Resonant soft X-ray scattering reveals cellulose microfibril spacing in plant primary cell walls

Cellulose microfibrils are crucial for many of the remarkable mechanical properties of primary cell walls. Nevertheless, many structural features of cellulose microfibril organization in cell walls are not yet fully described. Microscopy techniques provide direct visualization of cell wall organization, and quantification of some aspects of wall microstructure is possible through image processing. Complementary to microscopy techniques, scattering yields structural information in reciprocal space over large sample areas. Using the onion epidermal wall as a model system, we introduce resonant soft X-ray scattering (RSoXS) to directly quantify the average interfibril spacing. Tuning the X-ray energy to the calcium L-edge enhances the contrast between cellulose and pectin due to the localization of calcium ions to homogalacturonan in the pectin matrix. As a consequence, RSoXS profiles reveal an average center-to-center distance between cellulose microfibrils or microfibril bundles of about 20 nm.


Onion epidermis structure and scattering contrast of each layer
As shown in Figure S1a, the onion epidermis is composed of several layers. Based on profilometry measurements (Table S1), the unextracted epidermis is 1145 ± 131 nm thick and the Drislease-treated epidermis is 152 ± 40 nm thick. Most of the cell wall materials are removed with Drislease digestion. The top 52 nm could be a mixture of pectin and cuticle ( Figure S1b), if some pectin is intertwined with cuticle such that some pectin cannot be fully removed by Driselase. Alternatively, all pectin is digested completely by drislease, and some of the cuticle layer is left with holes ( Figure S1c).
For epidermal peels, AFM studies indicate that there is a thin layer of pectin sitting on top of cell walls 1 . Based on the AFM height profile ( Figure S2b) of the superficial pectin layer, the height distribution of pectin can be fitted to a Gaussian distribution with 43 nm as the mean and 14 nm as the standard deviation. The thicknesses of the cuticle layer and cuticle/pectin layer remain the same as the Driselase-treated sample, which leads to a 950 nm thick cell wall layer for untreated epidermal peels.
Based on the layered structure of untreated epidermis, the scattering contrast near the calcium L-edge was calculated for each layer of the epidermis (shown in Figure S1a). In the top 43 nm pectin layer, the contrast comes from pectin and vacuum. For the 950 nm thick cell wall layer, the scattering contrast is dominated by pectin and cellulose. Scattering contrast arises from pectin and cuticle in the mixed 52 nm pectin-cuticle layer. For our two models of the Driselase-treated sample shown in Figure S1b and S1c, the scattering contrast from the top 52 nm layer could either arise from cuticle and pectin ( Figure S1b) or cuticle and vacuum ( Figure S1c).

Comparing FFT of AFM image to cylindrical form factors
The Fourier transformed frequency spectra of the AFM image shown in Figure 1 of the main text has two features. A shoulder or broad peak is apparent near q = 0.04 Å -1 , and another peak is visible near q = 0.15 Å -1 . The feature near q = 0.15 Å -1 can be described with a cylindrical form factor with a diameter of 6.6 nm and length of 0.97 μm as shown in Figure S3.  Figure 1 to cylinder form factor. The fitted form factor has a diameter of 6.6 nm and a length of 0.97 μm.

Small angle X-ray scattering
SAXS data from unextracted epidermal peel and from the direct beam (scattering from air) are shown in Figure S4a. Scattering from the unextracted epidermal peel looks very similar to scattering from air, except for some differences in the low q region. Background correction of scattering data from the unextracted epidermal peel ( Figure S4b) shows a q -4 dependence up to q = 0.02 Å -1 but no other features. This indicates that scattering for the unextracted epidermal peel is mostly from large-scale structures, such as the sample roughness.

Effect of sample thickness on scattering intensity
Scattering intensity (I) is a function of sample thickness ( ) and sample transmittance ( ): Sample transmittance is related to attenuation length by where is the attenuation length that depends on the wavelength λ. Attenuation is a function of energy and it increases as energy increases. Assuming the cell wall has a density of 1.5 g/cm 3 , the attenuation length at 280 eV (near the carbon Kedge) is 1.75 μm ( Figure S5a). The scattering intensity is maximized when the sample thickness equals the attenuation length, which can be seen from taking the derivative of Equation 3 giving ′ ∝ 1 . Figure S5b shows the scattering intensity as a function of sample thickness for 10 keV X-rays. Figure S5b suggests the scattering from an epidermal peel with a thickness of 1.1 μm is much lower than the theoretical maximum scattering intensity that can be obtained for samples with larger thicknesses. Thus, we predict weak scattering from the onion epidermal cell wall at 10 keV that is potentially similar to background scattering, making it difficult to study these samples using SAXS.

NEXAFS and scattering contrast
We find that the calcium signal from commercially available pectin obtained from citrus (Sigma-P9436 with a 60% esterification level) is suppressed. Thus, to obtain the NEXAFS spectra for calcium-treated pectin, we used the carbon edge NEXAFS of pectin from citrus and the calcium edge NEXAFS of calciumtreated epidermal cell walls. The two datasets were merged at 320 eV ( Figure S6a). Because the cell wall is only about 40% pectin (by mass), 2 we remove the expected contributions of the other 60% of polysaccharides (e.g., cellulose and hemicellulose) to obtain a spectra representative of Ca-infused pectin. First, we assume that cellulose and hemicellulose do not complex with Ca ions, and therefore represent the NEXAFS spectra near the calcium L-edge (340-355 eV) as a line that connects the pre-edge (340 eV) and post-edge (355 eV), as shown in Figure S6b as a dashed black line. After subtracting the non-pectin NEXAFS ( Figure S6b, dashed black line) from the calcium-treated cell wall ( Figure S6b, solid blue line), the subtracted spectra were further divided by the mass fraction of pectin (0.4). The final spectra that we take as representative of calcium-treated pectin is shown in Figure S6b as a dashed red line. The scattering contrast can be related to the absorptive and dispersive components of the refractive index. The absorptive component β was determined from the NEXAFS spectra shown in Figure 4a of the main text using KKcalc. For these calculations, a density of 1.599 g/cm 3 was used for cellulose 3 and a density of 1.543 g/cm 3 was used for calcium-treated pectin 4 . The β spectra is merged and extend to the entire energy range from 10 eV to 10,000 eV. The dispersive component δ is calculated using the Kramer-Kronig theorem based on β. The calculated scattering contrast between cellulose and calcium-containing pectin is shown in Figure 4b of the main text and indicates that the scattering contrast near the calcium L-edge is two orders of magnitude higher than in the hard X-ray regime.

2D RSoXS Images
RSoXS 2D images for unextracted, calcium-treated, and pectate lyase-treated onion 11 th scale epidermis are shown in Figure S7. The scattering anisotropy (ellipsoidal shape) of the 2D RSoXS data is due to fiber alignment of cellulose microfibrils within the epidermal cell wall and has been previously observed in SAXS studies of hypocotyl of Arabidopsis thaliana 5 . Anisotropy has previously been observed for other semi-crystalline filament materials as well 6 .

Driselase digested calcium-treated epidermis
We performed RSoXS measurements near the carbon K-edge and the calcium L-edge on calcium-treated epidermis that had been digested with Driselase ( Figure S8). Near the carbon K-edge, a feature is observed around q = 0.03 Å -1 . Near the calcium L-edge, the on-resonance scattering profiles (349.3 eV and 352.6 eV) are the same as the scattering profiles collected at energies which are off-resonance (345 eV and 355 eV). Because Driselase digests cell wall components and leaves the cuticle intact, this data suggests that the scattering feature near the calcium L-edge around q = 0.03 Å -1 observed in the calcium-treated epidermis ( Figure 5b of the main text) is likely not from the cuticle.

Total scattering intensity and scattering contrast
Total scattering intensity from RSoXS data was calculated from scattering profiles using . For each sample, TSI at different energies was normalized to TSI at 345 eV near the calcium L-edge and 280 eV near the carbon K-edge ( Figure S9).

Comparing RSoXS data to cylindrical form factors
To explore whether the scattering feature near q = 0.03 Å -1 in RSoXS scattering profiles ( Figure 5b of the main text) is from the form factor of cellulose microfibrils, we calculated form factors for different cylindrical shapes. Figure S10 shows that a monodisperse 36 nm diameter cylinder has a peak around q = 0.03 Å -1 . If that is the case, higher order reflections should appear above q = 0.03 Å -1 . We also examined whether the addition of polydispersity (PD) in the diameter of the cylinder can help match experimental scattering profiles. Polydispersity is defined based on the Shultz distribution , where is the averaged diameter and is the variance of the distribution. By setting the polydispersity to 0.2, the feature around q = 0.03 Å -1 persists but there is a shoulder near q = 0.01 Å -1 that is not observed in RSoXS scattering profiles. When the polydispersity is increased to 0.3, no feature is apparent near q = 0.03 Å -1 . These data show that different cylindrical form factors cannot produce scattering profiles similar to the observed RSoXS scattering profile. Previous work has also shown that although interference between the structure factor and form factor can dampen higher order oscillations in scattering profiles, the peak width is nevertheless relatively narrow. 7,8 Thus, we conclude that the feature around q = 0.03 Å -1 in RSoXS data represents the averaged interfibril spacing and not the form factor of cellulose microfibrils.

NEXAFS spectra of driselase-treated onion epidermis
To obtain the refractive index of cuticle near the C K-edge, we acquire NEXAFS data of Driselase-treated onion epidermis as shown in Figure S11.  Figure S11. NEXAFS spectra of Driselase-treated epidermis. We use this spectra to calculate the optical properties of cuticle.