Wild-type (Y) and staygreen (G) lines of Lolium temulentum Ceres, near isogenic for an introgression from a senescence mutant of Festuca pratensis, were generated as described by Thomas et al. (1999). Plants were grown from seed on vermiculite, seven plants per 5 inch pot, in a controlled environment providing a constant 20°C and 8 : 16 h light : dark cycle (light flux 350 µmol m−2 s−1). Plants were fed with a nutrient solution as described by Gay & Hauck (1994).
Tissue was harvested from the fully expanded youngest (fourth) leaf 5–6 wk after seed sowing. Laminae were cut into 3 cm lengths and surface-sterilized by immersion for 5 min in 1.2% sodium hypochlorite solution containing four drops of Tween 20 per 250 ml followed by thorough rinsing with five changes of sterile water. Segments were incubated under aseptic conditions at 20°C in continuous darkness by floating them lower side down on water, 10 µm L-MDMP or 10 µm D-MDMP in 9-cm-diameter Petri dishes (usually seven segments per dish). The stereoisomers of MDMP (2-(4-methyl-2,6-dinitroanilino)-N-methyl-propionamide; Baxter et al., 1973) were chemically synthesized by Dr Richard Simmonds, Aberystwyth University. Freshly harvested tissue was used for 0 d samples. At 2, 4 and 6 d, control and MDMP-treated segments were rinsed with distilled water and, to avoid contributions from injured cells at cut surfaces, 2–3 mm of tissue at each end of each segment was discarded. Leaf segments were scanned for reflectance spectra as described in the following sections. Tissue for extraction was weighed, immediately frozen in liquid N2 and stored at −80°C for later analysis.
Recording and multivariate analysis of spectral data
Optical reflectance spectra of the leaf samples were obtained using an ImSpector direct sight imaging spectrograph (Spectral Imaging Ltd, Oulu, Finland). This is a passive optical device that, when interposed between a camera and a lens, permits acquisition of an individual spectrum at pixel resolution along a linear slice through the field of view. This results in an image in which the horizontal axis represents displacement along that line and the vertical axis corresponds to wavelength; each pixel in this image therefore represents the reflected intensity at a specific wavelength at a specific location on a line across the current sample. The ImSpector used was type V9, which has a nominal spectral range of 430–900 nm but this is subject to modification by the spectral characteristics of the camera used and the geometry of the ImSpector/camera combination.
For the present study, the ImSpector was attached to the C-mount of a Hitachi KP-M1EK monochrome CCD video camera with AGC (automatic gain control) disabled and gamma correction set to 1.0. The optical IR-cut filter fitted in front of the CCD in the camera was removed and this extended the spectral response from the normal cutoff of 700 nm towards the infrared. Wavelength calibration of the ImSpector/camera combination was effected with a mercury vapour lamp, aided by coloured LEDs to resolve possible ambiguity in identifying the mercury lines. This revealed that the effective spectral range was approx. 461–854 nm. A linear least-squares fit was applied to the relationship between pixel number on the spectral axis and mercury line wavelength (r2 = 0.999, with no improvement using a quadratic fit). This was used to provide the following empirical wavelength calibration, where wavelength is in nm and ‘pixel’ refers to pixel number on the spectral axis in the range 1–260:
Wavelength = (pixel × 1.51) + 461.65(Eqn 1)
Image capture was via a desktop personal computer fitted with a low-cost video capture card. Illumination was by means of two 12 V quartz-halogen lamps. Acquired images were in the form of raw binary files that were unpacked by means of a simple Perl script to provide ASCII decimal image files of 760 pixels in the spatial dimension and 260 pixels along the spectral axis. Lens aperture, exposure time and lighting were all kept constant throughout and data were recorded with nominal 8 bit intensity resolution, but maximum values were restricted to 7 bit resolution to avoid any possibility of saturation. Initial checks were made for the presence of any anomalous extra bright pixels in the leaf regions of the images (see later for definition) by calculating the ratio of mean to median reflectance, and as this only varied between 0.95 and 1.01, the absence of extra-bright pixels was confirmed.
Initial observations showed that there was a slight difference in illumination across the field of view (±1%), so to reduce systematic bias an average was taken over the illuminated leaf surface at each wavelength, containing 200 separate pixels on the spatial axis from the same part of the field of view for each sample. The ratio of the leaf value to that of the mean of two 60-pixel-wide regions on the spatial axis of the background paper on either side of the leaf (taken far enough away from the leaf to avoid direct shadows, and from fixed regions in the field of view) was then calculated, thus ensuring that the comparison between samples was unaffected by the slightly uneven illumination. Values were finally corrected to a known reflectance of substandard (Spectralon®, Labsphere Inc., North Sutton, NH, USA), which had previously been compared with the background paper. This procedure also accounted for the changes in the relative sensitivity of the ImSpector/camera combination with wavelength. For each group of five leaves, single spectra were produced for their upper and lower surfaces, and named according to genotype, days of senescence, chemical treatment and whether the upper or lower leaf surface was scanned. For tissue incubated on water or MDMP for 2, 4, 6 and 8 d, this gave four time points × two surfaces × two genotypes × two treatments = 32 data objects. For 0 d (unsenesced) tissue, observations were made on two surfaces × two genotypes = four data objects, making a total of 36 data objects, each with 260 spectral variables.
To investigate whether the spectra were able to distinguish the different groups, principal-components analysis (PCA, Hotelling, 1933) was applied to the set of 36 spectra, after scaling values at each wavelength to zero mean and unit variance, using MATLAB 7 (The MathWorks Ltd, Natick, MA, USA). The first six principal components were found to account for 70.36, 10.77, 5.45, 2.72, 1.54 and 1.34%, respectively, of the variance in the data.
Leaf colour was quantified by pixel analysis of high-resolution JPEG images of leaf segments. RGB (red-green-blue) profiles across a montage of segments were obtained using the ‘improfile’ tool in MATLAB.
Protein extraction and electrophoresis
Frozen leaf material was ground to a fine powder with liquid N2 and sand, in a mortar and pestle. The powder was homogenized with 5 ml g−1 FW extraction buffer (EB) comprising 0.2 mm tris, pH 8.0, containing 5% (v/v) glycerol, 1% (v/v) 2-mercaptoethanol, 1 mm PMSF and 1 mm monoiodoacetate, and allowed to thaw before adding 20% (w/v) lithium dodecyl sulphate (LiDS; 0.5 ml g−1 FW) and homogenizing further. The homogenate was heat-denatured by boiling for 2 min, cooled and centrifuged at 12 500 gav for 10 min. Protein in the supernatant was determined according to Lowry et al. (1951). Total LiDS-EB-extracted proteins were fractionated by 12.5% SDS polyacrylamide gel electrophoresis (SDS-PAGE; Laemmli, 1970) using a Bio-Rad mini-gel system. Aliquots of extracts were loaded on the basis of equal weight of leaf tissue. Gels were stained with Coomassie Brilliant Blue.
High-performance liquid chromatography (HPLC) of leaf pigments
Leaf segments were frozen in liquid nitrogen and ground to a fine powder in a pestle and mortar with quartz sand. The pigments were extracted with 80% acetone (1.5 ml per 100 mg tissue) and, after centrifugation at 10 000 gav for 10 min at 4°C, chlorophylls and related pigments were determined by HPLC (Roca et al., 2004). Pigments were identified from their spectral absorption maxima and peak ratios and by HPLC cochromatography with authentic samples. Identification of 132 hydroxy chlorophyllide a is tentative, based on spectral absorption maxima and peak ratios. Online UV-visible spectra were recorded from 350–700 nm with a photodiode array detector. HPLC was performed on a Nova-Pak C18 4 µm Radial-Pak cartridge 8 mm × 100 mm column using Waters 515 HPLC pumps and a Waters model 996 photodiode array detector. The manual injection valve (Rheodyne, model 7725I) was fitted with a 20 µl loop. Separation was carried out using an elution gradient (2 ml min−1) with the mobile phases (A) ion pair reagent/methanol (1 : 4 v/v) and (B) acetone/methanol (1 : 4 v/v). The ion pair reagent was 1 m ammonium acetate in water (Langmeier et al., 1993). The gradient was isocratic A 4 min, A to B 5 min, isocratic B 9 min, return to A 2 min (Siefermann-Harms, 1987), and detection was at 660 nm. Analytically pure samples of chl a and chl b were used to obtain the calibration slopes representing the area of the peak obtained with different injected volumes of pure solutions of known concentration. The same preparation was acidified and used for calibration with regard to phaeophorbides (phd). For quantification of chlorophyllides (chld) it was assumed that dephytylation does not change the spectral properties of the porphyrin moiety (Langmeier et al., 1993).