The existence of positional memory during appendage regenerative outgrowth comes largely from the work performed in amphibians. Blastema cells of an amphibian limb inherit a memory of their initial position and specify the proximal boundary of the regenerate. This boundary will prevent blastema cells to form structures proximal to their level of origin . The proximal-distal axis of a regenerating amphibian limb can be viewed as a series of such boundaries since the limb always regenerates the missing elements with the correct patterning and identity upon amputation at different levels . In the amphibian limb, positional information is easy to recognize since there is a clear proximal-distal succession of different bone elements.
In zebrafish, evidence for positional information comes from the fact that a caudal fin will regenerate a similarly shaped fin after being amputated at different proximal-distal levels . Moreover, a proximal amputation presents a greater growth rate, which correlates to the higher number of proliferating cells detected when compared to a distal amputation . However, how different proximal-distal amputation places will impact on the positional information of the caudal fin bony rays has not been studied probably because bone landmarks are scarce in the caudal fin. Since most bony rays bifurcate in a defined distal position, this is the only morphological feature that can be used as a reference of positional memory. Even though several reference points would be ideal for the study of positional information, the number of segments before bifurcation is an objective reference in the proximal-distal axis of the fin.
Using bony ray bifurcations as landmarks, we were able to show that, in contrast to what happens in amphibians, the amputation place influences the bony ray bifurcation position. Repeated amputations performed near the bifurcation will progressively induce a distal shift, changing the original position of the bifurcation and resetting its positional information. We have also observed that the distalization of the bifurcation is independent of the proximal-distal place of amputation before the second amputation. This is likely explained by the short distance between the amputation plane and the bifurcation observed before the first amputation even when the amputation is done near the base of the fin. This distance is increased after the first amputation, resulting in the elimination of the influence of the amputation in the bifurcation position when the amputation is done near the base of the fin. Our data is consistent with the previously reported increase in the number of segments formed before a bifurcation when an amputation is done 2–3 segments below the bifurcation . This means that a certain number of segments will need to form/differentiate before a bifurcation is signalled to form. Moreover, we show that the bifurcation position is modulated by neighbouring regenerating tissues. Thus, it is possible that upon amputation, the regenerating surrounding tissues, namely the blastemas and inter-rays will inhibit the signal(s) responsible to initiate the cell and molecular mechanisms of a bifurcation and consequently delay its formation.
Previous reports have shown that preceding the formation of a bony ray bifurcation during caudal fin regeneration, shh splits in two its single domain of expression in the basal layer of the epidermis . This indicates that Shh is a good candidate to signal the formation of a bifurcation [6, 9]. Therefore, we hypothesized that the place of amputation could modulate the dynamics of shh expression and therefore the proximal-distal level at which a bifurcation will form. However, we observed that the dynamics of shh expression do not change with different proximal-distal amputation places, being always expressed in two separate groups of cells in the basal layer of the epidermis. This new pattern of shh expression that we have uncovered was possible to determine due to a high cellular resolution analysis that we performed. Furthermore, in caudal fins without any bifurcations, after being submitted to several distal amputations, the expression of shh is always observed in two separate domains. Thus, our results suggest that Shh cannot be the instructing signal responsible for positioning the bony ray bifurcation in a regenerating caudal fin.
We propose however that Shh, may be important for the formation of bone at the right place, acting has an attractor of bone progenitors aligning them, directing bone growth and possibly controlling the width of the bony rays in the regenerating fin. This conclusion is based on our time-course analysis of Zns5-expressing cells in the context of a shh reporter line. This analysis revealed that soon after the detection of shh expression, osteoblasts in the bone growing tip start to align close to the basal layer of the epidermis next to shh-expressing cells. This interpretation is consistent with previous findings that proposed that Shh might play a role in the osteoblasts patterning during fin regeneration .
It has been previously demonstrated that Fgf targets show higher expression levels in proximal regenerates when compared to distal ones. This suggests the existence of an Fgf gradient in the regenerating fin, which indicates that Fgf signalling might be implicated in the regulation of positional memory during fin regeneration. . Moreover, Fgf signalling is required for the expression of the homeobox-containing gene, msxb which, accordingly to an earlier report, is differentially expressed along the proximal-distal axis of the fin . Thus, Fgf signalling would be a good candidate to modulate the position of the bony ray bifurcation.
In order to address a potential role of Fgf signalling in determining the bifurcation position, we made use of a heat-shock inducible transgenic to attenuate Fgf signalling in a time controlled manner. All the different protocols used to transiently attenuate Fgf signalling did not alter the position of the bony ray bifurcation when compared to the controls with unaffected Fgf signalling levels. This indicates that Fgf signalling is not likely to be the factor controlling the formation of a bony ray bifurcation in the zebrafish regenerating caudal fin.
Retinoic acid (RA) is an additional strong candidate to be involved in the regulation of positional information. Evidence for this comes from relevant work in the amphibian limb where a gradient of RA and of the cell surface protein CD59 was shown, with higher levels in more proximal blastemas when compared to the distal ones [17, 18]. In addition, treatment with RA stimulates regeneration of proximal structures in a concentration-dependent fashion [19, 20] by increasing the levels of CD59 .
In contrast, the role of RA in the positional memory of the regenerating zebrafish caudal fin remains poorly understood. It has been proposed that RA treatment distalizes the bifurcation point due to the fusion of fin rays [13, 14]. It is not clear though, whether this is caused by a proximalization of the regenerating tissue, by the downregulation of shh following RA treatment , which leads to defects in bone formation/patterning  or even toxicity . Therefore, the role of RA in the positional memory of the regenerating fin should be further investigated.
Positional memory is a complex process that is likely to involve local interactions between different cell types and domains and multiple signalling pathways. In fact, a crosstalk between blastema, distal epidermis and inter-ray tissue was demonstrated to be essential to signal the formation of a bifurcation in regenerating zebrafish fins . More recently, a mathematical model proposes that the regeneration of a fin with the correct shape and pattern requires the interplay of three morphogens . Future studies will be essential to uncover the signals that give positional information to the regenerating fin/intact fin tissue.