Skeletal progenitor LRP1 deficiency causes severe and persistent skeletal defects with WNT/planar cell polarity dysregulation

Low-density lipoprotein receptor-related protein 1 (LRP1) is a multifunctional endocytic receptor whose dysfunction is linked to developmental dysplasia of the hip, osteoporosis and osteoarthritis. Our work addresses the critical question of how these skeletal pathologies emerge. Here, we show the abundant expression of LRP1 in skeletal progenitor cells at mouse embryonic stage E13.5 and onwards, especially in the perichondrium, the stem cell layer surrounding developing limbs essential for bone formation. Lrp1 deficiency in these stem cells causes joint fusion, malformation of cartilage/bone template and markedly delayed or lack of primary ossification along with aberrant accumulation of some of the LRP1 ligands at as early as E16.5. These early abnormalities result in multiple and persistent skeletal defects including a severe deficit in hip joint and patella, and markedly deformed and low-density long bones leading to dwarfism and impaired mobility. Mechanistically, we show that LRP1 regulates core non-canonical WNT/planar cell polarity (PCP) components that may explain the malformation of long bones. LRP1 directly binds to Wnt5a, facilitates its cell-association and endocytic recycling. Using Xenopus as a model system we show that loss or gain of LRP1 function leads to shortened tadpoles similar to Wnt5a and Wnt11 overexpression, indicating a role for LRP1 in WNT/PCP signalling. Finally, we show the colocalisation LRP1 and Wnt5a in the developing mouse limbs and that Lrp1 deficiency diminishes graded distribution of Wnt5a and Vangl2. We propose that skeletal progenitor LRP1 plays a critical role in formation and maturity of multiple bones and joints by regulating morphogen signalling, providing novel insights into the fundamental processes of morphogenesis and the emergence of skeletal pathologies.


INTRODUCTION
The low-density lipoprotein (LDL) receptor-related protein 1 (LRP1) is widely expressed type 1 transmembrane protein in adult tissues (1,2) that regulates cellular events by modulating the levels of structurally and functionally diverse extracellular molecules via clathrin-dependent endocytosis (3,4).
Our previous studies demonstrated that LRP1 plays an important role in the turnover of extracellular matrix (ECM) components in articular cartilage by mediating endocytic clearance of cartilage-degrading proteinases and their inhibitors (5)(6)(7)(8)(9)(10).This endocytic process is impaired in cartilage under inflammatory conditions or in osteoarthritis (OA), the most prevalent age-related degenerative joint disease (11).LRP1 also participates in signalling pathways through interaction with membrane receptors and cytoplasmic adaptor proteins.The cytoplasmic NPxY motifs within the LRP1 intracellular domain also provide binding sites for a set of signalling proteins (12)(13)(14).
Global deletion of the Lrp1 gene in mice results in early embryonic lethality at E13.5 (15,16).
LRP1 single nucleotide polymorphisms are associated with a decrease in bone mineral density and content (31).A recent study by Yan et al (32) identified mutations in LRP1 including R1783W with developmental dysplasia of the hip (DDH) patients.In mice, mutant Lrp1 R1783W homozygote and heterozygote mice exhibited delayed Y-shaped triradiate cartilage and smaller acetabulum in 8-week and 16-week-old mice, respectively (32).Heterozygous global Lrp1 knockout (KO) mice also developed a hip dysplasia phenotype.In contrast, a bone and cartilage conditional KO mice (Lrp1 flox/flox /Col2a1 Cre ) showed shortened bones and cartilage growth plate (27).In vitro, siRNA-mediated Lrp1 gene-silencing reduced chondrogenesis of human mesenchymal stem cells (33).These studies suggest that function of LRP1 during synovial joint and bone formation is cell type-and time-dependent.However, LRP1 expression and distribution in the developmental stages remains incompletely understood.
In this study, we addressed the critical questions of when and where LRP1 is expressed during skeletal development, how deficiency of LRP1 leads to skeletal pathologies and how long it persists, and which molecular mechanisms underpin the defects.We showed that LRP1 is abundantly expressed in skeletal progenitor cells at embryonic stage E13.5 and onwards, in particular in the perichondrium (34).To investigate the role of skeletal progenitor LRP1, we generated a conditional KO of Lrp1 using paired-related homebox gene-1 (Prrx1), which is expressed in the early limb bud skeletal mesenchyme and a subset of craniofacial mesenchyme (35,36).Lrp1 flox/flox /Prrx1 Cre mice exhibited severe and persistent malformation of multiple bones and synovial joints, which were not evident in either the Lrp1 R1783W or Lrp1 flox/flox /Col2a1 Cre mice.Our exploration of molecular mechanisms showed unique regulation of non-canonical WNT/planar cell polarity (PCP) components and signalling by LRP1.We propose that LRP1 plays a critical role in skeletal development by regulating WNT morphogen signalling, which governs a myriad of biological processes underlying the development and maintenance of adult tissue homeostasis.

RESULTS
LRP1 is abundantly expressed in skeletal progenitor cells, in particular in the perichondrium.
We investigated the distribution of LRP1 during skeletal development, which had thus far remained unknown.Histological investigation of LRP1 protein in developing limbs showed that LRP1 is abundantly expressed in E13.These joint and cartilage/bone template defects have not been reported for the Lrp1 R1783W (32) or Lrp1 flox/flox /Col2a1 Cre mice ( 27) most likely because they reflect events that occurred earlier than Col2a1 transcriptional activation.To further evaluate the role of chondrocyte LRP1 in the cKO early skeletal phenotype, we generated a double transgenic mouse line Lrp1 flox/flox /Acan CreERT2 (39).In contrast to Prrx1 expression (36,37), Aggrecan expression was detected in the E13.0 forelimb and in all developing cartilage by E15.5 (39).Lrp1 was deleted at the different embryonic stages and E19.To investigate details of the multiple and severe skeletal defects in Lrp1 flox/flox /Prrx1 Cre mice, we histologically examined the bone and joint sections.H&E staining of 2-and/or 14-week-old Lrp1 flox/flox /Prrx1 Cre shoulder, elbow and knee revealed striking defects (Fig 5A-C).These included a deformed and disrupted growth plate with poorly stacked columnar chondrocytes, markedly delayed secondary ossification, impaired articulation as a result of fused bone ends and lack of articular cartilage formation.Safranin-O/fast green staining further revealed proteoglycan depletion in articular cartilage and the growth plate in adulthood at 14-week-old mice but not during growth phase in 2-week-old mice (Fig 5A-C).Similar results were also obtained by Toluidine blue staining.Notably, femoral heads were substantially deformed with an extra groove juxtaposed to a poorly developed acetabulum socket (Fig 5D).Sox9, a pivotal transcription factor expressed in developing and adult cartilage (40)(41)(42), was almost absent in the articular cartilage and growth plate of 14-week-old Lrp1 flox/flox /Prrx1 Cre compared with WT mice (Fig 5E), emphasising defects in cartilage development.
We then tested dynamic histomorphometry of bone formation using calcein-double labelling in 14-week-old mice and showed a comparable rate of bone formation rate between WT and Lrp1 flox/flox /Prrx1 Cre mice (Fig 5F and S3A-C).Further analysis using tartrate-resistant acid phosphatase (TRAP) staining for osteoclasts revealed approximately 1.9-fold increased osteoclast activity in 14week-old Lrp1 flox/flox /Prrx1 Cre compared with WT littermates (Fig 5GH and S3DE).These results were consistent with Lrp1 deficiency in osteoblasts (28) or osteoclasts (29) resulting in low bone-mass phenotype due to increased osteoclastogenesis.(43).For example, deletion of core components of WNT/PCP including Ror2 (44), Vangl2 (45), Ryk (46) and Wnt5a (47) as well as inducible Wnt5a overexpression (48) all resulted in shorter and thicker long bones.Our recent study identified >50 novel candidate binding partners for LRP1 in cartilage (9) including Wnt5a and Wnt11, key components of WNT/PCP signalling.We therefore investigated the potential for direct interaction of Wnt5a and Wnt11 with LRP1 using a solid-phase binding assay.The assay employed purified Wnts and full-length LRP1 (49).We Several secreted LRP1 ligands are degraded intracellularly following internalisation (6)(7)(8)(9)(10).Some ligands, however, are recycled back to the extracellular milieu, facilitating their distribution and availability (51,52).As mentioned above, Wnt5a was identified as LRP1 ligand candidate in our previous study based on that it was co-immunoprecipitated from the chondrocyte culture medium with recombinant soluble LRP1 ligand-binding cluster II (sLRP1-II) (9), which is considered to be responsible for most ligand binding (7,8,(53)(54)(55)(56).However, our secretome analysis found that Wnt5a is not increased in the chondrocyte culture medium upon inhibition of LRP1-mediated endocytosis (9), suggesting a possibility of its endocytic recycling.We thus monitored the level of exogenously added Wnt5a in the medium and cell lysate of WT and LRP1 KO MEFs by Western blot analysis.After 3-24 h incubation, exogenously added Wnt5a was detected in both the conditioned medium and cell lysate (Fig 6JK).No significant difference in Wnt5a levels were observed in LRP1 KO compared to WT MEFs and relatively large amounts of Wnt5a still remained even after 24-h incubation with the cells.Human chondrocytes express higher levels of LRP1 compared to WT MEFs but unlike other LRP1 ligands such as TIMP3 (5,6), MMP13 (7) and a disintegrin and metalloproteinase with thrombospondin motifs (ADAMTS)5 (10), rapid clearance of Wnt5 was not observed in human chondrocytes (Fig 6L).These results suggest that LRP1 facilitates cell-association and mediates internalisation of Wnt5a but not its intracellular degradation.

LRP1 regulates WNT/PCP signalling in Xenopus embryonic development.
To evaluate the significance of LRP1-Wnt5a interaction in the WNT/PCP signalling pathway, we employed Xenopus laevis (African clawed frog), which is an invaluable model system for studying the role of WNT signalling in development (57).WNT/PCP singnalling directly controls convergent extension Lastly, we investigated the interaction of LRP1 and Wnt5a and how LRP1 deficiency affects Wnt5a distribution and activity in the developing limbs.Wnt5a mRNA is expressed in the perichondrium in developing limbs (47,63,64) but Wnt5a protein distribution in the developing limbs remains incompletely understood.Our histological investigation of Wnt5a protein in E16.5 elbow showed a Wnt5a distribution pattern similar to that seen for LRP1; abundant expression of Wnt5a in perichondrium and proliferative flattened chondrocytes, but very weak expression in hypertrophic chondrocytes (Fig 8A).Immunofluorescent confocal microscopy analysis further revealed that partial colocalisation of Wnt5a and LRP1 in E16.5 hind limbs (Fig 8B).Notably, Wnt5a immunosignal was substantially reduced in E16.5 Lrp1 flox/flox /Prrx1 Cre compared to WT limbs.To evaluate Wnt5a activity, we examined expression and phosphorylation of Vangl2, a core component of WNT/PCP as mentioned above.It has been reported that the abundance of Vangl2 is tightly controlled by the ubiquitin-proteasome system through endoplasmic reticulum-associated degradation (65).Wnt5a activity prevents proteasomal degradation of Vangl2, facilitating its export from the endoplasmic reticulum to the plasma membrane.Wnt5a also dose-dependently induces phosphorylation of Vangl2 through Ror2 to establish WNT/PCP gradient (45).
Immunofluorescent confocal microscopy analysis revealed a remarkable reduction of Vangl2 in E16.5

Discussion
This study showed, for the first time, a critical role of LRP1 in skeletal progenitor cells and its regulation of the WNT/PCP pathway.Lrp1 deletion in early skeletal progenitors caused severe defects in multiple bones and joints which persisted into skeletally mature 14-week-old mice, indicating a non-redundant function in skeletal development and maturity.These observations of long bone and joint malformations were not evident in Lrp1 flox/flox /Col2a1 Cre mice (27) or Lrp1 flox/flox /Acan CreERT2 mice.In the Lrp1 flox/flox /Col2a1 Cre strain, Col2a1 Cre gene expression was confirmed in limbs at as early as E12.5 (66), whereas in our Lrp1 flox/flox /Prrx1 Cre mice, Prrx1 Cre gene expression starts at E9.5 (35)(36)(37)(38).These studies suggest that LRP1 in early skeletal progenitor cells prior to E12.5 is more intrinsically involved in guiding proper mesenchymal cell recruitment for the formation of multiple cartilage and bone elements.
Similarly, mice, in which Wnt5a overexpression is induced at E10.5 exhibited a much more severe long bone phenotype than mice induced at E12.5 (48).Indeed, overexpression of Wnt5a at E13.5 does not result in visible bone phenotype, indicating that the most critical period for Wnt5a in limb development is prior to E13.5.Our study using different models demonstrate the regulation of WNT/PCP signaling pathway by LRP1 during this critical early time point.
Skeletal progenitor LRP1 has at least two different functions: to remove molecules from the extracellular milieu and target them for intracellular degradation, and to capture, recycle and distribute molecules (Fig 9).LRP1 phosphorylation is likely to be dispensable for its function in skeletal development since knock-in mouse models of mutation of the proximal, and separately, distal NPxY motif to disable phosphorylation exhibited normal skeletal development (67).Our data clearly demonstrate that TIMP3 and CCN2 are tightly regulated LRP1 ligands (6,9), which accumulated throughout the limb in P0 when LRP1 was deleted in Lrp1 flox/flox /Prrx1 Cre mice.This indicated that in addition to the WNT/PCP pathway regulation, LRP1 is important in regulating extracellular concentrations of biologically active ligands such as TIMP3 and CCN2 in the developing limbs.
This study identifies LRP1 as a major receptor for Wnt5a.Higher levels of cell-associated Wnt5a were detected in WT compared with LRP1 KO cells, suggesting that LRP1 effectively captures Wnt5a from the cellular microenvironment.It has been reported that Wnt5a is internalised by cells via a clathrindependent pathway (68,69).We found that LRP1 mediates Wnt5a endocytosis but internalised Wnt5a is not degraded, and is most likely recycled.Our immunofluorescence staining further revealed a colocalisation of LRP1 and Wnt5a in bone template and perichondrium of E16.5 WT limbs.Strikingly, Wnt5a and Vangl2 was almost diminished in E16.5 Lrp1 flox/flox /Prrx1 Cre limbs, suggesting that LRP1 loss causes dysregulation of the WNT/PCP pathway.We are currently exploring two possibilities.One is that LRP1 facilitates Wnt5a binding to its cell-surface receptor complexes consisting of Ror1/2 and the frizzled receptors (Fig 9B) (70)(71)(72).The other is that LRP1 directly interacts with and stabilises Vangl2.
The perichondrium is a dense layer of fibrous connective tissue that covers cartilage in endochondral ossification.Two-way signalling between cells in the perichondrium and the underlying cartilage are essential for endochondral bone formation but questions remain regarding molecular mechanisms underpinning cellular communication between them (34,43).Our present study revealed an abundant expression of LRP1 in the perichondrium, strongly suggesting a role for LRP1 in the signaling pathways underpinning bond formation.Furthermore, the colocalization of LRP1 and Wnt5a in the perichondrium raises the possibility that these two molecules interact to control the recruitment of chondroprogenitor cells from the perichondrium into growth plate.We hypothesise that absence of their interaction in Lrp1 flox/flox /Prrx1 Cre limbs results in uncontrolled recruitment of chondroprogenitor cells leading to skeletal element widening.
It is worth noting that the data presented here differ significantly from other LRP family members associated with skeletal formation, such as LRP5 and LRP6, which function as Wnt coreceptors with the frizzled receptor and regulate canonical WNT/β-catenin signalling (73,74).LRP5/6 interact with many Wnts (Wnt1/2/3/3a/2b/6/8a/9a/9b/10b) but not Wnt5a or Wnt11 (75).Lrp6 KO mice die at birth and exhibit a variety of severe developmental abnormalities resembling those caused by mutations in Wnt1, Wnt3a and Wnt7a (73).Lrp6 KO limbs showed a consistent loss of the most posterior digit but did not exhibit the limb outgrowth defects observed in Wnt5a KO mice (47).Some Lrp6 KO mice exhibited deletion of additional digits and the radius, as well as malformation of the ulna.Therefore, Lrp1 flox/flox /Prrx1 Cre mice Lrp6 KO phenotypes are distinct.
In Xenopus, convergent extension movements are known to be controlled by non-canonical Wnt signalling (58,59).We clearly show that lrp1 knockdown using morpholinos or overexpression by using a well characterised mini-Lrp1 construct leads to convergent extension phenotypes similar to Wnt5a and Wnt11 (Fig 7 C-G and S4).These results suggest a clear interaction of lrp1 with non-canonical Wnt signalling.Nonetheless, accumulating evidence suggests a role of LRP1 in the regulation of canonical WNT signalling.Lrp1 deletion in mouse neural crest cells results in heart defects, which are associated with a decrease in canonical Wnt signalling (22).In vitro, LRP1 interacts with Frizzled1 and downregulates WNT/β-catenin pathway in HEK293T cells (76).In contrast, LRP1 stimulates WNT/βcatenin pathway and prevents intracellular cholesterol accumulation in fibroblasts (77).Macrophage LRP1 also increases WNT/β-catenin pathway by directly binding to and effectively removing secreted frizzled-related protein 5, which prevents Wnt binding to its receptor (78).These studies emphasise the developmental stage and tissue-specific nature of WNT signalling.Considering the importance of WNT/β-catenin pathway for synovial joint formation (79), LRP1 may selectively regulate a canonical WNT singalling during synovial joint formation, which requires further investigations.Since the 2.4 kb Prrx1 enhancer is reported to be expressed in skeletal muscle as early as E16.5 (37), we cannot exclude the possibility that skeletal muscle LRP1 plays a role in the observed phenotype.In particular twisted long bones, can be explained by altered mechanical adaptation due to abnormal skeletal muscle-bone interaction.Lrp1 flox/flox /Prrx1 Cre mice exhibited significant differences in limb length and femur thickness as early as at P0.One possibility to rationalise drastic skeletal changes at this stage is the mechanical stimuli generated by the embryo's active movements.Embryos with restricted movements in utero showed a skeletal phenotype of reduced ossification, especially in forelimbs (80).
In conclusion, we showed that abundant expression of LRP1 in early skeletal precursor cells, its critical role for synovial joint formation and accurate bone growth.We further demonstrated the regulation of WNT/PCP signaling pathway by LRP1 which may explain the malformation of long bones in Lrp1 flox/flox /Prrx1 Cre mice.We propose that combination of graded distribution of LRP1 and LRP1mediated endocytic clearance and recycling of extracellular signaling molecules provides a novel mechanism for appropriate morphogen gradient formation to ensure that bones and joints form correctly.
Investigations into the mechanisms underpinning formation of severe and persistent defects in skeletal elements caused by LRP1 loss are an essential for understanding the fundamental processes of morphogenesis, as well as the emergence of skeletal pathologies including DDH, osteoporosis and OA.

Mice
The Lrp1 flox/flox /Prrx1 Cre mice were generated by crossing the Lrp1 flox/flox (Strain 012604, the Jackson lab) and Prrx1 Cre (Strain 005584, the Jackson lab) mice.For post-natal analysis, 22 homozygotes, 14 heterozygotes and 21 wild-type littermates were examined for this study.For embryos, we examined 26 homozygotes, 12 heterozygotes and 24 wild-type littermates were examined for this study.

Immunohistochemistry
Slide sections were dewaxed and rehydrated by xylene and decreasing ethanol concentrations.Epitope unmasking was performed using a basic antigen retrieval reagent (R&D systems).The slides were immersed in the basic reagents and kept in a water bath for 10 minutes at 95°C.To block the endogenous peroxidase activity, 0.3% hydrogen peroxide was added to the slides and the slides were kept at 37°C for 15 minutes.Avidin/Biotin blocking kit (Vector Lab) was used to block endogenous biotin, biotin receptors and avidin.To block unspecific antibody binding, 10% goat serum was added to the slides and incubated for 3 hours at room temperature (RT).The primary antibody was then added to the slides and kept at 4°C overnight.The impress HRP Goat anti-Rabbit IgG polymer Detection Kit (Vector Lab) was used as a secondary antibody.One drop was used for each section, and the slides were kept at RT for 30 minutes.For signal enhancement, the VECTASTAIN ABC-HRP Kit (Vector Lab) was added to the slides and kept in the dark for 30 minutes.For signal visualisation, the DAB substrate kit (Vector Lab) was used to develop the brown signal.Fast green was used for counterstaining.Slides were dehydrated, cleared and mounted using increasing concentrations of ethanol, xylene and DPX (SIGMA), respectively.The primary antibodies used were as follows; rabbit monoclonal anti-LRP1 (1:200)(ab92544, Abcam), rabbit polyclonal anti-Sox9 (1:500)(AB5535, SIGMA), rabbit polyclonal anti-Wnt5a (1:200)(bs1948R, Bioss).The Rabbit IgG control antibody (I1000, Vector Lab) was used as isotype control.At least three mice/group were analysed for each staining.

Microscopic Examination
For microscopic examination and imaging, the Nikon Eclipse Ci microscope with the DS-Fi2 highdefinition colour camera head was used.All images were visualised using NIS-Elements imaging.The Zeiss Axio Scan.Z1 slide scanner system was used for the automated imaging.The resulting images were inspected using ZEN 3.0 (Blue edition), in which the image scale bar was added.

Histology and staining
All samples were fixed with neutral buffered formalin overnight, then kept in 70% ethanol until processing.EDTA was used for bone decalcification with different incubation times based on the samples age.Sample processing was automated using the Leica ASP300 tissue processor (Leica Microsystems, UK).The Leica EG1150 H embedding station was used for sample embedding.Sample sectioning at 5 µm thickness was performed using the Leica RM2245 microtome.For H&E staining, slides were deparaffinised in xylene and rehydrated was performed using decreasing ethanol concentrations.For nucleus staining and counterstaining, slides were stained with haematoxylin and eosin, respectively, for 5 minutes.The slides were then dehydrated in increasing ethanol concentrations before clearing them in xylene followed by mounting.For Safranin O staining, sections were deparaffinised, rehydrated and stained with haematoxylin for 30 seconds followed by counterstaining with 2% fast green for 2 minutes.The sections were then dipped in acetic acid for 20 seconds before staining with safranin O for 8 minutes.At least three mice/group were analysed for each staining.

Osteoblast isolation
Upper and lower limb bones from each mouse were chopped into small pieces, followed by collagenase digestion in a shaking water bath for 90 minutes at 37°C.After washing, the crushed bones were resuspended in 2ml of DMEM/F12 media with 10 % Fetal bovine serum (gibco, A384400-01) and then transferred to a T-25 culture flask.Clavaria osteoblasts were harvested the same way, but with collagenase digestion for 180 minutes.Osteoblasts were cultured and then subjected to SDS-PAGE followed by Western blot analysis.

Micro-computed tomography (μCT)
For ex vivo high-resolution μCT imaging, all samples were fixed in buffered and then kept in 70% ethanol until processing.µCT scanning was performed using the Skyscan 1272 (SKYSCAN, Belgium) for all developmental stages.Based on the sample type and age, different imaging parameters were applied.
For embryos, the parameters were as follows: 0.5 rotation step, 9μm isotropic resolution, 0.25 aluminium filter, 30 random movements and 4 average frames.For postnatal samples, the scanning parameters were as follows: 0.3 rotation step, 4.5μm isotropic resolution, 0.5 aluminium filter, 30 random movements and 2 average frames.The parameters for skull imaging were as follows: 0.5 rotation step, 9μm isotropic resolution, 0.5mm aluminium filter, 30 random movements and 2 average frames.Skyscan NRecon software was used to reconstruct the obtained images.Skyscan Data Viewer software was used to measure bone length and width.For trabecular bone, the Skyscan CT-analyser software was used to identify the trabecular bone in the tibia.CTvox software was used to visualise the obtained images in a 3D form.
For in vivo μCT, the University of Liverpool's centre of preclinical imaging provided the imaging services using the Quantum GX-2 system (PerkinElmer, Inc.Waltham, MA).Whole-body scan images were acquired with the protocol FOV72, high speed, 8 sec x 3. High-resolution scans for upper and lower limbs were performed with the protocol FOV36, high resolution, 4 min.At least five mice/group were analysed for each staining.

Locomotor activity monitoring
All mice were housed in a Digital Ventilated Cage (DVC®) rack, equipped with a home cage monitoring system capable of automatically measuring animal activity 24/7 (81).The DVC® rack is installed on a standard IVC rack (Tecniplast DGM500, Buguggiate, Italy) by adding sensing technologies externally to the cage, so that neither modifications nor intrusion occur in the home cage environment.Animal locomotion activity is monitored via a capacitance sensing technology by means of 12 contactless electrodes, uniformly distributed underneath the cage floor.The 6-week-old WT and cKO mice were monitored for 4 weeks and total distance and average speed of each mouse were measured.

Whole-mount skeletal staining
The whole-mount skeletal staining was done following the Rigueur and Lyons's protocol (82).Briefly, after dissection, adult mice skeletons were initially kept in 95% ethanol for 4 h before changing the solution and leaving in 95% ethanol overnight at RT.Then, skeletons mice were kept in acetone for two days at RT. Cartilage staining was performed by immersion in alcian blue for three days; then, mice underwent two changes of 95% ethanol for 4 hours and overnight for destaining.For pre-clearing, skeletons were kept in 1% KOH overnight at 4°C.Bone staining was performed by submersion in alizarin red for five days.Lastly, the final clearing was performed using 1% KOH prior to long-term storage in 100% glycerol.

Double calcein labelling
Double calcein labelling was performed as described previously (83).Two intraperitoneal (IP) calcein injections (150 µl/mouse) were given to 5-6 weeks old mice 4 days before culling at an interval of two days.Mice were dissected, and the tibia bones were fixed in natural buffered formalin and kept in 70% ethanol until processing.The bones were dehydrated at 4°C with xylene and decreasing ethanol concentrations.Sample infiltration was performed under vacuum for seven days using a solution containing 88.99% Methyl Methacrylate (MMA), 10% dibutyl phthalate, 1% Perkadox 16 and 0.01% Novoscave.Teflon blocks filled with MMA were used for bone embedding.The blocks were kept at 30 °C in a water bath to polymerise for 18 hours.After that, Historesin was used to attach the embedding rings.Sectioning at 5 µm was performed using the Leica RM2265 microtome.To analyse bone formation, sections were stained without de-plasticisation with 0.1% calcein blue for 3 minutes, dehydrated using different ethanol concentrations and cleared with xylene changes.Three mice/group were analysed.
Section clearing and rehydration were accomplished by xylene and decreasing concentrations of ethanol.TRAP solution was prepared by dissolving naphthol ASTR-phosphate (1.4 mg/ml) and fast red (1.4 mg/ml) in a 0.2 M acetate buffer (pH 5.2) containing 100 mM sodium tartrate.Bone sections were kept in the TRAP solution for 2 hours at 37°C.Then bone sections were counterstained with 0.33 g/l aniline blue and 6 g/l phosphotungstic acid for 15 minutes.The slides were washed with distilled water and then cover slipped with Apathy's serum.Three mice/group were analysed.

Enzyme-linked immunosorbent assay (ELISA)
Purified human full-length LRP1 (10 nM in 100 µl of 50 mM Tris-HCl (pH 7.5)/150 mM NaCl/10 mM CaCl2, TNC) was coated overnight at 4°C onto microtiter plates (Corning, NY).Wells were blocked with 5% bovine serum albumin in TNC for 24 hours at 4 °C and washed in TNC containing 0.1% Brij-35 after this and each subsequent step.Wells were then incubated with various concentrations of Wnts in blocking solution for 30 min at RT. Bound proteins were detected using anti-Wnt3a antibody (ab219412, Abcam), anti-Wnt5a antibody (MAB6452, R&D systems) or anti-Wnt11 antibody (ab31962, Abcam) for 1 hour at RT and then with a secondary antibody coupled to horseradish peroxidase for 1 hour at RT.
Hydrolysis of tetramethylbenzidine substrate (KPL, Gaithersburg, MA) was measured at 450 nm using a FLUOstar Omega (BMG Labtech).Mean values of technical duplicate were normalized by subtracting the amount of recombinant protein bound to control well that was not coated with LRP1.Extrapolated KD,app values were estimated based on one-phase decay nonlinear fit analysis using GraphPad Prism 9.

Monitoring exogenously added Wnt5a levels in the cell culture
For 0-24 hours incubation assay, WT and LRP1 KO MEFs, and human chondrocytes (5 x 10 3 /well) were grown in 24 well plate (pre-coated with 0.1% gelatin for overnight) until cells reach confluent.Cells were then incubated with DMEM/F12 for overnight.The medium was replaced with 0.5 ml of fresh DMEM/F12 containing polymyxin B (50 µg/ml), CT1746 (100 µM) and the protease inhibitor cocktail (1/1000) with or without 20 nM Wnt5a in the presence or absence of 500 nM RAP.After 0-24 hours, 0.  showing the frequencies of the identified phenotypes.

Figures 1 to 9
Figures 1 to 9 5-to newborn (P0) elbow joints, with the strongest immunosignals of LRP1 detected in the perichondrium layers, a dense stem cell layer surrounding developing limbs essential for bone formation (Fig 1A).To investigate the role of skeletal progenitor LRP1 in skeletal development in vivo, we have established a double transgenic mouse line Lrp1 flox/flox /Prrx1 Cre (Fig 1B).In this strain, a 2.4 kb Prrx1 enhancer directs the transgene expression in undifferentiated mesenchyme in the developing limb buds around embryonic day 9.5 (E9.5) (35-38).Transgene expression is extinguished in the condensing mesenchyme and chondrocytes, and the expression is confined to the perichondrium/periosteum of the limbs at E15.5.Immunohistochemistry of LRP1 confirmed specific deletion of LRP1 in Prrx1 expressing cells in the E16.5 knee but not in their ribs (Fig 1C).Conditional deletion of Lrp1 in skeletal progenitors impairs early bone and joint formation.Haematoxylin and eosin (H&E) staining of limbs at different embryonic stages revealed fusion of joints, malformation of cartilage/bone template and markedly delayed or lack of primary ossification in E16.5 Lrp1 flox/flox /Prrx1 Cre homozygote shoulder, elbow and knee joints (Fig 2A-D) and the defects become more severe in E18.5 and P0 neonates (Fig 2BC).Striking joint malformations were observed in P0 Lrp1 flox/flox /Prrx1 Cre hip joints (Fig 2D).These defects were not observed in heterozygote (Lrp1 flox/wt /Prrx1 Cre ) littermates.Histological investigation of LRP1 ligands including tissue inhibitor of matrix metalloproteinase 3 (TIMP3) and CCN2 revealed their aberrant accumulation in P0 Lrp1 flox/flox /Prrx1 Cre compared with WT limbs (Fig 2HI), indicating a critical role of LRP1 in their tissue availability.High resolution µCT scanning showed that femur and humerus length were shorter in P0 Lrp1 flox/flox /Prrx1 Cre compared with WT mice, whereas no significant difference was observed in E18.5 embryos (Fig 2J).No significant differences in body length were observed in P0 neonates (Fig 2K).

5 embryos
were examined (Fig S1).Lrp1 gene excision was confirmed by Lrp1 exon 2 PCR (Fig S1B) but histological analysis and bone length measurement confirmed normality of skeletal development in E19.5 Lrp1 flox/flox /Acan CreERT2 mice (Fig S1C-F).LRP1 deficiency in skeletal progenitors results in dwarfism, impaired mobility, abnormal gaits and fore/hind limb malformation.We next investigated the impact of early skeletal developmental defects in postnatal Lrp1 flox/flox /Prrx1 Cre mice.The mice survived into adulthood, but the weight of the homozygote mice was consistently lower than their WT and heterozygote littermates starting 3 weeks up to 14 weeks after birth (Fig 3A).Strikingly, Lrp1 flox/flox /Prrx1 Cre mice consistently showed altered posture and impaired mobility at birth which was persistent up until 14 weeks of age (Movies S1 and S2).A continuous automated home-cage monitoring system for a 4-week period from week 6 showed significantly reduced locomotor activity of Lrp1 flox/flox /Prrx1 Cre mice compared with WT littermates (Fig 3B).Whole-mount skeletal staining of 14-week-old postnatal adult mice with Alcian blue and Alizarin red showed that Lrp1 flox/flox /Prrx1 Cre mice had shorter limbs and smaller stature than their WT littermates (Fig 3C).Lrp1 flox/flox /Prrx1 Cre mice were unable to open digits in their hind limb (Fig 3D).As an occasional feature in Lrp1 flox/flox /Prrx1 Cre mice, fused and abnormal digits were observed in their fore limb (Fig 3EF).None of these phenotypes were observed in Lrp1 flox/flox /Prrx1 Cre heterozygote mice, suggesting an autosomal recessive phenotype.Severe and persistent defects in multiple bones and joints in Lrp1 flox/flox /Prrx1 Cre mice.In vivo µCT scanning of 2-week-old Lrp1 flox/flox /Prrx1 Cre mice revealed that long bones were not only shorter but markedly thicker and twisted, with much delayed or a lock of emergence of secondary ossification centres (Fig 4A).The upper limbs were abnormally twisted and the scapular were poorly defined compared with WT.The absence of crescent-shaped acetabulum sockets, rounded femoral heads and patella bone in Lrp1 flox/flox /Prrx1 Cre mice suggested an impaired mobility.Notably, these bone and joint defects remained defective in 14-week-old mice (Fig 4B).The observed difference in the length and width of femur and humerus bones between 6-week-old WT and Lrp1 flox/flox /Prrx1 Cre mice persisted in 14-week-old mice (Fig 4C and S2).High-resolution µCT scanning of a 14-week-old Lrp1 flox/flox /Prrx1 Cre tibia revealed a substantial reduction in trabecular bone density with virtual absence of primary spongiosa compared to the WT tibia (Fig 4D).Fusion and deformities of phalanges were observed in Lrp1 flox/flox /Prrx1 Cre forelimb with fused and abnormal digits (Fig 4E).Defects in growth plate, organisation of columnar chondrocytes, secondary ossification, articulation and cavitation of joints and proteoglycan turnover in Lrp1 flox/flox /Prrx1 Cre mice.

LRP1 mediates endocytic recycling
of Wnt5a, a core non-canonical WNT/planar cell polarity (PCP) pathway component.The observed skeletal defects could potentially arise from a range of mechanisms.Markedly thicker and shorter long bones (Fig 4), and disrupted growth plate with disorganised columnar chondrocytes (Fig 5) closely resemble phenotypes associated with defects in the non-canonical WNT/PCP signalling pathway (Fig 6A) found that Wnt5a and Wnt11 directly bind to immobilised LRP1 with high affinity (apparent binding constant (KD,app) of31 nM and 42 nM, respectively (Fig 6BC).In contrast, binding of Wnt3a, a key canonical WNT component, to LRP1 was negligible with KD,app of >200 nM (Fig 6D).Since Wnt5a dysregulation may explain the malformation of long bones, function of LRP1 in cellular trafficking of Wnt5a was investigated.To test whether LRP1 facilitates cell-association of Wnt5a through direct capture of Wnt5a, exogenously added Wnt5a in the cell-lysate of WT and LRP1 KO mouse embryonic fibroblasts (MEFs) (50) was monitored for incubation periods ranging between 0-60 min.The levels of cell-bound Wnt5a in WT MEFs after incubation for 5, 15 and 30 min were significantly higher than these in LRP1 KO cells with 1.6, 1.5 and 1.4-fold differences, respectively (Fig 6EF).We next examined internalisation of Wnt5a in these cells.Immunofluorescent confocal microscopy analysis revealed the co-localisation of Wnt5a and LRP1 inside WT MEFs (Fig 6G).The fluorescent signal of Wnt5a was markedly reduced in the LRP1 KO MEFs, indicating that LRP1 is responsible for cellular update of Wnt5a.The intracellular colocalisation of exogenously added Wnt5a and LRP1 was also observed in human primary chondrocytes (Fig 6H).This was markedly reduced in the presence of receptor-associated protein (RAP), which antagonises ligand binding by competitively occupying the ligand binding region of LRP1.Three-dimensional reconstruction of the confocal microscopy images using IMARIS (v8.1) further confirmed that the majority of internalised Wnt5a was associated with LRP1 (Fig 6I).
movements in the developing embryo.Loss of expression or overexpression of WNT/PCP components leads to embryos showing shortened trunks (Fig S4A)(58, 59).We confirmed that injection of Wnt5a or Wnt11 mRNAs into the dorsal marginal zone of 4-cell stage embryos results in shortened tadpoles (Fig S4BC).According to Xenbase, lrp1 gene expression in Xenopus laevis starts at Oocyte V-VI stage, peaks at Nieuwkoop and Faber stage 1 and then remains on throughout development (Fig 7A)(60, 61).Wholemount in situ hybridization using a lrp1 probe showed that lrp1 is expressed in the neural tube, branchial arches, somites and neuroadrenergic cells during the development (Fig 7B).Geneknockdown of lrp1 by injection of various doses of lrp1 morpholino into the dorsal side of 4-cell stage embryos caused conversion extension phenotypes of shortened trunks at all the concentrations tested (Fig 7CD).Compared to control morpholino, 20 ng of lrp1 morpholino increased frequency of conversion extension phenotype by ~2.2-fold (Fig 7E).We next examined the effect of overexpression of LRP1 mini-receptor consisting of the ligand-binding cluster II and the entire C-terminus, including the transmembrane domain and the cytoplasmic tail.This functional mini-LRP1 receptor maintains capacity to mediate clathrin-dependent endocytosis of LRP1 ligands (62).Injection of the mini-LRP1 increased frequency of conversion extension phenotype in a dose-dependent manner (Fig 7F).Compared to control, 5 pg of mini-Lrp1 injection increased frequency of conversion extension phenotype by ~8.2-fold (Fig 7G).These results suggest a role for LRP1 in regulation of the WNT/PCP pathway.LRP1 partially colocalises with Wnt5a and its deficiency diminishes graded distribution of Wnt5a and Vangl2 in the developing limbs.

Fig 7 .
Fig 7. LRP1 regulates WNT/PCP signalling in Xenopus embryonic development.A, lrp1 gene expression profile during Xenopus laevis embryonic development available in Xenbase.

Fig 8 .
Fig 8. LRP1 partially colocalises with Wnt5a and its deficiency diminishes graded distribution of

Fig 9 .
Fig 9.A novel and critical role for LRP1 in skeletal development and its deficiency in emergence