Reassessing the Roles of PIN Proteins and Anticlinal Microtubules during Pavement Cell Morphogenesis

Abstract

The leaf epidermis is a biomechanical shell that influences the size and shape of the organ. Its morphogenesis is a multiscale process in which nanometer-scale cytoskeletal protein complexes, individual cells, and groups of cells pattern growth and define macroscopic leaf traits. Interdigitated growth of neighboring cells is an evolutionarily conserved developmental strategy. Understanding how signaling pathways and cytoskeletal proteins pattern cell walls during this form of tissue morphogenesis is an important research challenge. The cellular and molecular control of a lobed cell morphology is currently thought to involve PIN-FORMED (PIN)-type plasma membrane efflux carriers that generate subcellular auxin gradients. Auxin gradients were proposed to function across cell boundaries to encode stable offset patterns of cortical microtubules and actin filaments between adjacent cells. Many models suggest that long-lived microtubules along the anticlinal cell wall generate local cell wall heterogeneities that restrict local growth and specify the timing and location of lobe formation. Here, we used Arabidopsis (Arabidopsis thaliana) reverse genetics and multivariate long-term time-lapse imaging to test current cell shape control models. We found that neither PIN proteins nor long-lived microtubules along the anticlinal wall predict the patterns of lobe formation. In fields of lobing cells, anticlinal microtubules are not correlated with cell shape and are unstable at the time scales of cell expansion. Our analyses indicate that anticlinal microtubules have multiple functions in pavement cells and that lobe initiation is likely controlled by complex interactions among cell geometry, cell wall stress patterns, and transient microtubule networks that span the anticlinal and periclinal walls.

Figure 1.

Pavement cells from pin1-1 null mutants are indistinguishable from wild-type (WT) cells. Representative images of wild-type (top) and pin1-1 (bottom) cotyledon pavement cells are shown. The time points at which the seedlings were imaged are labeled at top. Bars = 50 μm.

Figure 2.

PIN3, PIN4, and PIN7, but not PIN1, are expressed in expanding cotyledon pavement cells. A, PIN1 is expressed in root cells but not in 2-DAG cotyledon pavement cells. B, PIN3, PIN4, and PIN7 are expressed in 2-DAG cotyledon pavement cells. C, Expression of PIN3, PIN4, and PIN7 in the root cortex. FM4-64 was used to visualize the plasma membrane when necessary. Bars = 50 µm.

Figure 3.

Pavement cell shape change at high temporal and spatial resolution. A and B, Example image field at time point 1 (A) and time point 11 (B) ∼9 h later. For full time lapse, see Supplemental Movie S1. C and D, Example image field at time point 1 (C) and time point 11 (D) ∼10 h later. For full time lapse, see Supplemental Movie S2. Yellow labels are defined as segments that produce a new lobe(s). White labels are defined as lobed segments that grow but do not have a symmetry-breaking event. Bars = 10 µm.

Figure 4.

Development of a method to quantify the timing and location of lobe initiation in pavement cell segments that span three-way cell junctions. A and B, Live-cell images of the cell boundaries at time points 1 and 11, respectively. The total elapsed time between these two images is 9.8 h. Bar = 20 µm. C, Segmented cell boundaries illustrating different types of wall reshaping that occur in developing cells. Yellow boxes label the time points and locations of new lobe formation. Red and green segments correspond to the segments shown in A and B, respectively. D, A plot of lobe height defined as the orthogonal distance from the cell boundary to a reference line that connects the two three-way junctions. E, Absolute value of the lobe height in D color coded with a heat map to reflect differing lobe height along the segment length. F, Heat map of lobe height of segment III as a function of normalized segment length (x axis) over time (y axis). The red star shows new lobe formation.

Figure 5.

Quantification of microtubule signals and their positions along the segment. A, Live-cell image of a lobing segment: plasma membrane marker in magenta and microtubule marker in green. Cell volume is tilted to highlight the anticlinal wall. B, Projected face-on view of microtubules along the segment. C, The segment is digitally straightened and resliced to provide a face-on view of the microtubules along the anticlinal wall. D, The signal from C is summed as a function of segment location and normalized from 0 to 1. E, Heat map of the normalized microtubule signals as a function of normalized segment length (x axis) over time (y axis). The red star labels the time point and location of microtubule data that are displayed in C and D.

Figure 6.

The microtubule peak occupancy at the lobe apex is not significantly higher than that of the adjacent flank regions. A, Example of lobe subdivision based on the lobe width (W) at half-maximal lobe height. The apex is 1/2 W centered at the lobe apex. The flank region total width is 1/2 W surrounding the apex region. B, Population-level analysis of microtubule signal peak occupancy for recently formed lobes (forming) and lobes present at the start of the time series (formed). Letters represent statistically equivalent groups.

Figure 7.

Persistence plots of microtubule signals show unequal accumulation of microtubule signals along a segment. A, Microtubule persistence plot of a segment that forms two new lobes. B, Absolute value of lobe height along the segment analyzed in A. C, Microtubule persistence plot for a lobed segment. D, Absolute value of lobe height along the segment analyzed in C. PCC values in B and C correspond to the correlation analysis between the microtubule persistence signal and the final segment shape.

Figure 8.

The cortical density of periclinal microtubules does not differ in opposing cells prior to lobe formation. A, Image of periclinal microtubules at the interface of a segment that will form a lobe. B, Processed microtubule image using the cortical occupancy method of Higaki et al. (2010). C, Skeletonized version of the processed image showing segmented microtubules. D, Overlay of the GFP:TUB6 image from B and the microtubule skeleton in C. The ROIs of the future convex (V) and future concave (C) cells are shown at the lobing interface. E, Box plots of the mean cortical occupancy of microtubules in paired convex (V) and concave (C) cells at the lobing interface. ***, Significant difference between the paired cell regions by Student’s t test (P < 0.05).

Figure 9.

Microtubule signals are significantly lower near three-way junctions compared with a central region of the segment. A, Live-cell image of a segment with microtubule marker in green and plasma membrane marker in magenta. B, Microtubule signal intensity plot of the segment in A. Orange is the area under the microtubule signal curve 1 µm from the three-way junctions. Light blue is the area under the microtubule signal curve 2 µm from the cortex centered at the segment midpoint. C, Box plot of the microtubule signal intensity from three-way junctions and center of segments for both lobing and lobed segments. Letters represent statistically equivalent groups.

Figure 10.

The domains of the anticlinal wall that underlie anticlinal microtubule bundles (AMBs) are not thicker than adjacent cell wall domains lacking microtubules. A, TEM image of a paradermal section that was perpendicular to the anticlinal wall. B, Thresholded image of A in which the cell wall is color coded from thinnest (yellow) to thickest (white). Microtubules are indicated with white circles. C, Box plot of cell wall thickness values in cell wall domains that overlaid the AMBs (+) and in adjacent regions of the wall not associated with microtubules (−).