Visualization of cellulose synthases in Arabidopsis secondary cell walls

See allHide authors and affiliations

Science  09 Oct 2015:
Vol. 350, Issue 6257, pp. 198-203
DOI: 10.1126/science.aac7446

Secondary cell walls built with speed

Plant cell walls provide the cellulose that is integral for wood, cotton fiber, and many biofuels. Cellulose is synthesized outside the cell membrane by cellulose synthase enzymes. Much of the secondary cell wall, responsible for the sturdiness of wood, is formed by xylem cells embedded in the core of the plant. Watanabe et al. leveraged ectopic expression to bring xylem-style cellulose synthase activity to the epidermal surface of the plant (see the Perspective by Schneider and Persson). Combining this improved accessibility with fluorescent tagging showed that secondary cell walls are built faster than primary cell walls, perhaps due to increased velocity and density of cellulose synthase complexes.

Science, this issue p. 198, see also p. 156


Cellulose biosynthesis in plant secondary cell walls forms the basis of vascular development in land plants, with xylem tissues constituting the vast majority of terrestrial biomass. We used plant lines that contained an inducible master transcription factor controlling xylem cell fate to quantitatively image fluorescently tagged cellulose synthase enzymes during cellulose deposition in living protoxylem cells. The formation of secondary cell wall thickenings was associated with a redistribution and enrichment of CESA7-containing cellulose synthase complexes (CSCs) into narrow membrane domains. The velocities of secondary cell wall–specific CSCs were faster than those of primary cell wall CSCs during abundant cellulose production. Dynamic intracellular of endomembranes, in combination with increased velocity and high density of CSCs, enables cellulose to be synthesized rapidly in secondary cell walls.

Cellulose, the most abundant biopolymer on Earth, is a key biomechanical component of land plants and a valuable natural resource. Cellulose in the primary cell wall, which is laid down during plant growth, determines plant shape (1). However, the bulk of terrestrial biomass is composed of the cellulose in secondary cell walls, which are laid down after the cell has stopped growing to strengthen plant vasculature and structure (2). The strength of these walls is derived from the organization of cellulose microfibrils, which, relative to primary cell walls, possess cellulose with a higher degree of polymerization, increased microfibril crystallinity, and a higher degree of microfibril organization (2, 3).

Cellulose is synthesized at the plasma membrane by cellulose synthase (CESA) enzymes that are organized in multiprotein cellulose synthase complexes (CSCs) (4). In Arabidopsis thaliana, 10 CESA isoforms exist, with CESA1, CESA3, and CESA6 involved in primary cell wall synthesis (5, 6), and CESA4, CESA7, and CESA8 required for secondary cell wall production (7). At least three distinct isoforms are required for normal cellulose production (57). Live-cell imaging of primary cell wall CSCs clearly shows CESA velocity and distribution at the plasma membrane as well as intracellular trafficking (810). Secondary cell walls are produced in vasculature and fibers deep within plant tissues, limiting the resolution of live-cell imaging (11, 12). Here, we visualized cellulose synthesis in secondary cell walls using fluorescently tagged CESA7 in a unique system where live epidermal cells are induced to form secondary cell walls ectopically (13).

Arabidopsis plants were engineered to constitutively express a transcription factor controlling xylem tracheary element cell fate, VASCULAR NAC-DOMAIN7, fused to a glucocorticoid receptor (VND7::GR). When these plants are exposed to a glucocorticoid hormone such as dexamethasone, cells are induced to become protoxylem-like tracheary elements (13). After induction, the secondary cell wall cellulose synthase genes were transcriptionally up-regulated by a factor of 1.5 to 3.0, whereas primary cell wall–related cellulose synthase genes were transcriptionally unaffected (table S1). Although it is not possible to know whether the VND7::GR levels in these plants are comparable to endogenous VND7 in protoxylem tracheary elements, the induced cells have features typical of developing protoxylem, such as secondary cell wall thickenings alternating with primary cell wall domains, in spiral or annular patterns. To investigate cellulose deposition, we crossed plants carrying the VND7::GR induction system with cesa7/irx3-4 null mutants complemented with a functional fluorescently tagged CESA7 (YFP::CESA7) driven by its native promoter (fig. S1). Fluorescent signal from CSCs was detectable in induced epidermal cells, although identification of individual plasma membrane–localized CSCs was possible only with the use of an optimized spinning disk confocal imaging system (optimization parameters are described in fig. S2).

Use of this optimized system permitted the visualization of discrete YFP::CESA7 particles in the plasma membrane of induced protoxylem tracheary elements (Fig. 1). Their linear movement at slow and steady velocities meets the criterion of actively synthesizing CSCs (movie S1) (810, 14). CSCs moved in a bidirectional fashion along plasma membrane domains underlying secondary cell wall thickenings. Relative to labeled primary cell wall CSCs (810), secondary cell wall CSCs showed a higher density at the plasma membrane, with overlapping signals from individual particles largely indistinguishable over extended areas (Fig. 1, B and C, and fig. S3). Moreover, CSC distribution changed over the course of secondary cell wall development. Early in development, CESA7 signal was observed across the plane of the plasma membrane on tracks defined by bands of microtubules forming in the cell cortex (Fig. 1A, merged image, and table S2). As secondary cell wall synthesis progressed (Fig. 1B), CESA7 signal was enriched in regions of the plasma membrane associated with tight bundles of microtubules (Fig. 1B and table S2) (15). In late secondary cell wall development (Fig. 1C), the YFP::CESA7 signal was apparent as U-shaped plasma membrane furrows curving around the secondary cell wall thickenings. The majority of the CESA7 signal was evenly distributed and restricted to these curved domains (Fig. 1C, inset). When VND7::GR-induced cells were cryofixed and examined by transmission electron microscopy (TEM), tangential sections along the cell surface revealed the curved domain of plasma membrane around secondary cell walls, as well as their associated microtubules in the cell cortex (Fig. 1D).

Fig. 1 CESA7 and microtubules are constrained to secondary cell wall domains during protoxylem development.

(A) Early development: YFP::CESA7-labeled CSCs and RFP::TUB6-labeled microtubules localize diffusely across the plasma membrane. (B) Mid-development: Narrow microtubule bundles lie underneath tight tracks of membrane-localized CSCs. (C) Late development: Apparent concentrations of YFP::CESA7 and RFP::TUB6 at the edges of forming secondary cell walls are revealed as uniform signal in subcortical optical sections. YFP::CESA7-labeled Golgi (arrowhead) are visible because of the increased depth of imaging through secondary cell wall thickenings. (D) TEM micrographs highlight plasma membrane (PM) curvature over secondary cell walls (2CW), lined by cortical microtubule (MT) bundles. Scale bars, 10 μm [(A) to (C)], 500 nm (D).

The rate of cell wall deposition is the consequence of both the concentration of CSCs and the rate at which each CSC produces cellulose. The importance of the restriction of the CSCs to the curved membrane domains is that cellulose deposition is concentrated in a discrete area of intense production. Such concentration may help to explain why secondary cell walls are synthesized more quickly than primary cell walls (16). With live-cell imaging, we measured deposition rates of individual CSCs. We assumed that CSC movement through the plasma membrane is a function of glucan chain biosynthesis and its subsequent crystallization into cellulose microfibrils, such that faster CSC velocities would reflect faster rates of synthesis (3, 17). To facilitate this comparison, we quantified the velocities of YFP::CESA7–labeled CSCs at the cell membrane and primary cell wall CSCs (GFP::CESA3) (9, 10) under identical growth and imaging conditions, alternating data collection between plant lines in an imaging session to control for possible environmental effects (Fig. 2). Time projections over a 5-min data collection period showed CSC tracks across the plasma membrane (Fig. 2, A and B). These tracks defined lines for kymograph analysis of particle velocity. In primary cell wall CSC kymographs, the tracks of each GFP::CESA3 were distinct (Fig. 2A). The tracks of CESA7 were more difficult to discern because of their higher density (Fig. 2B and fig. S3). The slope of the lines in the kymograph (Fig. 2, A and B, arrows) represents the CSC velocities, with steeper slopes indicating faster CSC movement (more displacement on the spatial axis). The average velocity of CESA7-containing CSCs producing secondary cell walls was 265 ± 75 nm/min [mean ± SD, n = 36 cells (40 CSCs per cell) from 12 plants] (Fig. 2D). Velocity changed over the course of protoxylem differentiation, which was divided into stages according to microtubule banding and secondary cell wall features. Average CESA7 velocity was 293 ± 29 nm/min during early development, 327 ± 37 nm/min during mid-development, and 187 ± 38 nm/min during late development (Fig. 2E). In contrast, primary cell wall CESA3-containing CSCs displayed an average velocity of 231 ± 34 nm/min (n = 12; 4 cells from 4 plants; Fig. 2C). Analysis of variance (ANOVA) identified significant variation among conditions (F3,44 = 38.7, P = 2.133 × 10−12). Post hoc Tukey’s pairwise comparison tests indicated that primary cell wall CSC velocity was significantly different from all stages of secondary cell wall development. Secondary cell wall CSC velocity did not differ significantly between early and mid-development, but both were significantly different from late development (P < 0.01; Fig. 2E). This illustrates that secondary cell wall CSCs move more rapidly in the plane of the membrane during peak secondary cell wall synthesis, then slow as the cell nears maturity, prior to programmed cell death. Thus, both the high density and high velocity of synthesis by individual CSCs contribute to rapid cell wall synthesis during secondary cell wall formation.

Fig. 2 Secondary cell wall CSCs have a higher velocity than primary cell wall CSCs during peak cellulose deposition.

(A and B) CSC tracks in single frames and time projections for primary cell wall (A) and secondary cell wall (B) visualized using GFP::CESA3 and YFP::CESA7, respectively. Kymographs sampled along the yellow lines show CSC trajectories over time, from which CSC velocity was calculated. Scale bars, 10 μm. (C and D) Histograms of GFP::CESA3 (C) and YFP::CESA7 (D) velocities calculated from kymograph analysis. (E) Box plot of CSC velocities across stages of xylem cell development. Means with different letters represent statistically significant differences (Tukey’s pairwise comparison, P < 0.01). For each developmental stage, 480 CSC velocities were measured from 12 cells in four plants. In (D), 1440 velocities were pooled from all developmental stages. In (E), velocities were averaged for each cell before analysis.

Previous studies examining fluorescently tagged secondary cell wall CESAs, visualized through layers of tissue to the center of the root, could not resolve CSCs at the plasma membrane, although these authors made contributions to understanding the trafficking of CSCs in intracellular endomembranes (11, 12). The authors described “CESA7-containing organelles” actively streaming through the cytoplasm and pausing at domains of secondary cell wall deposition (12). However, because of technical limitations, the nature of the organelles was not clear. In the induced protoxylem tracheary element system of the VND7::GR lines, we found that the CESA7-containing organelles were both small CESA-containing compartments (SmaCCs) (10) and Golgi (Fig. 3; red arrows in movie S2). SmaCCs were both Golgi-independent SmaCCs (blue arrows in movie S3) and Golgi-associated SmaCCs (yellow arrows), and their behaviors were most easily identified using fluorescence recovery after photobleaching (FRAP) (Fig. 3, A and B, and movies S3 and S4). After bleaching the bright plasma membrane–localized CSC signal, both SmaCCs and Golgi repopulated the underlying cytoplasm (Fig. 3A). Golgi and SmaCCs often moved in a coordinated fashion (movie S3; 40 ± 15% of SmaCCs were associated with Golgi, n = 3 cells), and frequently Golgi and associated SmaCCs would pause (ranging from 15 s to 3 min) at secondary cell wall domains, consistent with previous observations (12). When the Golgi moved on, a CSC signal would persist and split within 70 ± 20 s (n = 8 events; Fig. 3B and movie S4). This stationary signal followed by slow steady movement (280 ± 30 nm/min, n = 16 particles) fits the criterion for an insertion event of multiple CSCs into the plasma membrane, as defined for primary cell wall CSCs (10). We were unable to quantify the number of insertion events at secondary cell wall domains. However, it was apparent that CSC insertion events were restricted to thickenings (in 12 photobleached regions of 10 μm2 in six cells, no CSC inserted in plasma membrane regions between thickenings), hence the targeting of these events is tightly limited to secondary cell wall domains. TEM of induced developing protoxylem cells revealed several vesicle populations associated with Golgi that may carry CSCs, including trans-Golgi networks, secretory vesicle clusters, and larger vesicles lacking internal content (Fig. 3C). High-resolution live-cell imaging of CESA7-containing organelles demonstrates the dynamic exchange of CSCs among the Golgi, SmaCCs, and the plasma membrane associated with microtubule bands.

Fig. 3 Golgi and SmaCCs densely populate and rapidly deliver CSCs to domains of secondary cell wall formation.

(A) Fluorescence recovery after photobleaching (FRAP) of YFP::CESA7 in the boxed area, overlying a secondary cell wall thickening. Abundant Golgi-independent SmaCCs (red and blue arrows) rapidly repopulate bleached regions. Additionally, Golgi (arrowhead) and closely associated SmaCCs (yellow arrow) can be seen moving from one secondary cell wall band and pausing at another. (B) FRAP and kymograph analyses demonstrating insertion at the plasma membrane of at least two YFP::CESA7-labeled CSCs from a SmaCC (yellow arrow). After the Golgi (arrowhead) moves away, the signal from the SmaCC splits into two distinct punctae with steady velocities (red and orange arrows). (C) TEM micrographs of cytoplasm around secondary cell walls showing a diversity of closely associated vesicles. Trans-Golgi networks (arrows), secretory vesicle clusters (arrowheads), and electron-lucent vesicles (asterisks) are indicated. Scale bars, 5 μm (A), 2.5 μm (B), 500 nm (C).

Previous studies of primary cell wall CSCs showed that disruption of cortical microtubules did not affect the rate of insertion of CSCs into the plasma membrane (10), although CSC distribution over the membrane was transiently disorganized (9, 10). To test the effect of loss of microtubule bundles on secondary cell wall CSC insertion events, we measured the velocities and trajectories of CESA7-containing CSCs in the VND7::GR induction system after oryzalin treatment. In contrast to the dimethyl sulfoxide (DMSO) control (Fig. 4A), where the tracks of CSCs were restricted to the secondary cell wall bands, the plasma membrane­–localized CSCs in oryzalin-treated cells were disorganized (Fig. 4A and movie S5). These CSC clusters were similar to the “swarms” of primary cell wall CSCs described after treatment with an intermediate (8) but not higher (18) concentration of oryzalin. In the absence of microtubule bands, the velocity of secondary cell wall CSCs at the plasma membrane was unaffected (313 ± 41 nm/min for control, 324 ± 33 nm/min for oryzalin-treated; n = 12 cells from 4 plants for each; Fig. 4B) while CSC insertion events continued (Fig. 4C and movie S6). Oryzalin treatment also did not affect the total cellulose content produced by the differentiating cells (fig. S4). In contrast, inhibiting cellulose biosynthesis with 2,6-dichlorobenzonitrile (DCB) treatment influenced only cellulose deposition, while microtubule formation was unaffected (fig. S5). Therefore, the banded microtubule pattern is independent from secondary cell wall formation, and although microtubules are important for the overall secondary cell wall banding pattern, they are not necessary for CSC delivery from endomembranes to the plasma membrane, nor for reaching peak CSC velocity.

Fig. 4 Secondary cell wall CSC distribution at the plasma membrane is disorganized by the loss of microtubules after oryzalin treatment while CSC delivery and motility are unaffected.

(A) VND7::GR-induced cells expressing YFP::CESA7 and RFP::TUB6 after oryzalin treatment. Plasma membrane–localized secondary cell wall CSCs follow aberrant tracks in the absence of microtubules. (B) CSC velocities were not significantly different between oryzalin- and DMSO-treated cells (Student’s t test, P = 0.49). For each condition, 40 CSC velocities were averaged for each of 12 cells in four plants. (C) FRAP of YFP::CESA7 signal revealed insertion of CSCs at the plasma membrane after microtubule loss (arrows). Scale bars, 10 μm [(A) and (B)].

We attribute the rapid formation of the secondary cell wall to both increased concentration and velocity of CSCs at the plasma membrane. Secondary cell wall CSCs are delivered to these domains by populations of Golgi-associated and independent SmaCCs. These insertion events are associated with, but do not require, microtubule bundles underlying secondary cell wall domains. These secondary cell walls provide structural support and the water-conducting functions necessary for plants to inhabit land. These data provide important insights into how land plants produce secondary cell walls with the ultrastructural features required for upright growth.

Supplementary Materials

Materials and Methods

Figs. S1 to S5

Tables S1 and S2

Movies S1 to S6

References (1922)

References and Notes

  1. Acknowledgments: Arabidopsis plants containing RFP::TUB6 were the kind gift of C. Ambrose and G. O. Wasteneys. Supported by CREATE and Discovery grants (L.S. and S.D.M.); Canadian Natural Sciences and Engineering Research Council doctoral postgraduate scholarships (Y.W., M.J.M., and L.M.M.); NSF grant MCB-1158372 (D.W.E. and H.N.C.); and Japan Society for the Promotion of Science KAKENHI grants 24114002 and 25291062 (T.D.). We thank the technical support staff of the UBC Bioimaging Facility and R. White for statistical advice. The pTA7001 plasmid (VP16-GR vector) is available from N.-H. Chua of The Rockefeller University under a material transfer agreement; the VND7-GR line is available fromT.D.; and the GFP::CESA3 and YFP::CESA7 VND7-GR lines are freely available from L.S. or S.D.M. for noncommercial purposes under a materials transfer agreement on the GVG system with N.-H. Chua. Further data are reported in the supplementary materials.
View Abstract

Stay Connected to Science

Navigate This Article