Extensins, hydroxyproline-rich repetitive glycoproteins with Ser–Hyp4 motifs, are structural proteins in plant cell walls. The leucine-rich repeat extensin 1 (LRX1) of Arabidopsis thaliana is an extracellular protein with both a leucine-rich repeat and an extensin domain, and has been demonstrated to be important for cell-wall formation in root hairs. lrx1 mutants develop defective cell walls, resulting in a strong root hair phenotype. The extensin domain is essential for protein function and is thought to confer insolubilization of LRX1 in the cell wall. Here, in vivo characterization of the LRX1 extensin domain is described. First, a series of LRX1 extensin deletion constructs was produced that led to identification of a much shorter, functional extensin domain. Tyr residues can induce intra- and inter-molecular cross-links in extensins, and substitution of Tyr in the extensin domain by Phe led to reduced activity of the corresponding LRX1 protein. An additional function of Tyr (or Phe) is provided by the aromatic nature of the side chain. This suggests that these residues might be involved in hydrophobic stacking, possibly as a mechanism of protein assembly. Finally, modified LRX1 proteins lacking Tyr in the extensin domain are still insolubilized in the cell wall, indicating strong interactions of extensins within the cell wall in addition to the well-described Tyr cross-links.
Hydroxyproline-rich glycoproteins are a large family of plant extracellular proteins that include extensins. These proteins are found in organisms throughout the whole plant kingdom, playing a mechanical role and influencing the properties of cell walls (Cassab, 1998). Hydroxyproline (Hyp) is the result of post-translational hydroxylation of proline residues, and is subsequently glycosylated (Lamport, 1963). The extent of glycosylation of the same peptides varies among plant species or among tissues within the same plant (Estevez et al., 2006). Extensins are characterized by Ser–Hyp4 repeats, frequently embedded in a higher-order repetitive structure with blocks of O-glycosylated Hyp residues. Contiguous Hyp residues are arabinosylated, and the Ser residue is monogalactosylated (Showalter, 1993; Shpak et al., 1999, 2001). Extensins form 80–100 nm rod-like structures with a polyproline II helical conformation, that are stabilized by the glycomodules (Heckman et al., 1988). Cross-linking of extensin peptides can result in larger structures that may be important to bridge polysaccharides (Qi et al., 1995; Cassab, 1998). (Di-)isodityrosines and pulcherosine (with a tri-tyrosine linkage) have been shown to form during insolubilization of extensins in the cell wall. This suggests that oxidative cross-linking via the hydroxyl group in the aromatic ring of Tyr is the chemical basis of extensin insolubilization (Fry, 1982; Epstein and Lamport, 1984; Brady et al., 1998; Held et al., 2004). In vitro cross-linking experiments with various extensins and glycine-rich proteins showed that cross-linking of the proteins depends on Tyr residues (Ringli et al., 2001; Held et al., 2004). However, glycoproteins can also be attached to other cell-wall components via sugar moieties, exemplified by extensin–pectin interaction (Iiyama et al., 1994; Qi et al., 1995).
Reports on characterization of extensins based on antisense strategies and knock-out mutants are scarce. The lack of knock-out mutants has also prevented the characterization of extensins in respect to motifs or individual amino acids crucial for protein function using genetic strategies. An exception is the Arabidopsis extensin mutant rsh (root shoot hypocotyl defective), which fails to properly form new cell plates during cytokinesis, resulting in embryo lethality (Hall and Cannon, 2002). In vitro, RSH has the propensity for self-assembly into a scaffold, in which Tyr-based intermolecular linkages are formed between the RSH peptides (Cannon et al., 2008). Pectins and subsequently other cell-wall polysaccharides are thought to aggregate onto this structure, and thereby build up a wall in a coordinated way.
Extensins can also occur in the context of chimeric proteins, e.g. extensin–arabinogalactan proteins (Lind et al., 1994), lectin extensins (Kieliszewski et al., 1994) or leucine-rich repeat (LRR) extensin proteins (Rubinstein et al., 1995a). LRR extensins (LRX) contain a signal peptide for protein export, an N-terminal LRR and a C-terminal extensin domain (Figure 1a) (Baumberger et al., 2003a). Extracellular localization has been demonstrated for PEX1 of maize (Rubinstein et al., 1995b) and LRX1 of Arabidopsis (Baumberger et al., 2001). The Arabidopsis gene LRX1 is expressed in root hairs, where it functions in cell-wall formation. lrx1 mutants develop aberrant root hairs, probably as a result of defects in the cell-wall structure. LRX1 is insolubilized in the cell wall, a property that is attributed to the extensin moiety. Expression of a deletion construct of LRX1 lacking the extensin domain failed to complement the lrx1 mutant, and, when expressed in wild-type plants, led to an lrx1-like root hair phenotype. This suggests that the LRR domain competes for and blocks the interaction partner of the endogenous LRX1. Furthermore, it shows that the extensin domain is essential for LRX1 function (Baumberger et al., 2001).
Using complementation of the lrx1 mutant and expression in wild-type plants, the extensin domain of LRX1 was characterized in vivo to identify motifs and individual amino acids important for protein function. The various repetitive motifs in the extensin domain were shown to be functionally redundant and could be mostly be removed without losing protein activity. In situ immunodetection in purified cell-wall fractions revealed that insolubilization of LRX1 is dependent on the extensin moiety. However, Tyr residues, although important for fully active LRX1, are not essential for insolubilization of LRX1 in the cell wall.
Functional redundancy within the repetitive extensin domain of LRX1
Based on the amino acid sequence, the 363 amino acid extensin domain of LRX1 (where the Ser residue of the first Ser–Hyp4 motif is position 1) can be divided into three repetitive motifs: (SP5S2KMSPSVRAY)3, (SP4SP4YVYS)6 and (SP4SPVY2P2VTP/Q)10. The extensin sequences N-terminal to motif I and C-terminal to motif III contain the typical Ser–Hyp4 motif but otherwise are not particularly repetitive (Figure 1a and Figure S1). To identify whether all the repetitive motifs are necessary for a functional LRX1, a series of LRX1 deletion constructs was produced. The deletions (and all subsequent modifications) were based on an LRX1:mycLRX1 construct encoding LRX1 with a c-myc tag in front of the LRR domain (Figure 1a). This construct, under the control of the LRX1 promoter and LRX1 terminator, was previously shown to complement the lrx1 mutation, and can be detected using an anti c-myc antibody (Baumberger et al., 2001). The protein encoded by mycLRX1ΔE296 terminates at position 296 of the extensin domain, immediately C-terminal to motif III. Correspondingly, mycLRX1ΔE153 terminates C-terminal to motif II, mycLRX1ΔE90 terminates C-terminal to motif I, and mycLRX1ΔE14 contains only the first 14 amino acids of the LRX1 extensin domain (Figure 1a and Figure S1). These constructs were transformed into lrx1 plants, and complementation of the lrx1 root-hair formation phenotype was assessed. Seedlings expressing mycLRX1ΔE296, mycLRX1ΔE153 and mycLRX1ΔE90 rescued the lrx1 phenotype, but those expressing mycLRX1ΔE14 failed to develop wild-type-like root hairs (Figure 2a–f and Table 1). Expression of the transgene-encoded LRX1 constructs was confirmed by Northern hybridization using a c-myc-specific probe for hybridization of total RNA from one representative transgenic line for each construct (Figure 2g). This experiment showed that mycLRX1ΔE90 is the LRX1 protein with the shortest functional extensin domain. Thus the deleted extensin sequences, including the well-conserved repetitive motifs II and III (Figure S1), appear to be dispensable for extensin function under the growth conditions used.
Table 1. Effect of LRX1 constructs in wild-type and lrx1 mutant seedlings, and insolubilization of the proteins in the cell wall
Complementation of lrx1
lrx1-like phenotype in wild-type
Insolubilization in the cell wall
The constructs were all expressed under the control of the LRX1 promoter and terminator.
Extensins are known to have wall-stabilizing properties. Thus, a minimal extensin domain may be required in an LRX1 protein for complementation of the lrx1 mutant merely due to the wall-reinforcing activity of this protein domain. To test this possibility, the effect of only the LRX1 extensin domain on the lrx1 phenotype was investigated. The sequences corresponding to the LRX1 promoter and N-terminal signal peptide important for protein export were fused to those corresponding to the full-length extensin domain and the terminator of LRX1. This construct was transformed into lrx1 plants and the root hair phenotype was analysed in four transgenic lines. The lines expressed the transgene as determined by Northern blotting but still developed the lrx1 root hair phenotype (Figure 3). This suggests that the extensin domain primarily functions as an anchor to properly position the LRX1 protein in the cell wall, rather than having a cell-wall stabilizing function.
Tyrosines in the extensin domain are functionally important
The extensin domain of mycLRX1ΔE90 contains seven Tyr (Y) residues, two in the context of the sequence Tyr-Val-Tyr N-terminal to motif I and five individual Tyr residues in or following motif I (indicated in Figure 1b). To assess the contribution of these Tyr residues to protein function, mycLRX1ΔE90 was mutagenized such that the Tyr codons were changed to Phe (F). Compared to Tyr, Phe lacks the hydroxyl group on the aromatic ring of the side chain involved in oxidative cross-linking of Tyr residues (Figure S2). Thus, Phe cannot undergo oxidative cross-linking but is otherwise identical to Tyr. In a first step, a mutant mycLRX1ΔE90 was established in which the five isolated Tyr residues were changed to Phe, leaving the Y-V-Y motif intact. This construct, referred to as mycLRX1ΔE90_YVY-5F, was transformed into lrx1 mutants, and analysis of transgenic seedlings revealed that it is able to complement the lrx1 phenotype (Figure 4b). Similarly, a construct with a modified Y-V-Y motif in which the individual Tyr residues were left unchanged (referred to as mycLRX1ΔE90_FVF-5Y) was also able to complement the lrx1 mutation (Figure 4c). In contrast, construct mycLRX1ΔE90_FVF-5F, in which all Tyr residues in the extensin domain were changed to Phe, was not able to complement the lrx1 mutation (Figure 4d and Table 1). To test for expression of mycLRX1ΔE90_FVF-5F in the lrx1 mutant, Western blotting of total root protein extracts was performed using a c-myc-specific antibody. This revealed similar abundance of mycLRX1ΔE90 and mycLRX1ΔE90_FVF-5F in the respective transgenic plants (Figure 4f). These experiments show that, under the growth conditions used, tyrosine residues are necessary for a functional LRX1 extensin domain, and consequently for functional LRX1. mycLRX1ΔE90 migrates at approximately 100 kDa, which is more than the 62.5 kDa predicted based on the amino acid sequence. This suggests that the extensin domain of mycLRX1ΔE90 is glycosylated, as shown for other extensins.
Aromatic amino acids are important for function of the extensin domain
Over-expression in wild-type Arabidopsis of an LRX1 deletion construct lacking the entire extensin domain has a dominant-negative effect, resulting in an lrx1-like root hair phenotype (Baumberger et al., 2001). This phenomenon was used as a second parameter to assess the function of the various mycLRX1 proteins. As expected, expression of the functional mycLRX1ΔE90 in wild-type plants resulted in a normal root hair phenotype (Figure 5a,b). However, expression of mycLRX1ΔE14, lacking almost the whole extensin domain, led to development of an lrx1-like phenotype (Figure 5e) that segregated in the T2 population as a dominant trait. This confirmed that mycLRX1ΔE14 contains a non-functional extensin domain, and hence has a negative effect on root hair formation. The expression of mycLRX1ΔE90_FVF-5F resulted in a predominantly wild-type root hair phenotype (Figure 5c and Table 1), even though the protein was present in root extracts as determined by Western blotting (Figure 5f). This indicates that lack of Tyr residues in the extensin domain results in a protein with attenuated activity that does not allow complementation of the lrx1 mutant but prevents a negative effect on root-hair development in wild-type plants.
Previous work has suggested that the aromatic nature of Tyr residues is important for establishment of a self-assembled extensin network via hydrophobic stacking (Cannon et al., 2008). To investigate this, the Tyr residues in mycLRX1ΔE90 were changed to Leu (L), a non-polar amino acid lacking an aromatic ring (Figure S2), resulting in mycLRX1ΔE90_LVL-5L. Expression of mycLRX1ΔE90_LVL-5L did not complement the lrx1 mutant (Figure 4e), even though accumulation of the protein was demonstrated (Figure 4f). When expressed in wild-type Columbia, mycLRX1ΔE90_LVL-5L interfered with root-hair development (Figure 5d). However, fully formed root hairs were still detectable, in contrast to wild-type plants expressing the non-functional mycLRX1ΔE14 (Figure 5e and Table 1). Western blotting of total root protein extracts of the various lines revealed similar protein levels of the mycLRX1ΔE90 variants (Figure 5f). mycLRX1ΔE14 was also detectable, but at a lower level, presumably due to the few intact root hairs present in these seedlings. The stronger effect of mycLRX1ΔE90_LVL-5L on root-hair development in the wild-type compared with mycLRX1ΔE90_FVF-5F suggests that the aromatic side chains of Tyr (and Phe) residues contribute to the function of the extensin domain. However, there is still residual activity in the extensin domain of mycLRX1ΔE90_LVL-5L, as mycLRX1ΔE14 has a stronger dominant-negative effect.
The extensin domain is necessary for insolubilization of LRX1 in the cell wall
In the next step, it was tested whether the extensin domain is required for the observed insolubilization of LRX1 in the cell wall (Baumberger et al., 2001). To this end, root cell-wall fractions of plants expressing mycLRX1ΔE90 and mycLRX1ΔE14 were extensively washed to remove all soluble proteins. Subsequently, the presence of the transgene-encoded proteins in the cell-wall fractions was assessed by immunolocalization using a c-myc-specific antibody. No signal was detected in non-transgenic wild-type plant extracts, confirming the absence of non-specific immunolabelling. mycLRX1ΔE90 protein was found in most root hair structures identified, but mycLRX1ΔE14 was not immunodetected in a large number of structures identifiable as intact root hairs (Figure 6a–c). Thus, this initial analysis indicated that the extensin domain of LRX1 is responsible for the insolubilization of LRX1 in cell walls. Interestingly, immunolabelling was consistently found in root hair cell-wall fractions of seedlings expressing mycLRX1ΔE90_FVF-5F and mycLRX1ΔE90_LVL-5L (Figure 6d,e), indicating that Tyr residues of the LRX1 extensin domain are not required for insolubilization of the protein in the cell wall.
Extensins are repetitive hydroxyproline-rich glycoproteins that have structural functions. Their insolubilization in cell walls under tensile stress and upon wounding indicates that they modify the physical properties of cell walls (Cassab, 1998). However, detailed in vivo functional characterization has been hampered by the lack of obvious phenotypes induced by modified extensin expression. As the extensin domain of LRX1 is necessary for proper function of the protein, the lrx1 mutant provides a convenient model to assess protein function, and thus allowed in planta functional characterization of this extensin domain.
The extensin moiety of LRX1 and other LRX-like proteins comprises several clearly distinguishable repetitive motifs (Baumberger et al., 2003a). This is in contrast to the many extensin proteins encoded in the Arabidopsis genome that contain one prominent repetitive motif (Cannon et al., 2008). In domain-swap experiments, the extensin domain of LRX2 successfully substituted for the LRX1 extensin domain even though the repeats are quite different (Baumberger et al., 2003b). Hence, LRX1 is not dependent on a particular sequence motif in the extensin domain. The functional analysis presented here shows that a sequence containing the first repetitive motif is sufficient to provide a functional extensin, indicating functional redundancy among the repeats. Nevertheless, the repeats have been conserved during evolution, suggesting biological relevance. A possible explanation for this apparent contradiction is the laboratory conditions under which the experiments were performed. The full-length LRX1 extensin domain may be necessary for root-hair development under natural conditions. Alternatively, the various repetitive motifs may serve as a back-up system in case of accumulating mutations in one of the repeats, or amino acid substitutions may interfere with the structural organization of the extensin domain and are therefore counter-selected.
Insolubilization of structural proteins has been shown to correlate with the accumulation of di-tyrosine, pulcherosine (involving three Tyr) and isodityrosine linkages, produced by oxidative cross-linking involving the hydroxyl group of Tyr, H2O2 and peroxidase activity. Tyr residues in the context of Y-X-Y motifs are thought to lead to intra-molecular isodityrosines, which subsequently can form inter-molecular di-isodityrosine or pulcherosine cross-links (Fry, 1982; Epstein and Lamport, 1984; Brady et al., 1996, 1998). The importance of Tyr for in vitro protein cross-linking has been shown for various structural proteins (Ringli et al., 2001; Held et al., 2004; Cannon et al., 2008). The mutational analysis presented here shows that Tyr residues indeed are important for function, but not for insolubilization of LRX1 in the cell wall. In addition, the Tyr residues do not have to be present in the context of the Y-V-Y motif, as mycLRX1ΔE90_FVF-5Y, containing five isolated Tyr residues, represents a functional protein. It may be speculated that Tyr residues are important for intra-molecular cross-links to stabilize the three-dimensional conformation of LRX1. However, the aromatic ring of Tyr residues (and of Phe residues in mycLRX1ΔE90_FVF-5F) also contributes to the function of the LRX1 extensin domain, as indicated by the fact that mycLRX1ΔE90_LVL-5L shows further reduced activity compared to mycLRX1ΔE90_FVF-5F. Substituting the aromatic amino acid Phe with the non-polar residue Leu interfered with root-hair development in the wild type. Thus aromatic amino acids have a positive influence on protein function, indicating that hydrophobic stacking might be an organizing principle in establishment of the proper structure of the LRX1 extensin domain. The propensity of the extensin RSH to self-aggregate into a complex structure has been shown previously (Cannon et al., 2008). Similar to collagen (Cejas et al., 2007), Tyr residues (and Phe residues in mycLRX1ΔE90_FVF-5F) may act as hydrophobic recognition sites enabling the amphiphilic extensin domain to organize into a three-dimensional aggregate.
The insolubilization of LRX1 is an important, functionally essential, aspect of the protein. The extensin-less protein mycLRX1ΔE14 is not detectable in root hair structures. The fact that the protein level is below the detection threshold could be due to the low abundance of the protein in total root extracts. However, the low amount of mycLRX1ΔE14 also reflects the small number of intact root hairs potentially accumulating the protein. Thus, it is more likely that the protein is not detected due to its solubility, indicating that cross-linking of LRX1 is dependent on the extensin domain. Furthermore, mycLRX1ΔE14 is dysfunctional, as reflected by the strong dominant-negative effect on root-hair development in wild-type seedlings. The weaker dominant-negative effect of mycLRX1ΔE90_LVL-5L compared to mycLRX1ΔE14 is possibly due to the insolubilization observed for this protein. As mycLRX1Δ90_LVL-5L lacks Tyr, insolubilization must involve other chemical groups present in the extensin domain. Extensins are highly glycosylated (Cassab, 1998), the nature and extent of which depends on the exact amino acid composition and the tissue in which the protein is expressed (Shpak et al., 1999; Estevez et al., 2006). The migration of the mycLRX1 deletion constructs in SDS–PAGE experiments suggests a higher molecular weight than indicated by the protein sequence, and hence possible glycosylation of the extensin domain. Glycoside moieties may be involved in the insolubilization of LRX1 in the cell wall. Such linkages have been suggested and demonstrated previously (Keegstra et al., 1973; Iiyama et al., 1994; Qi et al., 1995). Even though they are not required for protein insolubilization in the cell wall, Tyr residues in the extensin domain are essential for functional LRX1, and may be important for stabilizing the three-dimensional structure of the protein by covalent intra-molecular linkages.
This paper describes an in planta functional analysis of a plant extensin protein. This was possible as a result of the strong lrx1 mutant phenotype and the dominant-negative effect of non-functional LRX1 variants on root-hair development in wild-type. The analyses revealed functional redundancy among distinct repetitive elements in the extensin domain, and Tyr-independent insolubilization of the protein in the cell wall. Finally, this study provides supporting evidence for the hypothesis that hydrophobic interactions involving aromatic amino acids might be an organizing principle in establishing a fully functional higher-order extensin structure.
The full-length mycLRX1 (LRX1 promoter:mycLRX1:LRX1 terminator) construct has been described previously (Baumberger et al., 2001) and has the National Center for Biotechnology Information (NCBI) accession number GU235992. The deletion constructs were produced by taking advantage of restriction sites present in the LRX1 gene (ScaI for mycLRX1ΔE296, HinfIII for mycLRX1ΔE153, AccI for mycLRX1ΔE90 and Kpn2I for mycLRX1ΔE14) and an SpeI site at the stop codon of LRX1. The SpeI site was made compatible with the ScaI, HinfIII, AccI and Kpn2I sites by blunt-ending. All constructs were control-sequenced. For the modified versions of mycLRX1ΔE90 (e.g. mycLRX1ΔE90_FVF-5F), point mutations were introduced into mycLRX1ΔE90 by PCR using corresponding DNA oligonucleotides and mycLRX1ΔE90 as the template. For expression of the extensin domain, the LRX1 promoter including the signal peptide-coding sequence and the extensin coding sequence including the LRX1 terminator, respectively, were each amplified by PCR and ligated. The resulting construct encoded the first 29 amino acids and the entire extensin domain of LRX1, and contained the same promoter/terminator sequence as mycLRX1. For plant transformation, the various constructs were cloned into pART27 (Gleave, 1992) using NotI.
Plant growth and transformation
The lrx1 mutant used has been described previously (Diet et al., 2004). For all experiments, Arabidopsis seedlings were sterilized and grown on agar plates in vertical orientation for 5 days as described previously (Ringli et al., 2008). Transformation of Arabidopsis was performed as described previously (Diet et al., 2006). For each construct, transgenic lines segregating 3:1 for kanamycin resistance in the T2 generation, suggesting one T-DNA integration site, were selected. Homozygous transgenic plants were identified in subsequent generations and used for analysis. At least two independent transgenic lines were used for each construct described.
Northern hybridization, Western blotting and cell wall immunolabelling
For Northern hybridization, root total RNA was extracted using the TRIZOL method (Invitrogen, http://www.invitrogen.com/). Hybridization was performed using established standard protocols. As probe for hybridization, a 250 bp fragment corresponding to the 5′ end of the LRX1 transcript was used.
For Western blotting, root material of 200 seedlings was collected, ground under N2, and 50 mg (fresh weight) was extracted with 100 μl 1% SDS. After SDS–PAGE using 10 μl of each sample and semi-dry blotting onto nitrocellulose (Amersham, http://www5.amershambiosciences.com/), blocking of membranes and antibody incubation were performed according to standard protocols using JAC6 rat anti-c-myc antibody (Abcam, 1:3000 dilution) as the primary antibody, and horseradish peroxidase-labelled rabbit anti-rat antibody (Santa Cruz Biotechnology, http://www.scbt.com, 1:3000 dilution) as the secondary antibody. The ECL system (Pierce, http://www.piercenet.com) was used for detecting horseradish peroxidase activity.
For cell-wall immunolabelling, cell-wall material was obtained by extracting 100 mg (fresh weight) root material using phenol/acetic acid, as described by Fry (1988). For labelling with antibodies, the protocol described by Baumberger et al. (2001) was strictly followed. The final spreading of cell-wall material onto glass slides prior to immunolabelling allowed the identification of root hair structures by microscopy. At least 30 root hairs in total were identified per line in several independent experiments.
I would like to thank Marco Steiner for technical assistance, and Beat Keller and Benjamin Kuhn for critical reading of the manuscript. This work was supported by the Swiss National Science Foundation (grant numbers 31-61419.00 and 3100A0-103891).