactive xyloglucan became wall-bound, the (initially light) 3H-labelled polysaccharide had become covalently bonded to heavy chains, forming hybrid molecules that were, on average, 29% heavy : 71% light (Fig. 7). This was interpreted as molecular grafting (Fig. 4) due to integrational transglycosylation. It cannot be determined from these data whether the newly secreted xyloglucan chains were usually the donor or usually the acceptor substrates for transglycosylation. The [3H]xyloglucan then gradually increased in buoyant density until, by 11 h, the hybrid chains were 38% heavy. Since negligible new [3H]xyloglucan was secreted between 2 h and 11 h, and the only new nonradioactive xyloglucans being secreted were light, the increase in density demonstrated that ‘re-structuring’ reactions occurred between the now wall-bound (29% heavy) [3H]xyloglucan and other (mainly old, i.e. approx. 100% heavy) wall-bound nonradioactive xyloglucans (Thompson & Fry, 2001b).
Brefeldin A (BFA), which blocks the Golgi vesicular traffic (Driouich et al., 1993), inhibited [3H]xyloglucan secretion (Thompson & Fry, 2001a). However, BFA, added 2.5 h after the [3H]arabinose, did not prevent the increase in buoyant density of the [3H]xyloglucan that occurred between 2 h and 11 h (Fig. 7). Thus, wall-bound [3H]xyloglucans continued to undergo transglycosylation reactions with the limited pool of existing (heavy) wall xyloglucans. Such transglycosylation must have been of the restructuring type since secretion (which is necessary for continued integrational transglycosylation) had been blocked. By 7 d in the presence of BFA, the 3H was found in hybrid molecules in which, on average, 55% of the molecule was heavy, indicating quite a thorough restructuring of the wall's xyloglucan (Thompson & Fry, 2001a).
Exogenous 1 mm XGOs (which are competing acceptor substrates for XETs) were added before or after [3H]arabinose. They did not affect integrational transglycosylation but they blocked the restructuring process (Fig. 7) (Thompson & Fry, 2001a). This supports the conclusion that restructuring was caused by an XET-catalysed transglycosylation since no other known reaction would be interfered with by 1 mm XGOs. It was suggested that the lack of effect of exogenous XGOs on integrational transglycosylation was because that process occurs at a particular site (on the inner face of the cell wall, near the point where a Golgi-derived vesicle had recently discharged its contents by exocytosis) where the endogenous xyloglucan concentration is very high and therefore exogenous XGOs at 1 mm would have little chance to compete with the cell's intended (high-Mr xyloglucan) acceptor substrates.
Between them, the data indicate that both integrational and restructuring transglycosylation occur concurrently, and presumably in competition with each other, during the normal growth of cultured Rosa cells. Integrational transglycosylation occurs very soon after xyloglucan secretion; restructuring transglycosylation occurs more gradually over a period of hours and days. It is likely that, by catalysing both types of transglycosylation, XETs serve important roles in both the assembly and the loosening of the PCW, together enabling long-term plant cell expansion with minimal loss of wall strength. The proposed architectural role of restructuring transglycosylation is illustrated in Fig. 8.
Spatial localization of XET action in vivo Experiments of the type described above, using living cells fed 3H and stable heavy isotopes, provide the most reliable evidence for the natural occurrence of transglycosylation reactions in vivo because all the xyloglucan-related substrates present in the system are endogenous and chemically normal. However, such labelling techniques cannot be used microscopically to localize XET action on endogenous substrates in vivo: the 3H could certainly be localized by microautoradiography, but there would be no way of deducing whether endogenous [3H]xyloglucan chains had been cleaved and rejoined.
Instead, workers have compromised by observing in vivo transglycosylation between one endogenous and one (labelled) exogenous substrate, the latter a labelled XGO that can act as an acceptor substrate for XET activity. The label could be either 3H or a fluorescent group, although microscopic detection of fluorescence is simpler than that of radioactivity. At a set time after the labelled XGO was fed, all remaining unreacted XGO is washed out of the tissue (e.g. with 80% ethanol). The formation of a fluorescently labelled, high-Mr product (not washed out by ethanol) at a particular location within the specimen indicates the action of endogenous XET on endogenous donor xyloglucans at that location. This constitutes a localization of action rather than of mere activity. The only quibble is that there is no proof that any endogenous acceptor substrate would have been naturally present at the same subcellular site and within reach of the enzyme, and thus no definite proof that full-blown transglycosylation can occur in the absence of the artificially added XGO.
Valuable data regarding the action of XTH on its donor substrate have nevertheless been obtained by use of fluorescently labelled XGOs in this way. Ito & Nishitani (1999) used a fluorescein-labelled heptasaccharide (XXXG) as a probe (exogenous acceptor substrate) to visualize the action of wall-bound XET activity on endogenous wall-bound xyloglucan in suspension-cultured tobacco cells. Enzyme action, indicated by wall-bound green fluorescence, was reduced in transformant cells in which the expression of an XTH had been suppressed by antisense technology and was enhanced in transformants that over-expressed XTH. The fluorescent polysaccharide produced was not extractable by 0.6 m NaOH at 25°C for 10 h, showing that it was firmly integrated within the wall architecture. It cannot be determined whether the donor substrate used by the enzyme had been recently secreted or had been a long-standing part of the wall's architecture, and therefore it cannot be deduced whether the XET action detected would have been involved in xyloglucan integration or in the restructuring of the existing cellulose–xyloglucan meshwork.
Fry (1997) reported a method for preparing highly fluorescent sulphorhodamine conjugates of XGOs (XGO–SRs; Fig. 9) and demonstrated their use in tissue prints to map XET activity with exogenous donor substrates. Vissenberg et al. (2000) then used these XGO–SRs for detecting XET action on endogenous donor substrate in vivo. The XGO–SRs were infiltrated into pieces of living plant tissue, and high-Mr fluorescent products were formed in the cell wall (Fig. 10). The method was specific for XET action, as shown by competition experiments with nonfluorescent XGOs, which competitively inhibited the formation of high-Mr fluorescent products, and by negligible reaction with cello-oligosaccharide–SRs, which are not XTH acceptor substrates. Thin-layer chromatography of the remaining, unincorporated XGO–SRs showed that these substrates were not extensively hydrolysed during the assays; thus, detection of XET action was not compromised by degradation of the fluorescent acceptor substrate. In young celery petioles, XET action was particularly high in the thick-walled but still elongating cells of the collenchyma. In roots of both Arabidopsis and tobacco, XET action was most prominent in the zone of cell elongation (Fig. 10; Vissenberg et al., 2000). Thus, high XET action is correlated with rapid cell expansion.
Root hairs are formed in certain epidermal cells (trichoblasts) by hair initiation and subsequent tip growth. Use of XGO–SRs as described above allowed a description to be made of the spatial and temporal patterns of XET action during these processes (Vissenberg et al., 2001). Root hair initiation, which involves the bulging of an epidermal cell wall caused by highly localized wall modifications, was always accompanied by a highly localized increase in XET action. The fluorescence at sites of future bulge formation was much brighter than in the nonhair parts of the trichoblast, suggesting an important role of XET action in root hair initiation. Older root hairs exhibited high XET action uniformly over the hair wall and not only at the hair's tip, where growth occurs. Experiments in which roots were incubated at various pHs (4.5, 5.5 or 7.0) suggested that at least three distinguishable isoforms of XET acted in different parts of the root. It was suggested (Vissenberg et al., 2001) that the XTH which acts in the cortex and epidermis of the elongation zone is involved in wall loosening for root elongation, that which acts in immature trichoblasts is involved in bulge initiation, and that which acts in the side-walls of older hairs is involved in the integration of new xyloglucan material for wall strengthening. High XET action in root hairs is a feature of all vascular plants examined – even those as primitive as lycopodiophytes (Vissenberg et al., 2003).
A similar method has also been used to localize XET action in poplar wood (Bourquin et al., 2002), and it was noted that XET action here is high at the time when secondary wall deposition is initiated. This was proposed to indicate a role for transglycosylation in welding the outer surface of the secondary wall to the inner surface of the primary wall.
6. Feruloylation of polysaccharides
The addition of feruloyl groups to polysaccharide chains (Fig. 1) could, in principle, occur within the endomembrane system (Golgi cisternae or vesicles) or after secretion of the polysaccharide into the PCW. To track this process in vivo it is not convenient to 3H-label the feruloyl group since it would then be necessary to use electron microscopy to localize the incorporated 3H at the subcellular level. Instead, it is more convenient to label the Ara and Xyl residues of the polysaccharide by feeding of exogenous [3H]arabinose. It is well established that new sugar residues are added to elongating polysaccharide chains within the Golgi system (Northcote & Pickett-Heaps, 1966); there is then a 10–30 min transit time before the 3H-polysaccharide reaches the PCW. If polysaccharide-bound 3H-sugar residues acquire feruloyl side-chains < 10 min after the first polysaccharide-bound 3H-sugar residues appear, then the feruloylation must be occurring intraprotoplasmically. It is straightforward to determine whether a polysaccharide-bound 3H-sugar residue is feruloylated or not: for example, mild acid hydrolysis of [3H]arabinoxylan yields the monosaccharide, [3H]arabinose, from nonferuloylated residues and feruloyl-[3H]arabinose from feruloylated residues. These two compounds can be resolved from each other (e.g. by paper chromatography) and then separately assayed for 3H. It has been found that the feruloylation occurs < 2 min after 3H-sugar incorporation, indicating that the process occurs within the endomembrane system, rather than in the cell wall after secretion, both in a dicot (spinach; Fry, 1987) and in a monocot (fescue grass; Myton & Fry, 1994).
This is a situation where the action of the enzyme is better established than the activity. Only recently has a feruloyltransferase been characterized: in vitro it can use feruloyl-CoA as donor substrate and a trisaccharide, Ara–Xyl–Xyl, as acceptor substrate (Yoshida-Shimokawa et al., 2001). It seems likely, but is not yet proven, that this enzyme activity can use the polysaccharide (arabinoxylan) as acceptor substrate in vivo.
7. Oxidative cross-linking
Feruloyl side-chains attached to polysaccharides can undergo oxidative phenolic coupling and thus potentially crosslink the polysaccharide chains to which they are attached (Fig. 11). Coupling could be catalysed by peroxidase or possibly laccase (Wallace & Fry, 1999). Peroxidase activity often correlates negatively with the rate of cell expansion (Fry, 1979; Warneck et al., 1996; MacAdam & Grabber, 2002), suggesting that peroxidase does indeed catalyse a ‘wall-tightening’ reaction. There is a long-standing assumption that feruloyl dimerization occurs in the cell wall itself, catalysed by the peroxidases that exist there. However, critical experiments to test this dogma by tracking the action of the enzymes (i.e. the fates of wall components) in vivo have shown that it is an oversimplification.
Experiments involved the feeding of a ferulate precursor, [14C]cinnamate (which is expected not to be susceptible to oxidative coupling until it has been hydroxylated within the protoplast) to cultured maize cells (Fry et al., 2000). This work showed that many feruloyl residues undergo oxidative coupling as little as < 1 min after they have been attached to a polysaccharide chain (i.e. still within the endomembrane system rather than in the wall; see preceding topic, Feruloylation of polysaccharides). Intraprotoplasmic coupling was not prevented by extracellular antioxidants (ascorbate or dithiothreitol) or by Brefeldin A (an inhibitor of polysaccharide secretion). The occurrence of intraprotoplasmic feruloyl coupling indicates that some polysaccharides are secreted into the PCW already crosslinked, and this has significant consequences for the mechanism of wall assembly and for the physiological role of phenolic coupling. For example, the implication is that each ‘vesicleful’ of xylans secreted by a cultured maize cell is in effect a single macromolecular coagulum (clot) rather than a solution of individual polysaccharide chains. Such a coagulum would be unlikely to permeate the wall after exocytosis but would remain as an intact coagulum, squashed on to the inner surface of the wall by the turgid protoplast. A coagulum would seem more likely to lubricate the wall than to infiltrate between microfibrils and ‘clamp’ the wall.
The same study (Fry et al., 2000) also indicated that 14C-dimers are not the only products of [14C]feruloyl coupling: 14C-labelled trimers and higher oligomers substantially exceeded the dimers quantitatively. This finding also invites a reassessment of the role of oxidative phenolic coupling: in many studies, total ferulate coupling will have been underestimated because only the dimers were measured (Ralph et al., 1994; Grabber et al., 1995; Waldron et al., 1997). One specific trimer of ferulate has recently been characterized in detail (Rouau et al., 2003), and it seems likely that numerous trimers, and higher oligomers, will soon have been structurally elucidated.
In the context of the present article, an important advantage of in-vivo radiolabelling, and thus the ability to monitor the ‘careers’ of the wall components of interest, is that it permits the detection of all metabolic products – in this case of polysaccharide-bound feruloyl groups – whether or not the nature of these metabolites could have been guessed. The products are guaranteed to be radioactive; therefore, accurate ‘book-keeping’ of the radiolabelled materials present in the cells plus their medium (and gas space if necessary) will reveal their presence. Some of their properties can then also be elucidated by radio-chromatography, radio-electrophoresis, etc. (Fry, 2000).
8. Expansin action
Further molecular rearrangements befall polysaccharides after secretion into the cell wall. Two groups of proteins, the α- and β-expansins, can loosen the wall by locally breaking interpolymeric hydrogen-bonds (McQueen-Mason & Cosgrove, 1994; Cosgrove et al., 1997), apparently without catalysing any covalent reaction. It is, however, currently impossible to prove that expansin-mediated cleavage of hydrogen bonds is occurring in the walls of living cells. Thus, expansin action (as opposed to activity) cannot yet be demonstrated in vivo.
9. Action of enzymes encoded by foreign genes
One approach by which to look for enzyme action, as opposed to mere activity, is to transform a plant with a foreign gene and to observe some effect on the chemistry of the walls of the transformant (Atkinson et al., 2002; Esteban et al., 2003; Park et al., 2003). This type of experiment would indicate that the foreign gene product can act in vivo, although not necessarily that the same gene product normally acts in the plant from which the gene was taken or that there was necessarily any equivalent enzyme action in the host plant before its transformation. It is also important to establish whether the inserted gene causes an increase, decrease or no effect on the activity of the enzyme in question: in the case of potato plants transformed with a petunia pectin methylesterase gene, under the control of the CMV 35S promoter, elevated, unchanged and diminished pectin methylesterase (PME) activities were observed in different parts of the potato plant and at different stages of development (Pilling et al., 2000). Pilling et al. (2000) observed that a decreased PME activity (in young stems of the transformed potato plants) correlated with an enhanced growth rate.
IV. Evidence for the occurrence of nonenzymic polymer scission in vivo?
Do •OH radicals cleave wall polysaccharides in vivo? Since •OH radicals are so extremely short-lived and reactive, they cannot be isolated from the apoplast and quantified. However, it is still possible, in principle, to test for the in vivo production of •OH in the apoplast and the action of •OH on wall polysaccharides. To demonstrate •OH production in the apoplast, it is feasible to employ membrane-impermeant probes that react with •OH to give a quantifiable product. A selection of such probes ([3H]benzoylated hydrophilic polymers) have been prepared, and have been shown to release an easily measurable product (3H2O) in the presence specifically of •OH (Fry et al., 2002). These probes are now available to be used for detailed testing in the apoplast in vivo.
To demonstrate the action of •OH on wall polysaccharides in vivo, the polysaccharides can be analysed for the presence of peculiar chemical features (‘fingerprints’) that are diagnostic of recent •OH attack. These fingerprints have been described on the basis of studies with pure polysaccharides treated with •OH in vitro (Fry et al., 2001; Miller & Fry, 2001). Apparently similar fingerprints are also obtained from the wall polysaccharides of pear ‘fruit’— progressively more during ripening (Fry et al., 2001). This suggests that •OH radical attack may be a normal feature of wall-loosening during the softening of the fruit.
Work is currently under way in this laboratory to apply the fingerprinting method to various plant tissues and thus to show whether and how widely •OH radical action can be detected in the apoplast of healthy plants during bouts of wall loosening.
Despite impressive recent progress in the characterization of genes that may direct polysaccharide modification, large gaps remain in our knowledge of the timing, location, enzymology and chemistry of wall polysaccharide reactions actually occurring in vivo during plant development. Much remains to be done by the methods of macromolecular metabolism, following molecules’ careers in vivo, including isotopic labelling, to dissect out these processes and to provide a firm biochemical foundation for interpreting their biological significance.
I thank the BBSRC for funding the previously unpublished work reported.