Quantification of microfibril angle in secondary cell walls at subcellular resolution by means of polarized light microscopy



This article is corrected by:

  1. Errata: Corrigendum Volume 207, Issue 1, 248, Article first published online: 22 April 2015

Author for correspondence:

Rivka Elbaum

Tel: +972 8 9489335

Email: elbaum@agri.huji.ac.il


  • The cell walls constitute the mechanical support of plants. Crystalline cellulose building the walls forms rigid microfibrils that set the stiffness of the cell and the direction in which it expands during growth. Therefore, the determination of the directions of the microfibrils is important in both mechanical and developmental assays.
  • We adapted polarized light microscopy to estimate the cellulose microfibril orientations at subcellular resolution. The optical information supplements X-ray scattering data, Raman microspectroscopy, and electron microscopy.
  • We analyzed samples from three plant tissues: cells from an Araucaria excels branch, in which we revealed lower cellulose density in regions where the cell wall curvature becomes bigger, namely, the cell wall corners; a wheat (Triticum turgidum) awn's hygroscopically active region, which revealed a gradient in the cellulose microfibril angles that spans across four cell rows; and a stork's bill's (Erodium gruinum) coiling awn, which revealed that the cellulose in the cell wall is organized in two orientations seamed together, rather than in a continuous helix.
  • The unique spatial information is easily obtained from microscopic specimens and further illuminates new aspects in the mechanical tissues.


Plant cell walls function as a mechanical skeleton for the plant body. The wall is a layered construction, consisting of stiff crystalline cellulose microfibrils embedded in an amorphous matrix of noncrystalline cellulose, hemicelluloses, pectin, various aromatic compounds and proteins. The cellulose microfibrils in the cell walls are arranged in a helical configuration (Peng & Jaffe, 1976). In cells with a mechanical role, after the cell reaches its final size, a secondary layer is deposited which is the main layer of the wall. The mechanical properties of the cell wall depend mainly on the cellulose microfibril angle (MFA) in relation to the cell's long axis (Harris & Meylan, 1965; Cave, 1968; Preston, 1974; Cave & Walker, 1994; Butterfield, 1998).

The cellulose MFA is characterized for either single cells or bulk samples (reviewed by Donaldson, 2008). Polarized light microscopy was one of the earliest methods used to infer the cellulose structure at light microscopic resolution (Schmidt, 1924). This method is based on the ability of birefringent materials, such as crystalline cellulose, to rotate the polarization plane of light. The assessment of a sample's birefringence can be achieved by using a computerized system that calculates the light retardance at optical resolution based on four images taken with different polarizations (Oldenbourg, 1999; Oldenbourg et al., 2004). Such systems have been used extensively in cytoskeleton studies (Katoh et al., 1999; Oldenbourg et al., 2004; Oldenbourg, 2007; Kilani & Chapman, 2011), and, to a lesser extent, in the elucidation of cellulose orientation in plant cell walls (Baskin et al., 2004; Eder et al., 2010; Gu et al., 2010).

Here we address the plant cell wall layers and use a computerized polarized light-based system to assess the orientation of the crystalline cellulose microfibrils. We decipher spatially the cell wall nanostructure in three samples: a young Araucaria branch, a wheat awn, and a stork's bill awn. These three tissues clearly differ in their mechanical roles; the branch is adapted to carry weight, while the awns bend (wheat) or coil (stork's bill) when they dry as part of their function in hygroscopic seed dispersal mechanisms. The variation in performance is reflected in the micro- and nanostructure of the tissues.

Materials and Methods

Wild emmer wheat, Triticum turgidum L., plants were grown in the glasshouse of the Max-Planck-Institute for Molecular Plant Physiology, Golm, Germany, starting on the 21 February 2006 for c. 4 months. Dry seed dispersal units were collected and kept under ambient conditions.

Wild mature Erodium gruinum L. dispersal units were collected in the hills of Nes-Tziona, Israel, on the 1 May 2009. E. gruinum was germinated according to a previously described procedure (Young et al., 1974). Dispersal units were continuously collected from the mature plants and kept under ambient conditions.

A branch of Araucaria excelsa, c. 1 cm in diameter, was collected from a tree grown in The Faculty of Agriculture garden, Rehovot, Israel.

Sample preparation

The samples were dried under ambient conditions and embedded in polyethylene glycol (PEG) 2000 MW as described previously (Rüggeberg et al., 2008). Cross-sections (10 μm thick) were cut on a rotary microtome (Leica RM2255, Leica Biosystems GmbH, Nussloch, Germany), and then placed in water to remove the PEG. The rinsed cross-sections were placed on a glass slide with a drop of water and sealed with a cover-slip to prevent evaporation of water during the measurement.

Scanning electron microscopy (SEM)

Samples from mature awns were broken by hand. The samples were prepared by critical point drying, mounted on aluminum stubs, and sputter-coated with gold-palladium. The samples were examined in the environmental scanning electron microscope, XL 30 ESEM FEG (FEI), at 10–12 kV and 9–9.5 mm working distance, using the high vacuum mode.

Small-angle X-ray scattering (SAXS)

The inner and outer faces of the coiling region of the E. gruinum awn were separated using a razor blade. A narrow long strip of a young branch of A. excelsa was cut using a razor blade. The wet samples were inserted into 1.5 mm quartz capillaries, to which 10 μl distilled water was added to maintain their wet state. The capillaries were flame-sealed and mounted vertically, in a perpendicular orientation to the X-ray beam. Experiments were carried out at room temperature. Each sample was checked before and after the experiment to verify that no fluid was lost during the time of exposure, c. 1 h. Scattering experiments were performed using CuKα radiation (λ = 0.154 nm) from a Rigaku-MicroMax 007 HF X-ray generator operated at a power rating up to 1.2 kW. The beam size at the sample position was 0.7 × 0.7 mm2, as defined by a set of two scatterless slits (Li et al., 2008). The scattered beam went through a flight path filled with He, and reached a Mar345 Image Plate detector (MAR Research, Norderstedt, Hamburg, Germany). The sample distance to the detector was 1841.3 mm, calibrated using silver behenate. Background correction was verified by measuring the scattering of a capillary filled with distilled water and correcting for sample absorption. Integration of the scattering density was performed using FIT2D software (Hammersley et al., 1996). Scattered intensity was plotted as a function of the scattering vector q = (4π/λ) sinθ, where λ is the X-ray wavelength and θ is half the angle between the incident and scattered wave vectors. A more detailed description of the experimental setup has been given elsewhere (Nadler et al., 2011).

Raman microspectroscopy

Raman images were collected using a microspectrometer inVia Reflex (Renishaw, Renishaw Wooten-under-edge, UK) equipped with a green Nd:YAG laser excitation source (532 nm, 45 mW max) under a ×50 objective Leica lens (NA = 0.75). The grating was 1800 lines mm–1. Spectral maps were obtained in the raster mapping mode, using a step size of 1 μm in the x and y directions. The scattered light was detected with a Peltier-cooled charge-coupled device (CCD) detector with spectral resolution c. 2 cm−1. Spectra were recorded with the accumulation of five scans, 1 s exposure time, and 5% laser power in the wavenumber region of 275–2016 cm−1. Baseline adjustment, smoothing, peak picking, and mapping of peak integrated intensities were performed with the instrument control software (Renishaw WiRE 3.3). The mean values of the peak areas were obtained using ImageJ (Rasband, 1997).

Polarized light imaging

Sample birefringence was investigated using a LC-PolScope image processing system (CRi, Inc., Woburn, MA, USA) mounted on a microscope (Nikon Eclipse 80i, Tokyo, Japan) equipped with Plan Fluor ×20/0.5 OFN25 DIC N2, Plan Fluor ×40/0.75 OFN25 DIC M/N2, Fluor ×60/100w DIC H/N2 ∞/0 WD 2.0 objectives. The system includes a computer-controlled universal compensator made of two liquid crystal variable retarders. Retardance images were taken by a cooled CCD camera at high optical resolution. Retardance values were extracted manually using Abrio software tools (CRi, Inc., Woburn, MA, USA) from at least five cells for each sample. Average retardance values were used for the calculation of the MFA.

Calculation of the MFA

Birefringent characteristics of cellulose microfibrils

Crystalline cellulose microfibrils are strongly birefringent owing to their biaxial anisotropic structure. As linearly polarized light passes the microfibrils parallel to their long axis, the light will experience a single refractive index, no. This is the microfibrils' optical axis. On the other hand, when the light propagates perpendicularly to the microfibrils, the polarization component of the light that is vibrating perpendicular to the microfibrils (the ‘ordinary ray’) will experience refractive index no. The polarization component that is vibrating parallel to the microfibrils (the ‘extraordinary ray’) will experience a refractive index ne, which in cellulose is larger than no. This direction is the slow optical axis, also named ‘the retardation azimuth’. At any other incidence angle, the refractive index experienced by the extraordinary ray is math formula. The change in math formula is described by an ellipsoid, whose major and minor axes are ne and no, respectively (Fig. 1), (Iyer et al., 1968; Lipson et al., 2010).

Figure 1.

Cellulose birefringent properties. The optical axis of cellulose lies parallel to the long axis of the crystalline cellulose microfibrils (brown rod), while the retardation azimuth (the slow axis) lies perpendicular to it. The variation in the refractive index math formula as a function of the angle of incidence, θ, is given by an ellipse (red), with ne as the major axis and no as the minor axis. The value of no is independent of θ and is given by a circle (blue).

The birefringence (B) of a cellulose microfibril oriented at an angle θ to the incident light is defined as the difference between the refractive indices math formula and no, (Eqn 1). The variation in the refractive indices indicates on the different light propagation speeds of the ordinary and extraordinary rays inside the crystalline microfibril as a result of its anisotropy, which results in the retardation (R) of the rays in relation to each other. The retardance increases with the length of the light pathway through the birefringent material. We refer to this length as the effective thickness, t′ (Eqn 2). As the math formula changes with θ, the resultant dependence of R on the microfibril orientation allows us to estimate the MFA in a microscopic section:

display math(Eqn 1)
display math(Eqn 2)

Calculation of the MFA

In order to calculate the MFA, we obtain a LC-PolScope image of the samples' cross-section (Fig. 2a, left) under ‘Retardance’ mode. The birefringence (B) is then extracted for a known t′ value and subsequently math formula, using published values of no and ne (in this work we used the reported values for ramie fibers, no = 1.599, and ne = 1.529 (Iyer et al., 1968)). As the value of math formula is dependent on the angle of light incidence, θ, we are able to calculate the angle of incidence by the application of the ellipse formula (Preston, 1933). The MFA is simply 90° − θ (Fig. 2a, right).

Figure 2.

Calculation of the microfibril angle (MFA) from the retardance values. (a) A simplified scheme illustrating the angular relation between the incident polarized light and the cellulose microfibrils within the cell wall (right), as the light is perpendicular to the cross-section of the cells (left); (b) a theoretical graph showing MFAs calculated for rising retardance values at three different effective thicknesses (t′). The smaller the t′ of the sample, the steeper the resulting sigmoidal curve. Thus, the ability to resolve high values of MFA is increased for thicker samples.

The retardance value obtained is limited by a maximal value, Rmax, corresponding to the maximal MFA (90°). The sigmoidal plots in Fig. 2(b) illustrate the effect of the effective thickness (t′) on the calculated MFAs as a function of the retardance value. It shows that, for the same retardance value, the calculated MFA will be higher for a low t′ than that for a higher t′. It also shows that the higher the retardance, the more significant is the effect of t′ on the calculated MFA. Because the effective thickness (t′) is not necessarily the actual thickness of the sample, but rather the total thickness of the crystalline cellulose microfibrils, to calculate the actual MFA we need to know the fraction of crystalline cellulose. However, the methods to measure cellulose crystallinity are not accurate (Park et al., 2010). We thus used SAXS to estimate the MFA range, and then back-calculated t′.


The orientation of cellulose microfibrils in an A. excelsa branch

In order to determine the spatial variance in the cell wall of cells from a thin A. excelsa branch, the SAXS pattern was first obtained (Fig. 3a). The symmetrical pattern around the meridian and equatorial indicates that the cellulose microfibrils form a regular helix along the long axis of the cell (Fig. 3b; Heyn, 1948). The MFA ranges between 10° and 40°, with a mean angle of 24°, in agreement with the values obtained for a young spruce branch (Färber et al., 2001). The retardance image of a 10-μm-thick cross-section of the branch, showing retardance values per pixel, reveals spatial variation in the cellulose arrangement (Fig. 3c right). Low retardance values in the intercellular space indicate on a low cellulose concentration typical of the thin primary cell wall and the middle lamella which contains no cellulose (Esau, 1965). At the cell wall periphery, we detected a layer comprising the first secondary layer (I). This layer, as well as the inner layer (III), shows relatively high retardance values. On the other hand, the median part of the wall (II) displays lower retardance, corresponding to lower MFA in the middle of the cell wall (Fig. 3d). The structure of layer III also appears to be further laminated, showing lower retardance values towards the cell center, at the youngest layers (Fig. 3c inset, e).

Figure 3.

Comparison between the traditional small-angle X-ray scattering (SAXS) data and the retardance imaging. (a) SAXS pattern obtained for a branch of Araucaria excelsa. (b) Scheme showing the microfibril angle (MFA) of a single cell. In such an arrangement, the MFA remains constant at all azimuthal directions, χ. (c) Brightfield (left) and retardance (right) images of a cross-section of the branch (bar, 10 μm). Inset: a close-up of a cell wall showing microlayers in layer III (bar, 5 μm). Relatively constant retardance is measured for any azimuthal angle χ (retardance scale, 0–273 nm). Layer I is visible as bright cyan layer at the outer border of the cell. Layer III appears as a cyan layer at the inner border of the cell wall. The middle lamella (ML) shows very low retardance in agreement with the absence of cellulose there. At the cell wall corners (wc) Layer II shows lower retardance than in the cell wall body (WB). (d) A plot showing the average retardance values in the cell wall body (blue) and at the corners (red) at the three cell wall layers (values extracted from five cells; four points were sampled from each cell for every cell wall region). Error bars represent a standard deviation interval (*, < 0.05). (e) A representative plot of the retardance values (blue line) measured along a radial line from the outer layer of the cell inwards, illustrating the microlayers in layer III. Calculated MFA values (t′ = 3.7 μm) for the different layers are also shown (red diamonds).

Another phenomenon we observed is that the cell wall corners, that is, the curved cell wall part facing the meeting points between three neighboring cells, display lower retardance than the rest of the cell wall (the cell wall body; Fig. 3d).

We calculated the actual MFA at the cell wall body by setting the effective thickness (t′) of the sample so that the MFA range calculated would match the MFA range measured by SAXS (Fig. 3e). For the Aexcelsa branch, the value of t′ was set to 3.7 μm, equivalent to 37% by volume crystalline cellulose within the cell wall. This value is within the previously estimated range of the cellulose fraction in secondary cell walls (30–60%), depending on the methodology used (Thygesen et al., 2005).

Spatial microfibril angle variation in a wheat awn cross-section

The structure of the hygroscopically active tissue of a wheat awn was analyzed previously (Elbaum et al., 2008). This region, which is the narrow part of the cross-section (Fig. 4a), contracts when the awn dries, pulling it into a bent position. The retardance image of the active region shows a gradual reduction in the retardance from its center to its edge (Fig. 4c, lower panel). This gradient may result from a reduction in the MFA or in the cellulose content. We examined the latter option using Raman microscopy, as follows (Fig. 5). The cellulose content was estimated by comparing the 370 cm−1 peak area in the edge cells (Fig. 5a) and the center cells (Fig. 5b), as was done previously (Konnerth et al., 2009), assuming that the volume sampled by the laser is similar throughout the mapping. While other bands attributed to cellulose, such as 1091 and 1117 cm−1, also stem from lignin and/or hemicellulose, the 370 cm−1 band stems from cellulose only (Agarwal & Ralph, 1997; Agarwal, 2006). Also, this band is insensitive to the direction of the polarization of the laser (Wiley & Atalla, 1987). The mean values obtained are 170 ± 24 au for the edge cells and 180 ± 26 au for the center cells, showing no significant difference. This supports the possibility that the cellulose content in these two regions is similar. Therefore we conclude that the retardance picture stems from a reduction in the MFA from the awn center to its periphery.

Figure 4.

The structure of the wheat awn (Triticum turgidum) contracting cells. (a) Scanning electron micrograph of a cross-section at the bending part of the awn. The hygroscopically active (contracting) region is the narrow upper part of the cross-section (bar, 200 μm). (b) A close-up of a cell from the awn's active region, illustrating the cell wall plywood structure (bar, 10 μm). (c) Brightfield (top) and retardance (bottom) images of a cross-section of the active region (scale, 100 μm; retardance scale, 0–273 nm). Close-up of cells from the edge (d) and the center (e) of the active region (bar, 10 μm; retardance scale, 0–273 nm). (f) Plots of the retardance values (solid lines) and calculated microfibril angle (MFA) (symbols) (t′ = 3.9 μm) of cell walls from the center (maroon) and edge (blue) of the active region, measured along radial lines (marked on the images d and e), starting from the outer layer of the cell inwards.

Figure 5.

Raman images (88 μm × 70 μm) of the active region in the wheat awn (Triticum turgidum) cross-section showing the peak area of the c. 370 cm−1 band (cellulose, orientation-insensitive) taken from the edge of the region (a) and its center (b).

Previously, cell walls in the active region were claimed to display a plywood-like organization (Elbaum et al., 2008). These conclusions were based on X-ray scattering patterns showing preferred MFA of 90° and 0°, and scanning electron micrographs showing the submicron layered cell wall structure (Fig. 4b). Retardance imaging of the cells at the edge and center of the awn (Fig. 4d,e, respectively) showed two layers of relatively high retardance delimiting a middle layer with lower retardance. This is reminiscent of the cell wall arrangement we detected in the A. excelsa branch (Fig. 3e). A closer examination revealed greater cell wall width and lower MFAs in the cells at the active region's edge, as compared with those in the center (Fig. 4f). More subtle undulations in the retardance values, which correlate to micron-thick sublayers, were also visible. The MFA calculated for t′ = 3.9 μm shows a variation of c. 10° between the thin sublayers. Based on high-resolution SEM images (Elbaum et al., 2008) we suggest that these micron layers are built of yet thinner layers of alternating MFA.

Detection of tilted cellulose helix in the stork's bill's awn

The awns of stork's bill (E. gruinum) coil as they dry. This hygroscopic movement is based on coiling cells in which the cellulose microfibrils are arranged in a tilted helix with a small pitch (illustrated in Fig. 6a; Abraham et al., 2012; Aharoni et al., 2012). In such an arrangement, the MFA on opposing sides of the cell wall differ by their sign. This is in contrast to a normal helix, where the MFA remains constant for any azimuthal direction, χ (Fig. 3b). Indeed, the retardance pattern of the tilted helix is not symmetric around the cell (Fig. 6b, lower panel), reflecting changes in the absolute values of the MFA. In addition, we observed a continuous reduction of the MFA from the cell wall's periphery to its center, and low retardance points between cells, marking the middle lamella.

Figure 6.

The cellulose arrangement in the active coiling cells of the stork's bill (Erodium gruinum) awn. (a) Scheme showing the tilted cellulose helix configuration. The microfibril angle (MFA) in the tilted helix arrangement changes with the azimuthal direction, χ, having different signs at the opposite sides of the cell wall. (b) Brightfield (top) and retardance image (bottom) of the coiling cells cross-section (bar, 10 μm; retardance scale, 0–273 nm). The white arrow shows the direction of the short axis of the awn's cross-section. The black arrows show the points of abrupt change in cellulose direction (see panels d, e and f). Encircled are the middle lamella regions showing very low retardance. (c) Plots of the retardance values for the outer (blue) and inner (maroon) cell wall rings as a function of χ. (d) Plots of the calculated MFA values for the outer (blue) and inner (maroon) cell wall rings as a function of χ (t′ = 3.5 μm). (e) Schematic illustration of the cellulose orientations in the cell wall periphery in a longitudinally cut and flattened cell. (f) Scanning electron micrograph of the cross-section of the active coiling cells (bar, 10 μm). The arrows show cell wall fracture lines that may correlate with the points of microfibrilar angle change.

We expected the MFA to reach 90° somewhere around the cell wall, to allow for a gradual cellulose microfibril direction change. To our surprise, the MFAs calculated according to the retardance values do not reach 90°. This means that the microfibril direction change is abrupt. We chose the direction change at the minimal retardance values (Fig. 6c–e), in correlation to cracks in the cell walls observed in SEM images (Fig. 6f, orange arrows). We explain the appearance of these cracks as weak points in the cell wall, marking the sharp microfibrilar direction changes.

The retardance maxima appear at about χ = 20 and 190°, corresponding to maximal MFA of 53° and −51° (at the cell wall periphery). This information reveals that the tilting of the cellulose helix is not around the short axis of the awn cross-section (Fig 6b, lower panel), but at around χ = 20°.


In this work we combine data from polarized light microscopy with other methods, to illuminate new aspects of mechanical plant tissues from an A. excelsa branch and the awns of wheat and stork's bill.

Reduced retardance at cell wall corners

The cross-section of the A. excelsa branch reveals polygonal cells showing lower retardance values at the corners of the cell wall than in the cell wall body. We propose that the low retardance regions reflect changes in the cellulose organization: the geometrical constraint means that fewer cellulose fibrils accumulate in the corners. This may be a result of stretching of the deposited wall, or of a simple reduction of the cellulose concentration during its deposition, without a mechanical stretch. We also detected this effect in the wheat awn cells (Fig. 4d,e) but not in the stork's bill cells, probably because their shape is nearly oval (Fig. 6b, lower panel). We therefore conclude that the polygonal cells formed by pressing against each other, and note that the retardance in cell wall corners cannot be used for estimating the MFA.

Hygroscopic bending and coiling

The highly specialized hygroscopic tissue of the wheat awn is adapted to serve a dual function, as it responds to changing water content by creating a movement while its high stiffness enables the awn to push the seeds along the ground (Elbaum et al., 2007). Both roles are reflected in the structural aspects of the hygroscopic tissue (Elbaum et al., 2008). The contracting cells in the active region display variation in retardance, which corresponds to an MFA range of 20–80°. The lowest MFA is found in the cells at the awn edge, maximizing the second moment of inertia and the stiffness of the structure (Fig. 4d). Undulations in the MFA may correspond to the micron-level layering of the cell wall's plywood structure previously described (Elbaum et al., 2008).

While bilayer hygroscopic structures are very common and well studied (Fahn & Zohary, 1955; Fahn & Werker, 1972), the hygroscopic coiling cells are known to us only in the Geraniaceae family. We have shown that in order for a cell to coil as it dries, it is enough to restrict its contraction by a tilted nonextendable microfibril helix (Abraham et al., 2012). However, the effects of structural variations on the coiling are not characterized. The retardance images of the coiling cells from the stork's bill awn further reveal their structural complexity. The retardance values of the outer (older) cell wall are higher than those of the inner (younger) cell wall (Fig. 6b,c), indicating on a corresponding difference in the MFAs (Fig. 6d). The homogeneity of cells across the tissue resulting from precise cellulose deposition means there is a strict cellular control over the microfibril tilt direction and angle. Mathematical and material models are needed to understand the design details and to apply them in biomimetic devices.


We have shown that polarized light microscopy using the computerized LC-PolScope imaging system is a quick and relatively easy method for determining the cellulose organization in the plant cell wall. It gave us the ability to clearly and easily reveal the spatial complexity of the microfibril organization in three tissues at optical resolution. Previously, Raman microscopy has been used to this end, taking advantage of the orientation-sensitive cellulose bands to determine the cellulose MFA (Gierlinger et al., 2010). However, this method has lower spatial resolution, as it is based on nonlinear optics, with relatively low efficiency. Thus the laser spot is typically larger than 1 μm. In addition, the method is calibrated to cells with normal cellulose helix, assuming the MFA does not change as a function of the azimuthal directions (χ). The retardance high sensitivity of the LC-PolScope system allowed us to determine MFA to an accuracy of 1°. The data collected include relative MFA values as a function of the cell radius and χ. The main disadvantage of this method is that the calculated MFA depends on the thickness of the crystalline cellulose, which is difficult to measure directly. To circumvent this requirement, we measured the MFA range by X-ray diffraction, and used the extreme values as the boundaries for the calculated MFA. In addition, the cross-section thickness must be uniform, and the relation between the cells' long axis and the optical path must be known. The elucidation of local variations in the microfibril organization provides new insights into the mechanical roles of the cell wall.


We thank Carmen Tamburu and Uri Raviv (The Institute of Chemistry, The Hebrew University of Jerusalem, Israel) for the X-ray scattering measurements and their valuable insights. The electron microscopy studies were conducted at the Irving and Cherna Moskowitz Centre for Nano and Bio-Nano Imaging at the Weizmann Institute of Science. This Research was supported by the Israel Science Foundation grant 598/10.