The spatial orientation and interaction of cell wall polymers in bamboo revealed with a combination of imaging polarized FTIR and directional chemical removal

The mechanical and physical properties of lignocellulosic materials are closely related to the orientation and interaction of the polymers within cell walls. In this work, Imaging Polarized FTIR, combined with directional chemical removal, was applied to characterize the spatial orientation and interaction of cell wall polymers in bamboo fibers and parenchyma cells from two bamboo species. The results demonstrate the cellulose in bamboo fibers is nearly axially oriented whereas it is almost transversely arranged in parenchyma cells. Xylan and lignin are both preferentially oriented alongside cellulose, but with less orientation degree in the parenchyma cells. After lignin removal, the average orientation of xylan and cellulose is little affected, suggesting a strong interaction between cellulose and xylan. However, the alkaline treatment significantly weakens the orientation of lignin in both fibers and parenchyma cells, and more significant for the latter, indicating the easy-degradable nature of lignin in parenchyma cells. Additionally, it seemed the lignin and xylan in fibers were more difficult to remove as compared to parenchyma cells, supporting the assumption that stronger interaction exists between lignin and xylan in the fibers. In a word, it is believed parenchyma cells are more suitable for biorefinery owing to their less ordered and relatively loose molecular assembly, as compared to fibers.


Introduction
Higher plants feature an extremely complex supramolecular cell wall assembled with cellulose, hemicelluloses and aromatic polymers like lignin (Zhao et al. 2019;Tatjana et al. 2017). This structure provides the cell wall with mechanical strength (Dick et al. 2011), but also makes it inherently recalcitrant to enzymatic and chemical treatments in biorefinery scenarios (Ji et al. 2017). Decades of effort have been devoted to genetic engineering and chemical pretreatments, aiming at improving the accessibility and digestibility of lignocellulosic materials (Poovaiah et al. 2014; Vanholme et al. 2008;Loque et al. 2015). Those studies indeed improved the utilization efficiency of biomass. Nevertheless, the inadequate and incomplete understanding of the spatial arrangement and interaction of polymers, which significantly contribute to the properties like biomass recalcitrance and timber strength (Terrett and Dupree 2019), still delays the development of the biorefinery industry.
Furthermore, the anisotropy of wood cell walls arises from the spatial orientation of cell wall polymers (Hinterstoisser et al. 2001). An in-depth understanding of wood polymer spatial orientation has been established for wood fibers (Chang et al. 2014;Stevanic et al. 2011) using polarized FTIR microscopy. As expected, the cellulose, in the form of microfibrils, was highly oriented along its cell axis, providing strong longitudinal mechanical strength for the fibers (Stevanic et al. 2011). The hemicelluloses (glucomannan mainly in softwood and xylan mainly in wood) have been shown to be highly associated with cellulose and therefore preferentially oriented along the direction of microfibrils (Stevanic et al. 2011), whereas lignin was found much less oriented than glucomannan and xylan (Akerholm and Salmén 2003).
In addition, the spatial orientation of cell wall polymers is closely influenced by their interactions. Molecular dynamics simulation indicated abundant hydrogen bonds between cellulose microfibrils and xylan (Zhang et al. 2015). Solid-state nuclear magnetic resonance (NMR) revealed a two-fold helical screw symmetry with a regular pattern of acetate or glucuronate substitutions in cellulose-bound xylan (Simmons et al. 2016;Kang et al. 2019). This evidence indicated that part of xylan is attached to the surface of cellulose, which is responsible for the preferred spatial orientation of xylan. The interaction between hemicellulose and lignin occurs via two different mechanisms (Terrett and Dupree 2019;Kang et al. 2019). Hemicellulose may incorporate into lignin by covalent bonding, usually with ferulic acid as the linkage, which is also called lignin carbohydrate complex (LCC) (Grabber et al. 1995;Jung et al. 2011;Rennie and Scheller 2014). As a supplement, there are electrostatic interactions between lignin and xylan (Kang et al. 2019). Hemicellulose and lignin played an indispensable role in connecting cellulose microfibrils into aggregates, and had a profound effect on cell wall properties and material utilization, such as biomass recalcitrance and timber strength (Slabaugh 2014;Knox 2008).
Bamboo, consisting of stiff sclerenchyma fibers (account for 40% by volume) and soft parenchyma cells (account for 52% by volume) (Liese 1980), is an ideal renewable resource to fill the gap between wood consumption and supply. Bamboo fibers and parenchyma cells differ substantially in chemical composition and cell wall structure (Abe and Yano 2010;Jin et al. 2019), optimized for their mechanical and physiological functions, respectively. Specifically, fibers possess excellent mechanical properties, while parenchyma cells are more suitable for biomass conversion (Jin et al. 2019). It was therefore speculated that fibers should be more orderly assembled than parenchyma cells, which serve more for mechanical buffer and nutrition storage.
Recent research (Tsuboi et al. 2014;Tsuji et al. 2021) shows that bamboo-derived cellulose nanofibers prepared was likely to exhibit more hydrophobic properties than that of wood cellulose nanofibers without any chemical modification, further suggesting the great potential of bamboo in biorefinery industries. In this work, the spatial orientation and interaction of cell wall polymers from two bamboo species were revealed with Imaging Polarized FTIR combined with directional chemical removal. The IR spectra of fibers and parenchyma cells at different polarization angles were respectively collected for polymer orientation assessment. Directional chemical component removal was applied to better explore their interactions. The aim of this work is to reveal the structural discrepancy of bamboo fibers and parenchyma cells at the molecular scale, which will give new insights into their relationship of structure-property-function, as well as precise utilization.

Sample preparation
The 3-years-old Bambusa longispiculata (HM) and Bambusa pervariabilis (CG) used in this work were collected from the Hua'an Bamboo Garden in Fujian province, China. Bamboo blocks with sizes of approximately 10 (longitudinal) × 5 (tangential) × 5 (radial) mm 3 were collected from the 1.5 m height location of a bamboo culm. After softening in water at 80 °C for 8 h, longitudinal sections with thicknesses of 20 μm for bamboo fiber tissue and 35 μm for parenchyma cell tissue were prepared with a sliding microtome for Polarized FTIR microspectroscopy under transmission mode. Six sections of each thickness from each bamboo species were made, with a total of 24 sections prepared.

Delignification and alkaline treatment
The prepared sections were divided into three groups and soaked in deionized water, 1% (w/w) acidified sodium chlorite (acidified to pH = 4.5 with HAc) and 8% aqueous NaOH, respectively. They were kept in an incubator at 30 °C for 8 h. After that, the sections were washed to pH = 7.0 with deionized water. They were then sandwiched between two glass slides before being oven-dried at 60 °C for 16 h. The glass was removed after drying.

Polarized FTIR measurement
Polarized FTIR measurements were carried out using an imaging FT-IR system (Spotlight 400, Perkin-Elmer Inc., Shelton, CT, U.S.A.) equipped with a gold wire grid polarizer under transmission mode. Before scanning, the mercury cadmium telluride (MCT) detector was cooled with liquid nitrogen and the instrument was stabilized for over 30 min (Stevanic et al. 2011). The longitudinal sections were mounted on a sample stage, as parallel as possible to the orientation of the 0° polarization (Fig. 1). A CCD camera was used to locate the area of interest. And then, a 50 × 50 μm scanning area was selected with a spatial resolution of 6.25 × 6.25 μm. The IR spectra were recorded with a resolution of 4 cm −1 in the range of 720-1900 cm −1 . To improve the signalto-noise (S/N) ratio, the number of scans was set at 64 and a background signal was acquired for every sample. The IR radiation was polarized from 0° to 90° in relation to the fiber direction of samples with step intervals of 5°, therefore, 19 IR spectra for each sample were collected.

Data processing
Data processing was performed using the software MAGE-Spotlight developed by Perkin-Elmer, Inc and OMNIC 8.2.0.397 developed by Thermo Fisher Scientific, Inc. The corrections, including atmospheric correction, automatic basline correction and baseline offset correction, were carried out to standardize the IR spectra.
The IR spectra recorded at 0° and 90° polarization angles were processed using the following Eq. 1 to produce an average orientation spectrum.
where S o is the orientation spectrum, implying the orientation of polymers. S 0° and S 90° are the absorption spectra recorded at 0° and 90° polarization angle, respectively.
The relative absorption of a specific IR band (that represented a specific polymer was calculated using the following Eq. 2 (Stevanic and Salmén 2009).
where R A is the relative absorption of a specific IR band at a given polarization angle. I P is the absorption intensity of a specific IR band at a given polarization angle. I max and I min are the maximum and minimum absorption intensity of a specific IR band between 0° and 90° polarization angles, respectively. (1)

Results and discussion
Polarized IR spectra of bamboo fibers and parenchyma cells Polarized FTIR spectra were recorded at 0° and 90° with respect to the longitudinal cell axis of bamboo fibers and parenchyma cells, and the difference spectra were collected from the two directions as shown in Fig. 1. The dominant polymers in bamboo cell walls are cellulose, xylan and lignin. The associated FTIR characteristic bands of these polymers are displayed in Fig. 2. The band at 1161 cm −1 is associated with the C-O-C glycosidic bond vibration corresponding to the backbone of cellulose and xylan, while the bands at 1426 cm −1 and 1373 cm −1 are associated with the C-OH and C-H wagging of the cellulose side group (Marchessault 1962). Previous studies demonstrated that the main hemicellulose in bamboo was xylan, which is substituted by arabinose and glucuronic acid as the side chains (Peng et al. 2011). Furthermore, the bamboo xylan is highly acetylated (Terrett and Dupree 2019). The band at 1738 cm −1 is attributed to the C=O vibration of acetyl groups and carboxyl groups in xylan (Akerholm and Salmén 2001). Moreover, the band at 1461 cm −1 (C-H wagging) is also related to xylan (Marchessault 1962;Liang et al. 1960). For lignin, the bands near 1506 cm −1 and 1601 cm −1 are both assigned to the C=C stretching vibration of aromatic rings. In addition, 1601 cm −1 is also associated with the C=O asymmetric stretching vibration (Faix 1991;Atalla and Agarwal 1985). Furthermore, IR intensity changes of cellulose, xylan and lignin from 0 to 90° can be observed in Fig. 2, implying the distinct spatial orientation of these cell wall polymers.
Polymer orientation in the cell walls of bamboo fibers and parenchyma cells Polarized IR produces spectra at one given angle (Fackler and Thygesen 2013), resulting in a stronger absorption for the functional group that is more oriented with respect to the polarized IR light (Stevanic and Salmén 2009). Figure 3 exhibited the average orientation spectra of all the three cell wall polymers for fibers and parenchyma cells from both HM and CG. In the average orientation spectra, the positive Note Due to the IR absorption in the wavenumber interval between 1000 cm −1 and 1120 cm −1 being too high, this area has been masked off in the figure signals indicate the associated functional groups are arranged in a preferred orientation to the longitudinal cell axis, while the negative signals indicate a more perpendicular orientation (Stevanic et al. 2011;Peng et al. 2019). Furthermore, the intensity of average orientation spectra is positively correlated with the absolute value of the angle between a specific functional group and the incident IR light. It is interesting to note that polymers of fibers and parenchyma cells possessed completely opposite orientation spectra, indicating that the polymers these two types of cells were arranged in different orientations, with the polymers in the former oriented more parallel to the longitudinal direction of bamboo culm, and those in the latter more oriented towards the transverse direction. Ahvenainen et al (2017) measured the microfibril angle (MFA) of bamboo parenchyma cells to be about 65° with wide-angle X-ray scattering, which well supported our deduction.
For fibers cellulose, the strong and sharp positive bands at 1161 cm −1 , 1426 cm −1 and 1373 cm −1 demonstrated that the cellulose was arranged in a highly parallel orientation to the fiber axis. On the contrary, the three negative cellulose bands indicated the cellulose in the parenchyma cell wall was arranged more transversely. The band at 1161 cm −1 corresponding to C-O-C cellulose backbone (Marchessault 1962;Liang et al. 1960), therefore, indicates a parallel orientation of this functional group to cellulose polymers. The bands at 1426 cm −1 and 1373 cm −1 are cellulose-1161 cm −1 , 1373 cm −1 and 1426 cm −1 ; xylan-1461 cm −1 and 1738 cm −1 ; lignin-1505 cm −1 and 1601 cm −1 assigned to the cellulose side groups (Marchessault 1962;Liang et al. 1960), which are almost perpendicular to the cellulose polymers. However, due to the high crystallinity of cellulose, cellulose chains will squeeze each other, forcing the side group parallel to the main chains. This squeezing effect was related to the cellulose crystal d-spacing. Our previous work demonstrated that the cellulose in fibers had smaller crystal d-spacing than that in parenchyma cells (Ren et al. 2021), which could give a reasonable explanation for the stronger cellulose side group orientation in the fibers.
In the case of xylan, two significant bands at 1738 cm −1 and 1461 cm −1 were detected in Fig. 3. According to Marchessault (1962), the carbonyl group (1738 cm −1 ) had a transition moment at an angle of 54° to the polymer backbone. The C-H in the xylan (1461 cm −1 ) (Marchessault 1962) has a different vibrational energy compared to the C-H in cellulose (1373 cm −1 ) (Liang et al. 1960), which makes it possible to selectively analyze the orientation of xylan and cellulose. Furthermore, it was reported that the C-H in xylan is also perpendicular to the main chain (Marchessault 1962). Hence, the negative bands at 1738 cm −1 and 1461 cm −1 in fibers indicated that the xylan was arranged in an orientation perpendicular to the cellulose backbone. The xylan in the parenchyma cells also showed perpendicular orientation, but to a less extent as compared to that of fibers.
Lignin has two important vibration bands (1505 cm −1 , 1601 cm −1 ) suitable for analysis (Faix 1991). These vibrations occur parallel to the aromatic ring structure (Peng et al. 2019). The weak but still distinct negative bands at both 1505 cm −1 and 1601 cm −1 indicated lignin was also preferentially arranged in the cell walls of both fibers and parenchyma cells, but with less degree than the xylan. Furthermore, the lignin from fibers is more preferentially oriented than that from the parenchyma cells. Figure 4 showed relative absorption change of the correlative bands assigned to cellulose, xylan and lignin, with increased polarization angle. In this plot, the bands with vibration oriented parallel to the cell wall axis will have stronger relative absorption at a lower polarization angle, while the bands with vibration oriented perpendicular to the cell wall axis are stronger at a higher polarization angle (Peng et al. 2019;Olsson et al. 2011). Compared with orientation spectra, the angular dependence of absorption intensity contains more information (Peng et al. 2019;Chang et al. 2014). The relative absorption of all the three cell wall polymers showed distinct and stable angular dependence, valid for both fibers and parenchyma cells of the two tested bamboo species. This gives stronger evidence for the preferential orientation of cell wall polymers in bamboo. What is more, all the cell wall polymers from fibers and parenchyma cells exhibited opposite angular dependence, which further demonstrated the cell wall polymers of fibers are as a whole longitudinally oriented, while they are transversely oriented in the parenchyma cells. A closer view will further reveal the difference in angular dependence between bamboo fibers and parenchyma cells, with higher consistency for the former, indicating a more ordered polymer arrangement as a whole in the fibers, as compared to parenchyma cells.
Considering MFA is closely related to the cellulose orientation, the average MFA of both fibers and parenchyma cells can be estimated from Fig. 4. As shown in Fig. 4A, all the three cellulose bands (1161 cm −1 , 1373 cm −1 and 1426 cm −1 ) in fibers exhibited a consistent negative correlation between absorption intensity and polarization angle. Based on the high absorption of these three absorption bands at a small polarization angle (0-10°) for the fibers and large angle (80°-90°) for the parenchyma cells, a small MFA (~ 10°) and a large MFA (~ 80°) could be estimated for the fibers and parenchyma cells, respectively. This result agrees well with the previous results by XRD measurement (Zhang et al. 2020).

Changes of polymer orientation after lignin removal
Cell walls of higher plants are featured with a tough and relatively rigid reinforced composite structure (Zhao et al. 2019). That structure could be analogous to reinforced concrete, in which cellulose fibrils act as reinforcing steel bar and hemicellulose-lignin matrix acts as the concrete (Simmons et al. 2016). Therefore, the interaction between these polymers played a significant role in the mechanical behavior of bamboo cell wall. In the present study, the orientation change of cell wall polymers after directional chemical removal was proposed to understand their interactions.
Acidified sodium chlorite is capable of directionally removing aromatic ring substances and has little 1 3 Vol:. (1234567890) Fig. 4 The relative absorption of IR characteristic absorption bands related to cellulose (A, B), xylan (C, D), lignin (E, F), plotted against the polarization angle for HF, HP, CF, CP effect on polysaccharides (Zhang et al. 2018). Figure 5 shows the IR spectra of fibers and parenchyma cells (HM and CG) before and after lignin removal. The significant intensity reduction of the bands at 1601 cm −1 and 1505 cm −1 (the C=C aromatic ring vibrations plus C=O stretch) demonstrated lignin was partially removed, while the high and sharp band at 1738 cm −1 suggested xylan was much less affected. Figure 6 presented the polymer average orientation spectra of fibers and parenchyma cells (HM and CG) after lignin removal. The bands at 1601 cm −1 and 1505 cm −1 in the spectra of parenchyma cells almost disappeared after lignin removal. However, the residual lignin in the fibers still maintained a relatively stronger orientation, indicating an interaction between residual lignin and xylan still remained in the fibers. This interaction facilitates the xylan-guided orientation of residual lignin. Therefore, it is reasonable to infer that a stronger interaction exists between lignin and xylan in bamboo fibers. However, the average orientation of xylan (1738 cm −1 and 1461 cm −1 ) and cellulose (1161 cm −1 and 1461 cm −1 ) is little affected after lignin removal (Fig. 6), suggesting a strong interaction between cellulose and xylan. That is consistent with several previous studies (Simmons et al. 2016;Zhang et al. 2015). Xylan, the most prevalent non-cellulosic polysaccharide, was crosslinked to lignin, and also bound intimately to cellulose microfibrils in the cell wall (Simmons et al. 2016;Terrett and Dupree 2019). Advanced 13 C solid-state magic-angle spinning (MAS) 2D NMR spectroscopy directly demonstrated xylan is spatially close to cellulose via its twofold screw conformation (Simmons et al. 2016). Molecular dynamics (MD) simulation further demonstrated xylan interacted with cellulose through three typical binding modes including bridge, loop and random (Zhang et al. 2015).

Changes of polymers orientation after xylan removal
Alkaline treatment, an effective method for xylan extraction, have various effects on cell wall polymers including cleaving the ether and ester linkages the in lignin-hemicellulose complex, as well as the ester bonds between lignin and hydroxycinnamic acid (Wen et al. 2011). Figure 5 showed the IR spectra of fibers and parenchyma cells (HM and CG) before and after alkaline treatment. The disappearance of the band at 1738 cm −1 (C=O vibration in O=C-H of acetyl groups and O=C-O of carboxyl groups) demonstrated that O-acetyl groups and carboxyl groups in xylan were disrupted (Wen et al. 2011(Wen et al. , 2015, and xylan was partially removed. However, the band intensity at 1601 cm −1 and 1505 cm −1 only slightly decreased, suggesting lignin was much less affected by alkaline treatment. Figure 7 showed the average orientation spectra of cell wall polymers from fibers and parenchyma cells (HM and CG) after alkaline treatment. The complete disappearance of the band at 1738 cm −1 was attributed to the destruction of O-acetyl groups and carboxyl groups in xylan.
The complete disappearance of bands at 1601 cm −1 and 1505 cm −1 in the parenchyma cell average orientation spectra indicated the residual lignin lost its orientation after alkaline treatment. That further demonstrated the orientation of lignin was largely dominated by xylan in the parenchyma cell wall, again strongly supporting the strong interaction between  (Terrett and Dupree 2019;Jung et al. 2011). This interaction can be roughly divided into two types. One was the lignin-carbohydrate complex (LCC), which is featured with various chemical bonds including phenyl glycoside (PhGlc), benzyl ether (BE) and ester linkages (Fengel and Wegener 1984). The second interaction was proposed recently based on the evidence from advanced 13 C solid-state MAS 2D double-quantum (DQ) correlation NMR spectroscopy, which found cross bands in S3/5 (C3/5 in syringyl)-X4, revealing the electrostatic interactions between lignin and xylan (Kang et al. 2019). Moreover, weak electrostatic interactions between cellulose-lignin have also been proposed via this technology (Kang et al. 2019). Differing from parenchyma cells, the lignin in the fibers still maintained a certain orientation after alkaline treatment, indicating lignin in the fibers is tougher to alkaline treatment, and also implying the difference of lignin structure between the two types of bamboo cells. Our recent research (Zhu et al. 2022) also indicate that the lignin in bamboo parenchyma cells had a higher content of β-O-4 substructures and a higher S/G ratio, but less β-β and β-5 linkages than the lignin in bamboo fibers. More β-O-4 linkages imply the lignin in parenchyma cells is more easier to be disrupted under alkaline treatment. The cell wall structure evolution model after lignin and xylan removal Based on the above analysis, the cell wall structure evolution model after lignin and xylan removal is presented in Fig. 8. As a whole, the cellulose in bamboo fibers is nearly axially oriented whereas it is almost transversely arranged in the parenchyma cells. In the native fibers and parenchyma cells, xylan and lignin are both preferentially oriented alongside cellulose, but with different extents. The xylan is more oriented than the lignin because the former is closely associated with cellulose. The formed framework by cellulose and xylan then guided the orientation of lignin that is deposited latterly. The delignification will not significantly change the orientation of both cellulose and xylan. However, the bamboo fibers and parenchyma cells responded differently to the same procedure of delignification with more lignin remained in the former, highlighting the stronger interaction between lignin and xylan in the fibers. On the contrary, the removal of xylan significantly weakens the orientation of lignin in both fibers and parenchyma cells, and is more significant for the latter. It is therefore inferred xylan and lignin play distinct roles in the bamboo cell wall, where the former participated in the construction of the so-called cellulose-xylan framework in the cell walls, whereas the lignin mainly acts as a filler of this framework.
Furthermore, our work found little discrepancy between HM and CG, indicating that the cell wall polymer arrangement is not related to the specific Fig. 7 The average orientation spectra of HF (A), HP (B), CF (C), CP (D) after alkaline treatment bamboo species. The different polymer arrangement characteristics between bamboo fibers and parenchyma cells should be attributed to their different cell wall functions. In particular, the stiff fibers provide mechanical support for the bamboo stem, while the soft parenchyma cells contribute much less to axial mechanical strength (Malanit et al. 2011), but function as nutrition storage (Abe and Yano 2010). Therefore, all the polymers in the bamboo fibers are more orderly arranged in the longitudinal direction to achieve better mechanical properties, while they were oriented perpendicular to the cell axis in the parenchyma cells with relatively weaker polymer interaction, especially between lignin and xylan.

Conclusion
In this work, Imaging Polarized FTIR microspectroscopy, combined with directional chemical removal, were applied to reveal the polymer orientation and interaction in the cell walls of two bamboo species, and significant differences were observed between fibers and parenchyma cells. All the polymers in the cell walls of bamboo are more or less oriented. Cellulose, as expected, is the most oriented polymer, followed by xylan and lignin in order. All the polymers in bamboo fibers are as a whole axially oriented whereas it is almost transversely arranged in parenchyma cells. Furthermore, the orientation degree of xylan and lignin in the parenchyma cells is not as strong as that in fibers. It seemed the lignin and xylan in fibers were more difficult to be removed as compared to parenchyma cells, according to the orientation analysis after directional chemical removal. That gives indirect but rational evidence to support the assumption that stronger interaction exists between lignin and xylan in the fibers. The results also supported the viewpoint that cellulose and xylan are combined to form the framework in the bamboo cell wall, with lignin acting as fillers. Briefly, it was believed the cell walls of fibers are more orderly and compactly assembled in bamboo, as compared to parenchyma cells, which can give a reasonable explanation on the well-observed higher biomass recalcitrance of bamboo fibers.
Funding This work was financially supported by the National Natural Science Foundation of China (Grant No. 31770600 and 32001255).

Conflict of interest
The authors declare that there is no conflict of interest and that the manuscript has been approved by all authors .  Fig. 8 Schematic illustration showing the cell wall evolution of bamboo fibers and parenchyma cells after lignin and xylan removal. All the polymers in fiber are nearly axially oriented while they are almost transversely arranged in parenchyma cells. After lignin and hemicellulose removal, more lignin and xylan can be remained in fibers, indicating a stronger interaction between them, as compared to parenchyma cells