A nondestructive method to estimate the chlorophyll content of Arabidopsis seedlings

Chlorophyll content decreases in plants under stress conditions, therefore it is used commonly as an indicator of plant health. Arabidopsis thaliana offers a convenient and fast way to test physiological phenotypes of mutations and treatments. However, chlorophyll measurements with conventional solvent extraction are not applicable to Arabidopsis leaves due to their small size, especially when grown on culture dishes. We provide a nondestructive method for chlorophyll measurement whereby the red, green and blue (RGB) values of a color leaf image is used to estimate the chlorophyll content from Arabidopsis leaves. The method accommodates different profiles of digital cameras by incorporating the ColorChecker chart to make the digital negative profiles, to adjust the white balance, and to calibrate the exposure rate differences caused by the environment so that this method is applicable in any environment. We chose an exponential function model to estimate chlorophyll content from the RGB values, and fitted the model parameters with physical measurements of chlorophyll contents. As proof of utility, this method was used to estimate chlorophyll content of G protein mutants grown on different sugar to nitrogen ratios. This method is a simple, fast, inexpensive, and nondestructive estimation of chlorophyll content of Arabidopsis seedlings. This method lead to the discovery that G proteins are important in sensing the C/N balance to control chlorophyll content in Arabidopsis.


Background
The chlorophyll content of leaves is an indirect indicator of the health and nutritional status of the plant [1]. Traditional methods to calculate the chlorophyll content include a destructive chemical extraction and a nondestructive measurement of chlorophyll fluorescence. The former method, while direct, is tedious and unsuitable for continuous monitoring individual plants because of its destructive manner. The latter method needs expensive instruments of which none are presently suitable for small leaves such as the commonly used Arabidopsis cotyledons. It is important to develop a non-destructive method to estimate chlorophyll content for Arabidopsis because it is a genetic model plant, however traditional chlorophyll extraction is not useful due to the small size of the Arabidopsis leaves grown on agar plates. Recently, digital photographic imaging showed great promise for quantitating plant phenotypes [2]. Indirect methods are available but none are yet suitable for Arabidopsis. Sass et al. [3] developed a protocol to convert the RGB values of a color image into a hue saturation value (HSV), and showed that the hue value was correlated to the chlorophyll content estimated by a destructive method. A similar color-image method was used to assess the nitrogen status of rice under natural light condition [4]. Riccardi et al. [5] found that an exponential function model displays the best correlation between the RGB values and the chlorophyll content through single and multiple regression in quinoa and amaranth leaves. No similar color-image methods have been adapted for Arabidopsis chlorophyll content, in particular Arabidopsis seedlings grown on agar plates. The lack of a quantitative method for measuring chlorophyll of plate-grown Arabidopsis restricted previous studies on stress-induced phenotypes to subjective assessment without quantitation [6,7].
In this study, we describe a convenient and nondestructive method to estimate leaf chlorophyll of Arabidopsis seedlings grown on agar plates using calibrated-RGB images. We also provide instructions how to adapt it to other small leave samples. We quantitated chlorophyll content in small Arabidopsis seedlings grown on different C: N ratios in the agar medium. The results indicated that G proteins play important roles in sensing and/ or responding to the C/N balance in this chlorophyll response.

Note 2:
It is not necessary to use a square petri dish on which to arrange the seedlings; any rectangle background with samples in a matrix format will work. It is also possible to treat seedlings in liquid culture or on some matrix other than agar and then transfer them to the square agar plates for photography.
Note 3: For this study, seedlings were grown on the indicated media arranged on square plates as described above and photographed as will be described below. Note 5: Acquire images of the X-rite ColorChecker classic chart both immediately before and after acquiring images of the samples (Fig. 1a). The Color-Checker chart is to make sure the final data comparable despite different light conditions and cameras but may not be necessary. Before and after images of the chart are acquired to test whether the light condition is consistent during the photographing. Check the RGB values of the X-rite Colorchecker chart boards as described below. Should you find that the starting and ending values are different, it will be necessary to stabilize the light environment and re-acquire sample images. If the light condition is stable in the lab, it is not necessary to acquire two images every time. It is important to use the same settings and the same light conditions for all the images to be compared.

Note 6:
The image size for the plate is the same as for the ColorChecker chart, approximately 210 × 300 mm. The square plate is placed in the center of this field for imaging.
Note 7: The light should be uniformly distributed.
Tests for position effect in this field were determined to be a maximum of 4.8% (n = 36) (Additional file 5: S5) of the chlorophyll values. This value was calcu-had saved in the ImageJ 'plugins folder. A dialog box will appear asking for the number of rows and columns (Fig. 1c). Enter these values or use the default value of 6 rows × 6 columns. This step divides the images into 36 parts with 6 rows × 6 columns. The dialog box also will request the RGB values of the Enter these values. By clicking "OK", you will generate a table of the chlorophyll content as ng/mm 2 . The dialogue also asks if you want to make a normalized grayscale image (Fig. 1d). If so, click the box and a grey scale image will also be presented as output (see Additional file 7: S7 for an example). This greyscale image offer a spatial map of chlorophyll on a leaf; i.e. 2-dimensional information is provided.

Chlorophyll extraction
For validation purposes, we compared our method to extracted chlorophyll. Chlorophyll content was estimated by spectrophotometry of samples prepared by 80% acetone extraction. The leaves were incubated at room temperature in a 1.5-mL tube with 1 mL 80% acetone solution for at least 24 h then clarified by centrifugation for 5 min at 15,000g. In this study, absorbance of the supernatant was measured at wavelengths 645, 646, and 663 nm (A 645 , A 646 , and A 663 ) with a Shimadzu UV-3000 ™ dual-wavelength, double-beam spectrophotometer, although any spectrophotometer is suitable. Complete spectra were taken during development of this protocol in order to assure that the predominant absorbance was from chlorophyll; this is not routinely necessary. Samples having absorbance greater than 1 were diluted by half with 80% acetone and re-evaluated. Chlorophyll concentration was estimated following the Lichtenthaler's equations (A) [23] and the Arnon's equations (B) [24] as follows: A.
(1) The area of the leaves were measured by software Image J and chlorophyll content and the total chlorophyll content per leaf area was expressed as ng/mm 2 .

Fitting parameters for the function and "Do it yourself" validation for other types of chlorophyll containing samples
The default coefficient values used above are described here in the event that the user needs to modify this tool to obtain different coefficients for other types of chlorophyll samples such as leaf pieces. A least squares method was used to search for the coefficients for the exponential function equation to estimate chlorophyll contents from RGB values. If the user desires to validate this method on their own using their own images and chlorophyll samples, then follow these steps: Chl = EXP a1 * R * r/243 + a2 * G * g/243 + a3 * B * b/242 + a4 Having the new coefficients enables you to estimate chlorophyll in other samples nondestructively.

From RGB value to chlorophyll content
We adapted the method of Riccardi et al. [5] to Arabidopsis seedlings grown on culture plates and compared this method to biochemical extraction of chlorophyll. We extracted chlorophylls with acetone, measured absorbance spectra, then calculated chlorophyll content in the solvent extract using both the Lichtenthaler's [23] and Arnon's equations [24] (Eqs. 1, 2). Samples from 12-day-old Arabidopsis shoots from seedlings grown under different C/N treatments on square plates as described under plant growth were used to search for the parameters for the exponential function model (see Additional file 11: S11). To assure that the images taken by different cameras are comparable, we incorporated a standard to enable comparison of published data. The X-rite ColorChecker chart (Fig. 1a) was used to make a DNG profile and to adjust the white balance. Another critical variable to account for is the light intensity and color in the room, chamber, greenhouse or field. The X-rite ColorChecker chart solves these problems and makes the assay applicable to artificial and natural light. The software Image J was used to measure average R, G and B values of individual leaves. We used the exponential function model to estimate the chlorophyll content where the (r i , g i , b i ) represents the R, G or B value for each sample where i accounts for the sample index [5]. We used a biochemical extraction with Lichtenthaler's (Eq. 1) equations to measure the chlorophyll content and fitted coefficients for the equation. We took samples from 4 independent experiments (n = 234 samples; Additional file 11: S11) and determined the coefficients: a 1 through a 4 in the equation (Eq. 4) using the least squares method in the MAT-LAB environment (Fig. 2a, b).

(4)
Chl i = e a 1 r i +a 2 g i +a 3 b i +a 4

Robustness of the default parameters
Our method is optimized for Arabidopsis seedlings grown on agar plates, a common format for Arabidopsis researchers. Since the default parameter values were generated using chlorophyll extracted seedlings grown under one light condition, a concern is how well these values apply to other growth conditions. To test this, we compared chlorophyll estimation in two extreme growth conditions. In our lab, the thickness of 4-7 week-old leaves grown on soil under a long-day condition is almost twice as seedling leaves grown on plates under constant low light (176 vs. 100 µm, respectively). We compared chlorophyll estimation of these thicker leaf pieces using the default parameters to refitted parameters. The thicker leaf pieces were imaged as described above and the extracted chlorophyll used to refit the data to generate new parameters: The optimized parameters increased the R 2 correlation coefficient only from 0.85 to 0.89 (cf. Fig. 2c, d). This indicates that the default values are robust with regard to different growth conditions that may affect leaf thickness. Nonetheless, when extreme accuracy is required, we recommend calculating the parameters by the "Do it yourself " fitting method described above.

Proof of utility: chlorophyll content of Arabidopsis seedlings grown under different C/N ratios
In order to determine how plants respond to different C/N ratios, we tested six different glucose concentrations (0, 1, 2, 4, 5 and 6% d-glucose) and five nitrogen concentrations (0.1, 0.3, 0.5, 2 and 6 mM KNO 3 ) in a matrix format. Figure 3 shows plant growth under these different C/N ratios and Table 1 shows the plants area in response to different C/N ratios. Plant area did not change in response to nitrogen under 0% glucose, although the average leaf area changed. As quantitated in Additional file 12: S12A, plant area at 0% glucose varied greatly but was statistically unchanged. When the medium contains glucose, plant area increased with increasing nitrogen. The optimal glucose concentration at the highest nitrogen concentration at 3 mM was 1-2% (Additional file 12: S12 panel A).
The chlorophyll content correlated with the nitrogen concentration in the presence of carbon supply. We estimated the chlorophyll content at different C/N ratio. As  Table 2, the chlorophyll content of the Col-0 seedlings grown on 0% glucose did not change with increasing nitrogen in the medium. All the plates were grown under continuous dim light (35-50 µEm −2 s −1 ) at 23 °C, which decreases the photosynthesis. However, even a slight amount of glucose dramatically changed this relationship. For example, in the presence of 1% glucose, the chlorophyll content increased in response to the nitrogen concentration, and slightly decreased when the nitrogen concentration was raised further. This indicates that the chlorophyll content of the leaves is slightly influenced by nitrogen concentration but highly influenced by the carbon availability. Also, the chlorophyll content increased as the glucose concentration increased; the seedlings grown with 4% or 5% d-glucose had dark green leaves. Based on plant area, optimal growth was at 1 and 2% glucose with 6 mM nitrogen. This condition did not produce the highest chlorophyll content because the seedlings were stressed. The highest concentration of 6% glucose, with 0.1 mM nitrogen, stressed seedlings further, as was apparent by the red color of leaves caused by excessive anthocyanin pigments.

Function of G protein signaling pathway in C/N sensing
In order to analyze whether G protein signaling pathway is important for C/N sensing/responsiveness, the null mutants of the G protein under various C/N condition were assayed. The leaf area differences between the mutants of rgs1-2, gpa1-3 and agb1-2 is not uniform under different C/N conditions (Additional file 13: S13). For example, agb1-2 mutants grown under limited

Table 1 Plant size (mm 2 ) in response to glucose and nitrogen treatment
Growth condition and treatments are as described for Fig. 3. Data analysis is performed by software SAS8.0. Single factor analysis (n = 12). Capital letters represent similarity groups (p > 0.05) among glucose treatment and lowercase letters represent similarity groups among nitrogen treatments (p > 0.05)

Glucose (w/v) (%)
Nitrogen   nitrogen (0.1 mM) are slightly larger than the wild type under 1 and 4% glucose but no difference under 2% glucose; when the glucose increased to 5 and 6%, the plant size of agb1-2 is smaller than wild type. We estimated the chlorophyll content in the G protein mutants. The optimal concentration for the growth of Arabidopsis grown on agar is 1-2% glucose (Additional file 12: S12B). The G protein complex is involve in glucose sensing [25] and AtRGS1 is a component of a glucose sensor [11,12,18]. Previous studies showed that the rgs1-2 null mutant is tolerant to high glucose levels [7], however that report used a semi-quantitative method, the so-called "green-seedling" assay. In order to compare the present quantitative results to the published semi-quantitative results, we established a threshold value of 15 ng/mm 2 to distinguish yellow seedlings from green seedlings (Additional file 12: S12C; the threshold value was from inner fence of agb1-2 mutants' box plot). Taking rgs1-2 mutants as an example (Fig. 4a), the proportion of green leaves was fourfold greater than wild type when the nitrogen concentration was 0.1 mM (0.22 and 0.05 for rgs1-2 and Col, respectively). When the nitrogen concentration was increased to 1 mM, the proportion of green leaves increased to 0.98 while the wild type was 0.83. Consistent with the results of Chen and Jones [7], the rgs1-2 mutants showed more than 90% green seedlings versus wild type seedlings which scored less than 40%. In addition, as previously observed, the agb1-2 mutants contain less chlorophyll than wild type at high glucose and low nitrogen growth conditions.
We also examined the chlorophyll content of the G protein mutants under moderate glucose and high nitrogen (4% glucose, 6 mM nitrogen) stress. Interestingly, as shown in Fig. 4b, the chlorophyll content of the xlg and agb1 mutants were significantly higher compared to the wild type under slight stress, whereas the rgs1-2 mutant was not significantly different from wild type.

Conclusion
This method offer a non-destructive, sensitive, and quantitative way to estimate the chlorophyll content of Arabidopsis small seedlings grown on agar plates, and it is an effective way to evaluate the growth condition of the plants. This method provides a quantitative alternative to the qualitative 'green' versus not 'green' seedling assay [26,27]. The use of the X-rite Color-Checker chart eliminates differences between cameras and environment, however, caution should be applied in comparing images from highly different environments. Although this method is dependent on the area of the seedlings, rather than the volume, it still provides a good estimate of the relative chlorophyll content for the Arabidopsis.