Imaging method of tobacco cell wall structure components
By using confocal Raman microscopy and the MCR-ALS model, the problems of fluorescence interference and low signal-to-noise ratio in Raman imaging technology were solved, enabling high-precision imaging and analysis of plant cell wall structural components and revealing the microscopic changes of cellulose and pectin during plant growth and senescence.
Patent Information
- Application Number
- CN202410885805.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-03
- Publication Date
- 2024-11-22
AI Technical Summary
Existing Raman imaging techniques face challenges in analyzing the structural components of plant cell walls, including strong fluorescence interference and low signal-to-noise ratio, making it impossible to accurately analyze the microscopic changes in polysaccharide components such as cellulose and pectin.
By employing confocal Raman microscopy combined with a multivariate curve-resolved alternating least squares (MCR-ALS) model, Raman imaging data is decomposed through preprocessing and the MCR-ALS model to remove interference signals, enabling high-precision quantitative and qualitative analysis of cellulose and pectin.
It enables high spatial resolution imaging of cellulose and pectin components in complex biological samples, providing accurate analysis of the microscopic distribution and changes of cell wall components during plant growth and senescence, while avoiding potential interference and damage from fluorescent labeling.
Smart Images

Figure CN119006626A_ABST
Abstract
Description
Technical Field
[0001] The invention relates to an imaging method for tobacco leaf cell wall structural components, belonging to the technical field of cell wall structural component analysis. Background Art
[0002] Plant cell walls have many important functions, including controlling cell volume and determining cell shape, conducting water and nutrients, and defending against biotic and abiotic stresses. They are important for plant growth, development, maintenance and reproduction, as well as the economic use of plants. Plant cell walls are an ordered network structure formed by polysaccharides and other polymers connected by covalent and non-covalent bonds. Its structural components mainly include cellulose, pectin, lignin and structural proteins. The cell wall is the least known cell structure in plants. Studying the spatiotemporal changes of the main structural components of leaf cell walls at the microscopic level is of great significance for understanding the growth and aging process of plants. Tobacco was selected as a model plant to study the above issues. It is a herbaceous plant of the genus Nicotiana in the Solanaceae family. In addition to being a raw material for the cigarette industry, its aerial parts are used as insecticides, sedatives, antipyretics, anesthetics and anthelmintics in traditional herbal medicine.
[0003] The structural components of the cell wall vary greatly during cell growth. Generally speaking, cellulose and pectin account for about 25% and 35% of the dry weight of the primary wall, respectively. Cellulose is composed of superimposed cellulose fibrils (linear structures composed of (1,4)-β-D-glucans). Cellulose fibrils have structural characteristics such as high hardness, chemical stability and insolubility due to the large number of non-covalent bonds between adjacent glucans. They are the load-bearing polymers of plant cell walls. Pectin is the most abundant component of most primary cell walls. It is a heterogeneous group of polysaccharides containing acidic sugars (such as galacturonic acid) and neutral sugars (such as rhamnose, galactose and arabinose). These different polysaccharides are usually (but not always) covalently linked to each other to form large molecular structures. When pectin is synthesized in the Golgi apparatus, many acidic residues are esterified with groups such as methyl and acetyl groups. Once pectin is secreted into the cell wall, the ester groups are removed by pectinesterase in the cell wall, so that the charges on the carboxyl groups are exposed, enhancing the ability of pectin to form a rigid gel state. According to the degree of esterification, pectin can be divided into low methoxyl pectin and high methoxyl pectin. Pectin, as a hydrated gel substance, fills the cellulose-hemicellulose network structure, which can prevent the aggregation and collapse of the cellulose network structure and also determines the permeability of the cell wall to macromolecules. During the invasion of plant tissues by fungi, oligosaccharides released from pectin will cause a defense response to limit the invasion of pathogens.
[0004] In addition, the cellulose and pectin content in the cell walls of leaves of some plants (such as tobacco) is also closely related to the physical and chemical properties and economic value of the leaves. Tobacco stems account for 20%-30% of the total mass of tobacco leaves, and their chemical composition is basically the same as that of tobacco leaves. Previous studies have shown that during the growth of tobacco leaves, the pectin content of the cell walls gradually increases; during the aging (baking) of tobacco leaves, the water-soluble pectin content in the cell walls gradually increases, the protopectin content gradually decreases, and the total pectin content shows a gradual downward trend. However, the microscopic spatiotemporal changes in the above-mentioned cell wall structural components during the growth and aging process are not clear enough, and there is an urgent need to develop non-invasive, non-labeling, high-throughput, in situ imaging methods.
[0005] Classical analytical methods for detecting cell wall structural components such as cellulose and pectin include colorimetry, chromatography, and nuclear magnetic resonance, which are mainly used for quantitative analysis and require destructive dissolution. For imaging tasks at the cellular level, common techniques include mass spectrometry imaging, fluorescence imaging, and Raman microscopy. These techniques have their own characteristics and can meet the observation needs of scientific researchers for life phenomena at different levels and scales. Mass spectrometry imaging has high throughput capabilities but the spatial resolution is not high enough. Fluorescence imaging technology has high sensitivity but usually requires fluorescent labeling. Raman microscopy technology has the characteristics of non-labeling, high throughput, strong fingerprint characteristics of chemical substances, spatial resolution up to the micron level, non-destructive and in situ imaging, etc., and can generate molecular fingerprint spectra with high spatial resolution (<0.5μm, up to single cell or even cell wall level). Raman microscopy technology has unique advantages in the imaging research of plant cell wall structural components. It has been used to study the softening mechanism of apples during ripening and post-harvest storage, the microscopic changes in loquat fruit cells after harvest, the visualization of poplar cell wall components, the material changes at the cellular level during tea leaf infection with Fusarium wilt, the spatiotemporal changes of the main components of the cell wall during pear infection with Alternaria alternata, and the spatiotemporal changes of cellulose, low-methoxyl pectin, high-methoxyl pectin, lignin and phenolic substances in the cuticle and cell wall during apple infection with Botrytis cinerea.
[0006] Currently, for Raman imaging data, the integrated intensity filter is traditionally used to analyze imaging data, that is, a characteristic peak of a substance is selected, and the molecular imaging diagram of the substance is drawn based on the signal intensity integral value of the interval. However, if there are signal contributions from other interfering substances in the interval (which often occurs in complex biological systems), the integrated intensity filter method will have a large error and may even lead to incorrect interpretation of potential biochemical molecules.
[0007] Cellulose and pectin in plant cell walls have some characteristic bands. Table 1 shows the Raman bands of plant cell wall polysaccharides. Due to the structural similarity of the monomers that make up the polysaccharides, some Raman bands of different polysaccharides originate from the vibration of the same functional group, which causes these Raman bands to overlap and interfere with each other. For example, 1122 and 1097 cm -1 The bands at 1379 cm-1 are usually assigned to cellulose, due to the asymmetric and symmetric stretching vibrations of the COC in the glycosidic bonds of cellulose, but they also appear (although weaker) in the Raman spectrum of hemicellulose. The bending vibrations of HCC, HC=O, and HO=C in cellulose are 1379 cm-1. -1 The reason for the appearance of the band at 375cm -1 The bands at 836 cm-1 related to the ring deformation mode are characteristic bands of cellulose. Pectins with different methylation levels exist in plant cell walls. The most unique Raman band of pectin appears at 836 cm-1. -1 , originating from the vibration of the α-glycosidic bond in HMP, is very sensitive to o-acetylation; 856 cm -1 The COC skeleton pattern assigned to the α-isomer of LMP. The characteristic bands of both types of pectin appear in the Raman spectrum of tobacco leaf cell walls. Other polysaccharides in plant cell walls generally do not interfere with the above characteristic bands of pectin. The ester carbonyl C=O stretching vibration band (about 1742 cm -1 ) is the characteristic band of HMP, which can be used to distinguish it from LMP in theory. However, hemicellulose will cause spectral interference here, so 1742cm -1 The bands are not suitable for identification of HMP.
[0008] Table 1 Raman band assignments for plant cell wall polysaccharides based on literature
[0009]
[0010] Based on the above, it can be seen that when using Raman imaging data to analyze cell wall structural components, Raman signals are limited by strong fluorescence interference (which is affected by spectral interference in complex biological samples) and low signal-to-noise ratio, which makes it impossible to accurately analyze cell wall structural components. It is necessary to study and improve this. Summary of the invention
[0011] Based on the above, the present invention provides a method for imaging the structural components of tobacco leaf cell walls to overcome the deficiencies of the prior art.
[0012] The technical solution of the present invention is: a method for imaging the structural components of tobacco leaf cell walls, comprising:
[0013] S1 measures frozen section samples using a confocal Raman microscope to obtain raw imaging data cubes TX×Y×J ;
[0014] S2S2 expands the original imaging data cube into a column-wise expanded pixel spectrum matrix M XY×J , preprocessing each Raman imaging data in the column-wise expanded pixel spectrum matrix;
[0015] S3 selects the fingerprint interval of the analyte from the column-wise expanded pixel spectrum matrix to obtain the data matrix and constructs the MCR-ALS model, decomposing it to obtain the column-wise expanded relative intensity profile matrix C XY×N and the normalized spectral matrix S J×N ;
[0016] S4 is based on the standard spectrum of the analyte and its characteristic band attribution analysis, and the relative intensity profile matrix C is expanded column-wise. XY×N and the normalized spectral matrix S J×N The components corresponding to the spectrum in the image are resolved into analytes and interfering components, and the relative intensity profile matrix C is expanded column-wise XY×N The columns corresponding to the analytes in the matrix are refolded into a matrix, and the concentration distribution image of the analytes is drawn using it.
[0017] Preferably, in step S1, the confocal Raman microscopy imager uses a 50× visible microscope objective lens, NA=0.75, uses an air-cooled back-illuminated CCD to detect Raman signals, uses a λ=532nm linear polarized laser to generate incident light, has an output power of 10mW, and the grating is set to 600g mm -1 The exposure time is set to 3s. Before each measurement, the peak value is set at 520.70cm -1 Silicon wafer calibration system.
[0018] Preferably, in step S2, the preprocessing includes:
[0019] S21 removes the spikes caused by cosmic rays in the Raman spectrum;
[0020] S22 removes background through MPLS algorithm;
[0021] S23 performs relevant interval noise reduction on each pixel spectrum through discrete wavelet transform;
[0022] S24 soft-smoothes each spectrum by discrete Fourier transform.
[0023] Preferably, step S23 includes:
[0024] S231 determines the noise intervals of different levels in Raman spectra;
[0025] S232 uses Daubechies 2 wavelet for discrete wavelet transform;
[0026] S233 processes the wavelet detail coefficients of each layer using a hard threshold criterion;
[0027] S234 reconstructs the denoised signal using the approximation coefficients and the modified detail coefficients.
[0028] Preferably, step S24 includes:
[0029] S241 uses a series of Fourier series consisting of sines and cosines to represent the spectrum, and the spectrum in the time domain is transformed into an amplitude spectrum in the frequency domain;
[0030] S242 uses a window function to eliminate the amplitude of the noisy frequency;
[0031] The S243 signal is transformed from the frequency domain back to the time domain through inverse discrete Fourier transform, and a smoothed spectrum is obtained.
[0032] Preferably, in step S3, the construction and decomposition method of the MCR-ALS model is:
[0033] S31 model initialization
[0034] Initialize the model using the singular value decomposition algorithm and set the number of groups;
[0035] S32 Iteration Process
[0036] In each iteration, a non-negativity constraint is applied to the concentration profile;
[0037] Divide the spectral profile by the sum norm;
[0038] The algorithm converges after a set number of iterations;
[0039] S33 Results Matrix
[0040] Decompose the column-wise extended relative intensity profile matrix C XY×N and the normalized spectral matrix S J×N .
[0041] Preferably, the analytes include cellulose, pectin, lignin and structural protein.
[0042] Preferably, the fingerprint range of the cellulose is 200-450 cm -1 The fingerprint range of the pectin is 800-900cm -1 .
[0043] Beneficial effects of the present invention:
[0044] 1. The method based on label-free Raman microscopy combined with the multivariate curve resolution-alternating least squares (MCR-ALS) model significantly improves the quantitative and qualitative analysis capabilities of polysaccharide components such as cellulose and pectin in complex biological samples, providing an efficient, accurate and non-destructive technical means for studying the changes in cell wall components during plant growth, development and aging.
[0045] 2. The present invention uses confocal Raman microscopy technology, which can provide high spatial resolution imaging, even reaching the level of single cells or cell walls, which is essential for observing and analyzing the microscopic distribution and changes of cell wall components. At the same time, Raman microscopy does not require fluorescent labeling, thus avoiding the potential interference and damage of the marker to the sample, retaining the natural state of the sample, and is suitable for the analysis of sensitive biological samples. In addition, Raman spectroscopy provides rich chemical fingerprint information, which can accurately identify and distinguish different chemical components, such as cellulose and different types of pectin (high methoxy pectin and low methoxy pectin).
[0046] 3. The present invention uses the MCR-ALS model to separate the pure spectrum of each component from the complex Raman spectrum data, eliminate interference signals, and improve the accuracy and reliability of the analysis. In particular, through the qualitative analysis of Raman spectra and the quantitative resolution of the MCR-ALS model, the concentration distribution of cellulose and pectin in the cell wall can be accurately determined, providing important data support for studying their dynamic changes during plant growth and aging.
[0047] 4. The present invention establishes an imaging analysis method for cellulose, high methoxyl pectin and low methoxyl pectin in tobacco leaf cell walls based on Raman confocal microscopy imaging technology combined with multivariate curve resolution-alternating least squares algorithm, which can characterize the microscopic spatiotemporal changes of cellulose and pectin in leaf cell walls during the growth and aging (baking) process of model plant tobacco. The proposed imaging method has the advantages of non-labeling, high throughput and high specificity. BRIEF DESCRIPTION OF THE DRAWINGS
[0048] Figure 1 Preprocessing of Raman imaging data. (a) Measured pixel spectrum; (b) Raman spectrum after removing cosmic ray peaks; (c) Raman spectrum after background removal using MPLS; (d) interval of non-uniform noise; (e) discrete wavelet transform (represented by the 808th pixel spectrum); (f) eliminated non-uniform noise; (g) interval-correlated denoised pixel spectrum; (h) discrete Fourier transform; (i) smoothed pixel spectrum. These pixel spectrum matrices are shown with one strip every 10 pixels.
[0049] Figure 2 Illustration of the MCR modeling process and the precise imaging results (taking the R2 stage sample as an example). (a) Original imaging data cube T 40×40×J; (b) Column-wise expanded data matrix M after preprocessing 1600×J ; (c) Cellulose characteristic region (200-450cm -1 ) resolution; (e) and (c1)-(c3) are the resolution spectra and accurate imaging of the cellulose characteristic region; (d) the resolution of the pectin characteristic region (800-900cm -1 ); (f) and (d1)-(d4) are the resolved spectra and accurate imaging of the characteristic regions of pectin, respectively.
[0050] Figure 3 Photos and Raman spectra of tobacco leaves at different growth, maturity and curing stages. (a) Photos of tobacco leaves at the growth stage (G1-G4) and maturity stage (R1-R4); (b) Photos of tobacco leaves at the yellowing stage (Y0-Y3), color fixation stage (CF1-CF3) and dry rib stage (SD1-SD2). (c) Raman spectra of the midrib of tobacco leaves during growth and maturity; (d) Raman spectra of the midrib of tobacco leaves during curing.
[0051] Figure 4 Raman microscopy images of polysaccharides in the cell walls of midrib and lateral veins of leaves during growth based on the MCR model. G1, G2, R1, R2, and R4 were selected as representative nodes, and the red squares on the bright field images marked the studied areas; Raman microscopy images of cellulose, HMP, and LMP in the cell walls of midribs were plotted in (a1)-(a5), (b1)-(b5), and (c1)-(c5); Raman microscopy images of cellulose, HMP, and LMP in the cell walls of lateral veins were plotted in (d1)-(d5), (e1)-(e5), and (f1)-(f5).
[0052] Figure 5 Raman microscopy images of polysaccharides in the cell walls of midrib and lateral veins of leaves during baking based on the MCR model. Y0, Y3, CF3 and SD2 were selected as representative nodes, and the red squares on the bright field image marked the studied areas; Raman microscopy images of cellulose, HMP and LMP in the cell walls of midribs were plotted in (a1)-(a4), (b1)-(b4) and (c1)-(c4); Raman microscopy images of cellulose, HMP and LMP in the cell walls of lateral veins were plotted in (d1)-(d4), (e1)-(e4) and (f1)-(f4).
[0053] Figure 6 Comparison between the clamp peak method and the MCR model (taking samples from the R1-R4 stage as an example). (a) Bright field image; (b) The cellulose obtained by the clamp peak method (based on the characteristic interval 200-450cm -1 ) and pectin (based on the characteristic interval 800-900cm -1 ) imaging diagram; (c) imaging diagram of cellulose and pectin (HMP and LMP are distinguished) given by the MCR model. DETAILED DESCRIPTION
[0054] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific embodiments of the present invention are described in detail below in conjunction with the accompanying drawings. In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention can be implemented in many other ways different from those described herein, and those skilled in the art can make similar improvements without violating the connotation of the present invention, so the present invention is not limited by the specific implementation disclosed below.
[0055] Example 1: Method for imaging cellulose and pectin in tobacco leaf cell walls
[0056] 1. Sample preparation
[0057] Tobacco leaf samples (Yunyan 87 variety) were collected during the growing process, and the midrib and lateral veins of all samples were cryosectioned using a CryoStar NX50 cryostat with a slice thickness of 30 μm.
[0058] 2. Raman microscopy
[0059] The frozen section samples were measured using a confocal Raman microscopy imager to obtain the original imaging data cube. The confocal Raman microscopy imager used a laser confocal Raman microscope system alpha300R (WITec, Germany) equipped with a confocal microscope, using a 50× visible microscope objective lens, NA = 0.75, an air-cooled back-illuminated CCD to detect the Raman signal, a λ = 532nm linear polarized laser to generate the incident light, an output power of 10mW, and a grating set to 600gmm -1 The exposure time is set to 3s. Before each measurement, the peak value is set at 520.70cm -1 Silicon wafer calibration system.
[0060] Both Raman spectroscopy and microscopic imaging data were derived from the corresponding frozen section samples. For the midrib and lateral veins of the leaf, the small imaging area (used to observe a single cell) was 160 μm × 160 μm and 100 μm × 100 μm (the step size was set to 4 μm), and the large imaging area (used to observe multiple cells) was 320 μm × 320 μm (the step size was set to 8 μm), and 1600, 625, 1600, and 1600 pixel spectra were obtained in each measurement area, respectively.
[0061] 3. Data preprocessing
[0062] Raman imaging is a technology with great potential in plant cell system and tissue analysis. Large data sets are usually collected. However, data analysis can be very challenging. There are many unfriendly influences that negatively affect the analysis of Raman spectra. Raman imaging data contains spikes caused by cosmic rays, baselines (backgrounds) mainly due to fluorescence interference, and noise. These factors, especially the baseline, can seriously affect the already weak Raman signal. Cosmic rays (CR) can affect charge-coupled device detectors, resulting in very narrow spikes in Raman spectra. They have the characteristics of sharp peaks, high intensity, random positioning, and uncertain quantity. Raman spectra of original biological tissues usually have a strong fluorescence background. The fluorescence background of biological samples rich in autofluorescent substances is many times stronger than the Raman signal, which can almost cover up the Raman signal visually and produce the adverse effect of false Raman intensity. Noise is ubiquitous in Raman and has a great impact on the signal-to-noise ratio of Raman spectra. Reducing noise is an effective way to improve the signal-to-noise ratio.
[0063] Raw imaging data cube T X×Y×J Expand into column-wise expanded pixel spectrum matrix M XY×J , expand the pixel spectrum matrix M in the column direction XY×J Each Raman imaging data is preprocessed. The specific method is as follows:
[0064] First, the original imaging data cube T X×Y×J Expand into an expanded matrix M XY×J (denoted as imageM), drawn on Figure 1 (a). In the Raman spectrum, very narrow and very strong cosmic rays appeared. The peaks caused by cosmic rays were removed using the algorithm provided by the instrument software. The results are shown in Figure 1 (b). The strong background (baseline) in the Raman spectrum almost masks the Raman signal, so the MPLS algorithm is used to remove the background. The calculation process is as follows: (1) Obtain a preliminary estimate of the background through mathematical morphological opening operation; (2) Use the penalized least squares method to further optimize the background based on the preliminary estimated background and local minimum; (3) Subtract the optimized background from the original signal to obtain the background-corrected signal. The optimized algorithm parameters are as follows: the structural element width, penalty parameter, flat area ratio and penalty difference order in the morphological operation are set to 50, 10, 1 / 2000 and 2, respectively. The spectrum extracted by the MPLS algorithm (imageM-subBG) is plotted on Figure 1 (c), The background-removed Raman spectrum reveals clearer Raman bands.
[0065] Then, discrete wavelet transform (DWT) is used to perform relevant interval denoising on each pixel spectrum. In the first step, the optimal number and location of variance change points are detected to determine the noise intervals with different levels, such as Figure 1 As shown in (d), the noise levels in different intervals of the Raman spectrum are indeed different, and the noise interval distribution of the Raman spectrum at different pixels is different. In the second step, the Daubechies 2 wavelet is used to perform discrete wavelet transform (down to level 5). The wavelet tree, details and approximation, detail coefficients and approximation coefficients of the 808th pixel spectrum are shown in Figure 1 (e) As shown. In the third step, the wavelet detail coefficients of each layer are processed using the hard threshold criterion. In the fourth step, the denoised signal is reconstructed using the approximate coefficients and the modified detail coefficients. The removed non-uniform noise is plotted on Figure 1 (f), it can be seen that the noise level does vary in different variable intervals. The interval-dependent denoised pixel spectra (imageM-subBG-DWT) are plotted on Figure 1 (g).
[0066] Finally, each spectrum was further soft-smoothed using DFT. Figure 1 (h) shows the process of Fourier transform smoothing the spectrum. In the first step, the spectrum is represented by a Fourier series consisting of a series of sines and cosines (yellow), and the spectrum in the time domain is transformed into an amplitude spectrum in the frequency domain (blue). In the second step, a window function (red) is used to eliminate the amplitude of the noisy frequency (from low frequency to high frequency, the amplitude of each frequency before the 65th frequency is multiplied by 1, and the amplitude of each frequency after the 65th frequency is multiplied by 0). In the third step, the signal is transformed from the frequency domain back to the time domain through an inverse discrete Fourier transform, which is the smoothed spectrum. The preprocessed pixel spectrum is shown in Figure 1. Figure 1 (i) as shown.
[0067] 4. MCR modeling process and imaging
[0068] The Multivariate Curve Resolution-Alternating Least Squares (MCR-ALS) algorithm performs curve resolution in an iterative manner (adding specific constraints when necessary) and can be used to decompose the instrument total response matrix D (a mixed signal of the raw measurements of all components) into the contribution of each pure component in the system, as shown below:
[0069] D I×J =C I×N S J×N T +E I×J
[0070] In the formula, C I×N and SJ×N T are the pure component profile matrices in the row and column directions respectively, E is the residual matrix; N is the number of chemical components that have signal contributions in the data.
[0071] Since Raman imaging data is a three-dimensional data array with a size of Xpixels×Ypixels×Jwavelengths (image T X×Y×J ). First, expand it into a column-wise expansion matrix (imageM) of size X*Yrows×Jcolumns. XY×J ); then perform simultaneous MCR-ALS analysis as shown below,
[0072] M XY×J =C XY×N S J×N T +E XY×J
[0073]
[0074] Thereafter, a common pure spectral matrix S J×N T and several profiles containing independent concentrations (C 1 ,C 2 ,…,C Y ) are analyzed, and the nth columns of C and S represent the relative concentration information and Raman spectrum of the resolved nth pure component, respectively. XY×N Refolding into a three-dimensional concentration information array C X×Y×N ,. Its nth forward slice matrix can be used to draw a molecular imaging map of the resolved nth component.
[0075] In this embodiment, the raw imaging data cube based on the frozen section measurement T 40×40×J ,like Figure 2 As shown in (a), after being preprocessed, it is expanded into a column-wise extended pixel spectrum matrix M 1600×J ,like Figure 2 (b) Based on the band attribution analysis results of the Raman spectrum of tobacco leaf midrib, two characteristic spectral intervals were carefully selected to achieve multivariate curve resolution of cellulose, HMP and LMP in leaf cell walls. The two spectral intervals are 200-450cm -1 (variable number J1 = 58) and 800-900cm -1 (Variable number J2 = 43).
[0076] Cellulose fingerprint range (200-450cm -1) and the final analyte imaging map are plotted on Figure 2 The first step is to expand the pixel spectrum matrix M from the preprocessed columns 1600×J Select the cellulose fingerprint interval to obtain the data matrix M XY×J1 (i.e. M 1600×58 ), and the MCR-ALS model was constructed for it. The number of groups was selected as 3 based on the singular value decomposition algorithm. In each iteration, a non-negative constraint was applied to the concentration profile and the spectral profile was divided by the sum norm. The algorithm converged after 32 iterations, and the data variance explained by the model was 94.67%. MCR-ALS decomposed the column-wise extended relative intensity profile matrix C 1600×3 and the normalized spectral matrix S 58×3 T ,like Figure 2 (c) and (e). The second step is to analyze the standard spectrum of cellulose and its characteristic band attribution. Figure 2 The component corresponding to the sky-blue spectrum in (e) (the first column in matrices C and S) is qualitatively identified as cellulose, and the components corresponding to the orange-red and yellow spectra are resolved interference components. In the third step, the first column of matrix C is refolded into a 40×40 matrix and the concentration distribution image of cellulose is drawn with it, as shown in Figure 2 As shown in (c1), cellulose is continuously distributed on the cell wall, with a higher content in the cell corners; the images of the two interfering components are similarly drawn on Figure 2 In summary, through the mathematical resolution process of the MCR model, we eliminated the spectral signals of interfering components from the Raman spectra of the complex plant system, obtained the pure Raman imaging of the analyte cellulose, and achieved precise microscopic imaging.
[0077] Pectin fingerprint range (800-900cm -1 ) and the final analyte imaging map are plotted on Figure 2 Similarly, in the first step, the pectin fingerprint interval is selected to obtain the data matrix M 1600×43 , the MCR-ALS model was constructed for it, where the number of groups was selected as 4 based on the singular value decomposition algorithm, and the non-negative constraint was used on the concentration profile in each iteration and the spectral profile was divided by the sum norm. The algorithm converged after 9 iterations, and the data variance explained by the model was 81.67%. MCR-ALS decomposed the column-wise extended relative intensity profile matrix C 1600×4 and the normalized spectral matrix S 43×4 T ,like Figure 2 (d) and (f). The second step is to analyze the standard spectra of HMP and LMP and their characteristic bands. Figure 2The components corresponding to the sky blue and orange-red spectra in (f) (the first and second columns in matrices C and S) are qualitatively identified as HMP and LMP, respectively, and the components corresponding to the yellow and purple spectra are resolved interference components. In the third step, the first and second columns of matrix C are refolded into 40×40 matrices, and the concentration distribution images of HMP and LMP are drawn with them, as shown in Figure 2 As shown in (d1) and (d2), HMP is continuously distributed on the cell wall, with higher content in the middle layer and cell corners; LMP is less continuously distributed on the cell wall, with higher content in the cell corners, but the content of LMP is significantly lower than that of HMP. The images of the two interfering components are plotted on Figure 2 (d3) and (d4).
[0078] It can be seen that the present invention not only eliminates the spectral signals of interfering components from the Raman spectrum measured in the complex plant system through the mathematical resolution process of the MCR model, but also resolves the overlapping spectral bands of the two analytes HMP and LMP, reflecting the characteristics of precise imaging of multiple analytes.
[0079] Application examples:
[0080] 1. Experimental samples
[0081] The tobacco leaf samples (Yunyan 87 variety) in the growth process were picked from Pingba District, Anshun City, Guizhou Province from June 17, 2022 to August 20, 2022. The W sampling method was used to select plants with consistent growth for sampling, and 5 leaves were collected each time, with the fourth leaf from the upper part as the sampling site. The tobacco leaf samples in the baking process were sampled at 10 temperature points, including the start of baking, 36°C (early yellowing), 38°C (early yellowing), 40°C (middle yellowing), 42°C (late yellowing), 44-46°C (early color fixation), 48-49°C (middle color fixation), 52-54°C (late color fixation), 60°C (early dry tendon), and 68°C (late dry tendon), with 5 leaves collected each time. The midribs and lateral veins of all samples were frozen and sectioned using a CryoStar NX50 cryostat (Thermo Fisher Scientific, USA), with a section thickness of 30 μm.
[0082] 2. Photos and Raman spectra of tobacco leaves
[0083] Figure 3(a) and (b) were photographed to record the appearance changes of tobacco leaves during the growth and maturity process (G1-R4) and the baking process (Y0-SD2). During the growth and maturity of tobacco leaves, after the cluster stage, the tobacco leaves enter the stem pulling period (indicated by G), which is the period when tobacco leaves grow most vigorously and accumulate the most dry matter. The growth rate of tobacco leaves accelerates, the leaf area increases rapidly, and the tobacco leaves are thin during this period. After the tobacco plants bud, they gradually mature (indicated by R), and the physiological activities of the tobacco plants are mainly manifested in the synthesis, accumulation and transformation of dry matter. When the upper leaves also begin to mature, the leaf area of the upper leaves gradually decreases and the leaf color changes from light green to yellow-green.
[0084] During the tobacco leaf baking process, the yellowing stage (represented by Y) is an important stage for the degradation and transformation of macromolecular substances inside the tobacco leaves. Through biochemical changes, macromolecular substances such as starch and protein in the tobacco leaves are decomposed and transformed to form aroma precursors. The color of the tobacco leaves gradually changes from yellow-green to yellow, and the shape changes to withering, softening, hooked and curled edges, but most of the leaves have not been dried. In the color fixing stage (represented by CF), the aroma precursors condense to form aroma substances, most of the leaves are dried, and the leaves further curl toward the midrib. The leaves reach a fully curled state in the late color fixing stage (CF3). In the dry tendon stage (represented by SD), under high temperature conditions, some aroma components decompose, and the midrib and leaves are fully dried.
[0085] For representative stages of the growth and maturation process and the baking process, Raman spectra of the cell wall were collected on frozen sections of leaf midribs using a Raman microspectrometer, and are shown in Figure 3 (c) and (d) The Raman spectra of the cell walls in the G1 and G2 phases appear very similar at first glance, with cellulose having a peak at 375 cm -1 The bending vibration of the CCC ring at this band seems to be stable; however, some subtle differences between the two spectra are visible, with pectin being highly methylated (836 cm -1 ), as the tobacco leaves grow, the pectin content is 856cm -1 The band gradually becomes stronger, indicating that the low-methoxyl pectin gradually increases. This trend is more obvious in the Raman spectra of the R1-R4 stage. In the mature stage, the band intensity of high-methoxyl pectin gradually increases, and the characteristic band of cellulose is 375cm -1 It also shows a trend of gradual strengthening. During the entire baking process, including the yellowing period, color fixing period and dry period, the intensity of the characteristic bands of cellulose and pectin is decreasing, although the decrease is relatively slow.
[0086] 3. Microscopic spatiotemporal changes of polysaccharides in the cell walls of midrib and lateral veins during the growth process
[0087] Figure 4The Raman microscopy images of the cell wall polysaccharides of the midrib and lateral veins of the growing leaves based on the MCR model are shown. For the cells of the midrib, in the early stage of vigorous growth (G1), cellulose and pectin (HMP and LMP) first appear in the cell corners, see Figure 4 As the plant continues to grow, starting from the mid-growth phase (G2), HMP gradually appears in the cell wall and middle layer. Figure 4 (b2). This change is more obvious in the mature stage (R1, R2 and R4), the HMP content gradually increases, and the distribution outline in the cell wall, cell corner and intercellular layer gradually becomes complete, as shown in Figure 2. Figure 4 (b3)-(b5), this change is related to the cell division in the rapid growth and maturity stage of tobacco leaves, because one of the functions of pectin is to be responsible for cell adhesion. In the vigorous growth period (G1-G2), there is no obvious change in the temporal and spatial distribution of cellulose on the cell wall, see Figure 4 (a1)-(a2). In the mature stage (R1, R2 and R4), the change pattern of cellulose is consistent with that of HMP, see Figure 4 (a3)-(a5).
[0088] The microscopic temporal and spatial changes of polysaccharides in the cell walls of lateral veins during the growth process are basically the same as those in the midrib, but there are slight differences, see Figure 4 (d1)-(d5), (e1)-(e5) and (f1)-(f5). For the cells of the leaf lateral veins, in the early stage of vigorous growth (G1), cellulose and pectin (HMP and LMP) first appeared at the cell corners, as shown in Figure 4 (d1), (e1) and (f1). At the early stage of maturity (R1), the signals of cellulose and HMP decreased slightly in the cell corners, while the content of LMP increased and appeared in the entire cell wall area. In the mature stage (from R2 to R4), the content of cellulose and HMP in the cell wall gradually increased, especially in the cell corners; the content of LMP did not change significantly, but eventually reached the highest concentration.
[0089] 4. Microscopic temporal and spatial changes of polysaccharides in the cell walls of midribs and lateral veins during baking
[0090] Figure 5Raman microscopy images of cell wall polysaccharides in the midrib and lateral veins of leaves during the baking process based on the MCR model are shown. For cells in the midrib, cellulose, HMP, and LMP in the cell wall are all degraded during the yellowing stage (Y0-Y3), and the degradation is more obvious at the cell corners. From the late yellowing stage (Y3) to the late color fixation stage (CF3) and then to the late stem (SD2), the leaves curl up in morphology and gradually dry, so the cell wall becomes thinner (as shown in the bright field image), the cell changes from an ellipse to an irregular polygon, and the cell corner becomes smaller. From the late yellowing stage (Y3) to the late color fixation stage (CF3), as the baking process progresses (with the increase in temperature), the cellulose, HMP, and LMP on the cell wall are all degraded. However, from the late color fixation stage (CF3) to the late stem (SD2), the concentration and distribution of cellulose on the cell wall remain basically stable; the distribution of HMP on the cell wall continues to narrow, with a trend of accelerated degradation; while the concentration of LMP on the cell wall is increasing. This is because HMP is demethylated during the baking process, which increases the content of LMP. In the late dry gluten stage (SD2), the concentration levels of HMP and LMP are similar.
[0091] For cells in the lateral veins of leaves, the microscopic temporal and spatial changes of polysaccharides in the cell walls of lateral veins during the yellowing period (Y0-Y3) were basically the same as those in the midrib, but there were slight differences in distribution, e.g. Figure 5 (d1)-(d4), (e1)-(e4) and (f1)-(f4). From the late yellowing stage (Y3) to the late color fixation stage (CF3), as the baking process progresses, the cellulose and LMP on the cell wall are degraded; however, the HMP signal on the entire cell wall is enhanced, which is different from the decrease in the Raman intensity of HMP on the midrib cell wall. The microscopic spatiotemporal changes of polysaccharides in the cell wall of the lateral veins from the late color fixation stage (CF3) to the late dry tendon stage (SD2) are similar to the rules of the midrib.
[0092] Comparative Example:
[0093] Taking the leaf midrib slices of the samples at the R1-R4 stage in the application example as an example, a larger range is selected for imaging, and the results given by the peak clamping method and the MCR model are compared (except for the peak clamping method and the MCR model, the other methods are the same as in Example 1), as shown in FIG. Figure 6 The images of cellulose, HMP and LMP given by the MCR model are accurate images of the specific pure analytes themselves, as shown in Figure 6 As shown in (c), the MCR-based imaging reveals the change pattern of HMP content increasing with the maturation process, and clearly reveals the microscopic spatiotemporal change pattern of cellulose content with the maturation process. However, the peak-clip method fails to distinguish HMP from LMP and interference, nor does it distinguish cellulose from interference, and only gives a mixed imaging of them, such as Figure 6 As shown in (b), it is difficult to reveal the above-mentioned changing rules.
[0094] The above-mentioned embodiments only express several implementation methods of the present invention, and the description thereof is relatively specific and detailed, but it cannot be understood as limiting the scope of the patent of the present invention. It should be pointed out that, for ordinary technicians in this field, several variations and improvements can be made without departing from the concept of the present invention, which all belong to the protection scope of the present invention. Therefore, the protection scope of the patent of the present invention shall be subject to the attached claims.
Claims
1. A method for imaging the structural components of tobacco leaf cell walls, characterized in that: include: S1 measures frozen section samples using a confocal Raman microscope to obtain raw imaging data cubes T X×Y×J ; S2S2 expands the original imaging data cube into a column-wise expanded pixel spectrum matrix M XY×J , preprocessing each Raman imaging data in the column-wise expanded pixel spectrum matrix; S3 selects the fingerprint interval of the analyte from the column-wise expanded pixel spectrum matrix to obtain the data matrix and constructs the MCR-ALS model, decomposing it to obtain the column-wise expanded relative intensity profile matrix C XY×N and the normalized spectral matrix S J×N ; S4 is based on the standard spectrum of the analyte and its characteristic band attribution analysis, and the relative intensity profile matrix C is expanded column-wise. XY×N and the normalized spectral matrix S J×N The components corresponding to the spectrum in the image are resolved into analytes and interfering components, and the relative intensity profile matrix C is expanded column-wise XY×N The columns corresponding to the analytes in the matrix are refolded into a matrix, and the concentration distribution image of the analytes is drawn using it.
2. The method for imaging the structural components of tobacco leaf cell walls according to claim 1, characterized in that: step In S1, the confocal Raman microscope imager uses a 50× visible microscope objective lens, NA = 0.75, uses an air-cooled back-illuminated CCD to detect Raman signals, uses a λ = 532nm linear polarized laser to generate incident light, has an output power of 10mW, and the grating is set to 600g mm -1 The exposure time was set to 3s. Before each measurement, the peak value was set at 520.70cm. -1 Silicon wafer calibration system.
3. The method for imaging the structural components of tobacco leaf cell walls according to claim 1, characterized in that: step In S2, the preprocessing includes: S21 removes the spikes caused by cosmic rays in the Raman spectrum; S22 removes background through MPLS algorithm; S23 performs relevant interval noise reduction on each pixel spectrum through discrete wavelet transform; S24 soft-smoothes each spectrum by discrete Fourier transform.
4. The method for imaging the structural components of tobacco leaf cell walls according to claim 3, characterized in that: step S23 includes: S231 determines the noise intervals of different levels in Raman spectra; S232 uses Daubechies 2 wavelet for discrete wavelet transform; S233 processes the wavelet detail coefficients of each layer using a hard threshold criterion; S234 reconstructs the denoised signal using the approximation coefficients and the modified detail coefficients.
5. The method for imaging the structural components of tobacco leaf cell walls according to claim 3, characterized in that: step S24 includes: S241 uses a series of Fourier series consisting of sines and cosines to represent the spectrum, and the spectrum in the time domain is transformed into an amplitude spectrum in the frequency domain; S242 uses a window function to eliminate the amplitude of the noisy frequency; The S243 signal is transformed from the frequency domain back to the time domain through inverse discrete Fourier transform, and a smoothed spectrum is obtained.
6. The method for imaging the structural components of tobacco leaf cell walls according to claim 1, characterized in that: In step S3, the construction and decomposition method of the MCR-ALS model is: S31 model initialization Initialize the model using the singular value decomposition algorithm and set the number of groups; S32 Iteration Process In each iteration, a non-negativity constraint is applied to the concentration profile; Divide the spectral profile by the sum norm; The algorithm converges after a set number of iterations; S33 Results Matrix Decompose the column-wise extended relative intensity profile matrix C XY×N and the normalized spectral matrix S J×N .
7. The method for imaging the structural components of tobacco leaf cell walls according to claim 1, characterized in that: The analytes include cellulose, pectin, lignin and structural protein.
8. The method for imaging tobacco leaf cell wall structural components according to claim 7, characterized in that: The fingerprint range of the cellulose is 200–450 cm -1 The fingerprint range of the pectin is 800–900 cm -1 .
