A procedure for the simultaneous analysis of cell-wall polysaccharides, amides and aliphatic polyesters by transmission Fourier transform infrared microspectroscopy (FTIR) has been established for Arabidopsis petals. The combination of FTIR imaging with spectra derivatization revealed that petals, in contrast to other organs, have a characteristic chemical zoning with high amount of aliphatic compounds and esters in the lamina and of polysaccharides in the stalk of the petal. The hinge region of petals was particular rich in amides as well as in vibrations potentially associated with hemicellulose. In addition, a number of other distribution patterns have been identified. Analyses of mutants in cutin deposition confirmed that vibrations of aliphatic compounds and esters present in the lamina were largely associated with the cuticular polyester. Calculation of spectrotypes, including the standard deviation of intensities, allowed detailed comparison of the spectral features of various mutants. The spectrotypes not only revealed differences in the amount of polyesters in cutin mutants, but also changes in other compound classes. For example, in addition to the expected strong deficiencies in polyester content, the long-chain acyl CoA synthase 2 mutant showed increased intensities of vibrations in a wavelength range that is typical for polysaccharides. Identical spectral features were observed in quasimodo2, a cell-wall mutant of Arabidopsis with a defect in pectin formation that exhibits increased cellulose synthase activity. FTIR thus proved to be a convenient method for the identification and characterization of mutants affected in the deposition of cutin in petals.
Plant cells have a complex primary cell wall that consists of polysaccharides, such as cellulose, hemicellulose and pectin, and structural proteins (Cosgrove, 2005). In addition, secondary cell walls are further reinforced by the aromatic polymer lignin (Boerjan et al., 2003). Cell walls may also be impregnated by aliphatic polyesters, cutin and suberin, as well as their associated waxes (Jetter et al., 2006; Pollard et al., 2008). The composition of the extracellular matrix of plant cells depends on the organ, cell type and developmental stage of the plant (Pennell and Roberts, 1995).
The epidermis of the shoot has a specialized extracellular matrix as the cuticle, which consists of soluble waxes and cutin, a polyester of mainly oxygenated fatty acids and glycerol, forms an interface with the aerial environment. The cuticle covers the organs and protrudes deep into the primary cell wall (Jeffree, 2006). Its formation is tightly regulated during cell expansion, and when the epidermal cells are no longer expanding, cutin is most likely further structurally modified (Suh et al., 2005). Understanding of the tri-dimensional structure of cutin, as well as its structural modifications during plant growth and development, is still elusive (Pollard et al., 2008). Although the cuticle has a tight association with the cell wall, potential relationships between the cuticle and cell-wall composition have not been investigated.
During the last decade, a number of Arabidopsis mutants have been identified that have an altered cutin amount and composition, and the corresponding gene functions in cutin biosynthesis have been characterized (Beisson et al., 2012). However, only recently has characterization of regulators of cutin biosynthesis provided molecular evidence that cutin formation is directly linked to cell-wall formation (Shi et al., 2011). In Arabidopsis, the SHINE1/WAX INDUCER1 (SHN1/WIN1) transcriptional regulator directly regulates a number of genes involved in cutin biosynthesis (Kannangara et al., 2007; Shi et al., 2011). Down-regulation of expression of all three SHN homologs altered the expression of genes involved in pec- tin modifications as well as extracellular matrix proteins (Shi et al., 2011). However, simultaneous characterization of polysaccharides, polyesters and proteins in muro by chemical and immunological methods has not been performed.
Fourier transform infrared spectroscopy (FTIR) is a method that has been used for the molecular characterization of cell-wall polymers, cutin and proteins (Villena et al., 2000; Dokken et al., 2005). Chemical identity may be assigned to specific molecular vibrations for chemically pure substances. However, the complexity of the polysaccharide fraction in the cell wall is very high, and correct assignment of spectral features to certain cell-wall polysaccharides is rather difficult in muro. A wide variety of cell-wall fractions, as well as specific tissues or developmental stages of certain cell types, have been characterized by FTIR during recent decades, leading to assignment of characteristic vibrations to particular cell-wall polymers (Gorzsas et al., 2011). FTIR has also been successfully used for the identification and classification of cell-wall mutants in Arabidopsis (Chen et al., 1998; Mouille et al., 2003; McCann and Carpita, 2005). The development of FTIR microspectroscopy and imaging techniques further increases the possibility of exploring the composition and localization of extracellular matrix components. These techniques have been particularly useful when combined with chemical or genetic analyses (Carpita et al., 2001; Carpita and McCann, 2002). Although the simultaneous characterization of aliphatic polyesters, proteins and polysaccharides in muro may be feasible, FTIR methods for identification of mutants that are affected in formation of non-polysaccharide cell-wall components or characterization of the relationship between the cuticle and the cell wall have not been developed.
Here we characterize the distribution of spectral features associated with various classes of extracellular matrix components in Arabidopsis by FTIR microspectroscopy. A procedure based on FTIR imaging has been developed to assess the distribution of spectral features of petals and cotyledons. Studies of mutants that are impaired in cutin formation, such as long-chain acyl CoA synthetase 2 (lacs2) and permeable cuticle 1 (pec1), indicated that the peaks of aliphatic compounds and esters in petals may be attributed to a large extent to cuticular polyesters. Spectral features of the petals of several mutants affected in deposition of extracellular matrix components, such as lacs2, pec1 and quasimodo2 (qua2), are discussed. FTIR microspectroscopy has been identified as a very useful technique for chemical characterization of the extracellular matrix composition of petals, an organ that is difficult to access by chemical analyses.
Analysis of the FTIR spectrum of the lamina of Arabidopsis petals
Transmission FTIR microspectroscopy has been exploited to obtain information on the chemical composition of the entire Arabidopsis petal. The infrared spectrum of the delipidated petal lamina is shown in Figure 1. The most characteristic spectral features visible on the spectrum are connected with the presence of functional chemical groups of aliphatic compounds, esters and amide band vibrations as well as polysaccharides. As these chemical groups are potentially associated with three types of polymers, namely cutin, proteins and polysaccharides, methanolysis of polyesters as well as enzymatic digestion with proteinase K and/or pectolyase were performed. Spectra obtained after the treatments are presented in Figure 2.
The broad spectral band with a maximum at approximately 3400 cm−1 was assigned to the ν(OH) vibration of hydrogen-bonded hydroxyl functional groups. The two intense bands at 2919 and 2850 cm−1 were attributed to the asymmetric and symmetric stretching ν(C-H) vibrations of the methylene group of aliphatic compounds, respectively (Figure 1). Both peaks disappeared after acid-catalyzed methanolysis of aliphatic polyesters, confirming the assignment of band vibrations to aliphatic compounds (Figure 2).
A characteristic intense peak at 1734 cm−1 was assigned to the stretching ν(C=O) band vibration of the carbonyl group in esters (Johnson et al., 2007; Figure 1). Aliphatic polyesters and pectin methylesters mainly contribute to this band vibration in plant tissues. Methanolysis of aliphatic polyesters abolished the vibration at 1734 cm−1, while pectolyase treatment only diminished the width of the peak, indicating a minor contribution of pectin esters (Figure 2).
The strong broad band in the 1550–1700 cm−1 range, with a characteristic double peak with maxima at 1656 and 1627 cm−1, was assigned to amide I, resulting from (C=O) and (C–N) stretching vibrations of the peptide bond (Surewicz and Mantsch, 1988; Zhang et al., 1992). Similarly, a medium-intensity amide II peak originating from bending (N–H) and stretching (C–N) vibrations of the amide bond was identified at 1540 cm−1 (Figure 1). Amide I and amide II band vibrations were significantly reduced after treatment with proteinase K, demonstrating that the majority of such vibrations originate from proteins (Figure 2).
The massive band in the 900–1200 cm−1 range is associated to the ring vibrations, overlapped with the stretching vibrations of the (C–OH) side groups and the (C–O–C) glycosidic band vibrations of polysaccharides in bulk and de-waxed cutin (Kacurakova et al., 2000. Sequential treatment of petals with proteinase K and methanolysis of polyesters enabled us to obtain a spectrum of the polysaccharide skeleton of the Arabidopsis petal (Figure 2). Interestingly, the pectolyase treatment noticeably reduced the intensity of the IR signal in the polysaccharide region, indicating a relatively high content of pectins in the petal tissue.
In the next stage, the second derivative of the FTIR spectrum was calculated. Derivatization of spectra allows a more specific identification of small and nearby band vibrations. It also enables establishment of the exact position of peaks on the infrared spectrum, because the minima on the derivative plot correspond to the bands' maxima (Susi and Byler, 1983). The differentiation automatically enhances the number of spectral features that may be analyzed and helps to identify some peaks that were hidden under broad bands in the raw spectrum. Additionally, the derivatization process eliminates constant factors from the spectrum and reduces some baseline effects.
The overlap of the FTIR spectrum of the petal (thick line) and its derivative (thin line) is shown in Figure 1 (top part). The differentiation confirmed the position of all the previously mentioned band vibrations. The process enhanced the spectral representation of aliphatic compounds by two additional bands. An asymmetric stretching νas(CH) vibration at 2928 cm−1, which was hidden under a shoulder of the 2919 cm−1 band, and the symmetric bending δs(CH2) vibration at 1465 cm−1 were revealed. Furthermore, an additional amide II band vibration at 1515 cm−1 became visible. Particularly strong was the enrichment in spectral features in the part of the spectrum between 900 and 1200 cm−1, in which band vibrations typical of polysaccharides are located. The full list of resolved peaks and their assignment is provided in Table S1 and Figure S1.
Chemical zoning in Arabidopsis petals
In order to obtain information on whether the distribution of chemical compounds shows variations over the entire petal, the FTIR imaging technique was applied. The spectral data were differentiated, and absorbance maps were obtained and then analyzed according to the minima on the second derivative (see Figure S1). This procedure revealed a number of spatial distribution types for the previously identified spectral features of Arabidopsis petals, reflecting differences in the chemical composition across the petal. These distribution types are closely related to the morphology of the petal.
A high concentration of vibrating species characteristic of the (C–H) band vibrations at 2928, 2919, 2850 and 1465 cm−1 is dominant in the petal lamina. The intensity of a signal in the hinge region of the petal is twofold smaller. The same pattern on the map is obtained for the stretching (C=O) band at 1734 cm−1, i.e. for the carboxyl group vibration of esters. The similarity of the absorbance maps for the (C–H) and (C=O) bands indicates that the peaks listed above (2928, 2919, 2850, 1734 and 1465 cm−1), belong to the same compound class, i.e. esters of aliphatic compounds, potentially cutin. The map for the peak at 1734 cm−1 is shown in Figure 3(a), and maps obtained for the other bands are shown in Table S1.
A very characteristic map was obtained for the amide I and II band vibrations. The same ‘foot-like’ pattern with the maximal concentration of the compound in the stalk was obtained for any peak assigned to amides, namely at 1656, 1627, 1540 and 1515 cm−1. The concentration of vibrating species changed 10-fold from the lamina to the stalk of the petal, as shown in Figure 3(b) and Table S1. The relatively high contribution of proteins in the FTIR spectrum of petals led to a clear assignment of these peaks. The identification of other substances with less characteristic spectral features may be hindered by the presence of even low amount of amides. This is visible in the case of the band at 1439 cm−1, for example, which may be assigned to both amide III and pectin bending (C–H), and absorbance maps do not give an explicit answer.
In the ‘fingerprint region’ of the IR spectrum (900–1200 cm−1), several additional types of absorbance maps were observed and are expected to characterize polysaccharides. The maps for the peaks at 1167, 1106 and 1071 cm−1 have a flat distribution pattern as shown in Figure 3(c) and Table S1, and are potentially associated with cellulose (Gorzsas et al., 2011). Similar absorbance maps were also obtained for bands at 1368 and 1318 cm−1.
A very characteristic pattern of absorbance maps was found for another set of vibrations (1454, 1387 and 1148 cm−1) that are potentially associated with hemicellulose (Gorzsas et al., 2011). The absorbance maps displayed a high concentration of vibrating species on the stalk and hinge region of the petal without any contribution in the lamina, as shown in Figure 3(d) for 1387 cm−1 and Table S1.
Another particular absorbance map was obtained for the peak at 1508 cm−1, clearly showing a vein-related location of the compound, as shown in Figure 3(e). This pattern suggests that the peak may originate from the aromatic (C=C) band vibration of lignin (Gorzsas et al., 2011). The position of the peak may be found on the slope of the second derivative of the IR spectrum only, as shown in Figure S1. This band was not directly visible because of a high contribution of amide II vibrations at 1515 cm−1. The other peaks typical for aromatics were not observed as they were hidden under the intensive amide band vibrations.
Chemical zoning is not observable for cotyledons
Additional measurements utilizing FTIR imaging were performed for cotyledons in order to obtain absorbance maps similar to those characteristic for petals. A detailed list of band assignments and absorbance maps for cotyledons according to minima on the second derivative of the spectra (Figure S2) is shown in Table S2. The aliphatic compounds showed interesting differences in the pattern of absorbance maps in comparison to petals. Two separate groups of compounds were easily distinguishable in the stretching (C–H) region. The first group showed a significant increase in the signal on veins as visible on the maps obtained for the 2958, 2933 and 2872 cm−1 band vibrations, see Table S2. In the second group of aliphatic compounds, the absorbance is evenly distributed on the lamina without a higher concentration on veins, as it is visible for absorbance maps of 2916 and 2849 cm−1 band vibrations. The peak at 1734 cm−1 indicating the presence of esters, which shows a very characteristic pattern in petals, is not detectable in the spectrum of cotyledons. The pattern on the map obtained for the (C=O) band vibration at 1740 cm−1, potentially originating from pectin methylesters (Kacurakova et al., 2000), was uniform with only a slight increase of the signal in the stalk. The absorbance maps of amide bond vibrations were almost identical to those of the first group of aliphatic compounds. A characteristic map pattern was observed for a peak at 1508 cm−1 that may be attributed to lignin (Gorzsas et al., 2011). The signal appeared on veins and along the border of the lamina, but not on the lamina. Peaks that may be assigned to cellulose according to the 1369, 1314, 1168 and 1056 cm−1 band vibrations (Kacurakova et al., 2000) have the same distribution pattern, showing an increase in signal intensity in the petiole, as shown in Table S2. In summary, cotyledons did not show extensive chemical zoning as petals did. Only veins and petioles had a pattern distinct from the cotyledon blade.
Cutin may be assessed by FTIR spectroscopy in Arabidopsis petals
As the identical distribution pattern of aliphatics and esters indicated that FTIR may detect cutin in Arabidopsis petals, the lacs2 and the pec1/abcg32 mutants that have been previously described as having an altered amount of cutin were investigated in comparison to their wild-type (Schnurr et al., 2004; Bessire et al., 2007, 2011; Tang et al., 2007). The cell-wall mutant qua2 was also included in the comparison (Mouille et al., 2003, 2007). Because of the extensive chemical zoning in petals, measurements were only performed in a confined area of the lamina, 50–350 μm from the tip.
For detailed comparison of the various genotypes, the spectrotypes, i.e. average spectra including the standard deviation (SD) of intensities at each wavenumber, were calculated. Direct overlap of the spectra showed striking differences in some peak intensities for pec1 and lacs2 mutants in comparison to the wild-type. These differences were visible in both alleles of pec1, namely pec1–2 and pec1–3, and both alleles of lacs2, namely lacs2–3 and lacs2–4. The biggest changes were visible for bands that were assigned to esters and aliphatic compounds. The intensities of the (C=O) and (C–H) peaks were significantly reduced for cutin mutants in comparison to their intensity on the wild-type spectrum. The IR spectrotypes of all the genotypes are shown in Figure 4 and Figure S3.
More detailed information about differences in chemical composition between plants was obtained by digital subtraction of spectrotypes of the mutants from the wild-type. The presence of a set of positive peaks at 2919, 2850, 1734, 1455, 1240 and 1170 cm−1 on the difference spectra confirmed the aliphatic character of the spectral deficiency as a result of a reduction in cutin formation for the pec1 and lacs2 mutants. The observed differences were greater than the SD of absorbance intensities established for wild-type, as shown in Figure 5(a). Student's t tests were performed to determine the statistical significance of the observed differences. The differences between spectrotypes of wild-type and cutin mutants, visible in the form of peaks that appeared on the (wild-type minus mutant) subtraction plots, were statistically significant. Interestingly, the t-value plots indicated some additional significant spectral differences between wild-type and mutants in the range of stretching (C=O) and (C–H) band vibrations; these differences were not observable when digital subtraction method was applied. Based on the t value plot, the significance of the differences in intensity of band vibration at 1711 cm−1 is similar to the differences in intensities of the most pronounced ester peak at 1734 cm−1. The last peak is assigned to the double-bonded (C–O) vibration in (–COOR) groups, while the peak at 1711 cm−1, which was not detectable by subtraction of spectrotypes, represents the vibration of ester groups involved in hydrogen bonding (–COOR···HO–) within polyesters rich in hydroxylated fatty acids (Heredia-Guerrero et al., 2010). The results of application of Student's t tests to the spectral data set are shown in Figure 5(b).
In contrast to the mutants in cutin deposition, the mutant qua2 did not show any differences in the vibrations of aliphatic compounds and only a small reduction in the intensity of the ν(C=O) band vibration associated with the ester bond. The reduction in ester bond vibration was still within the range of the SD, but was significant according to the t value plot.
The overlap of the spectra of all genotypes as well as the difference spectra indicated that pec1 had more cutin than lacs2 (Figures 4 and 5). The spectrotype of lacs2–3 was therefore subtracted from that one of pec1–2. Maxima at 2919, 2850 and 1734 cm−1 on the difference plot as well as the t value plot confirmed that pec1–2 had significantly more aliphatic esters in petals than lacs2–3 (Figure S4).
Calculation of the areas under the ν(C=O) peaks was used to estimate cutin content in muro by FTIR. Figure S5 shows that both cutin mutants, lacs2 and pec1, had a significant reduction in esters in comparison to wild-type, but qua2 did not show a significant difference in ester content. The lacs2 mutant showed an average 30% reduction in esters, while pec1 showed only a 20% reduction. However, the lower ester content of lacs2 compared with pec1 was not significant by this evaluation method. FTIR imaging was also performed with the lacs2 and pec1 mutants. Absorbance maps for the stretching (C=O) band vibration at 1734 cm−1 associated with the ester bond were obtained to compare the polyester distribution of wild-type and cutin mutants. When the same intensity scale was applied, the major concentration changes in the polyester were easily visualized (Figure 6). When the intensity scale was reduced by a factor of two for the mutants, the characteristic distribution pattern of aliphatic esters became visible again, showing that the residual cutin in the petals of the mutants is also mainly present in the lamina.
Morphology of the epidermal cells of petals
In order to investigate whether the gradient of cutin amount as evidenced by FTIR imaging may be correlated to a morphological gradient, cryo-scanning electron micrographs were obtained from the entire Arabidopsis petal (Figure 7). On the adaxial side of the petal, epidermal cells have a conical shape, and the vertical nanoridges originating from the top of the cones have a substantial height of approximately 15–20 μm (Figure 7a,b). In contrast, in the stalk, epidermal cells are very elongated, having a flat surface with very thin parallel-oriented nanoridges (Figure 7a,d). In the middle of the petal, the hinge region, a transition between these two forms of epidermal cells is visible (Figure 7c,e). Epidermal cells have an intermediate morphology, being much shorter but having mainly a flat surface, while some of them form a flat cone (Figure 7c,e). In roundish cells, nanoridges are radially oriented and more prominent than in elongated cells (Figure 7c). This transition zone is rather narrow (approximately 150 μm), although the conical shape appears in cells above veins slightly more towards the tip of the lamina compared to cells between veins (Figure 7d).
FTIR detects increases in polysaccharide vibrations in qua2 and lacs2 mutants
A comparison of the qua2 and wild-type spectrotypes showed a difference in absorbance in the 900–1200 cm−1 range (Figure 4). A negative broad peak with an intensity that is greater than the SD level is visible on the difference spectrum as well as on the t value plot, as shown in Figure 5, demonstrating a significant increase in polysaccharide material in the mutant. Interestingly, the same negative broad peak on the difference spectrum was obtained for the lacs2–3 mutant. In the lacs2–4 mutant, similar but not identical spectral features were observed. Spectral differences were approximately at the limit of the SD, and differences were visible in both lacs2–3 and lacs2–4 mutants alleles, but not in both alleles of pec1. The t value plots showed that the increase in polysaccharides was significant in both lacs2 mutants as well as in the qua2 mutant. Consistent changes in polysaccharide content were not detected in the pec1 mutant alleles by this evaluation method.
There was an indication of an increase in protein content in pec1 and lacs2 mutants in comparison with wild-type. However, the peaks of amide band vibrations were below the significance threshold, and the t value plot pointed only to the statistical significance of the stretching (N-H) vibration, known as amide A, with minima at 3390 cm−1 on the difference plot. In addition, an irregular fluctuation of intensities of amide was observed in various series of experiments, in the wild-type spectra as well in those of different mutants.
Advantages and limitations of transmission FTIR
Transmission FTIR microspectroscopy has been identified as a suitable method for simultaneous characterization of various polymeric compounds in Arabidopsis. Limitations associated with measurements in the attenuated reflection (ATR) mode, the infrared technique that is often used in cutin analysis, were excluded. These include lack of refractive index values for various tissue fragments in the microscopic scale, and the changing depth of penetration of plant material by the laser beam (estimated to be as high as 1.5 μm), which is dependent on wavelength (da Luz, 2006). Although application of transmission FTIR spectroscopy is limited by transparency of the studied material to the electromagnetic spectrum in the mid-IR range, measurements may be undertaken in roots, cotyledons, juvenile leaves and various flower organs of Arabidopsis after a simple procedure of pigment and water removal.
The petal of Arabidopsis was identified as an ideal organ for FTIR microspectroscopy because of its delicate thickness and low amount of pigmentation. Spectra of the lamina of the petal are characterized by the presence of vibrations in the form of broad uncharacteristic peaks as typically seen in complex biological material (Figure 1). Derivatization of the FTIR spectrum, a commonly used procedure in analytical chemistry (Susi and Byler, 1983), revealed features that were hidden under broader peaks of the raw spectrum. FTIR imaging analyses were performed for all these uncovered spectral features (Figure S1). The imaging technique allowed assignment of the distribution of functional groups within the entire organ. However, all chemical components that form the tissue contribute to the FTIR spectrum. Therefore, assignment of particular bands to the functional groups of specific molecules is often difficult and requires additional investigations. FTIR imaging for a bigger fragment or whole tissue, in this study the entire petal or cotyledon, gives a unique opportunity for a detailed band vibration assignment on the basis of spatial distribution. By comparing the patterns of absorbance maps, it is possible to assign unknown or not clearly assigned peaks on the spectrum more accurately to a certain group of compounds as vibrations originating from the same molecules must have the same spatial distribution. In addition, FTIR imaging helped in identification of the part of petal in which the vibrations of the compound of interest are the least influenced by vibrations of other compounds having another distribution pattern. For example, vibrations of aliphatics and esters are most stable at the tip of the petal, as the high intensities of amide bond vibrations disturb their assessment in other parts of the petal.
FTIR imaging as a tool for establishing compound classes
FTIR imaging combined with spectra derivatization leads to identification of several distribution patterns in Arabidopsis petals (Figure 3). The pattern on the absorbance map seen in Figure 3(a) (very high absorbances in petal lamina) was found for all vibrations associated with functional groups of aliphatics (2928, 2919, 2850 and 1465 cm−1) as well as esters (1734 cm−1; Table S1 and Figure S6), suggesting a common chemical origin. Indeed, both treatment of petals with agents that depolymerize aliphatic polyesters, as well as investigation of mutants affected in the deposition of cutin, unequivocally attributed a major part of the intensities of these vibrations to cutin.
Interestingly, after depolymerization of aliphatic polyesters in the presence of acid and methanol, a small peak remained in the area of stretching (C=O) vibrations, with a maximum at 1740 cm−1, that was also present in the polysaccharide backbone after digestion with proteinase K/methanolysis of polyesters (Figure 2). Acid-catalyzed trans-esterification in the presence of methanol depolymerizes aliphatic polyesters, but maintains methylesters that are present in pectin. Thus, acid-catalyzed depolymerization of polyesters revealed the band vibration from pectin methylesters that is usually hidden under the dominant polyester peak at 1734 cm−1 (Figure S7). Subsequent digestion with pectolyase or cleaving of the ester bond by base treatment strongly diminished the peak and indicated a high contribution of pectin to the peak intensity (Figure S8). In contrast, in cotyledons, in which polyesters are below the detection limit, only a peak at 1740 cm−1 is visible, which disappears after treatment with pectolyase or base (Figure S9). The contribution of pectin methylesters to the (C=O) vibrations leads to an under-estimation of calculated reductions in the quantity of the polyester in cutin mutants (Figure S5).
The distribution pattern seen in Figure 3(b) (a characteristic ‘foot’ with an extremely strong intensity gradient) was visible for all peaks attributable to amide bond vibrations, as present in proteins. Indeed, proteinase K treatment abolished the amide bond vibrations in the FTIR spectra, demonstrating that proteins were the source of these vibrations (Figure 2). It will be particular interesting to determine whether these are mainly proteins of the extracellular matrix or cellular proteins.
The low specificity and similarity of the spectra of polysaccharides makes definitive assignment of particular peaks in the range of 900–1200 cm−1 to cellulose, hemicellulose or pectin difficult and often ambiguous (Kacurakova et al., 2000; Gorzsas et al., 2011). The IR spectra of an isolated or purified material may noticeably differ in comparison with the spectra of the same compound recorded in planta due to structural changes in polysaccharides during chemical treatments. The vibrations of compounds may also change due to different chemical environments, i.e. in different tissues. This problem may be illustrated by comparing peaks that may possibly be attributed to cellulose in Arabidopsis petals and cotyledons. The band vibrations that are often taken as a fingerprint of cellulose are present at 1160, 1105, 1060 and 1040 cm−1 (Carpita et al., 2001). In cotyledons, peaks that are probably derived from cellulose appear at 1168, 1155, 1105 and 1056 cm−1, while the peak at 1040 cm−1 is not detectable (Figure S2). In petals, we observe a shift of their positions to 1167, 1106 and 1071 cm−1 (Figure S1), and the absorbance maps generated for these band vibrations have the same pattern, as shown in Table S1. These peaks are thus prime candidates for cellulose-derived peaks in petals, as they are in the range for cellulose vibrations (Gorzsas et al., 2011). Thus, the tentative assignment of 1167, 1106 and 1071 cm−1 in petals to cellulose exemplifies the advantages of using imaging as support for peak assignments.
Biological relevance of the distribution of cutin in Arabidopsis petals
The distribution patterns obtained by FTIR imaging probably reflect differences in the extracellular matrix composition in the petal, but understanding of their biological significance is elusive. As cutin is the structural polymer of the cuticle on the surface of epidermal cells and contributes to nanoridge formation (Li-Beisson et al., 2009; Panikashvili et al., 2009), we directly compared the distribution pattern of aliphatic esters by FTIR with the shape of epidermal cells and the formation of nanoridges. The amounts of cutin measured in the petal nicely paralleled the number and size of nanoridges on epidermal cells, being high in the petal lamina and low in the stalk, with only a narrow transition phase in the hinge region. However, the shape of epidermal cells in both parts of the cell type may contribute to the increased intensity of the FTIR signal. In the stalk, epidermal cells are flat, thus FTIR collects a signal directly from the cuticle layer of the cell. In contrast, in the petal lamina, epidermal cells have a conical shape, thus their three-dimensional surface is projected into two dimensions during sample preparation and FTIR measurements. Thus, signals that originate specifically from the cone of epidermal cells are exaggerated, making estimation of polyester content in various mutants even more complex when the cell shape is modified by the mutation, e.g. changes in the height of the cones or the amount of nanoridges (Li-Beisson et al., 2009; Panikashvili et al., 2009). In addition, changes in the architecture of mutant petals, e.g. more internal cells, or alterations in pectin esterification may interfere with signals originating from aliphatic esters.
Estimation of the significance of spectral differences
For detailed comparison of various genotypes, a procedure has been established that comprises computation of average spectra, i.e. spectrotype, including determination of the SD of intensities. Because direct overlay of spectrotypes does not give a clear insight into the spectral variation between genotypes, difference spectra were obtained by subtraction of the spectrotype of the mutant from that of the wild-type. The calculation of SD, as criterium of statistical significance, for the subtraction results helped to detect the most pronounced spectral differences. By this approach, only a few observed differences between genotypes were found to be significant (Figure 5). The amount of aliphatic esters (cutin) was significantly reduced in lacs2 and pec1 mutants in comparison to wild-type, and the polysaccharide content was increased in the qua2 mutant, which has increased cellulose synthase activity in leaves (Mouille et al., 2003).
Interestingly, a few other spectral differences were strikingly close to the SD significance level, such as a reduction of ester content in the qua2 mutant, which has been found to have a lesion in pectin biosynthesis and a significant reduction in esterification of pectins (Mouille et al., 2007; Ralet et al., 2008). The significance of spectral changes was also evaluated based on Student's t test statistics. Interestingly, almost all spectral differences between genotypes were significant at the established confidence interval by this method. Therefore, comparison of both methods gives a more complete view of which spectral differences are significant. Another possibility is to take only spectral differences that are larger than half of the SD into account; this criterion is commonly used in medical sciences (Feinstein, 1999). For example, the increase in polysaccharides in the lacs2 mutant is significant according to this criterion. Therefore, FTIR suggests that some of the mutants affected in cutin composition may have also an altered cell-wall composition. Further studies are required to follow this observation.
In the literature, two approaches can be found using infrared spectroscopy as a tool in cuticle research. In the classical method, isolated cutins were dried, powdered and dissolved in KBr. Spectra of pellets recorded in transmission mode were used to analyze structural changes of cutin after various methods of depolymerization (Villena et al., 2000; Graça et al., 2002; Benitez et al., 2004). The use of ATR accessories enabled collection of the spectral data from the surface of studied tissues, leaves and fruits, as well as thin films of isolated cutin. Both methods provide valuable information about the chemical composition of cutin and its changes due to various influences, but do not provide information about the spatial distribution of the polymers in the plant organ.
FTIR as additional method for the characterization of cutin mutants in Arabidopsis
In this paper, we have shown that cutin may be measured by FTIR in muro in Arabidopsis. This is partly possible because the petal has a high ratio between epidermal and internal cell-wall quantity (Pyke and Page, 1998). In addition, a deposition pattern was determined in petals, with higher amounts in the petal lamina. This pattern was not disturbed in the Arabidopsis lacs2 and pec1 mutants investigated in this study. FTIR imaging thus provided the possibility of directly visualizing deposition of cutin over an entire organ. It gave evidence of an unbalanced distribution of cutin that may in part be explained by the morphology of epidermal cells and their surface structure.
In summary, in this paper we present a procedure for imaging the chemical composition of entire Arabidopsis organs, and show the advantages of FTIR imaging for assigning vibrations to various cell-wall polysaccharides. We furthermore demonstrate that cutin may be directly measured by FTIR in Arabidopsis petals, providing opportunities for studying the cuticle–cell wall continuum in Arabidopsis.
Arabidopsis thaliana ecotype Columbia (Col–0), lacs2–3 (Bessire et al., 2007), lacs2–4 (Tang et al., 2007), pec1–2 (Bessire et al., 2007), pec1–3 (Bessire et al., 2011) and qua2 (SALK line 122037C) were analyzed. Plants were grown on a pasteurized soil mix under 10 h light (20°C day temperature and 17°C night temperature, 60% humidity) for vegetative growth, and 16 h light (constant temperature of 20°C, 60% humidity) for flowering plants. Fully expanded cotyledons and open flowers were harvested and placed in a methanol bath to remove soluble lipids. Dried samples were moved onto KBr disks to record IR spectra.
Chemical treatments and digestions
Fresh or dried petals prepared as described above were treated. Polyesters were removed as described by Li-Beisson et al. (2010). Proteins were removed by digestion with proteinase K (Roche Diagnostics, Rotkreuz, Switzerland; 0.5 mg ml−1) in 0.1 m Tris/HCl buffer (pH 7.9) for 3 h at 37°C. Pectin was degraded by digestion with pectyolase Y–23 (Kyowa Chemicals, Kagawa, Japan; 5 mg ml−1) in 25 mm MES/KOH, 5 mm EDTA, 5 mm EGTA, pH 5.5, for 5 h at room temperature after pre-incubation of the enzyme for 10 min at 55°C to remove potential protease activity. Pectin esters were hydrolysed as described previously (Sene et al., 1994).
FTIR spectroscopy and data treatment
A Nicolet iN10 MX microspectrometer (Thermo Fisher Scientific Inc., Madison, WI, USA) equipped with a liquid nitrogen-cooled mercury/cadmium/telluride (MCT/A) detector and KBr beam-splitter was used to perform the IR measurements. Transmission spectra were recorded in the range of 4000–650 cm−1 at a resolution of 4 cm−1. The interferograms were averaged over 64 scans. Samples were placed between two KBr disks, and spectra at the aperture of 50 × 50 μm were collected from a predefined area of petals, omitting veins. Digital data were transferred into a matlab environment (version 7.9.0, R2009b; MathWorks Inc., Natwick, MA, USA). Spectra were baseline-corrected and normalized by dividing by the area under the curves, and the average spectrum was computed. The SD of absorbance was determined at each wavenumber to estimate the spectral variability of the genotype.
An array MCT detector was set up to perform the FTIR imaging measurements. A rectangle area, approximately 5000 points at a resolution of 25 μm in the x and y dimensions, covering the whole petal was defined. Spectra were collected from each point at an aperture of 25 × 25 μm with a resolution of 8 cm−1; 32 scans per spectrum were accumulated. The second derivative of the spectra was computed using a Savitski–Golay function (seven points, third-degree polynomial function applied) implemented in omnic picta software (Thermo Fisher Scientific). The 2D or 3D absorbance map may be displayed accordingly to selected wavenumber, showing the intensity of the chosen band vibration.
Determination of spectrotype
With the aim of defining a plant spectrotype, IR spectra were recorded for 24 petals originating from separate flowers, in three biological replicates (72 individuals in total). For each petal, ten spectra were collected from equally distributed points on the upper part of the lamina, from the area ranging from 50 to 350 μm from the tip of the petal. A matrix of intensities (720 spectra × 1725 wavenumbers) was constructed on the basis of corrected and normalized spectra. Hotelling's T2 statistics and principal component analysis (PCA; pls-toolbox version 5.8.2; Eigenvector Research Inc., Manson, WA, USA) were used for detection of outliers. Six hundred spectra, 83% of the pool, were averaged to determine the spectrotype.
Cryo-scanning electron microscopy
Three petals from Col-0 plants were investigated by cryo-scanning electron microscopy directly without solvent treatment as flower wax does not contribute to the conical cell shape and nanoridge formation in petals (Bessire et al., 2011). For cryo-scanning electron microscopy, we used a Quorum system PP3000T (Quorum Technologies Ltd, Ringmer, East Sussex, UK) attached to a Quanta 250 FEI scanning electron microscope (FEI Company, Eindhoven, The Netherlands). The petals were mounted on aluminium stubs using a mixture of Tissue-Tek (Sakura Finetek Europe, Alphen aan den Rijn, The Netherlands) and colloidal graphite (Agar Scientific, Stansted, Essex, UK), frozen in nitrogen slush at −210°C and then transferred to the preparation chamber of the Quorum system. The sample was freeze-dried at −80°C for 10 min, and then sputter-coated with platinum at 10 mA for 25 sec. After transfer on the cryostage at −140°C in the scanning electron microscope, imaging was performed at 5 keV using an Everhart-Thornley electron detector (Hayles et al., 2006).
We are grateful to the Swiss Plant Science Web for making this research possible. We thank Herman Höfte and Gregory Mouille from the Laboratory of Cell Biology, INRA, Versailles, France as well as Frédérik Beisson (Department of Plant Biology and Environmental Microbiology, Cadarache CEA Research Center, France) for helpful and constructive discussions. Furthermore, we thank the Faculty of Biology and Medicine, University of Lausanne for generous support of the Electron Microscopy Facility. Willy Blanchard (Electron Microscopy Facility, University of Lausanne) is thanked for help with the preparation of the figures. We thank the European Arabidopsis Stock Center for supplying us with seeds of the SALK and SAIL lines, and Roger Innes (Department of Biology, University of Indiana, USA) for lacs2–4 seeds. C.N. was supported by the Swiss National Science Foundation (grant 31003A_125009) and the Herbette Foundation of the University of Lausanne. S.M. was supported by the Swiss Plant Science Web.