The interactions occurring between host and graft leg epidermis at a non-congruent junction were studied in the cockroach, Blatella germanica. Graft and host tibia were cut perpendicular to the proximal-distal axis and two heteropleural combinations were used to reverse separately the two transverse axes of the graft relative to the host. Use of dark and light cuticle colour mutants gave a good indication of the graft or host origin of regenerated structures.

Graft/host junctions regenerated segmented structures in various spatial arrangements, always comprising two copies of all structures distal to the level of the junction.

It is concluded that the categories - two separate laterals, double lateral, completely and partially autonomous regeneration - reflect two processes.

  • If the graft tarsus is removed, graft and host may not heal together and interact, but form autonomous regenerates lying in mirror-image symmetry separating original graft and host levels.

  • If interaction occurs between graft and host (or their developing autonomous regenerates) two laterals of dual origin are produced, one from each point of transverse axis incongruity. These laterals may secondarily fuse together to form a double structure originating from a point of congruity. The orientation and composition of the component tarsi of the double structure depend on the site of origin and the extent to which the two laterals fuse.

It is argued that the four ‘faces’ and two ‘transverse axes’ of the leg are merely descriptive terms. A new model is developed whereby lateral regeneration arises directly from the circumferential organisation of the leg epidermis. Previous work has shown that position is specified continuously around the circumference, and that intercalary regeneration occurs by the shortest route between confronted positions. After reversal of one ‘transverse axis’ the shortest route between confronted graft and host positions is different on the two sides of each of the two points of ‘axis’ incongruity, and at these points the two halves of a complete circumference are formed. These lateral circumferences, like the terminal circumference exposed by amputation, cannot heal over by intercalary regeneration, and this leads to regeneration of distal structures.

The model accounts for lateral regeneration after reversal of both ‘transverse axes’ by 180° rotation of a homopleural graft.

The possibility is discussed that there may be clonal restrictions on the circumferential positions which the progeny of a cell may occupy.

The cuticle of insects is secreted by the underlying single layer of epidermal cells, and a study of the cuticular patterns formed after surgical operations which remove or spatially disturb part of the epidermis can give information about the systems of cellular interaction within the epidermis which must form some sort of ‘map’ from which the cells derive their ‘positional information’ (Wolpert, 1969; Lawrence, 1970; Wolpert, 1971; Bryant, 1974).

Grafting together two different proximal-distal levels of a segment (e.g. tibia) results in longitudinal intercalary regeneration of the intermediate levels (Bohn, 1967 & 1970; Bullière, 1971; French, 1976a). Similarly, after association of different circumferential positions, there is transverse intercalary regeneration of the intermediate positions (French & Bullière, 1975a, b).

Lateral regeneration of supernumerary legs can also occur from a graft/host junction. The leg has been considered to be organized along three mutually perpendicular axes (Bohn, 1965): a longitudinal (proximal-distal) axis running from the articulation with the body to the claws, and two transverse axes (anterior-posterior and internal-external). Two complete lateral regenerates have been shown to form from a non-congruent junction (i.e. host and graft transverse axes not aligned) in several species of cockroach: Periplaneta americana (Penzlin, 1965), Leucophaeamaderae (Bohn, 1965, 1972b),Leucopheael Gomphadorhina portentosa combinations (Bohn, 1972a), and Blabera craniifer (Bullière, 1970b). Similar laterals have also been produced in the stick insect, Carausius morosus (Bart, 1971a), and in Lepidoptera (Bodenstein, 1937), Hemiptera (Shaw & Bryant, 1975), Dermaptera (Furokawa, 1940) and in Arachnida (Lheureux, 1970, 1971).

There has been general agreement about the number, position and orientation of the laterals resulting from various graft combinations, but different conclusions have been reached concerning the tissue composition (graft or host origin) of the laterals. Grafts have been made between different legs of animals of the same species (Bart, Bodenstein, Bullière, Lheureux, Penzlin) or between the legs of animals of different species (Bohn). Laterals have been considered to be of pure graft or pure host origin, forming because of a failure of the two components to interact (Bodenstein, Bullière, Lheureux, Penzlin), or it has been concluded that the laterals are usually of dual graft and host origin, produced by an interaction between misaligned regions of host and graft (Bart, Bohn).

In the present study, non-congruent graft/host junctions were made at the tibia level in Blatella germanica to confirm lateral regeneration in this species of small cockroach, to provide more information about the range of possible regenerated structures, and to determine the composition of the regenerated structures by grafting between cuticle colour mutants (French, 1976a). The results obtained from tibial junctions can also be used for comparison with the regenerates produced from congruent and non-congruent junctions between non-homologous leg segments (French, 1976b).

The way in which the non-aligned graft and host tibial epidermis interacts (or fails to interact) to produce the regenerated structures, is discussed. Various earlier theories of the origin of lateral regenerates are found to be unsatisfactory and a new model is developed.

This model considers the leg epidermis to be effectively two-dimensional: the surface of a cylinder upon which cellular interactions can occur longitudinally and circumferentially. Lateral regeneration is shown to arise as a direct result of this spatial organization of the epidermis.

Laboratory colonies of Blatella germanica were maintained as described previously (French, 1976 a), and the grafting operations were performed on 3rd and 4th instar larvae 1 or 2 days after moulting. In all experiments some grafts were made on wild-type animals (usually between different legs of the same animal), and some were made between the dark cuticle colour mutant, Bl (Ross & Cochran, 1967) and the light cuticle colour mutant, br (French, 1976a). Grafting operations were performed under the dissecting microscope, using fine forceps and small spring scissors. Animals were immobilized with CO2. Graft and host tibiae were chosen to be of slightly different diameters so the graft could be pushed slightly into the end of the host stump, where it was secured by dried haemolymph. Experimental animals were kept until a few days after the 1st or 2nd post-operative moult (p.o.m.) and then the operated leg was removed fixed and cleared in Gum Chloral, and examined.

The structure of the leg of Blatella germanica and the labelling convention to be used have both been described previously (French, 1976a). The leg will be described in terms of four ‘faces’: anterior, posterior, internal and external (but see Discussion). The position and orientation of structures produced after the operations will be given with reference to the host axes. It should be noted that a regenerated tarsus is 4-segmented instead of 5-segmented (O’Farrel, Stock, Rae & Morgan, 1960).

The two basic operations involved reversing separately the two transverse axes of the graft with respect to the host, and are illustrated in Fig. 1. The right proor meso-thoracic donor leg was removed at proximal tibial level and grafted on to the host left meta-thoracic leg which had been amputated at mid-tibia level. The graft tarsus was either amputated (as in Fig. 1) or left intact.

Fig. 1

Structures regenerated from the graft/host junction following reversal of one transverse axis at the level of the tibia. (A) Reversal of the anterior-posterior axis; (B) reversal of the internal-external axis. All views are anterior (with respect to host axes) except A(b1), A(b4), B(b2) and B(b5) which are external, (a) Graft situation. (b1b5) Classes of result after 1st or 2nd post-operative moult, (b1) Two separate laterals; (b2) double lateral (one of the possible positions and orientations); (b3) completely autonomous regeneration; (b4, b5) partially autonomous regeneration. A, Anterior; I, internal; P, posterior; E, external.

Fig. 1

Structures regenerated from the graft/host junction following reversal of one transverse axis at the level of the tibia. (A) Reversal of the anterior-posterior axis; (B) reversal of the internal-external axis. All views are anterior (with respect to host axes) except A(b1), A(b4), B(b2) and B(b5) which are external, (a) Graft situation. (b1b5) Classes of result after 1st or 2nd post-operative moult, (b1) Two separate laterals; (b2) double lateral (one of the possible positions and orientations); (b3) completely autonomous regeneration; (b4, b5) partially autonomous regeneration. A, Anterior; I, internal; P, posterior; E, external.

Anterior /posterior (A/P) reversal

The graft was rotated by 180° about its longitudinal axis, leaving internal and external host faces adjacent to the corresponding graft faces, but confronting anterior and posterior host with posterior and anterior graft, respectively (Fig. 1A).

Internal/external (J/E) reversal

The graft was not rotated, leaving anterior and posterior host faces aligned with the corresponding graft faces, but confronting internal and external host with external and internal graft, respectively (Fig. 1B).

(A) Frequency of regeneration from the graft /host junction

As shown in Table 1, most operated legs which retained the graft regenerated segmented structures from the junction by the 1st p.o.m. and all had done so by the 2nd p.o.m.

Table 1

Frequency of regeneration from the tibial graft/host junction following reversal of one transverse axis of the graft

Frequency of regeneration from the tibial graft/host junction following reversal of one transverse axis of the graft
Frequency of regeneration from the tibial graft/host junction following reversal of one transverse axis of the graft

(B) Classification of regenerates from the graft /host junction

Structures produced from the junction are classified in Table 2, and it will be seen that the two experiments give rise to the same categories of structures with comparable frequencies. In each experiment, some of the animals with graft tarsus left intact at the time of operation subsequently lost the tarsus (and regenerated four tarsal segments by the 1st p.o.m.). These are therefore considered together with those animals which had the graft tarsus amputated at the time of grafting. Table 3 relates the state of the graft tarsus to the structure regenerated from the graft/host junction. Each category of regenerate will now be considered.

Table 2

Classification of regenerates from the tibial graft/host junction following reversal of one of the transverse axes of the graft

Classification of regenerates from the tibial graft/host junction following reversal of one of the transverse axes of the graft
Classification of regenerates from the tibial graft/host junction following reversal of one of the transverse axes of the graft
Table 3

Relationship between state of the graft tarsus and regeneration from the graft /host junction following reversal of one transverse axis of the graft (combined data from the two experiments)

Relationship between state of the graft tarsus and regeneration from the graft /host junction following reversal of one transverse axis of the graft (combined data from the two experiments)
Relationship between state of the graft tarsus and regeneration from the graft /host junction following reversal of one transverse axis of the graft (combined data from the two experiments)

(i) Two separate lateral regenerates

These structures could appear at the 1st or 2nd p.o.m. (Table 2), and at the junction between the host and a graft which had retained its tarsus intact or had lost and regenerated the tarsus (Table 3). The combination of graft proximal tibia and host mid tibia often gave rise to reversed orientation intercalary regeneration (French, 1976 a) and the lateral regenerates occurred at its proximal limit (NB, all positions and orientations refer to the host axes).

The two lateral structures each comprised a tibial apex (the coronet of spines and articulation with the condyle on the external side of the proximal tarsus), a 4-segmented tarsus and a set of two claws (Figs. 1 A(b1), B(b1), 2, 3).

Fig. 2

Separate lateral regenerates formed from the graft/host junction after reversal of the anterior-posterior axis (A and B) or the internal-external axis (C and D). A, I, P, E, Anterior, internal, posterior and external ‘faces’, gc, Claw of graft origin; gh, tarsal hair of graft origin; gs, coronet spine of graft origin; he, claw of host origin; hh, tarsal hair of host origin; hs, coronet spine of host origin.

Fig. 2

Separate lateral regenerates formed from the graft/host junction after reversal of the anterior-posterior axis (A and B) or the internal-external axis (C and D). A, I, P, E, Anterior, internal, posterior and external ‘faces’, gc, Claw of graft origin; gh, tarsal hair of graft origin; gs, coronet spine of graft origin; he, claw of host origin; hh, tarsal hair of host origin; hs, coronet spine of host origin.

Fig. 3

Camera lucida drawings of separate lateral regenerates formed from the graft/host junction after reversal of the anterior-posterior axis (A) or the internalexternal axis (B). A, I, P, E, Anterior, internal, posterior and external ‘faces’.

Fig. 3

Camera lucida drawings of separate lateral regenerates formed from the graft/host junction after reversal of the anterior-posterior axis (A) or the internalexternal axis (B). A, I, P, E, Anterior, internal, posterior and external ‘faces’.

The position of origin of the laterals around the circumference of the tibia depended on which transverse axis had been reversed by the operation. Following reversal of the anterior/posterior (A/P) axis of the graft, the laterals were usually (13/17 cases) positioned one anteriorly and one posteriorly (Figs. 1A (b1), 2A, B, 3 A); following reversal of the internal/external (I/E) axis, the laterals were usually (21/29 cases) found one internally and one externally (Figs. 1B (b1), 2C, D, 3B). Thus, in both experiments, a separate lateral was formed at each region of incongruity between the transverse axes of host and graft: at each confrontation of opposite faces of the leg.

The A/P orientation of the laterals could not be determined, but their I/E axis was almost always polarized in accordance with host and graft following reversal of the graft A/P axis (15/17 cases), and in accordance with the host following reversal of the graft I/E axis (27/29 cases).

(ii) Double lateral regenerates

Double lateral regenerates appeared at the 1st or 2nd p.o.m. (Table 2) on legs which had retained the graft tarsus intact or had lost and regenerated it (Table 3). As in the case of the separate laterals, the double lateral structure was regenerated at the proximal limit of any reversed orientation intercalary regenerate formed between host and graft.

In all cases there was a large tibial apex (probably two apices fused together but this could not be determined), articulating with a large double tarsal structure having two proximal condyles and two distal sets of two claws. The tarsus was usually a double structure along its entire length (Fig. 4, 1A (b2), I B (b2)) but sometimes had two separate distal parts (Fig. 5D).

Fig. 4

Double lateral structures regenerated from the graft/host junction after reversal of the anterior-posterior axis (A and B which are anterior and posterior views of the same specimen; C, D and E) or the internal external axis (F). A, I, P, E, Anterior, internal, posterior and external ‘faces’. A, B, C and F show a double lateral with one component tarsus more or less host-derived and the other more or less graft-derived. In D and E each component tarsus is of dual origin, c, c1, Denote the two sets of claws; gc, gc1, claw of graft origin; gs, coronet spine of graft origin; hc, hc1, claw of host origin; hh, tarsal hair of host origin; hs, coronet spine of host origin.

Fig. 4

Double lateral structures regenerated from the graft/host junction after reversal of the anterior-posterior axis (A and B which are anterior and posterior views of the same specimen; C, D and E) or the internal external axis (F). A, I, P, E, Anterior, internal, posterior and external ‘faces’. A, B, C and F show a double lateral with one component tarsus more or less host-derived and the other more or less graft-derived. In D and E each component tarsus is of dual origin, c, c1, Denote the two sets of claws; gc, gc1, claw of graft origin; gs, coronet spine of graft origin; hc, hc1, claw of host origin; hh, tarsal hair of host origin; hs, coronet spine of host origin.

Fig. 5

Camera lucida drawings of double laterals regenerated from the graft/host junction after reversal of the anterior-posterior axis (A, B) or the internal-external axis (C, D). A, I, P, E, Anterior, internal, posterior and external ‘faces’.

Fig. 5

Camera lucida drawings of double laterals regenerated from the graft/host junction after reversal of the anterior-posterior axis (A, B) or the internal-external axis (C, D). A, I, P, E, Anterior, internal, posterior and external ‘faces’.

The position of origin of the double lateral around the circumference of the tibia differed in the two experiments. Where the A/P axis of the graft had been reversed, the lateral occurred in an external (25/43 cases, Figs. 4A, B, C, 5A, B) or internal (17/43 cases, Figs. 4D, E) position; following reversal of the I/E axis, the lateral usually originated posteriorly (30/34 cases, Figs. 4F, 5C). Thus, in both experiments, the double lateral was usually formed at a region of congruity between the transverse axes of host and graft: at a position midway between the two regions of incongruity where the laterals form when they are separate.

Only the I/E orientation of the double tarsi could be determined and, as shown in Fig. 6, the orientations were very variable. The tarsi were always orientated in one plane, but this could be the plane of the graft and host I/E axes (Fig. 5 A) or perpendicular to it (Fig. 5B). The component tarsi were almost always oppositely orientated, with the two sets of claws facing directly towards (Fig. 5 A) or away from each other (Fig. 5B). The range of variation is shown in Fig. 6 as it is important in connexion with the composition and modes of origin of the double structures.

Figure 6

Position, orientation and composition of double lateral regenerates developing from the graft/host junction following reversal of one transverse axis. A, Results of experiment reversing the anterior-posterior axis; B, results of experiment reversing the internal-external axis. Figures denote number of cases.

‘Position’: schematic representation of the graft/host junction, distal view. The outer circle shows the orientation of the host; the inner circle represents the graft; the asterisk shows the position of the double lateral around the circumference. A, Anterior; I, internal; P, posterior; E, external.

‘Orientation’-‘end-on’ view of the double lateral structures, showing the orientation with respect to the host axes, with position of origin of the lateral at the top to facilitate comparison with Fig. 8. Tarsal claws curve from external to internal ‘faces’ on the component tarsi which are shown separated by a dashed line.

‘Composition’-‘end-on’ view of the double laterals showing approximate division into host-derived (stippled) and graft-derived parts. ‘?’-unknown composition; ‘1 host/1 graft’, one component tarsus of host origin and the other of graft origin; ‘both dual’ - each component tarsus of dual graft and host origin.

Figure 6

Position, orientation and composition of double lateral regenerates developing from the graft/host junction following reversal of one transverse axis. A, Results of experiment reversing the anterior-posterior axis; B, results of experiment reversing the internal-external axis. Figures denote number of cases.

‘Position’: schematic representation of the graft/host junction, distal view. The outer circle shows the orientation of the host; the inner circle represents the graft; the asterisk shows the position of the double lateral around the circumference. A, Anterior; I, internal; P, posterior; E, external.

‘Orientation’-‘end-on’ view of the double lateral structures, showing the orientation with respect to the host axes, with position of origin of the lateral at the top to facilitate comparison with Fig. 8. Tarsal claws curve from external to internal ‘faces’ on the component tarsi which are shown separated by a dashed line.

‘Composition’-‘end-on’ view of the double laterals showing approximate division into host-derived (stippled) and graft-derived parts. ‘?’-unknown composition; ‘1 host/1 graft’, one component tarsus of host origin and the other of graft origin; ‘both dual’ - each component tarsus of dual graft and host origin.

(iii) Completely autonomous regeneration

Completely autonomous regeneration of the host and graft surfaces occurred by the first p.o.m. but did not occur subsequently in animals which had not regenerated from the junction at the first p.o.m. (Table 2). The distal parts of the long and fragile structures were often broken off but, from the seven, cases where this had not occurred, autonomous regeneration was seen to have occurred only in association with loss of graft tarsus at or after the operation (Table 3).

Following reversal of either axis, the autonomous structure consisted of two regenerates each comprising distal tibia, four-segmented tarsus and two claws, lying in mirror-image linear sequence separating the original graft and host levels, and joined by their distal tips (Figs. 1A (b3), IB (b3), 7 A, B).

Figure 7

Autonomous structures regenerated from the graft/host junction after reversal of the anterior-posterior axis (B, C, D, E) or the internal-external axis (A). A, B, Completely autonomous regeneration; C, D, E, partially autonomous regeneration with a double lateral structure, with one component of host origin and the other of graft origin (C, D), or both of dual origin (E).

A,1, P, E, Anterior, internal, posterior and external ‘faces’, c, c1, the two sets of claws, g, graft; gc, gc1, claw of graft origin; gh, tarsal hair of graft origin; gr, autonomous regenerate from graft component of the junction; h, host; he, he1, claw of host origin; hh, tarsal hair of host origin; hr, autonomous regenerate from host component of the junction; It, double lateral structure; rt, tarsus regenerated from the distal end of the graft.

Figure 7

Autonomous structures regenerated from the graft/host junction after reversal of the anterior-posterior axis (B, C, D, E) or the internal-external axis (A). A, B, Completely autonomous regeneration; C, D, E, partially autonomous regeneration with a double lateral structure, with one component of host origin and the other of graft origin (C, D), or both of dual origin (E).

A,1, P, E, Anterior, internal, posterior and external ‘faces’, c, c1, the two sets of claws, g, graft; gc, gc1, claw of graft origin; gh, tarsal hair of graft origin; gr, autonomous regenerate from graft component of the junction; h, host; he, he1, claw of host origin; hh, tarsal hair of host origin; hr, autonomous regenerate from host component of the junction; It, double lateral structure; rt, tarsus regenerated from the distal end of the graft.

The I/E orientation of the more proximal regenerate conformed to that of the host, while the more distal regenerate was orientated like the graft.

(iv) Partially autonomous regeneration

Partially autonomous regeneration of the graft and host surfaces could occur by the first or second p.o.m. (Table 2). As in the case of the completely autonomous regenerates, distal parts of the long and fragile structures were often broken off, but, from the 14 complete structures, it seems that partially autonomous regeneration usually occurred in association with loss of the graft tarsus at or after the operation (Table 3).

The structures consisted of two partial regenerates lying in mirror-image linear sequence separating host and graft, and fused together at some level proximal to the end of the tarsus. This level could be the tibia apex or in any of the tarsal segments but was almost always the same for the two elements concerned. From this level of fusion were formed either two separate laterals or a double lateral branch, complete to two sets of claws (Figs. 1A (b4), (b5) 1B(64), (b5), 7C–E).

As in the case of the completely autonomous regenerates, the two axially situated partial regenerates were orientated (in the I/E axis) like the host and graft respectively. The lateral elements were positioned and orientated just like the laterals originating directly from the graft/host junction. Following A/P reversal, the separate lateral elements were positioned one anteriorly and one posteriorly, and were orientated like the graft and host; double lateral elements were positioned externally or internally and had a range of orientations similar to those in Fig. 6. After I/E reversal, separate laterals were positioned internally and externally, and were both orientated like the host; double laterals were usually positioned posteriorly, with various orientations.

(v) Other structures

In four cases only a single lateral was regenerated from the graft/host junction and, in another four cases the leg bore a poorly segmented lateral ending in two atypical claws. One of these animals was allowed to moult again, and produced a normal double lateral.

Structures regenerated from the junction - Summary

In all cases where operated legs had retained the graft until the second p.o.m., the graft/host junction developed segmented structures in various spatial arrangements but always comprising two copies of all structures normally lying distal to the level of the junction.

(C) Composition of regenerates from the graft/host junction

The cuticle colour difference between Bl and br Blatella proved to be of definite, though limited, use in determining the origin of the epidermis of regenerated structures (French, 1976a). Although the colour difference between the original host and graft tissues was usually clear, it was usually not possible to draw a precise boundary between them or their derivatives. This may just reflect an insufficient difference in colour and some cell mixing at the boundaries, or there may be a local interaction affecting cellular phenotype near the boundary (French, 1976a). In addition, lateral or autonomous structures regenerated from the graft/host junction were often fragile and poorly pigmented, reducing the number of analysable cases. Although only an approximate boundary could be drawn over areas of bare cuticle, the spines of the tibial apex, well-developed tarsal hairs and the tarsal claws could usually be easily identified as Bl or br, and the origin of the epidermis of the regenerates will be deduced from these criteria.

(i) Separate lateral regenerates

All cuticle on the axial member proximal to the bases of the laterals was host-derived and all axial cuticle distal to them was graft-derived. Hence the reversed orientation intercalary regenerate (if any) was of graft origin.

Grafts between Bl and br animals reversing the graft A/P axis gave eight analysable legs with separate laterals anteriorly and posteriorly. In one case one lateral appeared to be of dual origin and the other of host origin, but in all other cases both laterals were clearly of dual origin. They had host-derived spines in the tibial coronet and a host-derived claw on the side adjacent to the host, and graft-derived spines and claw on the side adjacent to the graft (Fig. 2A, B).

After reversal of the I/E axis there were 11 analysable legs with separate laterals internally and externally (or approximately internally and externally). In nine cases one lateral tarsus seemed host-derived with two claws of host origin, and the other seemed graft-derived with both claws of graft origin. However, many of these laterals were really of dual origin as indicated by the presence of both host- and graft-derived spines in the lateral coronets (Fig. 2C). In the other two cases, one of the laterals had one host-derived and one graft-derived claw (Fig. 2D). These laterals developing after I/E axis reversal are so orientated that a division into side-adjacent-to-host and side-adjacent-to-graft (which separated the claws of the A/P axis reversal laterals) divides them into an external half and an internal half, and hence does not separate the claws. The two claws of a lateral tarsus might be expected to have approximately the same composition, and the dual origin of such a tarsus would not be obvious.

(ii) Double lateral regenerates

The compositions of the double lateral structures are given in Fig. 6. In all cases the double structure was composed of both host and graft-derived tissue.

The double laterals regenerating from internal or external positions following reversal of the A/P axis had one of the component tarsi more or less host derived and the other more or less graft-derived when their I/E orientations were in the same plane as those of the graft and host (Figs. 4A, B, C). When the I/E orientations were perpendicular to that of the host and graft, each component tarsus was composed of both host and graft-derived tissue (on the sides adjacent to the host and graft respectively), as shown in Fig. 4D, E.

Following reversal of the I/E axis, the double laterals were usually composed of one more or less host-derived member and one more or less graft derived member, regardless of their point of origin on the circumference and their orientation (Fig. 4F). In two cases where the I/E orientations were perpendicular to those of the host and graft, component tarsi of the double lateral were each composed of both graft and host tissue.

(iii) Completely autonomous regenerates

As implied by the nomenclature, the autonomous regenerates were derived one from the host and the other from the graft (Fig. 7 A, B).

(iv) Partially autonomous regenerates

In the partially autonomous structure, the axial element lying between the original host level and the point of fusion was host-derived, and that lying between the point of fusion and the original graft level was graft-derived. There were a number of analyzable partially autonomous regenerates developed after reversal of the A/P axis and having a double lateral element. These were composed exactly as the double laterals developing directly from the corresponding graft/host junction. When they were orientated (in the I/E axis) in. the plane of the graft and host I/E axis, one of the components was more or less graft-derived and the other more or less host-derived (Fig. 7C, D). When they were orientated perpendicular to the graft and host, each component was composed of both graft and host tissue (Fig. 7E).

Composition of structures regenerated from the junction: Summary

Although use of the Bl and br mutants did not produce precise boundaries between hostderived and graft-derived parts of the structures regenerated from the non-congruent junction, it does allow certain conclusions to be made Lateral regeneration from the original graft/host junction or from the junction between autonomous graft-derived and host-derived partial regenerates involves both graft and host tissue.

When two separate laterals are produced they are usually of dual origin. This was especially obvious following reversal of the A/P axis, where the side adjacent to the host was host-derived and the other side was graft-derived. Double laterals formed after reversal of the A/P axis had either one host-derived and one graft-derived component tarsus, or each of dual origin depending on their orientation. When the I/E axis was reversed, one component tarsus was usually host-derived and the other graft-derived, regardless of orientation. The composition of the regenerated structures is clearly of great importance in considering their possible mode of origin, and this will be discussed below.

The results presented above will be compared with those of similar experiments performed on other insects by other workers. The available data will then be compared with the predictions of three theories which have been suggested to explain regeneration from a non-congruent junction. A new model will be developed for the spatial organization of the leg epidermis, and this will be used to explain regeneration from the non-congruent junction.

(A) Number, position and orientation of structures regenerated from the junction

The results of reversing the A/P or I/E axis at the tibia level of the Blatella leg agree very well with the results of the same experiment performed on other insects, in that the operation typically provoked the formation of two complete regenerates by the 1st or 2nd p.o.m. (Bohn (1965, 1972a) on Leucophaea and between Leucophaea and Gromphadorhina’, Bullière (1970b) on Blabera’, and Lheureux (1970) on the spider Tegenarid). This is in sharp contrast to the results from the congruent tibial graft/host junction which typically does not regenerate any segmented structures (Bohn, 1965, 1967, 1970; Bullière, 1970a, 1971; French, 1976a) although a congruent Blatella junction often regenerated partial structures if the graft tarsus had been amputated (French, 1976 a).

A/P and I/E axis reversal was also performed at the coxa level by Bohn (1972a), Bullière (1970 b) and by Bart (1971a) on the stick insect, with results comparable with those from the tibia level.

There is general agreement that the two separate laterals originate from the points of discontinuity between graft and host, and are orientated as shown in Fig. 1A (b)1), B (b1). The results from Carausius (Bart, 1971a) have two notable features: laterals appeared from exactly the points of discontinuity, and there was a relatively high frequency (20/57 cases) of formation of only a single lateral after reversal of either of the graft axes. Bullière (1970b) and Bart (1971a) found fusion between one of the laterals and the regenerated graft terminal tarsus (this was observed once in Blatella) but no cases of a double lateral. Bohn (1965, 1972a) obtained double laterals from a point of congruity (external from A/P reversal, and posterior from I/E reversal) as was found in Blatella, but he did not give details of their orientations.

Bohn found cases of partial autonomous regeneration at the tibial level (Bohn, 1965, fig. 7 a) and Penzlin (1965) found these structures in Periplaneta after reversal of both axes by 180° rotation of a homopleural graft. It is noticeable, however, that Bullière (from all of whose grafts the tarsus was removed) found no complete or partially autonomous regenerates, which were major categories of result in Blatella.

(B) Composition of separate lateral regenerates

Bullière’s grafts (1970b) were done between the pro- and meta-thoracic legs of Blabera. These legs can be distinguished by the relative sizes of their segments, and the presence of spines on the anterior/internal ridge of the femur of the pro-thoracic leg. Laterals regenerated from tibial level could not be identified as pro- or meta-thoracic, but Bullière concluded that laterals regenerated from coxa level were one of pure graft origin and the other of pure host origin. Lheureux (1971), grafting between pedipalps and hind legs of Tegenaria, reached the same conclusion using the criterion of one or two claws at the distal tip. In both of these experimental situations the criteria enable direct identification of only parts of the laterals.

Bohn (1972a) has shown that laterals of dual origin are the usual result of reversing one transverse axis at either tibia or coxa level. He grafted between differently pigmented species (Leucophaea and Gromphadorhina) and was able to draw boundaries between host- and graft-derived areas of a lateral. The boundaries almost always ran longitudinally down the lateral. After reversal of the A/P axis, Bohn found that nearly all lateral regenerates were of dual origin with the boundaries usually on the internal and external faces of the lateral. A lateral was divided into a host-derived half adjacent to the host, and a graft-derived half adjacent to the graft. This agrees completely with the composition of the corresponding Blatella laterals (Figs. 2 A, B). Separate laterals developed from an I/E reversal were less uniform, however, and were sometimes of pure graft or pure host origin (especially when produced at coxa level). Boundaries on dual origin laterals tended to be external and (in nearly every case) posterior, dividing the lateral into approximately three-quarters and one-quarter. These results correlate well with the tendency of the corresponding Blatella laterals to appear to be either host- or graft-derived (according to the limited criteria).

(C) The formation of double lateral regenerates

In both experiments the double laterals were formed in a region of congruity between the transverse axes of the graft and host or, in other words, midway between the sites where separate laterals would form. The existence of laterals which were double at the base but separated more distally (Fig. 5D) reinforces the view that the double laterals result from the secondary fusion of the two single laterals regenerating from the areas of incongruity.

Fig. 8 shows the various ways in which the two separate laterals could fuse, and comparison with Fig. 6 shows that most of the categories of double lateral can be explained.

Figure 8

Fusion of separate laterals regenerated from the graft/host junction after reversal of one transverse axis. Schematic representation of graft and host tibiae split internally (Aa, b1, b2), externally (Ac1, c2), posteriorly (Ba, b1, b2) or anteriorly (B c1, c2) and opened out flat, with the graft/host junction shown by the dashed line. A, I, P, E, Anterior, internal, posterior and external ‘faces’ of the host tibia;

(A), (I), (P), (E), ‘faces’ of the graft tibia. Laterals are shown ‘end-on’ developing out from the junction with host-derived tissue stippled (in A).

(A) Results from reversal of the anterior-posterior axis, (a) Two separate laterals; (b1, b2) fusion of the two laterals into a double lateral positioned externally. (b1) Incomplete fusion of the external ‘faces’ of the laterals, (b2) Complete fusion to give one double circumference with two sets of claws orientated from external to internal positions on the double circumference, (c1, c2) Fusion of the two laterals internally, (c1) Incomplete fusion of the internal ‘faces’ of the laterals, (c2) Complete fusion.

(B) Results from reversal of the internal-external axis, (a) Two separate laterals, (b1, b2) fusion of the two laterals into a double lateral positioned posteriorly. (b1) Incomplete fusion of the posterior ‘faces’ of the laterals, (b2) Complete fusion to give one double circumference, (c1, c2) Fusion of the two laterals anteriorly. (c1) Incomplete fusion of the anterior ‘faces’ of the laterals, (c2) Complete fusion.

Figure 8

Fusion of separate laterals regenerated from the graft/host junction after reversal of one transverse axis. Schematic representation of graft and host tibiae split internally (Aa, b1, b2), externally (Ac1, c2), posteriorly (Ba, b1, b2) or anteriorly (B c1, c2) and opened out flat, with the graft/host junction shown by the dashed line. A, I, P, E, Anterior, internal, posterior and external ‘faces’ of the host tibia;

(A), (I), (P), (E), ‘faces’ of the graft tibia. Laterals are shown ‘end-on’ developing out from the junction with host-derived tissue stippled (in A).

(A) Results from reversal of the anterior-posterior axis, (a) Two separate laterals; (b1, b2) fusion of the two laterals into a double lateral positioned externally. (b1) Incomplete fusion of the external ‘faces’ of the laterals, (b2) Complete fusion to give one double circumference with two sets of claws orientated from external to internal positions on the double circumference, (c1, c2) Fusion of the two laterals internally, (c1) Incomplete fusion of the internal ‘faces’ of the laterals, (c2) Complete fusion.

(B) Results from reversal of the internal-external axis, (a) Two separate laterals, (b1, b2) fusion of the two laterals into a double lateral positioned posteriorly. (b1) Incomplete fusion of the posterior ‘faces’ of the laterals, (b2) Complete fusion to give one double circumference, (c1, c2) Fusion of the two laterals anteriorly. (c1) Incomplete fusion of the anterior ‘faces’ of the laterals, (c2) Complete fusion.

Separate laterals formed after reversal of the A/P axis could fuse internally or externally (accounting for 42/43 of the structures obtained). If the laterals fuse incompletely in an external position (Fig. 8 A (b1)), each will become one of the component tarsi of the double structure and the claws will be orientated away from each other in the plane of the host A/P axis (6/25 cases, e.g. Fig. 5B). The component tarsi will each be of dual origin (5/5 of the analyzable results). Complete fusion of the laterals in an external position will make a double circumference (Fig. 8A(b2)). Claws curve from external towards internal so each component tarsus of the double structure will be formed from half of each of the original laterals. Hence the claws will be orientated towards each other in the plane of the host I/E axis (14/25 cases, e.g. Fig. 5 A), with one set of host origin and the other set of graft origin (8/8 cases).

[ufig]

Similarly, incomplete or complete fusion of laterals in an internal position (Fig. 8 A (c1), (c2)) will give claws orientated towards each other in the plane of the host A/P axis (13/17 cases) and each of dual origin (9/9 cases). Fusion of the separate laterals hence accounts for the position of 42/43, the orientations of 33/42, and the composition of 22/22 of the double laterals developed after A/P axis reversal.

Laterals developed after reversal of the I/E axis may fuse posteriorly or anteriorly (32/42 of the structures obtained). Incomplete fusion of laterals in a posterior position (Fig. 8B(b1)) will give claws side-by-side orientated in opposite directions in the plane of the host A/P axis (7/30 cases), as will incomplete fusion of laterals in an anterior position (2/2 cases, Fig. 5D, 8B(c1)). Complete fusion of the laterals posteriorly (Fig. 8B(b2)) will give claws orientated towards each other in a plane between the host A/P and I/E axes (13/30 of the observed double laterals were classified as approximately in the plane of the host A/P axis, e.g. Fig. 5C). Composition of these fused double laterals will depend upon the composition of the separate laterals: separate laterals developing after I/E reversal usually appear to be one host- and one graft-derived and this tendency is also seen in the components of the double structure. Fusion of the separate laterals hence accounts for the position of 32/42, and the orientations of 22/32 of the double laterals developed after I/E axis reversal.

Because of the correspondence between prediction and observation with respect to the position, orientation and (at least for the A/P reversal) the composition of the double laterals they will be assumed to have resulted from fusion of separate laterals.

(D) The formation of autonomous and partially autonomous regenerates

It was shown (French, 1976a) that completely and partially autonomous regeneration could occur from a congruent tibial junction (i.e. both transverse axes of host and graft in alignment) when healing together of host and graft was impeded by withdrawal of the graft epidermis from the cuticle, associated with the regeneration of an amputated graft tarsus. It was also argued that rotation of the graft further increased the chance that graft and host would not heal but would regenerate independently (at least initially). It is interesting that the autonomous categories of regeneration from a non-congruent junction also usually occur in conjunction with loss and regeneration of the graft tarsus (Table 3), suggesting that, in this situation also, they result from a failure of graft and host to heal together and interact.

The junction between the two axial elements of a partially autonomous regenerate resulting from a congruent graft regenerated no lateral or a single (usually incomplete) lateral, just like the original graft/host junction (French, 1976a). In the present study the junction of a partially autonomous structure developing after reversal of one transverse axis also behaved just like the corresponding graft/host junction. It regenerated two separate or a double fused lateral and, as described, these were orientated and probably composed exactly like the laterals regenerating from the original graft/host junction.

Thus the categories of lateral and autonomous regeneration reflect two different processes:

(i) If interaction occurs between host and graft (or between their developing autonomous regenerates) two laterals of dual origin are produced from the points of incongruity of the transverse axes. These laterals may subsequently fuse together.

(ii) If the graft and host surfaces do not heal together and interact, each regenerates independently. In Blatella these autonomous regenerates lie in linear sequence but in Blabera perhaps they may ‘slide’ past each other and project laterally as two ‘laterals’ of pure graft and pure host origin. This may explain why Bullière (1970b) found ‘pure origin’ laterals and no autonomous regenerates.

(E) Previous theories of the formation of separate lateral regenerates following the reversal of one transverse axis

There have been three major suggestions about the cause of lateral regeneration, and to be tenable they must be consistent with the number, position, orientation and composition of the laterals formed after reversal of one transverse axis and after similar operations.

(i) Lawrence (1970)

Regeneration of distal elements of the pattern may be inhibited at all levels by disto-proximal inhibition. This would obviously be relieved after amputation of distal structures, allowing the disinhibited area (the cut end) to regenerate. If information flow is not only polarized distoproximally but also can only occur between cells from similar positions on the circumference, reversal of a transverse axis will disinhibit two areas of the host cut surface, separated by two inhibited areas. This is shown in Fig. 9 A and would give two laterals of host origin, orientated like the host.

Figure 9

Theories of the origin of lateral regenerates after reversal of one transverse axis (the internal-external axis).

(a) Schematic representation with the tibia split posteriorly and opened out flat, with the graft/host junction shown by the dashed line. Hatching, area disinhibited (A and B) or induced (C) and able to regenerate, g, Graft; h, host. A, I, P, E, ‘faces’ of the tibia.

(b) Regenerated structures, anterior view. Stippling, host-derived lateral structures.

(A) Regeneration from two areas of the host where ‘tramlined’ disto-proximal inhibition cannot act.

(B) Regeneration from the host surface (isolated from normal disto-proximal contacts) and graft surface (isolated from normal proximo-distal contacts).

(C) Regeneration from the two regions of confrontation of opposite (I and E) faces of the tibia.

Figure 9

Theories of the origin of lateral regenerates after reversal of one transverse axis (the internal-external axis).

(a) Schematic representation with the tibia split posteriorly and opened out flat, with the graft/host junction shown by the dashed line. Hatching, area disinhibited (A and B) or induced (C) and able to regenerate, g, Graft; h, host. A, I, P, E, ‘faces’ of the tibia.

(b) Regenerated structures, anterior view. Stippling, host-derived lateral structures.

(A) Regeneration from two areas of the host where ‘tramlined’ disto-proximal inhibition cannot act.

(B) Regeneration from the host surface (isolated from normal disto-proximal contacts) and graft surface (isolated from normal proximo-distal contacts).

(C) Regeneration from the two regions of confrontation of opposite (I and E) faces of the tibia.

(ii) Bullière (1970 b)

Regeneration may be inhibited everywhere on the intact leg by the normal disto-proximal and proximo-distal short-range cellular interactions. Reversing one transverse axis effectively relieves the inhibition on the adjacent but non-communicating host and graft surfaces, so that they will behave independently and each will regenerate (Fig. 9B). The host- and graft-derived regenerates will grow past each other, retaining contact at a common region of the circumference and hence will project laterally from the original points of axis incongruity. Both will be orientated like the host (since the graft regenerates from a proximal-facing surface).

(iii) Bart (1971a, b)

Bart considers regeneration not to be disinhibited but actually to be induced by local confrontation of tissue from opposite faces of the leg. This confrontation occurs when the cut end of an amputated stump heals over before terminal regeneration, and it occurs at two positions following reversal of a transverse axis (Fig. 9C). A lateral of dual origin will be induced at each of the two regions of confrontation, and they will both be orientated like the host.

All three theories account satisfactorily for the number, position and orientation of the laterals, but only that of Bart accounts for the fact that they are usually of dual origin.

After reversal of one transverse axis, there are two areas of confrontation, two cut surfaces and ultimately two regenerates. One might hope to determine the relevant correlation by producing four areas of confrontation. The result of these experiments will now be given and discussed.

Reversal of both transverse axes

Both transverse axes can be reversed by 180° rotation of a homopleural graft, and the results are very variable. At tibial and coxal levels, Bullière (1970b) usually found no laterals or two laterals, and Bohn (1972a) usually found two laterals, often distally incomplete. Blatella usually produces no laterals or one incomplete lateral (French, 1976a). In all cockroaches there is a strong tendency for a rotated graft to rotate back into alignment with the host. Bart (1971 a) grafting at coxa level in the stick insect, found almost no tendency for grafts to rotate back. He obtained no laterals (1/63 cases), one lateral (15/63), two laterals (38/63) or three laterals (9/63).

Despite the difference in maximum number of laterals obtained, Bart’s results share many features with Bullière’s and Bohn’s. Laterals could form on any opposite or adjacent faces of the leg and, if two laterals were formed, one was a right and the other a left leg. The cases illustrated by Bullière (1970b, fig. 8) all conform to Bart’s illustration (1971 a, fig. 2) of the possible positions and orientations of the laterals.

The theory of Lawrence (Fig. 9 A) predicts one lateral, as the entire host cut surface would be disinhibited; that of Bullière (Fig. 9B) predicts one lateral from the host and one from the graft; only the theory of Bart (Fig. 9C) predicts more than two laterals (a maximum of four). It is difficult to assess the significance of the cases of three laterals in the stick insect. Bohn (1972b) suggests that two is the maximum possible number directly caused by the rotation, and that Bart’s triple laterals resulted from secondary wounding. However, Bart never found cases of three laterals developing from the very similar operations reversing only one transverse axis. Bohn also points to the failure to ever obtain four laterals, indicating that it argues against Bart’s hypothesis of the establishment of a ‘morphogenetic centre’ wherever opposite faces are confronted.

The differences in results between Blabera, Leucophaea, Blatella and Carausius could be the reflexion of relatively trivial factors (e.g. size, rate of wound healing, extent of back rotation), and are unlikely to be due to major differences of mechanism. It seems unwise to choose a model (Bullière’s) which restricts possible laterals to two when, even in only one insect, one seventh of operated legs bear three.

As Bart argues (1971a), it is likely that lateral regeneration occurs due to some positive interaction between opposite faces confronted at the junction, rather than just a lack of normal contacts. Grafts of sternite or tergite (Bart, 1966), coxo-pleural articulatory membrane, scape, or the same face of another segment (Bart, 1971b), or elimination of a part of a femur face (Bart, 1970), do not result in regeneration of the leg tissue thus deprived of its normal neighbours. Also, an interaction theory accounts more satisfactorily for most laterals being of dual origin. Such an interaction theory will now be developed and will be shown to account in detail for many of the features of lateral regeneration.

(F) Theory for the initiation of lateral regeneration from a non-congruent graft iho st junction

So far, in. describing the experiments and discussing the results, the leg has been considered to have two ‘transverse axes’ and four ‘faces’ (anterior, internal, posterior, external). Bart (1971 a, b) considers the four faces to be distinct and qualitatively different from each other: opposite qualities can interact in a unique way to induce a regenerate. After reversal of one ‘transverse axis’ of the graft, two host ‘faces’ will be in contact with the corresponding ‘faces’ of the graft, and the other two host ‘faces’ will each confront the opposite graft ‘face’: there will be interaction and induction of a regenerate at these two positions.

Bohn (1972a) suggests that ‘faces’ are nothing more than a ‘topographical characterization’, and that the cross-section of the limb could be arranged in a two-dimensional co-ordinate system ‘the axes of which coincide with the antero-posterior and dorso-ventral (external-internal) axes’. In this coordinate system every cell is different from any other. After reversal of one transverse axis, only at two points on the circumference will the normal contacts be made, and the laterals will develop from the two areas of maximum incongruity.

One might further suggest that the two ‘transverse axes’ are nothing more than a topographical characterization. There is no evidence that the anterior-posterior and internal-external planes are of any particular significance to the cockroach leg with respect to epidermal pattern or regeneration.

It is very likely that lateral regeneration is due solely to interactions occurring within the epidermis, since it can be provoked by grafts of epidermis not involving the translocation of any muscle or the sectioning of any major nerve (Bart, 1966, 1971a; Bohn, 1972 b; French & Bullière, 1975a, b). The leg epidermis, being a single celled layer, is effectively two dimensional; the surface of a cylinder upon which any position may be specified, not in relation to three axes, but by the two co-ordinates of proximal-distal level and circumferential position (Bullière, 1971; French & Bullière, 1975a, b)

Bohn (1972b) showed that circumferential position is not irreversibly determined since, in the course of lateral or terminal regeneration, a quarter of the circumference of the tibia often gave rise to more than the corresponding quarter of the tarsal circumference. French and Bullière (1975 a, b) showed that circumferential position (la génératrice) is an aspect of position independent of proximal-distal level and is specified in a continuous manner around the leg epidermis, rather than along two perpendicular axes going through the leg. By grafting a rectangular piece of cuticle plus epidermis to an abnormal position around the circumference, we confronted non-homologous positions along the lateral edges of the graft. In all such experiments (French, 1976 c) there was an intercalary regeneration of the structures which normally separate host and graft positions, as measured by the shortest route around the circumference (Fig. 10A, B). At the proximal and distal ends of the graft, different circumferential positions were placed adjacent to each other in a proximaldistal sense, and there was an identical intercalary regeneration of circumferential values to restore normal cellular contacts (Fig. 10C).

Figure 10

Transverse intercalary regeneration between different circumferential positions (adapted from French & Bullière, 1975 a).

(A) Schematic cross-section of the left femur, distal view. A, I, P, E, Anterior, internal, posterior and external ‘faces’. Twelve positions are marked around the circumference arbitrarily by numbers 0-12.

(B, C) Graft of internal face of the left femur into the anterior face of the left femur: graft situation (a) and result after two moults (b): transverse section through the graft (B) and anterior view (C). An internal face (r) is regenerated between positions 4 of the graft (g) and 8 of the host (h), and a part of the anterior face (r) is regenerated between positions 8 of the graft (g) and 9 of the host (A). At the ends of the graft, intercalary regenerates (r’) are formed between graft and host, and between regenerate (r) and host, to remove all discontinuities of position.

Figure 10

Transverse intercalary regeneration between different circumferential positions (adapted from French & Bullière, 1975 a).

(A) Schematic cross-section of the left femur, distal view. A, I, P, E, Anterior, internal, posterior and external ‘faces’. Twelve positions are marked around the circumference arbitrarily by numbers 0-12.

(B, C) Graft of internal face of the left femur into the anterior face of the left femur: graft situation (a) and result after two moults (b): transverse section through the graft (B) and anterior view (C). An internal face (r) is regenerated between positions 4 of the graft (g) and 8 of the host (h), and a part of the anterior face (r) is regenerated between positions 8 of the graft (g) and 9 of the host (A). At the ends of the graft, intercalary regenerates (r’) are formed between graft and host, and between regenerate (r) and host, to remove all discontinuities of position.

These experiments detected no ‘boundary positions’ having unique properties which could correspond to a boundary between ‘high’ and Tow’ values of a circumferential ‘gradient’ analogous to the postulated proximal-distal segment ‘gradient’ (Bohn, 1967). The circumferential position 12/0 in Fig. 10 does not imply a boundary; it arises inevitably when labelling a circle (e.g. a clock face) with numbers. Possible mechanisms for smoothly specifying position around a closed circle will be considered elsewhere (French, Bryant & Bryant, in preparation).

The position of an epidermal cell within a segment seems to be defined by its level within a linear proximal-distal ordering, and its position within a circular circumferential ordering. Intercalary regeneration occurs between different proximal-distal levels and between different circumferential positions.

It will now be shown that lateral regeneration is initiated at a non-congruent graft/host junction as a direct result of the spatial organization of the epidermis.

Reversal of one ‘transverse axis’

Consider the situation at the tibia graft/host junction following a graft between left and right legs. Fig. 11A shows the A/P axis reversal and the intercalary regeneration which will occur between the confronted graft and host positions, always forming the intervening positions, as measured by the shortest route. It will be seen that the shortest route between confronted graft and host positions is different on the two sides of each point of maximum incongruity (A and P), hence at these points the two halves of a complete circumference are formed, confronting each other. Fig. 11B shows the I/E axis reversal, forming the two halves of a circumference both internally and externally.

Figure 11

Initiation of lateral regeneration from non-congruent graft/host junctions.

(A) Reversal of anterior-posterior axis, giving laterals anteriorly and posteriorly.

(B) Reversal of internal-external axis, giving laterals internally and externally.

(C) Reversal of both axes by 180° rotation of a homopleural graft, giving no laterals (C1), or two laterals on adjacent faces (C2, C3) or on opposite faces (C4).

Schematic cross-section of graft/host junction; outer circle, host circumference; inner circle, graft circumference. A, I, P, E, Anterior, internal, posterior and external ‘faces’. Twelve positions are marked around the circumference by numbers 0–12 (as in Fig. 10) and numbers between the circles are the positions formed by intercalary regeneration (by the shortest route) between the different confronted positions of host and graft. Where the shortest route is different on the two sides of a point, the two halves of a complete circumference are formed, confronting each other; these are drawn spread apart to show the predicted orientation and composition (in A and B) of the regenerate forming at that point. Claws curve from external to internal positions on the laterally formed circumference, and stippling denotes host-derived tissue.

Figure 11

Initiation of lateral regeneration from non-congruent graft/host junctions.

(A) Reversal of anterior-posterior axis, giving laterals anteriorly and posteriorly.

(B) Reversal of internal-external axis, giving laterals internally and externally.

(C) Reversal of both axes by 180° rotation of a homopleural graft, giving no laterals (C1), or two laterals on adjacent faces (C2, C3) or on opposite faces (C4).

Schematic cross-section of graft/host junction; outer circle, host circumference; inner circle, graft circumference. A, I, P, E, Anterior, internal, posterior and external ‘faces’. Twelve positions are marked around the circumference by numbers 0–12 (as in Fig. 10) and numbers between the circles are the positions formed by intercalary regeneration (by the shortest route) between the different confronted positions of host and graft. Where the shortest route is different on the two sides of a point, the two halves of a complete circumference are formed, confronting each other; these are drawn spread apart to show the predicted orientation and composition (in A and B) of the regenerate forming at that point. Claws curve from external to internal positions on the laterally formed circumference, and stippling denotes host-derived tissue.

Consider the confrontation of the two half-circumferences formed at each point of maximum incongruity after the A/P axis (Fig. 11 A). The confrontations 10–8, 11–7, etc., will result in intercalary regeneration of the intervening positions, as measured by the shortest route. The shortest route is different on the two sides of the confrontation 12/0–6, hence at this point the two halves of a complete circumference will again be formed, confronting each other. Thus a round of intercalary regeneration produces new tissue (which will bulge laterally from the junction) but just recreates the confrontation of half circumferences.

After an amputation the epidermis from the circumference at the end of the stump migrates under the clot of dried haemolymph to re-establish epidermal continuity, resulting in the confrontation of cells from different circumferential positions. This situation can be simplified by considering epidermal migration to occur only in one plane, resulting in the confrontation of two half-circumferences which, exactly as in the lateral situation considered above, would be perpetuated by intercalary regeneration. The two half circumferences formed at each point of maximum incongruity after reversal of a transverse axis, and the circumference left after an amputation, are both cases where a complete circumference heals over because it is not sandwiched (as it is in the intact leg) between other circumferences overlying it proximally and distally. In these situations regeneration of distal structures occurs. Perhaps the tissue of a regeneration blastema is produced by successive rounds of intercalary regeneration between the different positions confronted by a circumference healing over at any level other than the distal tip of the tarsus. Clearly, this is only one component of the regeneration process since, within the blastema, the more distal levels of the leg must be specified. It is interesting that a bilaterally symmetrical, partial circumference, which could heal over and completely resolve confrontations between non-homologous positions by intercalary regeneration, often does not regenerate distal structures or forms distally incomplete partial regenerates (Bohn, 1965; French, 1976a).

Thus each of the two complete circumferences formed at the junction after an A/P or I/E ‘axis’ reversal will be the origin of a lateral regenerate. Both laterals will be orientated in the I/E axis like the host (I/E reversal) or like the host and graft (A/P reversal). Assuming that both components contribute to the intercalary regeneration which creates the complete circumference, the laterals will each be composed of host-derived and graft-derived longitudinal halves.

Reversal of both ‘axes’ by 180° rotation

After 180° rotation of a homopleural graft each host position around the graft/host junction will be confronted by the opposite graft position. A confrontation between opposite positions may be resolved by intercalary regeneration of either of the half-circumferences which separate them (French, 1976c). This can be visualized as going from the host position around a circumference in a clockwise or anticlockwise ‘direction’ to the opposite graft position. If intercalary regeneration occurs in the same ‘direction’ everywhere on the junction, no lateral regenerates will be formed (Fig. 11C,1). Any tendency of the graft to rotate back into alignment will increase the probability that no lateral will be formed by creating an unambiguous ‘shortest route’ in the same ‘direction’ at all positions on the junction. If, by chance, one sector of the junction forms the intercalary regenerate in one ‘direction’ while the remainder of the circumference of the junction regenerates in the opposite ‘direction’, a complete circumference will be created at each boundary between the two regions, and hence two laterals may be formed. This may occur anywhere on the circumference, forming two laterals situated opposite or adjacent to each other. One left-handed and one right-handed lateral will be formed. Consider the possible positions and orientations of these laterals. Considering only the I/E axis, a lateral will be orientated like the host if it is formed internally or externally (Fig. 11C2, C3), but either like the host or like the graft it formed anteriorly or posteriorly (Fig. 11C2, C3, C4). If one lateral is anterior and the other posterior, they will have opposite orientations (Fig. 11C4). These predictions, concerning handedness, position and orientation, which are not made by earlier models, are in good agreement with the data of Bullière, (1970b),Bart (1971 a) and Bohn (1972a).

More than two laterals may be formed if the ‘direction’ of intercalary regeneration in two sectors is opposed by that in the two sectors separating them. A complete circumference will be created at the four boundaries, making it possible for four laterals to form. Bart (1971a) found three laterals in 9/63 cases but no cases where four were formed. This is perhaps not surprising since analysis of Bart’s results for reversal of a single axis (Bart (1971a), table 4) indicates that a point of incongruity has a probability of only 0·7 of leading to the formation of a lateral. If cases of four changes of ‘direction’ around a single junction occur fairly infrequently (two changes being the usual situation) and if each circumference which forms at the site of such a change has only a 0·7 probability of leading to regeneration of a lateral, two laterals will be the most frequent result of the 180° rotation experiment and the incidence of three laterals will be much higher than that of the maximum, four.

It remains unclear why the laterals formed after 180° rotation were often distally incomplete (Bart, 1971 a; Bohn, 1972a). It was argued (French, 1976a) that a distally incomplete lateral could arise from a sector of a congruent junction if the graft and host did not initially heal together and interact. Incomplete laterals may arise from a rotated junction if there is a failure of healing at a position which is not a boundary between different ‘directions’ of intercalary regeneration.

Composition of the lateral regenerates

It has been assumed so far that position is specified smoothly and continuously around the leg, with no positions having unique properties and that all confrontations result in intercalary regeneration by the ‘shortest route’. This is consistent with the results of French & Bullière (1975a, b) and is the basis of the model for initiation of lateral regeneration. The model is consistent with the number, position and orientation of laterals formed after reversal of one transverse axis and also explains the main features of regeneration after 180° rotation of a homopleural graft.

It has further been assumed in Figs. 6, 8 and 11 that the intercalary regeneration results from proliferation of both of the confronted regions creating a complete circumference of dual origin, leading to a lateral regenerate composed of host-derived and graft-derived longitudinal halves. This assumes that there are no restrictions with respect to the circumferential position which the progeny of a cell can occupy. This assumption seems to be supported by the results of A/P axis reversal, where the laterals were usually of dual origin with borders between host- and graft-derived parts running approximately mid-external and mid-internal, as predicted (see Results, and Bohn, 1972a). After I/E reversal, however, the separate Blatella laterals were often largely host-derived or graft-derived, and Bohn’s interspecies grafts (1972 a) showed that dual origin laterals had one border on the posterior side (predicted) but the other border internal or external (not predicted). This suggests that an anterior or posterior region can produce tissue more internal and more external, but that an internal or external region may not always be able to produce tissue both anterior and posterior to it.

In this connexion it is intriguing that, after grafting an external sector of the tibia of Gromphadorhina to an internal position of the Leucophaea tibia, Bohn (1972b) found that the graft contributed more than one-quarter of the circumference of the lateral tarsi which were formed. The four illustrations (Bohn, 1972 b, figs. 1, 3b-e) all show the graft forming anterior and even internal portions of the tarsi, but not more posterior portions.

Clonal analysis has shown that the epidermal cells of the developing appendages of Drosophila are restricted in the positions which they can occupy. The imaginal wing disc becomes progressively subdivided into ‘compartments’ in the course of normal development (Garcia-Bellido, Ripoll & Morata, 1973; Garcia-Bellido, 1975). Clones initiated after a particular compartment boundary has formed do not cross it in normal development. Even if they are able to divide more rapidly than the rest of the tissue and nearly fill the compartment in which they arise, cells of the clone will not cross the compartment boundary. The disc seems to be divided into anterior and posterior compartments as soon as it is segregated from the rest of the blastoderm, but thereafter compartment boundaries form between dorsal and ventral, and wing and notum (during the 1st larval instar), between regions of the notum (during 2nd instar) and between proximal and distal wing (during 3rd instar). During regeneration from fragments of a disc, however, cells may cross the later compartment boundaries although they may not be able to cross the anterior/ posterior boundary (Bryant, 1975, and personal communication).

If the cockroach leg were divided into anterior and posterior compartments which were respected during regeneration, with the compartment boundaries approximately mid-internal and mid-external, this might account for the differences in composition between laterals resulting from A/P and I/E reversals. Experiments are in progress to explore this possibility.

(F) Comparison between lateral regeneration in insect and amphibian legs

When newt leg regeneration blastemas are grafted on to stumps in such a way as to reverse the A/P or dorsal/ventral axis (D/V) or both, supernumerary regenerates develop just as in cockroaches (Iten & Bryant, 1975; Bryant & Iten, 1976). Reversal of A/P or D/V axis gives two laterals originating from anterior and posterior, or from dorsal and ventral respectively, orientated like the stump. After reversal of both axes, one right and one left-handed regenerate are formed.

An interpretation of these newt laterals which is very similar to the present model of lateral regeneration in cockroach legs has been developed by Bryant & Iten (1976). It is suggested that position is continuously specified around the circumference of the limb, that confrontation of non-homologous positions results in intercalary regeneration by the shortest route, and that regeneration occurs from the complete circumference generated wherever the ‘direction’ of the shortest route changes over.

The similarities between insect and amphibian limbs will be developed elsewhere (French, Bryant & Bryant, in preparation), but the occurrence of similar lateral regeneration suggests that these anatomically quite different systems may be organized in fundamentally the same way.

This work was financed by an S.R.C. Postgraduate Studentship and a Royal Society European Exchange Research Fellowship. I thank Professor Sengel for the hospitality of his laboratory in Grenoble; and Professor MacGregor of the Department of Zoology, University of Leicester, for generously providing facilities for the preparation of this manuscript.

I am deeply indebted to Gerry Webster for preserving the sanity and guiding the efforts of his postgraduate student, and for advice on this manuscript.

Bart
,
A.
(
1966
).
Role de 1’hypoderme dans le déclenchement de la morphogenèse régénératrice chez Carausius morosus Br
.
C. r. hebd. Séanc. Acad. Sci
.,
Paris
262
,
1452
1454
.
Bart
,
A.
(
1970
).
Blessures et transformations trochantériennes chez Carausius morosus Br
.
Ann. Embr. Morph
.
3
,
379
398
.
Bart
,
A.
(
1971a
).
Morphogenèse surnuméraire au niveau de la patte du Phasme Carausius morosus Br
.
Wilhelm Roux Arch. EntwMech. Org
.
166
,
331
364
.
Bart
,
A.
(
1971b
).
Modalités de formation et de développement d’un centre morphogénétique chez Carausius morosus Br
.
Wilhelm Roux Arch. EntwMech. Org
.
168
,
97
124
.
Bodenstein
.
D.
(
1937
).
Beintransplantationen an Lepidopterenraupen. IV. Zur Analyse experimenten erzeugter Bein-Mehrfachbildungen
.
Wilhelm Roux Arch. EntwMech. Org
.
136
,
745
785
.
Bohn
,
H.
(
1965
).
Analyse der Regenerationsfahigkeit der Insektenextremitât durch Amputations und Transplantationsversuche an Larven der afrikanischen Schabe Leucophaea maderae Fabr. (Blattaria). II. Achensdetermination
.
Wilhelm Roux Arch. EntwMech. Org
.
156
,
449
503
.
Bohn
,
H.
(
1967
).
Transplantationsexperimente mit interkalarer Regeneration zum Nachweis eines sich segmentai wiederholenden Gradienten im Bein von Leucophaea (Blattaria)
.
Verh. dt. zool. Ges
. (1966). Suppl.
30
,
499
508
.
Bohn
,
H.
(
1970
)
Interkalare Regeneration und segmentale Gradienten bei den Extremitâten von Leucophaea-Lârven (Blattaria). I. Femur und Tibia
.
Wilhelm Roux Arch. EntwMech. Org
.
165
,
303
341
.
Bohn
,
H.
(
1972a
).
The origin of the epidermis in the supernumerary regenerates of triple legs in cockroaches (Blattaria)
.
J. Embryol. exp. Morph
.
28
,
185
208
.
Bohn
,
H.
(
1972b
).
The determination of the symmetry properties in the larval legs of cockroaches (Blattaria) Wilhelm Roux
.
Arch. EntwMech. Org
.
170
,
354
358
.
Bryant
,
P.
(
1974
).
Determination and pattern formation in the imaginai discs of Drosophila
.
Curr. Top. Devi Biol
.
8
,
41
80
.
Bryant
,
P.
(
1975
).
Pattern formation in the imaginai wing disc of Drosophila melanogaster fate map, regeneration and duplication
.
J. exp. Zool
.
193
,
49
78
.
Bryant
,
S.
&
Iten
,
L.
(
1976
).
Supernumerary limbs in Amphibians: experimental production in Notophthalmus viridescens and a new interpretation of their formation. (Submitted for publication. Devi Biol.)
Bullière
,
D.
(
1970a
).
Sur la déterminisme de la qualité régionale des régénérats d’appendices chez la Blatte, Blabera craniifer
.
J. Embryol. exp. Morph
.
23
,
323
335
.
Bullière
,
D.
(
1970b
).
Interprétation des régénérats multiples chez les Insectes
.
J. Embryol. exp. Morph
.
23
,
337
357
.
Bullière
,
D.
(
1971
).
Utilisation de la régénération intercalaire pour/l’étude de la détermination cellulaire au cours de la morphonenèse chez Blabera craniifer (Insecte Dictyoptere)
.
Devl Biol
.
25
,
672
709
.
French
,
V.
(
1976a
).
Leg regeneration in the cockroach Blatella germanica. I. Regeneration from a congruent tibial graft/host junction
.
Wilhelm Roux Arch. EntwMech. Org. (In the Press
.)
French
,
V.
(
1976b
).
Leg regeneration in the cockroach Blatella germanica. III. Regeneration from a graft/host junction between different segments. (In preparation.) To be submitted to
J. Embryol. exp. Morph
.
French
,
V.
(
1976c
).
Specification of the position of epidermal cells around the circumference of the cockroach leg. (In preparation.) To be submitted to
J. exp. Morph
.
French
,
V.
&
Bullière
,
D.
(
1975a
).
Nouvelles données sur la détermination de la position des cellules épidermiques sur un appendice de Blatte
.
C. r. hebd. Séanc. Acad. Sci
.,
Paris
280
,
53
56
.
French
,
V.
&
Bullière
,
D.
(
1975b
).
Étude de la détermination de la position des cellules épidermiques: ordonnancement des cellules autour d’un appendice de Blatte; démonstration du concept de génératrice
.
C. r. hebd. Séanc. Acad. Sci
.,
Paris
280
,
295
298
.
Furokawa
,
H.
(
1940
).
Transplantation experiments on appendages of Anisolabis marítima (Dermaptera). III. Transplantation of forceps
.
Jap. J. Zool
.
8
,
510
535
.
García-Bellido
,
A.
,
Ripoll
,
P.
&
Morata
,
G.
(
1973
).
Developmental compartmentalisation of the wing disc of Drosophila
.
Nature New Biol
.
245
,
251
253
.
García-Bellido
,
A.
(
1975
).
Genetic control of wing disc development
.
Cell Patterning, Ciba Fdn Symp
.
29
,
161
182
.
Iten
,
L.
&
Bryant
,
S.
(
1975
).
The interaction between the blastema and stump in the establishment of the anterior-posterior and proximo-distal organisation of the limb regenerate
.
Devi Biol
.
44
,
119
147
.
Lawrence
,
P.
(
1970
).
Polarity and patterns in post-embryonic development of insects
.
Adv. Insect Physiol
.
7
,
197
266
.
Lheureux
,
E.
(
1970
).
Étude de schizomélies expérimentales de l’appendice locomoteur chez l’araignée Tegenaria saeva. Bl
.
C. r. hebd. Séanc. Acad. Sci
.,
Paris
270
,
709
711
.
Lheureux
,
E.
(
1971
).
Récherches expérimentales sur la participation du greffon et du porte-greffe dans la morphogenèse d’appendices surnuméraires chez l’araignée Tegenaria saeva Bl. C
.
r. hebd. Acad. Sci
.,
Paris
272
,
3299
3301
.
O’farrel
,
A.
,
Stock
,
A.
,
Rae
,
C.
&
Morgan
,
J.
(
1960
).
Regeneration and development in the cockroach, Blatella germanica
.
Acta Soc. Entomol. Cechslov. (Cas. Csl. Spol. ent.)
57
,
317
324
.
Penzlin
,
H.
(
1965
).
Experimentelle Erzeugung von Dreifachbildungen bei einem heterometabolen Insekt
.
Naturwissenschaften
52
,
217
.
Ross
,
M.
&
Cochran
,
D.
(
1967
).
Genetic variability in the German cockroach. II. A description of new mutants and linkage tests
.
J. Hered
.
58
,
274
278
.
Shaw
,
V.
&
Bryant
,
P.
(
1975
).
Supernumerary regeneration in the large milkweed bug, Oncopeltus fasciatus
.
Devi Biol
.
45
,
221
230
.
Wolpert
,
L.
(
1969
).
Positional information and the pattern of cellular differentiation
.
J. theor. Biol
.
25
,
1
47
.
Wolpert
,
L.
(
1971
).
Positional information and pattern formation
.
Curr. Top. Devi Biol
.
6
,
183
224
.