Fasciclin-like arabinogalactan proteins: specialization for stem biomechanics and cell wall architecture in Arabidopsis and Eucalyptus


For correspondence (fax 61 2 6246 5000; email colleen.macmillan@csiro.au).


The ancient cell adhesion fasciclin (FAS) domain is found in bacteria, fungi, algae, insects and animals, and occurs in a large family of fasciclin-like arabinogalactan proteins (FLAs) in higher plants. Functional roles for FAS-containing proteins have been determined for insects, algae and vertebrates; however, the biological functions of the various higher-plant FLAs are not clear. Expression of some FLAs has been correlated with the onset of secondary-wall cellulose synthesis in Arabidopsis stems, and also with wood formation in the stems and branches of trees, suggesting a biological role in plant stems. We examined whether FLAs contribute to plant stem biomechanics. Using phylogenetic, transcript abundance and promoter–GUS fusion analyses, we identified a conserved subset of single FAS domain FLAs (group A FLAs) in Eucalyptus and Arabidopsis that have specific and high transcript abundance in stems, particularly in stem cells undergoing secondary-wall deposition, and that the phylogenetic conservation appears to extend to other dicots and monocots. Gene-function analyses revealed that Arabidopsis T-DNA knockout double mutant stems had altered stem biomechanics with reduced tensile strength and a reduced tensile modulus of elasticity, as well as altered cell-wall architecture and composition, with increased cellulose microfibril angle and reduced arabinose, galactose and cellulose content. Using materials engineering concepts, we relate the effects of these FLAs on cell-wall composition with stem biomechanics. Our results suggest that a subset of single FAS domain FLAs contributes to plant stem strength by affecting cellulose deposition, and to the stem modulus of elasticity by affecting the integrity of the cell-wall matrix.


The fasciclin (FAS) domain is an ancient cell adhesion domain that is found in proteins across various kingdoms: in insects (Zinn et al., 1988), humans (Escribano et al., 1994), algae (Huber and Sumper, 1994), bacteria (Ulstrup et al., 1995), birds (Kawamoto et al., 1998), plants (Schultz et al., 2000) and fungi (Miyazaki et al., 2007). Although the FAS domain is conserved across various phyla, it often occurs in extracellular matrix proteins expressed in a wide range of biological contexts, making understanding of their function(s) complex. Higher plants contain many different FAS proteins that are expressed throughout the plant (Johnson et al., 2003; Seifert and Roberts, 2007); however, their biological functions have not been clearly elucidated.

Progress in other phyla suggests that FAS proteins are involved in specific functions in diverse biological processes. In insects, the product encoded by Fasciclin-I (Zinn et al., 1988) is involved in axon fasciculation (bundling) (Snow et al., 1988) and nerve terminal arborization (Zhong and Shanley, 1995), and that encoded by Mid-line Fasciclin is likely to have multiple developmental roles including neuronal function (Hu et al., 1998). In the alga Volvox, Algal-CAM gene splice variants are important for cell–cell contact sites in embryos and stage-specific development (Huber and Sumper, 1994). The Mycobacterium protein MPB70 is thought to be involved in host–pathogen interaction, and is a critical antigen in tuberculosis vaccines (Ulstrup et al., 1995). Of the four known vertebrate FAS-containing proteins, periostin and TGFBI (βIg-H3) are highly expressed in connective tissues rich in collagen (Kannabiran and Klintworth, 2006; Norris et al., 2007), a protein that is important for tissue mechanical properties and architecture. Periostin knockout mice have impaired biomechanical properties in various tissues (Rios et al., 2005; Norris et al., 2007), and TGFBI mutations affect the human cornea, also a collagen-rich tissue, producing corneal dystrophies (Kannabiran and Klintworth, 2006).

Many FAS-containing proteins have been identified in higher plants. For example, 21 FAS-containing genes have been identified in the Arabidopsis genome (Johnson et al., 2003). Plant FAS-containing proteins are hydroxyproline-rich glycoproteins that contain sites for glycosylation with arabinose oligosaccharides and/or arabinogalactan (AG) polysaccharides to form FAS-like arabinogalactan proteins (FLAs) (Johnson et al., 2003; Seifert and Roberts, 2007). FLAs have been classified into four general groups (A–D) based on the number of FAS domains (one or two), location and number of AG domains (one or two) and whether or not they contain glycosyl phosphatidyl inositol (GPI) anchor addition sites (see Figure 1a) (Johnson et al., 2003).

Figure 1.

 Arabidopsis and eucalypt FLAs are related in sequence and transcript abundance patterns.
(a) Phylogenetic comparison of mature protein sequences of eucalypt and Arabidopsis FLAs (accession numbers in Figure S3). The diagrams indicate FLA classification into groups A–D (Johnson et al., 2003). The blue dotted lines distinguish the different FLA groups; the grey dotted line distinguishes the two structural types of group C FLAs. Red boxes, AG addition/hydroxyproline-rich regions; blue boxes, FAS domains; green boxes, GPI anchors. Within the group A FLAs, EniFLA1, 2, and 3 cluster in a subclade with AtFLA11 and 12.
(b) Quantitative real-time PCR transcript abundance analysis for AtFLA11 and 12 in (1) roots, (2) hypocotyls, (3) mature leaves, (4) base of stems, (5) mid-point of stems, (6) upper part of stems, (7) inflorescence flowers, and (8) siliques (all of mature plants). The transcript abundance of both AtFLA11 and 12 is high in stems. AtFDH transcript abundance was used as a normalization gene input for comparative quantification algorithms. Bars = 2 × SE of technical replicates.
(c) RNA blot transcript abundance of EniFLA1, 2 and 3 in (1) flowers, (2) leaves, (3) roots, (4) phloem, (5) lower branch xylem, (6) upper branch xylem, (7) vertical stem xylem, and (8) shoots. EniFLA1, 2 and 3 transcripts are abundant in stem and branch xylem, with EniFLA1 transcripts also abundant in phloem. Ribosomal RNA (rRNA) is shown as an RNA loading reference.

While the types of FLAs in higher plants have been determined, their biological functions are not well understood, partly because analyses of mutants have so far yielded few clear phenotypes. A point mutation in the FAS domain of Atfla4 of Arabidopsis gave rise to a root-swelling phenotype under salt stress, suggesting a role in cell expansion (Shi et al., 2003). In various analyses for gene networks that operate during cellulose synthesis in secondary-wall formation in Arabidopsis stems, a correlation was found between AtFLA11 and/or AtFLA12 transcript abundance and the onset of expression of secondary cell-wall cellulose synthases in Arabidopsis stems (Brown et al., 2005; Persson et al., 2005). FLA11 (Johnson et al., 2003) is also known as IRX13 (Persson et al., 2005). Atfla11 T-DNA mutants were examined for phenotypes (Brown et al., 2005; Ito et al., 2005; Persson et al., 2005), and a slight collapsed-vessel phenotype together with a reduction in stem cellulose content was reported (Persson et al., 2005). A related gene, ZeFLA11, is expressed specifically in the xylem of Zinnia elegans stems (Dahiya et al., 2006). In trees, differential FLA transcript abundance, based on microarray studies, was correlated with differences in wood properties in the xylem fibres of poplar (Lafarguette et al., 2004). Recently, we have also found similar results in Eucalyptus (Qiu et al., 2008), in which there was a correlation between FLA transcript abundance and cellulose microfibril (MF) orientation and wood properties. In radiata pine (Pinus radiata), a large difference was found in FLA EST abundance between juvenile and mature wood, which have very different wood properties (Li et al., 2009). The cellulose MF angle (MFA), i.e. the angle of the MFs relative to the longitudinal axis of the cell, is an important factor affecting wood stiffness and has important industrial consequences, especially for construction timber. The microarray-based transcript abundance findings in Arabidopsis and trees raise the interesting possibility of a causal relationship between FLA transcript abundance during secondary cell-wall formation in plant stems and the biomechanics of these tissues, and we have tested this hypothesis here.

Synthesis of woody tissues allows plants to produce structures that are both strong and flexible, such as tall trunks that can withstand stresses induced by wind and gravity, while also facilitating the transport of water and nutrients. Within woody cells such as xylem vessels and xylem fibres (both found in angiosperms) and tracheids (found in gymnosperms), the thick secondary cell wall can be considered as a fibre-reinforced composite matrix composed of strong cellulose (β-1,4-glucan) MFs embedded in a matrix of hemicelluloses (heterogeneous polysaccharides), lignin (phenolic polymer) and various proteins. As a composite material, this intricate network of molecules ultimately determines how much load a tissue can bear in terms of tension before breaking (tensile strength) and how elastic the tissue is (modulus of elasticity or stiffness).

Given the observed correlation between AtFLA11 and AtFLA12 transcript abundance and the onset of cellulose synthesis during secondary-wall deposition in Arabidopsis stems (Brown et al., 2005; Persson et al., 2005), and also the correlation of FLA transcript abundance with wood fibre properties and cellulose MF orientation in poplar and Eucalyptus (Lafarguette et al., 2004; Qiu et al., 2008), we have examined whether FLA transcript abundance is involved in plant stem biomechanics. We use Arabidopsis and Eucalyptus, which both produce stems that contain xylem vessels and fibres with significant amounts of secondary cell-wall deposition. These two plants also have some differences: Arabidopsis is an annual small herbaceous model plant, whereas eucalypts are large perennial woody trees, and their stem xylem and fibre cell ontogenies also differ. We have performed phylogenetic comparisons, tissue transcript abundance and reporter gene analyses, and phenotypic tests on knockout mutants. We found that the transcript abundance of a subset of FLA genes encoding proteins with a single FAS domain (group A FLA genes) is specific and high in Eucalyptus and Arabidopsis stems, and that similar group A FLA genes appear to be conserved across monocots and dicots. When FLA transcript abundance is reduced in Arabidopsis knockout mutant stems, plant stem biomechanics are altered, and there are changes in the molecular composition and architecture of the stem cell walls. Using principles of materials science, we describe how such molecular changes in the cell wall could affect plant stem biomechanics.


Related FLA genes from various angiosperms are stem-specific

We previously identified three eucalypt FLA genes, FLA1, 2 and 3, whose transcript abundance correlated with altered wood properties in Eucalyptus nitens wood stems (Qiu et al., 2008). Putative Arabidopsis orthologues were identified using a phylogenetic comparison of mature proteins of the three eucalypt FLAs, together with 21 Arabidopsis FLAs classified into groups A–D (Figure 1a). The N-terminal endoplasmic reticulum signal peptides and C-terminal hydrophobic tails were removed to leave the mature protein sequence. EniFLA1, 2 and 3 together with AtFLA11 and 12 form a distinct subclade of group A FLAs that contain a single FAS domain flanked by two AG regions, indicating that AtFLA11 and 12 are more closely related to EniFLA1, 2 and 3 than to the other Arabidopsis group A FLAs.

Both AtFLA11 and AtFLA12 are specifically and highly expressed in the base, mid and top sections of the inflorescence stem (Figure 1b), as determined by quantitative real-time PCR analysis. The transcript abundance was low in other tissues such as roots, leaves or flowers. These results are consistent with the Arabidopsis developmental data set of Schmid et al. (2005) obtained using the data extraction tool described by Toufighi et al. (2005), which shows that the abundance of AtFLA11 and AtFLA12 transcripts in stems is similar to that of ubiquitin (AtUbi10), secondary cell-wall cellulose synthases (AtCesA4, 7 and 8) and a laccase, with low transcript abundance in tissues that have predominantly primary cell walls, such as cotyledons and the shoot apex (Figure S1a). Although other FLA genes are expressed in stems, their transcripts are significantly less abundant than those of AtFLA11 or 12 (Figure S1b); furthermore, their transcript abundance is not stem-specific (except for AtFLA16), based on the data set of Schmid et al. (2005). Other FLAs are more abundant in tissues other than stems; for example, group A AtFLA9 transcripts are highly abundant in roots and vegetative rosette leaves, based on the developmental data set of Schmid et al. (2005), leading us to speculate that FLAs could perform similar functions in different tissues.

As found for AtFLA11 and 12, all three eucalypt FLA genes, EniFLA1, 2 and 3 show highest transcript abundance in vascular tissues of vertical stems and branches, as shown by Northern blot analysis (Figure 1c). Transcripts of the eucalypt FLA genes were localized to stem xylem, with no transcripts detected in floral, leaf, shoot or root tissues. A limited amount of EniFLA1 transcript was detected in stem phloem, but not to the same extent as observed in the xylem of stems. In addition to stem-specific expression, some FLA transcripts are highly abundant in stems and branches in trees (as for Arabidopsis), as seen in developing xylem cDNA libraries for Eucalyptus grandis stems, in which EgrFLA1 and 2 were among the most redundant transcripts (Qiu et al., 2008). The coding regions of the Eucalyptus grandis and Eucalyptus nitens FLA1, 2 and 3 genes are highly similar, sharing 99% identity at the nucleotide level.

To examine the tissue and cell specificity of Arabidopsis AtFLA11 and 12 expression, we produced transgenic Arabidopsis plants transformed with promoter–GUS fusion constructs. Both the AtFLA11 and 12 promoters demonstrated stem-specific GUS expression (Figure 2), and similar patterns were seen across all independent GUS transformants; empty binary-vector transformants showed no GUS staining. At the whole-plant level, ProAtFLA11:GUS and ProAtFLA12:GUS transformant lines showed GUS expression in the base, middle and upper regions of the stem, whereas leaves, inflorescence flowers and siliques showed little GUS staining (Figure 2a). In some lines, GUS staining was also visible in leaf petioles, stamen filaments of open flowers and the silique central vasculature, but this was not consistent across all lines (data not shown). Given the stem specificity of the GUS staining, we examined GUS localization within the stem, and found that both AtFLA11 and 12 promoters are highly active in cells that are depositing secondary walls and in the surrounding cells, such as inter-fascicular fibres and vessels, as well as parenchyma adjacent to inter-fascicular fibres and vessels (Figure 2b). ProAtFLA11:GUS transgenic plants showed GUS staining that was slightly more concentrated in stem vascular bundles than for ProAtFLA12:GUS transgenic plants, which showed GUS staining that was slightly more concentrated in the inter-fascicular fibres. In the base of ProAtFLA11:GUS transgenic stems, GUS staining was also often visible in phloem bundles. Interestingly, in transgenic tobacco plants heterologously expressing an EniFLA1promoter:GUS fusion construct, GUS staining was seen in phloem fibres, as well as the xylem (C.P. MacMillan, unpublished results). In some very strongly staining lines, GUS staining was also clearly visible in central stem parenchyma, a cell type that does not undergo secondary-wall deposition (data not shown).

Figure 2.

 Arabidopsis AtFLA11 and 12 promoter–GUS activity in various tissues. AtFLA11 and 12 promoter–GUS expression was monitored in transgenic Arabidopsis plants expressing either ProAtFLA11:GUS or ProAtFLA12:GUS constructs.
(a) GUS staining was examined in rosette leaves (post-flowering), cauline leaves, various stem parts (the upper stem of ProAtFLA11:GUS has been cut lengthwise), inflorescence flowers and siliques, and was clearly localized to all parts of the stem. Scale bars = 5 mm.
(b) Cellular localization of ProAtFLA11:GUS or ProAtFLA12:GUS staining in hand-cut cross-sections from the base, middle (mid) and upper parts of the stem. For both ProAtFLA11:GUS or ProAtFLA12:GUS constructs, GUS staining was localized to inter-fascicular fibres (if) and adjacent parenchyma (p), as well as vessels (v) and adjacent cells (arrowheads). Some staining was also seen in the phloem (ph), particularly in ProAtFLA11:GUS stem bases. ProAtFLA11:GUS staining appears to be slightly more pronounced in vascular bundles, and ProAtFLA12:GUS staining appears to be slightly more pronounced in inter-fascicular fibres. Scale bars = 50 μm.

The expression of this subset of group A FLA genes appears to be conserved, given the stem specificity of the Arabidopsis and eucalypt FLA tissue transcript abundance patterns. Phylogenetic comparisons of FLAs from monocots and dicots revealed further evidence that subsets of group A FLAs with distinct transcript abundance patterns have been conserved (Figure S2). One group A subclade contains AtFLA11, which shows expression coinciding with secondary cell-wall cellulose synthesis (Brown et al., 2005; Persson et al., 2005), and also contains FLAs from poplar (developing xylem; Lafarguette et al., 2004), zinnia (stem metaxylem; Dahiya et al., 2006), eucalypt (EgrFLA3, developing xylem), grape, wheat and rice. Another group A subclade contains the dicot FLAs AtFLA12, poplar FLAs (developing xylem; Lafarguette et al., 2004), eucalypt FLAs (EgrFLA1 and EgrFLA2; developing xylem and tension wood) and cotton FLAs (including GhFLA6, which is expressed specifically in cotton fibres undergoing secondary cell-wall deposition; Huang et al., 2008). A more distant group A subclade contains monocot and dicot FLAs together with AtFLA6, 9 and 13, which are not highly or specifically expressed in stems (Figure S1a,b).

Knockout of AtFLA11 and AtFLA12 affects stem tensile strength and stiffness

To investigate the functional roles of AtFLA11 and AtFLA12, T-DNA insertional lines were identified (Figure 3a,b), and an Atfla11/fla12 double mutant was generated that lacked mRNA expression of both AtFLA11 and AtFLA12 (Figure 3c). The morphology of Atfla11/fla12 plants appeared similar to that of Columbia wild-type plants (Figure 3d), including the inflorescence stems, which were of particular interest as they are the major site of AtFLA11 and AtFLA12 expression. At the cellular level, Atfla11/fla12 stems and hypocotyls also appeared similar to wild-type (Figure 4): the cell type, size and shape of inflorescence stem cells and the secondary cell walls appeared normal when stained with toluidine blue or phloroglucinol (for lignin), or when exposed to UV to visualize autofluorescence. When we compared the area in the stems that contained cells with secondary walls as a function of the stem diameter, there was no difference between wild-type and Atfla11/fla12 stems: 28% of the stem base cross-sectional areas contained cells with secondary walls. No difference was observed for cell-wall thickness in wild-type and Atfla11/fla12 stems.

Figure 3.

 Atfla11/12 T-DNA insertions and plant morphology. Homozygous double T-DNA insertion lines of (a) an Atfla11 allele and (b) an Atfla12 allele were generated. The mutants both contain T-DNA insertions (dotted lines) in the FAS domain. Coding regions are as indicated in Figure 1; yellow bars indicate signal peptides. Putative sites for addition of O-linked AG polysaccharides (feathers) and arabino-oligosaccharides (short bars; found only for AtFLA12) (predicted using http://cbs.dtu.dk/services/NetOGlyc/) (Julenius et al., 2005) as well as N-glycosylation sites (Y shape) (predicted using http://cbs.dtu.dk/services/NetNGlyc/) for the mature protein are indicated.
(c) AtFLA11 and AtFLA12 transcript abundance in Columbia and Atfla11/12 stems quantified by quantitative real-time PCR. Error bars = 2 × SE.
(d) Morphology of Atfla11/12 and wild-type plants. Scale bars = 1 mm.

Figure 4.

 Cellular morphology of Atfla11/12 and Columbia (wild-type) stems. Mutant and wild-type stem cells have similar histologies. Hand sections were cut from (a) hypocotyls, (b,c) stem bases, (d) mid stems and (e) upper stems, and were stained with toluidine blue (a,b,e) and phloroglucinol (d), or visualized under UV (c). Scale bars = 10 μm.

Because of the possible impact of FLA transcript abundance on stem biomechanics, we examined the tensile strength, modulus of elasticity and flexural (bending) strength of Atfla11, Atfla12 and Atfla11/fla12 stems. Tensile strength was determined using uni-axial tensile tests in which the sample is placed under tension, and the tensile strength of the sample is measured as the force required to break the sample as a function of the cross-section area of the sample. The modulus of elasticity (often referred to as stiffness) is a measure of the force required to deform the sample as a function of displacement, and also takes into account the cross-sectional area of the sample. For example, the higher the tensile stiffness, the greater the force required to stretch the sample to a given length. We also performed flexural tests, which are often used to measure the physical properties of stems (Turner and Somerville, 1997; Li et al., 2003).

The bases of both fresh and dried stems of Atfla11/fla12 stems were significantly weaker than wild-type, having approximately 70% of the tensile strength of wild-type stems (Figure 5a,b). The midpoints of Atfla11/fla12 dried stems also showed reduced tensile strength (Figure 5b). In comparison, the Atfla11 and Atfla12 single mutants did not show weaker tensile strength than wild-type (Figure 5a); sometimes Atfla11 and Atfla12 stems had slightly greater tensile strength than wild-type, but this was not consistent across experiments. Furthermore, the Atfla11/fla12 stems had reduced tensile stiffness compared to wild-type and the single mutants, for both fresh and dried stems (Figure 5c,d). The reduction in stiffness in Atfla11/12 stems was particularly pronounced in dried stem bases (26%), whereas fresh stem bases showed a 15% reduction compared to wild-type. No difference in relative stem water content was detected in the double mutant compared to wild-type (data not shown), excluding the possibility that the double mutants had reduced strength because of reduced turgor.

Figure 5.

 Biomechanical properties of Columbia (wild-type), Atfla11, Atfla12 and Atfla11/12 stems. Tensile strength (a,b), tensile stiffness (modulus of elasticity) (c,d), three-point flexural strength (e) and flexural stiffness (f) tests show that Atfla11/12 double mutants have lower tensile strength and tensile stiffness than wild-type or single mutants. Tensile strength tests measured the stress at yield (maximum load) of (a) fresh stems and (b) dried stems, and the tensile modulus of elasticity of (c) fresh and (d) dried stems. Three-point flexural tests measured (e) the stress at yield and (f) the modulus of elasticity of fresh stems at 15, 50 and 90 mm from the stem base. Black bars, wild-type Columbia; light-grey bars, Atfla11; dark-grey bars, Atfla12; white bars, Atfla11/12. Measurements sharing the same symbol are significantly different at the following levels of probability: a,b,c, P < 0.05; d,e, P ≤ 0.01; f,g, ≤ 0.005; h, P ≤ 0.001; i,j, P ≤ 0.00001). Error bars = 2 × SE. (a,c) n = 3–6; (b,d) n = 5–7; (e,f) n = 5–7.

Three-point flexural tests, which measure flexural strength and stiffness, were performed on Atfla11, Atfla12 and Atfla11/fla12 stems. Unlike tensile strength, the flexural strength of Atfla11/fla12 stems did not differ from that of wild-type as measured by the stress at yield (Figure 5e). No change in flexural stiffness was detected by the three-point flexural tests for Atfla11/fla12 stems compared to wild-type, across various parts of the stem (Figure 5f). The flexural stiffness in both single mutants was significantly increased compared to wild-type stems. The applied force in three-point flexural tests causes tension and compression forces in opposite sides of the same stem, and thus material failure can occur due to compression, tension or both.

AtFLA11 and AtFLA12 expression affect secondary cell-wall composition

The reduced tensile strength and stiffness phenotypes in the Atfla11/fla12 stems could be a consequence of an alteration in the composition of the secondary cell walls. To determine whether insertional mutagenesis in AtFLA11 and 12 affects the synthesis of cell-wall polysaccharides, we analysed the carbohydrate composition of stem cell walls. Table 1 shows that the Atfla11/fla12 stem cell walls have a significant reduction in the concentration of arabinose and galactose, consistent with AtFLA11 and AtFLA12 proteins being arabino- and galactosylated proteins and suggesting that these FLAs are indeed components of the secondary wall. Interestingly, the glucose content of the Atfla11/fla12 stem cell walls was also significantly reduced by 17%. Given the extent of the reductions in glucose moieties, it is reasonable to infer that cellulose deposition is reduced in these stems. In further experiments, cellulose content in Atfla11/fla12 stem bases was 27% lower than in wild-type and approximately 17% lower than in the single mutants (Table 2). Also, the single mutant stems contained 12–13% less cellulose than wild-type, confirming previous findings that Atfla11 stems have decreased cellulose content (Persson et al., 2005), a trend also seen by Brown et al. (2005).

Table 1.   Cell-wall monosaccharide content of Columbia (wild-type), Atfla11, Atfla12, Atfla11/12 stems
  1. Cell-wall monosaccharide contents were measured as nmol mg−1 dry weight in inflorescence stems. The Atfla11/12 double mutants had significantly less glucose, arabinose and galactose than wild-type. Data are means from two biological experiments (except for glucuronic acid, given as as mmol mg−1 dry weight, which are technical replicates from a single experiment) and are followed by SD values. t tests assuming unequal variance were used to calculate P values as a measure of statistical significance. Data that is signficantly different at  0.02 to a given genotype is indicated by an a for Columbia, b for Atfla11,c for Atfla12,d for Atfla11/12.

Arabinose110.8 ± 13.4d106.7 ± 8.7d103.0 ± 4.0d89.9 ± 6.9a,b,c
Galactose141.9 ± 14.8d139.1 ± 4.7d,c130.8 ± 3.5b118.2 ± 14.3a,b
Glucose1816.9 ± 27.5c,d1778.9 ± 60d1709.1 ± 37.2a,d1502.7 ± 131.7a,b,c
Xylose811.8 ± 121.6807.2 ± 51.4d788.1 ± 10.0720.5 ± 68.3b
Mannose203.7 ± 61.4214.9 ± 41.6233.0 ± 45.2186.1 ± 18.4
Rhamnose36.9 ± 4.933.1 ± 1.726.2 ± 10.426.9 ± 9.5
Fucose10.0 ± 5.38.1 ± 5.610.9 ± 6.67.1 ± 4.2
Glucuronic (mmol mg−1)1.57 ± 0.011.35 ± 0.011.41 ± 0.021.31 ± 0.04
Table 2.   Cellulose content of Columbia (wild-type), Atfla11, Atfla12, and Atfla11/12 stem bases
GenotypeCellulose content ± SE
  1. The cellulose content (μg cellulose mg−1 dry weight) of inflorescence stem bases, where the biggest differences in tensile strength and stiffness occurred, was measured. Values in parentheses are percentage cellulose. Atfla11/12 double mutants had significantly less cellulose than the other genotypes. Data are means from six biological replicates for each genotype followed by SE. t tests assuming unequal variance were used to calculate P values as a measure of statistical significance. Data that is signficantly different at  0.01 to a given genotype is indicated by an a for Columbia, b for Atfla11, c for Atfla12, d for Atfla11/12.

Columbia268.6 ± 5.9 (26.8%± 0.6)b,c,d
Atfla11237.1 ± 9.7 (23.7%± 1.0)a,d
Atfla12234.9 ± 6.0 (23.5%± 0.6)a,d
Atfla11/12196.2 ± 6.9 (19.6%± 0.7)a,b,c

Stem lignin content and composition were also measured. In stem cell walls, acid-insoluble lignin content was increased in the single mutants by approximately 3% and in Atfla11/fla12 stems by 1.7% (Figure 6a). As a proportion of the acid-insoluble lignin in wild-type stems, these changes equate to increases of 16–18% in the single mutants, and 9% in the double mutant. No other changes in lignin content were observed. Using Maule’s staining, no major qualitative difference in the deposition of syringyl lignin (fibres) or guaiacyl lignin (vessels) monomers was visible between wild-type, single or double mutant stems (Figure 6b), a finding reflected by the syringyl:guaiacyl monolignol ratios in the stem bases of the four genotypes (Figure 6c). The small increase in guaiacyl lignin in Atfla11 and Atfla11/12 stems may reflect the fact that AtFLA11 expression may have an impact on the levels of guaiacyl monolignols.

Figure 6.

 Lignin composition of Columbia (wild-type), Atfla11, Atfla12 and Atfla11/12 stem cell walls.
(a) Lignin content was measured in inflorescence stems. Atfla mutants had more acid-insoluble lignin than wild-type stems, with Atfla11/12 double mutants showing a smaller increase than that of the single mutants. No change in acid-soluble lignin was detected. Values are means for two technical replicates ± SD.
(b) Maule staining of stem cross-sections. Scale bars = 10 μm.
(c) Syringyl and guaiacyl monolignol monomers shown as percentages in dried stem bases (0–60 mm) ± SD as determined by thioacidolysis (n = 3). Only trace amounts of p-hydroxyphenyl (H lignin) were present (data not shown).

AtFLA11 and AtFLA12 expression affect cellulose MFA

Because wood stiffness is inversely correlated with MFA (e.g. Cave and Walker, 1994; Keckes et al., 2003), and because there was a correlation between MFA and FLA transcript abundance in eucalypts, we examined MFA in Atfla mutant stem tissues in which significant changes in stiffness and stem strength were observed. The Arabidopsis stem MFA was measured by developing a specialized X-ray diffraction method for Arabidopsis using a SilviScan-3, a machine that is routinely used to measure wood MFA in trees (Evans, 1999). Atfla11/fla12 stems had a significantly higher MFA than wild-type: a 6° difference in hypocotyl MFA (Figure 7a) and a 2° difference in MFA in stem bases (Figure 7b), but no detectable difference in mid-stems (Figure 7c). Furthermore, Atfla11/fla12 stems also had a significantly higher MFA than the corresponding single mutant stems (Atfla11 or Atfla12) (Figure 7b,c). Examples of X-ray diffraction patterns of high- and low-MFA stems (that differ by 5°) are shown in Figure 7(d). No difference in MFA was detected between Atfla11 and wild-type stem bases (Figure 7b). In contrast to the higher MFA of Atfla11/fla12, both Atfla11 and Atfla12 mid stems had a lower MFA than wild-type (Figure 7c), and this was also seen in Atfla12 stem bases (Figure 7b). The increase in MFA in Atfla11/fla12 stems suggests that the transcript abundance of both AtFLA11 and AtFLA12 may influence cellulose MF orientation.

Figure 7.

 MFA of Columbia (wild-type), Atfla11, Atfla12 and Atfla11/12 stems. The MFA was measured in (a) hypocotyls, (b) base of the stem and (c) middle of the stem using X-ray diffractometry as shown in (d), which shows X-ray diffraction patterns for samples with high and low MFA as seen by SilviScan-3 (Evans, 1999). Atfla11/12 double mutants had a significantly higher MFA than wild-type or single mutants in hypocotyls and stem bases. In contrast, MFA was frequently lower in single mutants than in wild-type and Atfla11/12 double mutant stems. Black bars, wild-type Columbia; light-grey bars, Atfla11; dark-grey bars, Atfla12; white bars, Atfla11/12. Measurements sharing the same symbol are significantly different at the following levels of probability: a,b, ≤ 0.005; c,d, ≤ 0.01; e,f, P ≤ 0.005), g, ≤ 0.001. Error bars = 2 × SE. (a) n = 16; (b,c) = 4.


In higher plants, a subset of single FAS domain FLAs appears to contribute to the biomechanical properties of stems through their impact on the synthesis and architecture of the secondary cell wall. In various angiosperms, there is high and specific transcript abundance of these genes largely in stems undergoing secondary cell-wall deposition, as demonstrated in Arabidopsis stems (Figures 1 and 2 and Figure S1a) (Brown et al., 2005; Ito et al., 2005; Persson et al., 2005), eucalypts (Figure 1), poplar (Lafarguette et al., 2004), zinnia (Figure S2) (Dahiya et al., 2006) and cotton (Huang et al., 2008) (Figure S2). The similarity in the tissue-specific FLA transcript abundance patterns observed for eucalypts and Arabidopsis was notable (Figure 1), as was the stem specificity of the Arabidopsis FLA promoter–GUS reporter constructs (Figure 2). To perform functional studies, we generated knockout Atfla11/12 plants, and found that these plants had no gross phenotypes (Figures 3 and 4). Because the transcript abundance of related FLA genes was correlated with wood properties in trees, we explored the possible involvement of this sub-group of group A FLAs in stem biomechanics. We performed tensile (uni-axial) and three-point flexural tests, determined MFA and measured the cell-wall composition in Arabidopsis knockout mutants. We found that Atfla11/12 stems had decreased tensile strength and stiffness, and that their cell walls had reduced cellulose, arabinose and galactose content, increased lignin content and an increased MFA. By relating the changes in biomechanical properties of the stem to the molecular changes in the plant stem extracellular matrix, we obtained insight into the potential molecular roles of these single FAS domain FLAs.

In a multi-component material such as the plant cell wall, tensile load is transferred from the matrix to the MFs by interfacial shear stress. The overall tensile stiffness is a global average of the stiffness of individual components. One important factor affecting stiffness is MFA (Cave and Walker, 1994; Keckes et al., 2003), together with matrix hemicellulose and lignin content (Kohler and Spatz, 2002). With increasing extension, the MFA decreases (Keckes et al., 2003), and as the load increases beyond the elastic regime, the weakest links in the sample begin to yield, leading to viscoelastic/plastic deformation. Increasing shear stress at the fibril/matrix interface can cause local rupture of the interface, resulting in the so-called ‘slip–stick Velcro’ mechanism (Keckes et al., 2003). Tensile strength is finally reached at the limit of the load-carrying capacity of the sample, and a major factor affecting tensile strength is cellulose content (Turner and Somerville, 1997; Li et al., 2003). Rupture either occurs in the matrix first, thus exposing MFs to higher loads, or in the MFs if the load-carrying capacity has decreased. At a low MFA, the cellulose molecules are aligned with the loading direction, resulting in high stiffness and strength. As the MFA increases, rupture is more likely to initiate in the matrix, the MFs peel apart and stiffness is reduced.

Stem tensile strength is affected by AtFLA11 and AtFLA12 transcript abundance because Atfla11/12 stems show a 30% reduction in tensile strength compared to wild-type (Figure 5). This is probably due to reduced cellulose synthesis, as cellulose is an extremely strong material, and stems of various mutants of Arabidopsis and rice with reduced cellulose content are significantly weaker (Turner and Somerville, 1997; Li et al., 2003). The 27% lower cellulose content in Atfla11/12 stem cell walls compared to wild-type (Table 2) can explain and account for the decreased tensile strength of Atfla11/12 stems. A likely site of action on cellulose synthesis is the plasma membrane, to which AtFLA11 and AtFLA12 are GPI-anchored in close proximity to the cellulose rosettes, as found for COBRA in Arabidopsis (Roudier et al., 2005). Interestingly, up-regulated FLA1 and 2 transcript abundance in eucalypt branches has been correlated with increased cellulose content (Qiu et al., 2008). It is unlikely that the reduction in tensile strength in the double mutant can be explained by the observed increase in MFA. As a 2° MFA increase is predicted to reduce tensile strength by only 1–3% (based on Burgert et al., 2002; Courchene et al., 2006), such a change in MFA in Atfla11/12 stems is more likely to be a consequence of altered cellulose/cell-wall deposition, rather than being a primary cause of the reduced tensile strength.

Stem tensile stiffness (modulus of elasticity) is also influenced by AtFLA11 and 12 transcript abundance, as Atfla11/fla12 stems show 15 and 26% decreases in stiffness (fresh and dry, respectively) compared to wild-type (Figure 5c,d). Similarly, in the upper side of eucalypt branches, in which EniFLA1,2 transcript abundance is up-regulated, wood stiffness is predicted to be higher. In this case, the prediction is based on the correlation between MFA and stiffness, and MFA is considerably reduced in the upper sides of branches (Qiu et al., 2008). We consider the possibility that these single FAS domain FLAs influence stem tensile stiffness partly by their effect on MFA, as well as their incorporation into the extracellular matrix. The 2° increase in MFA in dry Atfla11/fla12 stems (Figure 7) is likely to contribute to, but not fully account for, the 26% decrease in tensile stiffness, based on curves correlating MFA with stiffness (Page et al., 1977; Keckes et al., 2003). AtFLA11 and 12 could affect stem tensile stiffness through an influence of their AG side chains on the integrity of the hemicellulosic matrix, as, given the same MFA, a reduction in hemicellulose content by chemical maceration (Burgert et al., 2005) results in fibres with 25% of the stiffness of mechanically isolated fibres (Burgert et al., 2002). Also, extraction of hemicelluloses from Aristolochiamacrophylla schlerenchyma reduces the initial modulus of elasticity (Kohler and Spatz, 2002). The reduced cell wall arabinose and galactose content in Atfla11/fla12 stems could lower the shear strength of the H-bonded hemicellulose–MF interface, and be part of the proposed ‘stick–slip Velcro’ mechanism (Keckes et al., 2003).

FAS domain interactions between different group A FLAs, as shown between FAS domains 3 and 4 within the Drosophila Fasciclin-I protein (Clout et al., 2003), could contribute significantly to the modulus of elasticity properties of the extracellular matrix. Such interactions could generate an interlacing network providing shear strength. In animals, protein domain interactions are involved in mechanical stability, as found for the Ig domain in titin (the largest known protein in nature), a molecule that is critical in the elasticity of muscle (Kellermayer et al., 1997). In a similar way, the root swelling phenotype of Atfla4 Arabidopsis plants (Shi et al., 2003) may result from increased wall elasticity due to the point mutation in the FAS domain in Atfla4 located in the cell-wall matrix. Given that the FAS domain is approximately 150 amino acids long, and there is a high abundance of AtFLA11 and 12 transcripts in stems, it is possible that the FAS domain could be a significant component of stem cell walls. To examine this possibility, we attempted to visualize the subcellular localization of AtFLA11 and 12 proteins using transgenic plants expressing AtFLA11 or 12 GFP translational fusion proteins; however, the Arabidopsis stem autofluorescence at approximately 520 nm provided too much background interference (data not shown). In certain cells, heterodimeric interaction of AtFLA11 and 12 in stems could result in an optimal configuration for cellulose synthesis, MFA and matrix biomechanics. Homodimeric interactions could lead to quite different properties. This may partially account for the observation that Atfla11 and Atfla12 single mutant stems, in contrast to Atfla11/fla12, tended to have a decreased MFA and increased stiffness. In eucalypts, homodimeric interactions of the highly abundant EniFLA3 in the bottom side of branches could lead to reduced stiffness, and hetero- and homodimeric interactions of the highly abundant EniFLA1 and 2 in upper side of branches could lead to increased stiffness. Dimeric FAS protein interactions have been shown for periostin (Norris et al., 2007), and knockout periostin mice have reduced heart-tissue stiffness and skin with increased compliance (less stiffness), as well as altered bone and enamel deposition (Rios et al., 2005; Norris et al., 2007). It is intriguing that FAS-containing proteins are linked with tissue biomechanics in vertebrates and plants, leading to speculation that perhaps this function has been conserved to some degree across both plant and animal taxa.

Functionally, we consider that these FLAs affect stem biomechanics by affecting cellulose MF deposition, as well as by their presence within the extracellular matrix. However, certain biomechanical phenotypes observed in the mutants may be influenced by the increases in lignin content. Lignin is considered to be important in providing compressive and flexural (bending) strength (MacKay et al., 1997), which could explain why the single mutant stems, which had increased acid-insoluble lignin content (16–18% greater than wild-type) (Figure 6), also had increased flexural stiffness (Figure 5f). However, Atfla11/fla12 stems, which had a smaller increase in acid-insoluble lignin content (9% over wild-type) (Figure 6), showed no change in flexural stiffness (Figure 5f). In keeping with a role in resisting compressive forces, lignin has been considered to be a packing material in the spaces between the MF layers (Hepworth and Vincent, 1998). Lignin content has been observed to be proportional to tensile stiffness, for example antisense lignin (CCR) tobacco plants had less stiff wood in tensile tests (Hepworth et al., 1998). Thus, an increase in lignin content is predicted to cause an increase in tensile stiffness. However, although Atfla11/fla12 stems have increased lignin content, their tensile stiffness decreases (Figure 5c,d), indicating that this biomechanical phenotype is not explained by the change in lignin content. The lignin and cellulose pathways are often inversely proportional to each other, e.g. repression of lignin synthesis in transgenic poplars resulted in increased cellulose synthesis (Hu et al., 1999; Coleman et al., 2008). We consider that the increase in lignin content is more likely to be a consequence of the decrease in secondary-wall cellulose synthesis and/or the altered cell-wall three-dimensional architecture in Atfla stems.

The FAS domain is considered to be an ancient cell adhesion domain (Elkins et al., 1990; Huber and Sumper, 1994; Kim et al., 2000), and its conservation in plant cell-wall arabinogalactan proteins across diverse species raises questions about their mode of action. One possibility is that the group A FLAs described here may have become specialized to function as adhesion molecules between cellulose MFs within and between adjoining layers of the secondary cell-wall matrix. We propose that a subset of group A FLAs contribute to the strength and stiffness of load-bearing plant materials such stems in a partially redundant manner via their impact on cellulose synthesis and deposition, and also as structural components of the extracellular matrix. FLAs could be localized near the cellulose synthase rosettes through attachment to lipid rafts. Interestingly, FLAs and tubulins, another protein implicated in the orientation of cellulose MFs in secondary cell walls (Spokevicius et al., 2007), have been found to be attached to plasma membrane lipid rafts in Medicago truncatulata roots (Lefebvre et al., 2007). Alternatively, FLAs may affect MFA by performing a signalling role, linking newly forming cellulose MFs to the intracellular cortical microtubules, supporting frequent observations that MFs and microtubules are closely aligned. The predicted arabinose oligosaccharides of AtFLA12 (Figure 3b) may be important in this scenario as oligosaccharides may act as signalling molecules. After cleavage from the GPI anchor, the FLAs could contribute to cell-wall biomechanics as part of the extracellular matrix in which the cellulose MFs are embedded, thereby contributing to the shear strength of the matrix via FAS dimer interactions and H-bond interactions of the AG side chains with cellulose MFs and other matrix hemicelluloses.

Experimental procedures

Plant growth conditions

Arabidopsis thaliana (ecotype Col-0 CS60000) and T-DNA lines were germinated on MS medium (without sucrose), cold-stratified for 3–4 days, and seedlings were grown in a mixture of 35% peat moss, 46% vermiculite, 19% perlite, under quartz halide lamps (150 μmol photons m−2 sec−1 light), with a light/dark cycle of 8 h/16 h at 20°C, and watered every 2–3 days. Eucalyptus nitens was grown as previously described (Qiu et al., 2008).

Phylogenetic analysis

Mature protein sequences were obtained by removing signal peptides identified using the SignalP 3.0 server (http://www.cbs.dtu.dk/services/SignalP/) (Emanuelsson et al., 2007) and GPI lipid anchor sequences identified using the big-PI Plant Predictor (http://mendel.imp.ac.at/gpi/plant_server.html) (Eisenhaber et al., 2003). All accession numbers are listed in Figure S3. Protein sequences were aligned using Clustal W (Thompson et al., 1994), and the alignments were used as input into MEGA 4.0 (Tamura et al., 2007) to generate an unrooted phylogenetic tree using the neighbour-joining algorithm for Figure 1 and minimum evolution for Figure S2. Similar trees were obtained using algorithms other than those shown. All analyses were performed using the default parameters.

FLA transcript abundance

The tissue-specific transcript abundance of AtFLA11 and 12 was determined using quantitative real-time PCR analysis. Total RNA was extracted from various tissues from mature reproductive plants using an RNeasy plant mini kit (Qiagen, http://www.qiagen.com/), genomic DNA was removed using RQ1 RNase-free DNase (Promega, http://www.promega.com/), and RNA was confirmed to contain no genomic contamination by PCR using the oligonucleotides listed in Figure S4. First-strand cDNA was synthesized from 100 ng total RNA with oligo(dT)20 oligonucleotides using Superscript III (Invitrogen, http://www.invitrogen.com/). One-tenth diluted cDNA (5 μl) was used in 20 μl quantitative real-time PCR reactions performed in triplicate using gene-specific oligonucleotides (Figure S4) designed to have a Tm > 59°C and giving a product of approximately 150–200 bp and Platinum Taq DNA polymerase (Invitrogen), and performed in a Rotor-Gene 6000 machine (Qiagen). Cycling conditions were 5 min at 94°C, 45 cycles of 15 sec at 94°C, 15 sec at 60°C and 30 sec at 72°C, followed by a melting curve (55–99°C with a 5 sec hold at each temperature); fluorescence was acquired at the 72°C step and during the melting-curve program. Transcript abundance was calculated relative to the transcript abundance of the Arabidopsis FDH gene (Figures S3 and S4) (Bond et al., 2009), which shows little transcript abundance variation across the tissues tested, based on the Arabidopsis developmental data set of Schmid et al. (2005), and which has transcript abundance levels that fall within the range found here. The comparative quantification analysis method of the Rotor-Gene 6000 software was used for these calculations. Results are shown for three technical replicates of RNA from tissues of pools of six or seven plants per genotype; repeat experiments showed similar results.

Expression of EniFLA1, 2 and 3 was investigated by RNA blot analysis using tissues from a single 9-year-old Eucalyptus nitens tree (tree 347) using a probe generated from a Eucalyptus grandis cDNA as previously described (Qiu et al., 2008). Total RNA was extracted and Northern analysis was performed as previously described (Southerton et al., 1998). Blots were exposed to X-ray film for 6 h.

Generation of promoter–GUS transgenic plants and histochemical staining

The generation of ProAtFLA11:GUS and ProAtFLA12:GUS transgenic plants and histochemical staining was performed as previously described (Spokevicius et al., 2007), except that the Arabidopsis transgenic plants were generated using the floral dip method using the GV3101 strain of Agrobacterium tumefaciens (Clough and Bent, 1998). Gene-specific oligonucleotides used for the generation of these constructs are shown in Figure S4.

Isolation of T-DNA insertion mutants

Bulk Atfla11 seed (SALK_046976) (Alonso et al., 2003) was obtained from the Arabidopsis Biological Resource Center (ABRC), and bulk Atfla12 seed (SM.15162) (Tissier et al., 1999) from the Nottingham Arabidopsis Stock Centre (NASC). Lines were screened for homozygous T-DNA insertion plants by PCR genotyping using wild-type gene-specific and T-DNA oligonucleotides (Figure S4). The Atfla11/12 double mutant genotype was generated by crossing homozygous Atfla11 and Atfla12 plants and selecting homozygous double mutant progeny by PCR genotyping.

Stem histology

Freshly harvested stem and hypocotyl segments were hand-sectioned, stained using 0.05% toluidine blue O (pH 4.5), or stained for lignin using phloroglucinol (1% in 95% ethanol for 10 min, acidified with 37% HCl for 3 min, washed once briefly in water) or Maule staining (1% KMnO4 for 10 min, washed briefly in water, acidified with 37% HCl for 1 min, washed briefly in water, and incubated in 5% NaHCO3 for at least 5 min; Sibout et al., 2005). Sections were observed under a Leica DMR light microscope fitted with a Leica DC500 camera (Leica Microsystems, http://www.leica.com/).

Physical tests

Tensile and three-point flexural tests were performed using a 4500 series Instron universal testing machine (series IX automated materials testing system, http://www.instron.co.uk). Tensile strength was calculated as the maximum load required to break the stem within the gauge length divided by the cross-sectional area of each stem. The modulus of elasticity (stiffness) was calculated using Hooke’s law. Flexural three-point bending stiffness and strength were calculated according to standard equations. Figure S5 provides a detailed description of the tests, including equations and how the calculations were performed, and images of the custom-made apparatus.

Cell-wall analysis

Cell-wall monosaccharide as described by Coleman et al. (2008) was performed on dried stems harvested 0–120 mm from the base, unless otherwise stated. These same tissues were used to determine lignin monomer subunit composition, as per Robinson and Mansfield (2009). Determination of cellulose content, based on the method of Updegraff (1969), was performed in stems harvested 0–60 mm from the base. All the analyses were performed on dried stems harvested from plants that had completed their lifecycle, and that had senesced and dried out in their pots for more than 4 weeks.

MFA measurements

Microfibril angle measurements were performed using a using SilviScan-3 (CSIRO, http://www.csiro.au/) (Evans, 1999) and customized brass holders designed for Arabidopsis stems. The samples were clamped vertically across a window to allow passage of the X-ray beam. A tapered capillary was used to focus the beam to approximately 200 μm diameter, and the samples were scanned at 100 μm intervals. Gaps between the samples were used to assess the X-ray scattering background and to identify the individual samples. Approximately 20 samples were clamped in each holder.


We thank the Arabidopsis Biological Resource Center and Nottingham Arabidopsis Stock Centre for providing the Arabidopsis T-DNA insertion mutants. We thank Anthony A. Millar (The Australian National University), Robert Furbank, and Tony Ashton for critical suggestions on the manuscript. This project was supported by the Commonwealth Scientific Industrial Research Organization CSIRO (C.P.M., R.E., S.G.S.), Natural Sciences and Engineering Research Council (NSERC) of Canada grant (S.D.M.) and The Australian National University (Z.H.S).

Author contributions

CPM, SGS formulated the project; CPM RE SDM ZHS SGS conducted the experiments; all authors contributed to the interpretation of results, text and editing of the manuscript.