Interpolation of microtubules into cortical arrays during cell elongation and differentiation in roots of Azolla Finn Ata

Microtubules have been observed in the cortical cytoplasm of plant cells, adjacent to the plasma membrane, in many types of cell and tissue (see review by Hepler & Palevitz, 1974). In the majority of cases their orientation parallels that of the cellulose microfibrils being deposited in the wall, be the wall primary or secondary in nature. A few reports have also revealed the capacity of the cortical microtubule arrays to predict the form of microfibril deposition, the microtubules being deployed before similarly oriented microfibril deposition begins (Newcomb & Bonnett, 1965; Pickett-Heaps, 1967; Palevitz & Hepler, 1976). It is generally believed that cortical microtubules aid in orienting the cellulose microfibrils in the wall, possibly by influencing the alignment of cellulose synthetase enzyme complexes in, or on, the plasma membrane (Heath, 1974; Hepler & Palevitz, 1974). If the hypothesis that cortical microtubules participate in aligning cellulose microfibrils is correct, then the role of cortical microtubules is a fundamental part of the overall morphogenetic process in plants, the shapes of plant cells being governed largely by the cellulose component of the wall (Green, Erickson & Richmond, 1970). It seems therefore important to learn more about how the cell controls the development of its cortical microtubule arrays.

There are 3 phases in which interphase cortical microtubules are specifically deployed in the life of a cell: (i) during reinstatement of the arrays after cytokinesis; (ii) during maintenance of the arrays while the cell expands; and (iii) during cellspecific development of the arrays, as part of the terminal differentiation of the cell. The first of these phases has been considered in previous papers (Gunning, Hardham & Hughes, 1978a; Hardham & Gunning, 1978b) in which putative microtubuleorganizing regions are reported. The present paper is concerned with aspects of phases (ii) and (iii). Certain details of the organization of the arrays that are seen during these 2 phases have recently been obtained (Hardham & Gunning, 1978a; Seagull, 1978), but the underlying developmental processes are not understood. The cellular construction of the roots of the water fem, Azolla pinnata, R. Br., is such that the cells can be fully categorized in terms of their developmental history and stage of differentiation (Gunning, Hughes & Hardham, 1978b). Longitudinal sections provide views of files of cells, with their (mainly) transversely oriented microtubules visible against the side walls. The files are all identifiable, and traceable back to their respective initial cells. The initial cells are formed by a sequence of longitudinal cell-lineage divisions within the ‘merophytes’ that are the units of construction of the root, each merophyte being the product of one division of the apical cell. The initial cells elongate and pass through a cell-specific number of rounds of transverse proliferative division within their merophytes. The background information that is available has made it possible to select files of cells whose microtubules were undergoing ‘phase (ii)’ (above) developments, and others exhibiting features of ‘phase (iii)’. The present work describes changes in the microtubule arrays, along with features of cell differentiation that are necessary for purposes of interpretation.

Roots of Azolla pinnata R. Br. were fixed at room temperature in 2.5% glutaraldehyde in 0.025 M phosphate buffer, pH 7, for 2–5 h, rinsed in buffer, and post-fixed in 2% osmium tetroxide in the same buffer for 2 h. The tissue was dehydrated in graded acetone solutions, embedded in epoxy resin (Spurr, 1969) and sectioned for light and electron microscopy on a Reichert OMU3 ultramicrotome.

Roots were also cleared by boiling briefly in lactophenol, mounted on glass slides in fresh lactophenol and viewed in the light microscope using Nomarski interference-contrast optics. The roots were not squashed or flattened, the coverslip being held away from the slide by 2 strips of parafilm.

For low-temperature treatment, intact A. pimtata plants were transferred to culture medium in beakers surrounded by ice in a cold room. The temperature of the water in which the roots were immersed varied within the range 0–2 °C. Pressure treatments of 9.65 × 10⁴ kN m⁻² (14000 lb in. ⁻²) for 15 min were applied to intact plants using a French pressure cell. The pressure was released smoothly, but quickly, and the roots were immersed in fixative within 10 s of the onset of decompression. Deuterium oxide (D₂O) was used at 99.8% purity for 5 or 18 h before fixation in glutaraldehyde dissolved in D₂O. A 4-h treatment with colchicine at 5 × 10⁻³ M in culture medium was carried out in the dark.

Longitudinal sections of roots were collected on Parlodion-coated grids with 2-mm slots and stained with saturated uranyl acetate in 50% ethanol for 30 min, then in lead citrate for 10 min (Venable & Coggeshall, 1965). Electron micrographs of selected walls along cell files were taken at 10000–15000 times and printed to a final magnification of 25000—40000. Microtubule profiles adjacent to the longitudinal cell walls were counted, and expressed as the number per unit length of plasma membrane (Figs. 1, 9, 25) or as the total number present in the longitudinal dimension of each merophyte in the file (Figs. 2–5). The least squares method was used to find the slope, intercept and standard error of the lines of best fit through the values for the number of microtubule profiles per merophyte in merophytes of different lengths. Whether a given population of points was significantly different from other populations was determined by applying an F test (e.g. Edwards, 1961).

The cell files in the endodermis and pericycle differ from the remaining cell files in undergoing more (4–5) rounds of transverse proliferative division before differentiating. Their zones of proliferation therefore extend into basal regions of the root where the other files have ceased to divide and where differentiation is in progress. In both cases the major dimensional change is elongation, the diameter of the stele being nearly constant from its inception.

The endodermis forms a hexagon of 6 cells that delimit the stele (see fig. 6 of Gunning et al. 1978b). Its outer tangential wall is flanked by cells of the inner cortex, and its inner tangential wall lies adjacent to either a pericycle or an outer sieve element cell. Longitudinal sections of the root cut through the inner and outer tangential walls of 2 files of endodermal cells, on either side of the root, and counts were made of the microtubule profiles adjacent to the walls in both. The regions (up to 0.2 mm from the apex of the root) of the endodermal cell files in which microtubule counts were obtained encompassed up to 8 merophytes, commencing at the initial cell and its precursor, but were confined to the zone of the root containing only the first and second rounds of proliferative divisions. Five roots of lengths ranging from 0.27 to 0.63 mm were examined and for each root counts were made in 2 or 3 sections.

The data obtained have been used to construct graphs showing the number of microtubule profiles per unit length of plasma membrane (Fig. 1). The inner tangential wall is laid down before the outer, in the precursor cell that gives rise to the endodermis and the inner cortex, and in Fig. 1D, E, the first value plotted refers to this first representative of the inner tangential wall. The first pairs of values (Fig. 1 A-E) occur in the endodermis initial cell after the wall that separates the endodermis and inner cortex files has been laid down. Subsequent pairs of symbols refer to elongating cells along the files. It is evident that, despite being subjected to elongation by factors of up to 8-fold, the cells can maintain a reasonably constant number of microtubules per unit length of wall. The general pattern of microtubule distribution along the files is approximately the same in A-E, i.e. in the roots of different lengths. It is thus valid to normalize the data from different roots by using the lengths of the merophyte as an index of development, thereby pooling data and allowing information on the underlying developmental processes to be extracted.

Since a wall that is shorter than the shortest merophyte does not exist, the graphs do not begin closer to the origin than the length of the merophyte in which the 2 longitudinal walls were formed. However, mathematical extrapolation of the lines to their intercept on the vertical axes (Table 1) yields further information about the microtubule arrays. Two possible situations exist: (i) Mathematical extrapolation shows the lines to pass through the origin. This would mean that the number of microtubules μm⁻² is constant throughout, even at the first-formed wall, i.e. the rate of interpolation during elongation (microtubules μm^-1) = extent of original insertion at the first-formed wall (also expressed in microtubules μm^-1), (ii) Extrapolation shows the line to cut the vertical axis above the origin. The intercept then represents ‘extra ‘microtubules per merophyte that the first formed wall received over and above the rate of interpolation given by the slope of the line.

Three conclusions relating to the establishment and maintenance of the cortical microtubule arrays in the files of endodermal cells can be drawn from the information presented in Figs. 1 and 2, and Table 1.

1. Interpolation of microtubules into the arrays takes place throughout the zone of elongation that was examined, the rate of interpolation keeping pace with elongation. There is a decrease in the density of microtubules in the arrays during early stages of elongation (Fig. 1) associated with an ‘over-provision’ of microtubules adjacent to the newly formed walls (intercepts, Table 1). The density of microtubules inserted initially, after cytokinesis, is thus greater than is provided for by the subsequent rate of interpolation of microtubules as the wall elongates.

2. The number of microtubule profiles (Fig. 2) and the density of microtubules in the arrays (Fig. 1) is lower along the inner tangential wall than along the outer tangential wall. The slopes of the lines of best fit through the values along the inner and outer walls are 3.4 and 3.9 microtubules μm^-1 respectively (Table 1). Comparison of the 2 populations of points using an F test revealed that they were significantly different and not simply random samples from a common pool. The higher density observed along the outer tangential wall than along the inner wall is achieved by the initial establishment of a nearly 2-fold difference in the number of microtubules when the walls are first formed, and the subsequent enhancement of this difference by a slightly higher rate of interpolation of microtubules (Fig. 2, Table 1).

3. The six pairs of small symbols in the graphs in Fig. 1 refer to values of microtubule profiles μm^-1 for cells which had the majority of their cortical microtubules grouped in a pre-prophase band. These points are not markedly different from the points that represent cells in interphase, implying that the total number of microtubules is not very different at the 2 periods of the cell cycle.

During the sequences of formative divisions in the stele 6 cells lying adjacent and internal to the 6 endodermal cells are formed. Four of these become pericycle initial cells directly (designated as Targe pericycle’)-The 2 remaining outer cells, on opposite sides of the stele, divide radially, one product being an ‘outer sieve element’ initial cell and the other a pericycle initial cell (designated as ‘small pericycle’).

The numbers of microtubule profiles per merophyte adjacent to the inner and outer tangential walls of large and small pericycle cells are shown in Fig. 3. The F value calculated for the 2 populations of points for the large pericycle cells showed that they did not differ significantly, and the line of best fit through the combined points is drawn. This combined pool does, however, differ at a 95–99% confidence level from those along the outer wall of the small pericycle. The slopes and intercepts of the lines of best fit and the results of comparisons with the endodermis are given in Table 1. As in the endodermis cell files, the counts reveal that interpolation of microtubules occurs throughout merophyte elongation in the region of the roots examined, and, in the case of the small pericycle cells, follows the initial establishment of a high density of microtubules adjacent to the newly formed outer wall. By contrast the initial quota of microtubules against the young walls of the large pericycle cells is at almost the same density as the subsequent interpolations. There are significantly fewer microtubules at the outer walls of the small pericycle cell files than at either wall in the endodermis, but the combined values in the large pericycle do not differ from the combined values in the endodermis.

There are 2 layers of cortex cells in the root of A. pinnata: an inner ring of 6 cells (the inner cortex) which lies adjacent to the endodermis layer, and an outer ring of 12 cells (the outer cortex). In contrast to the cells of the stele, the diameter of the cells of both cortex layers increases markedly during elongation of the merophytes (see fig. 18 of Gunning et al. 1978b), this increase being the main factor contributing to the increase in the girth of the root.

The inner cortex cells display a very characteristic pattern of proliferative divisions. Two rounds of proliferative divisions generate 4 cells per merophyte. Then, unlike all other cell files in the root, where the individual cells in a merophyte appear to be equivalent, the cell at the acroscopic face of the merophyte divides transversely to give a mature complement of ‘5 cells in each merophyte. All proliferative divisions are completed relatively close to the root apex and the regions of the root examined in the present study extend beyond the zone of proliferation of the inner cortex cell files. It has thus been possible to detect a break in the pattern of microtubule distribution at the site of the second proliferative division, i.e. at the end of the proliferative zone for 3 of the 4 cells in each merophyte (Fig. 4). The points before and after this site (at a merophyte length of approximately 15 μm) were thus considered separately. The sets of points of the inner and outer walls in merophytes less than 15 μm in length did not differ signficantly when an F test was applied and a line has been fitted through the combined values, as in the case of the large pericycle cells. A high density of microtubules is established adjacent to the newly formed walls and is supplemented by an interpolation of approximately 2.7 microtubules μm^-1 in the zone of proliferative divisions of all cells in the merophytes. The combined pool does not differ significantly from that of the endodermis. About the time the majority of cells have completed their proliferative divisions the interpolation rate at the inner wall drops markedly and the points adjacent to both inner and outer walls become significantly different from those in the endodermis. All but one value at the outer wall display the same trend: the absence of other data points in the longer merophytes makes the significance of this single high value difficult to interpret. It would seem, however, that cessation of proliferative divisions is correlated with the onset of wall expansion that is not (at least along the inner wall) associated with augmentation of the numbers of microtubules.

Files of outer cortex cells complete 4 rounds of proliferative divisions. The data shown in Fig. 5 extend only into the zone of elongation following the second round of proliferative division, the third round occurring at a merophyte length of 50–60 μzm. As in the large pericycle and inner cortex cell files in the zone of proliferation, evidence for 2 separate populations of values along the inner and outer walls was not obtained and a single line has been fitted through all values. The density of microtubules at the newly formed wall is once again greater than the rate at which the microtubules are added into the arrays during cell elongation. The rate of interpolation, 2 microtubules μm^-1, is lower than observed in the other cell types and the set of values was found to be significantly different from that in the endodermis.

The final longitudinal division that generates the 2 outer sieve element initial cells on opposite sides of the stele has been described above. The initial cells complete only one round of proliferative division before they enter the phase of differentiation (Gunning et al. 1978b). As differentiation proceeds, the longitudinal walls become much thicker, in part through the deposition of transversely aligned microfibrils (Fig. 7).

The 2 files of outer sieve elements commence, and may complete, differentiation in the zone of the root that was examined in the preceding sections, i.e. in the zone in which the other cell types are undergoing rounds of proliferative divisions. Because differentiation proceeds acropetally along the files of cells, the length of the merophyte does not directly reflect the stage of development of the cells and data from roots of different lengths cannot be pooled as was possible when analysing cell files in zones of proliferative division.

The ultrastructure of sieve elements at an early stage of differentiation (Fig. 6), and at the stage of development showing the maximum numbers of microtubules (Figs. 7, 8) is illustrated. The transverse alignment of the microtubules is evident in both the transverse (Fig. 6) and the longitudinal sections (Figs. 7, 8) of the developing sieve elements. In Fig. 6, the microtubules are associated with small vesicles and electron-dense amorphous material near 2 of the corners of the cell.

As shown in Fig. 9, much more dense arrays of microtubules (up to 14 μm^-1) than those seen elsewhere develop during the wall-thickening phase of differentiation of sieve elements. Fig. 9 A shows data obtained from a young root, including developmental stages in which the density first increases and then declines. In the older root (Fig. 9B), corresponding phases of differentiation occur closer to the root apex (in shorter merophytes), and the data that were obtained show only the decline from a high density.

The distribution of microtubules in files of xylem elements in A. pinnata roots cannot be analysed without examining some of the complexities of the xylem development. The 2 files of xylem elements lie opposite each other in the centre of the stele (see figs. 5, 6 in Gunning et al. 19786). Initially the cells are triangular but after hydrolysis of the common wall the radial walls become less steeply angled and each element assumes a trapezoidal cross-section. A key feature is that the xylem initial cell does not divide before commencing differentiation. Stages of xylem development can be seen in Fig. 24 and are represented diagram-matically in Fig. 25 D.

Two vacuoles form at the ends of each xylem initial and remain in the cells until autolysis begins. Reconstructions from serial sections of young elements showed that one or both vacuoles may be composed of a complex series of membrane-bound channels and chambers with a lysosome-like appearance (Fig. 10). The endoplasmic reticulum (ER) that is observed between and over the developing wall thickenings may be tubular or may form 1-4 sheets (Fig. 10). At the end walls, 3-5 sheets of ER stack parallel to the cross-walls, being most prominent at the basiscopic end of the cell. ER remains close to the developing thickenings but the sheets at the end walls do not persist throughout differentiation. As development of the wall thickenings reaches completion and autolysis begins, the profiles of ER become dilated, containing granular material which appears more electron-dense than the surrounding cytosol, this difference in contrast being enhanced by the progressive autolysis of the cytosol (Fig. 11). During the later stages of formation of thickenings there is an elaboration of wall ingrowths resembling the ingrowths of transfer cells, firstly at the end walls (Figs. 11, 12, 24) and then also at many of the thickenings (Figs. 12, 24). Profiles of dilated ER often completely surround the profiles of the wall ingrowths (Fig. 11). The thickenings and the wall ingrowths both become lignified after early developmental stages when they stain metachromatically.

The secondary wall thickenings of the xylem are in the form of descrete rings, the shape of which can be seen in the cleared preparation in Fig. 13. The outer surface of the ring conforms to the triangular shape of the young xylem elements but greater deposition of wall material at the corners of the cell generates, in cross-section, a circular inner surface.

Secondary wall deposition begins at the corners of the xylem elements and progresses laterally around the cell. This pattern of initiation means that in longitudinal ultrathin sections of the faces of the walls of young elements, profiles of thickenings may be seen only on one side of the cell, or not at all, even though formation of thickenings may have commenced at all 3 corners of the cell.

Xylem differentiation, like that of the outer sieve element, progresses in an acropetal direction along the roots, i.e. in longer roots, the first xylem element with secondary wall thickenings is closer to the root apex, than in shorter roots (Figs. 14–20). It is evident from the graph in Fig. 20 that the rate of the progression of differentiation down the root is not constant in roots of different lengths. In a file of xylem elements in a very short root (less than approximately 0.5 mm) the first cell undergoing formation of thickenings is 5–6 cells closer to the apex than the first cell showing stages of autolysis. In longer roots, there are fewer cells spanning the stages of development. This acropetal advance means that, unlike the sites of proliferative divisions, xylem differentiation, like that of the outer sieve element, commences in cells (merophytes) of different lengths (Figs. 14–19, 21). Deposition of thickenings commences in progressively shorter cells until the root is about 1-4 mm in length (Fig. 21). Fewer wall thickenings are initiated in the shorter cells (Figs. 14–19). The linear relationship between the number of thickenings and the cell length at commencement of secondary wall deposition (Fig. 22) was fitted using the least squares method. There is thus a constant spacing, approximately 2 μm, between adjacent thickenings at the initial stages of deposition. Wall thickenings in the 2 adjoining xylem elements of a root are not necessarily ‘back-to-back’ (Figs. 14–19). Differential elongation along the root subsequently generates a graduation of separation of the rings along the root, the extent of separation increasing towards the base of the root. In mature regions of the roots where there has been a great deal of elongation, it is common to find that only fragments of each ring remain (Fig. 13). It is not known whether this results from incomplete fusion of the advancing edges of the thickening along the faces of the cell, or if the rings are broken during cell elongation.

As in the case of the outer sieve elements, the density of microtubules in differentiating xylem elements cannot be directly compared between roots of different lengths, as was possible for endodermal, pericycle and cortical cells in their zones of proliferative divisions. The developmental sequence is more condensed and is shifted more towards shorter merophytes, the older the root. In the xylem initials, and sometimes the next 1 or 2 cells (depending on the length of the root) the cortical microtubules are distributed relatively evenly along the longitudinal walls, their transverse orientation being evident in Figs. 10 and 23. As was the case in the cell types described in the preceding sections, the high density of microtubules seen adjacent to the newly formed longitudinal wall diminishes during the early stages of cell elongation (Fig. 25). In older cells, but still before any formation of thickenings is evident, the microtubules become grouped into bands, the number of which depends on the length of the cell. During this period there is an increase in the number of microtubules μm^-1, preceding the initiation of secondary wall deposition (Fig. 25). Deposition of secondary wall commences and proceeds beneath the groups of microtubules, which are sometimes surrounded by electron-dense, amorphous material. The maximum density of microtubules, about 7–8 microtubules μm^-1, occurs when the deposition of wall thickenings is approximately one half to two thirds complete. As wall deposition ceases and autolysis begins, the number of microtubules overlying the thickenings decreases (Fig. 25). During all stages of formation of thickenings, there are a few microtubules adjacent to the elongating primary wall between the thickenings.

The ultrastructural changes that occur during differentiation allow a series of arbitrary stages of development to be assigned to the cells in the files that are seen in each category of root length. These arbitrary stages in differentiation are shown diagrammatically below the graphs in Fig. 25, and related to trends in microtubule distribution. The overall picture is not unlike that shown by the sieve elements: cell differentiation involves a rise in the density of the microtubules, followed by a decline after the phase in which the walls become thickened. Had the thickening of the walls occurred throughout, instead of just locally, the maximum density of microtubules might well have matched that seen in the sieve elements.

The information given in previous sections enables a comparison to be made between the normal microtubule distribution in the roots of A. pinnata and that seen after treatments which affect the cortical microtubules. The endodermis was chosen for these comparisons and the number of microtubule profiles in the files of endodermal cells was counted in roots which had been subjected to a variety of treatments. In all cases the data are restricted to the zone of the root in which the initial cell is elongating before undergoing its first proliferative division. Untreated controls and treatments are therefore being compared using the same type of cell at the same stage of development.

Low temperature (Fig. 26 A) and high pressure (Fig. 26B) treatments both abolished the significant difference between the numbers of microtubules per merophyte against the inner and outer walls of the files of endodermal cells seen in untreated roots (Table 2). After 15 min at o °C the pattern of microtubule interpolation is similar to that in control roots but mathematical extrapolation of the lines of best fit shows the intercept to pass close to the origin (Table 2) reflecting the displacement of the curve to lower numbers of microtubules per merophyte. The values along the outer wall are significantly lower than in the control roots. After 15 min at 9.65 × 10⁴ kN m^-2 there is a highly significant decrease in the numbers of microtubule profiles μm⁻¹ accompanied by a halving of the apparent rate of interpolation of microtubules (Table 2).

After 4 h in a 5 ×10⁻³ M solution of colchicine the microtubule density was higher against the outer wall than against the inner wall (Fig. 26c) just as in controls (Fig. 2). The numbers of microtubules adjacent to the newly formed walls are approximately the same as in untreated roots, but the number per merophyte does not increase as much as it does in controls (Table 1). The populations of points for inner and outer walls are significantly different from those in control roots (Table 2).

When grown in D₂O for extended periods, intact plants of A. pinnata displayed a general inhibition of growth of both fronds and roots. The numbers of microtubules in the endodermal initial cells following 5 h treatment with D₂O do not differ significantly from those in untreated roots (Fig. 26D, Table 2). After 18 h in D₂O, the pattern along the outer wall is still the same as in controls but the distribution against the inner wall is highly irregular, in fact showing fewer microtubules in the older cells than in the younger (Fig. 26E, Table 2).

Although there are many reports of changes in orientation, and to some extent frequency, of microtubules (Chafe & Wardrop, 1970; Palevitz & Hepler, 1976; Robards & Kidwai, 1972; Robinson, Grimm & Sachs, 1976; Sachs, Grimm & Robinson, 1976; Sawhney & Srivastava, 1975; Schnepf, 1972, 1973, 1974; Schnepf, Deichgrâber & LjubeSié, 1976; Stetler & DeMaggio, 1972; Wada & O’Brien, 1975; Westafer & Brown, 1976), there have been no previous quantitative studies of microtubule production during the cell cycle in growing cells of higher plants. The counts presented here for endodermis, pericycle and cortex cells unequivocally demonstrate the interpolation of microtubules as the cells elongate in the zone of proliferative divisions. Two alternative hypotheses - that additional microtubules are produced post-cyto-kinetically at each round of division, and become ‘diluted’ as the cells expand, and that the short microtubules of which the arrays are built (Hardham & Gunning, 1977, 1978 a) elongate and therefore overlap more so that more cross-sectional profiles are encountered, are not supported by the data. There is no evidence for the former phenomenon, except when the longitudinal walls are first laid down, and the sieve and xylem elements in particular display addition of microtubule profiles in the absence of cell division. As to the latter hypothesis, elongation factors up to 8-fold would have to be invoked, and extensive serial sectioning and measurement of average lengths indicates that this does not occur (Hardham & Gunning, 1978 a). Rather, interpolation must occur more or less in step with cell elongation.

A comparable observation to that seen in A. pinnata has been made in Sphagnum leaflets (Schnepf, 1972, 1973, 1974), where immediately following cell division, the distribution of cortical microtubules was uneven around the cell profiles, the majority of microtubules lying adjacent to the new cell wall. Later stages of development showed more even distributions of the microtubules around the periphery of the cells. It was proposed that the ‘post-cytokinetic’ layer of microtubules in Sphagnum might ‘stiffen the young wall until it has attained sufficient stability’ (Schnepf, 1974). Expressing this in other terms, the greater initial number of microtubules could be correlated with the initiation of cellulose deposition and orientation at the new cell plate.

The number of microtubules initially established at the new cell wall and the rate of interpolation of microtubules during cell elongation together determine the density of the microtubules in the cortical arrays. Differences in one or both of these processes lead to the formation of arrays which can be characteristic not only of the cell type but also of the wall within the cell. While the combined points in the large pericycle and in the inner cortex cell files, before the second proliferative division, are similar to those of the endodermis, the values for the small pericycle outer wall and the outer cortex combined points differ significantly from those in the endodermis. Although the initial density is similar to that at the outer wall of the endodermis, a lower rate of interpolation during cell development in the small pericycle and outer cortex cells produces arrays with lower densities of microtubules. Within the endodermis cells, a higher rate of interpolation at the outer wall than at the inner, maintains the higher density at the outer wall that was initially established when the walls were formed.

These differences in microtubule densities characteristic of particular cell walls may correlate with the amount of cellulose that is being deposited: greater deposition of cellulose may require greater numbers of microtubules to control orientation. A 1 : 1 relationship between microtubules and microfibrils of chitin has been observed in Poteriochromonas (Schnepf, Rôderer & Herth, 1975) and in Oocystis, a ratio of microtubules to cellulose microfibrils of 1:25 has been estimated (Robinson et al. 1976). That the endodermal walls in A. pinnata are stronger than those of the cortex cells is indicated by the very limited increase in the diameter of the stele. The inner and outer cortex walls, on the other hand, undergo much expansion in all dimensions and this is the main contributing factor in the increase in girth of the root. From the time the outer cortex initial cell has completed its first proliferative division the density of its microtubule arrays is almost half that found at the outer wall of the endodermis cells. In the cells of the inner cortex in the zone of proliferative division the rate of interpolation is intermediate between that of the outer cortex and the endodermis, but following the completion of the proliferative division, the rate falls markedly and the density of the arrays diminishes as the cells continue to elongate and swell.

Examination of the zone of differentiation for 3 cell types in the A. pinnata root revealed 3 distinctive modes of development of the cortical microtubule arrays. In contrast to the situation in the inner cortex cell files, where the completion of the proliferative divisions heralds a decrease in the density of the arrays, the commencement of differentiation of the sieve and xylem elements is accompanied by a marked increase in numbers of cortical microtubules against the longitudinal walls. In the sieve elements, the dense microtubular array overlies the entire area of the longitudinal walls, whereas in the xylem elements the microtubules group and form discrete bands spaced along the wall. In both cases the increase in numbers of microtubules is correlated with an increase in the amount of wall deposition: in the sieve element cells the longitudinal walls are thickened along their entire length, while in the xylem elements bands of secondary wall material form beneath the grouped microtubules.

The grouping of the microtubules prior to wall deposition in A. pinnata, as in Triticum (Pickett-Heaps, 1967), is part of the mechanism that determines the sites where the secondary wall thickenings form in the young xylem elements. Further evidence that the microtubules are an essential part of the system comes from the effect of removal of the microtubules by colchicine treatment and the consequent abnormal thickening formation (Hepler & Fosket, 1971; Hardham & Gunning, in preparation). In A. pinnata xylem differentiation advances acropetally down the root and the back-to-back positioning of the thickenings in adjacent elements observed in Coleus (Hepler & Fosket, 1971), does not occur routinely. It was found instead that a constant spacing of the thickenings was the predominant feature: at different stages of root growth, formation of wall thickenings is initiated in cells of widely different lengths, but always the thickenings are spaced approximately 2 μm apart. Although different numbers of thickenings form in cells of different lengths, the maximum density of microtubules is the same.

It is of interest that Goosen de Roo (1973) found that the plasmalemma is pulled in at the sites of the groups of microtubules when developing xylem elements in Cucumis are plasmolysed. Arrays of cortical microtubules overlying xylem thickenings, as in other situations, are composed of short overlapping microtubules (Hardham & Gunning, 1978 a) and Goosen de Roo’s observation indicates that there are interactions of microtubules with one another and with the plasma membrane that tend to maximize the extent of overlapping of the microtubules. Further support for this explanation could be gained by application of the form of analysis described in the present paper. If plasmolysis allowed more extensive overlapping of the microtubules, the density of the arrays should increase and this would be reflected by a greater than normal number of microtubule profiles along the longitudinal walls.

Possible modes of action of low temperature, high pressure, colchicine and D₂O on cortical microtubule arrays in A. pinnata root tips emerged from studies which utilized the technique of serial sectioning to obtain data on the average length of microtubules, the number of very short microtubules, and the occurrence of C-shaped terminations in the arrays (Hardham & Gunning, 1978a). However, the interpretations were uncertain because of lack of information on the numbers of microtubules per cell and because it was not possible to compare treated and untreated cells of the same type at the same stage of development. The present work overcomes some of this uncertainty.

An agent which causes the depolymerization of microtubules might act upon the arrays in one of 3 ways: (i) No effect on microtubules that have been interpolated but selective destruction of the initial quota. This would be indicated by a lowering of the intercept while the slope of the graph remained unchanged, (ii) No effect on the initial quota of microtubules but selective destruction of later interpolations. The slope of the graph would be reduced but the intercept unaltered, (iii) Non-selective removal of both categories of microtubules. Both the intercept and slope would be reduced. The counts show a general reduction in the numbers of microtubules in the endodermal initial cells following the low temperature, high pressure and colchicine treatments, giving support to the previous interpretations of the effects of these treatments on microtubules in the roots of A. pinnata (Hardham & Gunning, 1978 a). All 3 treatments were found to reduce the average length of the microtubules in the arrays but each exerted a different effect on the number of C-shaped terminations and the proportion of very short microtubules. The present analysis reveals that the reduction in the overall numbers of microtubules may also be accomplished in a different manner by each agent. The distributions of microtubules following low-temperature, high-pressure and colchicine treatments approximate to the general cases (outlined above) (i), (iii) and (ii), respectively. There is thus, in addition, evidence for the selectivity in the action of each agent, as has been reported in other systems (Brinkley & Cartright, 1975; Salmon, Goode, Maugle & Bonar, 1976): the ‘initial quota’ and the ‘subsequent interpolations’ that have been detected by the counting procedures must be somewhat different as judged by their differential sensitivity. The arrays at the inner and outer faces of the cells may also be responding differently.

The unexpected complexity of the results makes comparisons between treated and control data very difficult. It is clear, however, that it is not sufficient to compare any cell at any stage of development if meaningful comparisons are to be obtained. In most previous work, random sections of random cells in root tips and other tissues have been used. The uncertainty as to the effects of D₂O on microtubules in plant cells (Burgess & Northcote, 1969; Waber & Sakai, 1975; Schnepf et al. 1976; Hardham & Gunning, 1978a) may be due, at least in part, to failure to meet this requirement. The data presented here show that D₂O, like the treatments which cause depolymerization, may be selective in its action. Increased numbers of microtubules in the ‘initial quota’ established soon after cytokinesis are observed against the outer wall, after the 5-h treatment and against the inner wall after the 18-h treatment.

It would be seen from these data that D₂O is capable of enhancing polymerization of microtubules during early stages of the cell cycle when the cell is reinstating its cortical microtubule arrays, which is in accord with the proposal that D₂O does not enhance assembly onto existing microtubules, but rather, the generation of new ones (Hardham & Gunning, 1978 a).

Comparison of the total lengths of microtubules in cells at different stages of development and knowledge of the age difference between adjacent merophytes (Gunning et al. 1978 b) then allows the rates of net production of microtubules to be calculated. A sample calculation using data obtained during early stages of secondary wall deposition in a file of developing xylem elements is given below.

Data for other roots, cell types and stages of development are given in Table 3. The method assumes that all faces of the cell possess uniformly dense arrays. The inner and outer faces of the endodermis, being significantly different, have been averaged in order to make the calculations.

The total length of microtubules per cell determined in this way exceeds previous estimates made in plant tissues: 4000–7000 μm in developing sieve and xylem elements in A. pinnata compare to 500–900 μm in Sphagnum leaflet cells in interphase (Schnepf, 1973), but it is similar to the total length of 6000–7000 μm estimated by indirect immunofluorescence microscopy with tubulin antibody in 3T3 cells in interphase (Hiller & Weber, 1978). Rates for the net production of microtubules during cell development in other systems are not available for comparison with the 1–2 μm min^-1 estimated here. It is, of course, not known when in the cell cycle tubulin polypeptides are formed, or whether more tubulin is made than becomes incorporated into microtubules, as is the case in 3T3 cells where about 60% of the cellular tubulin is not assembled during interphase (Hiller & Weber, 1978); however, the observed rate of growth of formed microtubules in A. pinnata is equivalent to the synthesis of 25–50 molecules of tubulin per second. Another unknown factor that is not taken into account in the calculations is whether the population of microtubules is subject to ‘turnover’.

Interpolation of microtubules throughout the arrays during cell elongation in zones of proliferative divisions and of differentiation implies that the capacity to initiate the polymerization of microtubules must exist along the length of the cells during post-cytokinesis stages of the cell cycle, or else that newly polymerized microtubules can migrate to become uniformly distributed. Specialized activity at cell edges, in the form of clusters of microtubules, vesicles and electron-dense amorphous material was observed in the present study, during the differentiation of both outer sieve and xylem elements. Formation of xylem thickenings was also observed to begin at the 3 corners of the cell. All of these observations accord with the previously published evidence that the initiation of cortical microtubules in cells of A. pinnata root tips may be concentrated at the cell edges (Gunning et al. 1978 a). The previous work rested upon ultrastructural observations made using cells that were undergoing the conspicuous process of reinstating their cortical arrays, but the present work indicates that the specialized edge zones may act as microtubule-organizing regions at later stages in the cell cycle, though perhaps not so intensively as immediately after cytokinesis.

The rates of net synthesis of microtubules during interpolation, up to 2.0 μm min^-1, can be compared with the rate of microtubule reinstatement after cytokinesis. Taking, for example, the reinstatement of the array following the first proliferative division of the endodermis initial cell, a total microtubule length of 1500 μm is polymerized in each daughter cell. If it takes approximately one quarter of the cell cycle, i.e. 1 h, to reinstate the array, then an overall rate of 25 μtm min^-1 is achieved. Obviously, the rates of polymerization at the level of individual microtubules will be less than this and will depend on the number being assembled at any one time; rates of elongation up to 1–2 μm min^-1 have in fact been estimated (see Hardham & Gunning, 1978 a).

The quantitative data presented here not only help to explain why initiation of microtubules is more conspicuous soon after cytokinesis than it is later, but also hint at the properties of the microtubule-organizing system of the plant cell. As pointed out elsewhere (Gunning et al. 1978a), it is not possible, with the present methods, to rule out the possibility that new microtubules can form against the faces of the cells: the available evidence merely indicates that they can form along the edges. It was envisaged that a form of feedback control of microtubule formation might exist, in which the rate of generation of new tubules would decelerate as the surface of the cell became progressively more occupied by the developing array. The present results are consistent with this concept, with cell elongation opening up gaps in the array, and thus allowing more polymerization to occur, so that formation of microtubules parallels wall expansion. Clearly, however, there are other controls. Continued polymerization during expansion need not occur, as shown by the behaviour of the cortex after completion of proliferative divisions. The fact that different walls within one cell can become equipped with arrays of different density (as in the endodermis) speaks of the existence of more than one active local control region (such as the edges that circumscribe the individual walls), and of the ability on the part of the cell to control quantitatively the activity of each control region. Quantitative controls that lead to differences between cells are also evident, as is the capacity to generate localized arrays such as seen in developing xylem. All of these attributes can be accommodated in the hypothesis that the edges of the plant cell are its microtubule-organizing regions, by invoking (i) selective activation of particular edges to generate arrays on selected faces, (ii) quantitative controls, and (iii) localized activation, as against activation along the whole length of the edge.