Visualization and Semiquantitative Study of the Distribution of Major

May 18, 2018 - This paper investigates changes in biochemical components of tissue during stages of elongation, booting, heading, flowering, grain-fil...
1 downloads 0 Views 18MB Size
Subscriber access provided by Kaohsiung Medical University

Article

Visualization and Semi-Quantity Study of the Distribution of Ma-jor Components in Wheat Straw in Mesoscopic Scale using FTIR Microspectroscopic Imaging Zengling Yang, Jiaqi Mei, Zhiqiang Liu, Guangqun Huang, Guan Huang, and Lujia Han Anal. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.analchem.8b00614 • Publication Date (Web): 18 May 2018 Downloaded from http://pubs.acs.org on May 18, 2018

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 10 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Analytical Chemistry

Visualization and Semi-Quantity Study of the Distribution of Major Components in Wheat Straw in Mesoscopic Scale using FTIR Microspectroscopic Imaging Zengling Yangab, Jiaqi Meia, Zhiqiang Liua, Guangqun Huanga, Guan Huanga and Lujia Hana a. College of Engineering, China Agricultural University, Beijing 100083, P. R. China b. Key laboratory of clean production and utilization of renewable energy, the Ministry of Agriculture, Beijing 100083, P.R.China. ABSTRACT: Understanding the biochemical heterogeneity of plant tissue linked to crop straw anatomy is attractive to plant researchers and researchers in the field of biomass-refinery. This study provides an in-situ analysis and semi-quantitative, visualization distributions of major composition distribution in internodal transverse sections of wheat straw based on FTIR microspectroscopic imaging, with a fast Non-Negativity-Constrained Least Squares (fast NNLS) fitting. The paper investigates changes in biochemical components of tissue during stages of elongation, booting, heading, flowering, grain-filling, milk-ripening, dough, and full-ripening. Visualization analysis was carried out with reference spectra for five components (microcrystalline cellulose, xylan, lignin, pectin and starch) of wheat straw. Our result showed that: a) the cellulose and lignin distribution are consistent with those from tissue-dyeing with safranin O-fast green; and b) the distribution of cellulose, lignin and starch are consistent with chemical images for characteristic wavelength at 1432 cm-1, 1507 cm-1 and 987cm-1, showing no interference from the other component analyzed. With the validation from biochemical images using characteristic wavelength and tissue-dyeing techniques, further semi-quantitative analysis in local tissues based on fast NNLS was carried out, and the result showed that: a) the contents of cellulose in various tissues are very different, with most in parenchyma tissue and least in the epidermis; and b) during plant development, the fluctuation of each component in tissues follows nearly the same trend, especially within vascular bundles and parenchyma tissue. Thus, FTIR microspectroscopic imaging combined with suitable chemometric methods can be successfully applied to study chemical distributions within the internodes transverse sections of wheat straw, providing semi-quantitative chemical information. One section for IR analysis

Spotlight 400 Imaging System

Spectrum 400 光谱仪

FTIR microspectroscopic image 

小麦茎秆切面全图

Sliced around the  middle of the second  above‐ground node 

5条纯光谱

红外总吸光 度图像

Starch    Pectin    Lignin    Hemicellulose    Cellulose 

One section for microscope observation  Wheat straw

Leica显微镜

Leica DM2500 Fast‐NNLS fitting

As Validation

Chemical imaging from fast‐NNLS Cellulose   Hemicellulose     Lignin           Pectin            Starch      

特征峰成像图

Chemical imaging at characteristic wavelength  1432 cm‐1, 1507 cm‐1, 987 cm‐1

Semi‐quantitative analysis based on fast NNLS fitting in selected histological structure

ACS Paragon Plus Environment

染色 图

Image of section dyed  with safranin O‐fast green

Analytical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Crop straw is a vast agricultural by-products and an important source of lignocellulosic biomass, which is now being developed as a renewable energy resource to address a serious energy shortage and environmental issues related to other energy sources. It has multi-scalar structure, spanning macroscopic (10-2 m) to microscopic (≤10-6 m) scales. Distribution of various biochemical components in straw affect the use of straw as a renewable energy resource. To use straw with high efficiency, its structure and composition must be properly and fully understood at various scales. “Seeing” the distribution of chemical components within various tissues and cells at a cellular level provides a context to of straw fiber separation and effective conversion processes. Macroscale study of its characteristics, using traditional chemical methods, can provide basic data on the different parts of straw, which promotes the development of straw-based technologies. But it cannot acquire accurate or in-depth information on its microscopic complexity. In contrast, microscale research of its material properties, using transmission electron microscopy, confocal fluorescence microscope, and two-dimensional nuclear magnetic resonance, enhances our understanding of this material at a cellular level. Such methods can be used to analyze the hierarchical structure, microfibrillar architecture of cell walls, information on the chemical structure of basic units and the linking of these structure units1-3. These methods provide highly accurate data and theoretical support for deep exploitation and use of straw. In the case of crops straw, its complex structure, source dispersion and low output per unit have limited microscale analysis to selected tissue, with low analytical efficiency. Mesoscopic analysis, covering analysis between microscopic and macroscopic scales, has the capacity to successfully combine analytical accuracy and efficiency to enhance our knowledge of crop straw characteristics at tissue level. At the mesoscopic scale, the biophysical structure of crop straw can be observed using optical microscope; while qualitative chemical information can be obtained by histochemical staining and observation with fluorescence or polarization microscopes. However, analyses of components are limited by staining and fluorescent agents. To facilitate analysis of the distributions of major components within tissue of crop straw, Fourier transform infrared (FTIR) microspectroscopic imaging as a method having high precision, simple operation and a high degree of automation is explored here. As a mature technique, FTIR spectroscopy links infrared absorption spectra at different wavelengths with specific functional groups and is suitable for chemical structure analysis4. It can provide identification of polymers and functional groups, as well as organization and structure information; it has been widely applied to analysis of cell-wall components and the connection between different functional groups5-7. FTIR microspectroscopic imaging combines microscopy imaging with infrared spectral information, providing visible imaging of objects as well as infrared spectral information in a spatial context. This leads to visualization of chemical components, with high precision and high sensitivity, which is highly suited to analyzing spatial distributions of complex compounds. FTIR microspectroscopic imaging has been increasingly used to study compositional distributions within complex spatial structures in many fields8,9. These include lipid and protein distributions in

Page 2 of 10

monkey brain tissue10, characteristic peak imaging of phosphate, amide and carbonate distributions on bone surface11, aromatic lignin, structural and non-structural carbohydrates distributions in pericarp, seed coat, aleurone layer, and endosperm in wheat seeds12, as well as compositional distributions in sunflower and maize roots13. Clearly, FTIR microspectroscopic imaging has potential as a technique for studying tissue feature of crop straw at a mesoscopic scale. Using FTIR microspectroscopic imaging, our research team has focused on investigating the feasibility of in-situ characterization of the complex crop straw transverse section for several years. We have shown that FTIR microspectscopic imaging combined with the characteristic wavelength method has the capability to reveal the relative concentration and distributions of cellulose, hemicellulose, and lignin linked to tissue structures of internodal transverse sections of cotton and corn stalks14. However, the characteristic wavelength method does not always accurately identify functional group locations, when target components having familiar spectra, like pectin and hemicellulose4. Thus, our team have resorted to more elaborate chemometric methods. The fast Non-Negativity-Constrained Least Squares (fast NNLS) algorithm is a multivariate curve resolution algorithm able to handle spectral data from complex samples. The algorithm minimizes the sum of the squares of the error15, reconstitute the matrix of raw data and calculate the concentration of the target components16. It can obtain relevant quantitative information for a sample at the scale of a pixel, using the same calculation. Therefore, pixel information can be compared. This study investigated the feasibility of in-situ characterization of a complex wheat straw transverse section using FTIR microspectroscopic imaging coupled with the fast NNLS algorithm to provide visualization and semi-quantitative distributions of five major components—cellulose, hemicellulose, lignin, pectin and starch. To verify the reliability of the methods, we analyzed changes in tissue structure and chemical composition for different growth stages of wheat, namely, elongation, booting, heading, flowering, grain-filling, milk-ripening, dough, full-ripening periods.

EXPERIMENTAL SECTION Sample collection and preparation Sample of Triticosecale straws (variety: Nongda 211) were collected from the experimental field at the western campus of China Agricultural University at different times to reflect different growth stages from April to June in 2015. Specific growth stages were sampled on the following dates: elongation: 15 April; booting: 22 April; heading: 30 April; flowering: 6 May; grain-filling: 19 May; milk-ripening:24 May; dough: 5 June; full-ripening:12 June. Samples were analyzed around the middle of the second above-ground node. A paraffin embedding method was used to prepare for sectioning17 (see Supporting Information, Fig. S1). A microtome (ESM-150S, ERMA Inc., Tokyo, Japan) was used to produce two 15μm-thick transverse sections for each sample, which were transferred onto ZnS windows for FTIR microspectroscopic imaging and microslides for microscope observation using tissue-dyeing, respectively.

2 ACS Paragon Plus Environment

Page 3 of 10

FTIR microspectroscopic image acquisition Deparaffined17, freeze-dried sections supported on ZnS windows were placed onto the stage of FTIR microspectroscopic imaging system (Spotlight 400, PerkinElmer Ltd., Beaconsfield, Bucks, UK) to obtain the FTIR microspectroscopic images with a liquid nitrogen-cooled mercury-cadmium-telluride (MCT) line array (16×1 element) detector. Visible images were obtained using a charge-coupled device (CCD) camera. All FTIR images were taken using Spectrum IMAGE R1.7.1 Software (PerkinElmer), collected in transmission mode in the region of 4000-750 cm-1 at 4 cm-1 spectral resolution and 6.25 μm × 6.25 μm spatial resolution. Eight scans were co-added for each pixel, with all morphological features was selected.

|←—————— 1362μm——————

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Analytical Chemistry

|←—— 768μm——→|

Microscope observation of Tissue-dyeing effect with safranin O-fast green After being deparaffined17 and dyed with safranin O-fast green11, sections on microslides were put onto a microsystems stage (Leica DM2500, Leica Microsystems CMS GmbH, Germany) to be examined using a CCD camera. All images were taken using LAS V3.8 Software (Leica Microsystems) and a 50x objective lens to observe morphological features. FTIR spectra acquisition of major components of wheat straw Five major components of wheat straw were prepared for FTIR spectra acquisition. To be specific, cellulose, hemicellulose, and starch were represented with microcrystalline cellulose, xylan and starch extracted from wheat kernels purchased from Sigma-Aldrich (http://www.sigmaaldrich.com/); while lignin and pectin were extracted using methods in Ref.18 Ref.19. KBr pellets mixed with powdered reference materials were prepared for the stage of the FTIR spectrometer (Spectrum 400, PerkinElmer). FTIR spectra were obtained using Spectrum 10.4.2 Software (PerkinElmer), collected in transmission mode in the region of 4000 – 750 cm-1 at 4 cm-1 spectral resolution. In this case, 32 scans were co-added for each pellet to carry out our characteristic wavelength analysis and to define pure component spectra for fast NNLS fitting.

DATA PROCESSING SECTION Data pre-processing for FTIR microspectroscopic images FTIR microspectroscopic images can be defined on m x n lattices with p spectral bands, as shown in Fig. 2. They require filtering for noise, background effects and baseline offset.

Noise reduction. After spatial unfolding, the transmission spectra obtained by FTIR microspectroscopic imaging were converted to absorbance spectra by the formulation 𝑨 𝒍𝒐𝒈 𝟏⁄𝑻 . To improve the signal to noise ratio, principal component analysis (PCA) noise-reduction was carried out with Spectrum IMAGE Software, decomposing spectrum into principal components and reconstructing them using the first 20 principal components Background removing. During spectra acquisition, non-sample area (background) occur in the center and on edges of the image (Fig. 1), which may affect the subsequent characterization of the sample image (foreground). To ensure the non-sample area does not affect sample analysis, it must be set to zero and removed from follow-up calculations. Considering that the sample was a wheat straw internodal transverse section, with obvious infrared spectral characteristics, while the non-sample area was mainly ZnS having no absorption of infrared, a marked difference exits between the average absorbance of these areas. Therefore, a local threshold method20 was used to abstract the foreground and to set the background to zero. Correction of baseline and offset. Before loading data for correction and calculation, the spectral region of the main characteristics bands between 1800 - 800 cm-1 was abstracted. Random and unavoidable interference factors cause spectra to have baseline drifts, while variation in sample thickness and density leads to differences in absorbance. Corrections carried out using PLS_Toolbox 8.0 (Eigenvector Research, Inc., Manson, WA, USA) included Savizky-Golay smoothing within five points, Standard Normal Variate Transformation (SNV), and Automatic Whittaker Filter Baseline.

3 ACS Paragon Plus Environment

Analytical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Data pre-processing for FTIR spectra For characteristic wavelength analysis and fast NNLS fitting, Savizky-Golay smoothing within five points, SNV and Automatic Whittaker Filter Baseline of characteristic bands between 1800 – 800 cm-1 were carried out, before data were abstracted for loading. Fast Non-Negativity-Constrained Least Squares fitting and calculation of target components concentration in pixels The linear relationship between absorbance and concentration is used in NNLS algorithm, based on a multi-component Beer-Lambert’s Law, given by: 𝑫

𝑻 𝒌 𝑺𝒌 𝐩 𝟎, 𝑺𝒌,𝒑

𝒁𝒎 𝒏 𝒎 𝒏 𝒑 While  𝑪𝒎 𝒏,𝒌

𝑬 𝟎 

𝒑 

𝒎 𝒏

 

𝑷

𝒂𝒎,𝐤 𝒃𝒏,𝐤 𝒄𝒌,𝐩 𝐩 𝟏

While  𝒂𝒎,𝒌

𝟎, 𝒃𝒏,𝒌

𝒆𝒎𝒏𝒑   𝟎, 𝒄𝒌,𝐩

𝟎 

(2) 

where 𝒅𝒎𝒏𝐩 is the element of 𝑫 in the mth row, nth column, and pth tube. The element of the mth row and kth column of loading matrix 𝐀 is called 𝒂𝒎,𝐤 . The loading matrices of the second and third modes are defined likewise giving 𝒃𝒏,𝐤 and 𝒄𝒌,𝐩 , while 𝒆𝒎𝒏𝒑 is the residual part of the mnpth element of 𝑫. The number of components is equal to k. The loading matrices 𝐀, 𝐁 and 𝐂 are estimated using a least-squares sense, based on iteratively solving conditional least squares problems21. The fast NNLS fitting was carried out with Matlab R2015a (MathWork Inc, Natick, USA) and PLS_Toolbox 8.0 (Eigenvector Research, Inc.). The target component concentration in one pixel was calculated using the following formula, based on the hypothesis that total content of tested components in each pixel is equivalent22,23: 𝒁𝒂 %

𝒁𝒂 ∑𝑲 𝒁

 

 

 

(3) 

𝒁𝒂 is the value of the target component in one pixel calculated using Eq. ⑴ and ⑵, 𝒁𝒂 % is the content of target component a in that pixel, while ∑𝑲 𝒁 is the sum of content of all k components. was caliFor comparison between growth stages, 𝑍 % brated according to the sum of content of all k components based on laboratory analysis (see Supporting Information, analysis result are given in Table S1). 𝑾% different during various growth stages: 𝒁𝒂 %

𝒄𝒂𝒍

𝒁𝒂 %

𝑾% 

 

 

Similarity analysis of chemical images Setting chemical images to analyze as 𝑨𝒎 𝒏 and 𝑩𝒎 𝒏 , the similarity of the gradient histogram was calculated using the following procedure with Matlab R2015a: a) calculate the matrix of the horizontal gradience 𝑨𝒉 and 𝑩𝒉 , the matrix of vertical gradience 𝑨𝒗 and 𝑩𝒗 ; b)

calculate the gradient histograms 𝑯𝑨𝒉 , 𝑯𝑨𝑽 , 𝑯𝑩𝒉 , 𝑯𝑩𝒗 ;

c)

calculate the similarity between gradient histograms based on the Bhattacharyya coefficient24, using: 𝑺𝒉

𝟏

𝟏

𝑺𝒗

𝟏

𝟏

(1) 

where a FTIR microspectroscopic image is defined on m x n lattices with p spectral bands (Fig. 2). For k pure component spectra, 𝒁 is a (mxn)×k matrix of mixing coefficients (concentrations), 𝑺𝑻 is a k×p matrix of k pure component spectra and 𝑫 is an (mxn)×p spectrum matrix of spatially unfolded FTIR image. 𝑬 is the residual matrix. The fast NNLS algorithm, based on a three-way PARAFAC model21, is an efficient algorithm built using the NNLS model, based on the formula: 𝒅𝒎𝒏𝐩

Page 4 of 10

∑ 𝑯𝑨𝒉 .∗𝑯𝑩𝒉

 

horizontal case

(5) 

 

vertical case

(6) 

∑ 𝑯𝑨 ∗∑ 𝑯𝑩 𝒉 𝒉 ∑ 𝑯𝑨𝒗 .∗𝑯𝑩𝒗 ∑ 𝑯𝑨𝒗 ∗∑ 𝑯𝑩𝒗

Calculation of target component concentrations in selected tissues The target component content in selected tissue (represented as S) was calculated using: 𝒁𝒂

𝒔

∑𝒔 𝒁𝒂 % 𝑵𝒔

𝒄𝒂𝒍

 

 

 

(7) 

where 𝒁𝒂 𝒔 is the content of target component a in selection S, is the 𝑵𝒔 is the number of pixels in the selection, ∑𝒔 𝒁𝒂 % sum content of target component a for pixels in the selection

RESULTS AND DISCUSSION Visualization analysis of distributions of the major components of wheat straw Chemical imaging from fast NNLS fitting. Following fast NNLS fitting and calculation based on Eq. (3) and (4) for the five major components, the resulting chemical images for a single pixel are presented in Fig.3. Panels from top to bottom represent growth stages, while images arranged from left to right are the visible image of the wheat straw section, the false color intensity image, and concentration images for cellulose, hemicellulose, lignin, pectin and starch. Color reflects the relative concentration of each target component, where red is high concentration and blue is low concentration. In this arrangement, comparisons can be made between various components horizontally as well as between different growth stages vertically. In general, cellulose mainly occurs in parenchyma tissue, with less in the vascular bundles, sclerenchyma tissue and the epidermis. Lignin is mainly distributed in the epidermis, sclerenchyma tissue, and vascular bundles, with small amounts in parenchyma tissue; during later stages of growth, lignin increases in all tissues to provide structural support to resist external environment stress. There are clearly contrasting distribution patterns of lignin and cellulose, where the core of vascular bundles and the epidermis have higher levels of lignification and lower levels of cellulose.

(4) 

After the above calculation, a maximum-minimum normalization (adjusted to [0,1]) was carried out on chemical images, with maximum and minimum values obtained from all images.

4 ACS Paragon Plus Environment

Page 5 of 10 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Analytical Chemistry

5 ACS Paragon Plus Environment

Analytical Chemistry Hemicellulose has a similar distribution to cellulose, but has lower overall levels. Tissues contain relatively low amounts of pectin, although it is higher in sclerenchyma tissue and the epidermis. Starch content is relatively uniform during growth, especially in sclerenchyma tissue and vascular bundles, with slight reductions and increases as it concentrates in the vascular bundles. Chemical imaging of tissue-dyed sections. Sections at different growth stages stained with safranin Ofast green are shown in Fig. 4; images from top to bottom and from left to right show growth stages in order. In these images, red represents for lignin stained by safranin O and green is cellulose stained by fast green. Clearly, sclerenchyma and parenchyma tissues are almost all green, indicating that cellulose is the major component; while the inside of vascular bundles are almost red, indicating they mainly contains lignin. The contrasting distribution patterns of lignin and cellulose also were observed in our fast NNLS analysis (Fig. 3). Staining showed corresponding results to the fast NNLS fitting, except in heading and flowering stages. In these stages, the fast NNLS contents of cellulose and lignin in sclerenchyma tissue and the epidermis were greater than any other stages. This may be an artifact caused by sampling. Thus, additional studies should be conducted to clarify their distributions at these stages.

Flowering

|———|

|———|

|———|

Dough

100μm

100μm

100μm

|———|

100μm

Milk-ripening

|———|

Heating

100μm

|———|

Booting

Grain-filling

100μm

|———|

Elongation

100μm

|———|

100μm

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Full-ripening

Figure.4 Microscopic images of stained internodal transverse sections of wheat straw at different growth stages. Red represents lignin stained by safranin O and green is cellulose stained by fast green

Chemical imaging at characteristic wavelength. FTIR spectra of pure samples of the five major components after pre-processing shows characteristics bands in range from 1800—800 cm-1 (Fig. 5), as listed in Supporting Information, Table S225-32. Overlaps are obvious. Therefore, only the bands at 1507cm-1 (aromatic ring frame stretching vibration), 1432 cm-1 (C-H antisymmetric deformation), and 987 cm-1 (C-O stretching vibration) are diagnostic bands for lignin, cellulose and starch, having little interference. In our former research14, the band at 1240cm-1 (corresponding to a C-O-C vibration33) was selected as the characteristic wavelengths for cellulose. However, a band at 1234 cm-1 in

Page 6 of 10

FTIR spectrum of lignin is very close to this band. To avoid interference from lignin, we selected the band at 1432 cm-1, corresponding to C-H antisymmetric deformation in this study. For lignin, the band at 1507 cm-1 was selected, corresponding to C=C stretching vibration of the aromatic rings in lignin26. This is nearly the same band (1504 cm-1) in our former research14.

Starch      Pectin      Lignin      Hemicellulose      Cellulose 

There is also a slight absorbance at 1507 cm-1 and 1432 cm-1 in the FTIR spectrum of pectin. However, according to our laboratory analysis (see Supporting Information, Table S1)., cellulose (about 40%), and lignin (about 20%), contribute about 60% of the dry weight of crop straw, while pectin contributes less than 2%. In this case, we assumed there was little impact on the analysis of lignin and cellulose using these bands. Similarly, the slight absorbance of hemicellulose at 1432 cm-1 also would have little impact on the analysis of cellulose. For starch, the band at 987 cm-1 was selected, corresponding to a C-O stretching vibration in starch. In FTIR spectra, cellulose and hemicellulose also contribute at 987 cm-1, but the band in starch is much stronger. Given that starch contributes about 15% of the dry weight of wheat straw, we surmised that this wavelength would be characteristic for starch. In the case of hemicellulose and pectin, the spectra of these two components are too alike to identify a characteristic wavelength for imaging having little interference from other components. The result of characteristic wavelength imaging was shown in Fig. 6 (the maximum-minimum normalization, adjusted to [0,1], was carried out on chemical images with maximum and minimum values obtained from images at the same wavelength), in which panels from top to bottom represent growth stages, while images from left to right are the visible image of wheat straw section, the false color intensity image, and spectral images at wavelengths of 1432 cm-1, 1507 cm-1 and 987 cm-1.

6 ACS Paragon Plus Environment

Page 7 of 10 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Analytical Chemistry

In spectral images, color reflects relative concentration, revealing the distributions of various components. In these images, red shows high concentration areas while blue depicts low concentration areas. Because there is no correspondence between intensities of different bands, comparison can only be made vertically, between different growth stages. In general, cellulose occurs within sclerenchyma and parenchyma tissues, with least in the inside of vascular bundles. Lignin mainly occurs in vascular bundles, with most in the core during heading and flowering stages, in contrast to cellulose. Lignin in parenchyma tissue increases during growth, reinforcing the straw’s mechanical properties. The contrasting distribution patterns of lignin and cellulose are best illustrated in Fig. 6, in which the core of the vascular bundles and epidermis have higher levels of lignification, and low levels of cellulose. Starch is relatively uniformly distributed in tissues, especially in sclerenchyma tissue and the outside of vascular bundles. During growth, starch reduces and but later increases, peaking around the flowering stage. Comparison analysis between chemical imaging methods. All three methods had similar results, with contrasting distribution patterns between cellulose and lignin: the core of the vascular bundles and epidermis have higher levels of lignification, and low levels of cellulose. To compare the results based on chemical images derived from characteristic wavelength versus chemical images from fast NNLS fitting, a similarity analysis of their gradient histograms was conducted in both horizontal and vertical directions, using Eq. (5) and (6). The result is list in Table 1. Table 1 Similarity between gradient histograms Growth stage elongation booting heading flowering grain-filling milk-ripening dough full-ripening average

Cellulose 𝑆 𝑆

𝑆

Lignin 𝑆

𝑆

Starch 𝑆

0.8261

0.8105

0.9186

0.8997

0.8114

0.8038

0.7840

0.7714

0.8874

0.8946

0.8208

0.8228

0.8518

0.8431

0.8655

0.8896

0.8221

0.8303

0.8165

0.8051

0.8574

0.8618

0.8360

0.8373

0.7590

0.7590

0.8487

0.8565

0.7849

0.8108

0.8247

0.7966

0.8782

0.8438

0.8381

0.8288

0.7497

0.7497

0.8135

0.8258

0.7400

0.7510

0.8475

0.8403

0.8778

0.8753

0.8266

0.8392

0.8261

0.8105

0.9186

0.8997

0.8114

0.8038

The images of lignin during various growth stages, almost having similarity values superior to 0.85, are classed as ‘highly similar’; Likewise, images of cellulose and starch, almost having similarity values between 0.7 and 0.85, are classed as ‘moderate similar’. Mostly, we can confirm that the chemical images from the characteristic wavelength for lignin and those from fast NNLS fitting are consistent. However, the similarity between images of cellulose and starch from both methods are slightly poor, may due to small interferential between bands. In contrast to characteristic wavelength imaging, the fast NNLS fitting makes it possible to analyze components with similar spectra, such as hemicellulose and pectin. Also, due to the same calculative process and the calibration with lab analysis, semi-quantitative analysis could be done from the perspective of NNLS content.

Semi-quantitative analysis based on fast NNLS fitting in selected histological structures To further clarify changes in the distribution of different components during growth of wheat straw, concentration of components in various tissues, namely the epidermis, sclerenchyma, vascular bundles and parenchyma. These are based on analyses of three profiles for each tissue type, as detailed in the Supporting Information, Fig. S2. The content was determined by fast NNLS fitting across growth stages for a selected tissue using Eq. (7). Our results are presented in Fig. 7, as well as in Supporting Information, Table S3. During the growth of wheat, the content of cellulose shows marked differences among various tissues types, with notable fluctuations between stages. It is concentrated in parenchyma and vascular bundles, with lesser in the epidermis and sclerenchyma tissue. In the case of lignin, its content increases, decreases and increases from elongation to full-ripening stages, with turning points at heading and milk-ripening. Although the content of lignin from fast NNLS fitting does not exactly correspond to our laboratory analysis (see Supporting Information, Table S1), they do share the same trend across growth stages34. Meanwhile, cellulose and other components show similar variation and trends with growth stages in vascular bundles and in parenchyma tissue. Fluctuations in images appear to be minimal or random, yielding semi-quantitative results containing valuable chemical information. In particular, we show that a) the content of hemicellulose is almost below 20% in most tissues and shows some fluctuations during growth; b) the content of pectin is lower than all other components, with highest levels during the heading stage; c) the content of starch shows decreases and then levels off during growth; and d) the content of cellulose presented a decreasing-increasing-decreasing cycle in the epidermis and sclerenchyma tissue, but decreased and then increased slightly in vascular bundles. However, Fig. 7 does show some unreasonable levels in the epidermis and sclerenchyma tissue, especially in the panels for cellulose and pectin (Fig. 7 (a) and (b)). This may reflect artifacts developed during sample collection or preparation. It requires further study to determine their cause.

CONCLUSIONS This article presents a novel study of the FTIR microspectroscopic biochemical imaging of internodal transverse sections of wheat straw, using a fast NNLS algorithm. This method is widely used for spectroscopic analysis of heterogeneous materials. Our results show that the FTIR microspectroscopic chemical images of cellulose and lignin based on fast NNLS analysis match information obtained from serial dyed sections to reveal morphology, and chemical imaging using characteristic wavelength. Though, the employ of fast NNLS routine does not alter the image analysis for the three major components of wheat (lignin, cellulose and starch), it does permit synchronously detection of lesser components, such as hemicellulose and pectin.

7 ACS Paragon Plus Environment

Analytical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 8 of 10

8 ACS Paragon Plus Environment

Page 9 of 10 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

 

Analytical Chemistry Cellulose

80

 

Hemicellulose

80

epidermis vascular bundles sclerenchyma parenchyma

70

70

60

60

60

50

50

50

40

40

40

30

30

30

20

20

10

10

20

epidermis vascular bundles sclerenchyma parenchyma

10

0

--

elongation

booting

heading

flowering

grain-filling milk-ripening

dough

full-ripening

--

0

elongation

booting

epidermis vascular bundles sclerenchyma parenchyma

70

0 --

Lignin

80

heading

Growth stage

flowering

grain-filling milk-ripening

dough

full-ripening

--

--

elongation

booting

Growth stage

heading

flowering

grain-filling milk-ripening

dough

full-ripening

Growth stage

 

Via the application of the fast NNLS results, additional semiquantitative analysis has been performed, providing information on major components distributions in various morphological tissues during growth. Although the contents from fast NNLS fitting are not consistent with laboratory analysis, they do share the same trend across growth stages. Given that fluctuations are minimal or random, these semi-quantitative results do provide valuable chemical information. However, some anomalous contents were recorded, which may be related to sample collection or destruction during preparation, or other reasons, requiring ongoing study to clarify.

*E-mail: [email protected]

ACKNOWLEDGMENT This work was supported by Natural Science Foundation of China (Nos. 31471407), the Innovation Team Project of the Ministry of Education (IRT_17R105), the China Agriculture Research System (CARS-36). We thank Dr. Trudi Semeniuk for editing the English text of a draft of this manuscript.

CONFLICTS OF INTEREST

ASSOCIATED CONTENT Supporting Information Supporting Information is available free of charge on the ACS Publications website. Table of Contents: 1. Experimental details: Paraffin embedding and deparaffinization methods, Fig. S1 2. Table of laboratory analyses, Table S1 3. Table of characteristics bands in the range 1800–800 cm⁻1, supplementary Fig. 5, Table S2 4. Details of local tissue selections, corresponding to Fig. 7, Fig. S2 5. Table showing Fast NNLS content in various tissues of the five major components, corresponding to Fig 7, Table S3 All above contents are included in a single PDF file.

AUTHOR INFORMATION

Corresponding Author

The authors declare no conflicts of interest.

REFERENCES (1) Chundawat S P S, Donohoe B S, da Costa Sousa L, et al. Energy & Environmental Science, 2011, 4(3): 973-984. (2) Ding S Y, Liu Y S, Zeng Y, et al. Science, 2012, 338(6110): 10551060. (3) Brunecky R, Alahuhta M, Xu Q, et al. Science, 2013, 342(6165): 1513-1516. (4) Ewelina S, Katarzyna M, Zeineb F, et al. Spectrochim Acta A Mol Biomol Spectrosc, 2016, 169:152-160 (5) Li Q, Tang Y, Yan Z, et al. Spectrochim Acta A Mol Biomol Spectrosc, 2017, 180:154~160 (6) Fang K, Gao S, Zhang W, et al. PLoS ONE, 2016,11(2): 1-14 (7) Corrales C F, Díaz S B, Ben A A, et al. Chemistry & Physics of Lipids, 2018, 210: 60-69 (8) S. L. Sukanya, J. Sudisha, H. S. Prakash, et al., Journal of Phytology, 2011,3(10): 26-32 (9) Ricci C, Bleay S, Kazarian S G. Analytical Chemistry. 2007,79(15):5771~5776

9 ACS Paragon Plus Environment

--

Analytical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 10 of 10

(10) Lewis E N, Gorbach A M, Marcott C, et al. Applied Spectroscopy, 1996, 50(2):263~269 (11) Kobayashi M, Hara K, Akiyama Y. JOURNAL OF BONE AND MINERAL METABOLISM. 2007, 25(1):12~18 (12) Peiqiang Y, H. Block, Z. Niu, et al, Journal of Synchrotron Radiation, 2007,14(4): 382-390 (13) Dokken K M, Davis L C. Journal of Agricultural and Food Chemistry. 2007, 55(26):10517~10530 (14) Cao C, Yang Z, Han L, et al. CELLULOSE. 2015,22(1):139~149 (15) Howard M, Jerome W. Spectroscopy. 2010, 25(5):16~21 (16) Gendrin C, Roggo Y, Spiegel C, et al. European Journal of Pharmaceutics and Biopharmaceutics. 2008,68(3):828~837 (17) Rossi S, Anfodillo T, Menardi R. Trephor: IAWA JOURNAL. 2006, 27(1):89~97 (18) Sun X, Sun R, Fowler P, et al. Agricultural and Food Chemistry. 2005(53):860~870 (19) Oliveira T Í S, Rosa M F, Cavalcante F L, et al. Food Chemistry. 2016,198:113~118 (20) Siqi H, Lei W. Systems Engineering and Electronics, 2002,06:9194+102 (21) Rasmus B Sijmen D Jong. Journal of Chemometrics, 1997,11:393~407 (22) Carsten R. [PhD thesis] University of Copenhagen, Denmark, 2009 (23) E. Neil Lewis, Linda H K. John Wiley & Son,Inc:Raman, Infrared, and Near-Infrared Chemical Imaging.2010 (24) D. Comaniciu, V. Ramesh, P. Computer Vision and Pattern Recognition. 2000, 2:142–149 (25) Abdel-Halim E S. Arabian Journal of Chemistry, 2014, 7(3):362~371 (26) C S J C, George N, Narayanankutty S K. Carbohydrate polymers, 2016,142:158~166 (27) N. A. Mamleeva, S. A. Autlov, A. N. Kharlanov, Russian Journal of Physical Chemistry A, 2015,89(8):1458~1463 (28) Faix O. Holzforschung. 1991,45S:21~27 (29) Xu F, Sun R, Zhai M, et al. JOURNAL OF APPLIED POLYMER SCIENCE. 2008, 108(2):1158~1168 (30) Liu J, Mukherjee J, Hawkes J J, et al. Journal of Chemical Technology & Biotechnology, 2013,88(10):1807~1814 (31) Dean A P, Sigee D C, Estrada B, et al. Bioresource Technology, 2010,101(12):4499~4507 (32) Oyeyinka S A, Singh S, Adebola P O, et al. Carbohydrate Polymers, 2015,133:171~178 (33) Na L, Peiqiang Y, Journal of Agricultural and Food Chemistry, 2010, 58(13): 7801-7810 (34) Howard M, Jerome W, Spectroscopy. 2012,27(10):12~17

10 ACS Paragon Plus Environment