Wide-Range Segmentation of Cotyledon Epidermal Cells for Morphometrical Analysis and Mechanical Simulation

Abstract

Cell segmentation from microscopic images is conventionally used to investigate cell morphology. However, the time expense for manual segmentation becomes extreme with increasing numbers of cells to be analyzed. Recent progress in automated image analysis techniques can facilitate efficient and accurate cell segmentation in wide-range confocal images. Pavement cells, which mainly comprise the epidermal tissue of plant leaves, show jigsaw puzzle-like shapes and provide a model for elucidating the mechanisms underlying the complex morphology of plant cells. This mini-review demonstrates the effectiveness of using a confocal image processing pipeline for morphometric analysis and mechanical simulation using Arabidopsis thaliana cotyledon pavement cells as an example. We examined A. thaliana cotyledon surfaces using wide-range confocal images and used an image processing pipeline in ImageJ software to extract epidermal cell contours. We then used the segmented epidermal cell images to provide examples of how this information can be used for morphometry and mechanical simulation. The use of this high-throughput segmentation method is not limited to plant epidermal tissue and can be applied to various biological materials. Therefore, our approach to microscopic image analysis will hopefully contribute to the advancement of quantitative cell morphology research.

Epidermal tissue covers and protects the inside of organisms and interfaces with the outside environment. It has also been suggested that epidermal tissue controls organ morphology, especially in plants (Zhou et al. 2020). Epidermal tissue of plant leaves is mainly composed of guard and pavement cells. The guard cells regulate stomatal movement, which is responsible for gas exchange for photosynthesis, respiration, and stress response (Higaki et al. 2014, 2021, Sato et al. 2021). Alternatively, pavement cells occupy most of the leaf surface and cover the inside of the leaf. During leaf growth, pavement cells exhibit rapid cell expansion by turgor pressure with jigsaw puzzle-like morphogenesis in most dicotyledonous plants (Fig. 1) (Higaki et al. 2016, 2017). It has been assumed that the intricate jigsaw puzzle-like shape contributes to reducing mechanical stress derived from turgor pressure in the cell wall (Sapala et al. 2018).

Fig. 1. Changes in epidermal cell shape during cotyledon growth. Manually traced images of cell contours on the abaxial surfaces of A. thaliana cotyledons were visualized using the plasma membrane marker GFP–PIP2a. Seedlings were photographed at 2 (left), 4 (middle), and 7 d (right) after sowing on half-strength Murashige and Skoog salt agar plates. The insets show rectangular cells gradually becoming interdigitated in the middle region and smoothly elongated in the basal region of cotyledons during expansion. Scale bar=100 µm. This figure is reprinted from Higaki et al. (2017) with permission of Oxford University Press (License Number: 5047481409892).

Pavement cells are an excellent model for understanding the mechanisms of complex morphogenesis in plant cells and have been extensively studied by molecular cell biological approaches (Akita 2020, Lin and Yang 2020). Molecular genetic approaches revealed that the members of a plant-specific Rho GTPase family regulate jigsaw puzzle-shaped morphogenesis through reorganization of microtubules and actin filaments that restrict and promote cell outgrowth, respectively (Fu et al. 2005). Recent advances in bioimaging techniques and mechanical simulations have indicated that geometry and mechanical stresses in cells may contribute to the jigsaw puzzle-shaped morphogenesis (Sampathkumar et al. 2014, Altartouri et al. 2019, Bidhendi and Geitmann 2019). For more information on pavement cell morphogenesis mechanisms, please refer to recent review papers (Eng and Sampathkumar 2018, Sapala et al. 2019, Du and Jiao 2020).

Cell segmentation from microscopic images is a basic technique for investigating cell morphology, and not only for jigsaw puzzle-shaped pavement cells. Conventionally, manual tracing of cell contours on microscopic images was a popular method for segmentation. In fact, we performed morphological measurements of pavement cells in A. thaliana cotyledons on the basis of manually segmented cell shapes (Akita et al. 2015, 2017, Higaki et al. 2016, 2017) (Fig. 1). However, the time expense for manual segmentation becomes enormous as the analyzed area increases, such as when comparing pavement cell shapes in the basal and central regions of a cotyledon using a whole cotyledon image (Fig. 1, insets on the largest cotyledon). Recent progress in image analysis techniques has allowed us to efficiently and accurately perform leaf epidermal cell segmentation using wide-range confocal images. In this mini-review, we briefly describe our confocal imaging method, which captures a wide range of A. thaliana cotyledon surfaces and the image processing pipeline to extract the epidermal cell contours using ImageJ software (Schneider et al. 2012). Additionally, we provide some examples for morphometry and mechanical simulation based on the segmented epidermal cell images.

Confocal imaging and image processing to extract pavement cell shapes

Fluorescent labeling of plasma membranes or cell walls is a standard method for visualizing leaf epidermal cell contours. We prefer using transgenic A. thaliana plants that stably express the plasma membrane marker GFP–PIP2a (Cutler et al. 2000, Higaki et al. 2017, Higaki and Mizuno 2020). Figure 2A shows an example of the single optical section of GFP–PIP2a-labeled plasma membranes in the cotyledon surface of a 15-day-old A. thaliana seedling that was captured by our confocal microscope [a fluorescence microscope (IX-70; Olympus, Tokyo, Japan) equipped with a low magnification (10×) objective lens (UPLXAPO10X, NA=0.4; Olympus), a spinning disk confocal laser scanning unit (CSU-W1; Yokogawa, Tokyo, Japan), a laser illumination homogenization unit (Uniformizer; Yokogawa), and a complementary metal-oxide-semiconductor camera (Zyla; Andor, Belfast, UK)].

Fig. 2. Segmentation of cotyledon epidermal cell contours. The bottom images of (A–C) are enlarged images of the insets in the top images of (A–C). (A) A confocal section of the cotyledon surface of a 15-day-old A. thaliana that expressed the plasma membrane marker GFP–PIP2. Because the cotyledon surface is not perfectly flat, not all cell contours could be captured with a single confocal section. (B) Scale bar=500 µm. Maximum intensity projection (MIP) image of the serial confocal sections at the same region of (A). This MIP image includes both epidermal cell contours and undesired signals from mesophyll cells. (C) Projection image of the surface signal extracted with the ImageJ macro SurfCut (Erguvan et al. 2019) in the same region of (A). (D) Stitched SurfCut images taken at multiple positions to cover the whole cotyledon. Scale bar=500 µm. (E) Segmented image of (D) with the ImageJ plugin Morphological Segmentation (Legland et al. 2016) and manual post-processing.

As shown in Fig. 2A, some epidermal cells were defocused because the cotyledon surface was not perfectly flat. Therefore, a projection image synthesized with the serial optical sections was needed to visualize pavement cell shapes in a whole cotyledon. However, a simple projection method [e.g., maximum intensity projection (MIP)] can only produce images that contain components that are undesirable for extracting the epidermal cell contours, such as signals derived from the plasma membrane of mesophyll cells (Fig. 2B). To obtain the signal of epidermal cells on the cotyledon surface, the ImageJ macro SurfCut is quite useful (Erguvan et al. 2019) (Fig. 2C). The projection image of the cell contour signal extracted with SurfCut (hereafter, SurfCut image) is preferred for cell segmentation than the MIP image (Fig. 2B, C).

When analyzing a large cotyledon that does not fit within the field of view, the images are taken at multiple locations so that the fields of view overlap. The SurfCut images of the serial sections taken at multiple positions are automatically stitched using normalized cross-correlation in the ImageJ plugin LPX-Registration (Nagata et al. 2016) (Fig. 2D). On the basis of the stitched SurfCut images, the epidermal cells are segmented using the ImageJ plugin Morphological Segmentation (Legland et al. 2016) (Fig. 2E). For accuracy, the segmented images should be visually checked by comparing them with the SurfCut images. If there are any segmentation errors, a visual check and manual correction of the automatically segmented images should be performed (Fig. 2E). In the case of our data, this manual post-processing required approximately 60–120 min per cotyledon.

Pavement cell morphometry

Using segmented cell images, measurement of morphological features such as circularity (Fig. 3A), aspect ratio (Fig. 3B), and solidity (Fig. 3C) is a basic way to quantitatively evaluate pavement cell shape (Akita et al. 2015, 2017, Higaki et al. 2017). These morphological features can be easily determined using the ImageJ Analyze Particles function (Schneider et al. 2012). Cell circularity is defined as 4π·Cell Area divided by the square of the cell perimeter length (Fig. 3A); circularity has a maximum value of 1 when the cell shape is a perfect circle and becomes lower as the shape becomes more complex. Complexity, another morphological feature, is defined as the reciprocal of circularity. Therefore, circularity and complexity have essentially the same meaning. These are appropriate indicators for evaluating the change of a cell into a complex shape; for example, the use of these features revealed that cytoskeleton or membrane trafficking inhibitor suppresses jigsaw puzzle-like morphogenesis in A. thaliana cotyledons (Akita et al. 2015, 2017). However, circularity and complexity are indicators of cell perimeter length per unit cell area and cannot distinguish whether the shape change is due to cell elongation or interdigitation. Therefore, we prefer to use aspect ratio, which is defined as the ratio of the major axis length to the minor axis length of the fitted ellipse, as an indicator of cell elongation (Higaki et al. 2017) (Fig. 3B). Aspect ratio shows a minimum value of 1 when the cell is not elongated and increases in value as the cell elongates. We also use solidity, which is defined as the ratio of the cell area to the convex hull area, as an indicator of cell interdigitation (Higaki et al. 2017) (Fig. 3C). Solidity shows a maximum value of 1 when the lateral cell wall is smooth and a smaller value as cell interdigitation proceeds. Aspect ratio and solidity showed that that exogenous cellulase treatment of hydroponically grown A. thaliana cotyledons switched pavement cell interdigitation to smooth elongated cell production (Higaki et al. 2017).

Fig. 3. Morphological features used for quantitative evaluation of jigsaw puzzle-shaped pavement cell shape. The black closed curve indicates a representative pavement cell shape. (A) Circularity, which is defined as the 4π·Cell Area divided by square of the cell perimeter length. (B) Aspect ratio, which is used to evaluate cell elongation. The aspect ratio of a cell is defined by the fitted ellipse shown in red. Long magenta and short green lines indicate the major and minor axes of the fitted ellipse, respectively. The aspect ratio was defined as the ratio of the major axis length to the minor axis length. (C) Solidity, which is used to evaluate cell interdigitation. Solidity is defined as the ratio of cell area to the area of the convex hull (blue closed curve). The solidity value decreases as cell interdigitation proceeds.

In the case of wide-area images, it is also possible to visualize the spatial distribution of epidermal cell morphological features. To visualize the morphological features, the ImageJ macro Shape Descriptor Maps in the BioVoxxel Toolbox (https://imagej.github.io/plugins/biovoxxel-toolbox) is useful. Figure 4 shows the spatial distribution of circularity (Fig. 4A), aspect ratio (Fig. 4B), and solidity (Fig. 4C) of epidermal cells. These feature maps allow us to examine epidermal cell shape distribution. For example, epidermal cells with high aspect ratio values are located near the petiole, and elongated jigsaw puzzle-shaped pavement cells are found throughout the leaf blade (Fig. 4B). In addition, there might be a tendency for interdigitated pavement cells with low solidity values to be more abundant in the lateral sides of the leaf blade (Fig. 4C). Therefore, we believe that these quantitative analyses of cell morphology in a wide field of view can help facilitate morphological analysis across the cell–organ hierarchy.

Fig. 4. Representative spatial distribution of morphological features of the epidermal cells from a 15-day-old A. thaliana cotyledon. The binary image Fig. 2E was used for feature mapping. Circularity (A), aspect ratio (B), and solidity (C) are shown. Scale bar=500 µm.

Mechanical simulation for biomimicry research

Recently, mechanical simulation approaches have enhanced our understanding of the jigsaw puzzle-shaped pavement cell morphogenesis mechanism and its physiological significance (Sampathkumar et al. 2014, Sapala et al. 2018, Bidhendi and Geitmann 2019). It is also important to understand the mechanical properties derived from the organismal cell shapes in biomimicry research because the fascinating morphology of plant leaf surfaces can be used to develop artificial structures, such as in architecture. Therefore, we briefly describe our ongoing research on mechanical simulation of an arched masonry structure built with pavement cell-shaped pieces as an analysis example of using the obtained segmented images (Fig. 5).

Fig. 5. Mechanical simulation analysis of pavement cells for architectural applications. (A) Analysis areas on the segmented image of 21-day-old A. thaliana cotyledon epidermal cells. Three rectangular regions were set up for patterns 1, 2, and 3; the numbers of pieces were 39, 38, and 34, respectively. (B) Pattern 1 arch structure designed on the basis of the catenary curve. The top is a three-dimensional visualization of the structure and the bottom is a lateral view for the boundary conditions. Both fulcrums were fixed for three translational directions and three rotational directions (blue), and gravity was applied to each piece and joint as the initial condition. In all patterns, the material of each piece was wood and the joints were made of steel. The mechanical response under a point load of 100 kN in the vertical direction applied to the center point of the arch (orange) was investigated by the finite element method using the structural analysis software Karamba (Preisinger and Heimrath 2014). (C) Visualization of the distribution of the principal stress in the pieces. The principal stresses are tensile when the value is positive (blue) and compressive when the value is negative (red).

This research is supported by the MEXT KAKENHI Grant-in-Aid for Scientific Research on Innovative Areas “Plant-Structure Optimization Strategy,” which aims to utilize the mechanical advantage of plants in engineering, especially in architecture (Demura 2020). Because the jigsaw puzzle shape of pavement cells might be caused by buckling due to compressive forces on the cell walls (Higaki et al. 2017), we investigated the use of arch structures based on catenary curves in which there are only compressive forces. To design an arch structure made of jigsaw puzzle-shaped pavement cell pieces, we used rectangular regions in the wide-range segmented images (Fig. 5A, B). Using the finite element method, it is possible to calculate the complex interplay of tension and compression when loading the arch structures (Fig. 5B, C; see the legend for detailed conditions). Therefore, we can compare these biomimicry architectural forms with conventional architectural forms through such mechanical simulation analysis. As shown in this example, segmented images obtained by microscopic image processing can be applied to other fields of analysis and be used for interdisciplinary research, such as biomimicry research.

Conclusion

Here, we presented a confocal imaging procedure and image processing pipeline for morphometric analysis and mechanical simulation of A. thaliana cotyledon epidermal cells. We focused on two-dimensional approximation analysis; however, it is possible to acquire three-dimensional structural information of these cells and their changes over time (i.e., four dimensions) (Higaki and Mizuno 2020). It is better to obtain three- or four-dimensional information, but there are many cases where two-dimensional approximations are sufficient. It is important to compress the data in an appropriate dimension according to research purposes. The high-throughput segmentation method described in this manuscript is not limited to plant epidermal tissue but can be applied to a variety of biological materials. These technical advances in microscopic image analysis will hopefully continue to advance the study of cell morphology from a quantitative perspective.

Acknowledgments

We thank Prof. Taku Demura of Nara Institute of Science and Technology, Ken’ichi Kawaguchi of The University of Tokyo, and Prof. Akitoshi Iwamoto of Kanagawa University for their helpful suggestions. This work was supported by the Japan Society for the Promotion of Science (JSPS) KAKENHI to T.H. (18H05492 and 20H03289). We thank Mallory Eckstut, Ph.D., from Edanz (https://jp.edanz.com/ac) for editing a draft of this manuscript.

References

• Akita, K. 2020. Quantitative imaging of pavement cell morphogenesis and stomatal differentiation. Cytologia 85: 101–106.
• Akita, K., Higaki, T., Kutsuna, N. and Hasezawa, S. 2015. Quantitative analysis of microtubule orientation in interdigitated leaf pavement cells. Plant Sig. Behav. 10: e1024396.
• Akita, K., Kobayashi, M., Sato, M., Kutsuna, N., Ueda, T., Toyooka, K., Nagata, N., Hasezawa, S. and Higaki, T. 2017. Cell wall accumulation of fluorescent proteins derived from a trans-Golgi cisternal membrane marker and paramural bodies in interdigitated Arabidopsis leaf epidermal cells. Protoplasma 254: 367–377.
• Altartouri, B., Bidhendi, A. J., Tani, T., Suzuki, J., Conrad, C., Chebli, Y., Liu, N., Karunakaran, C., Scarcelli, G. and Geitmann, A. 2019. Pectin chemistry and cellulose crystallinity govern pavement cell morphogenesis in a multi-step mechanism. Plant Physiol. 181: 127–141.
• Bidhendi, A. J. and Geitmann, A. 2019. Geometrical details matter for mechanical modeling of cell morphogenesis. Dev. Cell 50: 117–125.
• Cutler, S. R., Ehrhardt, D. W., Griffitts, J. S. and Somerville, C. R. 2000. Random GFP:cDNA fusions enable visualization of subcellular structures in cells of Arabidopsis at a high frequency. Proc. Natl. Acad. Sci. U.S.A. 97: 3718–3723.
• Demura, T. 2020. Preface to the special issue “Approaches for strategies of mechanical optimization in plants.” Plant Biotechnol. 37: 393–395.
• Du, F. and Jiao, Y. 2020. Mechanical control of plant morphogenesis: Concepts and progress. Curr. Opin. Plant Biol. 57: 16–23.
• Eng, R. C. and Sampathkumar, A. 2018. Getting into shape: The mechanics behind plant morphogenesis. Curr. Opin. Plant Biol. 46: 25–31.
• Erguvan, Ö., Louveaux, M., Hamant, O. and Verger, S. 2019. ImageJ SurfCut: A user-friendly pipeline for high-throughput extraction of cell contours from 3D image stacks. BMC Biol. 17: 38.
• Fu, Y., Gu, Y., Zheng, Z., Wasteneys, G. and Yang, Z. 2005. Arabidopsis interdigitating cell growth requires two antagonistic pathways with opposing action on cell morphogenesis. Cell 120: 687–700.
• Higaki, T., Hashimoto-Sugimoto, M., Akita, K., Iba, K. and Hasezawa, S. 2014. Dynamics and environmental responses of PATROL1 in Arabidopsis subsidiary cells. Plant Cell Physiol. 55: 773–780.
• Higaki, T., Kutsuna, N., Akita, K., Takigawa-Imamura, H., Yoshimura, K. and Miura, T. 2016. A theoretical model of jigsaw-puzzle pattern formation by plant leaf epidermal cells. PLOS Comput. Biol. 12: e1004833.
• Higaki, T. and Mizuno, H. 2020. Four-dimensional imaging with virtual reality to quantitatively explore jigsaw puzzle-like morphogenesis of Arabidopsis cotyledon pavement cells. Plant Biotechnol. 37: 429–435.
• Higaki, T., Sato, F. and Iba, K. 2021. Environmental responses of the membrane trafficking factor PATROL1 in the Arabidopsis stomatal complex. Cytologia 86: 101–102.
• Higaki, T., Takigawa-Imamura, H., Akita, K., Kutsuna, N., Kobayashi, R., Hasezawa, S. and Miura, T. 2017. Exogenous cellulase switches cell interdigitation to cell elongation in an RIC1-dependent manner in Arabidopsis thaliana cotyledon pavement cells. Plant Cell Physiol. 58: 106–119.
• Legland, D., Arganda-Carreras, I. and Andrey, P. 2016. MorphoLibJ: Integrated library and plugins for mathematical morphology with ImageJ. Bioinformatics 32: 3532–3534.
• Lin, W. and Yang, Z. 2020. Unlocking the mechanisms behind the formation of interlocking pavement cells. Curr. Opin. Plant Biol. 57: 142–154.
• Nagata, K. et al. 2016. In vivo 3D analysis of systemic effects after local heavy-ion beam irradiation in an animal model. Sci. Rep. 6: 1–9.
• Preisinger, C. and Heimrath, M. 2014. Karamba—A toolkit for parametric structural design. Struct. Eng. Int. 24: 217–221.
• Sampathkumar, A., Krupinski, P., Wightman, R., Milani, P., Berquand, A., Boudaoud, A., Hamant, O., Jönsson, H. and Meyerowitz, E. M. 2014. Subcellular and supracellular mechanical stress prescribes cytoskeleton behavior in Arabidopsis cotyledon pavement cells. eLife 3: e01967.
• Sapala, A., Runions, A., Routier-Kierzkowska, A. L., Das Gupta, M., Hong, L., Hofhuis, H., Verger, S., Mosca, G., Li, C. B., Hay, A., Hamant, O., Roeder, A. H., Tsiantis, M., Prusinkiewicz, P. and Smith, R. S. 2018. Why plants make puzzle cells, and how their shape emerges. eLife 7: e32794.
• Sapala, A., Runions, A. and Smith, R. S. 2019. Mechanics, geometry and genetics of epidermal cell shape regulation: Different pieces of the same puzzle. Curr. Opin. Plant Biol. 47: 1–8.
• Sato, F., Iba, K. and Higaki, T. 2021. Involvement of the membrane trafficking factor PATROL1 in the salinity stress tolerance of Arabidopsis thaliana. Cytologia 86: 119–126.
• Schneider, C. A., Rasband, W. S. and Eliceiri, K. W. 2012. NIH Image to ImageJ: 25 years of image analysis. Nat. Methods 9: 671–675.
• Zhou, L., Du, F., Feng, S., Hu, J., Lü, S., Long, M. and Jiao, Y. 2020. Epidermal restriction confers robustness to organ shapes. J. Integr. Plant Biol. 62: 1853–1867.