Mutation of the galectin‐3 glycan‐binding domain ( Lgals3‐R200S) enhances cortical bone expansion in male mice and trabecular bone mass in female mice

We previously observed that genomic loss of galectin‐3 (Gal‐3; encoded by Lgals3) in mice has a significant protective effect on age‐related bone loss. Gal‐3 has both intracellular and extracellular functionality, and we wanted to assess whether the affect we observed in the Lgals3 knockout (KO) mice could be attributed to the ability of Gal‐3 to bind glycoproteins. Mutation of a highly conserved arginine to a serine in human Gal‐3 (LGALS3‐R186S) blocks glycan binding and secretion. We generated mice with the equivalent mutation (Lgals3‐R200S) and observed a subsequent reduction in Gal‐3 secretion from mouse embryonic fibroblasts and in circulating blood. When examining bone structure in aged mice, we noticed some similarities to the Lgals3‐KO mice and some differences. First, we observed greater bone mass in Lgals3‐R200S mutant mice, as was previously observed in Lgals3‐KO mice. Like Lgals3‐KO mice, significantly increased trabecular bone mass was only observed in female Lgals3‐R200S mice. These results suggest that the greater bone mass observed is driven by the loss of extracellular Gal‐3 functionality. However, the results from our cortical bone expansion data showed a sex‐dependent difference, with only male Lgals3‐KO mice having an increased response, contrasting with our earlier study. These notable sex differences suggest a potential role for sex hormones, most likely androgen signaling, being involved. In summary, our results suggest that targeting extracellular Gal‐3 function may be a suitable treatment for age‐related loss of bone mass.

We previously observed that genomic loss of galectin-3 (Gal-3; encoded by Lgals3) in mice has a significant protective effect on age-related bone loss. Gal-3 has both intracellular and extracellular functionality, and we wanted to assess whether the affect we observed in the Lgals3 knockout (KO) mice could be attributed to the ability of Gal-3 to bind glycoproteins. Mutation of a highly conserved arginine to a serine in human Gal-3 (LGALS3-R186S) blocks glycan binding and secretion. We generated mice with the equivalent mutation (Lgals3-R200S) and observed a subsequent reduction in Gal-3 secretion from mouse embryonic fibroblasts and in circulating blood. When examining bone structure in aged mice, we noticed some similarities to the Lgals3-KO mice and some differences. First, we observed greater bone mass in Lgals3-R200S mutant mice, as was previously observed in Lgals3-KO mice. Like Lgals3-KO mice, significantly increased trabecular bone mass was only observed in female Lgals3-R200S mice. These results suggest that the greater bone mass observed is driven by the loss of extracellular Gal-3 functionality. However, the results from our cortical bone expansion data showed a sex-dependent difference, with only male Lgals3-KO mice having an increased response, contrasting with our earlier study. These notable sex differences suggest a potential role for sex hormones, most likely androgen signaling, being involved. In summary, our results suggest that targeting extracellular Gal-3 function may be a suitable treatment for age-related loss of bone mass.
In order to determine whether the extracellular function of Gal-3 is necessary for its effect on age-related bone loss, we generated a secretion deficient mouse knock in variant of Gal-3. Gal-3 can be mutated to selectively disrupt its glycan-binding and secretion functions while preserving its intracellular functions by mutating a highly conserved arginine to a serine in the glycan-binding domain of human Gal-3 (R186S) [30]. The R186S mutation lacks extracellular function in in vitro assays and is unable to bind any serum glycoproteins [30,31]. Mutation of the structurally equivalent arginine in galectin-7 blocks glycan-binding but still allows galectin-7 to function intracellularly [32,33]. Dissociation of intracellular interactions and regulation of mRNA splicing from extracellular glycanbinding has also been described for galectin-1 [34]. We created the equivalent R186S mutation in mouse Gal-3 (Lgals3-R200S) to prevent extracellular secretion of Gal-3 and assessed the consequences of this mutation on age-related bone loss in vivo.

Materials and methods
Experimental animals B6C3F1/J mice were obtained from Jackson Labs (Bar Harbor, ME, USA) and were used to generate our B6;C3-Lgals3 tm1Vari mutant model, which will be referred to as Lgals3-R200S.
Once this line was generated, we backcrossed it for two generations with purebred C57BL/6J mice. All animal procedures followed the protocol (PIL-16-01-002) approved by the Institutional Animal Care and Use Committee of the Van Andel Institute. Mice were housed in climatecontrolled conditions (25°C, 55% humidity, and 12 h of light alternating with 12 h of darkness) and fed a standard LabDiet Rodent Chow 5010 (Purina Mills, Gray Summit, MO, USA). Animals for our studies were euthanized at embryonic day 13 to collect mouse embryonic fibroblasts (MEFs) or at a 36-week time point for bone studies.

gRNA design and synthesis
Guides were designed to encompass 40 bp upstream and downstream of the target mutation site (Lgals3 codon 200 in exon 5) using the Zhang Lab's Optimized CRISPR Design Tool (crispr.mit.edu). A potential sgRNA binding site with a protospacer adjacent motif (PAM; AGG) was identified 8 bp downstream from codon 200. This sgRNA was selected for its proximity to the desired mutation site and low predicated probability for off-target binding (quality score 65).
A 20 bp DNA fragment (synthesized by Integrated  DNA Technologies (IDT), San Diego, CA, USA) representing the sgRNA plus BbsI complementary overhangs  (sgRNA  FWD: 5'-CACCTTTGCCACTCTCAAAGG GGA-3 0 and sgRNA REV: 5 0 -AAACTCCCCTTTGA-GAGTGGCAAA-3 0 ) was cloned into a BbsI digested pX330-U6-Chimeric_BB-CBh-hSpCas9 plasmid (42230; Addgene, Cambridge, MA, USA). Template DNA for in vitro transcription was generated by PCR amplification of the gRNA sequence using Phusion HF DNA polymerase and a primer set (synthesized by IDT) consisting of a FWD primer recognizing the cloned sgRNA sequence with a T7 RNA polymerase recognition sequence fused on the 5 0 end (5'-AATACGACTCACTATAGGGTTTGCCACTCT-CAAAGGGGA-3 0 ) and a REV primer that recognized the terminal end of the gRNA scaffold sequence on the pX300 plasmid (5'-AAAAGCACCGACTCGGTGCC-3 0 ). After verifying consistent production of a single product by agarose gel electrophoresis, the template was purified using the QIAquick PCR purification kit (Qiagen, Germantown, MD, USA) and eluted with RNAse free water.
In vitro transcription was performed on the purified gRNA template using a MEGAshortscript Kit (Ambion; Thermo Fisher Scientific, Waltham, MA, USA). The synthesized RNAs were pooled, treated with DNase to remove the remaining template, and purified using a MEGAclear Kit (Ambion; Thermo Fisher Scientific). The purity of eluted gRNA product was determined by formalin gel electrophoresis, and gRNA was quantified on a Nanodrop 2000c (Thermo Fisher Scientific). gRNA was stored at À80°C.

Microinjection of mice
Mouse zygotes were obtained by mating superovulated B6C3F1/J females with B6C3F1/J males. RNAs and ssODN were thawed and mixed just prior to injections for final concentrations of 30 ngÁlL À1 WT Cas9 mRNA (Sigma Aldrich, St. Louis, MO, USA), 15 ngÁlL À1 gRNA, and 50 ngÁlL À1 ssODN. The mix was microinjected into the pro-nuclei of zygotes and transferred into pseudopregnant females at the two-cell stage. After identifying founders, we backcrossed the line to C57BL/6J twice before intercrossing to generate animals for our study.

Genotyping
Genomic DNA was isolated from tail clips at weaning and necropsy by proteinase K digestion and ethanol precipitation. For the Lgals3-R200S reaction, wild-type mice (Lgal-s3 +/+ ) yielded a single 483 bp band, homozygous mutant mice (Lgals3-R200S KIKI ) yielded two 131 and 586 bp bands, and heterozygotes (Lgals3-R200S KI+ ) had three bands of 131, 483, and 586 bp. Sanger sequencing verified that the correction mutations and no spurious indels were present in the Lgals3-R200S allele. Tail tip DNA from Lgals3-R200S-positive animals was amplified by Phusion HF DNA polymerase PCR using primers encompassing exon 5 of Lgals3 (FWD: 5 0 -TTCAGGAGAGGGAATGA TGTTG-3 0 and REV: 5'-CTGAAGGAGCTGAAGGA CAC-3 0 ). The product of this reaction was purified using a QIAquick PCR Purification Kit (Qiagen), then cloned into pMiniT Vectors with the NEB PCR Cloning Kit and transformed into NEB 10-beta Competent E. coli (NEB, Ipswich, MA, USA). Colonies positive for the Lgals3-R200S allele by PCR were grown overnight at 37°C with nutation at 200 rpm. DNA from small aliquots of these cultures was then amplified using the TempliPhi DNA amplification kit and submitted to Genewiz (South Plainfield, NJ, USA) for Sanger sequencing using a forward primer upstream of the insertion site on the pMiniT Vector that came with the NEB PCR Cloning Kit. (5'-ACCTGCCAACCAAAGC GAGAAC-3 0 ).

Cell-surface biotinylation of mouse embryonic fibroblasts
Lgals3-R200S wild-type, heterozygous and mutant MEFs were harvested from 13-day mouse embryos. Cells were cultured on 10 cm dishes until they reached 70% confluency. Experiments were carried out at 0-4°C to reduce biotin internalization. Cells were washed three times with ice-cold PBS pH 8.0 and labeled with 1 mgÁmL À1 EZ-Link TM Sulfo-NHS-LC-Biotin (Thermo Fisher Scientific) for 30 min with gentle rocking. The cells were then washed three times with ice-cold PBS containing 100 mM glycine, and lysed in 50 mM sodium phosphate dibasic, 1 mM sodium pyrophosphate, 20 mM sodium fluoride, 2 mM EDTA, 2 mM EGTA, 1% Triton X-100, 0.5 mM DTT, and complete protease inhibitor cocktail (Roche, Basel, Switzerland) lysis buffer. A portion of the whole cell lysate was kept as an input control. Remaining lysate was incubated with NeutrAvidin agarose beads (Thermo Fisher Scientific) for 3 h with slow rotation prior to washing three times with lysis buffer. Samples were boiled in loading buffer containing DTT prior to SDS/PAGE and western blot. Antibodies used were as follows: goat anti-Galectin 3 (AF1197, R&D Systems, Minneapolis, MN, USA) diluted 1 : 10 000 and rabbit anti-Vinculin (#4650, Cell Signaling Technology, Danvers, MA, USA) diluted 1 : 1000.

Immunocytochemistry for Lgals3
Lgals3-R200S wild-type and mutant MEFs were harvested from e13.5 mouse embryos and grown directly on poly-llysine (2 lgÁmL À1 ) coated glass coverslips placed in six-well plates for 24 h prior to permeabilization and fixation with methanol for 15 min. Following three PBS rinses, cells were permeabilized with 0.1% Triton X-100 in PBS for 5 min, followed by three more PBS rinses, prior to blocking in 1% BSA/5% rat serum for 30 min. Coverslips were then incubated in anti-Gal3 antibody (R&D Systems, AF1197) diluted 1 : 1000 in 1%BSA/5% rat serum in PBS for 1-2 h, rinsed three times in PBS, incubated in secondary antibody conjugated to Alexa594 for 1 h, prior to three PBS rinses. DAPI (diluted 1 : 1000 in 1%BSA/PBS) was added to coverslips for 5 min and washed with 39 PBS and 19 water, before inverting onto a glass slide coated in mounting media. Optical imaging of cells was performed on a Nikon A1 + RSi laser scanning confocal microscope, equipped with DU4 high-sensitivity detectors. Images were captured using Nikon NIS-Elements AR 5.21.03 software with a 409 magnification plan fluor oil objective (NA 1.3), 402, 488 and 561 nm solid-state laser lines, and 450/50, 525/50, and 595/50 nm bandpass emission filters. Intensity of the immunocytochemical staining was quantified using FIJI software (National Institutes of Health, Bethesda, MD, USA). Briefly, cells were selected manually, and their mean and median fluorescence was automatically determined per unit area. We analyzed one biological sample per genotype and five slides per genotype were evaluated, with a total of n = 31 WT MEF cells and n = 40 Lgals3-R200S mutant MEF cells. These data are presented as the mean AE the standard error of the mean where the error bars show the error between slides.

Plasma isolation and enzyme-linked immunosorbent assays (ELISA)
Mice were euthanized at 36 weeks for all following experiments. Immediately following euthanasia, approximately 0.5 mL of whole blood was collected by heart puncture and transferred to a microcentrifuge tube containing 5 lL of 0.5 M ethylenediaminetetraacetic acid (EDTA) pH 8.0. To separate plasma, tubes were centrifuged at 6000 g for 6 min. Plasma Gal-3 from male and female wild-type and heterozygous mice was measured using a DuoSet ELISA Development System kit (D1197; R&D Systems). Wells were coated with 100 lL of 2 lgÁmL À1 Gal-3 capture antibody. 1 lL of plasma was added per well. Recombinant Gal-3 was used to generate a dilution curve for quantification.

Micro-computed tomography (lCT)
Right lower limbs and spines were defleshed and fixed in 10% neutral buffered formalin (NBF) for 72 h, rinsed with sterile distilled water, and stored in 70% ethanol at 4°C. Whole femurs and L3 vertebrae were imaged using a desktop SkyScan 1172 microCT imaging system (SkyScan, Kontich, Germany). Scans were acquired at 80 kV using a 5.98 lm voxel size. The femoral trabecular volume encompassed regions 0.25-1.75 mm from the distal growth plate. For analyses of trabeculae within the body of L3 vertebrae, a 1.5 mm volume centered on the midpoint was used. Cortical measurements were obtained from a 0.6 mm segment that was 45% of the distance proximal of the length of the diaphysis from the growth plate (diaphysis length = distance of femoral headdistance of the growth plate). Tissue mineral density and bone mineral density values were obtained using a standard regression line generated by converting the attenuation coefficients to mineral density from scans of hydroxyapatite standards with known densities (0.25 and 0.75 gÁcm À3 ).

Mechanical testing
Left femurs were defleshed, wrapped in PBS pH 7.2 soaked gauze, and stored at À20°C. To calculate tissue-level mechanical parameters, femurs were thawed at room temperature for 30 min in PBS pH 7.2 and analyzed by lCT at 80 kV with a 9.98 lm voxel size. Analyses were performed as described for cortical bone measurements for the right femurs-prior to À20°C storage. For mechanical testing, femurs were thawed and equilibrated to room temperature for 2 h in PBS pH 7.2. Following equilibration, a standard four-point bend test was performed using a Tes-tResources 570L axial-torsional screw-driven testing system (TestResources, Shakopee, MN, USA) with a displacement rate of 0.005 mmÁs À1 . The distances between the lower and upper supports were 7.3 and 3.5 mm, respectively. The supports had radii of curvature of 0.5 mm at each point of contact with the femur. Displacement was applied by the upper supports in the anterior-posterior direction such that the anterior of the femur was in compression and the posterior was in tension. Force and displacement were directly measured from the load cell and crosshead, respectively. Tissue level mechanical parameters (max stress and elastic modulus) were calculated as described for fourpoint bending [35] where max stress = (max force*a*c)/ (2I min ) and elastic modulus = stiffness*(a2/(12I min ))* (3LÀ4a); where a = the distance between an upper and lower support beam, L is the distance between the lower support beams, I min is the minimum calculated value of inertia, and c is the radius of the bone. I min and c were obtained by lCT.

Statistical analyses
A v 2 test (df = 5; a = 0.05) verified Mendelian distribution of pups born from Lgals3-R200S heterozygous crosses. For most comparisons, the differences between Lgals3-wild-type (+/+) and Lgals3-R200S (KI/+ and KI/KI) mice were determined using two-way ANOVAs within age groups (sex, genotype). The Holm-Sidak method was used in post hoc analyses to identify significant differences (a = 0.05). However, due to the large difference in plasma Gal-3 levels between male and female mice, one-way ANOVAs with Dunnett's post hoc tests were performed within sex between the genotypes.  9-15). Dunnett's post hoc analysis adjusted P-values compared to wild-type; bold values highlight *P < 0.05, **P < 0.01, ***P < 0.001. (E) Immunoblot of Gal-3 and vinculin on the cell surface of mouse embryonic cells isolated from WT (+/+), HET (KI/+), and MUT (KI/KI) embryos. Two biological replicates were evaluated per genotype and the assay was run once. (F) Gal-3 immunocytochemistry of permeabilized MEFs from Lgals3-R200S KI/KI and WT mice. Gal-3 is present throughout the cytosol and nucleus in both conditions. Gal-3 protein levels are overall reduced by approximately 30%. Scale bar = 100 lm. One biological sample per genotype and five technical slide replicates per genotype were evaluated, with a total of n = 31 WT MEF cells and n = 40 Lgals3-R200S mutant MEF cells. Quantification of Gal-3 staining intensity is indicated on the right, data is mean AE SEM. Asterisks indicate a significant difference, *P < 0.05.

Generation of Lgals3-R200S allele using CRISPR/ Cas9
We mutated the homologous arginine to serine in the mouse Lgals3 gene using CRISPR/Cas9 (Lgals3-R200S; Fig. 1A) to generate a version of Gal-3 that retains some intracellular functions but is deficient in glycoprotein binding, both extracellularly and to accumulate around disrupted vesicles intracellularly [36][37][38]. We identified the closest potential gRNA binding site with low predicted off-target binding sites with a protospacer adjacent motif (PAM) that was within 8 bp of codon 200 in Lgals3. To introduce our mutation via homologous recombination, we designed a 200 bp single-stranded oligo donor (ssODN) template by substituting the arginine codon (AGA) with a serine (TCA). To identify the Lgals3-R200S allele easier by PCR, we included additional silent mutations that disrupted a unique HpyCH4III restriction site and generated a novel Tsp45I restriction site (Fig. 1B). We identified three out of 120 pups that were heterozygous for our targeted mutation by allelespecific PCR of tail clip DNA from weanlings. These mice (one female and two males) were mated with wild-type C57BL/6J mice to assess for germline transmission. Only one of the males was confirmed to carry the mutant allele in his germline and produced offspring with~50% of the pups showing positive PCR results for the Lgals3-R200S allele. Sanger sequencing revealed that these pups had the desired mutant allele (Fig. 1B). In this study, heterozygous offspring from this F1 generation were crossed to generate the mice analyzed for skeletal phenotypes. PCR genotyping was used to identify wild-type (Lgal-s3 +/+ ), heterozygous (Lgals3-R200S KI/+ ), and homozygous (Lgals3-R200S KI/KI ) mutant mice (Fig. 1C). Consistent with a reduction in Gal-3 secretion, we observed significantly reduced Gal-3 protein levels in the plasma of adult heterozygous and homozygous mutant mice (Fig. 1D).
We generated mouse embryonic fibroblasts (MEFs) to look at cell-surface proteins, to confirm reduction of extracellular Gal-3 protein in Lgals3-R200S cells. Cell surface proteins were biotinylated and captured with NeutrAvidin Agarose. Western blot analysis showed cell surface Gal-3 levels were decreased in Lgals3-R200S KI/+ and Lgals3-R200S KI/KI cells compared to wild-type (Fig. 1E). The absence of the cytoplasmic protein, vinculin, from the pull-down lanes confirmed that the experiments worked to preferentially pull-down biotinylated cell surface proteins. Immunocytochemistry of MEFs from Lgals3-R200S KI/KI mice confirmed that Gal-3 is present in the cytosol and nucleus (Fig. 1F). Quantification of the mean and median amount of fluorescence per unit area indicated that Galectin-3 both on the cell surface and inside the cell was reduced, by approximately 30% and 26% respectively, in Lgals3-R200S KI/KI MEFs (P = 0.024). From these studies, we conclude that the R200S mutation in mice reduced cell surface Gal-3 and may have contributed to lower intracellular Gal-3 levels. Further studies will be necessary to determine if the 25-30% reduction in intracellular Gal-3 is biologically significant.

No major changes in body composition in aged
The observed ratios of wild-type, heterozygous and homozygous mice were observed in expected Mendelian ratios as determined by a v 2 test (data not shown). Like Lgals3-KO mice [39], Lgals3-R200S mice were grossly normal. Given that we previously observed significant bone phenotypes in Gal-3 knockout animals at 36 weeks [5], we chose this timepoint to analyze the Lgals3-R200S KIKI model. Since we are phenotyping aging bones, we are unable to give insight into how this mutation impacts early bone formation. Analysis of body composition changes by dual-energy X-ray absorptiometry (DEXA) revealed no statistically significant differences for male or female Lgals3-R200S mice in weight, areal bone mineral density (aBMD), or body fat percentage (%BF) as shown in Table 1. Values are expressed as mean AE SEM (n = 9-15). Dunnett's post hoc analysis adjusted P-values compared with wild-type; bold values highlight *P < 0.05, **P < 0.01, ***P < 0.001.

Enhanced cortical bone expansion in aged
Lgals3-R200S male mice We previously observed a significant increase in cortical bone expansion by 36 weeks in Lgals3-deficient mice, suggesting that loss of Gal-3 protects bone from age-related loss in mass [5]. Both male and female Lgals3-deficient mice had an increase in total area (T.Ar), while female mice also had significantly increased bone area (B.Ar) [5]. In the current study, we took cortical bone measurements from 36-week-old mice using lCT to evaluate changes in cortical bone volume (Fig. 3A,B). We used lCT to determine T.Ar, B.Ar., and Marrow Area (M.Ar) (Fig. 3C). Notably, only male Lgals3-R200S KIKI mice had cortical expansion (Table 3), demonstrated by significant increases in T.Ar (+9.3%; Fig. 3D) and M.Ar; (+11.9%; Fig. 3F). The effect size of the increased T.Ar and M.Ar in Lgals3-R200S KIKI males was twice as strong compared with what we observed in Lgals3 KOKO male mice [5].
Lgals3-R200S KI+ males had a significant increase in B.Ar (+9%; Fig. 3E) compared with wild-type males. The effect size in Lgals3-R200S KIKI females was roughly half of what we previously observed in Lgal-s3 KOKO females at this age [5]. This data suggests that the extracellular function of Gal-3 is not required in male mice for the protective effect on bone loss that occurs as a consequence of Gal-3 loss but is required in female mice.

Reduced bone quality in aged Lgals3-R200S male mice
Our previous study suggested there was a decrease in tissue level strength (max stress) and stiffness (elastic modulus) in 36 week Lgals3 KOKO male and female mice [5]. In the current study, male Lgals3-R200S KIKI mice, despite having significantly improved cortical geometry compared to wild-type males (MMI min: +14%, MMI max: +19%), also had significantly reduced tissue level strength (Max stress: À18.6%; Fig. 3H) but did not have reduced whole bone strength (Max force). Female Lgals3-R200S KIKI mice, on the other hand, had no reduction in max force (Table 3) or max stress (Fig. 3H). Stiffness results differed with what we observed in our earlier study (Table 3 and Fig. 3). Neither male nor female Lgals3-R200S mice had a change in elastic modulus (Fig. 3I) or whole bone stiffness (Table 3).

Discussion
In this study, we generated the Lgals3-R200S allele using CRISPR/Cas9 and a single-stranded DNA oligonucleotide as a template for homologous recombination. Mutation of the cognate arginine to serine in human Gal-3 (R186S) prevents Gal-3 secretion and glycan-binding [30,31,40]. Because mutation of the functionally equivalent arginine in galectin-7 also prevents glycan-binding [32,33], the R200S mutation in Gal-3 should be functionally equivalent. As confirmation, our surface biotinylation experiment demonstrated a dose-dependent reduction in surface Gal-3 in heterozygous and homozygous Lgals3-R200S mice. In our aged mouse bone studies, we observed a sexdependent increase in trabecular bone mass in female Lgals3-R200S mice. Yet only male Lgals3-R200S had significant increase in cortical bone expansion. The increased cortical bone expansion was coupled with reduced tissue quality (reduced max stress), and no change in tissue or whole bone stiffness values.
Similar to the findings presented here, we previously observed that female mice with genomic loss of Gal-3 (Lgals3-KO mice) had significant protection against age-related trabecular bone loss between 24 and 36 weeks of age [5]. But Lgals3-KO mice also had increased cortical bone expansion, whereas only male Lgals3-R200S did in this study. The effect size of the increases in cortical bone size was greater in Lgals3-KO females than males [5]. The similarities between Lgals3-R200S and Lgals3-KO mice (i.e., increased trabecular bone mass in female mice and increased cortical bone expansion in males) likely reflect the role of extracellular Gal-3 loss in increasing bone mass. Conversely, the differences between the two models (tissue stiffness and lack of female cortical bone expansion) could reflect the role of intracellular Gal-3.
The female dominance of the cortical bone expansion in Lgals3-KO mice was further supported using a separate Lgals3 null allele (Lgals3-Δ), where females once again had significantly increased trabecular and cortical bone mass at 36 weeks, but male Lgals3-Δ had slight reductions in both cortical and trabecular bone mass. The apparent sex-dependency of the bone phenotype was most likely due to diminished bone mass accrual in Lgals3-KO males before 12 week of age [5,41], which led us to speculate that Lgals3-KO mice might have reduced androgen-induced cortical bone expansion during puberty [42].
However, in Lgals3-R200S mice, we observed a male dominant phenotype in cortical bone expansion. Our gonadectomy study suggested that global loss of Gal-3 may lead to reduced bioavailability of androgens [6]. This would reduce the ability of androgen to support bone mass accrual during puberty [42]. Altered sexhormone regulation in the Lgals3-KO mother during fetal development might also explain why a different skeletal phenotype (increased age-related bone loss) has been reported when comparing Lgals3-KO mice to litters of background matched wildtype-mice [43]. Studies looking at systemic changes in hormones and growth factors in Lgals3-KO and Lgals3-R200S mice would help answer this question. In addition, conditional knockout of Lgals3 at later developmental stages, such as pre-and postpuberty or in aged mice, and studies of pre-and postpubescent Lgals3-R200S mouse cortical bone growth, will further clarify the timing of when extracellular Gal-3 affects bone mass expansion. Values are expressed as mean AE SEM (n = 7-14); Holm-Sidak post-hoc analysis adjusted P-values compared with wild-type; bold values highlight *P < 0.05.