Lymphatic vessels in bone support regeneration after injury

Blood and lymphatic vessels form a versatile transport network and provide inductive signals to regulate tissue-specific functions. Blood vessels in bone regulate osteogenesis and hematopoiesis, but current dogma suggests that bone lacks lymphatic vessels. Here, by combining high-resolution light-sheet imaging and cell-specific mouse genetics, we demonstrate presence of lymphatic vessels in mouse and human bones. We find that lymphatic vessels in bone expand during genotoxic stress. VEGF-C/VEGFR-3 signaling and genotoxic stress-induced IL6 drive lymphangiogenesis in bones. During lymphangiogenesis, secretion of CXCL12 from proliferating lymphatic endothelial cells is critical for hematopoietic and bone regeneration. Moreover, lymphangiocrine CXCL12 triggers expansion of mature Myh11+ CXCR4+ pericytes, which differentiate into bone cells and contribute to bone and hematopoietic regeneration. In aged animals, such expansion of lymphatic vessels and Myh11-positive cells in response to genotoxic stress is impaired. These data suggest lymphangiogenesis as a therapeutic avenue to stimulate hematopoietic and bone regeneration.


INTRODUCTION
Vasculature, a key component of the bone marrow microenvironment, provides signals for the maintenance and proliferation of hematopoietic stem and progenitor cells in bone. [1][2][3][4] Vasculature also regulates the differentiation of perivascular mesenchymal stem cells to generate bone cells. [5][6][7][8][9] However, due to technical difficulties with imaging calcified tissues, the organization of blood vessels in bone has, until recently, remained elusive. Advancements in imaging thick slices of murine bones have now provided insights into the heterogeneity of blood vessels in bone. [10][11][12] Together with functional studies, this work demonstrated that the skeleton and the bone marrow endothelium form a functional unit that is important during development, homeostasis, and aging. [12][13][14] Specifically, a specialized capillary termed ''type H'' regulates both angiogenesis and osteogenesis and couples the two processes within the bone. 12 Blood vessels also provide distinct niches for the maintenance and proliferation of hematopoietic stem cells (HSCs). 1,2 The lymphatic system regulates fluid homeostasis, waste clearance, and immune responses. [15][16][17] Until recently, it was believed that certain tissues, such as the brain, eye, and bone, lack lymphatics. However, recent work has revealed the presence of lymphatic vessels in the dura mater of the mouse brain and the spinal vertebral column. [18][19][20] Further, the eye tissue hitherto believed to lack lymphatic circulation harbors the Schlemm's canal, which is analogous to lymphatics. 21,22 Nevertheless, the current dogma is that the bone and bone marrow lack lymphatic vessels and that lymphatic vessel growth in the bone may be detrimental, as seen in Gorham-Stout disease, a rare bone disorder characterized by the improper growth of lymphatic vessels in bones. 23,24 Some historic studies have pointed to a different conclusion. For example, earlier investigations indicated that Indian ink injected into long bones reaches the lymph nodes. 25,26 Moreover, when injected into the bone marrow, high-molecular-weight molecules such as ferritin and horseradish peroxidase are able to reach the periosteal surface of the bone, 27-30 suggesting a path connecting the two regions.
The identification of lymphatic endothelial cell (LEC) markers, such as lymphatic vessel endothelial hyaluronan receptor 1 (LYVE1), prospero-related homeobox 1 (PROX1), and podoplanin, has accelerated the characterization of lymphatic vessels in several organs over the last decade. 31-37 However, a study using LYVE1 and podoplanin failed to identify lymphatic vessels in 2D analysis of thin human bone sections. 30 Immunolabeling and 3D imaging of intact skeletal tissues is technically challenging due to their calcified nature. The current methods for clearing and immunolabeling whole and intact bones are limited 25,26,[38][39][40] and time consuming or generate low-resolution data.
Here, to more definitively investigate whether lymphatic vessels exist within bones, we devised a method to immunolabel and image intact bones at high resolution on a light-sheet microscopy platform. We applied this light-sheet imaging and functional genetics to bone tissues. We found that lymphatic vessels do exist in bone and drive bone and hematopoietic regeneration.

Light-sheet imaging of intact skeletal elements reveals lymphatic vessels in bones
The current immunolabeling methods for intact skeletal elements are limited and time consuming and generate low-resolution images. We modified the existing methods and developed a pipeline for efficient clearing and immunolabeling of intact skeletal tissues that enable rapid single-cell resolution and quantitative panoptic 3D light-sheet imaging ( Figures 1A-1F; Video S1). The method enables ultrafast immunolabeling and clearing of calcified tissues within 3.5 days and will accelerate discoveries compared to other imaging pipelines.
The addition of collagenase digestion step after fixation and decalcification allows rapid and through antibody penetration. In line with this, immunostained bones showed antibody penetration and staining deep throughout the stained bones and calcified tissues (Figures S1A and S1B), while the negative controls without primary antibodies lacked staining (Figures S1C and S1D). Such an approach allowed visualization of the bone marrow microenvironment at a high resolution and in 3D across diverse murine whole bones and hard tissues (Figures 1A-1F and S2A-S2D) and human bone biopsies (Figure 1G).
To determine whether bones contained lymphatic vessels, we performed immunolabeling with several different lymphatic vessel markers. We observed LYVE1-positive lymphatic vessels on cleared murine whole bones ( Figures 1H-1L, S3A, and S3B; Video S2), both in the cortical regions and the bone marrow cavity, but more abundant in the cortical regions ( Figures S3C-S3F). Both whole-bone 3D rendering and single z-plane projections of the light-sheet images revealed LYVE1-positive lymphatic vessels ( Figures 1H-1L, S3A, and S3B). Indeed, we observed LYVE1-positive lymphatic vessels across different murine bones, including sternum, vertebral column, costal bones, femur, calvarium, and hip bones (Figures 2A-2C, S3E, S3F, and S4A; Video S2). LYVE1, used together with CD102 also called ICAM2, a member of an intracellular adhesion molecule family, is expressed by blood vessels but not by lymphatic vessels (Figures 1J, 1L, and 2A-2C). Another lymphatic endothelial marker, PROX1, also displayed lymphatic vessels in bones ( Figures 2D and S4B), and qPCR demonstrated Prox1 expression across diverse murine bones ( Figure S4C). qPCR analysis with Lyve1, Prox1, and Pdpn expressions in bones processed for light-sheet imaging did not detect Lyve1, Prox1, and Pdpn expressions in the outer surface of bone ( Figure S4D). For further analysis and identification of lymphatic vessels in bone, we used Evans blue dye, a functional tool to map lymphatic vessels. 12,41,42 Evans blue, a vital dye, binds with high-affinity tissue protein and is selectively and exclusively absorbed from interstitial space by initial lymphatic vessels. 12,[43][44][45] Evans blue was injected subcutaneously into the inner leg, medial to the tail and footpad. 12,41,43,46,47 Imaging of tibial bones showed the uptake of Evans blue, and its accumulation in tibial bones confirmed the presence of lymphatic vessels (Figure 2E). In addition to immunostainings, light-sheet imaging of intact bones from the lymphatic vessel reporter mouse line LYVE1-EGFP 48 also detected LYVE1-positive lymphatic vessels in bones (Figure 2F; Video S2). Finally, immunostaining with the LEC marker podoplanin 49 detected lymphatic vessel structures in bones (Figure 2G).
LECs were also detected in the single-cell suspension of bones using LYVE1 and podoplanin by flow cytometry. Since podoplanin is also expressed by a subset of mesenchymal cells and LYVE1 by a subset of macrophages, we identified doublepositive LYVE1 + podoplanin + cells ( Figure S4E). In addition, we confirmed the presence of LECs in single-cell suspension of bones as CD45 À LYVE1 + cells, this confirmed the presence of Figure 1. Light-sheet imaging of intact skeletal elements identifies lymphatic vessels in whole bones (A) Images of tibiae prior to and post-tissue clearing. 3D images were acquired on a light-sheet microscopy platform post-clearing with a high-magnification inset; labeled with the nuclear stain DAPI, immunostained with CD102 and a-SMA. (B) Murine knee joint prior to and after tissue clearing. 3D images of the knee joint; labeled with DAPI, a-SMA, and endomucin.  Article LECs in different bones including sternum, femur, and tibia ( Figure S4F). Furthermore, immunostaining with macrophage marker, F4/80, in intact bones confirmed that the LYVE1-positive vascular structures are negative for F4/80 ( Figures 1J and 1K). The density of lymphatic vessels in cortical bones was higher compared to in the bone marrow, and the diameter of these vessels was $19 mm ( Figures 2H and 2I).
To determine whether these findings were relevant to the human context, we also examined the presence of lymphatic vessels in human bones by performing enzyme-linked immunosorbent assay (ELISA) for PROX1 and immunostaining with LYVE1 on human bone biopsies. PROX1 expression was detectable in all the human bone biopsies analyzed but was undetectable in primary cartilage cells cultured from these bone samples ( Figure 2J). 3D imaging of LYVE1 immunostaining also showed lymphatic vessels in cleared human bones ( Figure 2K). Further, analysis of our single-cell sequencing data of human bone marrow demonstrated the presence of a small fraction of LECs ( Figures 2L, S4G, and S4H). Thus, our imaging data, combined with other analyses, establish the presence of lymphatic endothelial cells within bone tissue in both human and mice.
Lymphatic vessels in bones expand during genotoxic stress Our own and other studies previously indicated that blood vessels in bones are highly susceptible and severely affected during radiation and myeloablation. 6,[50][51][52] Specifically, the bone marrow sinusoidal endothelium dilates and type H endothelial cells expand, while the number of sinusoidal endothelial cells declines upon radiation injury. 11 To determine whether and how lymphatic vessels are affected, we analyzed the impact of radiation injury and myeloablation on the bone lymphatic endothelium ( Figure 3A). To do so, we lethally irradiated mice, followed by transplantation of one million bone marrow cells. Bones from irradiated mice exhibited a dramatic expansion of lymphatic vessels. This was revealed by light-sheet imaging of LYVE1-positive lymphatic vessels and quantification of PROX1 by ELISA ( Figures 3B, 3C, and S5A-S5F). Based on immunostaining by the proliferation marker Ki67, it seems likely that the observed post-radiation expansion of lymphatic vessels involves LEC proliferation ( Figure 3B). Furthermore, analysis of lymphatic vessel density showed that vessel expansion in the bone peaked at 15 days post-radiation, subsequently returning to a normal density by 55 days post-radiation ( Figure S5G).
Finally, to confirm the specificity of these observations, we examined mice treated with a lymphangiogenesis inhibitor. VEGFR3 is a known regulator of lymphangiogenesis that is selectively inhibited by SAR131675. 11,53 When we treated irradiated mice with SAR131675, we no longer observed expansion of bone lymphatic vessels, as demonstrated via imaging of LYVE1-expressing lymphatic vessels and ELISA quantification of PROX1 concentration ( Figures 3A-3C). Analogous to radiation, chemotherapy-driven genotoxic stress induced by 5-fluorouracil (5-FU) treatment also led to the expansion of lymphatic vessels in bones as demonstrated by imaging of LYVE1-expressing vessels and quantification of PROX1 concentration by ELISA ( Figures 3D  and S5C). Further, when we treated 5-FU treated mice with the lymphangiogenesis inhibitor SAR131675, we no longer observed expansion of bone lymphatic vessels, as demonstrated via imaging of LYVE1-expressing lymphatic vessels and ELISA quantification of PROX1 concentration ( Figure 3D). Together, these findings demonstrate that genotoxic stress leads to expansion of lymphatic vessels in the bone.
Lymphatic vessels in bones support hematopoietic regeneration Since genotoxic stress impacts vasculature, we examined the relationship between these two processes. We first asked whether the ability of mice to undergo lymphangiogenesis impacted hematopoietic regeneration. We compared hematopoietic reconstitution in irradiated mice with or without treatment with SAR131675. Analysis of cellular frequencies after radiation showed a decline in the number of bone marrow cells, lineage (Lin) À Sca-1 + c-Kit + (LSK) cells, and HSCs in SAR131675-treated mice ( Figure 3E). We confirmed the decreases in HSC frequency by competitive secondary transplantation of bone marrow cells. Bone marrow cells from radiation and SAR131675-treated mice showed a significantly reduced reconstituting activity compared to bone marrow cells from radiation-only treated mice ( Figure 3F). To interrogate whether the observed loss of cellularity and    (G) 3D images showing LYVE1 or Tomato and DAPI from Cre À iDTA and Cre + iDTA; Prox1 Cre ER T2 X R26-td Tomato X iDTA mice subjected to radiation. Quantification of lymphatic vessel density in bones from Cre À iDTA and Cre + iDTA mice subjected to radiation (n = 5). ELISA quantification of PROX1 concentration (n = 6). (H) Bone marrow cellularity, number of LSK cells, and HSCs in bones from Cre À iDTA and Cre + iDTA, Prox1 Cre ER T2 X iDTA mice were analyzed at 4, 14, and 29 days post-radiation and BM transplantation (n = 6). (I) 1 3 10 6 donor BM cells from Cre À iDTA or Cre + iDTA, Prox1 Cre ER T2 X iDTA primary donor mice (as detailed in H) were transplanted into secondary wild-type recipient mice. Overall, myeloid, B cells, and T cells reconstitution were assessed at 4, 8, 12, and 16 week post-radiation and BM transplantation (n = 6 overall, myeloid, and B cells; n = 5 for T cells). (J) Frequency of HSCs in Prox1 Cre ER T2 X iDTA bones from Cre À iDTA and Cre + iDTA mice (n = 5). Frequency of HSCs in SAR131675 inhibitor-treated wild-type mice compared to PBS-injected (sham) mice (n = 5). One-way ANOVA tests with Tukey's multiple comparisons tests ( B and D); two-way ANOVA with Sidak multiple comparisons tests (E, F, H, and I); two-tailed unpaired t tests (G and J). ns p > 0.05, *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001. Scale bars: white, 500 mm; yellow, 100 mm. n represents biological replicates. See also Figure S5. Article reconstitution activity in the lymphangiogenesis inhibitor-treated mice is due to defective recruitment of transplanted bone marrow cells to bone, we analyzed the recruitment after transplantation with and without inhibitor. Toward this, we performed qPCR analysis of Tomato expression after 24 and 36 h of transplantation of bone marrow cells from mTmG mice. This analysis showed that lymphangiogenesis inhibitor does not impact the recruitment of transplanted bone marrow cells to bone after genotoxic stress ( Figure S5H). Above results suggested that inhibition of lymphangiogenesis reduces HSC regeneration capability.
To further examine whether the presence of lymphatic vessels affected HSC regeneration, we employed an alternative strategy to target LECs directly. The Prox1-Cre ER T2 mouse line 37 crossed to the reporter Rosa26-td-Tomato, allowing visualization of PROX1-expressing lymphatic vessels. By intercrossing Prox1 Cre ER T2 mice with mice expressing a tamoxifen-inducible diphtheria toxin antigen (iDTA), we could specifically deplete LECs. Tamoxifen injections were performed in both Cre À and Cre + mice. As expected, Cre + iDTA mice demonstrated suppressed expansion of the lymphatic endothelium in irradiated bones, leading to reduced PROX1 expression ( Figure 3G). Consistent with the above results, Cre + iDTA mice also exhibited lower bone marrow cellularity, and reduced LSK and HSC frequency, compared to Cre À iDTA mice ( Figure 3H). In agreement with these results, competitive transplantation of bone marrow cells from these mice confirmed significantly reduced reconstituting activity of bone marrow cells derived from Cre + iDTA donors compared to Cre À iDTA mice ( Figure 3I). Importantly, although depletion of LECs during radiation impacted HSC frequency in Cre + iDTA mice, the depletion of LECs in unirradiated Cre + iDTA mice did not ( Figure 3J). Similarly, treatment with SAR131675 in adult, wild-type, and unirradiated mice did not alter HSC frequency ( Figure 3J). Thus, inhibition of lymphangiogenesis or loss of LECs reduces HSC regeneration after myeloablation, but HSCs are unaffected by depletion of LECs during homeostasis.
IL6 is required for radiation-induced lymphangiogenesis in the bone VEGF-C is a known critical driver of lymphangiogenesis. 23,54,55 SAR131675 inhibits VEGF-C/VEGFR3 signaling, and the above data therefore indicate that VEGF-C/VEGFR3 signaling is involved in lymphangiogenesis in the bone during injury-induced regeneration. To gain further insight into the molecular underpinning of lymphangiogenesis in the bone, we also examined interleukin 6 (IL6), another known driver of lymphangiogenesis. [56][57][58] Analysis of Il6 mRNA expression levels showed a significant increase in Il6 expression after irradiation ( Figure 4A). Furthermore, ELISA analysis of bone supernatants revealed a corresponding increase in IL6 protein expression upon irradiation ( Figure 4A), suggesting a role of this cytokine in bone lymphangiogenesis.
To determine whether IL6 was required for lymphangiogenesis, we examined IL6-knockout (KO) mice. The IL6-KO mice lacked expansion of lymphatic vessels after radiation or 5-FU treatment (Figure 4B). Interfering with the IL6 signaling by blocking the IL6 receptor (IL6R) in the wild-type mice inhibited lymphatic vessel expansion induced by genotoxic stress ( Figure S5I). Administration of IL6 in IL6-KO mice after irradiation or 5-FU treatment mice rescued this defect, restoring lymphangiogenesis and lymphatic vessel expansion and confirming that the defects were due to the loss of IL6 ( Figure 4B). As expected, given the effect on lymphangiogenesis, IL6-KO mice after myeloablation also exhibited significantly lower bone marrow cellularity and reduced LSK and HSC numbers compared to WT mice (Figure 4C). Competitive secondary transplantation of bone marrow cells from IL6-KO mice confirmed the significantly reduced reconstituting activity compared to those from wild-type mice ( Figure 4D). Together, these data establish that bone lymphangiogenesis requires IL6.

LEC-derived CXCL12 supports hematopoietic regeneration
The above data demonstrated that lymphatic vessels play a critical role during hematopoietic regeneration. We next wished to determine how lymphatic vessels support hematopoietic regeneration after myeloablation. Recent evidence has shown that LECs can regulate other cell types through secreted factors known as lymphangiocrine signals. [59][60][61][62] We therefore began by analyzing the expression of secreted factors in purified LECs from irradiated and sham mice bones. We focused on factors known to be critical for HSCs' maintenance and survival. We observed significant upregulation of Cxcl12 mRNA (but not other factors) in LECs after irradiation ( Figure 4E). Indeed, analysis of CXCL12-EGFP mice showed CXCL12-positive lymphatic vessels in bones from irradiated mice ( Figure 4F). Importantly, inhibition of lymphangiogenesis in Prox1 Cre + iDTA mice, or in IL6-KO mice, prevented this upregulation of CXCL12 in the bone marrow supernatant after radiation injury ( Figure 4F), demonstrating that upregulation of CXCL12 depends on the presence of LECs and on lymphangiogenesis and is not produced by other cells in the bone. CXCL12 is therefore a candidate lymphangiocrine signal.
To determine the functional significance of CXCL12 expression, we examined the bone marrow of LEC-specific CXCL12 loss-offunction mice. Unirradiated Prox1 Cre ER T2 X Cxcl12 fl/fl mice had normal bone marrow cellularity and HSC, LSK, myeloid, and erythroid cell numbers ( Figures 4G and S6A). They also exhibited normal expression of bone and bone progenitor markers ( Figure S6B), demonstrating that loss of CXCL12 did not impact the bone marrow under homeostatic conditions. However, after myeloablation and transplantation with bone marrow cells from wild-type mice, LEC-specific CXCL12 loss-of-function mice had significantly lower bone marrow cellularity and reduced numbers of LSK cells and HSCs compared to Cre À mice (Figure 4H). Further, competitive secondary transplantation of bone marrow cells from LEC-specific CXCL12 loss-of-function mice showed lower hematopoietic reconstitution ability ( Figure 4I). Thus, CXCL12 acts as a lymphangiocrine signal to promote hematopoietic regeneration.
Lymphatic vessels promote bone regeneration by promoting the expansion of Myh11-positive pericytes We next investigated whether lymphatic vessels also play a role in bone regeneration and bone formation after irradiation by comparing the expression of bone progenitor and osteoblast cell markers in LEC-depleted and sham mice. We found reduced expression of osteoprogenitor and osteoblast markers in long bones of irradiated mice treated with SAR131675 or in Prox1 Cre + iDTA mice, compared to sham mice or Prox1 Cre À iDTA mice ( Figure 5A). Calcein double labeling showed reduced bone formation rates after genotoxic stress in LEC-depleted mice, e.g., Prox1 Cre + iDTA mice compared to Cre À iDTA mice ( Figure 5B). Furthermore, microcomputed tomography (m-CT) showed that LEC depletion, e.g., in Prox1 Cre + iDTA mice, led to significantly decreased bone mass in regenerating bones after irradiation injury ( Figure 5C). These data suggested that lymphangiogenesis contributes to bone regeneration after irradiation.
We therefore sought to determine which specific bone cell progenitor population was impacted by the inhibition of lymphangiogenesis. Through lineage tracing approach with multiple mesenchymal and perivascular cell markers, we found that irradiation induced a significant expansion of myosin heavychain 11 (Myh11)-positive cells in bones but not in other organs ( Figures 5D, 5E, and S6C-S6F). Myh11 cells also express alpha smooth muscle actin (a-SMA) 63,64 ; these Myh11-positive cell populations represent mature pericytes. After irradiation, these cells gave rise to chondrocyte, adipocyte, and osteoblast lineages ( Figures 5D and 5F-5H). However, expansion and differentiation of this Myh11-positive cell population was not observed upon inhibition of lymphangiogenesis (treatment with SAR131675) (Figures 5D and 5E). In addition, lineage tracing of Myh11 cells during normal bone development and homeostasis indicated that Myh11-positive cells remained near large arteries and did not differentiate into other lineages as during irradiation ( Figure 5I). Unlike Myh11-positive cells, other mesenchymal cell types such as Pdgfrb + cells, Gli1 + cells, AdipoQ + cells, and Cspg4 + cells, which are also known to contribute to bone formation or regeneration, did not show any changes in abundance upon suppression of lymphangiogenesis ( Figures S7A-S7F). These results suggested that irradiation and lymphangiogenesis specifically impact Myh11-positive cells.
To test the functional relevance of the Myh11-positive cell population during bone regeneration, we crossed Myh11 Cre ER T2 mice with the iDTA mouse line described above to specifically deplete Myh11-positive cells. After irradiation, we observed reduced expression of osteoprogenitor and osteoblast markers in long bones of Cre + iDTA mice (lacking Myh11-positive cells) compared to Cre À iDTA mice ( Figure 5J), suggesting that Myh11-positive cells promote bone formation after injury. Consistent with this idea, immunostaining of osteopontin, which accumulates around trabecular and cortical bones, showed fewer bony elements in irradiated, Myh11 Cre + iDTA mice compared to Cre À iDTA mice ( Figure 5J). Immunostaining with osteocalcin, a marker for osteoblasts, showed reduced expression in bones from Myh11 Cre + iDTA mice ( Figure 5J  (legend continued on next page) ll OPEN ACCESS elements ( Figure 5J). Furthermore, m-CT examination showed that irradiated mice with depleted Myh11-positive cells had significantly decreased bone mass ( Figure 5K). These results are all consistent with the idea that Myh11-positive cells are required for bone regeneration. As shown above, after irradiation, LECs express CXCL12, and this chemokine acts as a lymphangiocrine factor to promote HSC reconstitution in the bone. CXCR4 is the CXCL12 receptor. We found that Myh11-positive cells express CXCR4, as demonstrated by CXCR4 immunostaining on bone sections in Myh11 Cre ER T2 X R26-td Tomato mice ( Figure 5L), raising the possibility that Myh11-positive cells could respond to CXCL12 signaling from LECs. Indeed, administration of CXCL12 in mice led to the expansion of the Myh11-positive cell population in bones even in the absence of genotoxic stress ( Figure 5M). Interestingly, analysis of Prox1 Cre ER T2 X Cxcl12 fl/fl mouse bones demonstrated a significant decrease in the Myh11-positive and CXCR4 + cells as quantified by flow cytometry (Figure 6A), indicating that CXCL12 derived from lymphatic vessels is required for the expansion of Myh11 + CXCR4 + cells.
Further, tracking and lineage tracing of Myh11-positive cells demonstrated that they expand in both the bone marrow, where they differentiate into adipocytes, and the endosteal region of bones, where they differentiate into perivascular osteoblasts ( Figures 5H, 6B, 6C, and S8A). Vasculature and osteoblasts in endosteal regions are known to be niche for HSC and critical sites for hematopoietic regeneration. 6,7,11,65,66 In line with this, Myh11 Cre ER T2 X iDTA, Cre + iDTA mice with depletion of Myh11-positive cells exhibited lower bone marrow cellularity and reduced LSK and HSC frequencies post-radiation (Figure S8B). Competitive transplantation of bone marrow cells from these mice confirmed significantly reduced reconstituting activity by bone marrow cells derived from Cre + iDTA donors compared to Cre À iDTA donors ( Figure S8C).
Next, we wondered whether CXCL12 may also mediate the effects of lymphangiogenesis on bone formation during genotoxic stress. This investigation revealed that the loss of CXCL12 expression in the lymphatic endothelium as in Prox1 Cre ER T2 X Cxcl12 fl/fl mice led to a decrease in bone mass and reduced expression of osteogenesis markers after radiation ( Figures 6D  and 6E). Together, these data demonstrate that Myh11positive cells are required for hematopoietic and bone regeneration and that the lymphangiocrine factor CXCL12 drives their expansion after genotoxic stress.
Aging inhibits the response of LECs to genotoxic stress Aging, which is associated with alterations in the bone marrow microenvironment, is known to impact bone and HSC regenera-tion. Given the role of LECs and lymphangiogenesis in bone and HSC regeneration that we showed above, we performed a series of experiments to test whether changes in lymphangiogenesis may contribute to aging-associated reduced regeneration abilities in the bone. Unlike bones from young mice, bones from aged mice did not exhibit lymphatic vessel expansion in response to genotoxic stress (Figures 6F-6H). Further, unlike the young mice, LECs purified from the bones of aged mice after radiation do not upregulate Cxcl12 (Figures 4E and 6I). In line with this finding, there is no expansion of Myh11-positive cells after radiation in aged mice as demonstrated by imaging of tibial bones ( Figure 6J). Also, we quantified Myh11-positive cells in young versus aged mice bones, which confirmed lack of expansion of these Myh11-positive cells in aged bones after irradiation ( Figure 6K). Thus, the responses of LECs and Myh11-positive cells to genotoxic stress change with aging. Finally, to understand the consequence of genotoxic stress on LECs, we performed apoptosis and proliferation analysis during radiation. Interestingly, LECs were damage resistant to genotoxic stress as illustrated by lack of changes upon radiation in the frequency of apoptotic LECs; at the same time, there was a dramatic increase in the proliferation of LECs ( Figure S8D). Further, lineage analysis with Lyve1 EGFP Cre X R26-td Tomato mice led to a surprising finding that the type H vessels in the bones have lymphatic origin during development as demonstrated by the presence of Tomato-positive, Endomucin + , LYVE1-EGFP À column-like vessels in the adult mouse bones ( Figure S8E). Type H endothelial cells represent the angiogenic subset of endothelial cells in bone and respond to radiation injury to recover the blood vasculature from radiation stress. 12 As expected, after the radiation due to rapid proliferation of type H endothelial cells, there was a rapid increase in Tomato-positive cells, of which only a small subset was LYVE1-positive and Tomato-positive, indicating that a small fraction of LECs during genotoxic stress have type H origin (Figures S8F and S8G).
Next, we asked whether senescence could explain the loss of regenerative ability in LECs and bone cells from aged mice. We therefore analyzed the expression of senescence markers in sorted LECs from young and aged mice. qPCR analysis of senescence markers showed upregulation of p16 and p27 ( Figure 6L) in aged mice LECs. Moreover, the LECs from aged mice demonstrated downregulation of proliferation marker Ki67 and lymphatic endothelial marker Vegfr3 ( Figure 6L). These data suggested that cell-intrinsic changes during aging may drive the lack of response to the genotoxic stress by LECs in aged mice as observed above. If that is the case, we reasoned that transplantation of LECs from young to aged mice might restore regenerative abilities. To test this hypothesis, we injected LECs into the bones of aged mice via intra tibial route, prior to radiation ( Figure 6M). Administration of LECs isolated from young but not aged mice led to substantial expansion of Myh11-positive cells ( Figure 6N) in irradiated aged mice and increased expression of several bone markers, including collagen I, osterix, and osteocalcin ( Figures 6O and 6P). The expression of osteoprogenitor and osteoblast markers also increased significantly in aged mice injected with young LECs after radiation injury ( Figure 6P). Furthermore, m-CT examination showed that administration of young LECs led to significantly increased bone mass ( Figure 6Q). Moreover, aged mice injected with young LECs showed increased hematopoietic regeneration after genotoxic stress relative to mice injected with aged LECs ( Figure 6R). Thus, aging leads to impairment in genotoxic stress-induced lymphatic expansion and lymphangiocrine signaling due to cell-intrinsic aging of LECs, and the administration of young LECs promotes bone and hematopoietic regeneration in aged bones.

DISCUSSION
In this work, we modified the existing methods to image intact skeletal tissue at a high resolution. Using in-depth 3D imaging, we establish the presence of lymphatic vessels in murine and human bones. We show that radiation-and chemotherapyinduced injury promote lymphangiogenesis and expansion of lymphatic vessels in the bone, and these, in turn, support the regeneration of HSCs within the bone and the bone itself. Lymphangiogenesis requires both VEGF-C and IL6 signaling, while the lymphangiocrine factor CXCL12 mediates the effect of LECs on HSCs and bone. These effects diminish with age due to changes in the ability of LECs to expand and support regeneration.
Lymphatic vessels play a vital role in facilitating the transport of essential fluids, macromolecules, and immune cells. 16,17,67-73 Recent work suggests an even broader array of functions. For instance, lymphatic vessels support cardiac development, and homeostasis, and help restore heart tissue following injury via secretion of the LEC-derived extracellular protein, relin. 37, [74][75][76][77][78] In the skin, lymphatic vessel-stem cell crosstalk drives wound repair. 79 Specifically, in the skin context, stem cells support lymphatic drainage and wound repair through secretion of the lymphangiocrine factor, angiopoietin-like protein 7 (Angptl7). 80 LECs have also been implicated in cancer metastasis, 81,82 where the LEC-derived ETS domain-containing protein 2 (ELK2) facilitates communication between the tumor and its surrounding microenvironment. 83 Here, we identify a fundamental role for LECs in mediating bone and hematopoietic regeneration after genotoxic stress. Finally, our findings show that cell-intrinsic changes in the bone lymphatic endothelium during aging underlie their lack of response to genotoxic stress in aged bones, thereby impacting bone and hematopoietic regeneration in aged animals. These findings raise further questions regarding LEC function, and it would be particularly interesting to examine the potential crosstalk between lymphatic vessels and bone-resident immune cells, for instance in the context of inflammatory diseases of the bone such as rheumatoid arthritis and osteoarthritis. These findings also raise the possibility of manipulating LECs and lymphangiocrine factors in clinical settings to accelerate regeneration in the skeletal system. Furthermore, these key findings will contribute to designing therapeutic approaches to treat blood and bone diseases.
Limitations of the study Transplantation of young LECs boosts bone and hematopoietic regeneration in aged mice. However, we did not characterize differences between young and aged LECs. Investigating age-dependent changes in LECs will provide a more in-depth understanding of bone LECs and strategies to manipulate them. Functions of lymphatic vessels include fluid transport and immunosurveillance. In this context, the limitation of this study is that we have not investigated where the lymphatic vessels in bones drain. Characterizing the locations where these lymphatic vessels drain will add to understanding broader lymphatic physiological, associated functions and drug interventions in bone. Also, we have not studied the organ-specific differences between bone LECs versus LECs from other organs.

STAR+METHODS
Detailed methods are provided in the online version of this paper and include the following:

Lead contact
Further information and request for resources and reagents should be directed to and will be fulfilled by the lead contact, Anjali Kusumbe (anjali.kusumbe@rdm.ox.ac.uk).

Materials availability
All materials generated in this study are available from the lead contact upon request.
Data and code availability d Data and supplementary tables and videos have been attached and are publicly available as of the date of publication. d RNA-seq data are available at E-MTAB-11560. d Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

EXPERIMENTAL MODELS AND SUBJECT DETAILS
Mice C57BL/6 mice (Charles River) were used as wild type mice for all analysis unless stated otherwise. Juvenile mice were aged between 3 and 6 weeks, adult mice between 8 and 12 weeks, and aged mice were >55 weeks. Both male and female mice were used. Details of the transgenic mouse lines are listed in Key resources table. In drug treatments, mice were randomly allocated for treatment and littermates were used as sham. For SAR131675 treatment, C57BL/6 mice received SAR131675 (Selleckchem) with a dose of 100 mg/kg via the intraperitoneal route. In this experiment sham mice were injected with PBS (PBS).
All animals were maintained following Principles of Laboratory Animal Care formulated by the National Society for Medical Research and the Guide for the Care and Use of Laboratory Animals (National Academies Press, 2011). According to institutional guidelines and laws, all experiments were performed following the protocols approved by the local University of Oxford and Imperial College London Animal Welfare and Ethical Review Board and the UK Government Home Office (Animals Scientific Procedures Group).

Human samples
Bone marrow biopsies were taken from the patients (male and female; age range: 61-77 years) pathologically diagnosed with lung cancer in Dazhou Central Hospital from June 2020 to January 2021. Iliac or pyramidal cancellous samples were collected by a professional orthopedic surgeon in the interventional room. The study was approved by the ethics committee, and informed consent of patients was obtained (IRB2020023). Tissue samples were obtained from the Oxford Musculoskeletal Biobank and were collected with informed donor consent in full compliance with national and institutional ethical requirements, the UK Human Tissue Act, and the Declaration of Helsinki (HTA Licence 12,217 and Oxford REC C 09/H0606/11). The research was supported by the National Institute for Health Research (NIHR) Oxford Biomedical Research Center (BRC). Disclaimer: The views expressed are those of the author(s) and not necessarily those of the NHS, the NIHR or the Department of Health. Male and female donors with age range 56-87 years were involved in the study.

METHOD DETAILS
Tamoxifen treatment for inducible gene deletion and genetic lineage tracing Genetic deletion and lineage tracing was performed as previously described. 12 Briefly, tamoxifen (Sigma-Aldrich, T5648) was freshly prepared by first dissolving in 100% ethanol and then suspended in corn oil to a final concentration of 5 mg/mL. For tamoxifeninduced genetic lineage tracing, Cre ER T2 mouse lines, as indicated in the figure legends, were used. To induce Cre activity, tamoxifen was administered orally at a dose of 50 mg/kg for three consecutive days. In all the experiments in this study tamoxifen injections were performed in both Cre + and Cre À mice as in previous studies. 1,2,12,84 At the indicated time, mice were euthanized by CO 2 asphyxiation, and tissues were collected for analysis.
Bone collection, fixation, and decalcification for light sheet imaging Freshly dissected bones were washed with PBS, (PBS, VWR, 437117K) and then quickly moved to an ice-cold fixative solution comprising 4% (w/v) paraformaldehyde, (PFA, Sigma-Aldrich, P6148) and 0.05% (v/v) glutaraldehyde (Sigma-Aldrich, 340,855) for 2.5 h. The fixative solution should be freshly made and should be ice-cold during use. Before the fixation step, the adjoining muscles and fat attached to the bone should be entirely removed by scraping as their presence interfere with generating whole bone overviews during imaging. An important step in the method is to remove the adjacent soft tissues and fat by scraping. This not only enables rapid clearing but most importantly provide accurate bone structural features for imaging. The soft tissue removal process with scraping and the collagenase digestion impacts periosteal cells. Thus, the method is suitable for analysis of bone marrow and cortical bones but not the periosteum.
The bone samples were then washed with PBS three times at room temperature on a rocker platform for 5 min for each washing step. The fixed bones can be stored for up to four days at 4 C before going forward with the decalcification. For decalcification, skeletal tissues were treated with 0.5 M EDTA solution at a pH of 7.4 and incubated at 4 C for 24 h under constant rotation. After decalcification, the samples were washed three times with PBS on a rocker (5 min at each wash).
Bones were submersed in an increasing ethanol gradient of 50%, 80%, and 100% for 30 min each for dehydration. 100% ethanol was changed twice after every 20 min. These dehydration buffers should be used ice-cold. Samples were then immersed in 5% (v/v) H 2 O 2 for 2 h (Sigma-Aldrich, H1009) for bleaching. Tissues were then rehydrated with the decreasing ethanol gradient followed by washing three times with PBS for 20 min.

Immunolabeling of whole bones
The samples, after fixation and bleaching, were subjected to antigen retrieval and permeabilization with an ice-cold solution containing 25% (w/w) Urea (VWR, 28,876.367), 15% (w/w) Glycerol (VWR, 24,388.260), 15% (w/w) Triton X-100 (Sigma-Aldrich, T8787), and double distilled water 45% (w/v) incubated for 5 h at 4 C. Samples can then be subjected to enzyme-based matrix digestion with 0.2% (w/v) Collagenase (Merck, 10,103,578,001) in PBS at 37 C for 30 min on constant shaking. The samples were then washed twice for 5 min with the wash buffer made of 2% (v/v) FBS (Sigma-Aldrich, F7524) in PBS on a rocker platform.
The samples were moved to a blocking solution (freshly prepared) containing 10% (v/v) donkey serum (Abcam, ab7475), 10% (v/v) DMSO (Sigma-Aldrich, D5879), and 0.5% (v/v) Triton X-100 in PBS, at 37 C for 20 min. After blocking, the tissues were then incubated with the Alexa fluor conjugated antibodies or with the unconjugated primary antibodies as per the need. Primary antibody (antibodies are listed in the Key resources table) solution was prepared in antibody dilution buffer containing 2% (v/v) donkey serum, 10% (v/v) DMSO, and 0.5% (v/v) Triton X-100 in PBS. The tissues were kept overnight or 14-16 h at 37 C in a water bath (Stuart, SBS40), shaking at 70 rpm after the incubation period, samples were washed with a solution composed of 2% (v/v) donkey serum and 0.5% (v/v) Triton X-100 in PBS for 3 h at 37 C in the water bath shaking at 70 rpm. The solution was changed every 15 min for the first hour and then after every 30 min during the washing. In the case of staining with unconjugated primary antibodies, after washing, the samples were incubated with dye conjugated secondary antibodies for 6-8 h at 37 C in a water bath shaking at 70 rpm. The secondary antibody is diluted with antibody dilution buffer. Following incubation, the samples were washed as per the protocol described for the primary antibody.

Dehydration and clearing of bones
After immunostaining, the tissues were dehydrated in an increasing gradient of 30%, 50%, and 80% of ethanol for 30 min each under gentle rotation at room temperature. The samples were then immersed in 100% methanol for 1 h with ethanol changes after every 20 min during this duration. The methanol was then completely removed, and the samples were rinsed twice with Ethyl cinnamate (ECi) (Sigma-Aldrich, 112,372) for 5 min each. Bones were then cleared with a clearing solution containing 80% (v/v) ECi and 20% (v/v) polyethylene glycol (PEG) (Sigma, 447,943) under gentle rotation at room temperature for 30-60 min.

Evans blue
We used a direct lymphatic visualization procedure to analyze murine lymphatic vessels in bone. Evans blue, a vital dye, binds with high-affinity tissue protein and is selectively and exclusively absorbed from interstitial space by initial lymphatic vessels. 12,43-45 Evans Blue (Sigma Aldrich, E2129) was injected subcutaneously into the inner leg, medial to the tail, and footpad. 12,41,43,46,47 The injection site was blebbed slightly before lymphatic vessels gradually take up the Evans blue. The dye was allowed to travel through lymphatics and accumulate in tibial bones and evaluated 3-6 h post-injection. Following that, mice were sacrificed, and bones were dissected and immediately placed in ice-cold 2% (v/v) PFA for 4 h. Cryosectioning was performed as previously described. 10,12,41,43,46,47 To image Evans blue in the thick cryosections, the 633 nm laser wavelength was used.
Irradiation C57BL/6J mice were whole-body irradiated with two doses of 1080 rad (Gammacell irradiator) at least 2 h apart. One million bone marrow cells from wild type or mT/mG mice were injected into the tail vein of anesthetized mice. The recipient mice were maintained on antibiotic water for 14 days after transplantation and then switched to standard water.
Long-term competitive reconstitution assay Cells were transplanted intravenously into the tail vein of anesthetized mice. For competitive reconstitution assays, 3310 5 donor bone marrow cells along with 3310 5 recipient bone marrow cells were transplanted. For secondary transplantation assays, 1x10 6 bone marrow cells from primary recipients and 1x10 6 compromised bone marrow cells which are the previously transplantedrecipient-type, were transplanted. Mice were maintained with antibiotic water for 14 days and then shifted to standard water. Recipient mice were regularly bled to assess the level of donor-derived blood cells, including B cells, T cells and myeloid cells. Blood was subjected to ammonium chloride/potassium-based red cell lysis before antibody staining. Antibodies including anti-CD45.2 (104, 1:100), anti-CD45.1 (A20, 1:100), anti-Gr1 (8C5, 1:800), anti-Mac-1 (M1/70, 1:400), anti-B220 (6B2, 1:800), and anti-CD3 Confocal imaging set-up and image acquisition Z-stacks of immunostained sections were imaged on the Zeiss Laser scanning confocal microscope LSM880 using the 20X Plan Apo/0.8 dry lens and 10X Plan Apo 0.45 WD = 2.0 M27 dry lens. The imaging set-up consisted of a Zeiss laser scanning microscope 880 equipped with Axio Examiner, laser lines: 405, 453, 488, 514, 561, 594, and 633 nm, Colibri 7 epifluorescence light source with LED (light-emitting diode) illumination, four objectives, fast scanning stage with PIEZO XY, 32-channel gallium arsenide phosphide detector (GaAsP) PMT (photomultiplier tube) plus two-channel standard PMT, acquisition, and analysis software including measurement, multichannel, panorama, extended manual focus, image analysis, time-lapse, z stack, extended focus, autofocus, and with additional modules: Experiment Designer and Tiles and Position.
Large regions through the thick sections were imaged using the tile scan function, and images were stitched with a 10% overlap using Zen Black (version 3.1, Zeiss) software. In order to visualize the boundary of the organ, the autofluorescence from the 405 channel was converted into greyscale, and the 30% opaque image was manually overlayed with the corresponding TIFF file generated from Imaris. Imaris, Adobe Photoshop and Adobe Illustrator software were used to generate, analyze and compile images.

Image analysis and quantifications
Slices/Z-stacks of images acquired on the light sheet and confocal microscope were processed and reconstructed in three dimensions with Imaris software (version 9.6.0). Imaris, Adobe Photoshop, and Adobe Illustrator software were used for image processing and analysis in line with the journal's guidance for image processing.
Maximum intensity projections were analyzed using Imaris (version 9.6.0). Vessel width was calculated using the distance tool in Imaris on single slices. Quantification of cell numbers was done on a z stack of images in Imaris using the automatic spot detection feature and manually annotated to remove any signal that was determined to be non-specific. For the analysis of lymphatic vessel density, the Imaris Surface Analysis XTensions tool was used. Briefly, using the Crop 3D tool, the total tissue volume was acquired via the Volume Statistics function in Imaris. Then, a single channel of a lymphatic endothelial marker was reconstructed in 3D using the Surface function, and the tissue volume of lymphatic vessels was measured using the Surface Statistics function. The lymphatic vessel density was calculated by dividing the tissue volume of lymphatic vessels in the numerator by the total tissue volume of the whole organ in the denominator.
3D surface reconstruction 3D surface rendering in images was applied using the surface module in Imaris. Briefly, the ROI was defined, and a single channel of lymphatic or blood vessels, perivascular cells, or matrix markers was reconstructed with surface segmentation. Followed by the smoothness of the ROI, the background subtraction option was used for the threshold settings. After the threshold adjustment, the manual option was set up to reach a proper value according to the preview. Last, the resulting images were visually inspected to manually remove small individual segmented components of high sphericity, which were regarded as noise.
RNA isolation for outer surface of bones and cortical bones Freshly dissected bones were prepared for light sheet imaging as described above by removing adjacent fat and scraping the cells around the outer surface of the bones. To isolate the RNA from the outer surface of bones, intact femurs were dipped in the RNA lysis buffer (RNeasy Plus Micro Kit, QIAGEN) for 20 min. After this the outer surface of bones in lysis buffer were scraped with a scalpel blade. To isolate the RNA from the cortical bones, femurs were cut from the top and bottom and then flushed with PBS using a syringe to remove the bone marrow. The flushed bones (cortical bones) were then crushed with mortar and pestle. Crushed bones were then subjected to digestion with 0.2% collagenase IV, dispase (1.25 U/ml) (ThermoFisher Scientific, 171,055-041), and DNase I (7.5 mg/mL) (Sigma-Aldrich, D4527-10KU) for 45 min at 37 C to prepare single-cell suspension for the isolation of RNA. RNA isolation was performed RNeasy Plus Micro Kit (QIAGEN, 74,034) according to the manufacturer's instructions.

RNA isolation from whole bones
Total RNA was isolated according to manufacturer's protocol (RNeasy Mini Kit, QIAGEN). A total of 100 ng RNA per reaction was used to generate cDNA with the iScript cDNA Synthesis System (Bio-Rad). Bones were crushed with mortar and pestle. 12 Crushed bones were then subjected to digestion with 0.2% collagenase IV, dispase (1.25 U/ml) (ThermoFisher Scientific, 171,055-041), and DNase I (7.5 mg/mL) (Sigma-Aldrich, D4527-10KU) for 45 min at 37 C for the isolation of RNA.
qPCR qPCR was done using TaqMan gene expression assays on the ABI PRISM 7900HT Sequence Detection System. 1,2,12,52,85 The FAM-conjugated TaqMan probes were used with the TaqMan Gene Expression Master Mix (Applied Biosystems, 4,369,510). Gene expression assays were normalized to endogenous VIC-conjugated Actb probes as standard. RNA samples were rapidly processed for cDNA preparation using the Super-Script IV First-Strand Synthesis System (Invitrogen, 18,091,200). To conduct qPCR, FAM-conjugated TaqMan probes were applied with TaqMan Gene Expression Master Mix (Applied Biosystems, 4,369,510). demultiplexed and converted to FASTQ format using cell ranger tool kit mkfastq. The raw data were aligned to the human reference genome (GRCh38), allowing filtering barcode and UMI count using cell ranger count (cellranger-5.0.0).

Cell-type clustering analysis and marker identification
The feature count matrix was further processed using Seurat (CRAN, version 4.0.1). 86 The mitochondrial and ribosomal reads were excluded from the analysis. Additionally, only PECAM1 positive cells are used for further analysis. After filtering, a total of 2305 cells were left for the following analysis. The data were normalized, scaled, and significant variable genes were identified using SC Transform. 87 Then, the dimensionality reduction technique was applied to the dataset and cell clusters were identified (resolution = 0.4). Further, the differentially expressed genes (DEG) were identified (Wilcoxon Rank-Sum test for genes with a minimum 0.5 log fold change with thresh = 0.01 between clusters and expressed in at least 25% of cells between clusters). Cell clustering was visualized using uniform manifold approximation and projection for dimension reduction (UMAP). UMAP gene expression overlays and violin plots for cell type-specific marker genes were plotted using Seurat-specific functions.

QUANTIFICATION AND STATISTICAL ANALYSIS
Panels usually represent multiple independent experiments performed on different days with different mice. Sample sizes were not pre-determined based on statistical power calculations. No randomization techniques were used. Mice were allocated to experiments randomly, and samples were processed in an arbitrary order. No blinding was performed. No animals were excluded from analyzes. Variation was always indicated using SD. Data represents mean ± SEM or SD To assess the statistical significance of differences between two groups, we generally performed two-tailed unpaired Student's t-tests. To analyze the statistical significance of differences among more than two groups, we performed one-way ANOVAs with Tukey's multiple comparisons tests. To determine the statistical significance of differences between multiple groups when the experimental design involved multiple conditions, such as time points or cell types, in addition to differences in genotypes, we performed two-way ANOVAs with Sidak's multiple comparisons tests. p < 0.05 was considered significant. ns: not significant, p > 0.05; *: p < 0.05; **: p < 0.01; ***: p < 0.001; ****: p < 0.0001. All statistical tests were performed using GraphPad with Prism7 (version 7), following its Statistics Guide. No statistical analysis was used to determine the sample size.   Cre ER T2 X iDTA mice were analyzed at 4-, 14-and 29-days post-radiation treatment and BM transplantation (n = 6 biological replicates). Statistical significance between each group was determined using two-way ANOVA with Sidak multiple comparisons tests (*p < 0.05, **p < 0.01, ***p < 0.001 and ****p < 0.0001).