{"paper_id":"0231d813-0ae7-4a76-befa-974dbe245cc3","body_text":"1\nTissue composition shapes differential skeletal integration strategies during axolotl limb 1 \nregeneration 2 \nRita Aires1, 2, *, Sean D. Keeley1,2, Kerstin Brandt2, 3, Mário Carreira1,4,5, Doğ a Berş an Güneş 6, Yagiz 3 \nSavci2, Ulrike Anne Friedrich3, 7, 8, Andreas Dahl7, Can Aztekin9, Tatiana Sandoval-Guzmán1, 2, 3, * 4 \n 5 \n1 Department of Internal Medicine III, Center for Healthy Aging, University Hospital and Faculty of 6 \nMedicine Carl Gustav Carus, Dresden University of Technology (TUD), 01307 Dresden, Germany. 7 \n2 Center for Regenerative Therapies Dresden (CRTD), Center for Molecular and Cellular 8 \nBioengineering (CMCB), Dresden University of Technology (TUD), 01307 Dresden, Germany. 9 \n3 Paul Langerhans Institute Dresden of the Helmholtz Center Munich, University Hospital and 10 \nFaculty of Medicine Carl Gustav Carus, Dresden University of Technology (TUD), 01307 Dresden, 11 \nGermany. 12 \n4 Abel Salazar Institute of Biomedical Sciences, University of Porto, 4200-465 Porto, Portugal. 13 \n5 Faculty of Engineering, University of Porto, 4200-465 Porto, Portugal. 14 \n6 Graduate School of Science and Engineering, Yıldız Technical University, 34220 Esenler/Istanbul, 15 \nTürkiye. 16 \n7 DRESDEN-concept Genome Center (DcGC), Center for Molecular and Cellular Bioengineering 17 \n(CMCB) Technology Platform, TUD Dresden University of Technology, 01062 Dresden, Germany. 18 \n8 German Center for Diabetes Research (DZD e.V.), 85764 Neuherberg, Germany. 19 \n9 Friedrich Miescher Laboratory of the Max Planck Society, Tübingen 72076, Germany 20 \n 21 \n* corresponding author/requests for reprints: rita.aires@tu-dresden.de, tatiana.sandoval_guzman@tu-22 \ndresden.de 23 \n 24 \n Keywords: Axolotl, Regeneration, Skeleton, Tissue Integration, Osteoclasts, AEC.   25 \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 27, 2026. ; https://doi.org/10.64898/2026.02.26.708175doi: bioRxiv preprint \n\n 2\nAbstract: 26 \nLimb regeneration requires not only rebuilding the missing structures, but also integrating them with 27 \nthe stump tissues. Osteoclast-mediated tissue resorption is essential for skeletal integration during 28 \nregeneration. However, given the cellular and structural heterogeneity along the limb skeleton, it is 29 \nunknown if skeletal tissue composition impacts resorption and, if so, how it is regulated. 30 \nHere, we show that osteoclast-mediated skeletal resorption is primarily activated in amputations 31 \ndamaging calcified regions of the skeleton, but not in cartilaginous areas. Using a combination of 32 \nspatial transcriptomics and bulk RNA sequencing, we found that amputations in calcified regions 33 \ntrigger the sustained expression of RANKL and the chemokine Loc138491483/Ccl24-like . We also 34 \ndemonstrate that Loc138491483/Ccl24-like is sufficient to induce osteoclast presence in non-resorbing 35 \namputations. Finally, our data suggests that the transcriptomic profile of the apical ectodermal cap is 36 \nmodified according to the underlying tissue types injured by the amputation. 37 \nOverall, our work reveals that tissue composition at the amputation plane directs important 38 \nadaptations of the regenerative program to the damaged tissues, particularly regarding integration 39 \nstrategies. These context-dependent responses will ultimately contribute to the near-seamless tissue 40 \nintegration of the regenerating axolotl limb regardless of the amputation position.  41 \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 27, 2026. ; https://doi.org/10.64898/2026.02.26.708175doi: bioRxiv preprint \n\n 3\nIntroduction: 42 \nLimb regeneration in the axolotl ( Ambystoma mexicanum ) is an important model for the study of 43 \ncomplex structure re-formation and occurs in sequential, yet largely overlapping phases 1,2. After an 44 \namputation, the wound quickly heals through the establishment of a wound epithelium (WE). The 45 \nimmune response is then established by the migration of immune cells into the stump, where they 46 \nrelease important factors and help to clear pathogens and cell debris. Meanwhile, cells in the WE 47 \nproliferate, forming the multilayered apical epithelial cap (AEC) 3. Together, immune cells and the 48 \nAEC contribute to tissue histolysis by secreting proteolytic enzymes that extensively remodel the 49 \nstump tissues3,4. Finally, a blastema forms under the AEC, which will then proliferate and ultimately 50 \ndifferentiate to regrow the missing tissues of the limb.  51 \nAlthough these processes are well established, the mechanisms by which newly formed tissues 52 \nintegrate with mature tissue in the stump remain largely elusive. Remarkably, tissue integration occurs 53 \nirrespective of the amputation position in the axolotl limb 2,5,6, which means that regenerative 54 \nintegration processes are equally effective regardless of the specific tissue composition affected by the 55 \ninjury. This poses an interesting challenge especially for the regenerating skeleton, as a typical limb 56 \nskeletal element comprises a variety of cell types arranged in distinct conformations along its length, 57 \nresulting in positional differences in thickness, stiffness, and relative proportions 7. Likewise, cell and 58 \ntissue heterogeneity in the skeletal element can greatly differ in the span of a few micrometers, as is 59 \nevidenced by the diaphyseal (i.e., center) and flanking epiphyseal (i.e., proximal and distal) regions. 60 \nEpiphyses in the axolotl limb are mostly comprised of chondrocytes, which contribute to the growth 61 \nof the limb and remain cartilaginous throughout the life of the animal 7. In contrast, the diaphysis is 62 \nwhere the primary ossification center first develops, which comprises hypertrophic chondrocytes, 63 \nosteoblasts, and osteocytes, as well as periskeletal cells and a surrounding calcified extracellular 64 \nmatrix (ECM)7. Moreover, skeletal integration often requires the amalgamation of nascent cartilage 65 \nwith the calcified remains of the amputated skeletal element of the stump 6. Yet, despite these obvious 66 \ndifferences, it remains unclear whether or not the mechanisms of skeletal integration differ according 67 \nto the tissue composition at the amputation site. 68 \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 27, 2026. ; https://doi.org/10.64898/2026.02.26.708175doi: bioRxiv preprint \n\n 4\nRecently, we demonstrated that osteoclast-mediated tissue resorption is important for integrating the 69 \nregenerated radius and ulna with the previously existing skeletal elements 6. Osteoclasts are large, 70 \nmultinucleated cells that have an immune origin 8. In mammals, these cells differentiate in a stepwise 71 \nmanner from common myeloid progenitors (CMPs) 9,10, which also originate both macrophages and 72 \ndendritic cell progenitors 8. Osteoclasts resorb skeletal tissue by adhering to the bone surface and 73 \ndegrading the calcified matrix by secreting protons (H +) and proteolytic enzymes such as Cathepsin K 74 \n(Ctsk) and Matrix Metalloproteinases (MMPs) 9. During axolotl limb regeneration, osteoclast-75 \nmediated tissue resorption occurs in a short and distinct window of time 6, which contrasts with the 76 \nlong-lasting resorption observed in mammalian bone fractures 11. This suggests that unique underlying 77 \nregulatory mechanisms might be employed in regenerative integration.  78 \nIn this work, we investigated how osteoclast-mediated skeletal resorption is regulated to promote 79 \nskeletal integration during limb regeneration. We found that this process is dependent on the 80 \ncomposition of the injured tissue, with resorption being triggered specifically by amputations through 81 \nthe calcified diaphysis and primarily used for its regenerative integration. Using a combination of 82 \nspatial transcriptomics and bulk RNA-seq, we show that the RANK/RANKL system likely 83 \norchestrates osteoclast differentiation after diaphysis amputations as early as 3 dpa. Moreover, we also 84 \ndiscovered that the previously undiscovered chemokine Loc138491483/Ccl24-like has an important 85 \nrole in promoting osteoclast differentiation and/or recruitment. Finally, we observed that the 86 \namputation site could induce significant transcriptomic differences in the AEC. 87 \nAltogether, our work exploring two amputation planes affecting different tissue types shows that 88 \ntissue composition at the injury site induces adaptations of histolysis, immune response, and in the 89 \nAEC, which may synergistically promote skeletal integration. This further demonstrates that early 90 \nregenerative mechanisms are tailored to the types of tissues affected by the amputation. These 91 \nadaptations ultimately make possible the successful re-formation of all missing limb tissues and their 92 \nintegration with the mature structure regardless of the amputation position within the limb. 93 \n  94 \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 27, 2026. ; https://doi.org/10.64898/2026.02.26.708175doi: bioRxiv preprint \n\n 5\nResults: 95 \nOsteoclast-mediated skeletal resorption is specifically activated after diaphysis amputations 96 \nTo explore the mechanisms controlling tissue resorption, we chose two amputation planes that, 97 \nwhile impacting the same skeletal elements, affected regions with different cellular and ECM 98 \ncompositions.  99 \nFor this, we amputated through the calcified diaphysis and the cartilaginous epiphysis, and 100 \ncompared the extent of tissue resorption after these two amputations. To accurately assess skeletal 101 \ntissue resorption in these two amputation planes, we utilized Sox9:Sox9-T2a-mCherrynls ( Sox9-102 \nmCherry) transgenic animals stained with calcein. This allowed for the clear identification of both the 103 \ncartilaginous skeleton, which expressed mCherry in chondrocytes, as well as the calcified diaphysis, 104 \nwhich was marked by the binding of calcein to its mineralized ECM (Fig. 1A). Using this strategy, we 105 \nwere able to simultaneously assess 1) resorption of the remaining stump skeleton (the length of 106 \nmCherry signal from the elbow joint until the amputation plane); and 2) resorption specifically of the 107 \ncalcified tissue (the length of the calcein-positive region).  108 \nForelimbs of Sox9-mCherry/calcein animals were amputated and followed for 18 days. In diaphysis 109 \namputated limbs, resorption of the calcified region was first observed at 7 days post-amputation (dpa) 110 \nin the form of gaps in the calcein staining (Fig. 1B, white arrowheads). By 9 dpa, the mineralized 111 \nmatrix had been significantly resorbed, after which this process considerably slowed up until 18 dpa 112 \n(Fig. 1B, top row). In contrast, in epiphysis amputations (Fig. 1B, bottom row), both stump skeletal 113 \nelements and their calcified regions remained relatively unchanged. Quantification of the remaining 114 \nstump skeleton and respective calcified region length in diaphysis amputations revealed that, on 115 \naverage, approximately 20% of the total length of the skeletal elements and up to 40% of the calcified 116 \nregion of radii and ulnas were resorbed by 18 days (Fig. 1C, D). Furthermore, most of this resorption 117 \noccurred between 7 and 9 dpa, which agreed with previous reports6.  118 \nEpiphysis amputations, on the other hand, did not exhibit such extensive resorption (Fig. 1B, D). 119 \nHowever, some variations in the length of cartilage and calcified regions at 9 and 11 dpa could be 120 \nobserved (Fig. 1C, D). To further visualize the cellular and ECM structure of tissues after diaphysis 121 \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 27, 2026. ; https://doi.org/10.64898/2026.02.26.708175doi: bioRxiv preprint \n\n 6\nand epiphysis amputations, we analyzed the histology of limb sections after these two amputations at 122 \n7 and 9 dpa. This showed that, in amputated epiphyses, the cartilaginous matrix of the radius and ulna 123 \nhad undergone considerable remodeling, and that a few of the most distal chondrocytes had even 124 \ndisappeared entirely (Fig. 1E). This explained the observed variations in the length of whole skeletal 125 \nelements, and matched our previous observations of cartilage undergoing histolysis12.  126 \nGiven their roles as the main cellular effectors of skeletal tissue resorption, we analyzed osteoclast 127 \nprevalence in the two amputations. As it was expected, diaphysis amputations showed robust 128 \nosteoclast presence surrounding the calcified region of both skeletal elements at 7 and 9 dpa. In 129 \ncontrast, only a few osteoclasts were observed asso ciated to the skeletal elements in epiphysis 130 \namputations at these time points (Fig. 1E, black arrowheads, Fig. E insets). Next, we decided to 131 \nfollow osteoclasts in vivo over time after diaphysis and epiphysis amputations. For that, we combined 132 \nthe reporter line Sox9-mCherry with Ctsk-eGFP transgenic animals, in which eGFP is driven by the 133 \npromoter of the  mature osteoclast marker Ctsk 6, to generate Sox9::Ctsk animals. In vivo imaging 134 \nconfirmed the extensive presence of osteoclasts at 7 and 9 dpa in diaphysis amputations, which 135 \nquickly decreased at 11 dpa, and was cleared by 15 dpa (Fig. 2A, top row). In contrast, minimal to no 136 \nosteoclast recruitment and/or differentiation was initiated in epiphysis amputations (Fig. 2A, bottom 137 \nrow).  138 \nFinally, we assessed the endogenous expression of the osteoclast marker Ctsk by Hybridization 139 \nChain Reaction (HCR) to determine when osteoclast presence is first differentially established in 140 \ndiaphysis and epiphysis amputations. While examining the Ctsk nucleotide sequence, we found that 141 \nanother gene, Loc138578972, was present in an adjacent region and annotated as Ctsk-like (Fig. S1A). 142 \nThe predicted nucleotide coding sequence of this gene was 73.6% identical to the one of Ctsk (Fig. 143 \nS1B). Moreover, the predicted protein sequence of Loc138578972 shared 73.4% and 73.7% sequence 144 \nidentity with the axolotl Ctsk (Fig. S1C, Fig. S1F) and human CTSK  peptide, respectively (Fig. S1E, 145 \nS1F). We thus concluded that the axolotl genome contains at least one additional Ctsk-related gene 146 \nand, in keeping with the most current axolotl annotation (UKY_AmexF1_1, GCF_040938575.1), the 147 \ngene annotated as Ctsk will continue to be referred to as “Ctsk”, whereas the gene Loc138578972 will 148 \nbe referred to as “Ctsk-like” for the remainder of this work. 149 \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 27, 2026. ; https://doi.org/10.64898/2026.02.26.708175doi: bioRxiv preprint \n\n 7\nAnalysis of both Ctsk and Ctsk-like expression showed that Ctsk-expressing cells were already 150 \npresent at 3 dpa in diaphysis and epiphysis amputations, both in close proximity to the skeletal 151 \nelements and in the mesenchyme (Fig. 2B). However, large multinucleated Ctsk-positive cells, which 152 \nare the hallmarks of mature osteoclasts, were almost exclusively associated to diaphysis amputations 153 \nat 3 dpa, and were seen in even larger numbers at 5 dpa (Fig. 2B, white arrowheads, insets). Ctsk-like 154 \nexhibited a similar expression pattern and was largely co-expressed with Ctsk at both time points.  155 \nAltogether, our results show that extensive osteoclast-mediated skeletal tissue resorption is a tissue-156 \ndependent event, being activated by amputations directly affecting, or in close proximity to, the 157 \ncalcified diaphyseal regions of the skeletal elements. Moreover, osteoclast recruitment and/or 158 \ndifferentiation starts early within regeneration, with Ctsk/Ctsk-like -positive multinucleated osteoclasts 159 \nappearing as early as 3 dpa at the injured skeletal elements in diaphysis amputations.  160 \nSystemic and local calcium do not significantly impact osteoclast presence in regenerating limbs 161 \nAs one major difference between diaphysis and epiphysis amputations is the damage to the sheath of 162 \ncalcified tissue surrounding the skeletal elements7, we explored the role of systemic and local calcium 163 \nin the activation of osteoclast-mediated skeletal resorption.  164 \nAt the systemic level, blood plasma measurements in intact animals were consistent with previously 165 \nreported values13. While we observed some variation in calcium concentrations between intact and 166 \namputated individuals especially in early time points, no significant difference between conditions 167 \nwas observed (Fig. 3A).  168 \nWe next investigated if, instead, local changes in calcium levels could have a role in osteoclast 169 \nrecruitment. We thus injected amputated limbs of Sox9::Ctsk animals with either BAPTA or CaSO4 to 170 \ndecrease or increase extracellular calcium levels in the tissue, respectively, and followed osteoclast 171 \ndynamics after diaphysis and epiphysis amputations (Fig. 3B). Injections of BAPTA 10mM (Fig. 3C-172 \nE) or CaSO4 15mM (Fig. 3F-H) had no effect on osteoclast presence in either amputation plane.  173 \nHence, these results indicate that systemic calcium concentrations do not significantly change due to 174 \nregeneration, nor due to different amputations affecting diaphyseal or epiphyseal regions. 175 \nAdditionally, in our experiments, changes in local calcium levels were not sufficient to trigger 176 \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 27, 2026. ; https://doi.org/10.64898/2026.02.26.708175doi: bioRxiv preprint \n\n 8\nosteoclast recruitment and/or differentiation into the damaged skeletal tissue, particularly in epiphysis 177 \namputations. Overall, this suggests that neither systemic nor local changes in extracellular calcium 178 \nconcentrations during regeneration are sufficient to induce significant osteoclast 179 \nrecruitment/differentiation. However, we cannot rule out that other factors work together with either 180 \nsystemic or local extracellular calcium levels to recruit or differentiate osteoclasts.  181 \nSpatial transcriptomics reveals differential gene expression profiles in diaphysis and epiphysis 182 \namputations 183 \nOur data so far suggested that skeletal tissue could be involved in osteoclast recruitment and/or 184 \ndifferentiation observed in diaphysis-amputated limbs. Thus, we asked if gene expression differences 185 \ncould inform us on regulatory factors underlying osteoclast-mediated differential resorption. To 186 \naddress this, we used spatial transcriptomics in diaphysis- and epiphysis-amputated limbs at 3- and 5 187 \ndpa (Fig. 4A) – the two time points in which differential osteoclast presence first becomes evident – to 188 \ninvestigate differences in gene expression between these two conditions. 189 \nClustering analysis of spatial expression dots from all samples combined revealed 19 clusters 190 \nrepresenting all major tissue types contained in our tissue sections, including epidermis (clusters 5 and 191 \n7), muscle (clusters 4, 9, and 13), cartilage (clusters 8, and 15), periskeleton/bone (cluster 18), and 192 \nnerves (cluster 17). We also detected regeneration-specific clusters, particularly a blastema cluster 193 \n(cluster 1) enriched in the expression of Kazald214,15, two clusters associated with tissue histolysis 194 \n(clusters 12 and 14), and one cluster representing the AEC (cluster 16) (Fig. 4A-C; Fig. S2A-B). 195 \nAnalysis of the expression of the resorption-associated factors Nfatc1, Ctsk, Ctsk-like, and Acp5 found 196 \nthat these genes were highly expressed in cluster 2 (Fig. 4D, Fig. S2B). Mapping these spots back to 197 \nthe tissue sections revealed that these genes were especially represented in diaphysis-amputated 198 \nskeletal elements at both 3 and 5 dpa (Fig. 4F). This, together with the fact that approximately 70% of 199 \nthis cluster was derived from diaphysis-amputated limbs, caused us to annotate this cluster as a 200 \nresorption cluster (Fig. 4E inset, Fig. S2B-C). 201 \nAs expected from having a spatial dataset with supra-cellular resolution combined with the high 202 \nmobility of immune cells, we could not isolate a specific immune spatial cluster. Instead, we found 203 \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 27, 2026. ; https://doi.org/10.64898/2026.02.26.708175doi: bioRxiv preprint \n\n 9\nimmune markers spread across multiple clusters, including Perforin1-like (Prf1-like), Proteoglycan 3 204 \n(Prg3), and C-X-C motif chemokine ligand 12 (Cxcl12) in the mesenchyme (cluster 6), and two genes 205 \nannotated as Macrophage expressed 1-like (Mpeg1-like) in the epidermis (cluster 7) and AEC (cluster 206 \n16). Notably, there was an enrichment of genes related to macrophage/monocyte function like 207 \nMacrophage receptor with collagenous structure  (Marco), a third Mpeg1-like gene, Cd68, and Cd14 208 \nin the resorption cluster (cluster 2). This points at the possibility that, similarly to mammals 16,17, these 209 \nmyeloid cells could act as possible sources of osteoclast progenitors during regeneration. 210 \nTaken together, we show that spatial transcriptomics is able to identify clear differences in gene 211 \nexpression between diaphysis and epiphysis amputations, and that a specific resorption cluster 212 \nenriched in myeloid markers is predominately present in diaphysis-amputated limbs.  213 \nThe RANK/RANKL system likely orchestrates the activation of osteoclast-mediated skeletal resorption  214 \nIn examined vertebrates, osteoclast progenitors differentiate into mature osteoclasts under the 215 \ninfluence of RANKL ( Tnfsf11) and RANK ( Tnfrsf11a)18–20. In our spatial dataset, we found that the 216 \nmajority of spatial dots in the resorption cluster (cluster 2) did indeed express RANKL and RANK (Fig. 217 \n5A, Fig. S3A), with RANK slightly upregulated in diaphysis amputations at both time points. In 218 \ncontrast, differences in RANKL expression were only observed at 5 dpa, being expressed in more 219 \nspatial dots and with overall higher expression levels in diaphysis amputations (Fig. 5B).  220 \nTo validate our spatial data and identify the cells expressing RANK and RANKL, we performed HCR 221 \nfor these two genes in diaphysis and epiphysis amputations at 3, 5, and 7 dpa (Fig. 5C). In epiphysis-222 \namputated limbs, RANKL was observed primarily in the cartilage cells closest to the AEC at 3 dpa. Its 223 \nexpression was decreased at 5 dpa and, by 7 dpa, only very low levels of RANKL were detected in 224 \nthese limbs (Fig. 5C, bottom row). This contrasted with diaphysis-amputated limbs, in which RANKL 225 \nwas robustly expressed, especially in periskeletal cells and hypertrophic chondrocytes at all analyzed 226 \ntime points (Fig. 5C, top row). Furthermore, HCR staining for RANKL and Ctsk showed that Ctsk + 227 \ncells were found in close proximity with cells expressing RANKL. 228 \nOn the other hand, RANK expression was more prevalent in diaphysis-amputated limbs at 3 and 5 229 \ndpa and was frequently co-expressed with Ctsk (Fig. S3B), suggesting that these cells were 230 \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 27, 2026. ; https://doi.org/10.64898/2026.02.26.708175doi: bioRxiv preprint \n\n 10\ndifferentiating osteoclasts. We further validated this by assessing the expression of Nfatc1 , the master 231 \ntranscriptional factor of osteoclastogenesis 21 that promotes the expression of Ctsk and components of 232 \nthe vacuolar V-A TPase 22. Indeed, we found that RANK-expressing cells co-expressed Nfatc1, 233 \nespecially in diaphysis-amputated limbs (Fig. S3C). Interestingly, in these limbs, we also observed 234 \ncells that co-expressed RANK and Nfatc1 , but not Ctsk, suggesting that these were osteoclast 235 \nprogenitors undergoing differentiation (Fig. S3C, white arrowheads). 236 \nOverall, our data indicates that RANKL expression is upregulated and sustained in periskeletal cells 237 \nand hypertrophic chondrocytes upon calcified diaphysis amputations, which may drive the 238 \ndifferentiation of RANK+/Nfatc1+ precursors into mature osteoclasts. In contrast, in epiphysis 239 \namputations, RANKL is only expressed in distal chondrocytes, and its levels quickly decrease. This 240 \nsuggests that, similar to mammalian osteoclastogenesis, the RANK/RANKL system has a key role in 241 \nosteoclast differentiation during axolotl regeneration, and that the differential activation of RANKL 242 \nexpression after injury to the calcified diaphyseal region likely triggers tissue-dependent skeletal 243 \nresorption. 244 \nLoc138491483/Ccl24-like is sufficient to induce osteoclast presence in non-resorptive amputations  245 \nAs osteoclast-mediated tissue resorption occurs in a relatively short and well-defined time window 246 \nduring limb regeneration, we next searched for regeneration-specific factors that could regulate this 247 \nprocess.  248 \nFor this, we reasoned that osteoclast progenitors and/or immature osteoclasts, identified by Nfatc1 249 \nexpression, would be in close proximity to any putative signal factor at 3 and 5 dpa. Our analysis 250 \nrevealed that the top 15 differentially expressed genes in Nfatc1-enriched spatial spots (Log 2 251 \nExpression > 2) were enriched in genes heavily involved in osteoclast function, such as V-ATPase 252 \nsubunit genes ( Atp6v0c, Atp6v0d2, Atp6v1b2) and MMPs/ECM remodeling genes ( Ctsk, Ctsk-like, 253 \nMmp9) (Fig. 6A). However, there was one gene, Loc138491483 (hereafter referred to as Loc483), that 254 \nwas seemingly not directly related to osteoclast function while still being associated with the 255 \nresorption cluster (cluster 2) (Fig. 6B, C) and to the skeletal tissue in diaphysis amputations (Fig. 6D). 256 \nWe then validated these findings by HCR at an earlier time point (1 dpa) and at 3 and 5 dpa. While no 257 \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 27, 2026. ; https://doi.org/10.64898/2026.02.26.708175doi: bioRxiv preprint \n\n 11\ndifferences were observed in the expression of this gene at 1 dpa, Loc483 became highly expressed in 258 \ndiaphysis-amputated limbs both at 3 and 5 dpa when compared to epiphysis-amputated limbs (Fig. 259 \n6E).  260 \nInterestingly, Loc483 is annotated in the axolotl genome as C-C motif chemokine 24-like . A 261 \nphylogenetic search via the webserver aLeaves 23 revealed that its most closely related protein 262 \nsequence matches were Monocyte Chemoattractant Protein-1/C-C motif chemokine 2 (Mcp1/Ccl2) in 263 \nthe caecilian Rhinatrema bivittatum  and C-C motif chemokine 4 (Ccl4) in Xenopus tropicalis . 264 \nAdditional matches could only be found in two species of sharks and in the paddlefish (Fig. S4), 265 \nsuggesting that, despite its annotation, this gene is probably not an ortholog to the human CCL24 266 \ngene, and that amphibians are, to date, the only tetrapods identified to possess Loc483.  267 \nNevertheless, being a chemokine made Loc483 a promising candidate as a chemoattractant for 268 \nmacrophages and monocytes, both previously reported as sources of osteoclast progenitors 16,17,24. 269 \nThus, to test if Loc483 would have the potential to promote osteoclast recruitment and/or 270 \ndifferentiation, we overexpressed this gene in Sox9::Ctsk animals. For that, we co-electroporated 271 \nblastemas with a construct containing the coding sequence of Loc483 under the control of the 272 \nubiquitous promoter CAGGS (CAGGS-Loc483), together with a reporter plasmid containing mCherry 273 \ndriven by the same promoter ( GAGGS-mCherry). The contralateral blastema was electroporated only 274 \nwith the GAGGS-mCherry construct as a control. We found that, in diaphysis amputations, 275 \noverexpression of Loc483 had no effect on the presence of osteoclasts (Fig. 6F, G). However, Loc483 276 \nwas able to ectopically induce the presence of osteoclasts in epiphysis-amputated limbs (Fig. 6F, H).  277 \nThus, these results show that Loc483 is likely a chemokine within the amphibian immune system, 278 \nand that its expression is sufficient to recruit and/or differentiate osteoclasts to the amputation site.  279 \nDiaphysis and epiphysis amputations differentially impact other processes to related regeneration 280 \nTo complement the spatial information of our datasets and get an overview of gene expression in 281 \ndiaphysis- and epiphysis-amputated limbs, we performed bulk RNA sequencing (RNA-seq) using the 282 \ndistal-most region of regenerating limbs. This found 100 and 80 differentially expressed genes 283 \n(DEGs) between the two amputation planes at 3 and 5 dpa, respectively. Unexpectedly, genes related 284 \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 27, 2026. ; https://doi.org/10.64898/2026.02.26.708175doi: bioRxiv preprint \n\n 12\nto bone resorption and osteoclast function such as Ctsk, Ctsk-like, Nfatc1 , and Acp5, although 285 \nupregulated after amputations, were not differentially expressed between diaphysis- and epiphysis-286 \namputated limbs in our bulk RNA-seq dataset (Fig. S5A-E). This was especially surprising for the 287 \nfirst three of these genes, given that our spatial transcriptomics and subsequent validation by HCR 288 \nstaining showed them as being upregulated mainly in diaphysis amputations as early as 3 dpa (Fig. 289 \n2B, Fig. S3B, C). The absence of differential expression in bulk RNA-seq could either be due to the 290 \ntechnical limitations of this technique, in which cellular resolution is lost and subtle changes in gene 291 \nexpression cannot be easily detected if genes are only present in a limited number of cells, or due to 292 \nthe expression of these genes being bolstered by their presence in processes unrelated to 293 \nosteoclastogenesis18,25–28.  294 \nMany of the detected DEGs, instead, were either uncharacterized genes (i.e., protein-coding genes 295 \nwith no further annotation) or predicted to be non-coding RNAs (ncRNAs). Indeed, a look into the top 296 \n15 to 20 upregulated DEGs in diaphysis and epiphysis amputations at the two time points showed that 297 \n30% to 60% of these genes were computationally annotated as ncRNAs (Fig. S5A-D). Consequently, 298 \nwe filtered out both ncRNAs and uncharacterized genes for all further downstream analyses. This 299 \nresulted in 60 and 30 upregulated protein-coding DEGs in epiphysis amputations at 3 and 5 dpa, 300 \nrespectively (Fig. S5A-B), while diaphysis amputations exhibited only 10 and 8 upregulated protein-301 \ncoding DEGs at these time points, respectively (Fig. S5C-D).  302 \nGene ontology (GO) analysis for biological processes with these genes showed that upregulated 303 \nDEGs in 3 dpa epiphysis amputations were enriched in terms associated with response against viruses 304 \nand with protein folding (Fig. S5F). In contrast, upregulated DEGs in 3 dpa diaphysis amputations 305 \nwere mostly associated with muscle tissue, as we saw Myl4, Loc138582818/Myh4-like, and 306 \nLoc138573356/Tpm1-like representing 3 out of the 10 upregulated DEGs. Importantly, Loc483 was 307 \nalso significantly upregulated in diaphysis amputations at 3 dpa (Fig. S5C, G), which agreed with our 308 \nspatial transcriptomics datasets, as well as our previous expression and functional validations.  309 \nAs for at 5 dpa, DEGs identified in epiphysis-amputated limbs were significantly enriched for GO 310 \nterms associated with epidermis and with ECM production, whereas only Cell adhesion  311 \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 27, 2026. ; https://doi.org/10.64898/2026.02.26.708175doi: bioRxiv preprint \n\n 13\n(GO:0007155) was found to be significantly associated to diaphysis-amputated limbs at the same time 312 \npoint, with upregulation of the genes Postn, Msln, and Thbs4 (Fig. S5F).  313 \nThus, these results suggest that the amputation plane, and specifically the types of tissues damaged 314 \nwithin it, might also affect other regenerative processes within the first stages of regeneration, 315 \nparticularly ones related to muscle tissue, protein folding, and epidermis development, as well as 316 \naspects of the immune response. 317 \nThe transcriptomic profile of the AEC is impacted by tissue composition at the amputation plane  318 \nWhen we performed GO analysis for cellular component (CC) and molecular function (MF) instead, 319 \nwe saw that the DEGs found in diaphysis-amputated limbs at 3 and 5 dpa were also significantly 320 \nenriched in ECM and muscle-related CC terms (Fig. 7A). Moreover, MF terms enriched for these 321 \namputations at 3 dpa were generally in agreement with the CC analysis, as they were similarly 322 \nassociated to muscle function, although no MF term  was significantly enriched at 5 dpa. In contrast, 323 \nDEGs in 3 and 5 dpa epiphysis-amputated limbs were enriched for CC terms associated to secretion 324 \nor extracellular space, as well as to epidermis. Meanwhile, enriched MF terms in this amputation 325 \nplane at 3 dpa were associated to protein folding, acetylcholine receptor signaling, and exosome 326 \nfunction (Fig. 7A). Interestingly, by 5 dpa, these MF terms changed to being enriched in protease 327 \ninhibitor-related ones, such as Serine protease inhibitor  (KW-0722) and Serine-type endopeptidase 328 \ninhibitor activity (GO:0004867).  329 \nGiven the prevalence of GO terms related to epidermis and secretion enriched in upregulated DEGs 330 \nin the bulk RNA-seq from epiphysis-amputated limbs, we hypothesized that these genes were mostly 331 \nbeing expressed in the AEC. To explore this possibility, we leveraged our spatial transcriptomics 332 \ndataset to locate the expression of these DEGs within the context of the tissue. For that, we first took 333 \nall DEGs from the bulk RNA-seq previously used for GO analysis, removed genes that were not 334 \nexpressed or were very lowly expressed in the spatial transcriptomics dataset (Log 2 Expression <1). 335 \nWe then used Loupe Browser to display the combined average expression levels of upregulated DEGs 336 \nin our spatial tissue sections at 3 and 5 dpa of epiphysis (n= 43 and 36 genes, respectively) and 337 \ndiaphysis amputations (n=8 and 7 genes, respectively). This approach revealed that, on average, 338 \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 27, 2026. ; https://doi.org/10.64898/2026.02.26.708175doi: bioRxiv preprint \n\n 14\nupregulated DEGs found by bulk RNA-seq in epiphysis amputations at 3 and 5 dpa (Fig. 7B) tended 339 \nto be expressed in the AEC. Indeed, visualization of representative upregulated DEGs in epiphysis-340 \namputated limbs at 3 ( Psat1) and 5 dpa ( Klf17 and Csdn), found that the spatial dots with the highest 341 \nexpression levels were localized to the WE/AEC, with variable numbers of low-expressing dots 342 \nscattered throughout the inner limb tissue (Fig. S6A). This was also observed when the average 343 \nexpression of DEGs annotated as ncRNAs was analyzed (Fig. S6B). However, no such trend was 344 \ndetected with upregulated DEGs in diaphysis limbs at either time point (Fig. 7C, Fig. S6C). The 345 \ncombined average expression of upregulated DEGs in this amputation plane was instead found 346 \nthroughout the limb tissue, which included the epidermis and AEC ( Msln), connective tissue ( Lum, 347 \nPostn), muscle (Myl4, Myh4-like, Tpm1-like), nerves (Mpz), and tendons/joints (Thbs4) (Fig. 7D). 348 \nSurprisingly, the combined average expression of genes in the spatial transcriptomics sections did 349 \nnot seem to reflect the differences in expression of DEGs found by bulk RNA-seq. In particular, 350 \nDEGs upregulated in epiphysis amputations at 5 dpa identified by bulk RNA-seq did not appear to be 351 \naltered in spatial transcriptomics compared to diaphysis amputations. This could be explained as 352 \neither resulting from the flattening of differences between diaphysis and epiphysis amputations caused 353 \nby averaging the expression of multiple genes with different expression levels, by saturation of the 354 \nspatial dots with highly expressed genes, or simply by the fact that the analyzed tissue sections did not 355 \ncontain the particular region of the AEC where gene expression of these DEGs was at its strongest.  356 \nAltogether, the transcriptomic profile differences of the AEC in diaphysis vs. epiphysis amputations 357 \nsuggest positional adaptations of its gene expression profile during limb regeneration, particularly in 358 \nthe secretory profile. However, more work is still needed to identify how these modifications 359 \ncontribute to tissue regeneration and integration according to the amputation position, or what the role 360 \nis of the many differentially expressed ncRNAs in limb regeneration. 361 \nDiscussion: 362 \nA crucial aspect of successful limb regeneration is the robust integration between the tissues of the 363 \nnewly regenerated body part and the previously existing structure. Skeletal tissue integration is 364 \nespecially complicated by its spatially and temporally dynamic tissue composition, in which a fully 365 \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 27, 2026. ; https://doi.org/10.64898/2026.02.26.708175doi: bioRxiv preprint \n\n 15\ncartilaginous skeleton at the end of development becomes progressively mineralized until achieving 366 \nadult patterns of ossification around sexual maturation 7. Recently, our lab showed that the tightly-367 \ncontrolled clearing of skeletal tissue by osteoclasts is essential for skeletal integration 6. However, it 368 \nwas unknown if osteoclast-mediated resorption promoted integration in amputation planes exposing 369 \ndifferent cell and matrix compositions. Moreover, it was also unclear how this process was regulated, 370 \nespecially given its fast-acting nature during regeneration that contrasts with the slow bone 371 \nremodeling in fractures.  372 \nHere we show that skeletal integration strategies during axolotl limb regeneration are customized to 373 \nthe tissues damaged by the amputation, and that these adaptations are initiated early on. We 374 \ndemonstrate that osteoclast-mediated skeletal resorption is primarily activated in amputations through 375 \ncalcified diaphyseal regions, but not through the cartilaginous epiphysis. We also demonstrate that the 376 \nexpression of the chemokine Loc138491483/Ccl24-like is specifically sustained in diaphysis-377 \namputated limbs and is sufficient to promote osteoclast presence in amputated limbs. Finally, we find 378 \nthat the transcriptomic profile of the AEC is modified by the amputation plane, suggesting that this 379 \nstructure may be adapted to the types of tissues injured by the amputation. 380 \nPrevious studies found that successful tissue integration in the axolotl is impacted by factors like 381 \ndefect size 29–31, vitamin D 32 and positional identity incompatibility 33. In this work, we set to 382 \ninvestigate the importance of the skeletal composition itself at the amputation plane. To address this, 383 \nwe performed amputations in two sites that were closely localized within the same limb segment, and, 384 \nthus mainly differed on the affected skeletal tissue composition. We found that, unlike with injuries 385 \naffecting cartilaginous epiphyses, amputations in the calcified diaphysis undergo extensive osteoclast 386 \nrecruitment and/or differentiation and tissue resorption. However, exactly how osteoclast-mediated 387 \ntissue resorption contributes to skeletal integration in diaphyseal amputations is still unknown. We 388 \npropose two non-mutually exclusive hypothesis, which may even act synergistically. 389 \nThe first hypothesis is that this is a specialized additional step of histolysis that facilitates the 390 \nattachment of nascent cartilage cells to the skeletal stump. Tissue histolysis is a key event during 391 \nregeneration4,34–36 in which tissue stiffness greatly decreases 12. Furthermore, ECM remodeling has 392 \nbeen hypothesized to contribute to tissue integration in regenerating newt joints 37. It is thus likely that 393 \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 27, 2026. ; https://doi.org/10.64898/2026.02.26.708175doi: bioRxiv preprint \n\n 16\nosteoclast-mediated tissue resorption primes the mature skeleton for tissue integration by changing the 394 \ncomposition and rigidity of the calcified ECM so that differentiating chondrocytes can be correctly 395 \nincorporated.  396 \nA second hypothesis is that skeletal resorption could help release diaphyseal periskeletal cells. These 397 \ncells, together with dermal and interstitial fibroblasts, contribute to the regeneration of proximal 398 \nskeletal tissues 38–41 and are thus promising candidates to orchestrate skeletal integration. This is 399 \nsupported by our observations that periskeletal cells are strongly influenced by the amputation plane, 400 \nas evidenced by their high and sustained expression of RANKL upon diaphysis amputations. The 401 \nactivation of RANKL in periskeletal cells would 1) help coordinate skeletal tissue integration in the 402 \ncalcified diaphysis by promoting the differentiation of myeloid immune cells into osteoclasts; and 2) 403 \nstimulate osteoclast maturation directly on the surface of the calcified skeletal element.  404 \nOur work also revealed that immune responses during limb regeneration are context-dependent and 405 \nadapted to promote skeletal tissue integration according to its composition. Osteoclasts derive from 406 \nthe myeloid lineage, in which CMPs ultimately undergo terminal differentiation into fully mature 407 \nosteoclasts8,9. Thus, the fact that osteoclast-mediated tissue resorption is primarily triggered by 408 \ndiaphysis amputations demonstrates that the immune response is adapted to the injured tissues. 409 \nAnother adaptation of the immune system was the sustained expression of the previously 410 \nuncharacterized chemokine Loc483 in diaphysis amputations, which we demonstrated to be sufficient 411 \nto promote the presence of osteoclasts in regenerating limbs. The closest match to Loc483 is the 412 \ncaecilian Mcp1/Ccl2 which, in mammals, is reported to be a potent chemotactic factor for 413 \nmonocytes/macrophages42–44 and a promoter of osteoclast maturation 45. It is thus possible that, upon 414 \ncalcified diaphysis injuries, Loc483 expression could have a similar role in the axolotl. Further studies 415 \nare needed to elucidate how Loc483 is regulated in diaphyseal amputations, as well as its role on 416 \nosteoclastogenesis during regeneration.  417 \nFinally, our study further revealed that differences in the tissues affected by the amputation plane 418 \nimpact the transcriptomic profile of the AEC, suggesting that the AEC is a dynamic structure that can 419 \nlikewise be adapted to specific injury contexts. This agrees with our previous study showing that the 420 \nAEC is important in osteoclast-mediated skeletal resorption6. However, it is still unclear how the AEC 421 \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 27, 2026. ; https://doi.org/10.64898/2026.02.26.708175doi: bioRxiv preprint \n\n 17\nimpacts osteoclastogenesis, and how, in turn, the underlying tissue influences the AEC. One way by 422 \nwhich an amputation-dependent AEC could influence regeneration is through the differential secretion 423 \nof serine protease inhibitors, which could moderate or even terminate the histolytic process by 424 \ninhibiting the conversion of MMPs into their active forms 46,47. Intriguingly, we also observed many 425 \ngenes annotated as ncRNAs being differentially expressed in the AEC in our two amputation models. 426 \nGiven their complex biological roles 48,49, ncRNA function in the AEC could become an exciting new 427 \nfield of study.  428 \nUltimately, the fact that tissue integration mechanisms can be customized to the damaged skeletal 429 \ntissues highlights both the robustness and adaptability of regeneration in the axolotl. It would thus be 430 \ninteresting to explore whether similar adaptations happen in other tissues that also display positional 431 \ndifferences in cell type and ECM composition, such as the muscle50.  432 \nFinally, the association between the amputation position and regenerative outcomes has been 433 \nreported in other models. In Xenopus laevis, regenerative potential depends on the tissues affected by 434 \nthe amputation, and regeneration efficiency correlates inversely with the ossification status of the 435 \nskeletal element51,52. In neonatal mice, only amputations through the long bones in the limb, but not 436 \nthrough joints, can activate chondrocyte proliferation 53. These reports and our study thus emphasize 437 \nthe need to, by better understanding customized regenerative responses, start challenging the 438 \nassumption that limb regeneration occurs using a one-size-fits-all molecular milieu.   439 \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 27, 2026. ; https://doi.org/10.64898/2026.02.26.708175doi: bioRxiv preprint \n\n 18\nAcknowledgements: 440 \nWe thank past and current members of the Sandoval-Guzmán lab for continuous support and 441 \ncompanionship during the development of this work. We are particularly grateful to Susanne Weiche 442 \nfor their excellent technical assistance, to Maximilian Krause for his valuable comments on the 443 \nmanuscript, and to Anja Wagner, Beate Gruhl, and Judith Konantz for their dedication to the axolotl 444 \ncare. This work was supported by core facilities of the Technology Platform of the Center for 445 \nMolecular and Cellular Bioengineering (CMCB) of the TU Dresden, namely the Genome Center, the 446 \nLight Microscopy Facility, and the Histology Facility. 447 \nFunding 448 \nR.A. was supported by an Alexander von Humboldt-Stiftung research fellowship (PRT 1208176 449 \nHFST-P) and a Deutsche Forschungsgemeinschaft (DFG) Eigene Stelle Grant (AI 214/1-1, Project 450 \nnumber 523178173). S.D.K. was supported by the Dresden International Graduate School for 451 \nBiomedicine and Bioengineering (DIGS-BB) graduate program. M.A.C. and D.B.G. were supported 452 \nby ERASMUS+ traineeship mobility program. The work at the TU Dresden is co-financed with 453 \ntax revenues based on the budget agreed by the Saxon Landtag. 454 \nConflicts of Interest 455 \nThe authors have no conflicts of interest to declare. 456 \nAuthor Contributions 457 \nR.A. and T.S-G. conceived the study. R.A., C.A., and T.S-G. and acquired funding. R.A. designed 458 \nand performed most experiments, analyzed most data, and wrote the manuscript. S.D.K., K.B., 459 \nM.A.C., D.B.G and Y.S. assisted with experimental work. R.A. and S.D.K. processed and analyzed 460 \nbulk RNA-Seq data. R.A., U.A.F., and S.D.K.., processed and analyzed spatial transcriptomic data. 461 \nC.A. advised on the project. T.S-G provided supervision, critically revised and edited the manuscript. 462 \nAll authors proofread and revised the manuscript. 463 \n  464 \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 27, 2026. ; https://doi.org/10.64898/2026.02.26.708175doi: bioRxiv preprint \n\n 19\nMethods: 465 \nAnimal Husbandry and lines 466 \nHusbandry and experimental procedures were performed according to the Animal Ethics Committee 467 \nof the State of Saxony, Germany. Animals used were selected by their size (snout to tail = ST; snout 468 \nto vent = SV). All experiments in this work were performed in axolotls between 6.0 and 9.0 cm ST. 469 \nAxolotl husbandry was performed in the CRTD axolotl facility using methodology adapted from 54 470 \nand according to the European Directive 2010/63/EU, Annex III, Table 9.1. Axolotls were kept in 18-471 \n19°C water in a 12 h light/12 h dark cycle and a room temperature of 20-22°C. Animals were housed 472 \nin individual tanks categorized by a water surface (WS) area and a minimum water height (MWH). 473 \nAxolotls of a size up to 5 cm SV were maintained in tanks with a WS of 180 cm 2 and MWH of 4.5 474 \ncm. Axolotls up to 9 cm SV were maintained in tanks with a WS of 448 cm2 and MWH of 8 cm.  475 \nWhite axolotls (d/d) were used for most of the experiments. Transgenic lines used included the 476 \npreviously published C-Ti t/+(Sox9:Sox9-T2a-mCherry)ETNKA (referred as Sox9-mCherry )55 and 477 \nTgTol2(Drer.Ctsk:eGFP)TSG (referred to as Ctsk-eGFP)6. 478 \nAnimal procedures 479 \nAmputations were performed in the lower arm  under an Olympus SZX16 stereomicroscope. For all 480 \namputations, animals were anesthetized with 0.01% benzocaine (Sigma-Aldrich, #E1501) solution. 481 \nDiaphysis amputations were performed in the middle of the calcified diaphysis region of the radius 482 \nand ulna, whereas epiphysis amputations were conducted in the distal cartilaginous epiphysis of the 483 \nsame skeletal elements. After surgical procedure, animals were returned to the benzocaine solution 484 \nand allowed to recover for 10 min prior to be transferred back to swimming water. 485 \nIn vivo skeletal staining was performed using calcein (Sigma-Aldrich, #A5533) before amputations. 486 \nA 0.1% solution of calcein in swimming water was prepared and animals were submerged in this 487 \nsolution for 5–10 min in the dark. After staining, axolotls were transferred to a tank with clean 488 \nswimming water, which was changed as many times until the water was clear, and amputated shortly 489 \nafter.  490 \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 27, 2026. ; https://doi.org/10.64898/2026.02.26.708175doi: bioRxiv preprint \n\n 20\nLive imaging was conducted at specific time points in anesthetized d/d or transgenic animals by 491 \nplacing them in a 100 mm petri dish, and positioning the limb accordingly. An Olympus SZX16 492 \nstereoscope microscope (SDF Plapo 0.75xPF) with fluorescence module was used to acquire images. 493 \nLength measurements of the stump skeletal elements and calcified region were performed in scaled 494 \nimages using the line measurement tool in FIJI. 495 \nTissue collection  496 \nTissue collection was performed by euthanizing animals by immersion in a lethal dosage of 497 \nanesthesia (0.1% benzocaine).  498 \nFor paraffin embedding and HCR, limbs were collected and fixed in 1x MEMFa (MOPS 0.1M pH 499 \n7.4, EGTA 2mM, MgSO4 × 7 H2O 1mM, 3.7% formaldehyde) for at least 1 overnight at 4°C. For 500 \nRNA-seq experiments, 1 to 1.5 mm of tissue proximal to the amputation plane was excised, flash 501 \nfrozen in liquid nitrogen, and stored at -80°C until processed for RNA extraction.  502 \nFor plasma measurements, blood was collected directly from the heart immediately after euthanasia 503 \nusing heparin-coated pipette tips into heparin-coated tubes on ice. Then, samples were centrifuged 10 504 \nmin at 3000 g, the upper phase was collected into a new heparin-coated tube, and were kept at -80°C 505 \nuntil further processing. 506 \nCalcium measurements in blood plasma  507 \nCalcium measurement from blood plasma were performed using a Dri-Chem NX600 (Fujifilm). 508 \n10µl of plasma from intact, diaphysis, and epiphys is amputations samples were pipetted into a Fuji 509 \nDRI-Chem Slide (Fujifilm, Ca-P III #2350) and calcium concentration was measured. 510 \nCtsk and Ctsk-like sequence alignments 511 \nSequences were aligned using EMBOSS Needle Pairwise sequence alignment56 and processed with 512 \nSequence Manipulation Suite57. 513 \nRNA extraction, library preparation and bulk RNA sequencing 514 \nSequencing was performed using 3 animals (biological replicates) per amputation site and per 515 \ntimepoint, and each biological sample resulting from the pooling of the two amputated forelimbs of 516 \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 27, 2026. ; https://doi.org/10.64898/2026.02.26.708175doi: bioRxiv preprint \n\n 21\neach animal. RNA extraction was performed using RNAeasy Mini Plus Kit (Qiagen, #74134) 517 \naccording to the manufacturer's instructions. Samples were disrupted and homogenized using a 518 \nPolytron tissue homogenizer (Kinematica, #PT1600E) in 350 µl of RLT Plus Buffer containing β -519 \nmercaptoethanol (Sigma, #M625). Extracted RNA was stored at -80 until processed for sequencing. 520 \nRNA sequencing libraries were prepared using Watchmaker mRNA Library Prep (Watchmaker 521 \nGenomics, #7K0078) on Biomek i7 with estimated fragment sizes of 300 - 400 bp. Poly-dT pull down 522 \nenrichment of mRNA was performed before sequencing 101 bp paired-end reads on an Illumina 523 \nNovaSeq 6000 (Illumina), generating between 50 million read pairs per sample. RNA-seq raw data 524 \n(fastq) has been deposited in NCBI under the Gene Expression Omnibus (GEO) accession code 525 \nGSEXXXXXX.  526 \nBulk RNA-seq read mapping and expression analysis 527 \nGenerated reads from diaphysis and epiphysis amputated limbs were mapped against the current 528 \naxolotl reference genome available from NCBI (UKY_AmexF1_1; GCF_040938575.1) using 529 \nHISAT2 v2.2.158. HISAT2 was run through the command line with default parameters, and a known-530 \nsplicesite-infile created from the corresponding gtf annotation file via the 531 \nhisat2_extract_splice_sites.py command. StringTie (version 2.2.1 59,60) was then run through the 532 \ncommand line with standard parameters and the option of assembling novel transcripts to produce a 533 \nMerged Transcripts annotation file that was used for transcript quantification. Finally, normalized 534 \ncounts per million (CPM) values for each sample were calculated using the Bioconductor package 535 \nedgeR (version 3.40.2 61), for R (version 4.2.2 62). Raw gene counts for mandible and limb can be 536 \nfound in Table S2. Normalized gene counts (CPM) are provided in Table SX. 537 \nGene Ontology (GO) analysis of differentially expressed genes 538 \nFrom the full list of genes found to be differentially expressed in each condition, ncRNAs and 539 \nuncharacterized genes were filtered. The remaining genes of interest were analyzed for significantly 540 \nenriched GO terms via DAVID v6.8 63 using default parameters, and are available in Tables SX-SX. 541 \nGO enrichment terms were considered statistically significant when p < 0.01.  542 \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 27, 2026. ; https://doi.org/10.64898/2026.02.26.708175doi: bioRxiv preprint \n\n 22\nSpatial transcriptomics sequencing and analysis  543 \nSpatial transcriptomics was performed using first generation Visium Spatial Gene Expression 544 \nSystem for Fresh Frozen tissue (10X Genomics) as previously described 64. Briefly, animals 8–9cm 545 \nsnout to tail were amputated in diaphysis and epiphysis regions and allowed to regenerate until 3 or 5 546 \ndpa. Limbs were then harvested at the level of the upper arm, flash frozen in OCT, and then stored at 547 \n−80 °C. Samples were cryosectioned at -11/12 °C (chamber)/ -31°C (blade) at a thickness of 10 μ m.  548 \nOptimization and gene expression assays were carried out according to the manufacturer’s 549 \ninstructions. Briefly, slides were fixed in −20 °C methanol, dried with isopropanol, and stained with 550 \nH&E. A tile scan of all capture areas was generated using an OlympusOVK automated slide scanner 551 \nsystem with a color camera and fluorescent module. For tissue optimization, enzymatic 552 \npermeabilization was conducted for 0–30 min, followed by first-strand cDNA synthesis with 553 \nfluorescent nucleotides. The slide was reimaged using the standard Cy3 filter cube. An optimal 554 \npermeabilization time of 20 min was determined by visual inspection to maximize mRNA recovery 555 \nwhile at the same time minimizing diffusion. For gene expression, the initial workflow was similar to 556 \nthe optimization procedure and was done according to the manufacturer’s instructions. After tissue 557 \nlysis and RT, amplification of cDNA and library preparation – involving fragmentation, dA-Tailing, 558 \nadapter ligation and a 18 cycles indexing PCR under following conditions: 98 °C 45 sec, 10/14 cycles 559 \n[98°C 20 sec, 67°C 30 sec, 72°C 20 sec], 72°C 1 min, 4°C hold – was performed based on the 560 \nmanufacturer’s protocol using the Library Construction kit (10X Genomics, #PN-1000190). Included 561 \nin this protocol was a double-sided SPRI bead (Beckman Coulter, #B23319) size selection (0.6x/0.8x) 562 \npurification and a final purification (0.8x). After checking the quality control and quantification with 563 \nFragment Analyzer (Agilent, NGS Fragment Kit #DNF-473), the libraries were sequenced on an 564 \nIllumina Novaseq6000 in paired-end mode (R1/R2: 100 cycles; I1/I2: 10 cycles), generating 50-100 565 \nmillion fragment pairs for each library. 566 \nThe raw sequencing data was then processed with the ‘count’ command of the Space Ranger 567 \nsoftware (v3.1.1) provided by 10X Genomics. The Space Ranger reference for the axolotl genome 568 \n(UKY_AmexF1_1) was built using the Merged Transcripts gene annotation file generated from the 569 \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 27, 2026. ; https://doi.org/10.64898/2026.02.26.708175doi: bioRxiv preprint \n\n 23\nbulk RNA-seq experiments (available upon request) as input for the ‘mkref’ command of Space 570 \nRanger.  571 \nFor clustering and gene expression analysis, we used Seurat v.5.2.1 and the following analysis 572 \npipeline: Barcodes (spots) are filtered by having 200 or more reads and a mitochondrial content of 573 \nless than 50% and features (genes) are filtered for having at least 3 cells with at least 1 read. The raw 574 \ndata as UMI counts were regressed to remove the effect of library size using the log normalization. 575 \nHighly variable genes were identified using the variance-stabilizing transformation (vst), selecting the 576 \ntop 3,000. These genes were scaled and used as input for the dimensionality reduction by PCA 577 \nanalysis. 50 principle components were calculated and the top 20 are used for further calculations. For 578 \nvisualization purposes, Uniform Manifold Approximation and Projection (UMAP, k nearest neighbors 579 \n= 30) was applied to project high-dimensional gene expression data into two dimensions. Graph-based 580 \nclustering (k nearest neighbors = 30, resolution = 1.0) used the Leiden algorithm 65. Marker genes that 581 \nwere differentially expressed for each cluster were identified using the Wilcoxon rank-sum test and 582 \nBenjamini-Hochberg method to correct for multiple comparisons (presto v1.0.0). Markers which are 583 \nexpressed in at least 25% of the cells in a cluster and with a Log2 fold change >2 were included.  584 \nRNA-seq raw data (fastq) has been deposited in NCBI under the Gene Expression Omnibus (GEO) 585 \naccession code GSEXXXXXX. 586 \nAnalysis of Nfatc1 enriched spatial spots 587 \nAnalysis of DEGs in Nfatc1 expressing spatial spots was performed in Loupe Browser v8.1.2. using 588 \nthe Advanced Selection tool to set up a threshold of Nfatc1 fold expression (Log 2 Exp) > 2. This 589 \napproach selected 108 barcodes/spots. These were used to run a differential expression analysis 590 \ncomparing the selected barcodes in the UMAP to the entire dataset. The resulting list of most 591 \ndifferentially expressed gene is supplied in Supplemental file XX. 592 \nParaffin sectioning and Movat’ s pentachrome stainings 593 \nHistological stainings were performed in intact limbs (no amputation), and in regenerating limbs 594 \nafter diaphysis and epiphysis amputations at 7-, and 9 dpa. 3 animals per time point were used. 595 \nAxolotl limbs were fixed in MEMFa and decalcified in 0.5 M EDTA for 2 weeks with daily changes 596 \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 27, 2026. ; https://doi.org/10.64898/2026.02.26.708175doi: bioRxiv preprint \n\n 24\nof the solution. Sample embedding, sectioning and staining was performed by the CMCB Histology 597 \nFacility, Dresden. Briefly, samples were dehydrated in a series of EtOH in RNase-free water until 598 \n100% EtOH, and then embedded in paraffin. Longitudinal sections of 4-5 µm were generated using a 599 \nmicrotome. Movat’s Pentachrome ( Morphisto, #12057 ) staining was performed according to the 600 \nmanufacturer’s instructions. Imaging was performed using an Olympus OVK automated slide scanner 601 \nsystem (UPLSAPO 20x/0.75).  602 \nHybridization Chain Reaction (HCR) staining 603 \nProbe sets for Ctsk, Ctsk-like, and Loc483 were designed using the HCR probe generator created by 604 \nthe Monaghan Lab (https://github.com/Monaghan-Lab/probegenerator ) 66 and purchased as oligo 605 \npools (oPools Oligo Pools) from Integrated DNA Technologies. Probe sequences can be found in 606 \nTable SX. 607 \nWhole mount HCR was performed according to 67 with some modifications. Briefly, limbs were 608 \nrehydrated through a series of MetOH in RNase-free water and washed three times in PBT (0.1% 609 \nTween 20 in PBS). Tissue was then delipidated in Delipidation Solution (200mM Boric acid, 4% 610 \nSDS, pH 8.5 in RNAse-free water) for 2 hours at 37°C. After three washes in PBT, limbs were 611 \npermeabilized with Permeabilization Solution (0.3M Glycine, 2% Triton X-100, 20% DMSO in PBS) 612 \nfor 1 hour at room temperature (RT). Limbs were washed again in PBT, incubated in pre-warmed 613 \nHybridization Buffer (Molecular Instruments, #BPH01726) for 5 minutes and then pre-hybridized in 614 \nnew Hybridization Buffer for 30 minutes at 37°C. After this, tissue was incubated overnight with 615 \nHybridization Buffer containing 2 pmol per 500 µl of probe solution. The following day, limbs were 616 \nwashed four times with agitation for 15 minutes with Wash Buffer (Molecular Instruments, 617 \n#BPW01726) at 37°C and two times for 5 minutes in 5× SSCT (3M NaCl, 300 mM sodium citrate, 618 \n0.1% Tween 20, in water) at room temperature. Pre-amplification was performed for 5 minutes at RT 619 \nin Amplification Buffer (Molecular Instruments, #BAM01826), followed by amplification for 16-24 620 \nhours at RT in Amplification buffer with 30 pmol of each hairpin. Finally, tissue was extensively 621 \nwashed in 5× SSCT, incubated overnight with Hoechst 33258 (Abcam, #ab228550) 1:1000 in PBS, 622 \nand cleared in EasyIndex (LifeCanvas Technologies, #EI-500-1.52) for a minimum of one overnight. 623 \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 27, 2026. ; https://doi.org/10.64898/2026.02.26.708175doi: bioRxiv preprint \n\n 25\nSamples were mounted in a glass bottom dish in EasyIndex, and then imaged using a Zeiss LSM 980 624 \ninverted confocal laser scanning microscope (Plan-apochromat 10x/0.45) with 10 µm between optical 625 \nplanes. Each HCR was performed in a minimum of 3 biological replicates. 626 \nOrthology inference of Loc483 with aLeaves 627 \nOrthology of Loc138491483 was examined using the aLeaves webserver 23 with default parameters 628 \nand selected databases #1 Human – Refseq, #4 Non-eutherian Mammals – Ensembl 104, #5 Non-629 \nmammalian Bony Vertebrates – Ensembl 104 and others, #6: Cartilaginous fish and cyclostomes, and 630 \n#7: All vertebrate entries except mammalians in NCBI Protein . 631 \nInjection of BAPTA and CaSO4 in regenerating limb blastemas 632 \nFor the injections of BAPTA (Abcam, #ab144924), a working solution was prepared by diluting a 633 \nstock solution of 25mM of BAPTA (in DMSO) in APBS (80% PBS in RNAse-free water) with 1% 634 \nFast Green (Sigma-Aldrich, #F7252) for easier visualization while injecting. For CaSO4, a 15mM 635 \nCaSO4 solution in water was directly injected in the blastemas, also with 1% Fast Green for easier 636 \nvisualization. 637 \nAnimals were first anesthetized and then injected using using a fine heat-pulled glass capilary to 638 \ninject 150 nL of solution in both limbs. Ctsk+ signal was quantified by first defining a ROI of 3.5 639 \nmm2, which encompassed most of the lower arm, and applying the maximum entropy threshold 640 \nmethod. The area Ctsk+ signal was then measured using the Area function in FIJI. 641 \nCloning, and injection and electroporation of Loc483 in regenerating diaphysis or epiphysis limbs 642 \nTo obtain and clone Loc483, first RNA was extracted of diaphysis amputated limbs as above. cDNA 643 \nwas prepared using Takara PrimeScript™ 1st strand cDNA Synthesis kit (Takara Bio Inc, #6110A) 644 \naccording the manufacturer’s instructions and using Random 6mers primers. The full coding sequence 645 \nof Loc483 was then amplified with Phusion® High Fidelity DNA Polymerase (New England Biolabs, 646 \n#M0530) from whole cDNA according to the manufacturer’s instructions and using the primers in 647 \nSupp file XX. The fragment contaning the coding sequence of Loc483 was then cloned into an vector 648 \ncontaining the CAGGS promoter for expression using standard cloning techniques.  649 \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 27, 2026. ; https://doi.org/10.64898/2026.02.26.708175doi: bioRxiv preprint \n\n 26\nPlasmid administration was conducted by first anesthesizing the animals and injecting a solution of 650 \n1.5 µg/µL of Loc483 plasmid + reporter plasmid (experimental conditions) or just the reporter 651 \nplasmid (control conditions) diluted in APBS (0.8X PBS) into regenerating limbs using a fine heat-652 \npulled glass capilary. Both solutions contained 1% Fast Green for easier visualization while injecting 653 \nAnimals were then immediately electroporated using a Super Electroporator NEPA21 TypeII (Nepa 654 \nGene) with 2 poring pulses of 80V , 50 msec length, 50 msec interval, 9% decay rate, and 5 transfer 655 \npulses of 40V , 50 msec length, 999 msec interval, 5% decay rate, using a tweezer electrode (Nepa 656 \nGene, #CUY650P3). After electroporation, animals were returned to their holding tanks containing 657 \nswimming water, and allowed to regenerate until imaged at specific time points as before. Ctsk+ signal 658 \nwas quantified by first defining a ROI of 3.5 mm2, applying the maximum entropy threshold method, 659 \nand then measured using the Area function in FIJI. 660 \nImage processing, analysis and quantification  661 \nAll images were processed using Fiji 68. Processing involved selecting regions of interest, merging, 662 \nor splitting channels, and improving brightness and contrast levels for proper presentation in figures. 663 \nMaximum intensity projections were done in some confocal images, and it is stated in the respective 664 \nfigure’s descriptions. When appropriate, stitching of tiles was done directly in the acquisition software 665 \nZen (Zeiss Microscopy, Jena, Germany).  666 \nData representation and statistical analysis 667 \nAll graphs and statistical analyses were performed with GraphPad Prism 10 (GraphPad Software, 668 \nSan Diego, CA, USA). Specific number of replicates, statistical tests and pos-hoc tests are indicated in 669 \nthe respective figure legends. All figures were generated with Affinity Designer (Serif Europe, West 670 \nBridgford, UK).   671 \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 27, 2026. ; https://doi.org/10.64898/2026.02.26.708175doi: bioRxiv preprint \n\n 27\nReferences: 672 \n1. Aires, R., Keeley, S. D. & Sandoval-Guzmán, T. Basics of Self-Regeneration. in Cell 673 \nEngineering and Regeneration (eds Gimble, J. M., Marolt Presen, D., Oreffo, R. O. C., 674 \nWolbank, S. & Redl, H.) 1–44 (Springer International Publishing, Cham, 2020). 675 \ndoi:10.1007/978-3-319-37076-7_66-1. 676 \n2. Stocum, D. L. Mechanisms of urodele limb regeneration. Regeneration  4, 159–200 677 \n(2017). 678 \n3. Aztekin, C. Tissues and Cell Types of Appendage Regeneration: A Detailed Look at the 679 \nWound Epidermis and Its Specialized Forms. Front. Physiol. 12, 771040 (2021). 680 \n4. Vinarsky, V ., Atkinson, D. L., Stevenson, T. J., Keating, M. T. & Odelberg, S. J. Normal 681 \nnewt limb regeneration requires matrix metalloproteinase function. Dev. Biol. 279, 86–98 682 \n(2005). 683 \n5. Otsuki, L. & Tanaka, E. M. Positional Memory in Vertebrate Regeneration: A Century’s 684 \nInsights from the Salamander Limb. Cold Spring Harb. Perspect. Biol. 14, a040899 685 \n(2022). 686 \n6. Riquelme-Guzmán, C. et al. Osteoclast-mediated resorption primes the skeleton for 687 \nsuccessful integration during axolotl limb regeneration. eLife 11, e79966 (2022). 688 \n7. Riquelme /i1 Guzmán, C. et al. Postembryonic development and aging of the appendicular 689 \nskeleton in  AMBYSTOMA MEXICANUM . Dev. Dyn. 251, 1015–1034 (2021). 690 \n8. Tsai, J., Kaneko, K., Suh, A. J., Bockman, R. & Park-Min, K.-H. Origin of Osteoclasts: 691 \nOsteoclast Precursor Cells. J. Bone Metab. 30, 127–140 (2023). 692 \n9. Ono, T. & Nakashima, T. Recent advances in osteoclast biology. Histochem. Cell Biol. 693 \n149, 325–341 (2018). 694 \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 27, 2026. ; https://doi.org/10.64898/2026.02.26.708175doi: bioRxiv preprint \n\n 28\n10. Takegahara, N., Kim, H. & Choi, Y . Unraveling the intricacies of osteoclast 695 \ndifferentiation and maturation: insight into novel therapeutic strategies for bone-696 \ndestructive diseases. Exp. Mol. Med. 56, 264–272 (2024). 697 \n11. Einhorn, T. A. & Gerstenfeld, L. C. Fracture healing: mechanisms and interventions. Nat. 698 \nRev. Rheumatol. 11, 45–54 (2015). 699 \n12. Edwards-Jorquera, S., Aires, R., Ulbricht, E., Taubenberger, A. & Sandoval-Guzmán, T. 700 \nAtomic Force Microscopy Measurements of Cartilage in Intact and Regenerating Axolotl 701 \nLimbs. J. V is. Exp. 66946 (2024) doi:10.3791/66946. 702 \n13. Takami, Y . & Une, Y . Blood Clinical Biochemistries and Packed Cell V olumes for the 703 \nMexican Axolotl (Ambystoma mexicanum). J. Herpetol. Med. Surg. 27, 104–110 (2017). 704 \n14. Bryant, D. M. et al. Identification of regenerative roadblocks via repeat deployment of 705 \nlimb regeneration in axolotls. Npj Regen. Med. 2, 30 (2017). 706 \n15. Keeley, S. D. et al. Differential fates of Kazald gene quartet: ancestral roles in 707 \nskeletogenesis and regeneration to putative innovations in fish and birds. iScience 114934 708 \n(2026) doi:10.1016/j.isci.2026.114934. 709 \n16. Takahashi, N. et al. Postmitotic Osteoclast Precursors Are Mononuclear Cells Which 710 \nExpress Macrophage-Associated Phenotypes. Dev. Biol. 163, 212–221 (1994). 711 \n17. Udagawa, N. et al. Origin of osteoclasts: mature monocytes and macrophages are capable 712 \nof differentiating into osteoclasts under a suitable microenvironment prepared by bone 713 \nmarrow-derived stromal cells. Proc. Natl. Acad. Sci. 87, 7260–7264 (1990). 714 \n18. Anderson, D. M. et al. A homologue of the TNF receptor and its ligand enhance T-cell 715 \ngrowth and dendritic-cell function. Nature 390, 175–179 (1997). 716 \n19. Nakagawa, N. et al. RANK Is the Essential Signaling Receptor for Osteoclast 717 \nDifferentiation Factor in Osteoclastogenesis. Biochem. Biophys. Res. Commun. 253, 395–718 \n400 (1998). 719 \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 27, 2026. ; https://doi.org/10.64898/2026.02.26.708175doi: bioRxiv preprint \n\n 29\n20. Ono, T., Hayashi, M., Sasaki, F. & Nakashima, T. RANKL biology: bone metabolism, the 720 \nimmune system, and beyond. Inflamm. Regen. 40, 2 (2020). 721 \n21. Tsukasaki, M. & Takayanagi, H. Osteoimmunology: evolving concepts in bone–immune 722 \ninteractions in health and disease. Nat. Rev. Immunol. 19, 626–642 (2019). 723 \n22. Takayanagi, H. et al. Induction and activation of the transcription factor NFATc1 724 \n(NFAT2) integrate RANKL signaling in terminal differentiation of osteoclasts. Dev. Cell 725 \n3, 889–901 (2002). 726 \n23. Kuraku, S., Zmasek, C. M., Nishimura, O. & Katoh, K. aLeaves facilitates on-demand 727 \nexploration of metazoan gene family trees on MAFFT sequence alignment server with 728 \nenhanced interactivity. Nucleic Acids Res. 41, W22–W28 (2013). 729 \n24. Y ahara, Y . et al. Erythromyeloid progenitors give rise to a population of osteoclasts that 730 \ncontribute to bone homeostasis and repair. Nat. Cell Biol. 22, 49–59 (2020). 731 \n25. Loser, K. et al. Epidermal RANKL controls regulatory T-cell numbers via activation of 732 \ndendritic cells. Nat. Med. 12, 1372–1379 (2006). 733 \n26. Müller, M. R. & Rao, A. NFAT, immunity and cancer: a transcription factor comes of age. 734 \nNat. Rev. Immunol. 10, 645–656 (2010). 735 \n27. Räisänen, S. R. et al. Macrophages overexpressing tartrate-resistant acid phosphatase 736 \nshow altered profile of free radical production and enhanced capacity of bacterial killing. 737 \nBiochem. Biophys. Res. Commun. 331, 120–126 (2005). 738 \n28. Zou, N., Liu, R. & Li, C. Cathepsin K+ Non-Osteoclast Cells in the Skeletal System: 739 \nFunction, Models, Identity, and Therapeutic Implications. Front. Cell Dev. Biol. 10, 740 \n818462 (2022). 741 \n29. Hutchison, C., Pilote, M. & Roy, S. The axolotl limb: A model for bone development, 742 \nregeneration and fracture healing. Bone 40, 45–56 (2007). 743 \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 27, 2026. ; https://doi.org/10.64898/2026.02.26.708175doi: bioRxiv preprint \n\n 30\n30. Polikarpova, A. et al. The specialist in regeneration—the Axolotl—a suitable model to 744 \nstudy bone healing? Npj Regen. Med. 7, 35 (2022). 745 \n31. Satoh, A., Cummings, G. M. C., Bryant, S. V . & Gardiner, D. M. Neurotrophic regulation 746 \nof fibroblast dedifferentiation during limb skeletal regeneration in the axolotl 747 \n(Ambystoma mexicanum). Dev. Biol. 337, 444–457 (2010). 748 \n32. Vieira, W. A., Wells, K. M., Milgrom, R. & McCusker, C. D. Exogenous Vitamin D 749 \nsignaling alters skeletal patterning, differentiation, and tissue integration during limb 750 \nregeneration in the axolotl. Mech. Dev. 153, 1–9 (2018). 751 \n33. Vieira, W. A. et al. Integration failure of regenerated limb tissue is associated with 752 \nincongruencies in positional information in the Mexican axolotl. Front. Cell Dev. Biol. 753 \n11, 1152510 (2023). 754 \n34. Huang, T., Zuo, L., Walczyń ska, K. S., Zhu, M. & Liang, Y . Essential roles of matrix 755 \nmetalloproteinases in axolotl digit regeneration. Cell Tissue Res. 385, 105–113 (2021). 756 \n35. Satoh, A., Makanae, A., Hirata, A. & Satou, Y . Blastema induction in aneurogenic state 757 \nand Prrx-1 regulation by MMPs and FGFs in Ambystoma mexicanum limb regeneration. 758 \nDev. Biol. 355, 263–274 (2011). 759 \n36. Y ang, E. V ., Gardiner, D. M., Carlson, M. R. J., Nugas, C. A. & Bryant, S. V . Expression 760 \nofMmp-9 and related matrix metalloproteinase genes during axolotl limb regeneration. 761 \nDev. Dyn. 216, 2–9 (1999). 762 \n37. Tsutsumi, R., Inoue, T., Yamada, S. & Agata, K. Reintegration of the regenerated and the 763 \nremaining tissues during joint regeneration in the newt Cynops pyrrhogaster. 764 \nRegeneration 2, 26–36 (2015). 765 \n38. Currie, J. D. et al. Live Imaging of Axolotl Digit Regeneration Reveals Spatiotemporal 766 \nChoreography of Diverse Connective Tissue Progenitor Pools. Dev. Cell 39, 411–423 767 \n(2016). 768 \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 27, 2026. ; https://doi.org/10.64898/2026.02.26.708175doi: bioRxiv preprint \n\n 31\n39. Gerber, T. et al. Single-cell analysis uncovers convergence of cell identities during 769 \naxolotl limb regeneration. Science 362, eaaq0681 (2018). 770 \n40. Kaucka, M. et al. Altered developmental programs and oriented cell divisions lead to 771 \nbulky bones during salamander limb regeneration. Nat. Commun. 13, 6949 (2022). 772 \n41. McCusker, C. D., Diaz-Castillo, C., Sosnik, J., Q. Phan, A. & Gardiner, D. M. Cartilage 773 \nand bone cells do not participate in skeletal regeneration in Ambystoma mexicanum 774 \nlimbs. Dev. Biol. 416, 26–33 (2016). 775 \n42. V olejnikova, S., Laskari, M., Marks, S. C. & Graves, D. T. Monocyte recruitment and 776 \nexpression of monocyte chemoattractant protein-1 are developmentally regulated in 777 \nremodeling bone in the mouse. Am. J. Pathol. 150, 1711–1721 (1997). 778 \n43. Rahimi, P. et al. Monocyte chemoattractant protein-1 expression and monocyte 779 \nrecruitment in osseous inflammation in the mouse. Endocrinology 136, 2752–2759 780 \n(1995). 781 \n44. Fuentes, M. E. et al. Controlled recruitment of monocytes and macrophages to specific 782 \norgans through transgenic expression of monocyte chemoattractant protein-1. J. Immunol. 783 \n155, 5769–5776 (1995). 784 \n45. Kim, M. S., Day, C. J. & Morrison, N. A. MCP-1 Is Induced by Receptor Activator of 785 \nNuclear Factor-κ B Ligand, Promotes Human Osteoclast Fusion, and Rescues 786 \nGranulocyte Macrophage Colony-stimulating Factor Suppression of Osteoclast 787 \nFormation*. J. Biol. Chem. 280, 16163–16169 (2005). 788 \n46. Duncan, M. E., Richardson, J. P., Murray, G. I., Melvin, W. T. & Fothergill, J. E. Human 789 \nmatrix metalloproteinase-9: activation by limited trypsin treatment and generation of 790 \nmonoclonal antibodies specific for the activated form. Eur. J. Biochem. 258, 37–43 791 \n(1998). 792 \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 27, 2026. ; https://doi.org/10.64898/2026.02.26.708175doi: bioRxiv preprint \n\n 32\n47. Wilkinson, D. J. et al. The serine proteinase hepsin is an activator of pro-matrix 793 \nmetalloproteinases: molecular mechanisms and implications for extracellular matrix 794 \nturnover. Sci. Rep. 7, 16693 (2017). 795 \n48. Chen, L.-L. & Kim, V . N. Small and long non-coding RNAs: Past, present, and future. 796 \nCell 187, 6451–6485 (2024). 797 \n49. Mattick, J. S. et al. Long non-coding RNAs: definitions, functions, challenges and 798 \nrecommendations. Nat. Rev. Mol. Cell Biol. 24, 430–447 (2023). 799 \n50. Diogo, R. & Tanaka, E. M. Anatomy of the pectoral and forelimb muscles of wildtype 800 \nand green fluorescent protein-transgenic axolotls and comparison with other tetrapods 801 \nincluding humans: a basis for regenerative, evolutionary and developmental studies. J. 802 \nAnat. 221, 622–635 (2012). 803 \n51. Nye, H. L. D. & Cameron, J. A. Strategies to reduce variation in Xenopus regeneration 804 \nstudies. Dev. Dyn. 234, 151–158 (2005). 805 \n52. Wolfe, A. D., Nye, H. L. D. & Cameron, J. A. Extent of ossification at the amputation 806 \nplane is correlated with the decline of blastema formation and regeneration in Xenopus 807 \nlaevis hindlimbs. Dev. Dyn. 218, 681–697 (2000). 808 \n53. Masaki, H. & Ide, H. Regeneration potency of mouse limbs. Dev. Growth Differ. 49, 89–809 \n98 (2007). 810 \n54. Khattak, S. et al. Optimized axolotl (Ambystoma mexicanum) husbandry, breeding, 811 \nmetamorphosis, transgenesis and tamoxifen-mediated recombination. Nat. Protoc. 9, 812 \n529–540 (2014). 813 \n55. Fei, J.-F. et al. Application and optimization of CRISPR–Cas9-mediated genome 814 \nengineering in axolotl (Ambystoma mexicanum). Nat. Protoc. 13, 2908–2943 (2018). 815 \n56. Madeira, F. et al. The EMBL-EBI Job Dispatcher sequence analysis tools framework in 816 \n2024. Nucleic Acids Res. 52, W521–W525 (2024). 817 \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 27, 2026. ; https://doi.org/10.64898/2026.02.26.708175doi: bioRxiv preprint \n\n 33\n57. Stothard, P. The sequence manipulation suite: JavaScript programs for analyzing and 818 \nformatting protein and DNA sequences. BioTechniques 28, 1102, 1104 (2000). 819 \n58. Kim, D., Paggi, J. M., Park, C., Bennett, C. & Salzberg, S. L. Graph-based genome 820 \nalignment and genotyping with HISAT2 and HISAT-genotype. Nat. Biotechnol. 37, 907–821 \n915 (2019). 822 \n59. Pertea, M. et al. StringTie enables improved reconstruction of a transcriptome from 823 \nRNA-seq reads. Nat. Biotechnol. 33, 290–295 (2015). 824 \n60. Pertea, M., Kim, D., Pertea, G. M., Leek, J. T. & Salzberg, S. L. Transcript-level 825 \nexpression analysis of RNA-seq experiments with HISAT, StringTie and Ballgown. Nat. 826 \nProtoc. 11, 1650–1667 (2016). 827 \n61. Robinson, M. D., McCarthy, D. J. & Smyth, G. K. edgeR/i1 : a Bioconductor package for 828 \ndifferential expression analysis of digital gene expression data. Bioinformatics 26, 139–829 \n140 (2010). 830 \n62. R Core Team. R: A language and environment for statistical computing. Foundation for 831 \nStatistical Computing, (2021). 832 \n63. Huang, D. W., Sherman, B. T. & Lempicki, R. A. Systematic and integrative analysis of 833 \nlarge gene lists using DA VID bioinformatics resources. Nat. Protoc. 4, 44–57 (2009). 834 \n64. Zhong, J. et al. Multi-species atlas resolves an axolotl limb development and regeneration 835 \nparadox. Nat. Commun. 14, 6346 (2023). 836 \n65. Traag, V . A., Waltman, L. & van Eck, N. J. From Louvain to Leiden: guaranteeing well-837 \nconnected communities. Sci. Rep. 9, 5233 (2019). 838 \n66. Stein, D. & Monaghan, J. davidfstein/probegenerator: Beta Release (v0.1.0). Zenodo 839 \nhttps://doi.org/10.5281/zenodo.3516447 (2019) 840 \ndoi:https://doi.org/10.5281/zenodo.3516447. 841 \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 27, 2026. ; https://doi.org/10.64898/2026.02.26.708175doi: bioRxiv preprint \n\n 34\n67. Choi, H. M. T. et al. Third-generation in situ hybridization chain reaction: multiplexed, 842 \nquantitative, sensitive, versatile, robust. Development 145, dev165753 (2018). 843 \n68. Schindelin, J. et al. Fiji: an open-source platform for biological-image analysis. Nat. 844 \nMethods 9, 676–682 (2012). 845 \n 846 \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 27, 2026. ; https://doi.org/10.64898/2026.02.26.708175doi: bioRxiv preprint \n\n 1 \n 2 \nFig. 1 Extensive osteoclast-mediated tissue resorption is specifically triggered by amputations 3 \nthrough the calcified diaphyseal region of the radius and ulna.  A. Schematic representation of 4 \namputation planes through the calcified diaphysis and cartilaginous epiphysis (left) and experimental 5 \nset up (right). B. Time course of tissue resorption during lower arm regeneration in diaphysis and 6 \nepiphysis amputations at 0, 7, 9, 11, 15, and 18 dpa. Representative images of an experiment with N= 7 \n5 animals per condition. Scale bar: 500 µm. C.  Quantification of radius and ulna length of the stump 8 \nafter diaphysis and epiphysis amputations over time. D. Quantification of the calcified region length 9 \nin radii and ulnas after diaphysis and epiphysis amputations over time. E. Movat’s pentachrome 10 \nstaining of representative longitudinal sections of intact lower arms (left) and diaphysis and epiphysis 11 \nlimbs at 7- and 9 dpa. R, radius, U, ulna. Dashed lines represent the contours of the radius and ulna. 12 \nBoxes represent the inset location. Black arrowheads indicate multinucleated osteoclasts. Scale bar: 13 \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 27, 2026. ; https://doi.org/10.64898/2026.02.26.708175doi: bioRxiv preprint \n\n500 µm, scale bar in insets: 100 µm. For C and D, N= 5 animals. The lines show mean values over 14 \ntime ± sd. #p<0.05, ##p<0.01, and ###p<0.001 for diaphysis vs epiphysis amputations in the ulna; 15 \n*p<0.05, **p<0.01 and ***p<0.001 for diaphysis vs epiphysis amputations in the radius (two-way 16 \nANOVA with Tukey’s post-hoc test).  17 \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 27, 2026. ; https://doi.org/10.64898/2026.02.26.708175doi: bioRxiv preprint \n\n 18 \nFig. 2 Diaphysis amputations specifically trigger osteoclast recruitment in axolotl limb 19 \nregeneration. A. Time course of osteoclast presence in diaphysis and epiphysis amputations at 0, 7, 9, 20 \n11, 15, and 18 dpa. Chondrocytes are labelled in white, osteoclasts in green. Scale bar: 1 mm. 21 \nRepresentative images of an experiment with N= 5 animals per condition. B. Hybridization chain 22 \nreaction (HCR) for Ctsk  (green) and Ctsk-like (magenta) at 3- and 5 dpa in representative diaphysis 23 \nand epiphysis amputated limbs. White arrowheads indi cate the multinucleated osteoclasts depicted in 24 \ncorresponding insets. Scale bar: 500 µm, scale bar in insets: 50 µm.   25 \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 27, 2026. ; https://doi.org/10.64898/2026.02.26.708175doi: bioRxiv preprint \n\n 26 \n 27 \nFig. 3 Osteoclast presence is not correlated with systemic calcium levels or impacted by local 28 \ncalcium changes. A. Quantification of calcium levels in the serum of diaphysis and epiphysis 29 \namputated animals in intact conditions, and at 12 hpa, 1-, 3-, 5-, and 9 dpa. The graph represents the 30 \nmean and sd of the combined results of 4 independent experiments using a minimum of 3 animals per 31 \ncondition and per time point. B. Schematic representation of the experimental set up for the injections 32 \nin diaphysis and epiphysis amputated limbs. C. Time course of osteoclast presence in representative 33 \ndiaphysis and epiphysis amputated limbs injected with DMSO (control) or 10mM BAPTA at 5-, 7-, 9- 34 \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 27, 2026. ; https://doi.org/10.64898/2026.02.26.708175doi: bioRxiv preprint \n\nand 11 dpa. Scale bar: 1 mm. D. Quantification of the area of Ctsk+ signal in BAPTA and DMSO 35 \ninjected animals in diaphysis amputations. E. Quantification of the area of Ctsk+ signal in BAPTA and 36 \nDMSO injected animals in epiphysis amputations. F. Time course of osteoclast presence in 37 \nrepresentative diaphysis and epiphysis amputated limbs injected with water (control) or CaSO4 at 5-, 38 \n7-, 9- and 11 dpa. Scale bar: 1 mm. G. Quantification of the area of Ctsk+ signal in water or CaSO4 39 \ninjected animals in diaphysis amputations. H. Quantification of the area of Ctsk+ signal in water or 40 \nCaSO4 injected animals in epiphysis amputations. C and F show representative images of an 41 \nexperiment with N= 4 animals per condition. For C-H, the 2 limbs of 4 animals were injected and 42 \nassessed per condition over time. For D, E, G and H, the lines show mean values over time.  43 \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 27, 2026. ; https://doi.org/10.64898/2026.02.26.708175doi: bioRxiv preprint \n\n 44 \n 45 \nFig. 4 Spatial transcriptomics reveals differences in gene expression between diaphysis and 46 \nepiphysis amputations. A. Hematoxylin and Eosin (H&E) staining of representative longitudinal 47 \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 27, 2026. ; https://doi.org/10.64898/2026.02.26.708175doi: bioRxiv preprint \n\nsections of 3 and 5 dpa diaphysis and epiphysis amputated limbs used for spatial transcriptomics. 48 \nScale bar: 500 µm. B. Spatial view of the 19 clusters identified by Seurat clustering analysis 49 \ncombining spatial dots from diaphysis and epiphysis amputated limbs at 3 and 5 dpa. C. UMAP plot 50 \nand cluster annotation of spatial transcriptomic dots from 3 and 5 dpa diaphysis and epiphysis 51 \namputated limbs. D. UMAP plots showing the expression of the resorption-associated markers Ctsk, 52 \nCtsk-like, Nfatc1 and Acp5. E) UMAP plot showing the contribution of each sample to the UMAP at 3 53 \ndpa (left), 5 dpa (center) and both timepoints (right). F. Spatial expression profiles of Nfatc1, Ctsk and 54 \nCtsk-like in representative spatial transcriptomics sections at 3- and 5 dpa. Expression levels in D and 55 \nF were calculated as Log2 fold expression. CT, connective tissue.  56 \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 27, 2026. ; https://doi.org/10.64898/2026.02.26.708175doi: bioRxiv preprint \n\n 57 \n 58 \nFig. 5 The RANK/RANKL system likely orchestrates tissue-dependent osteoclast-mediated 59 \nskeletal resorption. A. UMAP view of the expression of RANK and RANKL. Expression levels were 60 \ncalculated as Log2 fold expression. B. Spatial expression profiles of Tnfsf11 (RANKL) and Tnfrsf11a 61 \n(RANK) in representative spatial transcriptomics sections at 3 and 5 dpa. Scale bar: 500 µm. C. HCR 62 \nfor Ctsk (green) and RANKL (magenta) at 3, 5 and 7 dpa in representative diaphysis and epiphysis 63 \namputated limbs. Scale bar: 300 µm.   64 \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 27, 2026. ; https://doi.org/10.64898/2026.02.26.708175doi: bioRxiv preprint \n\n 65 \n 66 \nFig. 6 Loc138491483/Ccl24-like is sufficient to ectopically recruit osteoclasts into epiphysis 67 \namputated limbs. A. Top 15 most differentially expressed genes in Nfatc1-expressing spatial 68 \ntranscriptomics spots. B. UMAP plots showing the expression of Loc138491483/Ccl24-like (Loc483). 69 \nC. Dot plot showing expression of Loc483 in the 19 annotated spatial transcriptomics clusters. D. 70 \nSpatial expression profiles of Loc483 representative spatial transcriptomics sections at 3 and 5 dpa. 71 \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 27, 2026. ; https://doi.org/10.64898/2026.02.26.708175doi: bioRxiv preprint \n\nScale bar: 500 µm. E. HCR for Loc483 at 1-, 3- and 5 dpa in diaphysis and epiphysis amputated 72 \nlimbs. Scale bar: 300 µm. F. Time course of osteoclast presence in representative diaphysis and 73 \nepiphysis amputated limbs injected with control plasmid (CAGGs-mCherry) or control plasmid + 74 \nCAGGS-Loc483 at 7-, 9-, and 11 dpa. Scale bar: 1 mm. G. Quantification of the area of Ctsk+ signal 75 \nin CAAGS-Loc483 and control (Ctrl) animals in diaphysis amputations. H. Quantification of the area 76 \nof Ctsk+ signal in CAAGS-Loc483 and control (Ctrl) animals in epiphysis amputations. F shows 77 \nrepresentative images of an experiment with n= 5 animals per condition. For G and H, the graphs 78 \nrepresent the combined results of 2 independent experiments using a total of 5 limbs of 5 different 79 \nanimals per condition. The lines show mean values over time ± sd. *p<0.05 for CAAGS-Loc483 vs 80 \nCtrl electroporated animals (two-way ANOV A with Tukey’s post-hoc test).  81 \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 27, 2026. ; https://doi.org/10.64898/2026.02.26.708175doi: bioRxiv preprint \n\n 82 \nFig. 7 Diaphysis and epiphysis amputations induce changes in the transcriptomic profile of their 83 \nAEC. A. GO analysis for Cellular Component (CC) and Molecular Function (MF) terms in 84 \nupregulated DEGs found in bulk RNA-seq of diaphysis and epiphysis limbs at 3 and 5 dpa. B. Spatial 85 \ntranscriptomic profile of the average expression of upregulated bulk RNA-seq DEGs in epiphysis 86 \namputated limbs at 3 dpa (43 genes) and 5 dpa (36 genes). Scale bar: 500 µm. C. Spatial 87 \ntranscriptomic profile of the average expression of upregulated bulk RNA-seq DEGs in diaphysis 88 \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 27, 2026. ; https://doi.org/10.64898/2026.02.26.708175doi: bioRxiv preprint \n\namputated limbs at 3 dpa (8 genes) and 5 dpa (7 genes). In B and C, expression levels were calculated 89 \nas Log2 average expression. D. Dot plots showing expression of upregulated DEGs found in bulk 90 \nRNA-seq from diaphysis limbs in the 19 annotated spatial transcriptomics clusters at 3 and 5 dpa.  91 \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 27, 2026. ; https://doi.org/10.64898/2026.02.26.708175doi: bioRxiv preprint \n\nSupplementary Figures 92 \n 93 \n 94 \nFig. S1 Ctsk-like is likely a duplication of the Ctsk gene present in the axolotl genome. A)  95 \nLocalization of Ctsk (GCF_040938575.1) and Loc138578972 (Ctsk-like) genes in the axolotl 96 \ngenome. B) Alignment of predicted coding sequences of Ctsk and Ctsk-like, with sequence identity 97 \nand similarity scores. C) Predicted protein sequence alignment of axolotl Ctsk and Ctsk-like. D) 98 \nProtein sequence alignment of human Ctsk (hCtsk) and axolotl Ctsk (axCtsk). E) Protein sequence 99 \nalignment of human Ctsk (hCtsk) and axolotl Ctsk-like (axCtsk-like). E) Table depicting calculated 100 \nsequence identities and sequence similarities between human hCtsk, axCtsk, and axCtsk-like (axCtsk-101 \nL) peptides.  102 \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 27, 2026. ; https://doi.org/10.64898/2026.02.26.708175doi: bioRxiv preprint \n\n 103 \nFig. S2 Cluster annotation and expression of known marker genes in spatial transcriptomics. A. 104 \nDot plot showing the top marker genes in each of the 19 annotated clusters. B. Dot plot showing 105 \nexpression of known marker genes for epithelia, bone, cartilage, endothelium (Endot), blood, immune 106 \nsystem, resorption and blastema (Blast) in the 19 spatial transcriptomics clusters. C. Contribution of 107 \neach sample to each of the 19 annotated clusters.  108 \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 27, 2026. ; https://doi.org/10.64898/2026.02.26.708175doi: bioRxiv preprint \n\n 109 \nFig. S3 RANK and RANKL system are likely involved in osteoclastogenesis during axolotl limb 110 \nregeneration. A. Dot plot showing the expression of Tnfsf11 (RANKL), Tnfrsf11a (RANK), and 111 \nTnfrsf11b (Osteoprotegerin), in spatial clusters. B. HCR for Ctsk (green) and RANK (magenta) at 3 112 \nand 5 dpa in representative diaphysis and epiphysis amputated limbs. Asterisks indicate 113 \nautofluorescence. Scale bar: 300 µm C. Insets of B showing Ctsk (green), RANK (magenta), and 114 \nNfatc1(white). Arrowheads indicate cells that are positive for RANK and Nfatc1, but not Ctsk. Scale 115 \nbar: 50 µm.   116 \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 27, 2026. ; https://doi.org/10.64898/2026.02.26.708175doi: bioRxiv preprint \n\n 117 \nFig. S4 Loc483 is likely not an ortholog to the human CCL24. aLeaves generated neighbor-joining 118 \ntree using the predicted aminoacid sequence of Loc483 gene. Bootstrap support values are shown in 119 \nselected nodes. Blue box contains cartilaginous fish species (Chondrichthyes), green box comprises 120 \namphibian species.  121 \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 27, 2026. ; https://doi.org/10.64898/2026.02.26.708175doi: bioRxiv preprint \n\n 122 \n 123 \nFig. S5 Bulk RNA-seq reveals that Cart and Calc amputations affected other processes than 124 \nosteoclastogenesis A. Top 20 most upregulated differentially expressed genes (DEGs), total number 125 \nof upregulated DEGs, proportion of protein-coding vs non-coding DEGs, and number of DEGs after 126 \nfiltering used for Gene Ontology (GO) analysis in epiphysis amputations at 3 dpa. B. Top 20 most 127 \nupregulated DEGs, total number of upregulated DEGs, proportion of protein-coding vs non-coding 128 \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 27, 2026. ; https://doi.org/10.64898/2026.02.26.708175doi: bioRxiv preprint \n\nDEGs, and number of DEGs after filtering used for GO analysis in epiphysis amputations at 5 dpa. C. 129 \nTop 20 most upregulated DEGs, total number of upregulated DEGs, proportion of protein-coding vs 130 \nnon-coding DEGs, and number of DEGs after filtering used for GO analysis in diaphysis amputations 131 \nat 3 dpa. D. Top 15 most upregulated DEGs, total number of upregulated DEGs, proportion of 132 \nprotein-coding vs pseudo-genes vs non-coding DEGs, and number of DEGs after filtering used for 133 \nGO analysis in diaphysis amputations at 5 dpa. E. Gene expression levels of the osteoclast-associated 134 \ngenes Ctsk, Ctsk-like, Nfatc1 and Acp5 in bulk RNA-seq of intact, and diaphysis and epiphysis 135 \namputated limbs at 3 and 5 dpa. F. Significantly enriched biological process GO terms for diaphysis 136 \nand epiphysis amputated limbs at 3 and 5 dpa. G. Gene expression levels of Loc138491483/Ccl24-like 137 \nin bulk RNA-seq of intact, and diaphysis and epiphysis amputated limbs at 3 and 5 dpa.  138 \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 27, 2026. ; https://doi.org/10.64898/2026.02.26.708175doi: bioRxiv preprint \n\n 139 \n 140 \nFig. S6 AECs of diaphysis and epiphysis amputations exhibit differences in the expression of 141 \nnon-coding genes. A. Expression of representative genes upregulated in epiphysis amputations at 3 142 \ndpa (Psat1) and 5 dpa (Klf17 and Csdn) in bulk RNA-seq. B. Spatial transcriptomic profile of the 143 \naverage expression of non-coding bulk RNA-seq DEGs upregulated in epiphysis amputated limbs at 3 144 \ndpa (6 genes) and 5 dpa (11 genes). C. Spatial transcriptomic profile of the average expression of non-145 \ncoding bulk RNA-seq DEGs upregulated in diaphysis amputated limbs at 3 dpa (4 genes) and 5 dpa (3 146 \ngenes). Expression levels were calculated as Log2 average expression. In A, expression levels were 147 \ncalculated as Log2 average expression, in B and C expression levels were calculated as average Log2 148 \nfold expression. Scale bar: 500 µm. 149 \n 150 \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 27, 2026. ; https://doi.org/10.64898/2026.02.26.708175doi: bioRxiv preprint","source_license":"CC-BY-4.0","license_restricted":false}