Deficient sucrose synthase activity in developing wood does not specifically affect cellulose biosynthesis, but causes an overall decrease in cell wall polymers


  • Lorenz Gerber,

    1. Umeå Plant Science Centre, Department of Forest Genetics and Plant Physiology, Swedish University of Agricultural Sciences, Umeå, Sweden
    Search for more papers by this author
    • These authors contributed equally to this work.
  • Bo Zhang,

    1. Department of Biomaterials, Max Planck Institute of Colloids and Interfaces, Potsdam, Germany
    Search for more papers by this author
    • These authors contributed equally to this work.
  • Melissa Roach,

    1. Umeå Plant Science Centre, Department of Forest Genetics and Plant Physiology, Swedish University of Agricultural Sciences, Umeå, Sweden
    Search for more papers by this author
    • These authors contributed equally to this work.
  • Umut Rende,

    1. Umeå Plant Science Centre, Department of Forest Genetics and Plant Physiology, Swedish University of Agricultural Sciences, Umeå, Sweden
    Search for more papers by this author
  • András Gorzsás,

    1. Umeå Plant Science Centre, Department of Forest Genetics and Plant Physiology, Swedish University of Agricultural Sciences, Umeå, Sweden
    Search for more papers by this author
  • Manoj Kumar,

    1. Umeå Plant Science Centre, Department of Forest Genetics and Plant Physiology, Swedish University of Agricultural Sciences, Umeå, Sweden
    Search for more papers by this author
  • Ingo Burgert,

    1. Department of Biomaterials, Max Planck Institute of Colloids and Interfaces, Potsdam, Germany
    2. Institute for Building Materials, ETH Zurich, Zurich, Switzerland
    3. Empa – Swiss Federal Laboratories for Material Testing and Research, Applied Wood Materials Laboratory, Dübendorf, Switzerland
    Search for more papers by this author
  • Totte Niittylä,

    Corresponding author
    1. Umeå Plant Science Centre, Department of Forest Genetics and Plant Physiology, Swedish University of Agricultural Sciences, Umeå, Sweden
    Search for more papers by this author
  • Björn Sundberg

    1. Umeå Plant Science Centre, Department of Forest Genetics and Plant Physiology, Swedish University of Agricultural Sciences, Umeå, Sweden
    Search for more papers by this author


  • The biosynthesis of wood in aspen (Populus) depends on the metabolism of sucrose, which is the main transported form of carbon from source tissues. The largest fraction of the wood biomass is cellulose, which is synthesized from UDP-glucose. Sucrose synthase (SUS) has been proposed previously to interact directly with cellulose synthase complexes and specifically supply UDP-glucose for cellulose biosynthesis.
  • To investigate the role of SUS in wood biosynthesis, we characterized transgenic lines of hybrid aspen with strongly reduced SUS activity in developing wood.
  • No dramatic growth phenotypes in glasshouse-grown trees were observed, but chemical fingerprinting with pyrolysis-GC/MS, together with micromechanical analysis, showed notable changes in chemistry and ultrastructure of the wood in the transgenic lines. Wet chemical analysis showed that the dry weight percentage composition of wood polymers was not changed significantly. However, a decrease in wood density was observed and, consequently, the content of lignin, hemicellulose and cellulose was decreased per wood volume. The decrease in density was explained by a looser structure of fibre cell walls as shown by increased wall shrinkage on drying.
  • The results show that SUS is not essential for cellulose biosynthesis, but plays a role in defining the total carbon incorporation to wood cell walls.


Cellulose is the main component of plant cell walls. It is synthesized at the plasma membrane by the heteromeric cellulose synthase (cesA) complex, which uses UDP-glucose as a substrate (Somerville, 2006). Cellulose biosynthesis in Populus wood is dependent on sucrose transport from source leaves and active sucrose import to cytosol of developing wood fibres (Mahboubi et al., 2013). Evidence from cotton fibres has suggested a central role for sucrose synthase (SUS) in subsequent sucrose cleavage to UDP-glucose and fructose. A large proportion of SUS in cotton fibres is associated with the membrane fraction (Amor et al., 1995), and in situ immunolocalization has suggested that SUS is located around the plasma membrane and cell wall in a pattern similar to cellulose microfibril orientation (Amor et al., 1995; Haigler et al., 2001). These data gave rise to a widely accepted model in which SUS associates directly with the cesA complexes, channelling UDP-glucose for cellulose biosynthesis (Haigler et al., 2001).

There are several observations supporting the SUS and cesA association model. Immunolabelled SUS in Azuki bean (Vigna angularis) epicotyls was found to be associated with plasma membrane structures resembling those reported for cesA (Fujii et al., 2010). Immunoprecipitation of cesA complexes from developing wood of a Populus deltoides × canadensis hybrid identified two SUS isoforms (SUS1 and SUS2), among c. 60 other proteins, suggesting an association between cesAs and SUS (Song et al., 2010). Reduced SUS activity in transgenic cotton plants repressed seed fibre initiation and elongation (Ruan et al., 2003). However, strong evidence against an essential role of SUS in cellulose biosynthesis has been obtained recently by Barratt et al. (2009). They discovered that a quadruple Arabidopsis sus1sus2sus3sus4 mutant, retaining SUS proteins (SUS5 and SUS6) only in the phloem tissues, showed no obvious growth phenotypes or cellulose deficiency (Barratt et al., 2009). This conclusion was challenged by Baroja-Fernández et al. (2012b), who claimed that the Arabidopsis sus1sus2sus3sus4 mutant had sufficient SUS activity from the remaining SUS5 and SUS6 to explain the lack of a strong phenotype. Therefore, the function of SUS in cellulose biosynthesis in particular, and cell wall biosynthesis in general, remains unclear (Baroja-Fernández et al., 2012a; Smith et al., 2012).

In developing wood of trees, SUS activity was found to be high in Populus canadensis (Schrader & Sauter, 2002), Robinia pseudoacacia (Hauch & Magel, 1998) and Pinus sylvestris (Uggla et al., 2001), suggesting an important role for SUS in wood cell wall biosynthesis. To investigate the role of the native SUS activity in developing wood of hybrid aspen (Populus tremula × tremuloides), we used RNAi to specifically reduce the level of the main wood-expressed SUSs (PttSUS1 and PttSUS2). Reduction of SUS activity to 4% of wild-type (WT) levels in developing wood of transgenic lines was found to have very minor effects on height and diameter growth, wood anatomy and the dry weight (DW) percentage composition of cellulose, lignin and hemicellulose in wood. However, some mechanical properties of wood were strongly altered, demonstrating a modification in cell wall structure. Further, we found that the wood density was decreased significantly. The decrease in wood density with no significant change in the percentage composition of cellulose, lignin and hemicellulose means that the reduced SUS activity resulted in a decrease in all major wood polymers per volume of wood. The data demonstrate that SUS is not essential in UDP-glucose production for cellulose biosynthesis. However, SUS activity is involved in defining the total carbon (C) incorporation to wood cell walls, and in ensuring an intact structure of the fibre wall.

Materials and Methods

Plant material and growth conditions

Hybrid aspen (Populus tremula L. × tremuloides Michx.) was grown in a glasshouse in a commercial soil–sand–fertilizer mixture (Yrkes Plantjord; Weibulls Horto, Hammenhög, Sweden) at 20 : 15°C (light : dark) in an 18 h : 6 h light : dark photoperiod. Trees were fertilized with c. 150 ml of 1% Rika-S (N : P : K = 7 : 1 : 5; Weibulls Horto) once a week after 3 wk of outplanting and grown to a height of c. 1.5 m. Harvested plant material was immediately frozen in liquid nitrogen and stored at −70°C. Developing wood used for gene expression and enzyme activity analysis was obtained by peeling the bark and scraping the exposed wood tissues as described in Gray-Mitsumune et al. (2004). Scrapings from internodes 20–29 (counted from the top) were used for metabolite measurements and enzyme activities, whereas scrapings from internodes 30–39 were used for gene expression analyses. Wood from internodes 30–39 was cleaved to remove the pith, freeze–dried for 48 h and used for pyrolysis coupled to gas chromatography/mass spectrometry (Py-GC/MS) and wet chemistry analysis. The bottom internodes (10–20 cm above the soil) were used for wood anatomy analysis. All analyses were performed on the same set of trees. Arabidopsis WT Col-0 and sus1sus2sus3sus4 mutants (used by Barratt et al., 2009) were grown in a 16 h : 8 h light : dark period at 22 : 18°C (light : dark) on a 3 : 1 mixture of soil (Yrkes Plantjord; Weibulls Horto) and vermiculite (Agra-vermiculite; Pull Rhenen, TX Rhenen, the Netherlands) for 7 wk. Hypocotyls and a basal 5-cm stem section (after excluding the bottom-most 1 cm) were freeze–dried for 48 h and used for Py-GC/MS.

SUSRNAi vector construction, hybrid aspen transformation and quantitative PCR (qPCR)

A fragment of the SUS1 coding sequence (corresponding to PU01211, was cloned into the binary RNAi vector pK7GWIWG2(I) (Karimi et al., 2002). Hybrid aspen was transformed and regenerated as described by Nilsson et al. (1992). Total RNA from developing wood was isolated using the Aurum RNA extraction kit (Bio-Rad, Hercules, CA, USA) and treated with the DNA-free kit (Ambion, USA) to remove residual DNA. Complete DNA digestion was confirmed by RNA template-based PCRs with ubiquitin primers. Reverse transcriptase reactions were performed with an iScript kit (Bio-Rad). Genorm (Vandesompele et al., 2002) was used to test five reference genes for stability of expression, and the three most stable transcripts (a 26s ribosome transcript, a tubulin transcript and an actin transcript) were selected for normalization. Primers used for expression analysis are specified in Supporting Information Table S1. PCRs were performed by the CFX96 Real Time System (Bio-Rad) and double-stranded DNA formation was detected using SYBR® Green (Bio-Rad). The following PCR program was used: 95°C for 3 min, followed by 40 cycles of 95°C for 30 s, 58°C for 15 s and 72°C for 30 s.

Mechanical tests

For the mechanical tests, a protocol was followed similar to former studies on modified poplar wood (Bjurhager et al., 2010; Hoenicka et al., 2012). Wood blocks were embedded in polyethylene glycol (PEG) with molecular weight 2000 to stabilize the fragile wood of the juvenile trees during the cutting process, which was carried out with a rotating microtome (RM 2255; Leica Microsystems, Wetzlar, Germany) in the longitudinal/radial direction. After cutting, the slices were washed with fresh water to dissolve the PEG and kept wet during the entire test. The dimensions of the specimens were c. 0.08, 1.5 and 20 mm in the radial (R), tangential (T) and longitudinal (L) directions, respectively. Measurements were performed in tension parallel to the grain until rupture of the specimens. The micro-tensile tester consisted of a motorized positioning system (Owis, Staufen, Germany) equipped with a load cell possessing a capacity of 50 N (Honeywell, Morris Township, NJ, USA). The strain rate was 2.5 μm s−1 and the gauge length was c. 10 mm. Force and elongation were recorded during the tensile experiment and strain was calculated on the machine path basis.

Wet chemical analysis of wood

Wood chemistry analysis on glasshouse-grown trees was performed according to Ona et al. (1995) with some modifications. The freeze–dried samples were ground in a centrifugal mill with a sieve size of 0.5 mm (Z200; Retsch, Haan, Germany) and extracted for 7 h with 1 : 2 ethanol : toluene, dried, flushed with deionized water (Milli-Q Advantage 10; Millipore AB, Solna, Sweden) and finally dried under reduced pressure. This material represents the extractive free wood. Holocellulose was produced by NaClO2 delignification. The amount of alpha-cellulose was determined gravimetrically after alkali dissolution of holocellulose. Klason lignin was determined on extractive free wood. The hemicellulose fraction was calculated by subtraction of all direct determined values.

Py-GC/MS chemical fingerprinting

Extractive free wood prepared as described above, freeze–dried Arabidopsis hypocotyls and stems were ball milled (MM400; Retsch) for 2 min at 30 Hz in stainless steel jars (1 ml) with one ball (7 mm); 30–40 μg of plant material were then applied to the on-line pyrolyser (PY-2020iD and AS1020E; FrontierLabs, Saikon, Koriyama, Fukushima, Japan) mounted on a GC/MS apparatus (7890A/5975C; Agilent Technologies AB Sweden, Kista, Sweden). Pyrolysis was conducted at 450°C and the pyrolysate was separated on a 30-m-long DB-5 capillary column (250 μm, 25 μm film, J&W; Agilent Technologies AB Sweden). The GC oven temperature program started at 40°C, followed by a temperature ramp of 32°C min−1 to 100°C, 6°C min−1 to 118.75°C, 15°C min−1 to 250°C and, finally, 32°C min−1 to 320°C. During each 19-min run, the full-scan spectra from 35 to 250 m/z were recorded. Data processing and analysis were performed as described in Gerber et al. (2012).

Monosaccharide composition analysis

Extractive free samples prepared as above were treated by methanolysis and trimethylsilyl derivatization (Biermann & McGinnis, 1989) for analysis by gas chromatography-flame ionization detection (GC-FID). Inositol (50 μl) was added as internal standard to extractive free sample aliquots of 200 μg as well as to four calibration samples consisting of the pure monosaccharide (glucose (Glc), arabinose (Ara), rhamnose (Rha), fucose (Fuc), galactose (Gal), galacturonic acid (GalA), glucuronic acid (GlcA), mannose (Man), xylose (Xyl) from Sigma-Aldrich, Stockholm, Sweden) at four different concentrations (1, 5, 20, 50 μg). The samples were methanolysed in dry methanol with 2 M hydrochloric acid at 85°C for 24 h. After two consecutive methanol cleaning steps, the samples were silylated with 200 ml hexamethyldisilazane (HMDS) + trimethylchlorosilane (TMCS) + pyridine, 3 : 1 : 9 (Sylon HTP, Supleco; Sigma-Aldrich) for 20 min at 80°C. Samples were desiccated, re-dissolved in hexane, centrifuged at 18 620 g (Universal 320; Hettich, Tuttlingen, Germany) and 1 μl was injected onto a capillary column (length, 30 m; Ø250 μm; film thickness, 25 μm; J&W DB-5; Agilent Technologies AB Sweden) installed in a GC-FID apparatus (7890A; Agilent Technologies AB Sweden). Calibration curves were constructed for every monosaccharide except 4-O-methyl-glucuronic acid, which is not commercially available. Instead, the detector response from GlcA was used to construct a calibration curve.

Density measurement

Wood bulk density was measured on a wet volume of harvested wood and on an oven-dry weight basis. Length, width and thickness were measured for each sample under a light microscope. Weight was measured using an analytical balance (Sartorius ME and SE microbalances, with a precision of 0.01 mg; Sartorious AG, Goettingen, Germany).

Anatomical parameter evaluation

Fibre wall area fraction was determined from 5-μm-thick cross-sections cut with a rotating microtome, washed with water and stained in Safranin-O solution for 2 min at room temperature. Grey-level images were taken under a light microscope and analysed using Image J software (Abramoff et al., 2004). Cross-sections of 10 μm thickness were used for measurements of fibre wall shrinkage on drying. Wet sections were covered with a thin glass slip and oven dried at 65°C for 20 h. The fibre wall thickness in wet and dry samples was measured from the same region.

Microfibril angle (MFA) measurements

The MFA measurements were performed on the same specimens as used for the mechanical tests by wide-angle X-ray scattering (Nanostar; Bruker AXS, Karlsruhe, Germany) with a sample–detector distance of 4.9 cm using CuKα radiation (wavelength, 0.154 nm). The diffraction patterns were collected with a two-dimensional (2D) position-sensitive (Hi-star) detector, with a measuring time of 60 min. The intensity was plotted against the azimuthal angle. MFAs were determined at three points of each sample.

Detection of SUS enzyme activity

The activity of SUS was detected in the direction of sucrose breakdown using a two-step end point assay that couples sucrose and UDP-dependent UDP-glucose production to NAD+ reduction at 340 nm. For the assay, 20 mg of powder from developing wood was extracted in a 25-fold (w/v) dilution of extraction buffer containing 100 mM Tris-HCl, pH 7.0, 2 mM EDTA, 5 mM dithiothreitol (DTT) and 3% (w/v) polyvinylpolypyrrolidone (PVPP). The extracts were incubated for 5 min in 20 mM Tris-HCl, pH 7.0, 100 mM sucrose and 4 mM UDP (sample to assay mix ratio, 1 : 9), after which the reaction was stopped by heating to 95°C for 5 min. The UDP-glucose content of the samples was then quantified in 100 mM Tris-HCl, pH 7.5, 2 mM NAD+ and 0.02 units UDP-glucose dehydrogenase (MyBioSource Inc., San Diego, CA, USA) using a spectrophotometer. This SUS assay is a modified version of the assay by Baroja-Fernández et al. (2012b). It measures SUS activity under the initial velocity conditions in neutral pH and without MgCl2. Baroja-Fernández et al. (2012a,b) determined these factors to be critical for a reliable SUS assay. In our protocol, the UDP-glucose content was determined by a one-step enzymatic assay instead of the high-performance liquid chromatography used by Baroja-Fernández et al. (2012b).

Metabolite profiling

Soluble sugars from developing wood were analysed as described previously (Roach et al., 2012). Hexose-phosphates and UDP-glucose were quantified from trichloroacetic acid (TCA) extracts prepared as described by Geigenberger et al. (1998).

Microarray sample preparation and data analysis

Total RNA from powdered developing wood was isolated with the Aurum RNA extraction kit (Bio-Rad). RNA was concentrated with the RNeasy MiniElute Cleanup kit (Qiagen, USA). Samples were hybridized on the Populus 385K array (Roche-NimbleGen, Sweden). Differentially expressed genes in transgenic lines were filtered by significance (P ≤ 0.05, t-test) and fold change ≥ 1.5 and ≤ −1.5.

Accession numbers

Sequence data from this article can be found in Phytozome P. trichocarpa genome sequence version 2.0 under the following numbers: POPTR_0018s07380, SUS1; POPTR_0006s13900, SUS2. Microarray data were deposited into the ArrayExpress database ( under the accession number E-MEXP-3974.


RNAi of SUS1 and SUS2 reduces SUS activity in developing wood

The Populus genome encodes seven SUS isoforms (Zhang et al., 2011). Expressed sequence tag (EST) sequencing and microarray data showed that the two paralogues SUS1 and SUS2 are by far the most abundant SUS transcripts in developing wood, and both show increased relative expression in this tissue (Hertzberg et al., 2001; Andersson-Gunnerås et al., 2006; Geisler-Lee et al., 2006; Zhang et al., 2011; Fig. S1). The data show that the SUS3, SUS4, SUS5, SUS6 and SUS7 transcripts are also present in developing wood, but with much lower abundance in comparison with SUS1 and SUS2 (Fig. S1a,c). SUS3 shows an increased relative expression in mature xylem, where ray cells would largely contribute to extractable RNA (Zhang et al., 2011; Fig. S1b). Thus, SUS1 and SUS2 are the key SUS isoforms expressed during wood cell wall biosynthesis.

To study the function of SUS1 and SUS2 in sucrose metabolism and secondary cell wall formation in wood, transgenic SUSRNAi aspen (Populus tremula × tremuloides) lines targeting both SUS1 and SUS2 were created and grown under glasshouse conditions to a height of c. 1.5 m. qPCR analysis showed that the single SUSRNAi construct reduced the transcript abundance of both SUS1 and SUS2 in developing wood to only a few per cent of WT levels in three independent lines (Fig. 1a,b). The minor expression of SUS3 in comparison with SUS1 and SUS2 was confirmed (Fig. S2). SUS3 expression was similar to WT in lines 1 and 2, and decreased slightly in line 3 (Fig. S2). The reduced transcript abundance of SUS1 and SUS2 resulted in a strong reduction in SUS activity in developing wood with a decrease to 6%, 13% and 4% of WT activity in lines 1, 2 and 3, respectively, providing evidence that these SUS isoforms account for most of the SUS activity in developing wood (Fig. 1c). All transgenic trees developed normally. Lines 1 and 2 were slightly taller and line 3 slightly shorter than WT (Table 1). The wood anatomy appeared normal in all transgenic lines as judged from light microscopy of transverse sections.

Table 1. Height and diameter 10 cm above the soil of wild-type (WT) and SUSRNAi hybrid aspen (Populus tremula × tremuloides) lines at harvest
  1. Mean ± SE, = 14 (WT) and = 7 (SUSRNAi) biological replicates. Student's t-test comparison with WT: *, < 0.05; **, < 0.01; ***, < 0.001.

Height (cm)145 ± 3157 ± 1**156 ± 3*131 ± 4**
Diameter (mm)10.0 ± 0.210.1 ± 0.210.2 ± 0.38.7 ± 0.2***
Figure 1.

SUS1 and SUS2 expression, and total sucrose synthase (SUS) activity, in developing wood of hybrid aspen (Populus tremula × tremuloides) SUSRNAi lines. (a) Relative expression of SUS1 in wild-type (WT) and SUSRNAi lines 1, 2 and 3. (b) Relative expression of SUS2 in WT and SUSRNAi lines 1, 2 and 3. (c) SUS activity in developing wood of WT and SUSRNAi lines. Activity was assayed in the sucrose cleavage direction as measured by UDP-dependent UDP-glucose production at pH 7.0. Values are means ± SE. For expression values, = 5 biological replicates; for SUS activity, = 6 biological replicates. Student's t-test comparison with WT: ***, < 0.001.

Reduced SUS affects the cell wall chemotype in hybrid aspen and Arabidopsis

A chemical fingerprint of wood from the transgenic lines and WT trees was obtained by Py-GC/MS according to Gerber et al. (2012). The data were evaluated by multivariate Orthogonal Projection of Latent Structures-Discriminant Analysis (OPLS-DA), a method specifically developed for the analysis of complex biological/chemical data, which makes use of all the information in the spectra, not only selected peaks (Bylesjö et al., 2006). A two-class multivariate OPLS-DA of integrated peaks gave a good model that separated the combined transgenic lines from WT trees (Fig. S3). This means that all transgenic lines differed in their chemotypes (as described by their Py-GC/MS chromatograms) from WT in a similar fashion. Chemical fingerprinting of wood by Py-GC/MS detected c. 200–300 pyrolytic degradation products, which reflected both the cell wall polymer abundance and structural features of the sample. Because such spectra contain more detailed information on cell wall chemistry than traditional extraction and quantification of the main wall components, we re-investigated the Arabidopsis sus1sus2sus3sus4 mutant used in the study by Barratt et al. (2009), employing Py-GC/MS combined with OPLS-DA. Similar to hybrid aspen wood, we found that SUS deficiency in Arabidopsis stem and hypocotyl tissues from 7-wk-old plants clearly resulted in a modified cell wall chemotype (Fig. S4).

Different lignin monomers and carbohydrates were quantified by peak integration of Py-GC/MS spectra according to Gerber et al. (2012). In hybrid aspen, the syringyl (S) lignin was slightly but significantly increased in all lines, whereas guaiacyl (G) lignin was slightly decreased, but was only statistically significant in line 3. Consequently, the S to G ratio was increased in the transgenic lines. No significant difference was observed for lignin and carbohydrate content (Table 2). A similar result was obtained for the hypocotyl tissues of the quadruple Arabidopsis sus mutant, whereas the stem tissues showed a statistically significant difference in lignin and carbohydrate content, as well as S to G ratio (Table S2).

Table 2. Total carbohydrates, lignin and lignin monomers in wood from wild-type (WT) and SUSRNAi hybrid aspen (Populus tremula × tremuloides) lines derived from pyrolysis-GC/MS analysis
  1. Values are presented as percentages of the total peak area of peaks attributable to carbohydrates (C; cellulose and hemicellulose), lignin (L), syringyl (S) and guaiacyl (G) monomers, and resulting S to G ratio. Peak annotation according to Gerber et al. (2012). Mean ± SE, = 10 WT and = 7 SUSRNAi biological replicates. Student's t-test comparison with WT: *, < 0.05; **, < 0.01; ***, < 0.001.

C75.4 ± 0.873.9 ± 0.874.3 ± 0.774.6 ± 0.6
G8.8 ± 0.48.2 ± 0.38.3 ± 0.27.6 ± 0.1*
S11.8 ± 0.313.6 ± 0.5**13.4 ± 0.5**13.7 ± 0.5**
L22.6 ± 0.724.0 ± 0.823.7 ± 0.823.2 ± 0.7
S : G1.4 ± 0.01.7 ± 0.0***1.6 ± 0.0***1.8 ± 0.04***

Reduced SUS alters the structure of wood cell walls

Micro-tensile tests were performed to investigate whether reduced SUS alters wood structure. Stress–strain curves show a pronounced difference in the mechanical behaviour between SUSRNAi lines and WT trees (Fig. 2). Two mechanical parameters, stiffness and ultimate stress, were determined and both parameters were found to be reduced by c. 50% in all SUSRNAi lines (Table 3). The average stiffness value was c. 1.2 GPa in transgenic wood of SUSRNAi lines, compared with 2.4 GPa in WT. Ultimate stress ranged from 10.7 to 13.2 MPa in SUSRNAi lines, compared with 24.9 MPa in WT.

Table 3. Wood mechanical properties and microfibril angle of wild-type (WT) and SUSRNAi hybrid aspen (Populus tremula × tremuloides) lines
  1. Mean ± SE. For stiffness and ultimate stress, = 5 WT and = 3 SUSRNAi biological replicates. For microfibril angle, = 9 WT and = 4 SUSRNAi biological replicates. Student's t-test comparison with WT: *, < 0.05; **, < 0.01; ***, < 0.001.

Stiffness (GPa)2.4 ± 0.071.2 ± 0.08***1.2 ± 0.05***1.2 ± 0.1***
Ultimate stress (MPa)24.9 ± 1.6413.2 ± 3.71*11.3 ± 0.67**10.7 ± 1.53**
Microfibril angle (degrees)15.5 ± 0.3519.6 ± 0.66***18.9 ± 0.77***20.1 ± 1.46**
Figure 2.

Mechanical properties of wild-type (WT) and SUSRNAi hybrid aspen (Populus tremula × tremuloides) wood. (a) Typical stress–strain curves of WT (black line) and SUSRNAi (green line) wood loaded in uniaxial tension. Note the lower slope of the curve (stiffness), the lower strength (maximum stress level) and the lower strain to fracture (sample failure after less deformation) of SUSRNAi relative to WT. (b) Typical fracture surfaces of WT and SUSRNAi wood. Note the pronounced zigzag crack path of the WT sample with a combination of crack propagation parallel (solid arrow) and perpendicular (open arrow) to the fibre direction, whereas, in SUSRNAi, crack propagation occurred predominantly perpendicular to the fibre direction.

In terms of other aspects of cell wall structure, a significant change in the cellulose microfibril orientation was measured. The MFA in the wood fibres was increased by c. 4° in all transgenic lines compared with the WT (Table 3). WT samples possessed a mean MFA of 15.5°, whereas MFA in the SUSRNAi lines was in the range 18.9–20.1°.

Reduced SUS causes a decrease in the biosynthesis of all wood cell wall polymers, resulting in a decreased wood density

Wet chemical analysis of wood composition according to Ona et al. (1995) showed no significant changes in the proportion of α-cellulose, hemicelluloses and lignin in transgenic lines compared with WT trees, with a small corresponding increase observed in Klason lignin (Fig. 3a). However, similar to the Py-GC/MS data, this difference was not statistically significantly different. Total cell wall hydrolysis and measurement of monosaccharide content by GC/MS showed consistent, but small, differences, with a relative increase in Rha and Gal in the transgenic lines and a corresponding decrease in Xyl and GlcA (Fig. S5).

Figure 3.

Wood composition of wild-type (WT) and SUSRNAi hybrid aspen (Populus tremula × tremuloides) lines. (a) Proportion of wood fraction expressed as milligrams per gram dry weight (DW). (b) Amount of extractable cell wall polymer per volume of wood, derived from the difference in wood density between SUSRNAi and WT. The amount was calculated as g kg−1 DW of polymer × kg m−3 of wood. Mean ± SE,= 12 (WT) and = 7 (SUSRNAi) biological replicates. Student's t-test comparison with WT: *, < 0.05; **, < 0.01; ***, < 0.001.

To investigate whether SUS deficiency affects the overall C incorporation into wood cell walls, we measured wood density based on oven dry mass and wet volume. This showed a decrease in wood density in the three SUSRNAi lines, ranging between 13% and 18%, compared with WT (Table 4). Despite the similar diameter growth, a decrease in SUS activity in developing wood negatively affects the dry matter content of the wood cell walls. The wood density estimate was then used to calculate the content of cellulose, lignin and hemicellulose per volume of wood (Fig. 3b). This calculation provides an estimate of the absolute content of each polymer, as opposed to the calculation of the DW percentage composition. From the estimation of the absolute polymer content per volume of wood, all major wood polymers are decreased, with a more pronounced effect on cellulose and hemicellulose (Fig. 3b).

Table 4. Wood density and anatomical parameters in wild-type (WT) and SUSRNAi hybrid aspen (Populus tremula × tremuloides) lines
  1. Wall thickness was measured across adjacent fibre walls. Mean ± SE, = 4–5 (WT) and = 3 (SUSRNAi) biological replicates. For fibre wall thickness, mean ± SE is shown. Student's t-test comparison with WT: *, < 0.05; **, < 0.01; ***, < 0.001.

Density (kg m−3)295.8 ± 2.8243.9 ± 2.8***256.9 ± 7.6**246.2 ± 21.8*
Fibre area (% of total area)80.7 ± 0.780.9 ± 0.682.8 ± 1.480.5 ± 1.4
Fibre wall area (% of total fibre area)41.8 ± 0.638.0 ± 0.5**39.4 ± 0.7*36.0 ± 2.3*
Wet fibre wall thickness (μm)3.9 ± 0.033.7 ± 0.03**3.8 ± 0.063.9 ± 0.02
Dry fibre wall thickness (μm)3.5 ± 0.082.9 ± 0.07***3.0 ± 0.05**3.1 ± 0.04***
Fibre wall shrinkage after drying (%)10202020

Reduced SUS results in a looser cell wall structure

Variations in wood density can arise from differences in the overall lumen to cell wall ratio, and will therefore be affected by the proportion of vessels and fibres. However, the relative fibre area in SUSRNAi lines was similar to that in WT, indicating that the density differences were not caused by an altered ratio between the fibre and vessel fractions (Table 4). Variations in density can also arise from differences in cell wall thickness. Interestingly, measurement of the fibre cell wall area in the wet state showed a decrease between 6% and 13% in SUSRNAi lines compared with WT (Table 4). Because, the decrease in fibre cell wall area in the SUSRNAi lines was smaller than the decrease in density, and because the large majority of bulk density is dependent on the fibre cell wall, it is possible that the decrease in density of SUSRNAi can be explained by a decrease in the density of the fibre cell wall per se. In support of this, the quantification of the fibre wall shrinkage on drying by measurement of the thickness of wet and dried fibre walls showed that WT wall thickness was reduced by 10% and SUSRNAi lines by 20% on drying (Table 4). Thus, SUSRNAi fibre walls were not only thinner, but also had a looser cell wall structure, compared with WT.

It can be concluded from chemical fingerprinting, micromechanical and MFA analysis that reduced SUS activity results in a modification of chemistry and structure of the wood. Quantitative chemical analysis in combination with density measurements showed that SUS deficiency caused a decrease in all major wood cell wall polymers (cellulose, lignin and hemicelluloses; Fig. 3b, Table 4).

Soluble neutral sugars accumulate in the developing wood of SUSRNAi lines

The levels of central primary metabolism sugars were quantified in developing wood of lines 1 and 3 to investigate the effect of decreased SUS activity. Sucrose, Glc and fructose were all increased in the transgenic lines compared with WT (Fig. 4a). The levels of hexose phosphates and UDP-glucose were decreased in line 3 (Fig. 4b), and a similar trend was also observed in line 1. This suggests that the transgenic lines had less C available for wood polymer precursors.

Figure 4.

Sugar metabolites in developing wood of wild-type (WT) and SUSRNAi hybrid aspen (Populus tremula × tremuloides) lines. WT, black bars; SUSRNAi-1, light grey bars; SUSRNAi-3, dark grey bars. (a) Glucose (Glc), fructose (Fru) and sucrose (Suc) content. (b) Hexose phosphate and UDP-glucose (UDP-Glc) content. Glucose-6-phosphate (Glc6P), fructose-6-phosphate (Fru6P), glucose-1-phosphate (Glc1P). Mean ± SE,= 6 biological replicates. Student's t-test comparison with WT: *, < 0.05; **, < 0.01; ***, < 0.001.

Effect of SUSRNAi on the wood transcriptome

To investigate the effect of SUS1 and SUS2 suppression on the wood transcriptome, the whole-genome transcript profiles in developing wood of SUSRNAi lines 1 and 3 and WT were compared using Nimblegen 385K oligo arrays. The data confirmed that SUS1 and SUS2 were the most highly expressed SUSs in developing wood (Fig. S1c), and that their expression was reduced in the RNAi lines (Table S3).

Sixty genes were differentially expressed in both transgenic lines compared with WT (fold change cutoff of ± 1.5). Of these, 30 genes showed increased and 30 genes decreased transcript levels (Table S3). The differentially expressed genes were analysed according to functional class gene ontology (GO) annotation using AgriGO (Du et al., 2010), but this did not reveal significant enrichment in any functional class.

Among the 60 differentially expressed genes, some were related to cell death/stress or cell wall integrity sensing. These transcript changes pointed to an induction of signalling cascades in response to SUSRNAi and altered cell walls. Similar changes in defence/stress genes have been reported previously for transgenic aspen lines showing defects in wood cell walls as a result of reduced FRUCTOKINASE 2 (FRK2) activity (Roach et al., 2012). A closer look was taken into the differentially expressed transcripts related to carbohydrate metabolism and cell walls. Apart from SUS1 and SUS2, the transgenic lines did not show any altered expression of annotated genes related to sucrose metabolism, with the possible exception of a transcript annotated as an invertase/pectin methylesterase inhibitor family protein, which was reduced in both lines. No difference in expression was detected for genes known to be related to cellulose, hemicellulose or lignin biosynthesis. There was, however, a decrease in the expression of primary wall-modifying enzymes, an expansin and a pectin lyase, and an increase in a fucosyltransferase (FUT11). These results show that a strongly reduced activity of SUS, resulting in changes in the cell wall structure and composition, does not result in major changes in the transcriptome.


The discovery of a membrane-localized SUS in developing cotton fibres (Amor et al., 1995) led to the concept of a direct association between SUS and cesA complexes to provide UDP-glucose for cellulose biosynthesis (Haigler et al., 2001). However, we show here that there is no clear difference in xylem anatomy or in the DW percentage of cellulose in wood cell walls of transgenic hybrid aspen with only residual SUS activity in developing wood (Figs 1c, 3a, Tables 2, 4). Our data therefore support the observation that a quadruple Arabidopsis sus1sus2sus3sus4 mutant showed normal growth and no difference in xylem anatomy or in the DW percentage of cellulose in stems compared with WT plants (Barratt et al., 2009). The interpretation of the early results by Amor et al. (1995) is further complicated by the recent discovery that, instead of one SUS isoform, developing cotton fibres of tetraploid Gossypium hirsutum express four SUS isoforms (Brill et al., 2011), and those of diploid cotton Gossypium arboreum five SUS isoforms (Chen et al., 2012). Intriguingly, the cotton SUS isoform most closely associated with the timing of secondary cell wall formation and cellulose biosynthesis (SusC) was mainly localized to the cell walls of cotton fibres (Brill et al., 2011). This led Brill et al. (2011) to suggest that most of the previous SUS immunolocalization signal around the plasma membrane of cotton fibres should, in fact, be attributed to apoplasmic SusC. The observation that mutant Arabidopsis and Populus plants deficient in SUS activity are not specifically inhibited in cellulose biosynthesis does not exclude a function of SUS in cellulose biosynthesis of WT plants. However, the hypothesis of an essential function for SUS in providing UDP-glucose to the cesA complexes is yet to be supported. Rather than having a specific effect on cellulose biosynthesis, we found that SUS deficiency reduced wood density as a consequence of a decrease in all major wood polymers (Fig. 3b). We further demonstrated that the stem tissues of the Arabidopsis sus1sus2sus3sus4 mutant showed a cell wall chemotype similar to hybrid aspen SUSRNAi wood (Table S2). Therefore, we propose that SUS has a similar function in developing xylem tissues in both Arabidopsis and Populus.

The reduced wood density in the SUSRNAi lines (Table 4) establishes a role for SUS in wood cell wall biosynthesis. This conclusion is in line with previous observations from sink tissues of other plant species with reduced SUS activity. Examples include transgenic potato with reduced tuber SUS activity, causing a reduced starch content and tuber DW (Zrenner et al., 1995), and the rugosus4 (rug4) mutant of pea with reduced SUS activity in the seed, resulting in reduced seed starch content (Craig et al., 1999). Arabidopsis SUS isoforms 2 and 3 have been shown to be required for transient starch accumulation in developing Arabidopsis seeds (Angeles-Núñez & Tiessen, 2010). The shrunken1 (sh1) mutant of maize, with almost complete absence of SUS activity in developing endosperm, caused a reduction in starch content and cell wall thickness (Chourey et al., 1998). Although cellulose content was not measured, Chourey et al. (1998) interpreted the endosperm cell wall defects to be caused by the inhibition of cellulose biosynthesis. However, the SUS defects in developing maize endosperm may have equally well been caused by a general reduction in C incorporation into cell walls, instead of specific inhibition of cellulose biosynthesis.

The overexpression of a cotton (G. hirsutum) SUS under a constitutive 35S or vascular 4CL promoter in the Populus alba × grandidentata hybrid led to a small (2–6% per gram DW) increase in wood cellulose proportion, suggesting that it is possible to increase relative cellulose content through increased SUS activity (Coleman et al., 2009). However, these trees also had up to 27% increased wood density, suggesting a considerable increase in the absolute content of all polymers in the wood cell walls. However, the fact that total biomass was reported not to be affected suggests that the SUS-overproducing trees were smaller than the corresponding WT trees. In another study, expression of a modified mung bean (Vigna radiata) SUS under a 35S promoter in Populus alba did not result in a change in the percentage composition of cell wall polymers on a DW basis, but did enhance the incorporation of isotope-labelled sucrose into cellulose and xyloglucan (Konishi et al., 2004). These results suggest that increased SUS activity in developing wood only causes a modest increase in cellulose biosynthesis, whereas the major effect is on increased wood density.

The sink strength of developing wood is largely determined by its capacity to import sugars from the phloem, and subsequent incorporation of C into cell wall polymers. We observed an increase in sucrose and hexose pools in the developing wood of SUSRNAi trees, indicating that the C supply from the phloem was not inhibited (Fig. 4a). At the same time, hexose phosphates and UDP-glucose pools were decreased (Fig. 4b). Hence, one possibility is that the reduction in all main wood polymers in SUSRNAi trees may be caused by a reduction in the catabolism of sucrose and in the subsequent production of hexose phosphates and nucleotide sugars. However, the exact mechanism of wood density reduction in response to SUSRNAi remains to be discovered. In addition to SUS, sucrose in plants can also be catabolized by invertase (Sturm & Tang, 1999). Our observation that tree growth was maintained, despite the strong reduction in SUS activity, points to invertases as an important sucrose-cleaving mechanism in developing wood.

The reduced amount of C used for wood biosynthesis in SUSRNAi lines not only resulted in a decreased density, but also in an altered cell wall ultrastructure that was manifested by a change in cellulose MFA and in a different mechanical performance of the wood (Fig. 2, Table 3). Cell wall density per se was decreased, that is SUSRNAi cell walls possessed smaller amounts of cell wall polymers per volume than did WT wood. This was evident from the higher degree of cell wall collapse (shrinkage) from the wet to dehydrated state (Table 4). The decreased density, together with the increased MFA, contributed to the observed lower stiffness and ultimate stress of the samples (Salmén & Burgert, 2009). An increase in MFA, however, usually results in higher strain to fracture and in a less brittle fracture behaviour (Reiterer et al., 1999). However, the SUSRNAi lines broke in a very brittle manner at relatively low strain to fracture (see stress–strain curves in Fig. 2a). Under uniaxial tensile tests, cracks can propagate either parallel to the fibre direction (predominately along the middle lamella) or perpendicular to the fibre direction and fracture the cell walls. The image of the fractured WT sample (Fig. 2b) displayed a typical fracture pattern of hardwood under longitudinal tensile loading, with a combination of both crack propagation mechanisms, whereas SUSRNAi lines showed predominantly crack propagation perpendicular to the fibre. Perpendicular crack propagation occurs more easily through low-density than high-density tissues (Frühmann et al., 2003). Thus, the perpendicular crack pattern in SUSRNAi lines can most probably be explained by the lower cell wall density, which eases the transverse fracture of cell walls.

No major effects on tree growth and development, and hence no major effects on cell division or cell expansion, were observed in the SUSRNAi lines (Table 1). Studies with Arabidopsis primary cell wall mutants and inhibitors of cellulose biosynthesis have shown that plants possess specific survey mechanisms to detect changes in cell wall structure (Hematy et al., 2007; Hamann et al., 2009). The phenotypes of primary cell wall mutants are probably partly caused by these survey mechanisms. Our transcriptional profiling of SUSRNAi trees and, previously, FRK2RNAi trees (Roach et al., 2012) suggested that similar survey mechanisms exist in wood. However, the results also showed that they do not have a major systemic effect on the primary wall-forming expansion zone of wood or meristems in general. This is an encouraging observation in relation to current efforts to modify secondary walls for industrial purposes, in that even a dramatic change in structural properties of wood secondary cell walls has no major effects on tree growth and development.

The example of SUSRNAi wood illustrates the importance of the measurement of wood density when quantifying cell wall polymers. No difference in wood polymers between WT and SUSRNAi lines would have been observed if the data had only been normalized against DW, as is usually the case. From our observation, that is that a strongly reduced SUS activity in developing wood reduced wood density, SUS can be concluded to have a role in the metabolism of sucrose during cell wall biosynthesis. However, it is clear that 2-month-old trees grown under glasshouse conditions are able to use alternative pathways to produce cell wall biosynthesis precursors from sucrose, including UDP-glucose, for cellulose biosynthesis.


We thank Lenore Johansson for assistance with transmission electron microscopy, and Professor Alison Smith for kindly providing the Arabidopsis sus1sus2sus3sus4 seeds. We also thank SweTree Technologies for production of the SUSRNAi lines. This work was supported by grants from the Swedish Research Council FORMAS (FuncFiber/Bioimprove: Centre of Excellence in Wood Science), Vetenskapsrådet, the Swedish Energy Agency, EU Programs Renewal (FP7/2007–2013) and CASPIC (FP7/2007-2010), VINNOVA, the Kempe Foundation and Bio4Energy, the Swedish Programme for renewable energy.