The role of RA in zebrafish pelvic fin development

Via overexpression of Cyp26a1 using a transgenic Hsp70l:Cyp26a1 zebrafish line (Mayer, 2020; this study) or via inhibition of Aldhs by DEAB treatment (Breu, 2017) an reduction of RA level could be achieved in zebrafish larvae over a period of 4 - 6 weeks. Both approaches led to similar results, which are the complete or partial inhibition of pelvic fin formation in case the treatment was started from fin developmental stages S<1, S1 or to a lesser extend S2. This indicates that RA signalling is particularly important during the early stages of pelvic fin development. As soon as the first signs of the pelvic fin bud appear (S2) its influence seems to shrink. Overall, a complete inhibition of pelvic fin formation (SM5, FM3) was observed in 24 % (4/17) of Hsp70l:Cyp26a1^+/- individuals treated from S<1 and 7 % (1/14) treated from S1 (Fig. 17). All other larvae developed at least minimal pelvic fin structures, considering both, the endoskeletal pelvic girdle and the exoskeletal fin part.

In contrast to that, Breu, 2017, achieved a complete reduction of all pelvic fin structures in comparably higher percentages of treated animals, precisely 50 % (7/14) of S<1 larvae, 44 % (14/32) of S1 and 6 % (2/31) of S2 larvae (Fig. 12) (Breu, 2017). This different distribution is probably due to the different approaches of the experiments. The heat-shock treatment does not affect every individual at the same intensity. This is also reflected in the various severity of the eye lens destruction observed in several Hsp70l:Cyp26a1^+/- larvae (Fig. S5).

Some fish show only minor impairments and some exhibiting a totally destroyed eye structure. This could be due to differing copy numbers of the transgene present in the larvae. The transgenic line was created using the I-SceI meganuclease system (Blum &

Begemann, 2012; Kikuchi et al., 2011). Similar to the Tol2 transposon system, this technique mediates the insertion of transgenes in the genome, however the number of insertion events and the length of the concatemers are variable (Rembold et al., 2006; Thermes et al., 2002). A larva with a higher number of transgene copies will therefore experience enhanced Cyp26a1 overexpression. Moreover, the heat-shocks in this study were performed in 100 ml Petri dishes in a 38.5 °C tempered incubator. Naturally, the water in the dish does not warm up completely evenly and therefore some individuals are affected more than others, depending on their occupied positions. This is why it is necessary to use a larger number of larvae in order to get a comprehensive picture.

It is also interesting, that a low percentage of Hsp70l:Cyp26a1^-/- larvae, lacking the heat-shock promoter transgene, exhibited severe to medium malformations (SM4 - SM2) of the endo- and exoskeletal parts of their pelvic fins. This indicates that there is a background effect on pelvic fin development potentially caused by the stress arising with the heat-shock treatment. During the procedure, the fish are regularly caught, undergo the heat-shock treatment and are put back in the cooler facility water afterwards. Control fish that were left completely untreated showed a significantly higher growth rate within the four weeks of treatment period (data not shown).

Nevertheless, the differences in pelvic fin appearance between Hsp70l:Cyp26a1^+/- and Hsp70l:Cyp26a1^-/- larvae are striking. The vast majority of Hsp70l:Cyp26a1^+/- fish treated from S<1 and S1 developed at least minimal pelvic structures or exhibited severe to medium pelvic reductions (categorized in SM4 - SM2). The most frequent reductions were observed for the posterior process and the radials (Fig. 16E,F). These structures serve as anchors of the lepidotrichs, which is why incorrectly attached fins were often associated with it (Fig. 16E,E'). The structure, which was mostly little affected, was the anterior process. Even in fish assigned SM4, a basic anterior process structure was present, albeit shortened (Fig. 16C). Those fish in which this skeletal element was nearly lost (Fig. 16D) were the exception. The anterior process is the structure that is formed first in pelvic girdle development (see Fig. 3) and only afterwards the posterior processes and the fin base appear. This indicates that the formation of the early skeletal elements of the pelvic girdle might follow a set program, less prone to tampering, and that the formation and growth of the late structures are more variable.

In previous studies on fish of the Konstanz wild type (KN WT) line, a cleavage of the fin base was found after DEAB treatment (Fig. 9G) (Welte, 2011). This phenotype has not been documented in the long-term DEAB experiments conducted by Breu, 2017. However, in this study, it was observed again following heat-shock treatment in few Hsp70l:Cyp26a1 +/-individuals (Fig. 16E, arrowhead). Since it appears in another genetic context than KN WT, this suggests that it might actually be due to reduced RA signalling. The further observation made by Welte, 2011, concerning the mirror image duplication of the posterior process (Fig. 9F) could neither be reproduced by Breu, 2017, nor in this study. It is believed that this phenotype is due to effects of inbreeding that arose over several generations in the KN WT

Furthermore it was observed that RA deficiency had the most impact on the length and internal organization of the pelvic girdle (Fig. 18E-G). In contrast to that, the total width generally was little affected (Fig. 18E). The reduced length might be attributed to an impaired chondrogenesis. Several studies described and reviewed the participation of RA in chondrogenesis and bone formation (Adams et al., 2007; Draut et al., 2019; Jiang et al., 1995; Laue et al., 2008; Wang et al., 2014). Studies on facial and axial bones demonstrated Cyp26b1 expression in chondrocytes, osteoblasts as well as their associated precursor cells.

Increased RA concentration, either in the Cyp26b1 mutant dolphin (dol) or after Cyp26b1 knockdown led to defects of the facial cartilage with missing and fused structures, particularly affecting the midline elements (Laue et al., 2008). This was attributed to a mediolateral RA gradient with lower RA concentrations towards the midline (Laue et al., 2008). An in vitro study on isolated rat hindlimb bud mesenchymal cells postulated a dose-dependent inhibition of Pitx1 expression upon exposure to excess RA. The modulation of Pitx1 signalling caused downregulation of Sox9 and Col2a1 and therefore impaired chondrogenesis (Wang et al., 2014). Another in vitro study, based on cultured mouse limb bud mesenchymal cells is in conformity with this. It focused on manipulation of RA signalling by downregulation of Rar genes, in particular Rarβ2, which had a enhancing effect on chondrogenesis (Jiang et al., 1995). From this it can be concluded that well regulated RA signalling is needed to ensure correct skeletal development. Thus in this study, disrupted RA signalling might indeed be responsible for the defective pelvic girdle structures. Therefore, revealing the effects of RA deficiency on the proliferation of chondrocytes of the developing pelvic girdle was a project worth investigating in. However, the establishment of an EdU assay for detection and quantification of cell proliferation failed for zebrafish larvae of an age of 3 - 4 wpf (Jean Eberlein, data not shown), leaving this question unanswered. It would be worthwhile to resume this project, possibly using a commercial EdU kit. Alternatively, a BrdU assay, which is based on a similar functional principle, could be established.

Regarding the exoskeletal part of the fin, it was generally developed in the majority of treated Hsp70l:Cyp26a1^+/- individuals (Fig. 17B). However, significant reductions of the fin (bud) length and the number of the fin rays suggest that growth was nevertheless impaired.

This effect was still measureable even in treatment group S2 (Fig. 18C,D). This is comprehensible considering that the expression of Aldh1a2 was detectable earliest in S3 and increased with growing pelvic fin bud (Fig. 14, Fig. 48). In later stages, WISH experiments and

Aldh1a2:eGFP reporter constructs indicated the participation of RA signalling in lepidotrich formation and growth (Fig. 49) (Breu, 2017; Mück, 2018). This need for RA signalling during later stages of pelvic fin outgrowth most likely explains the reduction of size and number of fin rays following Cyp26a1 overexpression.

In this context, it would additionally be interesting to see what effect Cyp26a1 overexpression has on the appearance of the pectoral fins. Their metamorphosis from the larval to the adult form takes place during approximately the same time span and likewise involves RA signalling as revealed in the course of a side project of this thesis (Fig. S29) (Mück, 2018).

Overall, the combined results from this study, and from Breu, 2017, and Welte, 2011, are consistent and indicate the participation of RA during pelvic fin induction and patterning of the skeletal elements of the pelvic girdle. Nonetheless, the question remains of whether the observed effects on pelvic fin development are actually specific or just a combined result of overall growth retardation, stress and other side effects caused by heat-shock treatment and Cyp26a1 overexpression or DEAB treatment in case of the previous studies (Breu, 2017;

Welte, 2011). The growth retardation caused by the long-term RA deficiency situations was clearly significant (Fig. 18B) (Breu, 2017; this study). Moreover, side effects like the destruction of the eye lenses resulted in lower food intake and impaired the general condition of the Hsp70l:Cyp26a1^+/- larvae. Besides, since irregularities in pelvic girdle formation also occur frequently in wild type fish (Fig. 3) (Draut, 2020; Marzi, 2015), not all significant variations concerning the distances within the pelvic girdle (Fig. 18F,G) can in fact be attributed to reduced RA signalling. A comparison of the phenotypes shown in Fig. 16 with the pelvic girdles in Fig. 3 dissected from WT fish suggests that some appearances represent a younger developmental stage rather than a malformation due to manipulated RA signalling.

In this context, it also has to be mentioned that no changes in the expression pattern of the investigated pelvic fin specific genes, Pitx1, Tbx4, Fgf10a, Fgf8a, could be detected in the course of the performed gene expression studies (Fig. S6 - S11) (Weber, 2020). This is in line with the observations of Breu, 2017. Following short-time DEAB treatments, no changes in the expression of Pitx1 or Tbx4 were detectable as well. Therefore, a direct or indirect regulation of key factors of pelvic fin development by RA was not proven. However, even the

expression level of Aldh1a2 was unchanged following Cyp26a1 overexpression (Fig. S10), possibly indicating that the experimental setup of the expression studies was not suitable.

In several studies, a feedback mechanism between RA production and degradation has been described, agreeing that excess RA leads to upregulation of Cyp26a1 in combination with simultaneous downregulation of Aldh1a2 expression (Begemann et al., 2001; Dobbs-McAuliffe et al., 2004; Emoto et al., 2005; Kudoh et al., 2002). The other way round, in zebrafish nls mutants, which are bearing a point mutation in the Aldh1a2 gene that leads to an inactive protein, an upregulation of Aldh1a2 gene expression was documented (Begemann et al., 2001). Thus, following Cyp26a1 overexpression an upregulation of Aldh1a2 would be conceivable. More detailed investigation using a higher number of zebrafish larvae might be useful here. An upregulation of Aldh1a2 may actually occur, but it is only noticeable after dissection of the pelvic fin bud followed by a more close analysis of the intensity of the WISH staining.

It remains therefore open for future investigations to clarify whether the observed defects in pelvic girdle skeleton are actually due to RA deficiency and to elucidate the exact mechanism how they are caused. Surely, one of the most elegant ways to approach this question is to accomplish a local disruption of RA signalling, affecting only the pelvic region and not the entire organism. For this purpose, extensive work was carried out in this study to establish the Gal4-UAS system (see chapter 2.3 and 3.3). In principle, a local heat-shock treatment of Hsp70l:Cyp26a1^+/- larvae would also be possible (Shoji & Sato-Maeda, 2008), however, in practice this is probably a major challenge for fish of an age of 3 - 4 wpf when pelvic fin development begins.