Plant materials
Three rice (Oryza sativa) lines, namely, IR64 (IRGC #66970), Kinandang Patong (KP, IRGC #23364), and Dro1-NIL [38], were used in this study. KP has deep, thick roots, while IR64 has shallow, thin roots. Further, Dro1-NIL is a near-isogenic line of IR64 with a KP-type DRO1 allele, having intermediate, thin roots.
Growth conditions
We used Profile® Greens Grade™ (PROFILE Products, Buffalo, Illinois, USA) as a soil-like root growth substrate. Profile is calcined clay and has characteristics similar to those of Turface® (PROFILE Products, Buffalo, Illinois, USA) [50]. Turface is a popular substrate for root imaging using X-ray CT [45, 51]. Profile has the following advantages in studying plant roots: (1) it is hard and its volume is not affected by water content, maintaining a consistent RSA in the soil under both dry and well-watered conditions; (2) it is easily removed from the root surface, and root samples are simple to collect; (3) it can retain sufficient water and nutrients for plant growth.
Before use, Profile was rinsed with tap water three to five times and dried, as it has a quite variable labile (readily desorbable or readily plant-available) nutrient content [50]. The dried Profile was packed in the target pot, and saturated with a modified Kimura B hydroponic solution [52] consisting of 1.23 mM of NO3-, 0.41 mM of NH4+, 0.18 mM of H2PO4-, 1.00 mM of SO42-, 1.78 mM of K+, 0.55 mM of Mg2+, 0.37 mM of Ca2+, and 8.9 μM of Fe3+. The pH was adjusted to 5.5, using HCl and KOH, as required. The ratio of nitrate to ammonium was based on the ratio of these compounds in the soil collected from an upland field in the Institute of Crop Science (National Agriculture and Food Research Organization, Ibaraki, Japan; 36°02'89'' N and 140°09'97'' E) on 2018 May 24, which has typically been used in the characterization of rice plant roots [53]. Except where indicated, we used custom-made, 20 cm diameter, 25 cm high pots (TSP2530P, Tecs, Itako, Ibaraki, Japan). Rice seeds were immersed in water for one d, at 15 ℃ and with a fungicide, and then in water for two d, at 30 ℃. The germinated seeds were sown in the pot. Deionized water was supplied to the bottom of the pots during cultivation.
Rice plants were grown in a custom-made growth chamber (Nippon Medical & Chemical Instruments Co., Tennoji-ku, Osaka, Japan), whose temperature, light intensity, and humidity were strictly controlled. The light condition applied was to simulate 14 h of daylight, and the light intensity was set as a photosynthetic photon flux density (PPFD) of approximately 500 μmol m−2 s−1 at the top of the pot, with the exception that a PPFD of 250 μmol m−2 s−1 was applied to simulate dawn and dusk. The diurnal light intensity program was as follows: ZT0, PPFD of 250 μmol m−2 s−1; ZT1, PPFD of 500 μmol m−2 s−1; ZT13, PPFD of 250 μmol m−2 s−1; and ZT14, PPFD of 0 μmol m−2 s−1. The diurnal temperature program was based on the average diurnal temperatures for Tsukuba in July 2017, which were obtained from the Weather Data Acquisition System of the Institute for Agro-Environmental Sciences, NARO. The diurnal program was as follows: ZT0, 25 ℃; ZT2, 26 ℃; ZT3, 27 ℃; ZT4, 28 ℃; ZT5, 29 ℃; ZT6, 30 ℃; ZT12, 29 ℃; ZT13, 28 ℃; ZT14, 27 ℃; ZT15, 26 ℃; and ZT16, 25 ℃. Humidity was set to 50% in light conditions, and 60% in dark conditions. To maintain the CO2 concentration, some of the air in the chamber was periodically exchanged with outside air, due to the large size of the chamber (86.4 m3), and CO2 concentration during cultivation ranged between 400–500 ppm.
Soil chemical analysis
The ammonium and nitrate concentrations of the field soil were quantified by the indophenol and colorimetric methods respectively, based on diazotization with nitrate reduction. Four samples from different locations in the field were subjected to ammonium and nitrate quantification using a commercial service (Katakura & Co-op Agri Corporation, Chiyoda-ku, Tokyo, Japan).
X-ray CT imaging and 3-D reconstruction
Rice roots were imaged in the soil using the X-ray CT system inspeXio SMX-225CT FPD HR (Shimadzu Corporation, Nakagyo-ku, Kyoto, Japan). Except where indicated, each scan digitally obtained 1200 projections, using a signal averaging two frames over 360° without binning (pixel detector resolution: 3000 × 3000), at 4.0 fps. 860 horizontal, 1024 × 1024 pixel resolution slices were computed. The final spatial resolution was 300 μm, corresponding to a total volume of 30.72 × 30.72 × 25.8 cm3. Tube voltages of 150, 175, 200, and 225 kV, and tube currents of 100, 200, 300, 400, and 500 μA were used. Source–detector and source–rotation axis distances were 1200 mm and 900 mm, respectively. Three Cu filters, 0.5, 1.0, and 2.0 cm thick, were used to harden the X-ray beam. Beam hardening was approximately corrected by the operating software with a correction table calculated using the correct metal material.
Image Processing
In our study, CT images were processed using python script; the slices were loaded as 16-bit grayscale data and stacked as 3-D NumPy arrays [54]. Grayscale histograms for the CT stacks showed bimodal distributions, with one mode coming from the air fraction and the other from the soil fraction—and we roughly normalized these two modes as zero and 1024, respectively, with negative values rounded to zero. Outcomes from this were derived as follows: (1) The grayscale range for each scan was normalized, as the window width and level changed with each scan, particularly under high-power scanning conditions; (2) the normalization result was affected by the type of soil substrate, and if only one soil substrate was used, the grayscale ranges of the normalized CT volumes would all be the same; (3) the local soil fraction grayscales for single CT volumes should be different, in our study, as we used the modes for normalization. This was solved using the edge detection process described later, and allowed the soil fraction grayscales to be converted to ~ zero.
The normalized CT volume was filtered with a 3-D median, and inverted. For each slice, its blurred slice was subtracted, to isolate roots as edges. Negative values were rounded to zero, and to remove small objects, thresholding and 3-D size opening were applied to the computed 3-D volume. The processed 3-D volumes were saved as 8-bit, grayscale, 2-D images. The median, blur, and size opening filters were imported from scipy [55], opencv [56], and skimage [57] modules, respectively. Parallel processing was performed using a multiprocessing package [58], and the 25 cm high / 18 cm diameter, cylinder-shaped area was trimmed to remove roots touching the pot wall. The source code for this algorithm is available in Additional file 5.
Evaluating root-to-soil contrast
Root-to-soil contrast was evaluated using the CNR, which was defined as shown in Eq. (1):
[Please see the supplementary files section to access the equation.] |
(1)
|
Visualization of CT slices and 3-D volume
CT slices were visualized using either VG Studio MAX 3.2 software (Volume Graphics, Heidelberg, Germany), or python scripts with NumPy and matplotlib [59] modules. The 2-D image rendered from 3-D volume data was obtained using VG Studio MAX. Maximum intensity projection was obtained using python scripts, and the 3-D movies were constructed with VG Studio MAX.
Scanning time to reduce X-ray doses
Scanning time was determined by the projection number, signal averaging number, exposing time, and binning size. In a scanning time of 10 min, each scan digitally obtained 1200 projections, using a signal averaging two frames over 360°, without binning (pixel detector resolution: 3000 × 3000), at 4.0 fps. At the fastest scanning speed, each scan digitally obtained 600 projections, using no signal averaging over 360°, with 3 × 3 binning (pixel detector resolution: 1000 × 1000), at 18.0 fps. Under these conditions, scanning was completed in 33 s. Eight scanning parameter combinations were considered, consisting of either 600 or 1200 projections, using no signal averaging or a signal averaging of two frames over 360°, with or without binning.
Metal filters to reduce X-ray doses
Metal filters were used to reduce the X-ray doses applied to plant materials. Because longer wavelength X-rays have lower energy, their penetration ability is lower, and are thus absorbed at the material surface, resulting in higher X-ray dosage being needed there. To reduce the X-ray doses to plants, 0.5, 1.0, and 2.0 mm Cu filters were tested.
Growth test against X-ray exposure
KP was cultivated for two weeks and subjected to daily CT scanning for seven d. For X-ray treatment, KP was exposed to a 10 min X-ray scan, receiving approximately 0.09 Gy per scan. For mock treatment, KP was loaded onto the CT machine turntable, where it was left for 10 min without X-ray exposure. The positions of all plants in the growth chamber were shifted daily, to reduce the influence of position on growth. At 21 DAS, shoot and root traits were quantified: plant height was measured using a ruler, and shoot samples were collected. The shoot samples were dried, at 80 ℃ for three d, to facilitate dry weight measurement. Root samples were collected from the soil and scanned using a 400-dpi scanner (Expression 12000XL, Seiko Epson Corporation, Suwa, Nagano, Japan). Total root length was measured using WinRHIZOTM Pro 2017a software (Regent Instruments, Quebec, Canada). The root samples were dried at 80 ℃ for three d, to facilitate dry weight measurement.
Wired mesh basket method
The basket method was used in evaluating rice cultivar rooting angles, by counting the proportion of roots penetrating the bottom of the basket [60, 34]. As a modified approach, we used baskets (diameter: 15 cm, height: 6 cm), whose insides were woven into a mesh using φ 0.148 mm nylon monofilament, to keep plant RSA in situ when they were unearthed from the ground and the soil was removed. Two wire layers were included, at 1 cm and 3 cm under the top of the basket, with each wire layer consisting of seven parallel and seven perpendicular wires, spaced at intervals of ~ 1.5 cm, vertically and horizontally. The wired mesh basket was buried in the pot, and germinated rice seeds were sown on the top of the basket.
After cultivating for three weeks, the roots in the soil were scanned using an X-ray CT scanner, and the wired basket, including its plant and associated soil, was removed from the pot. All excavations were performed in water to avoid damaging the roots, and the soil was then removed through the basket mesh. The plant was shot from directly above, using a digital camera (Xperia Z5 Compact, Sony Corporation, Shinagawa-ku, Tokyo, Japan). To enhance root colors, the chroma of the blue and cyan colors—which were the colors of the basket, were adjusted using GIMP software (version 2.8.22, https://www.gimp.org/). By comparing the CT image and the picture of the wired basket method, we categorized all crown roots and radicles as being either detectable or undetectable by X-ray CT. The radicle and the crown roots were cut at approximately 2 cm and 5 cm from the shoot–root junction, and the resulting 3 cm root fragments were sampled. Each fragment was scanned with a 600 dpi scanner (Expression 12000XL). The average root diameter for each fragment was then calculated, using ImageJ plug-in, SmartRoot (version 4.21, [61]).
Statistical analysis
Student’s t test was performed with the “t.test()” function in the R software (version 3.5.1), and a P value < 0.05 was considered significant.