On the differential divisions and preprophase microtubule bands involved in the development of stomata of Vigna Sinensis

The stomata are bicellular epidermal formations exhibiting a structural symmetry and an integrated specialized function, characters which are the outcome of a precise developmental sequence and a complicated differentiation. The ontogeny of a stoma commences with one or more differential divisions of some protodermal cell(s), which separate the guard cell mother cell (GMC) and the subsidiary cell (s), and is completed by a non-differential (symmetrical) division which produces the two guard cells. In different plants the number and mutual orientation of the planes of these divisions vary considerably and result in the appearance of a wide spectrum of stomatai patterns (for a review see Fryns-Claessens & Van Cotthem, 1973).

In the cortex of dividing cells of Triticum, among which are those of developing stomata, Pickett-Heaps & Northcote (1966a, b) discovered an ordered and complex structure, the preprophase microtubule band (PMB), which prior to mitosis indicates the plane of orientation of the cell plate as well as its sites of junction with the parent walls. In asymmetrical divisions the PMB was localized at the polar end of the cell, surrounding the nucleus centrally or on one side, while in symmetrical ones it occupied an equatorial position in relation to both the cell and the nucleus. Although further work on dividing cells of higher plants confirmed the existence of the PMB and many of its distinguishing features (Pickett-Heaps, 1969a, b, c;Deysson & Benbadis, 1968; Cronshaw & Esau, 1968; Esau & Gill, 1969; Evert & Deshpande, 1970; Schnepf, 1973; Packard & Stack, 1976; Galatis & Apostolakos, 1977), some investigators raised doubts about the consistency of the properties of the PMB in foreshadowing the plane and the position of cell plate formation.

In root meristematic tissue of Phleum, Burgess & Northcote (1967) observed that the PMB always showed a symmetrical situation in premitotic cells, although asymmetrical divisions occurred in the rhizodermal cells. The authors suggested that the structure orientates the nucleus prior to mitosis (see also Burgess & Northcote, 1968; Burgess, 1970; Singh, 1977; Singh, Shaw & Hollins, 1977). In addition, Singh (1977) and Singh et al. (1977) reported that in the rectangular GMC of Saccharum the PMB is aligned vertically to the plane of their division and therefore it is not in accordance with the plane and the position of the cell plate.

As far as we know, fine-structural studies on stomata development have been carried out mainly on monocotyledons and especially on those of the family of Gramineae (Pickett-Heaps & Northcote, 1966a; Pickett-Heaps, 1969a; Kaufman, Petering & Soni, 1970; Kaufman, Petering, Yokum & Baic, 1970; Srivastava & Singh, 1972; Palevitz & Hepler, 1974, 1976; Ziegler, Shmueli & Lange, 1974; Singh, 1977; Singh et al. 1977), while dicotyledonous stomata have been studied only to a limited extent (Landré, 1969a, b, 1970, 1972; Singh & Srivastava, 1973). Concerning the role of the PMB in the development of stomata of dicotyledons, no information is available, to our knowledge, except for a brief mention of its existence in Sinapis alba (Landré, 1972).

In the present report two aspects of stomatai development in young primary leaves of Vigna sinensis were examined in detail: (a) the extent of manifestation of premitotic cell polarity and the asymmetrical character of the differential divisions which occur ; and (b) the organization, orientation and disposition of PMB in dividing cells involved in stomatai development, as well as in the ordinary protodermal cells. Among them particular attention has been devoted to the meristemoids, since they are capable of dividing differentially without any apparent polarization and

Seeds of Vigna sinensis were germinated in quartz sand in plastic containers, either in a darkroom at 25 ± 1 °C or under controlled conditions of light and temperature. In the latter, during 12-h photoperiods, the developing seedlings received a mixed fluorescence and incandescent illumination (light intensity 22000–24000 erg/cm’.s (2·2 —2·4 × 10² J m⁻².s)). The temperature was maintained at 25 °C during the dark and 20 °C during the light periods.

Small pieces of primary leaves of 2-to 6-day-old seedlings were fixed in 3–5% glutaraldehyde in 0-025 ^M phosphate buffer, pH 6·8–7·0, at room temperature for 3 h. After thorough washing in buffer the specimens were postfixed in 1% osmium tetroxide in the same buffer, pH 7·0, at 4 °C for 5 h. The samples were infiltrated and embedded in Epon 812 or Durcupan ACM (Fluka). Thin sections of material were stained with a 4% solution of uranyl acetate in 70% ethanol followed by lead citrate (Reynolds, 1963), and examined with a Philips 300 or a Hitachi HS-8 electron microscope. Sections (1–2 μm) cut from the same blocks on an LKB Pyramitome were stained with hot 1% toluidine blue in 1% borax solution and examined by light microscopy.

cp, cell plate; dv, dictyosome vesicle; er, endoplasmic reticulum; gmc, guard cell mother cell; m_I, meristemoid I; m_II, meristemoid II; n, nucleus; s, subsidiary cell; v, vacuole.

Examination of 3-to 5-day-old primary leaves of Vigna sinensis seedlings grown under both dark and light conditions revealed that mesoperigenous and mesogenous modes of stomata development operate while a limited number of large perigenous stomata has already been formed. The time of operation of each mode of stomatai development appears to be related to a particular developmental stage of the leaf.

In the mesoperigenous process two differential divisions result successively in the formation of meristemoids I and II and one subsidiary cell. Meristemoid II functions as a GMC (Figs. 1 C, D, 6, 7, 9). In the mesogenous process an additional differential division yields another subsidiary cell and meristemoid III, which will become a GMC (Figs, 1 A, B, 10). The perigenous stomata originate by one differential and one symmetrical division (Figs, 1E, 8). Apart from the above predominating stomatai developmental lineages, in 1–2 day-old leaves a few large stomata resulting directly from the symmetrical division of an ordinary protodermal cell were observed (Galatis, unpublished data). In 6-day-old seedlings grown in light, the mesogenous mode of stomata formation predominates, while in the ones grown in darkness mesoperigenous stomata are still formed. Mesogenous stomata arising through 4 differential divisions were also recorded in 8-day-old leaves, grown in light.

The plane of orientation of the second differential division varies among different meristemoids I (stomatai initials) on the same epidermis, particularly at the first stages of leaf development. Two distinct modes of alignment of the plane of cell division were recognized. In the first there is a curved cell plate, which in paradermal sections appears joined exclusively to the wall set up during the first differential division (Figs. 1,2B, D, 3). In the second mode one end of the cell plate meets the wall laid down during the first differential division, while the other meets with one of the older walls (Figs, 1 2A, C, 5). Lens-shaped meristemoids are formed in the first case and triangular ones in the second (Figs. 3, 5).

In most cases the third differential division of the mesogenous process is performed by a mitotic spindle lying more or less parallel to that of the second division. In addition, the GMC always divides in a plane nearly parallel to those of the second or third differential divisions. The majority of the mature stomata of mesoperigenous and all those of mesogenous origin are of paracytic form (Figs. 9, 10).

In young leaves of Vigna sinensis it was not always possible to identify the type of meristemoids or the developmental mode to which they belong, since (a) more than one developmental mode operates simultaneously, (b) the asymmetry of differential divisions varies considerably, and (c) the larger daughter cell of the first differential division divides again. For instance, if a meristemoid of a 4-day-old leaf is flanked on one side by a subsidiary cell and on the other by an ordinary proto-dermal cell, it may represent a meristemoid II of the mesogenous process or the cell in the mesoperigenous mode of development which will act as a GMC (Fig. 3).

The protodermal tissue among the veins consists of rectangular or polygonal cells, which contain the usual complement of organelles of meristematic cells and a rather developed vacuolar system (Fig. 11). The level of vacuolation varies considerably among plants of different ages as well as between plants of the same age grown in dark and light conditions. A significant number of subplasmalemmal microtubules are encountered in all interphase ordinary protodermal cells. In paradermal sections they are distributed along all the anticlinal walls and traverse the cytoplasm vertically to the leaf surface.

As the cells prepare to enter mitosis, the interphase microtubular population disappears and is replaced by a typical preprophase microtubule band (PMB). This band was present in all cells in which the nucleus showed signs of chromatin condensation (Fig. 11). In serial paradermal sections 2 groups of microtubules arranged in 2–5 layers were observed in the mid-region of the cell that is assumed to coincide with the position of the future cell plate (Fig. 11). Each group consists of about 30–60 microtubules of the same orientation which spread out usually 1–2 μm along the walls (Figs. 12, 13). When the sections pass through the cortical cytoplasm of the periclinal walls, the microtubules are cut obliquely and finally longitudinally. The parietal cytoplasm does not contain microtubules other than those of PMB. Ribosomes are excluded from the intermicrotubular spaces of the bands. In the close vicinity of the microtubules, as well as among them, vesicles of dictyosomal origin and portions of endoplasmic reticulum (ER) are localized (Figs. 12–15). The relationship of dictyosome vesicles to the microtubules appears to be significant. For the association with ER nothing can be inferred with certainty; the cortical cytoplasm is always very rich in ER elements.

Microtubules of the band were infrequently observed bridged to the plasmalemma or to one another. The cytoplasm between the band and the nucleus contained scattered microtubules displaying different orientations. In Vigna sinensis protodermal tissue the PMB is retained up to advanced prophase, by which time a prominent perinuclear microtubular frame has been organized (Fig. 14). In dividing highly vacuolated cells the position of the PMB coincides with that of the phragmosome.

A careful scrutiny of serial sections of telophase ordinary protodermal cells dividing non-differentially sometimes revealed detectable differences in the size of the daughter cells and an unequal distribution of vacuoles between them (Fig. 28 A, p. 24).

Before the first differential division the nucleus migrates to one end of the cell, while the vacuoles are displaced to the other (Figs. 16, 19); however, some small vacuoles may remain at the polar end of the cell and after completion of cell division are included in meristemoid I. The cytoplasm in these cells seems to be unequally distributed to the 2 cell poles. Most of the plastids and mitochondria are located in the perinuclear cytoplasm.

In most protodermal cells undergoing the first differential division the PMB shows an obviously asymmetrical disposition, surrounding the nucleus in an equatorial or an eccentric position (Figs. 16–18, 19–21). In few instances of smaller ordinary protodermal cells the nucleus occupied a more or less central position, and the PMB was the only cytoplasmic structure exhibiting asymmetry. The structural features of the PMB are similar to those described in non-differential divisions.

The mitotic spindle of the first differential division is formed in the region in which the nucleus has been previously located and is therefore asymmetrically positioned. No structural differences between the anaphase groups of chromosomes or the ER distribution at the poles of the cells were observed. Decondensation of the telophase chromosomes occurs at the same time in the 2 daughter nuclei. From the first differential division 2 unequally sized cells are derived; a smaller one which represents the initial cell of a stoma (meristemoid I) and a larger protodermal cell (Figs. 2, 4). In young leaves the latter retains meristematic potential and redivides either differentially to give rise to other meristemoids of stomata or non-differentially to form additional protodermal cells (Figs. 6, 9, 10). In older leaves this cell may differentiate directly to a typical epidermal cell, rapidly becoming vacuolate. On the other hand the meristemoid I acquires protoplasm mainly. The plastids and mitochondria appear equally distributed between the 2 daughter cells.

In the highly vacuolated ordinary protodermal cells dividing differentially, the plane of the PMB corresponded in position to that of the phragmosome (Figs. 19–21).

It was of particular interest that in leaves of Vigna sinensis meristemoids I appeared able to induce the formation of other meristemoids close to them. The nuclei in the neighbouring ordinary protodermal cells dividing differentially were observed to lie close to the inducing meristemoid. As a result the new meristemoid I was adjacent to the pre-existing one. In this manner, groups of three, four or five adjacent meristemoids were commonly formed (Fig. 30, see also Fig. 10).

After the conclusion of the first differential division the subplasmalemmal microtubules reappear in both the meristemoid I and the other daughter cell. In the former cell they are evenly distributed along its walls and oriented perpendicularly to the leaf surface. In the latter cell a relatively higher number of microtubules in comparison to the ones of the older walls lines the cytoplasm adjoining the recently formed wall. The microtubules are frequently linked by bridges to the plasmalemma.

The meristemoid I is a small cell the nucleus of which fills a significant part of the cell interior (Fig. 22). It usually possesses some small vacuoles which are placed close to the periclinal walls, especially the external one. In premitotic cells the vacuoles are restricted in the region which will be included in the subsidiary cell after the termination of cell division; therefore, some structural polarity becomes evident (Fig. 25).

In meristemoid I the orientation of the PMB anticipates the arrangement of the cell plate, already described. The first mode of alignment of the cell plate (Fig. 1,2B, D) is marked in advance by a PMB occupying a highly asymmetrical position. In sections 2 groups of microtubules abutted to the daughter wall of the first differential division, a short distance from its ends, were invariably observed (Figs. 22–24). Examination of some of these PMB in serial paradermal sections showed that the structure is not restricted to the cortical cytoplasm of the anticlinal wall but bends slightly in order to invade the cortical cytoplasm of the periclinal walls. This suggests that the band predicts the intersections of the cell plate with the anticlinal and both the periclinal walls. In this case the PMB does not lie in the plane of the cell plate.

During cytokinesis the cell plate becomes organized in a curved manner and joins the mother wall in the area previously occupied by the microtubule band (Fig. 29; see also Fig. 3). In the cytoplasmic region where the hemispherical cell plate formation starts, the phragmoplast microtubules are oriented at right angles to the partition of the first differential division. As the cell plate grows laterally, the microtubule orientation changes continuously and at the marginal ends of the cell plate they tend to become parallel to the above wall. This curved cell plate organization is similar to that described in other differential divisions (Pickett-Heaps & Northcote, 1966b; Heslop-Harrison, 1968; Cutter & Hung, 1972).

The second type of division of meristemoid I (Figs. 1,2A, C, 5) is predicted by a PMB, of which one group of microtubules, in all paradermal sections, lies in the cortical cytoplasm of the daughter wall of the first differential division, while the other group lies close to one of the older walls (Figs. 25–27). In this case the arrangement of the PMB varies from nearly symmetrical to highly asymmetrical and points to the plane of the cell plate as well as to the positions where it will fit to the parent wall. It is notable that sometimes premitotic signs of polarization cannot be detected.

The mitotic spindle in meristemoids I is either central or slightly eccentric and their daughter cells may be of the same size or may differ obviously (Figs. 3, 5). The daughter cells of the meristemoid I illustrated in Fig. 28B are almost equal in size. In addition, the differences in vacuole distribution are comparable to those occurring in non-differential divisions. The meristemoid II will function as a GMC of a mesoperigenous stoma (Figs. 6, 7) or will divide differentially once more.

From the above it is clear that the asymmetrical character of the second differential division is sharply reduced in comparison to that of the first division.

The meristemoid II of the mesogenous process is a cell smaller than the meristemoid I and may lack vacuoles entirely or contain only some minute ones (Figs. 31, 34). The greater part of the cell is occupied by the nucleus. During interphase, a dense system of subplasmalemmal microtubules oriented perpendicularly to the leaf surface borders its walls (Fig. 37). In this cell the PMB is mostly strictly organized in the cortical cytoplasm of the wall set up during the second differential division. In paradermal sections it appears as 2 clearly demarcated groups of microtubules localized a short distance from the corners of the cell, in both triangular and lensshaped meristemoids II (Figs. 31–33, 34–36). A comparison of the arrangement of the cell plate in these cells with that of the PMB shows that the latter indicates the exact locations where the future cell plate will fuse with the parent wall (compare Figs. 31, 34 with Figs. 39, 38 respectively). However, the plane of the PMB does not always coincide with that of the subsequent cell plate (i.e. in triangular meristemoids II; Figs. 31-33). In the meristemoid II the PMB does not surround the nucleus to a significant extent and does not traverse the cell. The formation of the cell plate takes place hemispherically. Its morphogenesis is controlled by a radially disposed phragmoplast (Figs. 38, 39).

In most meristemoids II observed in division, the mitotic spindle occupied a central position, while the space available for chromosome separation was limited. However, it must be noted that in a few triangular meristemoids II the division was slightly asymmetrical.

As far as the structural polarization of the lens-shaped meristemoids II is concerned, the only evident premitotic asymmetry was the position of the PMB. If small vacuoles exist, they are preferentially placed in that part of the cell which will be distributed to the second subsidiary cell. Attempts to detect other differences between the cell poles in ER or ribosome distribution or in something else were unsuccessful. In addition, the daughter nuclei are identical in appearance, and no significant differences in structure or size between daughter cells can be detected (Figs. 38, 39).

The microtubules in the meristemoids which will behave as GMC form a prominent frame lining the whole cortex of the cell. The PMB in GMC of mesogenous and mesoperigenous stomata is consistently centrally situated. In paradermal sections 2 groups of microtubules were detected in the corners of the cell extending out laterally on the same wall, symmetrically on both walls, or in complementary positions on both walls (Figs. 40-42). In the triangular GMC one group of microtubules of the PMB are localized in the corner of the cell, extending to the one or both of the long walls, while the others are spread along the whole or a significant portion of the short wall (Figs. 44, 46, 47). The PMB in the perigenous GMC occupy positions that anticipate the subsequent location of the cell plate.

A close examination of over one hundred GMC possessing a PMB revealed that the part of the anticlinal wall marked by the microtubules is preferentially thickened, a phenomenon not recorded in other dividing meristemoids or ordinary protodermal cells (Fig. 44). This wall deposition is not the usual one in process of formation at the junctions of walls, but constitutes a particular differentiation of the GMC in all the developmental modes. The appearance of the PMB precedes that of the wall thickening (Figs. 40–42). The 4 distinct PMB arrangements in the GMC described above evidently resemble 4 particular patterns of deposition of thickenings along the walls, clearly visible in young stomata (Figs. 45, 48–51).

The above observations suggest that the PMB is directly implicated in the formation of the differential wall thickenings of the GMC.

Associations between microtubules of the band and the dictyosome vesicles are very common in the GMC. Smooth as well as rough dictyosome vesicles are localized among the microtubules, fusing with the plasmalemma (Figs. 41, 42, 46, 47). It must also be noted that the GMC exhibits a remarkable dictyosome activity.

The wall separating the guard cells is normally joined with the parent walls in the middle of the thickening (Figs. 43, 48–51).

Most plant cells arise as groups from more or less symmetrical divisions of meristematic cells and their differentiation has been considered as related to known or unknown factors connected with the position of the cells in the plant body (Bloch, 1965). Structural differences between the products of division are usually slight, if they exist at all. However, in some exceptional divisions, the daughter cells diverge immediately in structure and destination (Bünning, 1952, 1957; Sinnott, 1960). Numerous differentiations of the plant body, among which are the stomata, arise through these divisions which have been called unequal, asymmetrical or differential (Strasburger, 1866; De Bary, 1877; Miehe, 1899, 1901; Bünning & Sagromsky, 1948; Bünning & Biegert, 1953). The establishment of premitotic polarity and the structural asymmetry of the succeeding differential divisions have been extensively studied by means of light microscope (see reviews by Bünning, 1957; Bloch, 1965), and recently with the electron microscope (Avers, 1963; Pickett-Heaps & Northcote, 1966a; Heslop-Harrison, 1968; Cutter & Hung, 1972; Pickett-Heaps, 19696). In contrast to what was expected, the application of electron microscopy did not significantly extend existing knowledge regarding polarity and differential divisions.

Although the daughter cells of the differential divisions involved in the ontogeny of stomata of Vigna sinensis proceed along divergent lines of differentiation, the extent to which premitotic polarity on a structural level and accompanying asymmetry are manifest varies considerably. The observations presented here substantiate that two extreme forms of differential divisions operate; one of them is characteristic mainly of ordinary protodermal cells. In this type of division, the establishment of premitotic polarity is reflected in many cell activities, i.e. the concentration of the cytoplasm at the one end of the cell, the displacement of the nucleus and the vacuoles in opposite directions, the asymmetrical disposition of the PMB and the eccentric organization of the mitotic spindle. The differences in size and organization between the daughter cells of this division are conspicuous. This type of differential division we designate ‘asymmetrical differential division’. The other type of division characterizes meristemoids. In some of these cells the above-mentioned epigenetic sequences are recognized with difficulty or are not manifest (see also Fig. 28 B). A careful examination of the premitotic meristemoids revealed neither a difference in the electron density of the cytoplasm between the cell poles nor a particular distribution of ribosomes or the ER or any other structure which could be considered as underlying the differential division or causing polarity. In all divisions, the daughter chromosome sets at anaphase and the reformed nuclei at telophase are identical in structure. The remarkable differences between telophase nuclei recorded by Stebbins & Shah (1960) and Stebbins & Jain (1960) in dividing cells of monocotyledonous stomata were not observed in Vigna sinensis.

From the above it becomes clear that none of the observed premitotic asymmetries is ubiquitous in differential divisions; in addition, no difference was observed in the structure of the daughter cells which can be considered as responsible for their divergent differentiation. The unequal distribution of cytoplasm in the cells which are going to divide differentially can be manifested only in large cells. On the other hand, the size of the daughter cells of the non-differential divisions of the ordinary protodermal cells, as well as the distribution of the vacuoles within them may sometimes vary considerably (Fig. 28A). The absence of conspicuous premitotic asymmetries from differential divisions has already been noted by Avers (1963) and Burgess & Northcote (1967), as well as by Cutter & Hung (1972) in the case of trichoblast mother cells.

Bünning (1952) has already stressed that the primary phenomenon of differential divisions is the synthesis of protoplasm at the polar end of the cell. He also points that ‘the establishment of some other gradient may well be the first step that would then lead to the difference in the density of protoplasm’ between the poles of the cells dividing differentially which was usually observed by light microscopists. Recently, Stebbins (1973) has gone on to suggest that cell polarization could be achieved by a self-electrophoresis mechanism like the one described by Jaffe (1968) in Fucus eggs. The proteins of the cell migrate differentially across the self-electrophoresis gradient to which the nucleus and other organelles respond. It has been suggested that the divergent differentiation of the daughter cells of a differential division may result from feedback into the nucleus of particular proteins synthesized in the cytoplasm (Stebbins, 1973; Gurdon, 1974). In the present case the absence of any obvious sign of polarization in some meristemoids dividing differentially and the close similarity in appearance of the daughter cells suggest a differential distribution of particular regulatory substances in the polarized cells and later in their derivatives, which are invisible with the electron microscope.

A detailed consideration of the present observations supports strongly the view that there exists a strict positional correspondence between the PMB and the cell plate and that the band does not itself determine the differential division. The results, as well as the observation of polarized cells without a PMB, support Pickett-Heaps’ (1969b) contention that the PMB is a result and not a cause of polarity. In a significant number of differential and non-differential divisions occurring in different planes to each other we verified that this positional correspondence is primarily a matter of indicating in advance the position of intersection of the succeeding cell plate with the parent walls. The PMB in meristemoids II and some meristemoids I is organized apposed to the one wall set apart from the nucleus and does not lie in the plane of the cell plate. The above conclusions do not imply that the PMB determines the position of the cell plate formation.

The results presented here agree with those of Burgess & Northcote (1967) in suggesting that the PMB in differential divisions may occupy a symmetrical position. However, our findings do not support their observations, as well as those of Singh (1977) and Singh et al. (1977), that the PMB does not predict the position of the cell plate.

Although the morphology of the PMB has been investigated to a considerable extent in a number of plants, a definite function cannot be attributed to it. Pickett-Heaps & Northcote’s (1966a) attractive suggestion that the microtubules of the band constitute a pool of tubulin which becomes incorporated in the organizing spindle seems reasonable (for a detailed discussion see Pickett-Heaps, 1974); our results do not offer any additional support. The other proposed function, that the PMB orientates the nucleus (Burgess & Northcote, 1967) has been questioned by Pickett-Heaps (1969b, c, 1974). Data bearing on such an activity for the PMB did not arise from the present study.

Regardless of the above proposed functions, the PMB of GMC appears to be implicated in cell wall deposition, an activity which has already been attributed to the PMB in dividing root cells of Allium (Packard & Stack, 1976). In Vigna sinensis GMC the appearance of the PMB is followed by a local deposition of material at the wall sites indicated by the band. The microtubules of the band appear to mediate a directed exocytosis of dictyosome vesicles. The question arises, why the PMB behaves in this way in GMC only. In our view, the PMB must display the same activity in all dividing cells, but this activity is transformed to a structural effect in GMC only, since in these cells the Golgi apparatus is possibly more active and material is made available to non-expanding walls. It is also possible that the PMB functions in GMC longer. The transportation of dictyosome vesicles by the microtubules is widely accepted in the case of the developing cell plate and has been repeatedly reported in the case of growing cell walls (see reviews by Newcomb, 1969; Brower & Hepler, 1976). Whatever the answer to the above question is, the unequal thickening of the wall of GMC represents a divergent differentiation of these cells not occurring in other meristemoids. It allows the definite identification of GMC in Vigna sinensis leaves in which more than one developmental mode operates at the same time.

It has been generally accepted that differential division is an important factor in pattern formation in the epidermal tissue of dicotyledons; however, according to Biinning (1952, 1965) the phenomenon which in fact underlies pattern formation in this tissue is the mutual inhibition of meristemoid formation. One meristemoid does not permit the initiation of any other in its neighbourhood (see also Korn, 1972; Korn & Fredrick, 1973; Lang, 1973). In the plant examined in this paper, the mesogenous and mesoperigenous meristemoids I do not exert an inhibitory influence on each other. On the contrary, they induce the formation of other meristemoids close to them, determining at least the sites of polarization of their neighbouring ordinary protodermal cells dividing differentially. This short-distance induction affects the epidermal pattern formation in Vigna sinensis, but in the opposite way to what is generally accepted.

Our data appear to rule out the possibility that the adjacent formation of meristemoids in Vigna is a response to any other factor(s). In fresh tissue we observed that there is no positional relationship between the meristemoids I and the underlying air spaces. The substomatal cavities appear and grow at a later stage.

In the epidermis of dicotyledons the orientation of the divisions is very irregular. Although Smith (1935) observed in some dicotyledons that the axes of stomata tend to become parallel to the axes of the main veins, the factors controlling them remain obscure. In the leaves of Vigna sinensis there are some observations which suggest that the orientation of the plane of some divisions is controlled by cell interactions. In the case of adjacent formation of meristemoids, the meristemoids I seem to define the orientation of the mitotic spindle and the PMB. Another case in which a cell determines the plane of division of another is that of the large perigenous stomata. Most of the divisions occurring in their surrounding cells take place along their long axis and are parallel to the dorsal walls of the guard cells (Fig. 52). No correlation can be made between shape and organization of the meristemoid I and the plane of cell division; cells of the same form divide in different planes.

The apposition of the PMB close to one of the anticlinal walls before the second and third differential divisions (Figs. 22, 31, 34) suggests the presence of some critical factors, among which is the microtubule organizing centre of the PMB, in the cortical cytoplasm of this wall or the plasmalemma. This local differentiation of the cytoplasm or the plasmalemma is a polarization effect and could be regarded as a response to a short-distance chemical or mechanical induction directed by one of the neighbouring cells. In the case of meristemoids I and II the inducers appear to be respectively the protodermal and the subsidiary cells which have a common origin with them. It is interesting that at the advanced stages of leaf development when the meristematic activity of the protoderm weakens, the first type of orientation of the cell plate in the meristemoid I predominates (Fig. 1,2B).

A similar disposition of the PMB during the second differential division of the development of Graminaceous stomata, according to Green, Erickson & Richmond (1970), ‘reflects strain alignment in the responding cell provided that the GMC can modify (expand, contract) the cell membrane near it while compensating changes occur elsewhere in the responding cell’. In Vigna sinensis protoderm the daughter cells of the differential divisions show different rates of growth (the meristemoids do not become vacuolated). Therefore, if the larger cell of these divisions expands along an axis parallel to the recently established wall, it may induce changes to the plasmalemma and/or the adjoining cortical cytoplasm of the meristemoid. However, whether in the protodermal tissue, mechanical strains or chemical stimuli, or both of them, are responsible for the short distance cell interactions is quite unknown.