A gentle touch-force loading machine is built from human hair brushes, robotic arms and a controller. The hair brushes are driven by robotic arms installed on the machine and move periodically to apply touch-force on plants. The strength of machine-driven hair touches is comparable to that of manually applied touches.
Plants responding to both intracellular and extracellular mechanical stimulations (or force signals) and develop special morphological changes, a called thigmomorphogenesis. In past decades, several signaling components have been identified and reported for being involved in the mechanotransduction (e.g., calcium ion binding proteins and jasmonic acid biosynthesis enzymes). However, the relatively slow pace of research in the study of force signaling or thigmomorphogenesis is largely attributed to two reasons: the requirement for laborious human hand-manipulated touch induction of thigmomorphogenesis and the force strength errors associated with people’s hand-touch. To enhance the efficiency of external force loading on a plant organism, an automatic touch-force loading machine was built. This robotic arm-driven hair brush touches provide a labor-saving and easily repeatable touch-force simulation, unlimited rounds of touch repetition and adjustable touch strength. This hair touch-force loading machine can be used for both large scale screening of touch-force signaling mutants and the phenomics study of plant thigmomorphogenesis. In addition, touch materials such as human hair, can be replaced with other natural materials like animal hair, silk threads and cotton fibers. The automated moving arms on the machine may be equipped with water sprinkling nozzles and air blowers to mimic the natural forces of rain drops and wind, respectively. By using this automatic hair touch-force loading machine in combination with the hand-performed cotton swab touching, we have investigated the touch response of two force signaling mutants, MAP KINASE KINASE 1 (MKK1) and MKK2 plants. The phenomes of the touch-force loaded wild type plants and two mutants were evaluated statistically. They have exhibited significant differences in touch response.
Plant thigmomorphogenesis is a term that was coined by Jaffe, MJ in 19731. It is a plant tropism but different from the well-known phototropism or gravitropism caused by stimuli of sunlight or gravity2,3. It describes phenotypic alterations associated with periodic mechanical stimulations, which have been frequently observed by botanists in earlier times4,5. Raindrops, wind, plant, animal and human touches, even animal bites, are all considered to be different types of mechano-stimuli that trigger the force signaling in plants4,5. Characteristics of plant thigmomorphogenesis include the delay of bolting, a shorter stem, smaller rosette/leaf size in herbaceous plants, and thicker stem in woody plants6,7,8. This is unlike the thigmonastic or thigmotropic response often found in the Mimosa plant or other mechano-sensitive vines, where these rapid touch responses are easier to be observed1,9,10. Thigmomorphogenesis, on the other hand, is relatively difficult to be observed because of its slow growth response. Thigmomorphogenesis is usually observed following weeks or even years of continuous force-loading stimulation. This unique nature of plant touch response makes it difficult to perform a forward genetic screen using human hand touch stimulation to isolate the touch-force signaling resistant mutants in a robust manner.
To elucidate the force signal transduction pathways and the molecular mechanisms underlying the thigmomorphogenesis6,11, molecular and cellular biological experiments have been performed in the past6,12,13,14. These studies have proposed that the plant force signal receptors mainly consist of mechanosensitive ion channels (MSC) and the tethered MSC complexes composed by multimeric complexes of membrane-spanning proteins11,14,15. The cytoplasmic Ca2+ transient spike generated within seconds of the initial touch. Wind-, rain-, or gravi-stimulation may interact with the downstream calcium sensors to transduce the force signals to nuclear events14,16,17,18. In addition to molecular and cellular studies, the forward genetic screen with manual finger touching of plants has found that phytohormones and the secondary metabolites are involved in the consequent touch-inducible (TCH) gene expression following the touch-force loading13,19. For examples, aos and opr320 mutants have been identified thus far from the genetic studies. However, the major problem associated with application of the forward genetics in the study of thigmomorphogenesis is still the intensive labor required for quantitating the level of touch response and touching a large population of genetically mutated individual plants. The time-consuming issue also persists in the hand touching-based mutant screen14,20. For an example, to complete one round of touch-force stimulation, a person needs to touch 30-60 times (one touch per second) on an individual plant. In order to have enough number of plants for statistical phenotype analysis, 20-50 individual plants of the same genotype are normally required for the touch-force loading process. This touch-force loading regime means that a person needs to repetitively perform 600-3,000 touches on one genotype of choice. This type of touch normally needs to be repeated 3 to 5 rounds a day, which equals roughly 1,800-15,000 finger or cotton swab touches per day per genotype of plants. A well-trained person is normally required to maintain the strength and force of multiple touches within a desirable range throughout many rounds of repetition in a day to avoid the large variation in force and strength. As it is well known that thigmomorphogenesis is a saturable and dose-dependent process6,21, touch force/strength becomes critical to a success in triggering touch response of a plant.
To remove the person-dependent touch-force loading and to maintain mechanical application within an acceptable error range14, we therefore designed an automatic touch-force loading machine to replace the hand-manipulated touches. The machine has 4 moving arms built, each of which is equipped with one human hair brush. This version is named Model K1 to specify its feature of human hair touch-force loading. If 4 genotypes are measured quantitatively for their thigmomorphogenesis or touch response under one machine, 40-48 individuals per genotype can be measured. Each round of touch repetition (less than 60 times of touch per plant) lasts less than 5 minutes using a moving speed adjustable robotic arm. Thus, plants on a Model K1 touch machine can be mechanically stimulated for multiple rounds a day either with a constant touch-force loading or different levels of strengths as initially programmed.
Arabidopsis thaliana, a model plant organism, was therefore chosen as the target plant species for testing the fully automatic hair touch-force loading machine application. Because there are several large seedbanks available for retrieving the various germplasms of mutants and the size of flowering, Arabidopsis fits well to the space available in the growth shelf mounted with the Model K1 touch machine.
The Model K1 automatic touch machine consists of three major components: (1) the H-shape metal rack composed by two belt-driven linear actuators, (2) robotic metal arms equipped with hair brushes, and (3) a controller. For a customized Model K1 touch machine, each X/Y axis module is composed of one belt-driven guide-rail, two slide blocks (red) and one 57 stepper motor (pre-installed and dismountable) (Figure 1A,B). The upper horizontal actuator allows the robotic metal arm to move left and right horizontally, the lower vertical belt-driven linear actuator allows the robotic metal arm to move up and down vertically (Figure 1B, Figure 2A). Four dismountable robotic arms were installed on the vertical actuator (Figure 1C, Figure 2B). Four human hair brushes were bound to four robotic arms, respectively (Figure 1C, Figure 2B). All mechanical parts to construct the Model K1 touch machine in bolded font below are marked in Figure 1C (also see the Table of Materials).
1. Seed preparation
NOTE: Arabidopsis seeds of both wild type (Col-0) as well as mkk1 and mkk2 loss-of-function mutants used were purchased from the Arabidopsis Biological Resource Center (ABRC, https://www.arabidopsis.org, Columbus, OH).
2. Plant growth
3. Growth condition
4. The construction of touch-force loading machine
NOTE: This robotic hair touch-force loading machine (Model K1) is designed to serve purposes of both touch-force signaling mutant screening and plant thigmomorphogenesis generation (Figure 1, Figure 2).
5. Touch-Force loading machine setting
NOTE: All of controlling parameters to set the Model K1 touch machine in bolded font below are shown in the control panel (Figure 2F).
6. Physiological data collection and analysis
The automatic hair touch-force loading machine
For observation of morphological changes on plants, both the reproducible growth conditions and treatment methods are key to obtaining repeatable results. This high-throughput and automatic touch-force signaling mutant screening is achieved by the newly built hair touch-force loading machine, Model K1 (Figure 1, Figure 2). These hair brushes can touch a maximum of 4 trays of plants simultaneously. There were 24 cups placed in a tray, and 12 cups of plants in a group used as both control and the treated plants (Figure 2C,D). In each cup, four plants were grown and a total of 48 or less plant individuals were touched by the same hair brush, which guarantees enough plants for later statistical analysis. A maximum of 4 genotypes of plants can be touch-treated simultaneously on one Model K1 touch machine. One of the key points is the setting of the touch machine arm/hair height because thigmomorphogenesis is dose-dependent6,21. Different hair positions with respect to the plant rosette leaf position generate different touch forces, which may generate totally different thigmomorphogenesis results. In our experiments, the plant-contacting tip of hairs should be placed 0.5 cm lower than the cup rim (Figure 2E), which generates forces that are similar to the previously published touch force14. A programmable controller installed into a touch panel is used to control the whole touch-force loading machine (Figure 2F, see Table of Materials).
The comparison of two different touch methods
To compare this automatic machine-driven hair method with the conventional manually cotton swab touching method, two independent experiments were performed on Col-0 (Figure 3). In the cotton swab touch group, touching started from 12-day-old plants. Each round had 40 touches (1 touch/s). In total, 3 rounds were performed each day (Figure 3A). It showed 1.7 days of delay in bolting after a continuous cotton swab touch treatment (22.1 ± 0.2 days vs. 23.8 ± 0.2 days). Similarly, for the automatic machine-driven hair touch, the touch-force loading initiated from 14-day-old plants and 40 times of touch (within 3 min) were applied for a round. In total, 3 rounds of touches were performed a day with exactly 8 hours of interval (Figure 3B). Delayed bolting was observed for Col-0 plants. The average bolting time was 23.0 ± 0.3 days, while the bolting time of the Model K1 touch machine-treated plants was 24.7 ± 0.2 days. The differences between control and touch-treated plants were consequently analyzed with the univariate Cox proportional-hazards model. It offered the estimated Hazard Ratio (HR) of 0.31 (cotton swab touch) and 0.52 (machine-driven hair touch), respectively (Figure 3C), which means that the bolting risk/probability of plants in the touched group is 31% and 52% compared with plants in the control group, respectively. This indicates that the possibility of bolting of the touched wild type plants is around half as compared to that of the untouched control plants regardless of whether it is manual touch with a cotton swab or the automated hair touch.
The prospective results on different touch mutants
Recent preliminary data suggested that MKK1 and MKK2 might play an important role in the touch response of Arabidopsis14. We selected these two mutants and conducted touch experiments on these putative touch response mutants using the automatic hair touch-force loading machine (Figure 4, Table 1). The wild type control plants showed 1.8 days of bolting delay (24.1 ± 0.3 days vs. 25.9 ± 0.2 days, Figure 4A) just as the previous report14 while this bolting delay was not observed on T-DNA insertional mutants, mkk1 (24.6 ± 0.2 days vs. 24.4 ± 0.3 days, Figure 4B and Table 1) and mkk2 (23.9 ± 0.1 days vs. 24.2 ± 0.2 days, Figure 4C and Table 1). By analyzing these data with the univariate Cox proportional-hazards model, only wild type Col-0 exhibited a significant difference between control and touched plants with an estimated HR of 0.41 (Figure 4D). These touch-force loading experiments conducted by the automatic hair touch-force loading machine demonstrated that mkk1 and mkk2 mutants are touch response mutants.
The measurement of other morphological indexes
Morphological changes associated with thigmomorphogenesis are not limited to the delaying of bolting. Both shorter stem and smaller rosette leaf size are also the components of thigmomorphogenesis6,7,9,14. Hence, we reported here two additional types of measurements on morphological indexes of touch response, rosette radius/leaf length and rosette (projected) area (Figure 5). Similar to the previously observed phenotype change, the wild type Col-0 plant showed significantly smaller rosette radius and shorter leaf length after 3 days of constant and repetitive automatic machine-driven hair touch (1.77 ± 0.05 cm vs. 1.50 ± 0.04 cm, Figure 5A). The projected rosette area was changed from 20.32 ± 0.53 cm2 to 16.19 ± 0.48 cm2 after 13 days of touch (Figure 5B). Both mkk1 and mkk2 had the similar reduced rosette radius and area. Taken together, these data demonstrated that MKK1 and MKK2 proteins are important for the bolting delay of Arabidopsis and not required in shaping the rosette size and rosette area.
Statistical Analysis
As to the box and whisker plots shown in Figure 2 and Figure 3 and the column charts shown in Figure 5, the statistical significance was analyzed by two-tailed student’s t-test, with significance represented by *** and n.s. at p < 0.001 and p > 0.05, respectively. For the Kaplan-Meier plots shown in Figure 2 and Figure 3, a univariate Cox hazard analysis was used to analyze the effect of touch treatment on the bolting event23,24. The Hazard Ratio (HR), 95% Confidence Interval (95% CI) and p value are offered in the tables below. For instance, HR = 0.5 means that on a specific day, the bolting risk/probability of plants in the touched group was 0.5 or 50% compared with those plants in the control group.
Figure 1. The construction and parameters of the automatic hair touch-force loading machine. (A) Default schematics of the linear actuator. The upper left panel is the lateral view and the lower left panel is the dorsal view. The total lengths of the X axis module and Y axis module are 843 mm and 1,038 mm, respectively. Each default X/Y module is composed of one guide-rail, one slide block and one 57 stepper motor (pre-installed and dismountable). For a customized Model K1 touch machine, each X/Y module is composed of two slide blocks (red). The junction plate of the X module is enlarged from 56 mm to 100 mm to offer better connection and support. The upper right panel is the cross-section of the guide-rail and the lower right panel is the 57 stepper motor. (B) Schematics of the constructed double X axis and double Y axis linear actuators. This is the major part of the touch-force loading machine. The lower left panel is the dorsal view of constructed linear actuators. The upper left panel is the lateral view of the X axis module (843 mm). The middle panel is the lateral view of the Y axis module (1,038 mm). The upper right panel is the dorsal view of 4 slide blocks on the Y module and Y auxiliary girder. The lower right panel is the dorsal view of the junction plate on the X module. (C) The flowchart of machine part assembly. Different parts are marked and named in the figure. Detailed assembling processes were described in the protocol. The unit shown is this figure is mm. Please click here to view a larger version of this figure.
Figure 2. The overall design of the automatic hair touch-force loading machine. (A) The finished Model K1 touch machine. The photo was taken from the front side. The upper linear actuator controls the robot arm moving horizontally and the lower linear actuator controls the robotic arm moving vertically. (B) The lateral view showing dismountable robotic arms. Hair brushes were clamped onto the robotic arms. (C and D) Photos showing how human hairs brushes touch the plants, which were taken from the front side and the lateral side, respectively. (E) The lateral view showing how to set the height of the hair brush against the cup rim. Both the machine arms and hair brushes are visible. (F) The operation interface of the Model K1 touch machine. A programmable controller (AFPX-C30T) linked to a touch panel (MT6070i) is used to control the whole machine. Detailed settings and operating procedures are described in the protocol. Please click here to view a larger version of this figure.
Figure 3. Comparing the effects of two touch methods on thigmomorphogenesis. (A-B) The comparison of manual cotton swab touch (A) and human hair touch driven by the automatic touch-force loading machine (B), respectively. Box and whisker plots are shown in the left panel, which show the comparison of average bolting day between the control group and the touched group. Means ± SE are shown. Statistical analysis was performed by a Student’s t-test. Significance at p < 0.001 is shown as ***. Kaplan-Meier plots are shown in the middle, which are the percentage of bolting plants over the growth time (days after sowing). The right panel shows representative individuals of untouched control and touched plants that show the difference in bolting time and inflorescence stem height. (C) The summarized table: the numerical numbers in control and touched columns are the plant number used for statistical analysis. The Hazard Ratio (HR), 95% Confidence Interval (95% CI) and p value under the section of univariate Cox hazard analysis are offered. The bolting risk and probability of plants in the touched group were 31% and 52% as compared to the untouched group, respectively. The univariate Cox hazard analysis was estimated by SPSS. Please click here to view a larger version of this figure.
Figure 4. The thigmomorphogenesis of mkk1 and mkk2 mutants as well as the wild type plant (Col-0) induced by the automatic hair touch. (A-C) The prospective touch response of Col-0 (A), mkk1 (B), and mkk2 (C) generated through repetition of human hair touches driven by the automatic touch-force loading machine. Box and whisker plots are shown in the left panel, which are the comparison of average bolting day between the control group and the touched group. Means ± SE are shown. Statistical analysis was performed by a Student’s t-test. The *** and n.s. represent p < 0.001 and p > 0.05, respectively. Kaplan-Meier plots are shown in the middle, which are the percentage of bolting plants over the growth period (days after sowing). The right panel shows representative individuals of the untouched control and the touched plants that show the bolting difference. Data of mkk1 (B) and mkk2 (C) were compiled from two and three biological replicates, respectively. Detailed plant numbers used in each replicate were shown in Table 1. (D) The summarized table: the numbers under control and touch columns were the plant number used/analyzed in these two groups, respectively. The HR, 95% CI and p value under the section of univariate Cox hazard analysis were offered. The bolting risk/probability of wild type plants in the touched group is 41% as compared with the control group. The univariate Cox hazard analysis was estimated by SPSS. Please click here to view a larger version of this figure.
Figure 5. The rosette radius and area measurement for defining thigmomorphogenesis. (A-B) The rosette radius and the rosette area of the wild type were measured at day 17 and day 27 after seed sowing, respectively. Bars shown in the upper left panel are the comparisons of either rosette radius or rosette area between the control group and the touched group, respectively. Means ± SE are shown. Statistical analysis was performed by Student’s t-test; ***p < 0.001. Photos shown in the upper right panel are representative individual plants. The summarized tables below show the plant number analyzed in the control group and the touched group. Both the rosette radius (cm) at day 17 and rosette area (cm2) at day 27 are also shown. Please click here to view a larger version of this figure.
Table 1. The bolting data of different biological replicates. The summarized table contains two biological replicates of mkk1 and three biological replicates of mkk2.
Thigmomorphogenesis is a complex plant growth response towards mechanical perturbations, which involves a network of cellular signaling and action of phytohormones. It is a consequence of adaptive evolution of plants to survive under the undesirable environmental conditions25,26. Mechanical touch, especially human finger touch and hand-held cotton swab touch, have been selected to study this morphological changes in previously thigmomorphogenetic studies14,20. This simplified version of touch-force loading to trigger plant touch response is easier to control and apply. In addition, this type of touch-force loading method can in some way mimic the wind- and rain drop-stimulated force signals produced in the natural environment19. The touch force is able to trigger calcium spikes, induce protein phosphorylation14 and the downstream gene expression mediating touch response19. Similarly, human hair brushes mounted on automated moving arms can also generate the plant touch response by mimicking the human hand-manipulated touches. To diversify the types of force application, water sprinkling nozzles and/or wind blowers can also be installed onto the robotic arms of the machine and used for a physiological experiment (Figure 2). The unique feature makes the automatic mechanical-force loading machine more versatile in the morphogenetic and physiological studies. The biggest advantage of this automatic mechanical-force loading machine is probably its labor-free, repeatable and time-saving feature, which makes it possible to perform a specific mutant phenotype selection from a large number of mutagenized individuals. In contrast to hours of human hand-manipulated touches, the Model K1 touch machine can touch various mutants simultaneously and complete a round of touch within 3 to 5 min. The time frame for a round of touch largely depends on the program setting at the beginning of treatment. If each individual plant would be touched 40 times in a round, the Model K1 machine would only need 9-15 min to finish three rounds of touch treatment within a day. The interval time between each round of touches can be precisely controlled; it is less likely for human beings to achieve such a precision.
Another important issue regarding the touch treatment is which stage of plant growth the touch force needs to be applied upon. In our practice, touching started 14 days after seed sowing for both the wild type and two mutants as the growth rates of these three genotypes are similar. For those mutants that have a significant difference in development time from the wild type, one may choose a different initial day to start the touching. Performing the one-way ANOVA test on the bolting data of both wild type plant and mutants for multiple comparisons can help14. This statistical analysis can offer the proper conclusion about the differences of bolting time generated by genotypes. In this case, a multivariate Cox proportional hazard analysis should be used to consider two variable parameters.
To set the touch-force level of the human hairs mounted on the Model K1 touch machine, we adjusted both the height (vertical force) and the speed (horizontal force) of the hair brushes (Figure 2E). The right settings were determined based on the preliminary data collected from many rounds of force level tests on an Arabidopsis plant placed on an electronic scale. As we have found, keeping both the hair height and the speed unchanged throughout the entire touch response experiment will produce a similar and constant thigmomorphogenetic phenotype among replicates for an Arabidopsis line. Too heavy a touch-force may kill the young seedlings as the fast moving hair brushes may lead to wounding on the surface of a leaf. In contrast, too light a touch-force may not be enough to trigger the delay of bolting within 2 weeks of repetition of touching. In our previous experiment, we have determined the appropriate touch-force loading to be 1-2 mN per touch14,19. The hair length of 0.5 cm lower than the cup rim is used to generate a similar vertical touch force on Model K1 machine-based hair touch with a gentle horizontal moving speed 5000 mm/min (Figure 2E). This fixed setting of Model K1 machine reduces the variation of force strength resulted from the human error.
Overall, the hair touches performed by the automatic touch-force loading machine provide only an average touch-force loading on plants. The precise touch-force applied, especially the horizontal force loaded, is difficult to calculate either for a single hair or a group of hairs on a brush. In addition, the variance of plant shape and stem height can interfere with the application of horizontal force. Measuring this kind of physical strength or stress needs a more precise pressure sensor linked to a hair or a group of hairs. It is believed that more precise pressure sensor and mathematical modelling will be applied to improve the automatic touch-force loading machine in the future. The growth conditions, such as light intensity, moisture of soil and temperature of the greenhouse as well as nutrients supplying, all play a crucial role in the touch response phenotype development. Any stress conditions, such as drought, weak light condition with less than 90 μE∙m-2∙s-1, and a higher or a lower temperature that may affect the normal growth of Arabidopsis will interfere with the measurement of touch response of both wild type and mutants.
In short, this automatic touch-force loading machine can offer more labor-saving and uniformed average touch-force loading than human finger touch and cotton swab touch. It is expected that the Model K1 touch machine will be applied in various high-throughput touch-force signaling mutant screening and touch response analysis among agricultural crops or probably animal models with some modifications of the touch-force loading machine.
The authors have nothing to disclose.
This study was supported by the following grants: 31370315, 31570187, 31870231 (National Science Foundation of China), 16100318, 661613, 16101114, 16103615, 16103817, AoE/M-403/16 (RGC of Hong Kong). Authors would like to thank Ju Feng Precision and Automation Technology Limited (Shenzhen, China) for their offering of several schematics shown in Figure 1.
Authors would also like to thank S. K. Cheung and W. C. Lee for their contribution to the development of the touch-force loading machine.
4 hair brushes | customized | ||
4 robot arms with one holder | customized | 1000 mm length holder and 560 mm length robot arm | |
57 stepper motor | 57HS22-A | ||
All purpose potting soil | Plantmate, Hong Kong | ||
Arabidopsis plant seeds | Arabidopsis Biological Resource Centers, Columbus, OH | For arabidopsis seed purchase | |
BIO-MIX potting substratum | Jiffy Products International BV, the Netherlands | 1000682050 | Two soils were mixed together to grow Arabidopsis. The ratio of All purpos potting soil and BIO-MIX is 1:2 |
IL 1700 research radiometer | International Light, Newburyport, MA | The light intensity of both full-wavelength and photosynthetic active radiation can be measured. | |
ImageJ | https://imagej.nih.gov/ij/download.html | Free downloaded software | |
Ju Feng Precision and Automation Technology Limited | Shenzhen, China | For belt-driven linear actuators and other mechanical modules purchase | |
Junction plate of the slide block | To fix the Y guide-rail module or Y auxiliary girder onto backs of slide blocks | ||
Junction plate of the X axis module | customized | To connect the X guide-rail module and X auxiliary girder | |
Slide block | |||
WDT4045 X axis guide-rail module | 843 mm, customized | Pre-installed with two slide blocks and one 57 stepper motor | |
WDT4045 Y axis guide-rail module | 1038 mm, customized | Pre-installed with two slide blocks and one 57 stepper motor | |
X axis auxiliary girder | 843 mm, customized | Pre-installed with two slide blocks | |
Y axis auxiliary girder | 1038 mm, customized | Pre-installed with two slide blocks |