One of the most frequently used methods in the concept of “seeing is believing” is in situ hybridization, which allows the visualization of the spatio-temporal expression pattern of a gene of interest in the cells of whole organisms or tissues (Herrmann and Neidhardt, 2006). This method involves the hybridization of a labeled antisense RNA probe with the corresponding endogenous mRNA in the sample, followed by a label-determined staining step. The importance of this method among developmental molecular biologists cannot be overstated.

Despite the emergence of high-throughput methods such as single-cell or bulk RNA sequencing, spatial transcriptomics, and proteomics, the results obtained by in situ hybridization remain a staple in many publications, providing crucial validating data. However, detecting mRNA by WISH can be challenging when mRNA expression levels are very low, the transcripts are localized in hard-to-reach or hard-to-view areas, or tissue samples are prone to background staining, which decreases the signal-to-noise ratio.

One of such case is WISH in regenerating tail of Xenopus frog tadpoles, which can regenerate an amputated tail in a week and a hind limb in a month (Beck C.W. 2012). This model along with other models of fins, tails and limbs regeneration in fishes and amphibians is widely used to study mechanisms of regeneration in amniotic species, which are distinguishing by their amazing ability to restore these body appendages (Murawala et al., 2012). The study of genes associated with regeneration in model anamniotic species is of great interest not only for basic science, but also for the development of approaches that could enhance regeneration in amniotes, especially in human. The point is that amniotes (reptiles, birds and mammals) cannot regenerate body appendages because such important stages of regeneration as the formation of the regenerative epithelia and blastema cannot be realized in them. In particular, this is attributed by the destruction of gene network and loss of many genes responsible for these processes occurred during amniotes evolution (Zaraisky et al., 2024).

Here we describe additional treatments for Xenopus laevis tadpoles' regenerating tail samples, which allow us to obtain clear, high-contrast images of target RNA-containing cells by WISH. The most challenging aspect of this method is minimizing background staining in the sample to achieve high-sensitivity detection of cells expressing the target gene. For wild-type X. laevis tadpole tails at different stages of regeneration, this presented a double challenge.

Firstly, melanosomes (pigment granules) actively migrate with cells to the amputation site and can therefore interfere with the BM Purple stain signal. Additionally, due to the numerous melanophores, visualization and photodetection of the staining signal are very difficult (El Mir et al., 2024). If melanophores are not the focus of the investigation, we suggest adding a bleaching step for the tadpoles, which has proven effective for decoloring both melanosomes and melanophores after fixation and before the pre-hybridization stages.

Secondly, tail fins are very loose tissues, and the main problem during in situ hybridization is strong background staining, especially when the target RNA is not highly expressed and requires long staining incubation. We have observed that tadpole samples fixed immediately after amputation (0 hpa, hours post amputation) exhibited the lowest background staining of fins. To address the background problem, we recommend making fin incisions in a fringe-like pattern at some distance from the area of interest in the regenerating tail. This tail fin notching procedure improved the washing out of all solutions, preventing BM Purple from getting trapped in the loose fin tissues and causing non-specific autocromogenic reactions. Even after 3–4 days of staining, no background staining was detected. The combination of these procedures before hybridizing the regenerating tail samples allowed us to obtain high-contrast images of gene expression patterns without background interference.

To test various WISH protocols, we chose the gene encoding Zn2+-dependent extracellular matrix metalloproteinase 9 (mmp9), whose expression pattern was firstly detected in X. laevis albino embryos and tadpoles during normal development and hindlimb bud and lens regeneration (Carinato et al., 2000). Recently mmp9 was identified as one of the genes specific to the population of reparative myeloid cells that plays a key role in tail regeneration in X. laevis tadpoles (Aztekin et al., 2019; Aztekin et al., 2020). This lineage of cells is essential for the early steps of regeneration, as it quickly replaces the inflammatory myeloid lineage and induces subsequent processes necessary for regeneration progression: apoptosis and tissue remodeling, which ultimately lead to the relocalization of regeneration-organizing cells responsible for progenitor proliferation (Aztekin et al., 2019; Aztekin et al., 2020). According to yet unpublished data, mmp9, along with mpepa1, junb, mmp1 and mmp8, are unique markers of regeneration-inducing cells (RICs), which are formed transiently from basal epidermal cells. These markers are critical for modifying the surrounding extracellular matrix to facilitate the migration of other cell types, such as regeneration-organizing cells that further promote regeneration. These markers were identified through bulk, single-cell, and spatial RNA-seq (Radek et al., 2023).

As a result, we demonstrate, for the first time to our knowledge, the detailed expression pattern of mmp9 during the early stages (0, 3, 6, and 24 h post-amputation) of tail regeneration in tadpoles at stage 40. Additionally, we show that the activity of mmp9 is closely linked to regeneration competency. Its expression pattern significantly differs in tails amputated during the refractory period (stages 45–47), when regeneration is temporarily blocked as part of the normal developmental program (Beck et al., 2003). The high-quality images of regenerating tails, stained for mmp9-expressing cells using our optimized WISH protocol, allowed us to observe the behavior of these cells during the early stages of regeneration and to clarify and supplement data obtained by high-throughput methods, such as bulk- and sc-RNAseq.

To optimize the in situ hybridization staining in the regenerating tails of X. laevis tadpoles, we tested several additional treatments, as shown in Figure 1. We used tadpoles at stage 40 with tails regenerating for 0 or 6 hpa to evaluate different protocol variants. Samples at 0 and 6 hpa (12–15 tadpoles) for each protocol variant were collected in at least three independent experiments.

Treating samples with proteinase K helps to remove nucleases and makes the tissue more permeable to reagents. Lengthening the incubation time with proteinase K solution for samples at later developmental stages can increase the sensitivity of in situ hybridization and reduce non-specific staining.

In the first variant, we extended the proteinase K incubation time for tadpole tail samples to 30 min, but the staining results were unimpressive, with mmp9+ cells overlapping with strong background staining.

In the second variant, we partially notched the edges of the fin to help wash out reagents causing background staining from the loose tissues. This approach allowed us to observe many more mmp9+ cells. To reduce overlapping of the BM Purple stain with melanosomes and melanophores, we added a post-staining photo-bleaching step, as recommended by Harland (1991). The result was improved imaging, but melanophores only faded to brown.

In the third variant, we moved the photo-bleaching step to the very beginning of the protocol, immediately after fixation in MEMPFA and the dehydration step. This resulted in perfectly albino tails, but some samples developed large bubbles in the tail fin area filled with non-specific BM Purple staining after in situ hybridization.

Finally, in the fourth variant, we combined the early photo-bleaching step with caudal fin cutting before sample hybridization, resulting in very clear images of the specific staining of mmp9+ cells. The step-by-step procedures for this optimized protocol are described below.

Our laboratory has historically used the in situ hybridization protocol described by Richard Harland (Harland et al., 2004) as it is optimized for X. laevis embryos. However, in regenerating tail samples, this protocol resulted in strong background staining of the tail fins. Below, we describe our additional sample treatments and modifications to the in situ protocol for regenerating tadpole tail samples that significantly improved the signal-to-noise ratio.

The first step in preparing for an in situ hybridization experiment is to clone the gene of interest and synthesize the corresponding antisense RNA labeled with dig-/fluorescein-/dnp-uridine residues and prepare a hybridization probe (Long and Rebagliati, 2002). Hybridization with labeled sense RNA, which does not bind to the RNA of interest and allows detection of the presence of basic background staining, was used as a control.

Although the expression activity of mmp9 during early Xenopus development, wounding, and hindlimb regeneration was demonstrated as early as 2000 (Carinato et al., 2000), the first data on its spatiotemporal expression pattern in the regenerating tadpole tail tip appeared relatively recently. However, these data are quite contradictory and not complete. For the first time, mmp9 expression during Xenopus tadpole tail regeneration was reported at 1, 2, and 3 dpa in reparative myeloid cells using scRNA-seq (Aztekin et al., 2019). As the authors of this study showed, after amputation at stage 40 (0 dpa), the number of mmp9 transcript counts decreased up to 1 dpa with a subsequent increase to 3 dpa. However, in the work of other authors, an increase in mmp9 expression over background level was detected using bulk HiSeq already at 0.5-1 hpa, followed by further rise in expression up to 6 hpa and a subsequent drop down to 3 dpa (Radek et al., 2023). These authors also validated mmp9 expression by in situ hybridization and spatial transcriptomics only at 1 dpa, and both methods showed rather blurred images of expression regions along the trunk and on the amputated tail tip.

To clarify the data on the temporal-spatial distribution of cells with mmp9 expression, we tested mmp9 gene activity in tissues of the regenerating tail of X. laevis tadpoles using our new in situ hybridization protocol optimized for these samples. Using the fourth variant of treatments (see Figure 1), we performed in situ hybridization on tadpole tails amputated at stage 40 (regeneration-competent) or at refractory stages 46–47 (regeneration-incompetent) and fixed at 0, 3, 6, and 24 hpa (Figure 2). Samples for each time point (3, 6, 24 hpa, 12–15 tadpoles) were collected in three independent experiments. Images in Figure 2 show one tadpole out of ≥75% of tadpoles from each group at a given time point that have the same expression pattern of mmp9 mRNA in cells of regenerating tails.

As shown in Figure 1, taken immediately after tail amputation, mmp9+ cells are predominantly scattered in the area of the dorsal and ventral tail fins. When comparing mmp9 expression patterns in the regenerating tail at 0, 3, 6, and 24 hpa, a sharp increase in mmp9+ cells can be observed in the distal region of the caudal fins, as well as in the notochord and neural tube below the amputation line (Figures 1, 2A). The staining of the notochord in the proximal region of the tail is background, as it appears during hybridization both without RNA-probe and with the sense-RNA probe (compare Figures 2B, H). Considering that in early development, mmp7, mmp9, and mmp18 are active in macrophages and are required for their migration (Tomlinson et al., 2008) new data on the mmp9 expression pattern during regeneration are likely the result of active migration of myeloid cells towards the injury. At the same time, we cannot exclude the possibility that, in addition to the migration of mmp9+ cells, there is also an induction of mmp9 expression in the cells of the tail stump. Furthermore, by 24 h post-amputation, the orderliness of migrating mmp9+ cells increases greatly, resulting in a concentration of these cells along the main vascular pathways. Notably, the temporal mmp9 expression profile revealed here using WISH is consistent with the mmp9 temporal profile obtained by other authors using bulk RNA-seq of the regenerating tail in the more rapidly developing species Xenopus tropicalis, in which the expression peak of this gene was observed as early as 15 hpa (Chang et al., 2017).

To determine in which tissues mmp9+ cells are localized, we made frontal and sagittal cryo-sections (20 µm thick) of tails stained by WISH. Figure 2 shows that at the level of the notochord, cells expressing mmp9 appear near its distal tip, and their number increases significantly at 24 hpa, filling the space between the tip of the notochord and the wound epithelium. This area is usually defined as a blastema (Figures 2A’, C’). At first glance, it appears that the mmp9+ cells in the fin region lie in the outer epithelial layer. However, a frontal section at the level of the ventral caudal fin shows that mmp9+ cells are distributed underneath it in the stroma or vessels of the fin (Figures 2C’, C’’). Additionally, the sagittal section at 24 hpa demonstrates stained cells concentrated above and below the notochord and spinal cord. According to the literature, these cells correspond to the dorsolateral anastomosing vessel and posterior cardinal vein (Levine et al., 2003; Senevirathne et al., 2020). It is also clear that the closer the myeloid cells are to the distal end, the more actively they spread to the area of muscle tissue and the spinal cord (Figure 2C’’’).

In vivo tracing of labeled myeloid lineages, including granulocytes and macrophages, during wounding and infection in Xenopus tadpoles has shown that myeloid cells near the injury site start migrating as early as 20 min post-injury. The myeloid cells roll along the vascular endothelium or extravasate from vessels, migrating toward the wound area (Paredes et al., 2015). Mmp9 (gelatinase B) is a secreted matrix metalloproteinase that degrades proteins of the extracellular matrix, particularly collagen IV, VII, and X (Fu et al., 2009). The distribution of type IV collagen was detected at stage 56 in the basement membrane within the skin, between muscle cells, and around the spinal cord and notochord lamella (Nakajima and Yaoita, 2003). The detected pattern of the reparative myeloid marker mmp9 during the first day after tail amputation can be explained by combining all these data. Myeloid cells, likely attracted by chemokines or other agents, move toward the damage (amputation line) along vessels and also between cells of tissues containing collagen IV, VII, or X, cutting a path through the ECM using mmp9 as a machete. Such active ECM disruption near the amputation line could induce the release of ECM-linked factors, alter the cellular connectome and signaling repertoire, stimulate cellular dedifferentiation for blastema formation, and provide spatial freedom for further proliferation.

The next question we addressed was whether and how the expression pattern of mmp9 changes if the tail is amputated during stages when it is incompetent for regeneration, i.e., during the so-called refractory period (stages 45–47). Indeed, WISH staining of mmp9 transcripts at 3, 6, and 24 hpa at refractory stages differed significantly from that at the regeneration-competent stage 40 (compare Figures 2A–F). As shown in Figures 2D–F, the number of mmp9+ cells in refractory tadpoles is catastrophically low beneath the amputation line as well as in the tail fin tissues. By 6 hpa, mmp9+ cells are concentrated in the tail tip, and by 24 hpa, the number of cells decreases. At 2 and 3 dpa, mmp9 expression drops to zero levels, similar to 0 dpa (data not shown). These results suggest an interconnection between myeloid cell activity and regeneration progression.

Comparing the obtained results with scRNA-seq data, some inconsistencies in expression levels can be seen. The mmp9 expression, as determined by the scRNA-seq transcriptomic atlas, is downregulated by 1 dpa (Aztekin et al., 2019), in contrast to the apparent upregulation detected using our optimized in situ hybridization protocol. The expression level of mmp9 in the refractory period according to scRNA-seq data is higher than at stage 40, while we demonstrate the opposite results. Undoubtedly, the large volume of new omics data is indispensable and very useful for guiding research. However, validating the data using proven old-fashioned methods is still a necessary step. The obtained high-quality imaging data of regenerating tails stained for mmp9-expressing cells by the optimized WISH protocol made it possible to observe the life of these cells at the early stages of regeneration and to clarify and supplement the data obtained by high-throughput methods. It is likely that the proposed protocol will be suitable for addressing signal-to-noise ratio problems in similar samples, such as the regenerating fin of Danio rerio, the tail of X. laevis during normal development or the regenerating hindlimb, regenerating tail, or limb of Ambystoma mexicanum. In addition, we believe it is worth testing whether our additional sample treatments are applicable to variations of in situ hybridization, such as two-color WISH or fluorescent multichannel WISH (Long and Rebagliati, 2002; Vize et al., 2009).