Targeting Longevity Gene SLC13A5: A Novel Approach to Prevent Age-Related Bone Fragility and Osteoporosis

Reduced expression of the plasma membrane citrate transporter SLC13A5, also known as INDY, has been linked to increased longevity and mitigated age-related cardiovascular and metabolic diseases. Citrate, a vital component of the tricarboxylic acid cycle, constitutes 1–5% of bone weight, binding to mineral apatite surfaces. Our previous research highlighted osteoblasts’ specialized metabolic pathway facilitated by SLC13A5 regulating citrate uptake, production, and deposition within bones. Disrupting this pathway impairs bone mineralization in young mice. New Mendelian randomization analysis using UK Biobank data indicated that SNPs linked to reduced SLC13A5 function lowered osteoporosis risk. Comparative studies of young (10 weeks) and middle-aged (52 weeks) osteocalcin-cre-driven osteoblast-specific Slc13a5 knockout mice (Slc13a5cKO) showed a sexual dimorphism: while middle-aged females exhibited improved elasticity, middle-aged males demonstrated enhanced bone strength due to reduced SLC13A5 function. These findings suggest reduced SLC13A5 function could attenuate age-related bone fragility, advocating for SLC13A5 inhibition as a potential osteoporosis treatment.


Introduction
The SLC13A5 gene is the mammalian orthologue of the I'm not dead yet (Indy) gene in Drosophila melanogaster [1].In mammals, the solute carrier family 13 member 5 (SLC13A5) protein, also called sodium-dependent citrate transporter (NaCT), is an electrogenic, sodium coupled tricarboxylate plasma membrane transporter with a preference for citrate [2].Slc13a5 knockout in mice protected high-fat-diet-fed and aged animals from adiposity and insulin resistance [3] and lowered blood pressure through sympathetic inhibition [4].Thus, SLC13A5 is a promising therapeutic target for metabolic-associated fatty liver disease (MAFLD) and associated cardiometabolic diseases [5][6][7].
The SLC13A5 substrate citrate is essential for the formation, structuring, and stabilization of apatite crystals in bone, as citrate comprises ~1.6% of the bone content, and about 80% of the total body citrate resides in bone [15,16].Furthermore, SLC13A5 is expressed in osteoblasts [9,17] and affects osteogenic differentiation of human mesenchymal stem cells into osteoblasts [17].Osteoblasts take up citrate from circulation through SLC13A5 [9] but can also produce their own citrate for bone formation in the TCA [16], and several dietary supplement studies showed the beneficial effects of a citrate supplementation diet on osteoporosis [18].Overall, there is strong evidence that SLC13A5 may affect bone health in human [8,9].However, due to the substantial species differences between mouse and human, translational studies are urgently needed to understand the role of SLC13A5 in human bone development and diseases.
Causal inference tools such as Mendelian randomization [19] (MR) or the recently published SynTrial workflow [20] are robust and accessible tools to examine the causal relationship between an exposure variable and an outcome from GWAS summary statistics as well as from biobank data at the individual level.Specifically, MR leverages randomly allocated genetic variants as instrumental variables for studying the effect of varying an exposure.The random allocation of genetic variants at conception means that this paradigm is less vulnerable to the environmental confounding and reverse causation bias that can hinder causal inference in traditional epidemiological study designs.A more recent extension of MR allows it to be applied to study the effects of drug target perturbation [21,22].
Our very recent study using MR for SLC13A5 already identified 13 uncorrelated SNPs as instruments for SLC13A5 inhibition and found associations between genetically proxied SLC13A5 inhibition and improved kidney function as well as with higher plasma calcium levels and lower fasting glucose [23].We here used MR and those 13 SNPs as an instrument to investigate the role of SLC13A5 in human bone health.Furthermore, in order to establish a connection between the bone-related data obtained from our current and previously published studies conducted on young Slc13a5-deficient mice and our human genetic analysis, we conducted further investigations on bones obtained from middle-aged mice.These middle-aged mice are likely to be more representative of the population found in human genomic databases.

Mendelian Randomization
Mendelian randomization (MR) is a powerful method used in epidemiology and genetics leveraging genetic variants to act as natural experiments and allowing scientists to explore causal relationships between exposures and outcomes, providing valuable insights into disease mechanisms and potential therapeutic targets [21,22].
For genome-wide Mendelian randomization experiments, a p-value threshold of 5 × 10 −8 was applied to the exposure SNPs.Independent SNPs were identified by performing linkage disequilibrium (LD) clumping to ensure that genetic instruments were minimally correlated (r 2 < 0.001) based on the European reference panel from the 1000 Genomes Project (https://www.internationalgenome.org/home, accessed on 8 May 2020) using the ld_clump() function of the R package ieugwasr (https://github.com/MRCIEU/ieugwasr,published on 8 May 2020), which provides a wrapper around PLINK [24].For drug-target MR experiments, the 13 significant and independent SNPs were used as instruments [23].
The MR-Rücker framework [25] was used to decide whether IVW or MR-Egger is best supported by the data.Otherwise, the inverse-variance weighted (IVW method) for two SNPs or the Wald ratio method (one SNP) were used.All methods were implemented in the TwoSampleMR package [26].

Causal Biomarker Analysis
Complementary to MR analysis, we performed a causal biomarker analysis using synthetic clinical trial analysis, leveraging the rapidly expanding medical and genetic databases, computer modeling and simulation to replicate the processes of traditional clinical trials to get novel insights into disease mechanisms and potential therapeutic targets.Causal biomarker analysis, SynTrial, is described in detail in Baukmann et al. [20].
Using the rich information made available by the UK Biobank project [27], we defined osteoporosis cases as individuals that had reported ICD-10 M80 ("Osteoporosis with pathological fracture"), M81 ("Osteoporosis without pathological fracture"), or self-reported osteoporosis.Cases and controls were filtered for European ancestry ("White", "British", "Irish", and "Any other white background"), and individuals with missing age and sex information were discarded.Controls were then matched to the same number of cases based on age and sex.
Animals were euthanized by isoflurane inhalation and subsequent cervical dislocation.All breeding and experiments were performed according to Institutional Animal Care and Use Committee (IACUC) of the Johns Hopkins University and University of Maryland, Baltimore, guidelines and approved protocols.

Serum Analysis
Serum citrate (MAK333) and calcium (MAK022) levels were measured using a commercially available colorimetric assay and according to manufacturer's protocol (Millipore Sigma, Saint Louis, MO, USA).

Whole Bone Biomechanical Testing
To analyze the biomechanical properties of the femur, a low-force mechanical testing system (TA ElectroForce 3200 Series 3-Point Bending device, New Castle, DE, USA) and mat-lab software (version R2023b) was used.At dissection, the femur was harvested and frozen in −20 • C wrapped in a PBS-soaked gauze to keep it moist.The day before the test, specimens were thawed to room temperature in PBS, the mechanical properties of the femoral midshaft were assessed by 3-point bending by applying a flexion moment in the anterior-posterior plane with a span length of 7 mm.Preload was applied to the bones before testing to "train" the tissues.The force was applied at a constant displacement rate (0.05 mm/s) for 100 s, generating 2000 data points.The obtained force displacement data were used to determine yield load, failure load, stiffness, work-to-failure, and post-yield displacement.The material properties (Young's modulus, stress, strain, and modulus of toughness) were obtained by adjusting the force displacement data for the appropriate femoral mid-shaft area moment of inertia, as measured by micro-CT.

Mendelian Randomization
Citrate is the main substrate for SLC13A5 and is proposed to affect overall citrate homeostasis.Gill et al. [23] described genetic variants at the SLC13A5 locus which are robustly associated with circulating citrate levels at genome-wide significance, and which served as plausible genetic instruments for studying its effects in our study.We applied these 13 uncorrelated variants as instruments in drug target MR analyses.
We conducted two MR approaches: first, a genome-wide approach using all genomewide significant genetic variants, and then a drug target approach in which we only used SNPs in the SLC13A5 locus.In both cases, we utilized summary statistics on plasma citrate levels from Nightingale Health as the exposure and summary statistics on osteoporosis diagnosed by clinicians as well as on self-reported osteoporosis, both provided by the UK Biobank, as outcome.

Figure 1.
Mendelian randomization estimate for the causal effect of citrate on hospital-diagnosed and self-reported osteoporosis, restricted to variants in the SLC13A5 gene and genome-wide, respectively.Causal effects were estimated using the inverse-variance weighted method.Green data points represent significant effects without pleiotropy.Estimates are scaled per 1 standard deviation (SD) increase in plasma citrate, i.e., every 1-SD higher genetically proxied plasma citrate through SLC13A5 inhibition was associated with a 0.0034 lower log odds ratio for self-reported osteoporosis.

Causal Biomarker Analysis
Complementary to MR analysis, we performed a causal biomarker analysis using a previously developed synthetic clinical trial (SynTrial) workflow [20] to osteoporosis (Figure S1).Using UK Biobank data, we identified 9828 individuals with osteoporosis and used regression modelling to investigate the effect of 321 candidate predictive traits.A total of 89 traits significantly predicted osteoporosis with a Bonferroni-corrected significance threshold of p < ɑ/n = 0.05/321 (Table S1).After removing collinear traits, we applied drop-one analysis to compare all possible models that can be constructed by dropping a single model term and evaluating its impact on the regression model.The analysis revealed that three traits explain unique variance in osteoporosis status to a Bonferroni-corrected significance threshold of p < ɑ/n = 0.05/14 (Table S2).Propensity score analysis is a technique for estimating the treatment effects on an outcome independent of covariates.We employed propensity score stratification using the propensity function of Imai and van Dyk [29] to estimate treatment effects on osteoporosis independent of age, sex, BMI, and the other two traits, respectively.Heel BMD (estimate = -5.2377,p < 2.314 × 10 -177 ), vitamin D (estimate = 0.0127, p < 2.323 × 10 -69 ) and neutrophil/lymphocyte ratio (estimate = 0.1446, p < 2.546 × 10 -30 ) were causal traits (Table 1).As expected, Heel BMD is a highly significant causal protective factor.Remarkably, like in MR analysis, there is no significant impact of citrate in general in causal biomarker analysis (Table S2).Causal effects were estimated using the inverse-variance weighted method.Green data points represent significant effects without pleiotropy.Estimates are scaled per 1 standard deviation (SD) increase in plasma citrate, i.e., every 1-SD higher genetically proxied plasma citrate through SLC13A5 inhibition was associated with a 0.0034 lower log odds ratio for self-reported osteoporosis.

Causal Biomarker Analysis
Complementary to MR analysis, we performed a causal biomarker analysis using a previously developed synthetic clinical trial (SynTrial) workflow [20] to osteoporosis (Figure S1).Using UK Biobank data, we identified 9828 individuals with osteoporosis and used regression modelling to investigate the effect of 321 candidate predictive traits.A total of 89 traits significantly predicted osteoporosis with a Bonferroni-corrected significance threshold of p < Mendelian randomization estimate for the causal effect of citrate on hospital-diagnosed and self-reported osteoporosis, restricted to variants in the SLC13A5 gene and genome-wide, respectively.Causal effects were estimated using the inverse-variance weighted method.Green data points represent significant effects without pleiotropy.Estimates are scaled per 1 standard deviation (SD) increase in plasma citrate, i.e., every 1-SD higher genetically proxied plasma citrate through SLC13A5 inhibition was associated with a 0.0034 lower log odds ratio for self-reported osteoporosis.

Causal Biomarker Analysis
Complementary to MR analysis, we performed a causal biomarker analysis using a previously developed synthetic clinical trial (SynTrial) workflow [20] to osteoporosis (Figure S1).Using UK Biobank data, we identified 9828 individuals with osteoporosis and used regression modelling to investigate the effect of 321 candidate predictive traits.A total of 89 traits significantly predicted osteoporosis with a Bonferroni-corrected significance threshold of p < ɑ/n = 0.05/321 (Table S1).After removing collinear traits, we applied drop-one analysis to compare all possible models that can be constructed by dropping a single model term and evaluating its impact on the regression model.The analysis revealed that three traits explain unique variance in osteoporosis status to a Bonferroni-corrected significance threshold of p < ɑ/n = 0.05/14 (Table S2).Propensity score analysis is a technique for estimating the treatment effects on an outcome independent of covariates.We employed propensity score stratification using the propensity function of Imai and van Dyk [29] to estimate treatment effects on osteoporosis independent of age, sex, BMI, and the other two traits, respectively.Heel BMD (estimate = -5.2377,p < 2.314 × 10 -177 ), vitamin D (estimate = 0.0127, p < 2.323 × 10 -69 ) and neutrophil/lymphocyte ratio (estimate = 0.1446, p < 2.546 × 10 -30 ) were causal traits (Table 1).As expected, Heel BMD is a highly significant causal protective factor.Remarkably, like in MR analysis, there is no significant impact of citrate in general in causal biomarker analysis (Table S2)./n = 0.05/321 (Table S1).After removing collinear traits, we applied drop-one analysis to compare all possible models that can be constructed by dropping a single model term and evaluating its impact on the regression model.The analysis revealed that three traits explain unique variance in osteoporosis status to a Bonferronicorrected significance threshold of p < Metabolites 2023, 13, x FOR PEER REVIEW Figure 1.Mendelian randomization estimate for the causal effect of citrate o and self-reported osteoporosis, restricted to variants in the SLC13A5 gene a spectively.Causal effects were estimated using the inverse-variance weighte points represent significant effects without pleiotropy.Estimates are scaled pe (SD) increase in plasma citrate, i.e., every 1-SD higher genetically proxied p SLC13A5 inhibition was associated with a 0.0034 lower log odds ratio for self-r

Causal Biomarker Analysis
Complementary to MR analysis, we performed a causal biomarker viously developed synthetic clinical trial (SynTrial) workflow [20] to oste Using UK Biobank data, we identified 9828 individuals with osteop gression modelling to investigate the effect of 321 candidate predictiv traits significantly predicted osteoporosis with a Bonferroni-corrected old of p < ɑ/n = 0.05/321 (Table S1).After removing collinear traits, w analysis to compare all possible models that can be constructed by model term and evaluating its impact on the regression model.The an three traits explain unique variance in osteoporosis status to a Bonfe nificance threshold of p < ɑ/n = 0.05/14 (Table S2).Propensity score an for estimating the treatment effects on an outcome independent of ployed propensity score stratification using the propensity function o [29] to estimate treatment effects on osteoporosis independent of age other two traits, respectively.Heel BMD (estimate = -5.2377,p < 2.314 (estimate = 0.0127, p < 2.323 × 10 -69 ) and neutrophil/lymphocyte ratio ( < 2.546 × 10 -30 ) were causal traits (Table 1).As expected, Heel BMD is causal protective factor.Remarkably, like in MR analysis, there is no s citrate in general in causal biomarker analysis (Table S2)./n = 0.05/14 (Table S2).Propensity score analysis is a technique for estimating the treatment effects on an outcome independent of covariates.We employed propensity score stratification using the propensity function of Imai and van Dyk [29] to estimate treatment effects on osteoporosis independent of age, sex, BMI, and the other two traits, respectively.Heel BMD (estimate = −5.2377,p < 2.314 × 10 −177 ), vitamin D (estimate = 0.0127, p < 2.323 × 10 −69 ) and neutrophil/lymphocyte ratio (estimate = 0.1446, p < 2.546 × 10 −30 ) were causal traits (Table 1).As expected, Heel BMD is a highly significant causal protective factor.Remarkably, like in MR analysis, there is no significant impact of citrate in general in causal biomarker analysis (Table S2).In a recent study by Dirckx et al., global or osteoblast-specific Slc13a5 deletion caused increased mineral citrate levels and elicited mineralization defects resulting in reduced femur cortical thickness with increased fragility [9], which was in contrast to the genetic data showing protective effects on osteoporosis.However, these results were obtained in young mice (6 and 10 weeks old, respectively) that had not yet reached their peak bone mass.To elucidate age-related effects of Slc13a5 deletion on bone, we phenotyped one-year-old male and female skeletal-specific knockout mice using the osteocalcin-cre driver strain, further referred to as Slc13a5 cKO (Figure S2A for efficient recombination of Slc13a5 in full bone RNA extracts) and compared them to control littermates.Although 1-year-old mice are only considered middle-aged, we did observe significant age-related bone loss in our model at this age (Figure S2B).
In both middle-aged female (Figure 2A,B) and male (Figure 3A,B) mice, we did not observe changes in serum citrate and calcium levels between controls and mutants.In both female (Figure 2C,D) and male mice (Figure 3C,D), age-related weight gain and adult growth was unaffected, as bodyweight and femur length did not differ between control and Slc13a5 cKO mice.In female mice, detailed micro-CT analysis revealed that trabecular bone mass and bone mineral density (BMD) did not differ between control and Slc13a5 cKO mice (Figure 2E-J), and was in line with what was observed in 10-weekold female mice [9], though a trend towards increased bone volume/tissue volume was observed in middle-aged Slc13a5 cKO females (Figure 2F,G).Interestingly, cortical thickness, which was reduced by 20% in 10-week-old female Slc13a5 cKO mice [9], was normal in oneyear-old female Slc13a5 cKO mice (Figure 2K) while the width of bone and medullary space remained unchanged (Figure 2L-N).However, similar to 10-week-old female mice [9], the cortical TMD was significantly reduced in one-year-old female mice (Figure 2O).
In one-year-old male mice, we did not observe differences in trabecular BMD or bone mass (Figure 3E-J), despite a reduction in trabecular thickness (Figure 3H) between control and Slc13a5 cKO mice similar to 10-week-old male mice (Figure S3A-E).Cortical thickness was similar in both control and Slc13a5 cKO mice in the young and middle-aged group (Figures 3K and S3F), but femur width was significantly larger in middle-aged Slc13a5 cKO mice compared to younger ones and controls (Figures 3L-N and S3G-I).In young animals, we observed significant reductions in TMD (SI Figure S3J), which was normalized when the mice were middle-aged (Figure 3O).
Since our previous studies showed that altered citrate affects bone mass but more so bone quality [9], we next assessed bone mechanical properties by 3-point bending mechanical testing in young and middle-aged female and male mice.
In both, young (6 weeks, Table 2, left, and 10 weeks [9]) and middle-aged female Slc13a5 cKO mice (Table 2, right), the femurs were more elastic (assessed using reduced Young's modulus, reduced ultimate stress, and increased ultimate strain) than their control littermates.However, in young mice (6 weeks, Table 2, left, and 10 weeks [9]), increased elasticity was associated with a higher fragility as the ultimate moment (or load to fracture) was significantly reduced.Presumably, due to the substantially increased elasticity in the middle-aged female Slc13a5 cKO mice (Table 2, right) and the reverted cortical thickness, the ultimate moment (or load to fracture) did not differ from their control littermates.In young male mice, we did not observe differences in mechanical strength (Table 3, left) while middle-aged male mice even showed a significantly increased ultimate moment (Table 3, right), which suggests that the bones were less fragile in the middle-aged mutants versus controls.sexes, our data suggest that Slc13a5 deficiency in middle-aged mice is rather beneficial compared to what was observed in young and growing mice with respect to their resistance to fracture.These findings align with the MR results (Figure 1).In summary, there are several structural differences between young and middle-aged Slc13a5 cKO mice, and these differences also seem to be sex dependent.However, in both sexes, our data suggest that Slc13a5 deficiency in middle-aged mice is rather beneficial compared to what was observed in young and growing mice with respect to their resistance to fracture.These findings align with the MR results (Figure 1).

Discussion
Using Mendelian randomization (MR) with human data and complementary animal studies, we gained novel mechanistic insights for citrate transporter SLC13A5 and its role in bone diseases.The study suggests that reduced SLC13A5 function may attenuate age-related bone fragility and underscores the potential of pharmacological inhibition of SLC13A5 as an approach for the treatment of osteoporosis.
A recent study by Gill et al. [23] identified genetic variants at the SLC13A5 locus which are robustly associated with circulating citrate.Therefore, citrate represents a biologically plausible biomarker to estimate SLC13A5 activity, and these variants were used as genetic instruments for drug target MR analyses.Using drug target MR, we showed that genetically proxied SLC13A5 inhibition is associated with a lower risk of both clinically diagnosed and self-reported osteoporosis in UK Biobank participants.The association was not attributable to altered citrate levels more generally, despite greater statistical power of this measurement.In accordance with the results of the genome-wide MR experiment, circulating citrate was not a causal factor for osteoporosis in the SynTrial.This observation may suggest that the effect on osteoporosis is mediated through SLC13A5 and citrate uptake by the relevant cells independently of actions on plasma citrate levels, or that plasma citrate is a heterogeneous trait that is affected through several distinct pathways, and that at least some of these may be beneficial for bones.Similar observations in MR studies were described for the association of lipids and cancer [30,31].Variants in the HMGCR locus were associated with breast cancer outcomes, but not with genome-wide variants for LDL-cholesterol, indicating there may be an off-target non-LDL-C-based mechanism regulated by HMGCR.
In our causal biomarker analysis using a SynTrial method, we identified heel bone mineral density (Heel BMD), vitamin D, and neutrophil/lymphocyte ratio (NLR) as causal traits for osteoporosis.Previous MR studies showed that bone mineral density is a highly significant causal protective factor, is the gold standard for clinical osteoporosis assessment, and, therefore, serves as a positive control of our method [32,33].In contrast, epidemiological as well as genetic data for vitamin D are conflicting, despite widespread prescription in osteoporosis prevention and treatment [33,34].The discrepancies may result from heterogeneity of study populations and widely used vitamin D supplementation.Indeed, in the UK Biobank, 21.34% of osteoporosis cases were on vitamin D supplements, compared to 6.91% in control persons.The impact of NLR on osteoporosis found in our synthetic clinical trial confirms data from several prospective and cross-sectional clinical studies in osteoporosis patients [35][36][37].However, no novel traits could be identified.
Several citrate dietary supplement studies in patients showed beneficial effects on osteoporosis [18] which was not supported by our genetic data analysis.In fact, our study did not reveal any beneficial effects of altered citrate.Possibly, the discrepancy results from differences between small, lifelong changes in genetically predicted citrate levels and large changes in citrate availability on supplements.Furthermore, other variants causing changes in citrate plasma level may have possible opposite or compensatory effects on bone metabolism.Furthermore, few studies showed no or only limited effects of citrate diet on bone turnover [18].Finally, most clinical studies testing the impact of citrate diet on postmenopausal osteoporosis were performed in small cohorts and often combined with calcium and vitamin D supplementation [18].Therefore, further well-controlled and larger clinical studies are needed to understand the impact of citrate diet on bone health.
The particularly high hepatic SLC13A5 expression in humans (www.proteinatlas.org,accessed on 24 August 2023) suggests that the liver is an important organ for citrate elimination from circulation, and therefore, regulation of plasma citrate concentrations.The idea is supported by a previous metabolomics analysis in homozygous patients with lossof-function mutations who showed 3-fold higher plasma citrate concentrations compared with healthy control persons [38].In Slc13a5-deficient mice, plasma citrate was elevated in some but not in all studies by 10-100%, which may be explained by much lower hepatic SLC13A5 expression in mice than in human beings [3,9,39].Furthermore, modelling of plasma citrate flux into the liver based on pharmacokinetic data from healthy human subjects suggested that the liver is the major organ of citrate clearance from plasma in humans [40].Other studies showed that hepatic clearance can be further increased under certain conditions such as surgery causing hypocitricemia [41].However, the closely related transporters SLC13A2 and SLC13A3, mainly expressed in the kidney, may also contribute to citrate clearance.In Slc13a2 knockout mouse, urinary citrate excretion was increased with unchanged plasma citrate concentration [42].A recently reported human SLC13A3 variant with lower transporter function showed no effects on urinary citrate levels, but plasma citrate levels were not reported [43].Furthermore, SLC13A2 and SLC13A3 were not associated with changes in plasma citrate concentration in AstraZeneca PheWAS Portal [44].
The protective effects of altered SLC13A5 for osteoporosis proxied by genetic variants (SNPs) appear to contradict to previously published findings in Slc13a5-deficient mice which showed reduced bone mineral density and impaired bone mineralization leading to more fragile bones [8,9].The discrepancy might be explained by biological differences between small, lifelong changes in carriers of genetic variants in the UK Biobank and complete loss of function in Slc13a5-deficient mice.Impaired bone health was only observed in homozygous knockout, whereas heterozygous animals were unaffected.Furthermore, the time point of observations may play a role as well.UK Biobank data are collected in adults of an average age of 56 years, while Slc13a5-deficient mice have previously been studied at a young age.Indeed, a previous study showed negative effects on bone mineralization at 13 weeks but not at week 32 [8,9].Another mouse model of bone disease, osteogenesis imperfecta (OI), showed 2.5-fold increased SLC13A5 expression levels in bone and abnormal mineralization [45] and may provide another hint for beneficial effects of SLC13A5 inhibition on bone health.In line herewith, another recent publication showed that increased expression of Slc13a5 caused a progeria-like phenotype in both male and female mice with reduced bone density [46].
Intriguingly, our studies on one-year-old skeletal-specific Slc13a5-deficient mice reveal similarities with what was predicted with inhibitory SLC13A5 gene variants in humans.Although we only observed mild improvements regarding bone mass in middle-aged male and female Slc13a5 cKO mice, changes at the ultrastructural level determining bone quality were more pronounced.Indeed, while young female Slc13a5 cKO mice (6 weeks and 10 weeks) show reduced resistance to fracture, middle-aged female Slc13a5 cKO mice appear to have changed their material properties and developed enhanced elasticity in their bones, leading to a resistance to fracture that is comparable to their wildtype littermates.Young male Slc13a5 cKO mice had a similar bone strength to their control littermates, while middle-aged Slc13a5 cKO mice had a significantly increased ultimate moment compared to controls.As such, SLC13A5 appears to differentially affect young versus middle-aged and male versus female mice.The mechanisms behind the differential phenotype in one-year-old male and female Slc13a5 cKO mice remain speculative and are presumably attributed to physiological [18] and endocrine differences [47] in males and females that accumulate with age and differentially affect both citrate homeostasis and bone quality in both sexes.However, despite the morphological differences, in both sexes, the mechanical properties become more favorable with age upon Slc13a5 deletion and are, therefore, in agreement with our genetic data obtained in human subjects.We previously reported that young Slc13a5-deficient mice showed increased levels of mineral citrate which proved to be detrimental for bone mineralization and strength during growth [9].However, it has been shown that osteoporosis was associated with reduced citrate levels in plasma and bone [48] in both human and mice.Therefore, we speculate that the increased citrate accumulation in bone minerals from Slc13a5 cKO mice serves a more protective role against age-induced bone fragility in the one-year-old study group.It is noteworthy that these middle-aged Slc13a5 cKO mice were mutant since birth.Follow up studies with inducible Slc13a5 knockouts at an older age or administration of the SLC13A5 inhibitor in osteoporotic mice will provide further valuable insights into the beneficial effects of and mechanisms behind SLC13A5 inhibition in osteoporosis management.
Patients with loss of function mutations are not routinely tested for bone disease.Yet, investigations on non-neurologic health of patients with autosomal recessive SLC13A5 Citrate Transporter (NaCT) Disorder reported tooth abnormalities caused by hypomineralized dentin and enamel [9] in most patients without overt skeletal disease (e.g., deformities or spontaneous fractures), except for a trend of slower growth [49].Our study provides an impetus studying implication of SLC13A5 loss of function in humans in more detail.An ongoing NHS study (NCT04681781) may provide further insights.However, signifi-cant species differences in target biology may result in different phenotypes in humans and mice with genetic Slc13a5 deficiencies.In particular, differences in transport kinetics could lead to different outcomes associated with Slc13a5-deficient mice or mutations or other variants in human Slc13a5 [13,50].The high affinity/low capacity transporter in mice is completely saturated under physiological plasma citrate concentrations of approximately 150-200 µM and cannot respond to further increases in circulating citrate, whereas the human low affinity/high capacity transporter is not saturated, and therefore, can respond to changes in citrate levels [2,51].Additionally, there are significant differences in gene expression level and tissue distribution.In humans, SLC13A5 is mainly expressed in the liver at levels several orders of magnitude higher than in all other tissues (https://gtexportal.org/home/gene/SLC13A5, accessed on 24 August 2023) whereas in mice, the transporter is mainly expressed in incisors, bone, brain, and testis [9].Data on expression in bone or bone cells in both species are limited; however, the available data indicate that bone expression of SLC13A5 appears to be higher in mice than in human beings (https://genevisible.com/tissues/MM/Gene%2520Symbol/Slc13a5, accessed on 24 August 2023), but it also changes during bone development [52].We have recently shown that in vitro mouse osteoblasts expression of SLC13A5 increases more than 20-fold over the course of osteoblast differentiation compared to only 3-fold in humans [9].
Overall, our data suggest that pharmacological SLC13A5 inhibition could have utility in preventing or treating osteoporosis.Furthermore, epidemiological studies suggest a link between metabolic diseases, such as non-alcoholic fatty liver disease, and osteoporosis [53].

Conclusions
Overall, based on the presented data, pharmacological inhibition of SLC13A5 function may be considered as a promising new approach to treat osteoporosis.Although several drugs are already approved for osteoporosis with significant improvements in the last few years, there is still a high medical need for well-tolerated and long-term-efficient drugs [54][55][56].Interestingly, there are several epidemiological studies linking metabolic diseases in particular non-alcoholic fatty liver disease and osteoporosis [53,57].Those studies showed consistently that the prevalence and risk of osteoporosis or osteoporotic fractures were significantly associated with NAFLD in men and women.As SLC13A5 inhibitors are under development for fatty liver and NASH [7], synergies may be possible for patients with metabolic diseases and osteoporosis co-morbidity.However, further mechanistic studies are needed to better understand the impact of SLC13A5 inhibition in human bone metabolism.

Figure 1 .
Figure 1.Mendelian randomization estimate for the causal effect of citrate on hospital-diagnosed and self-reported osteoporosis, restricted to variants in the SLC13A5 gene and genome-wide, respectively.Causal effects were estimated using the inverse-variance weighted method.Green data points represent significant effects without pleiotropy.Estimates are scaled per 1 standard deviation (SD) increase in plasma citrate, i.e., every 1-SD higher genetically proxied plasma citrate through SLC13A5 inhibition was associated with a 0.0034 lower log odds ratio for self-reported osteoporosis.

15 Figure 1 .
Figure 1.Mendelian randomization estimate for the causal effect of citrate on hospital-diagnosed and self-reported osteoporosis, restricted to variants in the SLC13A5 gene and genome-wide, respectively.Causal effects were estimated using the inverse-variance weighted method.Green data points represent significant effects without pleiotropy.Estimates are scaled per 1 standard deviation (SD) increase in plasma citrate, i.e., every 1-SD higher genetically proxied plasma citrate through SLC13A5 inhibition was associated with a 0.0034 lower log odds ratio for self-reported osteoporosis.

Table 1 .
Results from Propensity Score Analysis (PSA), the last step of the Synthetic Clinical Trial workflow.

Table 1 .
Results from Propensity Score Analysis (PSA), the last step of the Synthetic Clinical Trial workflow.

Table 1 .
Results from Propensity Score Analysis (PSA), the last step of the S workflow.

Table 1 .
Results from Propensity Score Analysis (PSA), the last step of the Synthetic Clinical Trial workflow.Age-Induced Changes in Bone Morphology and Mechanical Properties in Skeletal-Specific Slc13a5 cKO Female and Male Mice

Table 2 .
Table representing all 3-point bending mechanical testing parameters in 1-year-old female control and Slc13a5 cKO mice (right, n = 8-9) compared to 6-week-old female control and Slc13a5 cKO mice (left, n = 8-11).Parameters that are significantly increased or decreased are represented in orange and green, respectively, Student's t-test versus control.

Table 3 .
Table representing all 3-point bending mechanical testing parameters in 1-year-old male control and Slc13a5 cKO mice (right, n = 9-11) compared to 10-week-old male control and Slc13a5 cKO mice (left, n = 8-9).Parameters that are significantly increased or decreased are represented in orange and green, respectively, Student's t-test versus control.