Targeting local lymphatics to ameliorate heterotopic ossification via FGFR3-BMPR1a pathway

Acquired heterotopic ossification (HO) is the extraskeletal bone formation after trauma. Various mesenchymal progenitors are reported to participate in ectopic bone formation. Here we induce acquired HO in mice by Achilles tenotomy and observe that conditional knockout (cKO) of fibroblast growth factor receptor 3 (FGFR3) in Col2+ cells promote acquired HO development. Lineage tracing studies reveal that Col2+ cells adopt fate of lymphatic endothelial cells (LECs) instead of chondrocytes or osteoblasts during HO development. FGFR3 cKO in Prox1+ LECs causes even more aggravated HO formation. We further demonstrate that FGFR3 deficiency in LECs leads to decreased local lymphatic formation in a BMPR1a-pSmad1/5-dependent manner, which exacerbates inflammatory levels in the repaired tendon. Local administration of FGF9 in Matrigel inhibits heterotopic bone formation, which is dependent on FGFR3 expression in LECs. Here we uncover Col2+ lineage cells as an origin of lymphatic endothelium, which regulates local inflammatory microenvironment after trauma and thus influences HO development via FGFR3-BMPR1a pathway. Activation of FGFR3 in LECs may be a therapeutic strategy to inhibit acquired HO formation via increasing local lymphangiogenesis.

6. The authors report that FGFR3 is involved in lymphatic migration and proliferation. However, the data do not exclude that FGFR3 is also involved in the differentiation from COL2-positive mesenchymal cells. Fig. 4, the authors claim that the reporter labeling indicates a high degree of lymphatic Fgfr3 deletion. However, reporter activity does not necessarily reflect gene deletion efficiency. To demonstrate this, the authors need to perform qPCR or immunoblot analyses.

In supplementary
Reference: Álvarez-Aznar, A et al., Tamoxifen-independent recombination of reporter genes limits lineage tracing and mosaic analysis using CreERT2 lines. Transgenic Research 2019 Reviewer #2: Remarks to the Author: In this study, authors explored the role FGF signaling in acquired HO development. Their lineage tracing experiment showed that Col2+ cells are adopted fate of lymphatic endothelial cells during HO development. FGFR3 cKO in Prox1-positive LECs increased HO formation. FGFR3 deficiency in LECs resulted in decreased local lymphatic formation with increased inflammatory levels. Local administration of FGF9 in Matrigel inhibited heterotopic bone formation. This study revealed Col2+ lineage cells as a novel origin of lymphatic endothelium in HO. This is an interesting novel finding. The experiments in general are well designed and executed. The data are convincing. I have the following comments to improve the manuscript: 1. Heterotopic ossification is a very complex process involved in many factors. HO could occur only if all conditions are satisfied. This is why there are so many factors were identified in inhibition of HO. In the introduction, it did not describe the overall scheme of HO development and the potential role of FGF signaling in the process. For example, TGFbeta levels are significantly increased at both initial phase and late stage as well. And it is also critical for chondrogenesis and progression of HO. The information is missing in the Introduction.
2. There is no evidence to show the process of acquired heterotopic ossification is different from the other types of HO. AHO is already used for acute hematogenous osteomyelitis. AHO for acquired HO used here causes confusion in the field and literature.
3. Authors claim "we still have limited knowledge about the cellular and molecular mechanism of AHO development", but the manuscript did not even review the current understanding of four different stages of HO development and did not discuss their finding of FGF signaling in HO relative to the four stages of HO development.
4. BMP signaling is known to determine cell lineage fate. What is the function of BMP signaling in fate of lymphatic endothelial cells under normal physiology? 5. "Sustained high-level inflammation after trauma is related to impaired local lymphatic drainage in FGFR3-deficient mice, which may aggravate AHO development". Apparently, increase of local inflammatory levels subsequently in elevation of AHO is an indirect effect. Inflammatory is at early stage of HO, which promotes TGFbeta level for chondrogenesis. The authors should examine whether increase of TGFbeta activity for HO development.
6. The Diagram and Discussion should include overall outline of HO development and relative position of FGF signaling in LECs in HO. 7. The overall writing about HO and interpretation of the results need to be improved Reviewer #3: Remarks to the Author: The manuscript by Zhang et al. introduces a compelling relationship between local lymphangiogenesis and acquired heterotopic ossification (AHO) in various sophisticated mouse models that underwent Achilles tenotomy. Specifically, the authors identified Col2+ resident progenitors of the peritendineum as a potential novel source of lymphatic endothelial cell (LEC) renewal post-tenotomy. The capacity for these progenitors to promote lymphangiogenesis posttenotomy was directly associated with the severity of AHO development in a FGFR3 dependent manner. Conditional knockout (cKO) of FGFR3 in Col2+ progenitors and Prox1+ LECs led to increased AHO formation post-tenotomy, and this pathologic change was associated with an increase in BMPR1a and p-Smad1/5. Moreover, cKO of BMPR1a in these models reversed this phenotype. The authors propose that reduced lymphatic function promotes local inflammation that eventually dysregulates the FGFR3-BMPR1a signaling pathway leading to AHO formation, and thus targeting FGFR3 may promote lymphangiogenesis to ameliorate disease. While the manuscript presents a convincing story with data from both mice and humans to support their claims, there are concerns about some of the data and interpretation of some results that need to be addressed. There are also some minor concerns that the authors should consider. Major Comments: 1. The current presentation of the images makes colocalization of markers difficult to assess. For example, the authors write, "Immunostaining revealed abundant expressions of canonical LEC markers LYVE1 and VEGFR3 in tdTomato labeled Col2+ lineage cells…" .
However, in the associated Figures 2g,h the colocalization of immunostain (green) and lineage trace (red) is questionable. As the authors point out, lymphatic vessels (LVs) were only present within the tendon after injury, so if Col2+ cells are truly the predominant progenitor for LECs in these circumstances, one would expect all (or most, dependent on tamoxifen efficiency) to be Col2-derived. Instead, in Figure 2h the VEGFR3 immunostaining appears to be mostly independent of the lineage traced cells besides a select few colocalized (yellow) cells. Moreover, for Figure 2g the presence of Col2-derived LYVE1+ cells is difficult to interpret as the lineage traced red fluorescence appears to only be present in the nuclei. How did the authors verify that these nuclei are specific to the LEC, and not the nuclei for the presumably directly adjacent lymphatic muscle cells? Is this lineage tracing mouse model expected to only show nuclear fluorescence, as it appears in other images (i.e. Figure 2h) to be non-specific to the nucleus and cytoplasm? An explanation for the lacking colocalization of many cells, or an alternative presentation of the fluorescence (i.e. split channels with a composite image) not just in Figure 2, but throughout the manuscript, is needed. 2. As presented, the conclusions of Figure 3 are misleading. The authors write, "Collectively, FGFR3 cKO in LECs tremendously promoted AHO development, which further supports that the aggravated AHO formation in FGFR3Col2 mice is strongly related to the disturbed LECs derived from Col2+ cells in the tendon after trauma" (Lines 293 -296). While there may be a connection associated with the similar reduction in lymphangiogenesis when FGFR3 is deleted in both the proposed Col2-derived LEC progenitors and Prox1+ LECs, a direct mechanistic relationship between these two cells cannot be made as presented. The strongest conclusion that can be made is that there appears to be a relationship between lymphangiogenesis and AHO development represented in both models. Additional studies are needed to demonstrate that the associated cellular changes following FGFR3 deletion mediate the same lymphangiogenic disruption in both affected Col2+ progenitors and Prox1+ LECs. For instance, do Col2-derived Prox1+ LECs demonstrate continued disruption in the FGFR3-BMPR1a pathway by protein expression in FGFR3f/f-Col2tomato compared to Col2tomato control mice? It may be possible that the reduced lymphangiogenesis in FGFR3Col2 animals functions by a similar, but unrelated mechanism compared to the FGFR3Prox1 model in which Col2+ progenitors are unable to proliferate and differentiate into LECs without FGFR3, and any lymphangiogenesis in FGFR3Col2 animals is due to ineffective deletion of FGFR3 in certain Col2+ progenitors. To make the claims as written, further experiments are needed to confirm that LECs derived from Col2+ progenitors in the FGFR3Col2 model indeed have disturbed FGFR3-BMPR1a signaling leading to the reduced lymphangiogenesis via mechanisms similar to the FGFR3Prox1 construct. 3. The authors write, "Lineage tracing of FGFR3f/f; Prox1CreERT2; R26RtdTomato mice (FGFR3f/f-Prox1tomato) revealed that almost all LYVE1+ LECs in the tendon were labeled by tdTomato at 8 weeks post tenotomy, indicating a high efficiency of FGFR3 deletion in LECs in the repaired Achilles tendon" (Lines 277 -280). Similar to Comment 1, the associated Supplementary Figure 4 seems to indicate very little colocalization between Prox1 lineage traced (red) and LYVE1 immunostained (green) cells. This finding brings into question the reliability of either the lineage tracing model or the immunostaining in these experiments, as Prox1 and LYVE1 should be colocalized as canonical LEC markers. Moreover, the title to Supplementary Figure 4, "Prox1+ lineage traced cells contribute to local LECs in the repaired tendon" is an inaccurate representation as LECs themselves are Prox1+. 4. LYVE1 is also a known marker of certain M2 polarized macrophages, especially surrounding smooth muscle cells to promote regulation of collagen content (PMID: 30054204). The authors demonstrate that under conditions where LYVE1 staining increases (depicted in the manuscript as equivalent to LECs, and used as a measure of LV area and length) that the number of M1 polarized macrophages decreases (Figures 4,Supp 5,5,6,Supp 7). Concurrent F4/80 or Prox1 staining ought to be performed with LYVE1 to confirm that LYVE1+ cells are truly representing LECs and the results are not confounded by LYVE1+ M2 polarized macrophages. 5. The near infrared indocyanine green (NIR-ICG) clearance as depicted in Figure 4j does not seem to match the quantified clearance results in Figure 4k. With the current images, it looks as if FGFR3Prox1 actually has greater or similar clearance compared to FGFR3f/f controls. A clarification of the analysis method depicted in the figures and more representative images in Figure 4j are needed. 6. As presented, the Western blot in Figure 5e is questionable. It seems as if siFGFR3 #1 had relatively ineffective knockdown of FGFR3, but the pSmad1/5 levels appear increased similar to the other siFGFR3 lanes compared to the control. An explanation for this discrepancy is warranted. In addition, the authors write, "FGFR3 knockdown in mouse LEC line led to upregulated BMPR1a…" (Lines 434 -435) however, as presented, it is difficult to see a noticeable increase in BMPR1a in the siFGFR3 conditions on the blot. Relative quantification of the protein levels would be helpful for interpreting these results. 7. Essential controls for the mouse models used in this study are missing. To validate accurate representation of the lineage tracing, Cre-negative and Cre-positive without tamoxifen induction (PMID 31641921) controls are necessary. 8. Clarification on the methods for immunostaining image analysis are needed. Were exact cell counts determined using an automated process or counted manually? How were the regions of interest for analysis determined in a representative and unbiased manner? Were the observers blinded to the conditions? "Image J software was used for quantifications. 5-8 independent images of 3-5 sequential sections in each sample were used for quantitative analysis" (Lines 733 -735) is not sufficient to allow repeatability of this study. 9. For the human specimen collection, the methods note that, "HO specimens were collected from male patients who had previously sustained a femur or elbow fracture…" (Lines 786 -787). In Supplementary Figure 5, are the human specimen data pooled to include both femur and elbow fractures? Is it expected for AHO formation in these two different conditions to behave similarly? What was the breakdown of subjects with elbow versus femur injuries that were assessed at the osteogenesis versus maturation stages? 10. Additional information on the mice used for the study is needed. In connection with Comment 9 in which it was noted for human subjects that only males were used, were both male and female mice used for this study? If so, were the sexes distributed evenly between the groups? Moreover, Jackson Laboratory stock numbers ought to be provided for the animals used in this study. For instance, according to Jackson Laboratories, the only Prx1-CreERT2 animal available is also tagged with a GFP (Prx1CreER-GFP; Stock No 029211). Was a different strain used for this study? If not, how was the Prx1-driven GFP fluorescence controlled in the analysis in Supplementary Figure 3, especially since the antibodies assessed were also labeled green? In addition, there are many Rosa26-tdTomato reporters offered by Jackson Laboratories, so which strain was used? This is also important to acknowledge given the differences in basal CreERT2 activity noted in the source for Comment 7. Minor Comments: 1. There are potentially misleading comments that do not match the data as presented. For example, in reference to Supplementary Figure 5f, the authors write, "immunostaining revealed significantly increased numbers of F4/80+iNOS+ inflammatory macrophages in HO lesions at maturation stage relative to osteogenesis stage" (Lines 372 -373). However, the figure demonstrates no significant difference in the % F4/80+iNOS+ cells relative to total F4/80+ cells between the two stages. The authors should ensure that all written explanations accurately depict the data as presented. 3. The manuscript ought to be thoroughly proofread for grammar and typos.
We would like to submit the revised manuscript entitled "Targeting local lymphatics to ameliorate heterotopic ossification via FGFR3-BMPR1a pathway" (NCOMMS-20-16420-T). We have addressed all reviewers' questions in the revised manuscript and provided 'point-to-point' replies. All changes were marked in red in the manuscript. We appreciate the opportunity allowing us to revise our manuscript.

Review 1
In the manuscript "Targeting local lymphatics to ameliorate heterotopic ossification via FGFR3-BMPR1a pathway" Zhang et al. study mechanisms that underlie acquired heterotopic ossification (AHO). As a result of their study, the authors propose FGF signaling as a promising therapeutic target to treat the disorder. Overall, the manuscript is preliminary and confusing and the presented data are not convincing. Especially, the lineage-tracing studies are of concern given the known leakiness of the genetic reporter used.
Response: Thank you very much for your valuable comments. We checked our manuscript according to your suggestions one by one. We carried out related experiments and hope our supplemental results or explanations might address your concerns.
Other comments: 1. All the histology and immunofluorescence (IF) data are not properly labeled, making it difficult to evaluate the data. Also, the quality of IF images must be improved. 3. It has been reported that the R26tdTom reporter mouse shows Tamoxifen-independent recombination (Álvarez-Aznar, A et al., 2019). Consequently, the authors need to examine this possibility before concluding that LECs are of mesenchymal-origin.

Response:
We appreciate the reviewer's important suggestion. According to the referred paper 1 , we tested tamoxifen-independent Cre recombination of reporter using tomato and Col2 tomato mice without tamoxifen induction as controls (S Fig3b). We did not find tomato positive cell in the repaired Achilles tendon of these control mice at 4 weeks post surgery, though co-staining of LYVE1 revealed that lymphatics were already formed in these tendons. Furthermore, as mentioned in this reference 1 , mTmG reporter mice are more suitable for lineage tracing as they have lower recombination susceptibility.
Therefore, we also used Col2 mTmG mice to further confirm the LEC identity of Col2-derived cells in the tendon after surgery in vivo and in vitro. GFP Fig5a,b). These results demonstrated that Col2 + cells adopted the fate of LECs rather than macrophages in the tendon after injury.
Meanwhile, we tried to sort Col2-derived cells in the repaired tendon of Col2 mTmG mice.
However, the amount of primary GFP-labeled Col2 + lineage cells is too low to do sorting and transcriptome profiling. We also tried to amplify these primary GFP + cells before sorting, but these GFP + cells were amplified in a much slower rate than GFP cells in vitro, even though we used endothelial cell medium (ECM, ScienCell). Since in vitro evidence can help further confirm the LEC identity of Col2 + lineage cells in the repaired Achilles tendon, we isolated the primary cells in the repaired tendon of Col2 mTmG mice and confirmed that GFP + Col2-derived cells were immunostained by LEC markers including LYVE1, Prox1 and VEGFR3 (S Fig4a). We agree that the transcriptome profiling of Col2 + lineage cells is an important study, and we will carry it out using emerging new approaches in the future research. Thank you again for your important suggestion.
6. The authors report that FGFR3 is involved in lymphatic migration and proliferation. However, the data do not exclude that FGFR3 is also involved in the differentiation from COL2-positive mesenchymal cells.
Response: Thank you for your suggestion. Previous findings reported that FGFR3 is essential for lymphangiogenesis by regulating LEC proliferation as well as migration 2,3 . It was also reported that FGFR3 is an initial target of Prox1, which is known as a master regulator inducing lymphatic differentiation 2 . Therefore, it was speculated that FGFR3 1. Heterotopic ossification is a very complex process involved in many factors. HO could occur only if all conditions are satisfied. This is why there are so many factors were identified in inhibition of HO. In the introduction, it did not describe the overall scheme of HO development and the potential role of FGF signaling in the process. For example, TGFbeta levels are significantly increased at both initial phase and late stage as well. And it is also critical for chondrogenesis and progression of HO. The information is missing in the Introduction. inducing type H vessel formation coupled with osteogenesis. Locally increased blood vessels transport more oxygen, nutrients and minerals for HO development 5 and 691-693 (TGF-β produced by macrophages has been found to contribute to HO development and TGF-β signaling remains activated in the osteogenesis stage of HO before a reduction till 15 weeks after injury 5 ). The potential role of FGF signaling in HO was mentioned in lines 93-104 (Fibroblast growth factor (FGF) signaling plays an essential role in skeletal development 6 . Activation mutations of fibroblast growth factor receptor 3 (FGFR3) in human cause chondrodysplasia including achondroplasia, hypochondroplasia as well as thanatophoric dysplasia through inhibiting chondrocyte proliferation and differentiation 7 . Meanwhile, FGFR3 also plays a vital role in the regulation of lymphatic formation. FGFR3 is expressed in human and mouse lymphatic endothelial cells (LECs) and is essential for LEC proliferation and migration 2 . 9-cis retinoic acid (9-cisRA) promotes LEC proliferation, migration and tube formation via activating FGF signaling. 9-cisRA-induced proliferation of LECs is coupled with increased FGFR3 expression, which is suppressed by soluble FGFR3 recombinant protein that sequesters FGF ligands 3 . All these findings suggest the possible involvement of FGFR3 in acquired HO development, although the accurate role and detailed underlying mechanisms remain to be clarified). 2. There is no evidence to show the process of acquired heterotopic ossification is different from the other types of HO. AHO is already used for acute hematogenous osteomyelitis. AHO for acquired HO used here causes confusion in the field and literature.

Response:
Thank you for your suggestion. We replaced 'AHO' with 'acquired HO' or 'HO' to avoid confusion.
3. Authors claim "we still have limited knowledge about the cellular and molecular mechanism of AHO development", but the manuscript did not even review the current understanding of four different stages of HO development and did not discuss their finding of FGF signaling in HO relative to the four stages of HO development.

The overall writing about HO and interpretation of the results need to be improved
Response: Thank you for your suggestion. We thoroughly reviewed the whole manuscript and improved the interpretation of the results. For example, improved descriptions highlighted in red are shown in the results of our manuscript. Further discussions about the influence of FGFR3 on LEC differentiation were added in Lines 709-719.

Reviewer 3
The manuscript by Zhang et al. introduces a compelling relationship between local lymphangiogenesis and acquired heterotopic ossification (AHO) in various sophisticated mouse models that underwent Achilles tenotomy. Specifically, the authors identified Col2+ resident progenitors of the peritendineum as a potential novel source of lymphatic endothelial cell (LEC) renewal post-tenotomy. The capacity for these progenitors to promote lymphangiogenesis post-tenotomy was directly associated with the severity of AHO development in a FGFR3 dependent manner. Conditional knockout (cKO) of FGFR3 in Col2+ progenitors and Prox1+ LECs led to increased AHO formation post-tenotomy, and this pathologic change was associated with an increase in BMPR1a and p-Smad1/5. Moreover, cKO of BMPR1a in these models reversed this phenotype. The authors propose that reduced lymphatic function promotes local inflammation that eventually dysregulates the FGFR3-BMPR1a signaling pathway leading to AHO formation, and thus targeting FGFR3 may promote lymphangiogenesis to ameliorate disease. While the manuscript presents a convincing story with data from both mice and humans to support their claims, there are concerns about some of the data and interpretation of some results that need to be addressed. There are also some minor concerns that the authors should consider.
Major Comments: 1. The current presentation of the images makes colocalization of markers difficult to assess. For example, the authors write, "Immunostaining revealed abundant expressions of canonical LEC markers LYVE1 and VEGFR3 in tdTomato labeled Col2+ lineage cells…" (Lines 200 -202). However, in the associated Figures 2g,h the colocalization of immunostain (green) and lineage trace (red) is questionable. As the authors point out, lymphatic vessels (LVs) were only present within the tendon after injury, so if Col2+ cells are truly the predominant progenitor for LECs in these circumstances, one would expect all (or most, dependent on tamoxifen efficiency) to be Col2-derived. Instead, in Figure 2h the VEGFR3 immunostaining appears to be mostly independent of the lineage traced cells besides a select few colocalized (yellow) cells. Moreover, for Figure 2g the presence of Col2-derived LYVE1+ cells is difficult to interpret as the lineage traced red fluorescence appears to only be present in the nuclei. How did the authors verify that these nuclei are specific to the LEC, and not the nuclei for the presumably directly adjacent lymphatic muscle cells? Is this lineage tracing mouse model expected to only show nuclear fluorescence, as it appears in other images (i.e. Figure 2h) to be non-specific to the nucleus and cytoplasm? An explanation for the lacking colocalization of many cells, or an alternative presentation of the fluorescence (i.e. split channels with a composite image) not just in Figure 2, but throughout the manuscript, is needed. Moreover, we also stained the primary cells isolated from repaired Achilles tendon of Col2 mTmG mice with Prox1, LYVE1 and VEGFR3 in vitro (S Fig4a) and found that Col2 + lineage cells in the tendon after surgery were labeled by these LEC markers.
2. As presented, the conclusions of Figure 3 are misleading. The authors write, "Collectively, FGFR3 cKO in LECs tremendously promoted AHO development, which further supports that the aggravated AHO formation in FGFR3Col2 mice is strongly related to the disturbed LECs derived from Col2+ cells in the tendon after trauma" (Lines 293 -296). While there may be a connection associated with the similar reduction in lymphangiogenesis when FGFR3 is deleted in both the proposed Col2-derived LEC progenitors and Prox1+ LECs, a direct mechanistic relationship between these two cells cannot be made as presented. The strongest conclusion that can be made is that there appears to be a relationship between lymphangiogenesis and AHO development represented in both models. Additional studies are needed to demonstrate that the associated cellular changes following FGFR3 deletion mediate the same lymphangiogenic disruption in both affected Col2+ progenitors and Prox1+ LECs. 5. The near infrared indocyanine green (NIR-ICG) clearance as depicted in Figure 4j does not seem to match the quantified clearance results in Figure 4k. With the current images, it looks as if FGFR3Prox1 actually has greater or similar clearance compared to FGFR3f/f controls. A clarification of the analysis method depicted in the figures and more representative images in Figure 4j are needed.
Response: Thank you for your suggestion. We replaced the previous NIR-ICG image of FGFR3 Prox1 mice with a more representative image (Fig4j). A clarification of NIR-ICG analysis was explained in the methods and materials as follows: The signal intensity in the footpad was recorded immediately after ICG injection as the initial signal intensity. ICG imaging was collected again 24 hours after ICG injection. Conditions including exposure time, focus and position of the mouse hindlimbs need to be consistent for all imaging and overexposure needs to be avoided. The images were analyzed using Evolution-Capt v18.02 software. In brief, regions of interest (ROI) defining the injection site of the footpad was identified. ICG clearance was quantified as the percentage of reduced ICG signal intensity in the footpad 24 hours post injection relative to the initial ROI signal intensity (Lines 843-850).
6. As presented, the Western blot in Figure 5e is questionable. It seems as if siFGFR3 #1 had relatively ineffective knockdown of FGFR3, but the pSmad1/5 levels appear increased similar to the other siFGFR3 lanes compared to the control. An explanation for this discrepancy is warranted. In addition, the authors write, "FGFR3 knockdown in mouse LEC line led to upregulated BMPR1a…" (Lines 434 -435) however, as presented, it is difficult to see a noticeable increase in BMPR1a in the siFGFR3 conditions on the blot. Relative quantification of the protein levels would be helpful for interpreting these results.
Response: Thank you for your suggestion. The seemingly inconsistency of the western blot result might be due to the variable FGFR3-knockdown efficiency among these three siRNAs. To obtain a more stable and efficient knockdown of FGFR3, siFGFR3#1, #2 and #3 were pooled together for a combined FGFR3 knockdown in mLEC line. As shown in Fig5g, FGFR3 level was evidently knocked down and BMPR1a/pSmad1/5 levels were remarkably upregulated.
7. Essential controls for the mouse models used in this study are missing. To validate accurate representation of the lineage tracing, Cre-negative and Cre-positive without tamoxifen induction (PMID 31641921) controls are necessary.
Response: Thank you for your suggestion. According to the referred paper 1 , tamoxifen-independent Cre recombination of reporter mice was tested using tomato and Col2 tomato mice without tamoxifen induction as controls (S Fig3b). We did not find tomato positive cell in the repaired Achilles tendon of these control mice at 4 weeks post surgery, though co-staining of LYVE1 revealed that lymphatics were already formed in these tendons. Furthermore, as mentioned in this reference