Probing and visualizing the heterogeneity of fiber cell wall deconstruction in sugar maple (Acer saccharum) during liquid hot water pretreatment

Abstract

Liquid hot water (LHW) pretreatment is an effective method to improve enzymatic digestibility of lignocellulosic feedstock by removing hemicellulose/lignin. To better understand how LHW pretreatment reduces plant cell wall recalcitrance, we applied a combined approach using multiple microscopic techniques and chemometric methods to monitor microstructural and topochemical changes in the fiber cell wall of sugar maple (Acer saccharum). The heterogeneity of deconstruction in various cell wall layers was easily visualized based on chemical characterizations through confocal Raman microscopy, combining with principal component analysis and cluster analysis. Interestingly, after LHW pretreatment, the S2 layer was differentiated into two regions, namely a heavy-damaged region (outer and thin inner S2) with more polysaccharides removed, and a light-damaged region (middle S2) which still remained relatively intact. Our results have established a direct correlation between microstructural and topochemical changes of a cell wall following LHW pretreatment. The removal of polysaccharides (mainly hemicelluloses) rather than lignin played a critical role in the visible damage of the cell wall, like cavities, gaps and collapses.

---

1 Introduction

Biofuel is a sustainable and renewable energy resource. It has received much attention and been extensively studied in recent years because of high worldwide demand for energy and diminishing supplies of fossil fuels.¹ As a readily available, low-cost and large-scale feedstock of biofuels, non-food plant biomass can capture huge amounts of solar energy. This feedstock plays a crucial role in the development of second generation bioethanol.^2–4 However, the resistance of biomass to enzymatic and microbial deconstruction is a major obstacle for biomass conversion to ethanol.^5,6 It is hypothesized that the recalcitrance of biomass is caused by the structural and chemical complexity of plant cell walls.⁷

To overcome this barrier and provide a low-cost bioconversion process, a number of different pretreatment methods involving biological, chemical, physical, and thermal approaches have been investigated over the years.^10–15 Liquid hot water (LHW) pretreatment, one of the most promising pretreatment process of lignocellulosic material, is an effective and environmentally friendly technology.¹⁶ This method is able to break the intricate structure of biomass and therefore increase the cellulose accessibility to enzymes.

To clarify the mechanism of LHW pretreatment and improve the conversion process, number of prior research studies focused on changes of chemical compositions, efficiency of enzymatic hydrolysis, yield of main degradation products, and comparison of pretreatment conditions, etc.^17–20 During pretreatment, hot water under pressure penetrates the cell structure of biomass, hydrates cellulose, and dissolves hemicellulose and lignin.²¹ Li et al.²² pointed out that hemicellulose removal helped eliminate or minimize the enzymatic barrier of cellulose and then improved the efficiency of ethanol production from Miscanthus lutarioriparious. With the development of analytical microscopy approaches, microstructural and topochemical characterizations of cell walls after LHW pretreatment have been investigated.^23,24 Kim observed the anatomical features of sugar maple cell walls after hot water treatment using light microscopy (LM) and scanning electron microscopy (SEM); collapsing of cell wall, changes in pore size and formation of lignin droplets were illustrated.²⁵ Ma et al.^26,27 investigated the chemical composition distribution of the poplar cell walls untreated and pretreated with LHW. Confocal Raman Microscopy (CRM) images exhibited the lignin and carbohydrate distribution, and transmission electron microscopy in combination with immunogold labeling revealed the topochemical changes of xylan during pretreatment. However, a direct correlation between components removal and microstructural changes of cell walls seems to be still unestablished. The interaction of hemicellulose dissolution and structural changes such as the formation of pores and gaps in the cell walls is still unclear. The compositional and structural heterogeneity among different regions of cell walls before and after treatment, which is crucial to establish the above correlation, remains to be further evaluated.

Sugar maple, also known as hard maple or rock maple, is one of the largest and most prominent hardwoods in eastern North America, and is extensively employed in the North American paper industry,⁸ and is also a potential carbon source for ethanol production.⁹ In the present work, we investigated the morphological and chemical changes in sugar maple cell walls after LHW pretreatment at 170 °C by using multiple visualization techniques, such as LM combined with staining, SEM, and fluorescence microscopy (FM). More importantly, we explored the topochemical changes of pretreated fiber cell walls by CRM. Chemometric tools (i.e. principal component analysis and cluster analysis) were applied to interpret the Raman datasets for visualizing the heterogeneity of cell wall deconstruction. Our work combining the morphological and topochemical information of the cell walls is expected to reveal the heterogeneity among cell wall regions and thus provide further insights into the mechanism of cell wall deconstruction during LHW pretreatment.

2 Materials and methods

2.1 Materials

Sugar maple (Acer saccharum) samples in this study were collected from one 6 year-old living tree (1.3 m above ground) in the State University of New York and then manually cut into small blocks (15 × 10 × 5 mm³). After degased by boiling deionized water, these blocks were transversely sectioned into 10 μm (ref. 28) slices by a sliding microtome (Leica 2010R; Leica Microsystems, Wetzlar, Germany) for following liquid hot water pretreatment.^27,29

2.2 Liquid hot water pretreatment

Liquid hot water pretreatment process was carried out in a high-pressure reactor with a total volume of 250 ml. About 10 pieces of wood sections and 100 ml deionized water were placed in the stainless steel reactor with a programmable temperature controller. The reactor was heated to 170 °C with a heating rate of approximately 3 °C min⁻¹. After the designed reaction time (10, 30 min), it was immediately cooled with ice to a temperature lower than 100 °C and slowly depressurized.³⁰ Most sections were then carefully collected between two glass slides and stored in a refrigerator at 4 °C for the subsequent experiments.

2.3 Light microscopy and fluorescence microscopy

After dehydration through a graded series of ethanol solution, the original and pretreated transverse sections were mounted in glycerol and covered with a coverslip (0.17 mm thickness). Subsequently, the sections were examined with a Leica DM 2000 fluorescence microscope with an ultra-pressure mercury lamp for illumination. The excitation wavelength was 435–480 nm and the emission wavelength at 495–600 nm was used for imaging lignin autofluorescence.^31,32 In addition, cross sections were positioned on glass microscope slides and stained with 0.01% toluidine blue (TB). Then the images were captured using a light microscope (Leica, DM 2500).

2.4 Field emission scanning electron microscopy

The surface morphological features of untreated and pretreated sugar maple sections were determined by an FE-SEM apparatus (Hitachi S-4300) at an accelerating voltage of 5 to 10 kV. After sectioned into thin slices and treated with LHW, samples were prepared by vacuum drying. Subsequently, they were mounted on carbon tape-adhered aluminum stubs and sputter coated with 10–12 nm gold particles by using a vacuum sputter prior to acquiring images.

2.5 Confocal Raman microscopy

After washed with deionized water, the native and pretreated sections were placed on a glass slide with a drop of ultrapure water and then covered with a coverslip of 0.17 mm thickness for Raman detection. Raman spectra and Raman mapping were acquired with a LabRam Xplora confocal Raman microscope (Horiba Jobin Yvon, Longjumeau, France) equipped with a confocal microscope (Olympus BX51; Olympus, Tokyo, Japan) and a motorized x, y stage. A linear polarized laser (diode-pumped green laser, λ = 532 nm), focused with a diffraction-limited spot size (0.61λ NA⁻¹), and a high numerical aperture (NA) microscope objective (Olympus 100×, oil, NA = 1.40) were used to conduct measurements to achieve high spatial resolution. The laser power on the sample was approximately 8 mW. The Raman light was detected by an air-cooled, front-illuminated spectroscopic charge-coupled device (CCD) behind a grating (1200 grooves per mm) spectrometer. For mapping, an integration time of 2 s was chosen and every pixel corresponds to one scan with a spectrum acquired every 0.6 μm by averaging 2 s cycles.

LabSpec 5 software (Horiba Jobin Yvon) was used to obtain Raman spectra in specific spectral ranges (600–3200 cm⁻¹) and correct baseline. This software was also used for the calculation of average spectra in certain lamellas and the acquisition of Rama mapping data sets. The mathematical software Matlab 2014b (MathWorks) was used to generate Raman mapping images.

2.6 Principal component analysis and cluster analysis

The Raman date sets were further analyzed by different multivariate methods (principal component analysis (PCA) and cluster analysis (CA)). The mathematical software Matlab was also used for data processing. The Raman spectra were baseline corrected and smoothed using the adaptive iteratively reweighted penalized least-squares (airPLS) algorithm.³³ The PCA algorithm and the K-means CA algorithm were performed by applying the self-built Matlab functions and procedures as described in the literature.³⁴

3 Results and discussion

3.1 Anatomical and chemical characteristics of two types of fiber cell walls

The secondary xylem of sugar maple is illustrated in Fig. 1. We observed three types of cells: xylem ray parenchyma, xylem fiber, and vessel. Here, we focused on xylem fiber cells due to their prevalence and superior value in end use applications. The xylem fiber cells could be further divided into the thick-walled and the thin-walled ((A) and (B) respectively in Fig. 1a). These two types of fiber cells showed the heterogeneity in cell wall thickness and lignin concentration. The thickness was measured by using Image-Pro Plus software (Media Cybernetics) and the results are shown in Fig. S1 of ESI.† The average cell wall thickness of thick-walled fibers was 1.298 times that of thin-walled fibers. As TB stains lignin (polyphenols) blue-green,³⁵ the coloring degree suggests the varied content of lignin in cell walls. The thick-walled fiber cells have higher concentration of lignin than thin-walled fiber cells, which can be further supported by the fluorescence image in Fig. 1b.

Fig. 1 (a) Cross-section of untreated sugar maple under light microscope, toluidine blue staining. (b) Fluorescent image of cross-section of untreated sugar maple. Circle (A) thick-walled fiber cells. Circle (B) thin-walled fiber cells.

To accurately quantify the intensity of fluorescence and further analyze the discrepancy between two types of fiber cell walls, the fluorescent image (Fig. 2a) of untreated transverse section of sugar maple was analyzed across multiple cells along a straight line using Matlab. This image is an RGB image and the values of red and blue components are zero. The green-component values of all pixels on the line through five double-cell-walls (as shown by the broken-line arrow in Fig. 2b, and the double-cell-walls are numbered from 1 to 5) were presented in Fig. 2c. As we can see from the green intensity distributions, peak value at the center of each peak correspond to compound middle lamella (CML) region and as expected the intensity is zero at the lumen region. The distance between two adjacent peaks indicates the cell size and the peak width indicates the thickness of cell wall. Based on the peak width, we hold that double-cell-wall 3, 4, and 5 were from thick-walled fiber cell walls, and double-cell-wall 1, 2 were form thin-walled fiber cell walls. Although these five double-cell-walls had different degrees of lignification, the characteristics of lignin distribution around the CML was similar. The detailed variation of fluorescence intensity on both sides of the CML was shown in Fig. 2d. The double-cell-wall 1 and 2 had much lower fluorescence intensity, however, there was still a significant correlation between all five fluorescence intensity profiles as demonstrated by the Pearson correlation coefficient (P < 0.01, data not shown).

Fig. 2 (a) Fluorescent image of cross-section of untreated sugar maple. Images (b) is a higher magnification of the region boxed in (a). (c) Green-component values of all pixels on the line through five double-cell-walls (as shown by the broken-line arrow in (b)). (d) Detailed profile of five peaks. DCW, double cell wall.

Additionally, the pattern of lignin distribution among the different morphological regions of the two types of fiber cells is very similar. As demonstrated by the fluorescence intensity and staining degree, the cell corner (CC) and CML had a higher concentration of lignin than the S2 layer. Ji et al.³¹ determined the distribution of lignin in wood tracheids of Pinus yunnanensis using fluorescence microscopy and showed that the CC and CML regions were more highly lignified than the S2 layer, which is consistent with our results. As a result, only thick-walled fiber cell walls were then further investigated due to their representativeness.

3.2 Morphological changes of fiber cell walls after LHW pretreatment

To show the morphological changes of sugar maple sections after LHW pretreatment, a series of images was obtained using FE-SEM. Compared with the flat and intact cell walls before treatment, pretreated cell walls were seriously disrupted (Fig. 3). After the sections were pretreated with hot water for 10 min, lamellas started to be detached and became easier to be distinguished (Fig. 3c and d). Cavities of large size appeared in the outer part of SW; thin lamina in the boundary of CC and CML after delamination was also observed. Prolonging the pretreatment time to 30 min, the boundary layer of cavities disappeared and the delamination was completed (Fig. 3e and f, arrow). This visible structural damage with the formation of pores and gaps may be due to the removal of some of chemical components (mainly hemicelluloses); alternatively, as shown earlier, the structural and topochemical heterogeneity in the CC and CML regions may have contributed to their varying degrees of structural changes and separation of the lamella due to LHW pretreatment.

Fig. 3 Morphological changes of various regions of fiber cell walls monitored by SEM before and after treatment for 10, 30 min with hot water at 170 °C. (a, c and e) The CML region; (b, d and f) the CC region. S, secondary cell wall; CML, compound middle lamella; CC, cell corner. Arrow: pores and gaps. Circle: thin lamina after delamination. Scale bar = 1 μm.

Fig. 4 shows more detailed information about morphological changes for the different cell wall regions after LHW pretreatment for 30 min. Although the CML and CC became rough and slightly folded, they still kept intact and distinguishable (Fig. 4a and d). In contrast, the SW has suffered damage. Interestingly, this kind of deconstruction was not homogeneous in different part of SW. Gaps (Fig. 4b) and holes of large size in the outer SW (Fig. 4c) were observed. Some of holes also appeared in the inner part (lumen side) and middle part of SW after treatment, but they were smaller in size. These results clearly indicate that different cell wall regions of the same cell have different degrees of resistance to hot water, which is a reflection of biomass heterogeneity. The outer SW layer was suggested to be easily destroyed during LHW treatment, whereas the CC and CML represented a higher ability to resist degradation. This may be related to the higher lignification of the CC and CML. As for different regions of the SW layer, the variation of pore size and the number of pores formed may be due to factors such as the spatial distribution of polymer matrix in the cell wall and their accessibility to hot water rather than the average chemical composition. Similar differences in pores formation and size of the pores have been observed earlier: large cavities formed at S1/S2 boundary and smaller cavities distributed throughout the S2 region were observed in the pine wood after nitric acid pretreatment and iodine staining.³⁶

Fig. 4 SEM images of thick-walled fiber cell walls pretreated with hot water at 170 °C for 30 min (a–e). (a) Pores and gaps. Images (b) and (c) are the higher magnification of the regions boxed in (a). (d) The internal morphology of the gap. (e) The surface morphology of pores.

Due to the existence of gaps, morphological features of the cell walls can also be observed. As shown in Fig. 4b, d and e, regular tube-like pores extending into the cell walls can be observed. The SEM images here, however, cannot confirm the shape of the pores as they extend into the cell walls from the surface. Donohoe et al.³⁷ reported lignin coalescence and migration through maize cell wall after dilute acid pretreatment. The presence of regular, narrow pores extending into the cell walls was also shown using TEM tomogram. Despite detailed information on the pore internal architecture, they showed the formation and extrusion of lignin droplets and redeposition on the cell wall surface. Similar migration of lignin is also observed in this study using FE-SEM as illustrated in Fig. 5. After LHW pretreatment for 10 min, the SW region adjoining the boundary of the SW and CML on was severely damaged; numerous protuberances of different size, gaps and cavities appeared. The pronounced papillary in the CML (Fig. 5c, white arrow) showed the extrusion of lignin droplet. Fig. 5b showed two round lignin droplets on the surface of SW, which migrated out of the other regions of cell wall to redeposit here.

Fig. 5 SEM images of thick-walled fiber pretreated with hot water at 170 °C for 10 min (a–c). (a) Severe damage on the boundary of the SW and CML, and round droplets on the surface of cell wall. Images (b) and (c) are the higher magnification of the regions boxed in (a).

3.3 Topochemical characterization of fiber cell walls before and after LHW pretreatment

3.3.1 Average Raman spectra of different fiber cell wall lamellas. Fig. 6 shows the average Raman spectra for the CC (Fig. 6a), CML (Fig. 6b), and SW (Fig. 6c) of fiber cell walls untreated and treated with LHW at 170 °C for 10 and 30 min. The spectral range from 550 to 3150 cm⁻¹, which includes bands from the wood components such as cellulose, hemicelluloses, and lignin, is of interest. However, no spectral features were found in the region from 1800 to 2750 cm⁻¹ neither before nor after LHW treatment; therefore, this region is not presented here. There is little difference between the Raman spectral features of cellulose and hemicelluloses; the hemicelluloses bands are broader and can reside beneath the cellulose bands as a result of their lower amount and amorphous character.³⁸ Therefore, the bands from cellulose and hemicelluloses cannot be distinguished and we refer to them as polysaccharides (PSs) in this study.

Fig. 6 Average Raman spectra for the CC (a), CML (b), and SW (c) of the fiber cell walls before and after treatment with hot water at 170 °C for 10 and 30 min.

The bands at 1096, 1377 and 2889 cm⁻¹ are assigned to C–O–C stretch (asymmetric), HCC, HCO, HOC bend, and C–H, C–H₂ stretch, respectively. These mainly originate from PSs. The bands at 1331, 1596, 1656, 2850, and 2933 cm⁻¹ mainly from lignin are assigned to aliphatic O–H bend, aromatic ring stretch (symmetric), C=C stretch of coniferyl alcohol, C–H stretch in OCH₃ (symmetric), and C–H stretch in O–CH₃, respectively. The band at 1454 cm⁻¹ from both PSs and lignin is assigned to HCH and HOC bend, CH₃ bend in OCH₃.^39–41

Before pretreatment, the fiber cell walls of sugar maple had prominent Raman peaks with high intensity. As shown by black line in Fig. 6a–c, the spectra from different lamellas of untreated cell walls presented different features. As for the CC, the band intensities at 1331, 1596, and 2933 cm⁻¹ were high, which was in marked contrast to the very low band intensity at 2889 cm⁻¹. This demonstrated the highest concentration of lignin in the CC. However, the highest band intensity in the SW is at 2889 cm⁻¹, indicating SW mainly contains PSs. As the transition, the CML has a medium concentration of lignin and PSs.

After pretreatment with hot water at 170 °C, a decrease in band intensity was observed at all Raman bands for the whole cell wall area; however, the degree of change in intensity of these bands differed across the three cell wall regions. The intensity of bands at 1331 cm⁻¹, 1596 cm⁻¹ and 1656 cm⁻¹, which are only assigned to lignin, decreased significantly for the CC during pretreatment, in contrast with only a slight decline of intensity for the SW. However, a significant decline in the band intensity of 2889 cm⁻¹ for the SW was observed, which indicted significant removal of PSs from the SW region after the LHW pretreatment. As for the CML, the decline in Raman intensity for both lignin bands and PSs bands, which indicates the decrement of compositional concentration, was observed. This kind of component changes was further accurately quantified by chemical components analysis (as shown in Table S1 of ESI†). As expected, a large amount of hemicelluloses (59.24%, w/w) were solubilized after LHW pretreatment for 40 min. At this treatment severity, about 20.09% Klason lignin and 48.78% acid-soluble lignin were also removed by hot water.

3.3.2 Chemical imaging of fiber cell walls by CRM. To obtain more information about chemical variation in the cell walls over a large area during LHW pretreatment, and enhance visualization of compositions (i.e. lignin and polysaccharides) distribution, we used Raman mapping technique. Chemical images of sugar maple cell walls obtained by CRM are presented in Fig. 7. The transverse sections untreated and treated with hot water for 10 min, 30 min were investigated. Rectangles in bright field images represent the selected areas of mapping.

Fig. 7 Raman mapping of thick-walled fiber cell walls before and after LHW pretreatment: (a–c) bright field images; (d–f) lignin maps; (g–i) polysaccharides maps. Rectangles in bright field images (a–c) indicate the selected areas of mapping.

In order to ensure accurate assessment of lignin content and distribution, Raman images of lignin distribution were generated using peak area of both the 1596 and 1656 cm⁻¹ bands. This is justified because both bands are characteristics of lignin features (1596 cm⁻¹ band assigned to aromatic ring stretch) and they partially overlap each other.²⁸ Therefore, we used the combined band region of 1546–1706 cm⁻¹ and calculated the band area by the sloping baseline method. A remarkable decrease in image intensity can be seen in Fig. 7d–f confirming the removal of lignin from lamellas of the cell corner. However, we did not observe significant destruction of the CC in SEM images shown earlier. This indicates that lignin in cell corner was removed in such a way that it did not generate obvious gaps or cavities on the surface that can be seen by SEM.

The band at 2889 cm⁻¹ was selected instead of 1096 cm⁻¹ to investigate the PSs distribution, because the 1096 cm⁻¹ band is sensitive to cellulose chain orientation.⁴² Additionally, considering that the FWHM (full width at half maximum) at 2889 cm⁻¹ is correlated to cellulose crystallinity⁴³ and this band is partially overlapped by lignin band at 2933 cm⁻¹, Raman images of PSs distribution (Fig. 6g–i) was obtained using peak height of the band at 2889 cm⁻¹. The height was calculated by a baseline method. In these images, a remarkable decrease of image intensity was also observed. Compared with the untreated sample (Fig. 7g), the Raman intensity of the CC at 2889 cm⁻¹ in the cross section treated with LHW at 10 min (Fig. 7h) increased, which may be due to the removal of lignin from the CC and that more PSs was exposed to the detector. At prolonged pretreatment time (30 min), the concentration of PSs in all of the regions decreased and as expected the width of the CML region increased (Fig. 7i). In fact, this broadening of the CML was not an anatomic alteration. During LHW treatment, the certain region of SW adjoining the CML was damaged more severely and then more PSs was removed. As a result, the image intensity of this region was similar to that of the CML. This result, together with above SEM images, demonstrated that the part of outer SW (in fact, S1 layer and outer S2 layer) was easier to be deconstructed and significant amount of polysaccharides (mainly hemicelluloses) were removed during LHW pretreatment.

3.4 Multivariate data analysis for Raman imaging

Multivariate statistical analysis is a powerful tool to process very large datasets, like Raman spectra in this study. Principal component analysis is an effective tool for data reduction from large data sets while retaining relevant details of the image.⁴⁴ Additionally, cluster analysis is able to take large data sets and group the cell walls into sublayers based on their chemical composition.⁴⁵ It is well-known in the literature that normally SW can be divided into S1, S2 and S3 layers. Since the S2 layer is the thickest (75–85% of the total thickness of the cell wall) and most important for mechanical stability,⁴⁶ its further-layered characteristics should be focused. Therefore, the above two multivariate methods are used to further investigate the heterogeneity within the S2 layer of cell walls before and after LHW pretreatment.

3.4.1 Similarity of Raman spectra from different regions of the S2 layer before LHW pretreatment. Due to the relative uniformity of the S2 layer, it is hard to further divide the S2 layer into several sublayers just based on the morphological features. To solve this problem and obtain the Raman spectra from different regions of the S2 layer, the method for automatically identifying and extracting spectra of different wood cell wall layer was used.³⁴ This method is based on cluster analysis, which is able to reveal the natural compositional segregations in the cell walls.⁴⁵ As illustrated in the Fig. 8a, the S2 layer was partitioned into two regions (cluster 2 and cluster 3) based on its chemical features determined by Raman spectra. However, the region contours were not regular, which demonstrated that the S2 layer was relatively homogeneous. Average Raman spectra were calculated (Fig. 8b) to further investigate the difference of composition between these two regions. We can see that the two spectra were similar, little difference in peak height and other spectra properties was observed. This indicated that there was no significant difference in chemical composition between the regions within S2 layer of cell walls before LHW pretreatment.

Fig. 8 (a) Cluster map of untreated cross-section; the number of clusters is set to k = 6. Cluster 1–6 correspond to six sublayers respectively. The S2 layer is partitioned into two sublayers: cluster 2 and 3. (b) Average Raman spectra of clusters 2 and 3.

3.4.2 Two regions of the S2 layer with different degrees of deconstruction after LHW pretreatment. The S2 layer of sugar maple cell walls obtained different degrees of deconstruction after LHW pretreatment for 30 min. Based on the differences observed in the Raman spectra, the S2 layer can be further divided into two kinds of regions by cluster analysis (Fig. 9a). The outline of these regions was consistent with that of the morphological regions based on structural features revealed by SEM. As shown in Fig. 9c, pores and gaps were abundant in the regions of inner and outer S2, in contrast with the middle part of S2 layer (red arrow) with light damage. Therefore, we can roughly divide the S2 layer into the light-damaged region and the heavy-damaged region. Furthermore, we investigated the chemical heterogeneity of these two kinds of regions after pretreatment by Raman spectra (Fig. 9b). Compared with the average spectra of untreated S2 layer, the spectra of S2 layer after LHW pretreatment changed significantly. However, changes in the intensity of some Raman bands differed significantly between the light-damaged region and the heavy-damaged region. This difference in Raman bands may be related to the nonuniform deconstruction in wood cell walls. Therefore, in order to further evaluate this heterogeneity, three spectral regions including the above Raman bands were highlighted and illustrated in Fig. 9d–f.

Fig. 9 (a) Cluster map of cross-section of sugar maple pretreated with hot water at 170 °C for 30 min; the number of clusters is set to k = 6. (b) Average Raman spectra of untreated S2 layer and two regions in (a). The inset (c) is the SEM image of fiber cell wall pretreated with hot water at 170 °C for 30 min. (d–f) Zoom into regions of interest of average Raman spectra in (b).

As for the region from 875 to 1200 cm⁻¹ (Fig. 9d), a decrease in band intensity was observed at 903, 970, 1039, 1094, 1117, and 1151 cm⁻¹ for the whole S2 layer after LHW pretreatment. However, the extent of the decline for the two sub regions of the S2 layer was different. After treatment, the Raman intensity of PSs bands at 1094, 1117, and 1151 cm⁻¹ in the light-damaged region was higher than that in the heavy-damaged region, which indicated that more PSs was removed from the heavy-damaged region during LHW pretreatment. This conclusion can be further confirmed by the Raman spectra in the region from 2800 to 3000 cm⁻¹ (Fig. 9f). As illustrated, the degree of intensity decline of the representative PSs band at 2889 cm⁻¹, compared with the untreated S2 layer, differed significantly for the light-damaged region and the heavy-damaged region. In the region from 1250 to 1500 cm⁻¹ (Fig. 9e), the PSs band at 1377 cm⁻¹ presented the same tendency of intensity change. However, the Raman intensity of the lignin band at 1331 cm⁻¹ for the two regions was almost uniform after treatment. The same situation was also found for other lignin bands like 1596 and 1656 cm⁻¹ (as shown in Fig. 9b). This indicated that the removal of lignin did not account for the heterogeneity of the S2 layer deconstruction.

By combining the obtained morphological and chemical information, a potential explanation for the observed heterogeneity with varying degrees of damage in the cell wall is offered. During LHW pretreatment, it is hypothesized that different layers of cell wall undergo varying degrees of swelling or structural changes. This potentially causes delamination at the boundary between CML and SW layers. Due to the high lignification and compact structure for the CC and CML, these two lamellas remained intact and complete after treatment. Interestingly, although the middle S2 layer has lower concentration of lignin and loose structure, it suffered very light destruction during pretreatment. However, the outer S2 layer and the thin inner S2 layer were damaged more heavily, because this region may have a higher accessibility to hot water or ultrastructure was easily damaged due to adjacent delamination at the CML-S2 interface.

4 Conclusion

The morphological and topochemical changes of sugar maple were investigated before and after liquid hot water pretreatment, using multiple visualization techniques like field-emission scanning electron microscopy, fluorescence microscopy and confocal Raman microscopy. Additionally, chemometrics tools (i.e. principal component analysis and cluster analysis) were applied to interpret the Raman datasets and to further visualize the heterogeneity of cell wall deconstruction. Although lignin was removed from the CC and CML regions during liquid hot water pretreatment, interestingly, no visible structural changes such as formation of pores, cavities, gaps were observed in these regions. Majority of the structural changes were observed in the outer S2 region adjacent to the CML and inner S2 layer. It is hypothesized that this could be due to differential swelling between chemically heterogeneous cell wall layers and easier accessibility to hot water penetration. This study provides not only new insights on the mechanism of liquid hot water pretreatment, but also a helpful procedure to further investigate the heterogeneity of cell wall deconstruction.

Acknowledgements

The authors gratefully acknowledge the financial support from the Fundamental Research Funds for the Central Universities No. 2015ZCQ-CL-03, the Chinese Ministry of Education 113014A, and the US National Science Foundation CNIC grant IIA 1352699.

Notes and references

1. A. P. C. Faaij, Energy Policy, 2006, 34, 322–342.
2. S. P. S. Chundawat, B. S. Donohoe, L. da Costa Sousa, T. Elder, U. P. Agarwal, F. Lu, J. Ralph, M. E. Himmel, V. Balan and B. E. Dale, Energy Environ. Sci., 2011, 4, 973.
3. M. Foston and A. J. Ragauskas, Ind. Biotechnol., 2012, 8, 191–208.
4. V. Menon and M. Rao, Prog. Energy Combust. Sci., 2012, 38, 522–550.
5. X. B. Zhao, L. H. Zhang and D. H. Liu, Biofuels, Bioprod. Biorefin., 2012, 6, 465–482.
6. J. D. DeMartini, S. Pattathil, J. S. Miller, H. Li, M. G. Hahn and C. E. Wyman, Energy Environ. Sci., 2013, 6, 898.
7. M. E. Himmel, S. Y. Ding, D. K. Johnson, W. S. Adney, M. R. Nimlos, J. W. Brady and T. D. Foust, Science, 2007, 315, 804–807.
8. J. Marvin, M. Morselli and F. Laing, For. Sci., 1967, 13, 346–351.
9. R. M. Stoutenburg, J. A. Perrotta, T. E. Amidon and J. P. Nakas, BioResources, 2008, 3, 1349–1358.
10. X. Zhou, J. Ma, Z. Ji, X. Zhang, S. Ramaswamy, F. Xu and R. Sun, Ind. Crops Prod., 2014, 62, 196–203.
11. Z. Ji, Z. Ling, X. Zhang, G. H. Yang and F. Xu, BioResources, 2014, 9, 4159–4172.
12. X. Zhang, J. Ma, Z. Ji, G. H. Yang, X. Zhou and F. Xu, Microsc. Res. Tech., 2014, 77, 609–618.
13. J. Zhang, Z. Zhu, X. Wang, N. Wang, W. Wang and J. Bao, Biotechnol. Biofuels, 2010, 3, 26.
14. J. Zhang, X. Wang, D. Chu, Y. He and J. Bao, Bioresour. Technol., 2011, 102, 4480–4488.
15. W. Zeng and H. Z. Chen, Bioresour. Technol., 2009, 100, 1371–1375.
16. J. Lu, X. Li, R. Yang, J. Zhao and Y. Qu, Biotechnol. Biofuels, 2013, 6, 159.
17. H. Q. Li, W. Jiang, J. X. Jia and J. Xu, Bioresour. Technol., 2014, 153, 292–299.
18. H.-Q. Li, C.-L. Li, T. Sang and J. Xu, Biotechnol. Biofuels, 2013, 6, 76.
19. C. Cara, I. Romero, J. M. Oliva, F. Saez and E. Castro, Appl. Biochem. Biotechnol., 2007, 137, 379–394.
20. T. Y. Zhang, R. Kumar and C. E. Wyman, Carbohydr. Polym., 2013, 92, 334–344.
21. G. Yu, S. Yano, H. Inoue, S. Inoue, T. Endo and S. Sawayama, Appl. Biochem. Biotechnol., 2010, 160, 539–551.
22. H.-Q. Li, C.-L. Li, T. Sang and J. Xu, Biotechnol. Biofuels, 2013, 6, 76.
23. J. K. Ko, Y. Kim, E. Ximenes and M. R. Ladisch, Biotechnol. Bioeng., 2015, 112, 252–262.
24. R. Ibbett, S. Gaddipati, S. Hill and G. Tucker, Biotechnol. Biofuels, 2013, 6, 33.
25. H. S. Kim, Microscopic evaluation of the cell wall in sugar maple chips after hot water treatment and fungal decay, ESF, State Univ. of NY, 2012.
26. J. Ma, Z. Ji, J. C. Chen, X. Zhou, Y. S. Kim and F. Xu, Planta, 2015, 242, 327–337.
27. J. Ma, X. Zhang, X. Zhou and F. Xu, BioEnergy Res., 2014, 7, 1358–1368.
28. U. P. Agarwal, Planta, 2006, 224, 1141–1153.
29. P. Prislan, G. Koch, K. Čufar, J. Gričar and U. Schmitt, Holzforschung, 2009, 63, 482–490.
30. C. S. Goh, H. T. Tan and K. T. Lee, Bioresour. Technol., 2012, 110, 662–669.
31. Z. Ji, J. Ma, Z.-H. Zhang, F. Xu and R.-C. Sun, Ind. Crops Prod., 2013, 47, 212–217.
32. Z. Ji, X. Zhang, Z. Ling, X. Zhou, S. Ramaswamy and F. Xu, Biotechnol. Biofuels, 2015, 8, 103.
33. Z.-M. Zhang, S. Chen and Y.-Z. Liang, Analyst, 2010, 135, 1138–1146.
34. X. Zhang, Z. Ji, X. Zhou, J. Ma, Y. H. Hu and F. Xu, Anal. Chem., 2015, 87, 1344–1350.
35. E. C. Yeung, Tested studies for laboratory teaching, 1998, vol. 19, pp. 125–141.
36. L. Donaldson and A. Frankland, Holzforschung, 2004, 58, 219–225.
37. B. S. Donohoe, S. R. Decker, M. P. Tucker, M. E. Himmel and T. B. Vinzant, Biotechnol. Bioeng., 2008, 101, 913–925.
38. U. P. Agarwal and S. A. Ralph, Appl. Spectrosc., 1997, 51, 1648–1655.
39. H. G. Edwards, D. W. Farwell and D. Webster, Spectrochim. Acta, Part A, 1997, 53, 2383–2392.
40. U. P. Agarwal, Adv. Lignocellul. Charact., 1999, 201–225.
41. U. P. Agarwal, J. D. McSweeny and S. A. Ralph, J. Wood Chem. Technol., 2011, 31, 324–344.
42. U. P. Agarwal and S. A. Ralph, Proc. XIVth Int. Symp. Wood, Fiber, Pulping Chem., Durban, South Africa, June, 2007, pp. 25–28.
43. U. P. Agarwal, Fiber Pulping Chem., 2009, p. 53.
44. N. Gierlinger, T. Keplinger and M. Harrington, Nat. Protoc., 2012, 7, 1694–1708.
45. P. N. Perera, M. Schmidt, P. J. Schuck and P. D. Adams, Anal. Chim. Acta, 2011, 702, 172–177.
46. D. Fengel and G. Wegener, Wood: chemistry, ultrastructure, reactions, Walter de Gruyter, 1983.

---

Footnote

† Electronic supplementary information (ESI) available: Additional data and graphics. See DOI: 10.1039/c6ra18333f

---