RIPPS: A Plant Phenotyping System for Quantitative Evaluation of Growth Under Controlled Environmental Stress Conditions.

High-throughput and accurate measurements of plant traits facilitate identification of gene function. Along with recent advances in quantitative genomics, there is a growing need for precise quantification of multiple traits in plants. However, it is difficult continuously to quantify plant adaptive responses to environmental stress responses such as drought because multiple environmental factors are intricately involved in the phenotype. To solve this problem, we developed an automatic phenotyping system for evaluating the growth responses of individual Arabidopsis plants to a wide range of environmental conditions. The RIKEN Integrated Plant Phenotyping System (RIPPS) controls soil moisture for single plants by automatically weighing and watering 120 continuously rotating pots under controlled light, humidity and temperature growth conditions. RIPPS also records individual rosette size and expansion rate by photographing plants every 2 h. We used RIPPS to establish phenotype evaluation methods for Arabidopsis growth response and water use efficiency under various water conditions, and analyzed the involvement of ABA metabolism in determining water use efficiency. We also used RIPPS to analyze salinity tolerance in Arabidopsis plants.

Introduction

Plants have developed highly flexible adaptive mechanisms to cope with environmental changes, which span multiple organizational levels from molecular to morphological (Fujita et�al. 2006, Yamaguchi-Shinozaki and Shinozaki 2006, Skirycz and Inze 2010, Zhu 2016). Plant adaptive responses can vary depending on the developmental stage and target organs (Granier and Tardieu 2009). To deepen our understanding of the complex mechanisms underlying plant responses to environmental stimuli, it is important to measure plant growth responses quantitatively under different environmental conditions by accurately controlling growth conditions, regulating the timing and strength of the adverse environmental stimuli, and conducting time-course observations of multiple parameters of plant growth. Functional gene analyses can be accelerated by analyzing mutants and natural genetic variants, and performing high-throughput genetic mapping and large-scale analysis of gene expression pathways, metabolite levels or plant hormones levels (Koornneef et�al. 2004, Oksman-Caldentey and Saito 2005, Weigel 2012, Assmann 2013). Quantitative genomics can be facilitated by improving phenotyping accuracy and throughput.

Several phenotyping systems have been established recently at different organizational scales (Granier and Vile 2014, Humplik et�al. 2015, Tardieu et�al. 2017). Automated phenotyping systems provide powerful tools for plant research by employing large-scale, high-throughput, non-destructive, real-time phenotype acquisition methods and measurements (Granier et�al. 2006, Lefebvre et�al. 2009, Tisne et�al. 2013, Clauw et�al. 2015). To quantify plant growth responses to environmental stresses, we developed a new automated phenotyping system named the RIKEN Integrated Plant Phenotyping System (RIPPS). RIPPS is an integrated platform that enables us simultaneously to cultivate 120 individual plants in soil pots and automatically to analyze time-dependent changes in plant growth. We utilized the advantages of existing automated platforms such as PHENOPSIS, PHENOSCOPE or WIWAM (Granier et�al. 2006, Skirycz et�al. 2011, Tisne et�al. 2013, Clauw et�al. 2015), e.g. an automated weighing and watering station and contiguous rotation of plants during the course of the experiment to develop RIPPS. Two major benefits are conferred by the automated rotation: (i) the micro-environmental conditions experienced by individual plants are homogenized, and (ii) additional imaging devices can be easily set to any position. RIPPS captures digital images of individual plants in the daytime and during the night-time dark conditions using infrared light-emitting diode (LED) light (peak wavelength 950 nm). Time-dependent changes in plant size and detailed analyses of temporal and spatial effects can be derived and quantified from the time-lapse growth images. The RIPPS phenotyping platform is operated under precise control of growth parameters including temperature, light, nutrient availability and water conditions. We used RIPPS to establish quantitative evaluation methods for determining growth responses and water use efficiencies of Arabidopsis plants under different water conditions. We also analyzed the involvement of ABA metabolism in water use efficiency (WUE) using RIPPS. These proof-of-concept studies demonstrate that RIPPS accurately facilitates the measurement and evaluation of plant growth responses to precisely controlled hydric environments.

Results and Discussion

Construction of the RIPPS phenotyping platform

The RIPPS automated phenotyping platform contains a belt conveyor with 120 pot trays, an irrigating station and imaging cameras in a steel structure (Fig.�1A−C). It includes a lighting device with a dimmer control, which work together with the cameras. The system continually rotates up to 120 pots in a stepwise movement, and the pots stopped on the irrigating station are automatically weighed and watered (Fig.�1B). One step takes approximately 1 min, and one full circuit takes approximately 2 h. All functions are executed by the operating program on a personal computer (Fig.�1D). For water deficit experiments, three irrigation mode patterns can be selected to design different experimental conditions, including the timing of water withholding or re-watering, and the soil water content (SWC) (Supplementary Fig. S1). The amount of water irrigated can be controlled in units of 0.05 g, which enables precise adjustment of the SWC in individual pots every 2 h. The watering frequency can be freely set by increasing the minimum irrigation amount so as to skip a small amount of watering. Each camera is mounted on a flexible arm of a movable pole on the outer frame of the RIPPS platform, so the camera position is easily adjusted. In this study, we used two color cameras to capture top and side view images under light conditions, and a black and white camera for night imaging under dark conditions (Fig.�1E−H). Night imaging was conducted with a 950 nm LED light, which is reported to have only minimal effects on plant growth under dark conditions ( Xie et�al. 2007, Takano et�al. 2009) (Fig.�1F).

Fig. 1

Overview of RIPPS. (A) The RIPPS platform: up to 120 plants are continuously rotated and automatically weighed, watered and photographed. (B) Weighing and watering station. When a pot is placed in the weighing position, the balance is lifted and weighs the pot together with the conveying tray. (C) Digital camera and LED light (peak wavelength 950 nm). (D) Flowchart of the system control software. (E) Top-view color image. (F) Night-time image using 950 nm LED light. (G, H) Side view images in the morning (G) and evening (H). Scale bar = 1 cm.

RIPPS achieves high positioning accuracy with a small position shift of <0.5 mm. This enables the RIPPS platform to capture time-lapse movies of sequential plant growth without any adjustments or corrections (Supplementary Movies S1−S3). Side view imaging can capture the timing of flowering and leaf hyponastic movement (Fig.�1G, H;Supplementary Movie S3). RIPPS takes images of each pot every 2 h; this time resolution is higher than other phenotyping systems. The high time resolution is suitable for analysis of fast growth responses, such as hyponastic movement affected by environmental light conditions (Supplementary Movie S3). The rotating system allows extension of cameras or analyzing devices. Projected leaf area can be analyzed using a customized leaf segmentation algorithm based on previously reported software (Tanabata et�al. 2010) (Supplementary Fig. S2).

Continuous pot rotation homogenizes plant growth conditions

During plant growth experiments, pot position is a significant factor that affects the results due to slight differences in growth conditions (e.g. light, air flow and temperature) for each plant. This problem is avoided with the RIPPS platform because continuous rotation effectively homogenizes the micro-environmental conditions (Tisne et�al. 2013). The RIPPS platform includes a belt conveyor system for pot rotation. To determine the effect of pot rotation on micro-environmental conditions, we compared plant growth on RIPPS with or without rotation. Plants were grown for 8 d under the same conditions, separated into two groups for fixed-position or rotated experiments, and then grown for 2 weeks under well-watered conditions (3.5 g water g–1 dry soil) (Fig.�2A, B). In the fixed-position and rotated experiments, pots were not rotated or subjected to continuous rotations, respectively, and plants were watered once a day. Plants were weighed at the end of the experiment. Fig.�2C presents the spatial distribution of plant growth rates. Plant sizes were more variable in the fixed-position experiment than in the rotated experiment (Fig.�2C;Table�1). We also compared the uniformity of plant growth under mild drought conditions (Fig.�2D;Table�1). SWC was kept at a low level (1.25 g water g–1 dry soil) (Fig.�2A, B). Under mild drought conditions, we confirmed that pot rotation homogenized the micro-environmental conditions and resulted in uniform plant growth (Fig.�2D;Table�1). The RIPPS system is operated in a growth chamber to control conditions including temperature, humidity and light. Therefore, the plants are subjected to uniform and controlled growth conditions for precise plant phenotyping (Supplementary Fig. S3).

Fig. 2

Effect of pot rotation on plant growth. (A) Transition of soil water content (SWC) of pots under well-watered (blue line) or mild-dry (red line) conditions. (B) Representative plant images at the final point. Scale bar = 1 cm. (C, D) Schematic presentation of plant growth distribution under well-watered (C) or mild-dry (D) conditions. Plants were grown for 2 weeks in pots with (lower panel) or without (upper panel) continuous rotation. Each pot on RIPPS was designated with specific colors indicating the range of plant fresh weights.

Table 1

Average and SD of fresh weight of plants grown under the indicated conditions (n = 60)

		Well watered		Mild dry
		Average FW (mg)	SD	Average FW (mg)	SD
Exp1	Fixed	112.4	22.2	87.7	14.3
	Rotated	120.4	11.3	93.5	11.8
Exp2	Fixed	142.9	22.7	127.9	25.5
	Rotated	135.7	16.6	115.6	19.1

Salt tolerance assay

High salinity affects plant growth and yield, and salt pollution in arable land is a serious agricultural problem. Plant salt tolerance can be evaluated on agar plates (Verslues et�al. 2006, Papdi et�al. 2010) to measure the effect of salt stress on root growth. However, the effects of salinity on shoot growth need to be evaluated by growing plants in soil because of space limitations and overhumidity in agar plates.

The accurate maintenance of salt concentration is crucial to assess plant responses to salinity in soil. The RIPPS irrigation system maintains specific salt conditions by preventing soil water reduction and salt condensation. Plants were grown under control conditions for 6 d, and NaCl solution was added to adjust soil salt concentrations to specific target levels (Fig.�3). After addition of NaCl, soil moisture was kept constant for 15 d. The final rosette area declined in an NaCl concentration-dependent manner. In the presence of 200 mM NaCl, the rosette area was reduced to approximately 50% of that of control plants. As a negative control, we analyzed the sos1 mutant, which lacks a plasma membrane Na+/H+ antiporter that transports sodium ions out of cells (Shi et�al. 2000). Consistent with previous observations, the sos1 mutant showed higher sensitivity to salinity stress. The sos1 rosette area was <50% smaller than that of control plants in the presence of 100 mM NaCl (Fig.�3B).

Fig. 3

Evaluation of salt stress response using the RIPPS platform. Salt stress response of sos-1 mutants (A, B) and three Arabidopsis accessions (C, D). Seeds were germinated and grown in soil for 6 d. Then, the indicated NaCl concentration was added and the soil water content was maintained for 2 weeks. (A) Images of 3-week-old wild-type and sos-1 mutants grown in NaCl-containing soil. Scale bar = 1 cm. (B) Final rosette area (mm2) of the wild type and sos-1 mutant grown in NaCl-containing soil. (C, D) Fresh weight (C) and fresh weight relative to the value obtained at 0 mM NaCl (D) of Col-0 (filled circle, solid line), Cvi-1 (filled triangle, dashed line) and Bur-0 (open circle, solid line). Error bars indicate the SD (n = 5).

The same salinity tolerance assay method was used to evaluate three Arabidopsis wild-type accessions: the salt-tolerant accession Bur-0 (Katori et�al. 2010), the salt-sensitive accession Cvi (Borsani et�al. 2001, Vallejo et�al. 2010) and Col-0. As expected, Bur-0 grew better than Col-0 under salt stress conditions, consistent with previous studies (Katori et�al. 2010) (Fig.�3C, D). Cvi growth was not affected by the salt stress conditions used in this study. Cvi was reported to display salt-sensitive germination and seedling growth in agar plates (Borsani et�al. 2001, Vallejo et�al. 2010), but recent phenotyping studies of salinity stress in soil reported that Cvi leaf area was not affected by salinity stress (Awlia et�al. 2016). This suggests that plants exhibit a different response to salinity stress depending on the culture system. These combined results indicate that RIPPS can be used for quantitative analysis of plant salinity responses in soil culture. Plant responses to other chemicals or nutrient conditions can also be evaluated using RIPPS.

Water use efficiency

Agriculture is the predominant consumer of the available freshwater, and increasing demand for irrigation together with a growing population causes the severe competition between agriculture and other human activities (Peleg and Blumwald 2016). Thus, more efficient management of available water resources is required to increase agricultural production. Greater WUE (defined as dry matter or yield produced per unit of water applied) is crucial to maximize plant productivity under limited water supply.

To understand the effect of soil moisture on plant water use, wild-type Col-0 plants were grown under various soil moisture conditions (Fig.�4A). Plant growth rate increased depending on the soil moisture level (Fig.�4B). We also analyzed leaf surface temperature, which is considered as an indicator of stomatal closure and leaf transpiration (water evaporation causes heat loss) (Chaerle and Van Der Straeten 2000, Merlot et�al. 2002). The results indicated that leaf temperature was inversely correlated with transpiration rate per unit leaf area (Fig.�4C, D). Leaf temperature was unchanged between 4.0 to 2.0 g water g–1 dry soil, but declined significantly under lower soil moisture conditions (<1.5 g water g–1 dry soil) (Fig.�4C, D). This suggests that stomatal closure occurs at an SWC lower than 1.5 g water g–1 dry soil. The relationship of leaf water content to pot SWC at the end of the experiment was similar to that of leaf water content to transpiration rate per unit area (Fig.�4E). Dry weight declined proportionally with pot SWC (Fig.�4F), indicating that growth arrest in response to low SWC starts earlier than the stomatal response. Finally, the integrated WUE was compared among pots. WUE was significantly elevated under low SWC levels (1.0 and 0.8 g water g–1 dry soil) (Fig.�4G). ABA accumulation was elevated at an SWC lower than 1.5 g water g–1 dry soil (Fig.�4H). WUE was correlated with ABA content, indicating that the principal determinant of plant water use is ABA regulation of stomatal aperture.

Fig. 4

Growth response and water use efficiency of Col-0 plants under different SWCs. (A) Transition of SWC. (B) Projected leaf area was measured in the middle of the day. Corresponding SWC is indicated at the right of each line. (C) Thermal images of plants at 20 DAG. The numbers indicate the SWC of each pot. Scale bar = 1 cm (D) Transpiration rates per unit leaf area (filled circle, solid line) and leaf temperature (open square, dotted line) at 20 DAG. (E, F) Leaf water content (E) and dry weight (F) were analyzed at the end of the experiment. (G) Water use efficiency was calculated as described in the Materials and Methods. (H) ABA accumulation in Col-0 and nced3 mutant plants under different soil water contents (SWCs). ABA content was measured in the fourth and fifth leaves of 25 DAG plants. Error bars indicate the SD [n = 5 for (B, D–G) and n = 3 for (H)].

ABA metabolism is involved in water use efficiency

Consistent with previous studies (Munemasa et�al. 2015, Kuromori et�al. 2016, Yang et�al. 2016), we observed that WUE is largely determined by stomatal aperture, which is regulated by ABA signaling (Fig.�4). To understand the impact of ABA levels on WUE, we used RIPPS to analyze the integrated WUE of the ABA biosynthesis mutant nced3-2 (Urano et�al. 2009) and the ABA catabolic enzyme mutant cyp707a3-1 (Umezawa et�al. 2006). The NCED3 gene encodes a 9-cis-epoxycarotenoid dioxygenase, a key enzyme of ABA biosynthesis that is highly responsive to drought stress (Iuchi et�al. 2001). The nced3 mutants displayed higher transpiration rates and drought-sensitive phenotypes (Iuchi et�al. 2001). The CYP707A gene encodes ABA 8′-hydroxylase, a key enzyme in ABA catabolism (Saito et�al. 2004, Umezawa et�al. 2006). The nced3-2 mutants had higher transpiration rates than Col-0 plants when grown under both well-watered and water-limited conditions (Fig.�5A, B), whereas cyp707a3 had lower transpiration rates under water-limited conditions. Under well-watered conditions, there were no significant differences in plant sizes of mutants and wild-type plants (Fig.�5C). Under water-limited conditions, the growth of nced3-2 mutants was drastically reduced, which resulted in lower WUE (Fig.�5D). Although cyp707a3 mutants also displayed slight growth retardation under water-limited conditions, the WUE of cyp707a3 was higher than that of the wild type due to the lower transpiration rate.

Fig. 5

Water use efficiency of ABA biosynthesis mutants. (A) Thermal images of the wild type (WT), nced3 and cyp707a3 mutants. Rosette leaves in 16-day-old plants grown on RIPPS under well-watered (WW) or mild drought (MD) conditions were imaged by an infrared thermography device. Scale bar = 1 cm. (B) Leaf surface temperature of the wild type (solid bar), nced3 (open bar) and cyp707a3 (gray bar). (C) Transpiration rates per unit leaf area of the wild type (filled square), nced3 (open circle) and cyp707a3 (open triangle) are plotted under well-watered (solid line) and mild drought (dashed line) conditions. Dry weight (D) and WUE (E) of the wild type (solid bar), nced3 (open bar) and cyp707a3 (gray bar) were calculated as described in the Materials and Methods. Error bars indicate the SD (n = 5).

These combined results indicate that the RIPPS phenotyping system provides an accurately regulated, homogenized environment and time-course information of plant growth under different environmental stresses. Integration with other genome-wide analyses such as the transcriptome, metabolome and proteome will be a powerful tool for identifying genes and understanding complex mechanisms of plant adaptation to changing environments. The RIPPS may capture aspects of plant growth responses to various environmental conditions that were not previously observed, and may facilitate mutant screening or genetic analyses such as genome-wide association studies (GWAS) or analysis of quantitative trait loci (QTLs) of growth- and stress-related traits.

Materials and Methods

Construction of the RIPPS platform

The RIPPS contains a belt conveyor with 120 pot trays, a balance (FZ-1200i WP, A&D Instruments), a syringe pump (X1000, Cavro Scientific Instruments) and an illumination device containing 10 straight-tube fluorescent lamps with a dimmer control in a steel structure (2.6 m�0.7 m) built by the TECS Inc. (http://www.tecs.ne.jp) (Fig.�1A−C). For imaging analysis, two color cameras at a resolution of 3,376�2,704 pixels (Grasshopper3 9.1 MP Color USB3 Vision, FLIR) for top and side view images under light conditions, and a black and white camera at a resolution of 4,240�2,824 pixels (Grasshopper3 12.0 MP Mono USB3 Vision, FLIR) for night imaging (Fig.�1E–H) were used. In the usual condition, spatial resolutions are about 0.08, 0.25 and 0.06 mm pixel–1 for the top-view color images, the top-view black/white images and the side view images, respectively. Night-time imaging under dark conditions was conducted using 950 nm LED lights (SFH420-Z, OSRAM GmbH), which have only minimal effects on plant growth under dark conditions (Takano et�al. 2009) (Fig.�1F). Projected leaf area was analyzed with a customized leaf segmentation algorithm that was based on previously reported software (Tanabata et�al. 2010) (Supplementary Fig. S2). Thermal images were obtained using an infrared camera (FLIR T640, FLIR Systems, Inc.). Leaf temperature was calculated using the spotmeter tool of the FLIR Tools software (FLIR Systems). The temperature of six spots on a plant was measured and averaged.

Plant materials and growth conditions

Arabidopsis thaliana wild-type accessions, the sos1-1 mutant and the cyp707a3-1 mutant were obtained from RIKEN BioResource Center (http://www.brc.riken.go.jp/lab/epd/Eng/), and the Arabidopsis Biological Resource Center (ABRC). The nced3-2 mutant (Urano et�al. 2009) was obtained frome the GABI-Kat project at the Max Planck Institute. Seeds were germinated on soil-filled 96-well plates after they were subjected to 6 d of stratification at 4�C in the dark as described previously (Clauw et�al. 2015). At 5 days after germination (DAG), the cotyledon area was analyzed and seedlings with average cotyledon sizes were transferred to pots filled with Jiffy mix (SAKATA Seed Corporation) and 20 ml of 0.1% Hyponex. The soil was dried at 105�C for 8 h and was weighed to determine the SWC (the ratio of weight water to the weight of dried soil). SWC at retention capacity was 4.2 g H2O g−1 dry soil. After acclimation treatment for 2 d, pots were put on the RIPPS and grown under long-day conditions (16 h light/8h dark) with 40% relative humidity, daytime temperature of 22�C and night-time (dark) temperature of 18�C.

Five-day-old seedlings were planted in pots and grown for 6 d under dry soil conditions (3.0 g H2O g−1 dry soil). NaCl solution in pure water was added until the SWC reached 3.5 g H2O g−1 dry soil, and the NaCl concentration in total water in soil reached the pre-specified level. To homogenize NaCl distribution within the pot, the salt solution was added at the surface and the bottom of the pot.

Water use efficiency assay

For well-watered (WW) treatments, the SWC of the pots was adjusted every 2 h to 3.5 g H2O g−1 dry soil for 21 d. For mild drought stress (MD) treatments, the SWC of the pots was adjusted to 3.0 g H2O g−1 dry soil for 7 d, and then watering was stopped until the SWC reached 1.25 g H2O g−1 dry soil. We confirmed that the growth stages of wild-type, nced3-2 and cyp707a3 mutant plants at the beginning of the dry treatment were the same (sixth leaf initiation). To prevent water evaporation from the soil surface, each pot was covered during the period of transpiration measurement with a disk of polystyrene board and masking tape (Supplementary Fig. S4). The transpiration rate was calculated based on the gravimetric water loss rate. Water loss in treated plants was normalized by subtracting the water loss from pots containing soil but no plants. Transpiration per unit area was calculated by dividing the daily transpiration amount by the leaf area at the end of the day. The increase in dry weight was determined from the difference between the estimated shoot dry weight at 9 DAG (for WW conditions) or at 23 DAG (for MD conditions), and shoot dry weight at the end of the experiment. The shoot dry weight halfway through the experiment was estimated based on the leaf area. WUE was calculated as the increase in dry weight divided by the total transpiration during the measurement period.

Funding

This research was supported by the Ministry of Education, Culture, Sports, Science and Technology (MEXT) [a special fund for agricultural projects and Grants-in-Aid for Scientific Research]; the Program for Promotion of Basic and Applied Research for Innovations in Bio-Oriented Industry (BRAIN); the Ministry of Agriculture, Forestry and Fisheries (MAFF); the Japan Society for the Promotion of Science (JSPS); and JSPS [KAKENHI grant No. JP16KT0031].

Click here for additional data file.

Click here for additional data file.

Click here for additional data file.

Click here for additional data file.