Osteoblast-intrinsic defect in glucose metabolism impairs bone formation in type II diabetic mice

Skeletal fragility is associated with type 2 diabetes mellitus (T2D), but the underlying mechanism is not well understood. Here, in a mouse model for youth-onset T2D, we show that both trabecular and cortical bone mass are reduced due to diminished osteoblast activity. Stable isotope tracing in vivo with 13C-glucose demonstrates that both glycolysis and glucose fueling of the TCA cycle are impaired in diabetic bones. Similarly, Seahorse assays show suppression of both glycolysis and oxidative phosphorylation by diabetes in bone marrow mesenchymal cells as a whole, whereas single-cell RNA sequencing reveals distinct modes of metabolic dysregulation among the subpopulations. Metformin not only promotes glycolysis and osteoblast differentiation in vitro, but also improves bone mass in diabetic mice. Finally, targeted overexpression of Hif1a or Pfkfb3 in osteoblasts of T2D mice averts bone loss. The study identifies osteoblast-intrinsic defects in glucose metabolism as an underlying cause of diabetic osteopenia, which may be targeted therapeutically.


Introduction 35
Bone quality is maintained by the coordinated functions of osteoblasts and osteoclasts. During 36 bone formation, polarized osteoblasts secrete extracellular matrix (ECM), consisting of type I 37 collagen and non-collagenous matrix proteins, that mineralizes over time 1,2 . The production of 38 ECM by osteoblasts is an energy-demanding process and likely requires metabolic 39 reprogramming, but the metabolic regulation of osteoblast differentiation and function is largely 40 unknown 3-5 . Interestingly, prominent signaling pathways that promote or suppress osteogenesis 41 alter glucose metabolism. For instance, parathyroid hormone (PTH), Wnt, insulin-like growth 42 factors (Igf) and Hif1a stimulate glycolysis while promoting osteogenesis, whereas Notch 43 signaling suppresses glycolysis and osteoblast differentiation in mesenchymal progenitors [6][7][8][9][10] . In 44 addition, a number of studies have identified glucose as the main energy substrate for 45 osteoblasts [11][12][13][14] . Thus, accumulating evidence to date supports an important role of glucose 46 metabolism in the regulation of osteoblast development and function. 47 48 Type 2 diabetes (T2D) is characterized by insulin insensitivity in target tissues resulting in 49 hyperglycemia. Adults with T2D have increased fracture risk despite having normal or increased 50 areal bone mineral density (aBMD) [15][16][17] . The poor bone quality has been often attributed to 51 abnormal accumulation of advanced glycation end products in the bone matrix, but could also 52 result from reduced bone turnover as indicated by suppression of both bone resorption and 53 formation in T2D [18][19][20] . Youth-onset T2D has emerged as a significant and increasing health 54 burden in adolescents and young adults 21 . Distinct from adult T2D, T2D in youth has been 55 reported to reduce bone mass due to impaired bone anabolism 22 . Overall, both types of T2D 56 diminish bone formation, but the underlying mechanisms are not fully understood. 57 The distal end of the femur was scanned by μCT (μCT45; SCANCO Medical) at 4.5 μm 111 isotropic voxel size. For trabecular bone analyses, regions of interest were selected at 0.45-2.25 112 mm below the growth plate. For cortical bone, a total of 70 slices at the femur midshaft located 113 at 5.4 mm away from the distal growth plate were analyzed. Parameters computed from these 114 data included bone volume (BV), total volume (TV), BV/TV, trabecular bone thickness (Tb.Th), 115 number (Tb.N), separation (Tb. Sp), and connectivity density (Conn. Dens) at the distal femoral 116 metaphysis and total area (TA), bone area (BA), BA/TA and cortical thickness (Ct.th) at the mid-117 diaphysis of the femur. 118 119 Dynamic histomorphometry was conducted by sequential injections of calcein green (5mg/kg 120 body weight (bw)) and alizarin red (15mg/kg bw) at 7 and 2 days, respectively, prior to sacrifice. 121 Femurs or tibias were fixed with 4% PFA in PBS for 48 hours and then kept in 70% ethanol with 122 protection from light; the fixed bones were switched to 30% sucrose in PBS overnight before 123 OCT embedding and cryostat sectioning. Longitudinal 10 µm cryostat sections of the 124 undecalcified bones were collected with Cryofilm type II membrane (Section Lab, Co. Ltd., 125 8 The supernatant (40 µl) was collected for liquid chromatography-mass spectrometry (LC-MS) 134 analysis. For bone metabolite extraction, frozen bones (1 tibia and 1 femur) were transferred to 135 2 ml round-bottom Eppendorf Safe-Lock tubes on dry ice. Samples were then ground into 136 powder with a cryomill machine (Retsch, Germany) for 30 s at 25 Hz, and maintained at a cold 137 temperature using liquid nitrogen. For every 20 mg tissues, 800 μl of −20 °C extraction solvent 138 was added to the tube, vortexed for 10 s, and then centrifuged at 21,000 ×g for 20 min at 4 °C. 139 The supernatants (500 μl) were transferred to another Eppendorf tube, dried using a N2 140 evaporator (Organomation Associates, Berlin, MA), and redissolved in 50 μl of −20 °C 141 extraction solvent. Redissolved samples were centrifuged at 21,000 ×g for another 20 min at 142 4 °C. The supernatants were then transferred to plastic vials for LC-MS analysis. A procedure 143 blank was generated identically without tissue, and was used later to remove the background 144 ions. 145 146 For metabolite measurements by LC-MS, a quadrupole-orbitrap mass spectrometer (Q Exactive 147 Plus, Thermo Fisher Scientific, San Jose, CA) operating in negative mode was coupled to 148 hydrophobic interaction chromatography (HILIC) via electrospray ionization. Scans were 149 performed from m/z 70 to 1000 at 1 Hz and 140,000 resolution. LC separation was conducted on 150 a XBridge BEH Amide column (2.1 mm x 150 mm x 2.5 mm particle size, 130 Å pore size; 151 Water, Milford, MA) using a gradient of solvent A (20 mM ammonium acetate, 20 mM 152 ammounium hydroxide in 95:5 water: acetonitrile, pH 9.45) and solvent B (acetonitrile). Flow 153 rate was 150 mL/min. The LC gradient was: 0 min, 85% B; 2 min, 85% B; 3 min, 80% B; 5 min, 154 80% B; 6 min, 75% B; 7 min, 75% B; 8 min, 70% B; 9 min, 70% B; 10 min, 50% B; 12 min, 155 50% B; 13 min, 25% B; 16 min, 25% B; 18 min, 0% B; 23 min, 0% B; 24 min, 85% B. 156 Autosampler temperature was 5°C, and injection volume was 15 μL. For improved detection of 157 fructose-1,6-bisphosphate and 3-phosphoglycerate, selected ion monitoring (SIM) scans were 158 added. For 3-phosphoglycerate and fructose-1,6-bisphosphate, scans were performed from 180-159 190 m/z and 336-350 m/z, respectively, from 13-15 min of the 25-min gradient run with 70,000 160 resolution and maximum IT of 500 ms and AGC target of 3 x 10 6 . Data were analyzed using the 161 El-MAVEN (Elucidata) software 27 . Carbon Enrichment was calculated as follows: Carbon 162 Enrichment = (m0*0+m1*1+m2*2+….mn*n)/n. 163 164

BMSC isolation and purification 200
Tibias and femurs were dissected clean from adjacent tissues, and the epiphyses containing the 201 growth plates were excised off with scissors. The bone marrow cells were flushed out with a 202 . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted January 18, 2023. ;https://doi.org/10.1101https://doi.org/10. /2023 syringe filled with MEM-α media containing 10% FBS and then passed through a cell strainer 203 (35 µm). The cells for all four bones of one mouse were seeded into one 100 mm dish. The 204 culture medium (MEM-α containing 10% FBS) was changed every day after 4 days following 205 seeding, and the cells normally became confluent at day 6-8, at which time they were dissociated 206 using Collagenase II (4 mg/ml) and 0.05 % trypsin with EDTA for 5 mins at 37°C. 207 Hematopoietic lineage cells were immuno-depleted using the MACS technique (Miltenyi Biotec, 208 USA). Briefly, the cells were mixed with CD45 antibody-coated magnetic beads (CD45 209 MicroBeads, and then passed through a MACS LD column (Miltenyi Biotec, 042-901) on a MidiMACS TM separator (Miltenyi Biotec, attached to a multistand. 211 The purified cells were cultured and passaged once for osteogenic differentiation and metabolic 212 assays. 213 214 For osteoblast differentiation, when the BMSCs reached 100% confluency, MEM-α containing 4 215 mM ß-glycerol phosphate (Sigma, G9422) and 50 ug/ml ascorbic acid (Sigma, A4544) was 216 added and then changed daily. Cells after 4 or 7 days of differentiation were dissociated with 4 217 mg/ml collagenase I in PBS for 45 mins, then with 0.05% trypsin for 10 mins for RNA 218 extraction or metabolic assays. 219 220

In vitro metabolic assays 221
For Seahorse assays, BMSC purified with MACS were seeded at 2 × 10 5 cells/cm 2 into XF96 222 tissue culture microplates (Agilent) at 24 hours prior to experiments. Complete Seahorse medium 223 was prepared from Agilent Seahorse XF Base Medium (Agilent, 102353) containing 5.5 mM 224 glucose, 2 mM glutamine and 1 mM pyruvate, with pH 7.4. The cells were incubated in 180 μl 225 . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted January 18, 2023. ;https://doi.org/10.1101https://doi.org/10. /2023 complete Seahorse medium at 37°C for 1 hour in CO 2 -free incubator before measurements in 226  Statistics was conducted with unpaired Student's t-test, one-way ANOVA followed by Student's 268 t test between multiple groups or two-way ANOVA with either Sidak's multiple comparisons or 269 Fisher's LSD test, as indicated in the figure legends, by using Prism Software 9.0. p value < 0.05 270 is considered significant. All quantitative data are presented as mean ± SD. 271 . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted January 18, 2023. ;https://doi.org/10.1101https://doi.org/10. /2023 Results 273

Modeling youth-onset T2D in the mouse 274
To generate a mouse model for human youth-onset T2D, 6-week-old C57BL/6J male mice were 275 fed a high-fat diet (HFD) for 6 weeks and then injected with a low dose of streptozotocin (STZ) 276 once daily for three consecutive days, followed by continuous HFD feeding until harvest at 22 277 weeks of age (Fig. 1A). Random glucose levels were checked two weeks after the first STZ 278 injection and mice with a glucose level lower than 200 mg/dL were excluded from the study. At 279 harvest, the T2D mice had a significantly higher body weight due to increased fat mass with no 280 change in lean mass ( Fig. 1B, C, S1A). They exhibited notable hyperglycemia and significant 281 impairment in glucose handling when compared to the controls, as revealed by glucose tolerance 282 test (GTT) and insulin tolerance test (ITT) (Fig. 1D-I). Fasting serum insulin and Igf1 283 concentrations were significantly higher in the T2D mice, confirming that they were less 284 sensitive to insulin and reflecting an early phase of T2D ( We next examined bone phenotypes in the T2D mice. Dual-energy x-ray absorptiometry (DEXA) 291 at harvest detected a clear decrease in bone mineral density (BMD) of both whole body (minus 292 the head) and the hindlimb in comparison with the control (Fig. 2A). The body length from nose 293 tip to tail base and the femur length were slightly longer in T2D than normal, perhaps due to 294 . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted January 18, 2023. ; https://doi.org/10.1101/2023.01.16.524248 doi: bioRxiv preprint increased Igf1 levels (Fig. S1B, C). Imaging and quantification with µCT revealed that T2D 295 significantly reduced trabecular bone volume (BV) and its ratio over tissue volume (BV/TV) in 296 the femur, without altering the tissue volume (TV) itself ( Fig. 2B upper, Fig. 2C, Fig. S1D). The 297 trabecular bone loss was due to reduced trabecular number (Tb.N ) and connectivity (Conn. Dens) 298 concurrent with increased trabecular spacing (Tb. Sp) (Fig. S1E-G). Similarly, in cortical bone, 299 T2D decreased bone area relative to total area (BA/TA) as well as cortical bone thickness (Ct. Th) 300 at the diaphysis of the femur (Fig. 2B lower, Fig. 2D, Fig. S1H-I). Thus, like human patients, 301 mice with youth-onset T2D exhibit osteopenia. 302

303
We next investigated the cellular basis for diabetic osteopenia. Serum P1NP (procollagen type I 304 N-terminal propeptide) and CTX-1 (collagen type I C-telopeptide) levels were both lower in 305 T2D than control (Fig. 2E), indicating that T2D suppressed the overall bone turnover whereas 306 reduced bone formation was responsible for the net bone loss. To further quantify osteoblast 307 activity, we performed dynamic histomorphometry with dual dyes (calcein followed by alizarin 308 red) that labeled active bone-forming surfaces ( (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted January 18, 2023. ; https://doi.org/10. 1101/2023 To explore the metabolic basis for impaired osteoblast activity in T2D, we conducted stable 318 isotope tracing in bone with uniformly labeled 13 C-glucose ( 13 C 6 -Glc). Catabolism of 13 C 6 -Glc 319 was expected to produce fully labeled pyruvate Pyr(m+3) through glycolysis, which could then 320 convert to fully labeled lactate Lac(m+3) in the cytosol or enter mitochondria to produce m+2 321 labeled TCA cycle metabolites in the first cycle (Fig. 3A). For the experiment, 13 C 6 -Glc was 322 injected into conscious mice (22 wks of age) via tail vein at 60 mins before the mice were 323 between T2D and CTRL, while glucose was increased in T2D as expected. The fractional 326 enrichment of 13 C 6 -Glc in bone reached approximately 20% with no difference between T2D and 327 CTRL mice (Fig. 3C). However, among the bone metabolites detected, the labeling enrichments 328 of Glu(m+2) and Gln(m+2) were significantly reduced in T2D (Fig. 3C). Moreover, when 329 normalized to bone Glc(m+6), the relative enrichments of bone Pyr(m+3), Glu(m+2) and 330 Gln(m+2) were decreased in T2D (Fig. 3D). Further examination of the TCA cycle metabolites 331 in bone revealed substantial enrichment of the m+1 isotopomers, among which Suc(m+1), 332 Asp(m+1), Glu(m+1) and Gln(m+1) were diminished in T2D, with or without normalization to 333 Glc(m+6) in bone (Fig. S3A, B). Therefore, to capture the overall contribution of 13 C 6 -Glc we 334 calculated the total carbon enrichment ((m0*0+m1*1+m2*2+….mn*n)/n) for each of the TCA 335 cycle metabolites in bone. The result showed that 13 C enrichments in succinate, glutamate and 336 glutamine, with or without normalization to bone Glc(m+6), were significantly reduced in T2D 337 compared to the CTRL (Fig. 3E). The bone-intrinsic metabolic defect was further supported by 338 Western blot analyses showing a marked decrease in Glut1 and Hk2 in the bone extracts of T2D 339 . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted January 18, 2023. (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted January 18, 2023. ; https://doi.org/10.1101/2023.01.16.524248 doi: bioRxiv preprint within individual clusters detected notable differences in metabolic pathways between T2D and 363 CTRL. In particular, T2D suppressed the OXPHOS genes in Adipo-CAR clusters 1 and 5 as well 364 as osteoblast cluster 15, but increased them in Osteo-CAR cluster 0 and osteoblast cluster 3 ( Fig.  365 4C, D) (Supplementary Table S2). In addition, T2D increased glycolysis genes in cluster 3 but 366 suppressed them in cluster 5 (Supplementary Table S2). Thus, T2D appears to disrupt normal 367 energy metabolism among osteoblasts and other mesenchymal cells in the bone marrow. 368

369
To assess the actual changes in energy metabolism, we purified bone marrow stromal cells 370 (BMSC) from the central marrow for metabolic assays in vitro. Immuno-depletion with MACS 371 beads effectively eliminated CD45 + hematopoietic cells from the mesenchymal cells ( Fig. S4A, 372 B). In the T2D BMSC, both glucose consumption and lactate production were markedly reduced 373 ( Fig. 5A). Seahorse assays detected a significant reduction in both oxygen consumption rate 374 (OCR) and extracellular acidification rate (ECAR) in T2D BMSC (Fig. 5B, C). The ATP 375 production rate calculated from either glycolysis (glycoATP) or OXPHOS (mitoATP) was 376 reduced in the T2D cells (Fig. 5D). Thus, T2D suppresses both glycolysis and mitochondrial 377 respiration in BMSC. 378 379 We next evaluated the potential effect of bioenergetic defects on osteoblast differentiation in 380 T2D. We first confirmed that MACS-purified BMSC underwent robust osteoblast differentiation 381 in vitro. BMSC from normal mice exhibited no Alizarin red staining before differentiation (day 0) 382 but formed intensely stained nodules on day 4 and day 7 (Fig. S5A). qPCR showed that the 383 osteoblast makers Ibsp, Bglap2, Alpl, Spp1, Bmp2, Col1a1, Sp7, and Runx2 were greatly 384 induced after four days of differentiation, with some further upregulated on day 7 (Fig. S5B). 385 . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted January 18, 2023. ; https://doi.org/10.1101/2023.01.16.524248 doi: bioRxiv preprint Importantly, when cultured in parallel with the normal cells, T2D BMSC expressed markedly 386 lower levels of virtually all osteoblast markers examined on both day 4 and day 7 of 387 differentiation (Fig. 5E, F). The differentiation defect was further supported by reduced Alizarin 388 red staining of the minerals after 4 or 7 days of differentiation ( Figure 5G). Overall, the data 389 indicate that T2D causes cell-intrinsic defects in both metabolic and osteogenic properties of 390 BMSC. 391 392

Enhancing glycolysis with metformin ameliorates bone loss in T2D 393
To test if impaired glucose utilization drives osteopenia in T2D, we sought to boost glycolysis 394 and examine its effect on bone mass. Metformin is a widely prescribed anti-diabetic and has been 395 proposed to boost glycolysis through a variety of mechanisms 32-35 . We therefore administered 396 metformin to the T2D mice through drinking water and examined the effects on metabolic and 397 bone parameters (Fig. 6A). GTT and ITT showed that glucose tolerance and insulin sensitivity 398 were modestly improved by the metformin regimen although fasting glucose levels remained 399 higher than normal ( Fig. 6B-E). Metformin did not affect body weight, body composition or 400 circulating insulin levels in the T2D mice ( Fig. S6A-C). Importantly, µCT showed that 401 metformin notably increased trabecular bone mass (BV/TV) mainly due to increased trabecular 402 number (Tb. N) and connectivity (Conn. Dens) coupled with decreased trabecular spacing (Tb. 403 Sp), with no changes in trabecular thickness or cortical bone parameters (Fig. 6F, Fig. S6D, E). 404 Double labeling revealed that metformin increased mineralizing surfaces (MS/BS) without 405 altering the mineral apposition rate (MAR), resulting in a 50% increase in bone formation rate 406 (BFR) in the treated T2D mice (Fig. 6G, H, Supplemental Fig. S6F). Thus, metformin mitigates 407 bone loss in T2D by promoting bone formation. 408 . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted January 18, 2023. ; https://doi.org/10.1101/2023.01.16.524248 doi: bioRxiv preprint

409
To assess the potential effect of metformin on glucose metabolism in bone, we next performed 410 stable isotope tracing with 13 C 6 -Glc in the T2D mice with or without metformin treatment 411 immediately before harvest at 23 weeks of age (Fig. 6A). Both Pyr(m+3) and Lac (m+3)  412 normalized to Glc(m+6) in bone exhibited a trend of increase in response to metformin, but the 413 differences did not achieve statistical significance, likely due to insufficient power of the sample 414 size (Fig. S6G p=0.05). On the other hand, the total carbon enrichment of aspartate and 415 glutamate relative to Glc(m+6) was significantly increased by metformin, indicating increased 416 glucose entry into the TCA cycle in bone (Fig. 6I). The data, therefore, support that metformin 417 promotes bone glucose metabolism in T2D mice. 418 419 We next examined whether metformin has a direct effect on osteoblast-lineage cells. For 420 metabolic effects, BMSC from T2D mice were treated with metformin for 24 hours before 421 Seahorse assays. Metformin had negligible effect on OCR but significantly elevated ECAR (Fig.  422 6J, K). Moreover, in osteoblast differentiation assays, metformin increased the expression of 423 multiple osteoblast markers and key glycolysis genes in T2D BMSC (Fig. 6L, M). Similarly, 424 metformin notably improved the formation of mineralized nodules by the T2D cells following 425 osteoblast differentiation (Fig. 6N). Together, these findings indicate that metformin directly 426 enhances both glycolysis and osteoblast differentiation in T2D BMSC. 427 428

Genetic activation of glycolysis reduces bone loss in T2D 429
To corroborate the direct effect on bone, we utilized a transgenic mouse model that increased 430 glycolysis specifically in the osteoblast-lineage cells. The transcription factor Hif1α is known to 431 . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted January 18, 2023. ; https://doi.org/10.1101/2023.01.16.524248 doi: bioRxiv preprint stimulate the expression of numerous genes in the glycolytic pathway, thus promoting glycolysis. 432 Here, we generated mice with the genotype of Osx-rtTA; tetO-Cre; Hif1dPA (Hif1OE hereafter) 433 to allow for Cre-mediated expression of a degradation-resistant form of Hif1a (Hif1dPA)  bone area fraction (BA/TA) and thickness (Ct. Th) were all increased in Hif1OE (Fig. 7D, Fig.  443 S7A, B). Double labeling experiments detected notable increases in MS/BS and BFR/BS but not 444 MAR in trabecular bone (Fig. 7E, Fig. S7C). Overall, Hif1a overexpression in osteoblast-lineage 445 cells increases bone mass in T2D mice. 446

447
We further investigated the rate-determining steps in the glycolytic pathway, which could be 448 therapeutically targeted for promoting bone formation in T2D mice. Overexpression of Glut1, a 449 major glucose transporter in osteoblasts, has been shown to promote bone formation in normal 450 mice 36 . We, therefore, generated mice with the genotype of Osx-rtTA; TRE-Glut1 (Glut1OE) to 451 induce Glut1 overexpression in osteoblast-lineage cells with Dox. Both Glut1OE males and their 452 sex-matched littermates (Ctrl) missing one of the two transgenes were rendered diabetic and then 453 exposed to Dox according to the same regimen as described for the Hif1OE mice. Western blot 454 . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted January 18, 2023. ; https://doi.org/10. 1101/2023 analyses demonstrated successful overexpression of Glut1 in the bone shaft of long bones in 455 Glut1OE (Fig. S8A). However, µCT analysis detected no significant change in either trabecular 456 or cortical bone parameters between Glut1OE and Ctrl mice (Fig. S8B, C). Likewise, serum 457 P1NP and CTX-1 levels were similar between the genotypes (Fig. S8D). Thus, increased Glut1 458 expression does not improve osteopenia in T2D mice. 459 460 We next tested the potential effect of phosphofructokinase-2/fructose-2,6-bisphosphatase 461 (Pfkfb3). Pfkfb3 synthesizes fructose-2,6-bisphosphate (F2,6P2), a potent allosteric activator of 462 6-phosphofructo-1-kinase (Pfk1) which controls a rate-limiting step of glycolysis 37 . To this end, 463 we generated Osx-rtTA; TRE-Pfkfb3 mice (Pfkfb3OE) along with their control littermates (Ctrl) 464 missing at least one of the two transgenes. Cohorts of Pfkfb3OE and Ctrl mice were each divided 465 into a normal or T2D group. In the T2D group, the mice were rendered diabetic and then treated 466 with Dox for the remainder of the experiment, whereas those in the normal group were 467 maintained on regular chow before being treated with Dox in the same way (Fig. 8A). Upon 468 harvest of the mice, µCT analyses detected no difference in any of the trabecular bone 469 parameters between Pfkfb3OE and Ctrl mice in the normal group (Fig. 8B, C, Fig. S9A). 470 However, within the T2D group, Pfkfb3OE significantly increased trabecular bone volume (BV), 471 trabecular bone fraction (BV/TV), coupled with increased connectivity density (Conn. Dens), 472 increased trabecular number (Tb. N) and reduced trabecular spacing (Tb. Sp) (Fig. 8B, C, Fig.  473 S9A). Similarly, Pfkfb3 overexpression did not change the cortical bone parameters in the 474 normal group, but it increased cortical bone fraction (BA/TA) in the T2D group, thanks to the 475 correction of total cross-sectional area (TA) (Fig. 8D, Fig. S9B). Dynamic histomorphometry 476 revealed that Pfkfb3 overexpression in T2D mice increased bone formation rate (BFR) due to a 477 . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted January 18, 2023. ; https://doi.org/10.1101/2023.01.16.524248 doi: bioRxiv preprint marked increase in bone mineralizing surface (MS/BS) without affecting the mineral apposition 478 rate (MAR) (Fig. 8E, Fig. S9C). Serum biochemical assays showed that Pfkfb3 overexpression 479 elevated P1NP levels without changing CTX-I in the T2D mice (Fig. 8F). RT-qPCR detected a 480 modest increase of Pfkfb3, Hif1a, Glut1, and Ldhb mRNA in the bone shaft of Pfkfb3OE 481 diabetic mice versus Ctrl T2D mice (Fig. 8G). Overall, the results demonstrate that bone-specific 482 activation of glycolysis is an effective anabolic strategy to counter the bone loss in T2D mice. 483 . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted January 18, 2023. ; https://doi.org/10.1101/2023.01.16.524248 doi: bioRxiv preprint

Discussion 484
We have investigated the mechanism underlying the bone anabolic defect in a T2D mouse model. 485 Evidence from multiple approaches including stable isotope tracing in vivo, scRNA-seq of bone 486 marrow mesenchymal cells, as well as in vitro flux assays supports that T2D causes metabolic 487 dysregulation intrinsic to osteoblast lineage cells. Importantly, activation of glycolysis either 488 pharmacologically or by genetic means effectively enhances bone formation and preserves bone 489 mass in diabetic mice. The results therefore support osteoblast-intrinsic impairment of glucose 490 metabolism as a pathogenic mechanism for diabetic osteopenia in T2D. 491

492
Although adult-onset T2D differs from youth-onset T2D in that the net bone mass is either 493 normal or even increased in adults, they both suppress bone formation. Multiple mechanisms 494 including hyperglycemia and decreased signaling by Wnt, adipokine, insulin, or Igf1 have been 495 proposed to impair osteoblast differentiation and function in diabetes, but their functional 496 relevance in vivo remains elusive 38 . Nonetheless, as Wnt and Igf1 signaling have been shown to 497 promote glycolysis directly in osteoblast lineage cells, it is reasonable to speculate that 498 downregulation of the aforementioned signals may be a proximate cause for impaired glycolysis 499 in diabetic bone 6,7 . Our mouse model recapitulates the low bone turnover as seen in human 500 patients, but we have chosen to focus on the bone anabolic defect as it drives the osteopenia 501 phenotype. The model can be further studied in the future to determine the mechanism for 502 impaired bone resorption in T2D and its potential contribution to impaired bone formation 503 through a coupling mechanism. Nonetheless, it is worth noting that osteoblast-directed activation 504 of glycolysis increased bone formation activity without any "coupled" activation of bone 505 resorption, indicating that osteoblasts can be selectively targeted to increase bone mass in T2D. 506 . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted January 18, 2023. ; https://doi.org/10.1101/2023.01.16.524248 doi: bioRxiv preprint 507 Our data support the functional contribution of osteoblast glucose metabolism to the bone 508 anabolic defect in T2D. Metformin notably increased bone formation in the T2D mice, 509 concurrent with stimulation of glucose metabolism in bone. Similarly, in vitro, metformin 510 enhanced glycolysis together with osteoblast differentiation in BMSC derived from T2D mice, 511 indicating a direct effect on osteoblast lineage cells. Targeted overexpression of Hif1a, known to 512 activate multiple target genes in the glycolysis pathway, markedly increased bone formation and 513 restored bone mass in T2D mice. Finally, overexpression of a single glycolysis-promoting gene 514 Pfkfb3 was sufficient to overcome the bone anabolic defect caused by T2D. In contrast, 515 overexpression of Glut1 failed to increase bone formation, indicating that glucose uptake was not 516 responsible for the reduced glycolysis flux or impaired bone anabolic activity in T2D. The result 517 further supports the notion that downregulation of Glut1 in diabetic bones likely represents a 518 protective mechanism against cellular damages, as intracellular accumulation of glucose is 519 known to be cytotoxic through multiple mechanisms 39 . 520 521 In conclusion, the study identifies impaired glucose metabolism in osteoblasts as a key mediator 522 for the bone anabolic defect in T2D. Furthermore, osteoblast glycolysis offers a promising target 523 for developing future bone therapies. 524 525 Acknowledgements 526 The work is partially supported by NIH grant R01 DK125498 (FL). 527 528 Data availability 529 . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made  Enrichment of specific isotopologues of metabolites relative to own pool. CTRL, n=8; T2D, n=6. 664 . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted January 18, 2023. (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted January 18, 2023. ; https://doi.org/10.1101/2023.01.16.524248 doi: bioRxiv preprint (P1NP) and resorption (CTX-1). Ctrl T2D, n = 12; Pfkfb3OE T2D, n = 9. (G) qPCR analyses of 711 glycolysis related genes in bone. Ctrl T2D, n = 7; Pfkfb3OE T2D n = 4. Data presented as mean 712 ± SD. *: P < 0.05, Two-way ANOVA followed by Fisher's LSD test. (C, D) or Student's t test 713 (all others). 714 . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted January 18, 2023. (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted January 18, 2023. ;https://doi.org/10.1101https://doi.org/10. /2023 marker genes. n=3. Data are represented as mean ± SD. *P < 0.05 by one-way ANOVA followed 738 by Student's t test. Quantification of mineral apposition rate (MAR) by double labeling in T2D mice with or without 760 . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted January 18, 2023. ;https://doi.org/10.1101https://doi.org/10. /2023 Pfkfb3OE. Ctrl T2D, n = 4; Pfkfb3OE T2D, n = 5. Data are represented as mean ± SD. *P < 0.05 761 Two-way ANOVA followed by Fisher's LSD test (A, B)  . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted January 18, 2023. ;https://doi.org/10.1101https://doi.org/10. /2023