ABSTRACT
The confrontation of cells from the anterior region of an abdominal segment of Oncopeltus with those from the posterior region of the same or the adjacent segment results in the generation of a segment border. The behaviour of epidermal cells during this regulation is described. It consists primarily of cell division and transverse elongation of cells at the site of confrontation. This behaviour can be separated from any associated purely with wound healing because a similar-sized wound to that used to ablate the segment border, performed within the segment, does not result in any cell division or elongation. The results are consistent with the view that there is a discontinuity in positional values at the segment border. The stability of such a discontinuity and the regeneration of segment borders are discussed in terms of there being a special population of cells at the segment border that have the property of isolating other cells with the maximum difference in positional values.
Introduction
The confrontation of cells from the anterior region of an abdominal segment of the milkweed bug, Oncopeltus fasciatus, with those from the posterior region results in the generation of a segment border at the site of confrontation (Wright & Lawrence, 1981). Such a confrontation can be created by ablating the segment border itself, resulting in its regeneration, or by the removal of most of the central portion of the segment, resulting in a supernumerary border in the centre of the segment. This regulation has been explained in terms of a model in which the abdominal pattern is specified by a repeating series of positional values, each series being equivalent to a segment length. The segment borders do not constitute discontinuities in the series so that the difference in positional values between two anteroposterior levels can be expressed as an angle (Wright & Lawrence, 1981; Russell, 1985). This model follows the rules of the polar coordinate model for limb regeneration where circumferential values are similarly arranged in a continuous series (French, Bryant & Bryant, 1976). In that case confrontation of cells with different positional values leads to intercalation of the missing values via the shortest route. Similar arguments can be used to explain regeneration of segment borders.
This model contrasts with the more traditional view of the insect abdomen in which there is a repeating series of gradient systems; each gradient corresponds to a segment so that each segment is considered to be an independent developmental field (Locke, 1959, 1960; Stumpf, 1966; Lawrence, 1966,1973). Here, the so-called ‘intersegmental membranes’, the boundary between adjacent segments, are assumed to form a barrier between adjacent segmental gradients. The gradients can be represented as series of positional values along the anteroposterior axis (Lawrence, Crick & Munro, 1972) so that there is a discontinuity in positional values at the segment border. According to this model, the discontinuity produced by ablation of the segment border could be removed by the intercalation of a series of intrasegmental levels with reversed polarity. However, except very rarely, that does not occur and a segment border reforms (Lawrence, 1966; Locke, 1960; Nübler-Jung, 1979; Wright & Lawrence, 1981). This was explained by immigration of preexisting cells of the intersegmental membrane over the ablated area to restore their continuity and isolate the separate gradient fields (Lawrence, 1966). However, this explanation is invalid because segment borders can reform from normal segmental cells (Wright & Lawrence, 1981). Therefore, in its original form, the gradient model is unable to account for the regeneration of segment borders.
This paper describes the behaviour of epidermal cells during the regeneration of segment borders. This consists largely of localized cell division and cell elongation. The results are discussed in terms of the two models described above and the latter gradient model is favoured. However, although this model will explain cell behaviour, it has to be modified to account for the regeneration of segment borders and formation of supernumerary borders.
Materials and methods
(A) General
Stocks of Oncopeltus fasciatus (Lygaeidae, Hemiptera) were kept at 24°C in a D: L cycle of 12:12 and fed sunflower seeds and water. Wounds were performed on animals that had been starved soon (at most 14 h) after moulting. They were starved for about 48 h before any operation. Animals were anaesthetized either with ice (for operations lasting under one minute) or CO2 (for experiments requiring more time). The integument was cut with slivers of razor blade and cells were scraped from the underside of the cuticle by inserting a blunted tungsten needle through a cut. To help plot the positions of mitotic figures, 4th instars were injected through the base of a metathoracic leg with 0·5 ml of a 1·0mgml-1 solution of colchicine (about 5·0mgg-1 body weight) 12 h before fixation. For whole mounts, integument was fixed-for about l·0h in Camoy’s fixative. Hydration through an alcohol series was augmented with 880 ammonia at the 70 % and 50 % stages to depigment the epidermal cells. The whole mounts were stained in Hansen’s trioxy-haematin (Pantin, 1969). Integument for scanning electron microscopy was fixed in acetone, critical-point dried with CO2, coated with gold and viewed and photographed on a Cambridge S100 SEM.
For a quantitative measure of cell division, the total number of mitotic figures was counted in preparations (treated with colchicine) over a length (mediolateral distance) of 1·0 mm in the centre of the wound. At the times that measurements were made, virtually all of the mitotic figures occurred within an anteroposterior width of 0·1 mm. 12 preparations were analysed to give the final mean. The final cell density in the centre of a scrape was determined by the following procedure. Three sample areas, each measuring 0·l×0·lmm, were randomly chosen from within the centre of the scrape (at different mediolateral positions) in preparations fixed after the time at which cell division was known to have ceased. The total number of cells in each of these areas was counted and the mean cell density was calculated for that preparation. Ten preparations were analysed to give the final mean for the centre of that class of scrape.
All means are given with their standard errors. To determine whether the amount of cell division or the final cell density after any particular scrape was significantly different from that following another scrape, the results were treated with a Student t-test.
(B) The operations
Cells from different anteroposterior levels were confronted by an indirect method. Strips of epidermis between two levels were scraped from the cuticle with a needle and the cells on either side were allowed to migrate across the wound and meet in the centre. The area from which cells have been removed is easily discerned because the epidermal cells of Oncopeltus are orange in colour (due to the presence of pigment granules) and the cuticle is transparent.
Cell behaviour, especially the amount of cell division, was observed in the period following the operation by preparing stained, whole mounts of integument at different times after the operation. For this study 4th instars (Oncopeltus has five larval instars) have been used that were starved following a moult. This eliminates any modification of behaviour by hormones secreted during the moult cycle, because the moult cycle of Oncopeltus is normally initiated by feeding (Nijhout, 1979). No epidermal cell divisions are found in unwounded, starved larvae (Campbell, 1987a). Identical confrontations were produced in 3rd instars but were scaled down because of the smaller size of this instar. The 3rd instars were fed and the pattern generated following the operation was recorded two moults later at the beginning of the 5th instar.
The area of integument used was the 3rd tergite (Fig. 1A), together with the 4th when the scrapes extended across a segment border. The scrapes extended transversely across the segment from an anteroposterior incision (required to introduce the needle) level with the muscle insertions on one side to past the muscle insertions on the other side of the segment. This distance is about 1·7 mm in 4th instars. The anteroposterior locations and widths of the scrapes are given below and in Fig. 2. In a rectangular scrape, the length is the mediolateral distance. The exact dimensions of these scrapes are for 4th instars. When repeated on 3rd instars these dimensions have to be scaled down. The length of the segment in 4th instars is just under 0·6 mm.
(A) A recently moulted fourth instar of Oncopeltus fasciatus, dorsal side. The abdomen is divided clearly into segments. Segment J is marked. The sites of the lateral muscle insertions (m) that occur approximately midway between the segment borders indicate the mediolateral extent of the scrapes shown in Fig. 2. (B) Unstained, whole mount of the segment border between 2nd and 3rd tergites in a 4th instar. The cells at the anterior of the 3rd segment are clearly more darkly pigmented than surrounding cells. (C,D) The cuticular pattern in and around the segment border between the 2nd and 3rd tergites of a 5th instar. (C) Scanning electron micrograph which shows that there is a groove in the surface of the cuticle at the segment border. (D) Whole mount, depigmented and photographed under phase contrast. At the top and bottom, this photograph shows the normal cuticular pattern found over most of the segment. At the segment border, the pattern of polygons is obscured. Here it appears that adjacent polygons fuse laterally and their posterior margins are aligned and continuous over long distances. Anterior at top. Bars: (A) 0·5 mm, (B) 0·1 mm, (C) 0·05 mm, (D) 0·02mm.
(A) A recently moulted fourth instar of Oncopeltus fasciatus, dorsal side. The abdomen is divided clearly into segments. Segment J is marked. The sites of the lateral muscle insertions (m) that occur approximately midway between the segment borders indicate the mediolateral extent of the scrapes shown in Fig. 2. (B) Unstained, whole mount of the segment border between 2nd and 3rd tergites in a 4th instar. The cells at the anterior of the 3rd segment are clearly more darkly pigmented than surrounding cells. (C,D) The cuticular pattern in and around the segment border between the 2nd and 3rd tergites of a 5th instar. (C) Scanning electron micrograph which shows that there is a groove in the surface of the cuticle at the segment border. (D) Whole mount, depigmented and photographed under phase contrast. At the top and bottom, this photograph shows the normal cuticular pattern found over most of the segment. At the segment border, the pattern of polygons is obscured. Here it appears that adjacent polygons fuse laterally and their posterior margins are aligned and continuous over long distances. Anterior at top. Bars: (A) 0·5 mm, (B) 0·1 mm, (C) 0·05 mm, (D) 0·02mm.
The location and widths of scrapes designed to confront cells originally found at different levels in the anteroposterior axis of the 3rd and 4th tergites and a summary of the results (cell behaviour and final pattern). Each scrape extends laterally from the incision on one side (required to introduce the needle used to remove the cells) to bey-ond the muscle insertions (Fig. 1A) on the opposite side. The 3rd tergite appears normal two moults after 0·1 mid scrapes. The segment border usually regenerates after 0·1 bord scrapes, but occasionally it does not and a narrow region of reversed polarity is found instead. Supernumerary segment borders are usually found after 0·5 mid scrapes. The polarity of the central region which contains this border is reversed. Following 0·6 bord scrapes, in half of the cases the 3rd and 4th segments become fused; in the remaining half there is cell elongation and distortion of the cuticular sculpturing pattern where the 3/4 border was originally located.
The location and widths of scrapes designed to confront cells originally found at different levels in the anteroposterior axis of the 3rd and 4th tergites and a summary of the results (cell behaviour and final pattern). Each scrape extends laterally from the incision on one side (required to introduce the needle used to remove the cells) to bey-ond the muscle insertions (Fig. 1A) on the opposite side. The 3rd tergite appears normal two moults after 0·1 mid scrapes. The segment border usually regenerates after 0·1 bord scrapes, but occasionally it does not and a narrow region of reversed polarity is found instead. Supernumerary segment borders are usually found after 0·5 mid scrapes. The polarity of the central region which contains this border is reversed. Following 0·6 bord scrapes, in half of the cases the 3rd and 4th segments become fused; in the remaining half there is cell elongation and distortion of the cuticular sculpturing pattern where the 3/4 border was originally located.
The scrapes are as follows.
(1) 0·1 mid and 0·1 bord
These operations remove a 0·1 mm wide strip of cells in the middle of the segment (0·1 mid) and across the border between segments 3 and 4 (0·1 bord). The width of these scrapes corresponds to about a sixth of a segment length.
(2) 0·5 mid
This operation removes most of the epidermal cells between the segment borders. The anteroposterior width of this scrape is about 0·5 min so that it leaves 0·05 mm strips of cells behind the anterior border and in front of the posterior border.
(3) 0·6 bord
In this case, cells are removed from the middle of the 3rd segment to the middle of the 4th. The scrape is about 0·6 mm wide.
At the outset it was assumed that the cells migrating over a scrape maintain their original positional values. The validity of this assumption is discussed below.
(A) The Oncopeltus abdomen
The abdomen of Oncopeltus is clearly divided into segments (Fig. 1A). The borders between the segments are characterized by a groove in the surface of the cuticle and by transverse elongation of the cells (Lawrence & Green, 1975; Wright & Lawrence, 1981; Fig. 1C). The cells situated at the segment border are usually more darkly pigmented than surrounding cells, especially in 4th instars (Fig. 1B). The surface of the cuticle over most of the segment is sculptured into a pattern of irregularly packed polygons each representing the area of cuticle secreted by a single epidermal cell (Hinton & Gibbs, 1971; Fig. 1D). This pattern is polarized because the posterior margin of each polygon is raised slightly and represents the light areas visible in phase-contrast microscopy.
(B) Cell behaviour in fourth instars
(1) Cell behaviour after 0·1 scrapes
The cells spread over the wound and cover it within
24 h. Following 0·1 mid scrapes there is virtually no cell division (four mitotic figures in 43 preparations fixed 2, 3 and 4 days after wounding) and no visible build up of cell numbers in the centre of the wound (Fig. 3A). Cell behaviour after an identically sized wound in other locations within the segment is very similar (data not given).
0·1 scrapes. Stained whole mounts of dorsal, abdominal integument from 4th instars fixed 4 days after wounding and injected with colchicine 12 h before fixation. (A) 0·1 mid. An area of low cell density has been created by the wound. No mitotic figures can be found among the activated cells that have migrated over the region originally scraped free of cells (s). (B,C) 0·1 bord. There is an area of higher cell density in the centre of the wound (z). At higher magnification, (C), numerous mitotic figures (m) can be seen here, and many of the nuclei are transversely elongated. Anterior at top. Bars: (A,B) 0·1 mm, (C) 0·02 mm.
0·1 scrapes. Stained whole mounts of dorsal, abdominal integument from 4th instars fixed 4 days after wounding and injected with colchicine 12 h before fixation. (A) 0·1 mid. An area of low cell density has been created by the wound. No mitotic figures can be found among the activated cells that have migrated over the region originally scraped free of cells (s). (B,C) 0·1 bord. There is an area of higher cell density in the centre of the wound (z). At higher magnification, (C), numerous mitotic figures (m) can be seen here, and many of the nuclei are transversely elongated. Anterior at top. Bars: (A,B) 0·1 mm, (C) 0·02 mm.
The behaviour following the same-sized wound but removing the cells located at the segment border between segments 3 and 4 (0·1 bord) is very clearly different from that described above (Figs 3,4). Often there is quite extensive cell division in the centre of the wound, leading to a local increase in cell density (Fig. 3B,C). Here the cells also show a change in morphology from the usual polygonal shape with round nuclei to a spindle-shaped form with elongated nuclei orientated transversely. This is an exaggeration of the morphology normally found at the segment border (Lawrence & Green, 1975). The number of divisions found increases from day 3 to peak 4 days after wounding (Fig. 4). All this division is confined to a band usually no wider than 0·05 mm found in the centre of the wound and there is no spread of cell division from the site of confrontation. The high division rate is reflected in a final cell density in the centre that is significantly larger (P < 0·001) than that at the centre of a 0·1 mid scrape.
Cell division and final cell density after 0·1 mm scrapes in 4th instars. (A) The number of mitotic figures per sample area per preparation 3 and 4 days after the operation. Animals were treated with colchicine 12h before fixation. No mitotic figures are found after 0·1 mid scrapes in contrast to 0·1 bord scrapes which ablate the segment border. (B) Final cell density in the centre of the scrapes after the cessation of cell division. The cell density in the centre of the scrapes is greater after 0·1 bord scrapes than after 0·1 mid scrapes.
Cell division and final cell density after 0·1 mm scrapes in 4th instars. (A) The number of mitotic figures per sample area per preparation 3 and 4 days after the operation. Animals were treated with colchicine 12h before fixation. No mitotic figures are found after 0·1 mid scrapes in contrast to 0·1 bord scrapes which ablate the segment border. (B) Final cell density in the centre of the scrapes after the cessation of cell division. The cell density in the centre of the scrapes is greater after 0·1 bord scrapes than after 0·1 mid scrapes.
However, it should be noted that the increased cell division following a 0·1 bord scrape is variable and sometimes does not occur. Then the behaviour is very similar to that following other 01mm scrapes within the segment. Thus, in a minority of cases (2/12) no mitotic figures were found in 0·1 bord preparations from animals treated with colchicine and fixed after 4 days. In all other cases, there was evidence of cell division and elongation.
(2) Cell behaviour after 0·5 mid and 0·6 bord scrapes
A 0·5 mid scrape confronts cells originally located at the same levels in the segment as confronted by a 0·1 bord scrape, but differs from the latter scrape in that the cells originate from the same segment rather than adjacent segments and in the wound being much larger. After a 0·5 mid scrape, two processes, wound healing and pattern regulation, take place. The component of cell behaviour due to wound healing alone is detectable in scrapes of a comparable size where the confronted cells have the same positional value. The 0·6 bord scrape is of this type:
0·5 mid scrapes are rapidly repaired and in most cases have been covered with cells within 48 h of the operation. By 72h numerous mitotic figures are present in all preparations. Injection of colchicine reveals a central band of extensive mitotic activity in a zone 0·1mm-0·3mm wide. After 4 days, this band has narrowed significantly to a strip usually no more than 0·07 mm wide located in the centre of the scrape (Fig. 5). Again, like the pattern of mitosis following 0·1 bord scrapes, mitotic figures are confined within this narrow band. The final cell density in the centre is much higher than in the sparse zone immediately adjacent to the zone of higher cell density (Fig. 6).
Stained whole mounts of dorsal, abdominal integument after 0·5 mid scrapes in 4th instars. In (A) and (B), the animal was fixed 4 days after the operation and was treated with colchicine. There is a zone of higher cell density (z) in the centre of the area of low cell density created by the scrape (s). Numerous mitotic figures (in) are found in this zone, as shown in B. (C) 7 days after the operation (no colchicine), at the site of confrontation in the centre of the scrape. Some nuclei are transversely elongated and a mitotic figure is present in which the orientation of the metaphase plate is clearly parallel to the anteroposterior axis. Some cell death (d), revealed by small darkly staining nuclei that gradually become more and more diffuse, is also present. Anterior at top. Bars: (A) 0· 05 mm, (B) 0· 02 mm, (C) 0· 02 mm.
Stained whole mounts of dorsal, abdominal integument after 0·5 mid scrapes in 4th instars. In (A) and (B), the animal was fixed 4 days after the operation and was treated with colchicine. There is a zone of higher cell density (z) in the centre of the area of low cell density created by the scrape (s). Numerous mitotic figures (in) are found in this zone, as shown in B. (C) 7 days after the operation (no colchicine), at the site of confrontation in the centre of the scrape. Some nuclei are transversely elongated and a mitotic figure is present in which the orientation of the metaphase plate is clearly parallel to the anteroposterior axis. Some cell death (d), revealed by small darkly staining nuclei that gradually become more and more diffuse, is also present. Anterior at top. Bars: (A) 0· 05 mm, (B) 0· 02 mm, (C) 0· 02 mm.
Cell division and final cell density after 0·5 mid and 0·6 bord scrapes in 4th instars. (A) The number of mitotic figures per sample area/preparation 4 days after the operation. Animals were treated with colchicine 12 h before fixation. The number of mitotic figures after a 0·5 bord scrape is considerably greater than the number found after a 0·6 bord scrape. (B) Final cell density in the central region of the scrapes and in the sparse zone to either side of it after the cessation of cell division. The cell density in the centre of a 0·5 mid scrape is greater than that in the sparse zone surrounding it. The central cell density is also greater than that following the 0·6 bord scrape. The density in the centre of 0·6 bord scrapes is not much greater than that in the area surrounding it.
Cell division and final cell density after 0·5 mid and 0·6 bord scrapes in 4th instars. (A) The number of mitotic figures per sample area/preparation 4 days after the operation. Animals were treated with colchicine 12 h before fixation. The number of mitotic figures after a 0·5 bord scrape is considerably greater than the number found after a 0·6 bord scrape. (B) Final cell density in the central region of the scrapes and in the sparse zone to either side of it after the cessation of cell division. The cell density in the centre of a 0·5 mid scrape is greater than that in the sparse zone surrounding it. The central cell density is also greater than that following the 0·6 bord scrape. The density in the centre of 0·6 bord scrapes is not much greater than that in the area surrounding it.
In preparations not treated with colchicine, the spindles of the mitotic figures are orientated predominantly perpendicular to the anteroposterior axis (Fig. 5C) as judged from alignment of the metaphase plates. Daughter cells will therefore separate to lie alongside each other laterally. Cells in the vicinity of the band of high mitotic activity are transversely elongated. All 0·5 mid preparations show evidence of cell division and elongation in the centre.
The pattern of cell behaviour during the first 72 h after 0·6 bord scrapes is similar to that after 0·5 mid scrapes. Many are repaired within 48 h and after 72 h mitotic figures are often found in a broad band in the centre of the scrape zone. However, 4 days after wounding, in contrast to 0·5 mid scrapes, a region of high mitotic activity in the centre of the wound is not usually found (Fig. 7). The number of mitotic figures found is significantly smaller (P < 0·001) than found 4 days after a 0·5 mid scrape (Fig. 6).
Stained whole mounts of dorsal abdominal integument from animals fixed 4 days after 0·6 bord scrapes in 4th instars treated with colchicine 12 h before fixation. The most common result is shown here. (A) The area originally scraped free of cells (s) has relatively uniform, low cell density with no increase in the centre. (B) The cells in the centre are obviously activated with large round (no elongated) nuclei. An, anterior; Po, postenor. Bars: (A) 0·05 mm, (B) 0· 01 mm.
Stained whole mounts of dorsal abdominal integument from animals fixed 4 days after 0·6 bord scrapes in 4th instars treated with colchicine 12 h before fixation. The most common result is shown here. (A) The area originally scraped free of cells (s) has relatively uniform, low cell density with no increase in the centre. (B) The cells in the centre are obviously activated with large round (no elongated) nuclei. An, anterior; Po, postenor. Bars: (A) 0·05 mm, (B) 0· 01 mm.
If the increased cell density, division and transverse elongation at the site of confrontation is termed a ‘reaction’, then 5 days after the 0·6 bord operation, the majority (8/12) shows no reaction while the minority (4/12) shows a varying amount. Where no reaction occurs there is an almost uniform cell density over all the wounded area with no increase in the centre (Figs 6, 7A). The nuclei of the cells in the centre of these preparations are round (Fig. 7B). The cell density in the centre is significantly smaller (P < 0·001) than that in the centre of 0·5 mid scrapes (Fig-6).
C) Cell behaviour and pattern one and two moults after the operation
Cell behaviour at later moult stages
When 0·1 bord scrapes are fixed as 4th instars soon (1–3 days) after the postoperative moult, the cells in the regenerating border region are often activated (having large round nuclei and prominent nucleoli (Wigglesworth, 1937)), and on a minority of preparations (5/16, not treated with colchicine) some mitotic figures are present (Fig. 8A). These animals were also starved following the postoperative moult to prevent initiation of the next moult cycle (during which all the epidermal cells would become activated). The extent of this activation is variable and sometimes it is absent. At this stage, activation and cell division are never observed following 0·1 mid scrapes.
Whole mounts of 0·1 bord scrapes fixed after the first or second postoperative moult when the animal was wounded as a 3rd instar. (A) Fixed 3 days after the first postoperative moult and stained with trioxyhaeniatin (animal not treated with colchicine). The animal has been starved since the moult. The cells in the regenerating border region are activated (ac) and mitotic figures (m) are found here. (B–D) Fixed soon after the second postoperative moult, illustrate the two possible outcomes of this operation. (B) Scanning electron micrograph of the regenerated border region showing the usual outcome. There is a groove in the cuticle in the centre of the field of view (arrowed), which probably corresponds to the regenerated border. However, there are also additional grooves running more or less parallel and anterior to this one. Regenerated borders are therefore often somewhat abnormal. (C) Whole mount, under phase contrast showing the minority outcome. There is a band of polarity reversal of about 0·1 mm in width over the region where the 3/4 border was originally situated (pr). The polarity change is apparent in the locally reversed orientation of the hairs and in the reversed pattern of the cuticular sculpturing. The sculpturing does not resemble that found at a segment border (compare with Fig. 1D). (D) The epidermis of the regenerated region (pr) in C, shown stained with trioxyhaematin, revealing the transverse elongation of the cells in this region. This change in cell shape may be responsible for the slightly distorted cuticular pattern shown in C. Anterior at top. uc, unactivated cells. Bars: (A) 0 ·05mm, (B) 0· 05mm, (C) 0·03mm, (D) 0 ·02mm.
Whole mounts of 0·1 bord scrapes fixed after the first or second postoperative moult when the animal was wounded as a 3rd instar. (A) Fixed 3 days after the first postoperative moult and stained with trioxyhaeniatin (animal not treated with colchicine). The animal has been starved since the moult. The cells in the regenerating border region are activated (ac) and mitotic figures (m) are found here. (B–D) Fixed soon after the second postoperative moult, illustrate the two possible outcomes of this operation. (B) Scanning electron micrograph of the regenerated border region showing the usual outcome. There is a groove in the cuticle in the centre of the field of view (arrowed), which probably corresponds to the regenerated border. However, there are also additional grooves running more or less parallel and anterior to this one. Regenerated borders are therefore often somewhat abnormal. (C) Whole mount, under phase contrast showing the minority outcome. There is a band of polarity reversal of about 0·1 mm in width over the region where the 3/4 border was originally situated (pr). The polarity change is apparent in the locally reversed orientation of the hairs and in the reversed pattern of the cuticular sculpturing. The sculpturing does not resemble that found at a segment border (compare with Fig. 1D). (D) The epidermis of the regenerated region (pr) in C, shown stained with trioxyhaematin, revealing the transverse elongation of the cells in this region. This change in cell shape may be responsible for the slightly distorted cuticular pattern shown in C. Anterior at top. uc, unactivated cells. Bars: (A) 0 ·05mm, (B) 0· 05mm, (C) 0·03mm, (D) 0 ·02mm.
Pattern two moults after 0·1 scrapes
Following 0·1 mid scrapes in the 3rd instar virtually no alteration can be found in the tergite pattern of the 5th instar. The cuticular pattern of 5th instars, two moults after a 0·1 bord scrape, is variable, but generally preparations can be separated into two classes. The majority (20/25) appears to have regenerated the segment border, which is often indistinguishable from that at an unwounded segment border (Fig. 1) and polarity is normal. However, occasionally it is imperfect and more than one cuticular groove is found (Fig. 8B). Preparations in the second class (5/25) have a band of reversed polarity in the region where the 3/4 border used to be. This extends anteroposteriorly for about 0·1–0·2 mm (Fig. 8C). In this region, a groove is absent and the cuticular sculpturing more closely corresponds to that over most of the normal segment rather than at the segment border (Fig. 8C).
Pattern two moults after 0·5 mid scrapes
When 0·5 mid scrapes are made on segment 3 of 3rd instars and the animals are examined as 5ths, a supernumerary segment border (characterized by a cuticular groove and characteristic sculpturing pattern, Fig. 1C,D) is present in 80% of cases (32/40) (Fig. 9). The polarity pattern is also altered so that the polarities of bristles and cuticular sculpturing are normal immediately posterior to the original 2/3 border and immediately anterior to the original 3/4 border, but the central region containing the supernumerary border has reversed polarity (Fig. 9B). The pattern appears to consist of a mirror-image duplication of the anterior-most part of the segment in front of the supernumerary border and a mirrorimage duplication of the posterior-most part of the segment behind the new border (Fig. 9B). The final pattern of the remaining 20 % is variable and difficult to interpret.
0·5 mid scrapes. Recently moulted 5th instars two moults after the operation. The resulting pattern has a supernumerary border (su.b) in the 3rd segment. The polarity of hairs and cuticular sculpturing is reversed in the vicinity of this new border. Within the 3rd segment, the area anterior to the supernumerary border is always larger than that posterior to it. (A) In situ. The area posterior to the supernumerary border has a lower level of pigmentation than the area anterior to it. (B) Unstained, depigmented whole mount, showing the polarity reversal. The points at which polarity reversal begins and ends, anterior and posterior to the supernumerary border, is approximately half-way between this border and the normal (2/3 and 3/4) borders. The polarity reversal is clearer anterior to the supernumerary border than posterior because the former region is larger and more hairs are present here. (C) Scanning electron micrograph. All three borders are represented by a groove in the cuticle, p, pigment spot around stink gland situated at the 4/5 border. Bars: (A) 0· 5 mm, (B) 0·2mm, (C) 0·2mm.
0·5 mid scrapes. Recently moulted 5th instars two moults after the operation. The resulting pattern has a supernumerary border (su.b) in the 3rd segment. The polarity of hairs and cuticular sculpturing is reversed in the vicinity of this new border. Within the 3rd segment, the area anterior to the supernumerary border is always larger than that posterior to it. (A) In situ. The area posterior to the supernumerary border has a lower level of pigmentation than the area anterior to it. (B) Unstained, depigmented whole mount, showing the polarity reversal. The points at which polarity reversal begins and ends, anterior and posterior to the supernumerary border, is approximately half-way between this border and the normal (2/3 and 3/4) borders. The polarity reversal is clearer anterior to the supernumerary border than posterior because the former region is larger and more hairs are present here. (C) Scanning electron micrograph. All three borders are represented by a groove in the cuticle, p, pigment spot around stink gland situated at the 4/5 border. Bars: (A) 0· 5 mm, (B) 0·2mm, (C) 0·2mm.
Apart from the polarity patterns, the pigmentation of the segment also helps interpretation. The posterior region of the tergites in 5th instars is often less heavily pigmented than more anterior regions so that it appears lighter in colour. Thus, at normal segment borders there is a discontinuity in the level of pigmentation (in addition to the heavily pigmented cells that are usually found at the segment border (Fig. 1B)). Where a supernumerary border forms, the region posterior to it often has the typical pigmentation of the posterior segment. The region anterior to the supernumerary border has a higher level of pigmentation characteristic of anterior levels of the segment, so that there is also a discontinuity in the level of pigmentation at this border (Fig. 9A).
Pattern two moults after 0·6 bord scrapes
In theory, 0·6 bord scrapes confront cells with the same positional value from adjacent segments, so that these operations should result in a segment extending from the 2/3 segment border to the 4/5 border. Out of 37 0·6 mid scrapes performed on 3rd instars and analysed in 5ths, approximately half (19) have patterns resembling that expected, i.e. a large segment extending from the 2/3 border to the 4/5 border (Fig. 10). No grooves in the cuticle are found where the 3/4 segment border would normally occur and the cuticular pattern resembles that found over most of an unoperated segment (Fig. 10B). There is also no sign of transverse elongation of cells at this site.
0·6 bord scrapes. Recently moulted 5th instars two moults after the operation. (A,B) These represent the expected pattern in which the 3/4 segment border is missing in the central area resulting in a large segment extending from the 2/3 to the 4/5 border. The polarity in the central area is normal. (A) In situ. (B) Unstained, depigmented whole mount. The 2/3 segment border is clearly distinguishable, but the 3/4 border is missing, p, pigment spot. Bars: (A) 0·5 mm, (B) 0·2 mm.
0·6 bord scrapes. Recently moulted 5th instars two moults after the operation. (A,B) These represent the expected pattern in which the 3/4 segment border is missing in the central area resulting in a large segment extending from the 2/3 to the 4/5 border. The polarity in the central area is normal. (A) In situ. (B) Unstained, depigmented whole mount. The 2/3 segment border is clearly distinguishable, but the 3/4 border is missing, p, pigment spot. Bars: (A) 0·5 mm, (B) 0·2 mm.
The pattern in the remaining 50% of scrapes (18/37) is very variable and difficult to interpret (Fig. 11). Here, in contrast, the cells do show transverse elongation in the centre of the original wound at the level where the ablated 3/4 segment border was located. However, its extent may vary from a single line of very modestly elongated cells to a 0·4 mm wide band of extremely elongated cells (Fig. 11B). The cuticular sculpturing pattern associated with this cell elongation is often very modified and the polarity in this region can be abnormal (Fig. 11A).
0·6 bord scrapes. Recently moulted 5th instars two moults after the operation. (A,B) These represent the unexpected irregular patterns that are found after about half of the operations in the region where the 3/4 segment border would have been situated in an unwounded animal (arrows). The cuticular pattern is very variable. The common feature is some cellular elongation in the centre of the scrape. (A) Depigmented whole mount viewed under phase contrast of a very extreme example. The distorted cuticular surface is quite different from the pattern of irregular hexagons found over most of a normal unwounded segment (Fig. 1D). (B) Cellular elongation is very clear in the centre of this stained whole mount preparation. In others, a much wider band of elongated cells may be present. Anterior at left. Bars: 0·025 mm.
0·6 bord scrapes. Recently moulted 5th instars two moults after the operation. (A,B) These represent the unexpected irregular patterns that are found after about half of the operations in the region where the 3/4 segment border would have been situated in an unwounded animal (arrows). The cuticular pattern is very variable. The common feature is some cellular elongation in the centre of the scrape. (A) Depigmented whole mount viewed under phase contrast of a very extreme example. The distorted cuticular surface is quite different from the pattern of irregular hexagons found over most of a normal unwounded segment (Fig. 1D). (B) Cellular elongation is very clear in the centre of this stained whole mount preparation. In others, a much wider band of elongated cells may be present. Anterior at left. Bars: 0·025 mm.
Discussion
(A) Cell behaviour during the generation of segment borders
Cells located in the anterior region of a segment can be confronted with those found in the posterior region either by removing the epidermis at a segment border (0·1 bord scrape), confronting cells from adjacent segments, or by ablating most of the segment (0·5 mid scrape), confronting cells from the same segment. Both these operations normally result in localized cell division and transverse elongation of the cells at the confrontation site (Figs 3, 5). Two observations show that this behaviour cannot be attributed simply to wound healing. Firstly, when an identically sized wound to that used to remove the segment border is performed in the centre of the segment (0·1 mid) no cell division is found (Fig. 4). Secondly, 0·6 bord scrapes, which confront cells from the same position in adjacent segments, result in much less cell division than after 0·5 mid scrapes even though the wounds are similar in size (Fig. 6).
Two possible explanations for the cell behaviour pattern after 0·1 bord and 0·5 mid scrapes are that it may result from the interaction of cells with large differences in positional values or that it may be directly associated with features peculiar to regeneration of the segment border. The second possibility is ruled out because similar cell behaviour patterns can be found after confrontations that do not result in segment border formation (0·3 post scrapes in Campbell, 1987b). Therefore, it seems likely that the cell division and cell elongation found after these two operations are associated with a large discontinuity in positional values at the site of confrontation.
The confrontation of cells with different positional values has been shown to result in localized cell division at the site of confrontation during regulation in the circumferential axis of the cockroach leg (Anderson & French, 1985) and the Drosophila imaginai disc (Dale & Bownes, 1980; Adler, 1984; Kiehle & Schubiger, 1985; O’Brochta & Bryant, 1987). Such localized cell division has also been demonstrated indirectly following the confrontation of cells from different levels in the anteroposterior axis of the insect segment (Nübler-Jung, 1977). Direct evidence for such localized division and evidence that the amount of division stimulated is related to the difference in positional values between confronted cells is also presented in the accompanying paper (Campbell, 1987b). If this is true, the cell behaviour results presented here are explicable in terms of the traditional concept of the segment in which there is a segmentally repeated gradient of positional values with a discontinuity in values at the segment borders (Locke, 1959, 1960; Stumpf, 1966; Lawrence, 1966, 1973; Fig. 12). In this model, a much greater discontinuity would be created by a 0·1 bord scrape than any other 0·1 scrape within the segment and therefore more division would be expected.(Fig. 12).
To explain the cell behaviour results the segment is represented by a gradient in positional values with a discontinuity in positional values at the segment border. It is suggested that the amount of division found after a scrape depends on the difference in positional values between confronted cells. Thus, the high mitotic activity found after 0·1 bord and 0·5 mid scrapes is explained because these operations result in large discontinuities (Xb and Xc).
The regeneration of segment borders and the formation of supernumerary borders is explained if the discontinuity at the segment border is maintained by’ a special population of border cells and if these cells are generated at the site of a very large discontinuity in positional values. According to the classic gradient hypothesis, the discontinuity produced by the ablation of the segment border (0·1 bord) should result in intercalation of intermediate intrasegmental levels and a region of reversed polarity (Bii). Although such a region is occasionally found, normally’ a segment border regenerates. This is explained if the large discontinuity is stabilized by the evolution of the special border cells. A similar explanation can be used to explain the formation of a supernumerary’ border after 0·5 mid scrapes. In this case, if border cells are polarized and the new ones adopt reversed polarity, the regions of reversed polarity anterior and posterior to the supernumerary border are explained by intercalation between the new border cells and the remaining cells of the segment.
The small amount of division usually found after 0·6 bord scrapes is explained if cells at the same level in adjacent segments have the same positional value. The region of high mitotic activity’ that is occasionally found after this scrape is explained if there is some change in positional value before confrontation. This would also explain why half of these scrapes result in an irregular pattern rather than the fusion of the two segments.
To explain the cell behaviour results the segment is represented by a gradient in positional values with a discontinuity in positional values at the segment border. It is suggested that the amount of division found after a scrape depends on the difference in positional values between confronted cells. Thus, the high mitotic activity found after 0·1 bord and 0·5 mid scrapes is explained because these operations result in large discontinuities (Xb and Xc).
The regeneration of segment borders and the formation of supernumerary borders is explained if the discontinuity at the segment border is maintained by’ a special population of border cells and if these cells are generated at the site of a very large discontinuity in positional values. According to the classic gradient hypothesis, the discontinuity produced by the ablation of the segment border (0·1 bord) should result in intercalation of intermediate intrasegmental levels and a region of reversed polarity (Bii). Although such a region is occasionally found, normally’ a segment border regenerates. This is explained if the large discontinuity is stabilized by the evolution of the special border cells. A similar explanation can be used to explain the formation of a supernumerary’ border after 0·5 mid scrapes. In this case, if border cells are polarized and the new ones adopt reversed polarity, the regions of reversed polarity anterior and posterior to the supernumerary border are explained by intercalation between the new border cells and the remaining cells of the segment.
The small amount of division usually found after 0·6 bord scrapes is explained if cells at the same level in adjacent segments have the same positional value. The region of high mitotic activity’ that is occasionally found after this scrape is explained if there is some change in positional value before confrontation. This would also explain why half of these scrapes result in an irregular pattern rather than the fusion of the two segments.
(B) Segment border generation and the gradient hypothesis
A regenerated or supernumerary segment border is usually found two moults after 0·1 bord and 0·5 mid scrapes respectively (Figs 8B, 9). These results confirm the observations of Wright & Lawrence (1981) who produced similar confrontations by burning and excising rather than by scrapes. They explained their results by suggesting that the segment border could be intercalated (via the shortest route) like any other pattern element within the segment, i.e. that the series of positional values was continuous across the segment border. This interpretation is at odds with the results described above because their model would predict a similar cell behaviour pattern after both 0·1 mid and 0·1 bord scrapes and this is not so. Because of this we would like to propose an alternative hypothesis to explain the regeneration of segment borders.
The regeneration of segment borders is difficult to explain in terms of the classic gradient model because the confrontation of cells with different positional values should result in the intercalation of intermediate positional values rather than the formation of the proposed point of discontinuity (Fig. 12). Although this can occur after the ablation of the segment border (Fig. 8C), normally the border is regenerated. This can be explained by considering how the discontinuity in positional values might be maintained at the segment border. Normally, an isolating mechanism must be in operation because no cell division is found at an unwounded segment border in starved larvae. Such an isolating mechanism may involve the special population of cells present at the segment border that have been identified in Oncopeltus by their special permeability properties (Blennerhassett & Caveney, 1984). These are the transversely elongated cells found at the anterior margin of each abdominal segment. Under certain in vitro conditions they present a barrier to the spread of fluorescent dye injected into cells located either anteriorly or posteriorly. This dye will spread from cell to cell within the segment. If these cells are able to isolate two populations of cells with the maximum difference in positional values then the discontinuity produced by confrontation of cells with a large difference in positional values (such as produced by the ablation of the segment border) may be removed by the formation of such special cells (from normal segment cells), rather than by intercalation of intermediate values. This would explain the regeneration of segment borders after 0·1 bord scrapes and the formation of supernumerary segment borders after 0·5 mid scrapes (Fig. 12). The polarity pattern anterior and posterior to a supernumerary border appears to consist of mirror-image duplications of the anterior-most and posterior-most parts of the segment (Fig. 9B). If the hypothetical border cells are polarized then this pattern can be explained by intercalation between the new border cells in the centre of the segment and the cells at each end of the segment, assuming the new border cells adopt a reversed polarity (Fig. 12).
(C) Cell behaviour and final pattern after 0-6 bord scrapes
If the sequence of positional values is repeated within each segment, cells located at the same position in adjacent segments should have the same positional value. This is supported by the observations that 0·6 bord scrapes usually result in little cell division and for half the cases in a final pattern that consists of an abnormally long, but correctly patterned, segment extending from the 2/3 to the 4/5 border (Figs 7,10). However, occasionally a region of high mitotic activity is found at the site of confrontation. In the remaining half that were allowed to moult, the final pattern is irregular and difficult to interpret (Fig. 11). A complete explanation for the morphology of these animals is not possible, but an explanation of why they do not form a normal but double-length segment extending from the 2/3 border to the 4/5 border, is required. One possible explanation is that the positional values of cells at the leading edges of the migrating epidermal sheet may change as they migrate over the wound (Fig. 12). If this process occurs before confrontation, it would result in a discontinuity that is resolved by local intercalation. The observed pattern irregularities may be associated with such a process and would represent either abnormal intrasegmental levels or an abnormal border. If the extent of positional value change varies from animal to animal this would explain why the final pattern is also variable. Such a process could also explain the minority of preparations which show a region of high mitotic activity and cell elongation in the centre of these scrapes in 4th instars. This idea of positional value change by a process other than intercalation in the cells at the free edge of a wound has also been suggested to occur in other systems, including Drosophila imaginai discs (Karpen & Schu-biger, 1981) and crustacean limbs (Shelton, Truby & Shelton, 1981).
(D) Is segment border generation an epimorphic process?
The accompanying paper (Campbell, 1987b) presents evidence that favours the view that pattern regulation within the abdominal segment of Oncopeltus is largely epimorphic. However, although the cell division during the regeneration of intrasegmental levels may correspond to the intercalation of intermediate positional values, the situation during the generation of segment borders may be different. Here, according to the present model, protracted cell division occurs until special border cells isolate regions of positional value disparity. It would be expected that such episodes of prolonged mitotic activity would generate new positional values. This process could reduce the positional value disparity and bands of reversed polarity adjacent to the border would be expected in many cases. Since that only rarely occurs an explanation is required. There are two possibilities. First, the cell division associated with border formation is not associated with the generation of new positional values, but has some unknown function. Second, the regenerated border has organizing properties (Spe-mann, 1938) and repolarizes any region of reversed polarity formed during the regeneration process.
(E) Segments as fields
The suggested discontinuity in positional values at the segment border is probably maintained by the special population of cells that have been identified at this position because of their unique permeability properties (Blennerhassett & Caveney, 1984). If the present interpretation is correct these cells may well form the boundaries in a segmentally repeated series of developmental fields. The two faces of a special border cell would then have opposite and extreme positional values. Recently it has been shown that there is a special population of cells at the boundary between leg segments in the grasshopper embryo (Caudy & Bentley, 1986) and these cells may have a similar function to that suggested for the cells at the Oncopeltus segment border. If segment borders do represent field boundaries they might be expected to be responsible for establishing the segmental gradient (Wolpert, 1969, 1971). The formation of segment borders (the proposed site of discontinuity and field boundary) within the segment (the field) due to the interaction of cells with large differences in positional values (Campbell, 1987b) has implications for other systems. The Wright & Lawrence (1981) model explains the regeneration of segment borders by suggesting the series of positional values is continuous even across the segment border. Because the present results suggest that this is incorrect, there is the possibility that other models which explain regulation in terms of a continuous series of positional values may actually be overlooking sites of discontinuity. For example, positional values in the circumferential axis of the cockroach leg have been described as a continuous series in the polar coordinate model (French et al. 1976), but in fact may contain discontinuities, perhaps at the compartment borders (French, 1980).
ACKNOWLEDGEMENTS
G.L.C. would like to thank the MRC for a research studentship.