Bone 73 (2015) 98–104

Contents lists available at ScienceDirect

Bone

j ourna l homepage: www.e lsev ie r .com/ locate /bone

Original Full Length Article

HIF-1α regulates bone formation after osteogenic mechanical loading

Ryan E. Tomlinson ⁎, Matthew J. Silva ⁎⁎Departments of Orthopaedic Surgery and Biomedical Engineering, Musculoskeletal Research Center, Washington University in St. Louis, 425 S. Euclid Avenue, St. Louis, MO 63110, USA

⁎ Correspondence to: R.E. Tomlinson, 720 N. Rutland Av21211, USA. Fax: +1 443 287 4428.⁎⁎ Correspondence to: M.J. Silva, 425 S. Euclid Avenue, C63110, USA.

E-mail addresses: ryan.tomlinson@jhmi.edu (R.E. Tom(M.J. Silva).

http://dx.doi.org/10.1016/j.bone.2014.12.0158756-3282/© 2014 Elsevier Inc. All rights reserved.

a b s t r a c t

a r t i c l e i n f o

Article history:Received 18 June 2014Revised 22 November 2014Accepted 16 December 2014Available online 23 December 2014

Edited by: David Burr

Keywords:Hypoxia inducible factorHIF-1αMechanical loadingOsteogenesis

HIF-1 is a transcription factor typically associated with angiogenic gene transcription under hypoxic conditions.In this study, mice with HIF-1α deleted in the osteoblast lineage (ΔHIF-1α) were subjected to damaging or non-damagingmechanical loading known to produce woven or lamellar bone, respectively, at the ulnar diaphysis. BymicroCT,ΔHIF-1αmice produced significantly lesswoven bone thanwild type (WT)mice 7 days after damagingloading. This decrease in woven bone volume and extent was accompanied by a significant decrease in vascular-itymeasured by immunohistochemistry against vWF. Additionally, osteocytes, rather than osteoblasts, appear tobe the main bone cell expressing HIF-1α following damaging loading. In contrast, 10 days after non-damagingmechanical loading, dynamic histomorphometry measurements demonstrated no impairment in loading-induced lamellar bone formation in ΔHIF-1α mice. In fact, both non-loaded and loaded ulnae from ΔHIF-1αmice had increasedbone formation comparedwithWTulnae.When comparing the relative increase in periostealbone formation in loaded vs. non-loaded ulnae, it was not different between ΔHIF-1α mice and controls. Therewere no significant differences observed between WT and ΔHIF-1α mice in endosteal bone formation parame-ters. The increases in periosteal lamellar bone formation inΔHIF-1αmice are attributed to non-angiogenic effectsof the knockout. In conclusion, these results demonstrate that HIF-1α is a pro-osteogenic factor for woven boneformation after damaging loading, but an anti-osteogenic factor for lamellar bone formation under basal condi-tions and after non-damaging loading.

© 2014 Elsevier Inc. All rights reserved.

Introduction

Hypoxia-inducible factor 1 (HIF-1) was initially reported as a nucle-ar factor for transcriptional activation in response to reducedO2 concen-tration (hypoxia) [1]. Subsequently, it was identified as a basic helix–loop–helix heterodimeric protein consisting of two subunits designatedHIF-1α and HIF-1β [2,3]. HIF-1α is expressed ubiquitously throughoutthe body. In normoxic conditions, HIF-1α is rapidly degraded by theubiquitin–proteasome pathway, but in conditions of hypoxia it remainsstable [4]. HIF-1 stability is one of the methods by which all nucleatedcells in the body sense and respond to O2 availability [5]. As a result,HIF-1 has been called the “highly involved factor” [6], since it has arole in awide variety of hypoxia-related processes required for develop-ment and homeostasis, including angiogenesis, erythropoiesis, andvasomotor control [7].

HIF-1α plays an important role in coupling angiogenesis and osteo-genesis, particularly in skeletal healing and development [8–11]. Duringfracture healing, HIF-1 target genes such as VEGF, PGF, and SDF-1 are

enue, Ross 209C, Baltimore,MD

ampus Box 8233, St. Louis, MO

linson), silvam@wustl.edu

both spatially and temporally regulated [12–15]. VEGF in particular isalso responsible for orchestrating the vascular invasion of hypertrophiccartilage that establishes the primary ossification center in developingmouse long bones [16,17]. Mice lackingHIF-1α in the osteoblast lineagehave narrow, poorly vascularized long bones at 3 weeks of age, despitenormal expression of HIF-1α in the surrounding tissues [18]. Moreover,thesemice do not produce adequate vascularity to support woven boneinduction during distraction osteogenesis [19], consistent with the re-quirement of angiogenesis in this process [20]. However, by 24 weeksof age, bones from mice lacking HIF-1α in the osteoblast lineage havenormal cortical thickness and moment of inertia as well as significantlyincreased bone area compared with wild type controls [21]. Consistentwith increased bone apposition in adult mice, these mice have en-hancedperiosteal lamellar bone formationwhen subjected tomild tibialmechanical loads [21]. Taken together, these results indicate that HIF-1α plays a complex role in postnatal osteogenesis.

In general, repetitive mechanical loading of the skeleton is a potentstimulus of bone formation. When applied at hyperphysiological strainlevels for many cycles, mechanical loading produces fatigue damagethat can progress to a non-displaced stress fracture and stimulate peri-osteal woven bone formation [22]. In this setting, woven bone forma-tion is associated with the upregulation of angiogenic genes (Vegf,Pecam1, Hif1α) and a dramatic downregulation of the Wnt antagonistSclerostin (Sost) [23,24]. Robust expression of HIF-1α has been ob-served in the inflammatory cells located in the expanded periosteal

99R.E. Tomlinson, M.J. Silva / Bone 73 (2015) 98–104

region as soon as 1 day after loading, with a peak in gene expression atday 7 [22,25]. Additionally, woven bone formation after stress fracture ispreceded by increased periosteal vascularity [25,26] and is impaired byangiogenic inhibition [27]. In contrast to fatigue loading, mechanicalloading applied near physiological strain levels for fewer cycles doesnot produce damage, and stimulates lamellar bone formation. Loading-induced lamellar bone formation is preceded by modest upregulationof angiogenic genes (Vegf, Hif1α) [25]. A small increase in vascularity isdetectable after lamellar bone formation has occurred, but loading-induced lamellar bone formation does not depend on angiogenesis [28].

In this study, mice with HIF-1α selectively removed from the osteo-blast lineagewere subjected to either damaging or non-damaging ulnarmechanical loading that triggered the formation of woven or lamellarbone, respectively. Bone formation was assessed using microCT anddynamic histomorphometry, and vascularity was quantified by immu-nohistochemistry. The overall goal of this study was to determine theinfluence of osteoblastic HIF-1α on the postnatal formation of wovenand lamellar bone using a single mechanical loading model.

Materials and methods

ΔHIF-1α mice

Mice with HIF-1α selectively removed from the osteoblast lineagewith high efficiency (N90% by mRNA) have been previously described[18]. Briefly, mice with the second exon of HIF-1α flanked with loxPsites (floxed) [29] were crossed with mice expressing cre recombinaseunder the control of the osteocalcin (OC) promoter [30]. HIF-1αfl/fl;OC-cre+ mice are referred to as ΔHIF-1α mice, and HIF-1αfl/fl; OC-cre− mice are considered wild type (WT) controls. PCR was used todetermine expression of OC-cre using 5′-CAAATAGCCCTGGCAGATTC-3′ (forward) and 5′-TGATACAAGGGACATCTTCC-3′ (reverse) primers.Additionally, PCR was also used to identify the loxP site on the secondexon of HIF-1α with 5′-TGATGTCCCTGCTGGTGTC-3′ (forward) and 5′-TTGTGTTGGGGCAGTACTG-3′ (reverse) primers.

Study design

Female mice were housed under standard conditions until18–22 weeks of age, to ensure skeletal maturity [31,32] and allow ap-proximate normalization of any cortical bone phenotype [21]. At thetime of loading, there was no significant difference in weight betweenWT (23.1 ± 1.4 g) and ΔHIF-1α (22.6 ± 2.0 g) mice. The right forelimbof eachmouse was mechanically loaded using one of two loading proto-cols designed to induce woven bone formation (WBF) or lamellar boneformation (LBF). For each animal, the contralateral (left) forelimb wasused as a non-loaded control. Euthanasia was by CO2 asphyxiation.WBF forelimbs were analyzed by microCT to assess woven bone volumeand density, then decalcified and embedded in paraffin for histologicalanalysis. LBF forelimbs were embedded in PMMA and sectioned for dy-namic histomorphometry to assessmeasures of lamellar bone formation.All animal protocols were approved by the Animal Studies Committee ofWashington University in St. Louis. Results are given as fold changes(loaded limb/non-loaded limb) and plotted as mean ± standard devia-tion. Statistical evaluation was performed using t-tests (Statview 5.0,SAS Institute Inc.) with p-value b 0.05 considered significant.

Mechanical loading protocols

Mechanical loading of the right ulna of each animal was performedas previously described [33]. Briefly, the right forelimbwas axially com-pressed by placing the olecranon process and the flexed carpus into spe-cially designed fixtures. Animals were anesthetized using isoflurane gas(1–3%) during loading. A material testing system (Instron ElectroPuls1000) was used to apply force and monitor displacement. Similar toprevious studies in mice [22] and rats [34], loading parameters were

selected as a function of ultimate force and total displacement to fatiguefracture asmeasured during preliminary work. First, ultimate force wasdetermined using axial monotonic compression to failure by displace-ment ramp (0.1 mm/s). Next, total displacement to fatigue fracturewas determined using a cyclic haversine waveform of 3.5 N (80% ofthe ultimate force) at 2 Hz. Animals were euthanized immediately fol-lowing either procedure. There were no significant differences betweenWT and ΔHIF-1α mice in ultimate force (4.38 ± 0.17 N vs 4.45 ±0.21 N) or average total displacement to failure (0.89 ± 0.08 mm vs.0.90 ± 0.13mm), so loading parameters for subsequent survival exper-iments were the same for both genotypes.

For WBF loading of experimental animals, a 0.3 N compressive pre-load was applied followed by a cyclic haversine waveform of 3.5 N at2 Hz until a total displacement of 0.5 mm relative to the 10th cyclewas achieved. In this loadingmodel, the magnitude of displacement in-crease is an external index of internal ulnar damage [22]. This amount ofdisplacement was equivalent to 55% of the average total displacementto fatigue fracture (0.9 mm), reliably producing an ulnar stress fracturecentered at 1 mm distal to the ulnar midpoint. Although the number ofcycles required to reach 0.5mmof displacement variedwidely betweenanimals (range: 743 to 10257 cycles), there was no significant dif-ference between WT (6066 ± 2771 cycles) and ΔHIF-1α (4358 ±3239 cycles) mice. For LBF loading, a 0.3 N compressive pre-load wasapplied followed by a cyclic rest-inserted trapezoidal waveform with apeak force of 3.0 N at 0.1 Hz for 100 cycles, similar to multi-day ulnarloading protocols used previously in the mouse [35–38]. The singleloading bout protocol used here wasmodified from a similar procedurein rats [25], and stimulates strain-adaptive bone modeling via lamellarapposition. After loading, mice were given an intramuscular injectionof analgesic (0.05 mg/kg buprenorphine) and allowed unrestrictedcage activity.Micewere euthanized 3, 7, or 10days after loading for sub-sequent analysis.

MicroCT analysis

Woven bone formation was analyzed using ex vivomicro computedtomography (μCT40, Scanco Medical AG) inWBF loaded animals 7 daysafter damaging loading, a timepoint when abundant woven bone isobserved in this model [22]. The central 9 mm of each loaded ulnawas scanned separately at 70 kV and 114 μA with 200 ms integrationtime. The scan tube diameter was 12.3 mm, and medium resolutionwas used to obtain a 12 μm voxel size. Scan slices were acquired inthe transverse plane by placing the forelimb parallel to the z-axis ofthe scanner. Hand drawn contours (sigma = 1.2, support = 2, lower/upper threshold = 150/1000) were used to manually segment bonewith Scanco imaging software. Woven bone volume was calculatedby subtracting the original cortical bone volume from the total bonevolume in the entire scan.Woven bone extentwas quantified bymeasur-ing the axial length ofwovenbone formation along the ulna.WovenboneBMD was calculated by analyzing only woven bone in the middle 20slices of the woven bone extent. Finally, the crack extent following WBFloading was measured as the axial length of apparent cortical cracking.Previous studies have demonstrated that microCT assessment of wovenbone corresponds well with dynamic histomorphometric assessment[22], so separate histomorphometry was not performed for WBF groups.

Immunohistochemistry

HIF-1α expression and vascularity was visualized using immunohis-tochemistry inWBF loaded limbs at 3 and 7 days. Intact forelimbs wereharvested and fixed overnight in 10% NBF, then decalcified in 14% EDTAfor 14 days. Following this, each bone was embedded in paraffin togenerate thin (5 μm) sections from 1 mm distal to the ulnar midpoint.Sections were deparaffinized in xylenes and rehydrated in graded etha-nol solutions. Antigen retrieval was performed by overnight incubationin 0.33M boric acid (Sigma, B6867) at 55 °C. A 20-minute incubation in

100 R.E. Tomlinson, M.J. Silva / Bone 73 (2015) 98–104

3% H2O2 was used to block endogenous peroxidase activity, then sec-tions were incubated in normal goat serum (sc-2043, Santa Cruz —

1.5% in PBS) to reduce nonspecific background staining. Followingthis, slides were incubated in 1:200 dilution of rabbit polyclonal anti-body againstHIF-1α (sc-10790, Santa Cruz) or vWF (AB7356,Millipore)at 4 °C overnight. Negative control slides were prepared by substitutingnormal goat serum for the primary antibody. To visualize binding, bio-tinylated goat anti-rabbit (sc-2018, Santa Cruz) secondary antibodywas applied for 30 min followed by avidin–biotin–peroxidase complexfor 30 min. Finally, slides were developed using diaminobenzidine for60 s, dehydrated, and mounted. Digital images of these sections werecaptured using bright field microscopy (Olympus BX-51) with a 20×or 40× objective. Imaging stitching and manual quantification wereperformed using FIJI [39].

Dynamic histomorphometry

Lamellar bone formation rates in LBF loaded animals were quantifiedby dynamic histomorphometry. Mice were given two intraperitoneal in-jections offluorescent bone formationmarkers. Calcein (10mg/kg, SigmaC0875) was administered 3 days after loading, and Alizarin Complexone(30 mg/kg, Sigma A3882) was administered 8 days after loading; andanimals were euthanized on day 10. Following fixation, forelimbs wereembedded in poly-(methyl methacrylate). 100 μm thick transversesections were cut (SP 1600, Leica Microsystems) at 1 mm distal to theulnarmidpoint, then polished to 30 μmandmounted on glass slides. Dig-ital images of these sections were captured using fluorescence microsco-py (Olympus IX-51) with fluorescein isothiocyanate (FITC — excitation465–495 nm, emission 515–555 nm) and tetramethylrhodamine isothio-cyanate (TRITC — excitation 515–565 nm, emission 550–660 nm) filtersfor visualization of calcein (green) and alizarin (red), respectively. Imageswere analyzed in BioQuant Osteo for endocortical (Ec) and periosteal(Ps) bone formation rate (BFR/BS), mineral apposition rate (MAR), andmineralizing surface (MS/BS), as defined by the ASBMR Committee forHistomorphometry Nomenclature [40].

Results

Woven bone formation after damaging loading is impaired inΔHIF-1αmice

Immunohistochemistry was used to localize HIF-1α expression fol-lowing damaging (WBF) loading as well as verify the efficacy of the

Fig. 1. HIF-1α expression following damaging mechanical loading. Immunohistochemistry agaiexpression ofHIF-1αwas robust in osteocytes on days 3 and 7 (black arrows) aswell as the inflarelatively few osteoblasts were positive. In contrast, ΔHIF-1αmice (B, D) had diminished expreremained strong (bracket). Images are representative of 5–8 samples per group.

knockout. Previouswork using qPCR has shown that HIF-1α expressionis maximal on days 3 and 7 following WBF loading [22]. Here, HIF-1αwas strongly expressed at day 3 in the inflammatory and stromal cellspresent in the expanded periosteal region in both WT and ΔHIF-1αmice (Figs. 1a, b — bracket). HIF-1α was also expressed by bone cells,with 63 ± 4% of osteocytes in the original cortical bone positivelystained (Fig. 1a). As expected, less than half as many osteocytes inΔHIF-1α mice were positively stained (29 ± 8%, p b 0.05 vs. WT,Fig. 1b). Similarly, 51 ± 6% of the osteocytes in the original corticalbone of WT mice had strong expression of HIF-1α at day 7, with someosteoblasts in the newly formed woven bone also positive for HIF-1α(Fig. 1c). Again, less than half as many osteocytes in ΔHIF-1α micewere positively stained for HIF-1α at day 7 (19 ± 3%, p b 0.05 vs. WT,Fig. 1d). By day 7, limited HIF-1α expression was observed in the skele-tal muscle adjacent to bone in bothWT and ΔHIF-1αmice, a stark con-trast to the inflammatory cells present at day 3, and similar to HIF-1αexpression routinely observed in normoxic skeletal muscle [41,42].

MicroCT and immunohistochemistry were used to quantify wovenbone formation and vascularity, respectively, in WT and ΔHIF-1α mice7 days afterWBF loading. The extent of the cortical crackwas not signif-icantly different between WT (0.37 ± 0.08 mm) and ΔHIF-1α (0.42 ±0.13 mm) mice, indicating that each genotype received a similar levelof fatigue damage by WBF loading. In response to this injury, both WT(Fig. 2a) and ΔHIF-1α (Fig. 2b) mice produced abundant woven bone,but ΔHIF-1α mice had significantly less (−22%) woven bone volumecompared with WT mice (Fig. 2c). Similarly, woven bone extent wassignificantly less (−21%) in ΔHIF-1α mice compared with WT mice(Fig. 2d). On the other hand, there was no difference in woven boneBMD between WT and ΔHIF-1α mice (Fig. 2e), indicating normal min-eralization. WBF loading stimulated significant angiogenesis in bothΔHIF-1α and WT mice (Fig. 3a). However, ΔHIF-1α mice had signifi-cantly fewer (−35%) blood vessels in the expanded periosteal regioncompared with WT mice (Fig. 3b).

Lamellar bone formation after non-damaging loading is enhanced inΔHIF-1α mice

Dynamic histomorphometry was used to quantify bone formationrates in WT and ΔHIF-1α mice after a single bout of non-damaging(LBF) loading (Fig. 4). No woven bone was observed in the LBF groups.Periosteal lamellar bone formation was greatly increased as a result ofmechanical loading in both genotypes. Ps.MS/BS, Ps.MAR, and Ps.BFR/

nst HIF-1αwas performed 3 and 7 days after damaging (WBF) loading. InWTmice (A, C),mmatory and stromal cells present in the expandedperiosteumon day 3 (bracket), thoughssion of HIF-1α in bone cells (white arrows), but expression in the expanded periosteum

Fig. 2.Woven bone formation after damaging mechanical loading. 7 days after loading, A)WTmice and B) ΔHIF-1αmice were imaged using microCT to quantify woven bone formation.The original cortical bone (cb— red), the newly formedwoven bone (wb— green), and the cortical crack (arrow) aremarked in each cross section. C)ΔHIF-1αmice had significantly lesswoven bone volume (−22%) and D) woven bone extent (−21%). *p b 0.05 vs. WT. (n = 8 per group).

Fig. 3. Blood vessel formation after damagingmechanical loading. A) 7 days after loading, vascularitywas quantified using immunohistochemistry against vonWillebrand factor. B)ΔHIF-1αmice had significantly fewer vessels (−35%) compared with WT mice. *p b 0.05 vs. WT. (n = 8 per group).

101R.E. Tomlinson, M.J. Silva / Bone 73 (2015) 98–104

Fig. 4. Dynamic histomorphometry after non-damaging mechanical loading. Calcein (green) and Alizarin (red) fluorescent bone formation markers were administered at days 3 and 8,respectively, following non-damaging mechanical loading. Wild type loaded (A) and non-loaded (B) limbs were quantitatively compared with ΔHIF-1α loaded (C) and non-loaded(D) limbs. Inset is magnified 200%. (n = 6 per group).

102 R.E. Tomlinson, M.J. Silva / Bone 73 (2015) 98–104

BS were all significantly increased in both WT and ΔHIF-1α loadedlimbs compared with the non-loaded control limbs (Fig. 5). Comparinggenotypes, ΔHIF-1α loaded limbs generated more lamellar bone thanWT loaded limbs, with increased Ps.MAR and Ps.BFR/BS comparedwithWT loaded limbs. ΔHIF-1α non-loaded limbs also had significantlyincreased periosteal bone formation (Ps.MS/BS, Ps.MAR, and Ps.BFR/BS)compared with WT non-loaded limbs, in which no double labels wereobserved. The differences in bone formation parameters between WTand ΔHIF-1α non-loaded limbs indicate a different baseline rate ofbone formation. Therefore, relative (r) parameters were also consid-ered, computed as the difference between loaded and non-loadedlimbs (Table 1). By this measure, no periosteal bone formation parame-ters were significantly different between genotypes, although rPs.BFR/BS was still increased (+9%) in ΔHIF-1α mice compared with WT. Onthe endocortical surface of the bone, loaded limbs had significantlyincreased Ec.MS/BS, but there were no differences between WT andΔHIF-1αmice. No other endocortical parameters were significantly dif-ferent between loaded and non-loaded limbs.

Fig. 5. Lamellar bone formation after non-damaging mechanical loading. Bone formation at theciated with increases in Ec.MS/BS, Ps.MS/BS, Ps.MAR, and Ps.BFR/BS in both WT and ΔHIF-1α m

Discussion

In this study, the role of HIF-1α in loading-induced osteogenesiswasexplored usingmicewith a targeted deletion of HIF-1α in the osteoblastlineage. Each animal was subjected to a single bout mechanical loadingprotocol of either damaging or non-damaging axial compression of theforelimb. The results demonstrate that ΔHIF-1α mice have attenuatedwoven bone formation following damaging mechanical loading (stressfracture), but enhanced lamellar bone formation following non-damaging mechanical loading (strain adaptive modeling). Thus, thisstudy shows that HIF-1α is a pro-osteogenic factor for woven boneformation following damaging loading, but an anti-osteogenic factorfor lamellar bone formation after non-damaging mechanical loading.

The diminished response to damaging mechanical loading is attrib-uted to the role of HIF-1 as a transcription factor associated with angio-genesis. One of themajor targets of HIF-1 is vascular endothelial growthfactor (VEGF), a well-characterized pro-angiogenic factor that plays animportant role in the vascularization of bone [43] and is upregulated

ulnar mid-diaphysis was quantified using dynamic histomorphometry. Loading was asso-ice. *p b 0.05 vs. non-loaded, +p b 0.05 vs. WT. (n = 6 per group).

Table 1Relative (loaded–non-loaded) periosteal bone formation parameters. There are no signif-icant differences between WT and ΔHIF-1α mice by this measure.

rPs.MS/BS (%) rPs.MAR (μm/day) rPs.BFR/BS (μm/day)

WT 25.4 ± 5.7 0.67 ± 0.28 0.35 ± 0.15ΔHIF-1α 20.2 ± 4.1 0.68 ± 0.38 0.38 ± 0.10p-Value 0.23 0.49 0.43

103R.E. Tomlinson, M.J. Silva / Bone 73 (2015) 98–104

after damaging bone loading [23,25]. Previous studies have shown thatHIF-1α is critical for angiogenesis and subsequent bone formation dur-ing murine distraction osteogenesis [19,44]. Additionally, inhibition ofangiogenesis led to diminished woven bone formation following dam-aging mechanical loading in rats [27]. Here, a significant decrease inperiosteal vascularity due to decreased HIF-1 expressionwas associatedwith less woven bone formation. This result demonstrates that HIF-1αexpression in the osteoblast lineage is specifically required for completeangiogenesis, despite robust expression of HIF-1α in inflammatory cellsand tissues immediately adjacent to the site of bone formation.

In contrast, lamellar bone formation was not impaired in ΔHIF-1αmice following non-damaging osteogenic forelimb loading, consistentwith previous work that found loading-induced lamellar bone forma-tion to be independent of angiogenesis in rats [28]. The significantincreases (+50%) in HIF-1α gene expression observed 3 days afterloading [25] suggested that HIF-1α may be a positive regulator in theaccrual of lamellar bone, but the results from the currentwork illustratethatmicewith HIF-1α removed in the osteoblast lineage have increasedbaseline bone formation. When challenged with non-damaging fore-limb compression,ΔHIF-1αmice respondnormally, resulting in a great-er absolute level of bone formation in ΔHIF-1α loaded limbs comparedwithWT. This finding is similar to the results from a previous study thatshowed ΔHIF-1αmice generated more bone following 3 weeks of tibialbending thanWTmice, although they found no differences in bone for-mation in sham limbs [21]. The increased bone formation in non-loadedlimbs observed in this study is consistent with the age-related changesin bone size previously reported — ΔHIF-1α mice begin life with smallbones that become comparable with their wild type littermates inadulthood [21].

Increased bone formation in ΔHIF-1α mice has previously been at-tributed to HIF-1α negatively regulatingWnt signaling [21], a pathwaycentral in the differentiation and function of osteoblasts [45]. AlthoughHIF-1α has been shown to bind to β-catenin, potentially inhibiting ca-nonical Wnt signaling [46], Wnt target genes are upregulated followingosteogenicmechanical loading alongwith HIF-1α, whereas knownWntinhibitors such as SOST and Dkk are downregulated [24,47,48]. Theseobservations suggest that a strong negative effect of HIF-1α on canoni-cal Wnt signaling in bone following loading is unlikely. In addition,the increase in bone apposition in non-loaded limbs of ΔHIF-1α micewas only observed here on the periosteal surface — in scenarios ofincreased Wnt signaling, bone formation is also dramatically increasedon the endocortical surface [49–51]. Thus, the mechanism of the non-angiogenic effects of HIF-1α on periosteal bone formation remainsunclear, although we have shown here that osteoblastic HIF-1α is notrequired for a strong induction of lamellar bone formation by mechan-ical loading.

Previous studies of damaging mechanical loading of the mouse ulnaobserved 3-fold increases in HIF-1α expression on day 3 after loading,based on qPCR of whole bone with attached periosteum [22]. Here,immunohistochemistry indicates that HIF-1α is primarily expressed inosteocytes and inflammatory cells adjacent to the bone, whereas fewosteoblasts present at the site of woven bone formation in WT miceexpressed HIF-1α. Thus, the effect of HIF-1α in the osteoblast lineagemay bemediated largely by osteocytes. A role for osteocytes in perioste-al angiogenesis andwoven bone formation after bone injury has not yetbeen established. On the other hand, osteocytes have been shown toplay an important role in lamellar bone formation [48,52].

Conclusion

In this study, mice with HIF-1α deleted from the osteoblast lineagewere subjected to damaging and non-damaging osteogenic mechanicalloadings. Following damaging loading, HIF-1α deficiency results in de-creased angiogenesis and reduced woven bone formation. In contrast,HIF-1α deficiency does not impair lamellar bone formation inducedby non-damaging loading, with overall bone formation increased inknockout mice. Immunohistochemistry against HIF-1α suggests thatosteocytes are the osteoblast lineage cells driving these results. In sum-mary, this study demonstrates that HIF-1α is a factor that regulates bothdamaging and non-damaging osteogenic mechanical loadings.

Acknowledgments

Founding members of the mouse colony used in this study weregenerously donated by Drs. Thomas Clemens and Ryan Riddle (JohnsHopkins University). The authors are grateful to Dr. Deborah Novackfor assistance with histology. This study was funded by a grant fromthe National Institutes of Health (R01 AR050211). This work was per-formed in facilities supported by the Washington University Center forMusculoskeletal Research (P30 AR057235).

References

[1] Semenza GL, Wang GL. A nuclear factor induced by hypoxia via de novo proteinsynthesis binds to the human erythropoietin gene enhancer at a site required fortranscriptional activation. Mol Cell Biol 1992;12:5447–54.

[2] Wang GL, Jiang BH, Rue EA, Semenza GL. Hypoxia-inducible factor 1 is a basic-helix–loop–helix–PAS heterodimer regulated by cellular O2 tension. Proc Natl Acad Sci U S A1995;92:5510–4.

[3] Wang GL, Semenza GL. Purification and characterization of hypoxia-inducible factor1. J Biol Chem 1995;270:1230–7.

[4] Salceda S, Caro J. Hypoxia-inducible factor 1alpha (HIF-1alpha) protein is rapidlydegraded by the ubiquitin–proteasome system under normoxic conditions. Its stabi-lization by hypoxia depends on redox-induced changes. J Biol Chem 1997;272:22642–7.

[5] Semenza GL. HIF-1: mediator of physiological and pathophysiological responses tohypoxia. J Appl Physiol 2000;88:1474–80.

[6] Semenza GL. HIF-1 and human disease: one highly involved factor. Genes Dev 2000;14:1983–91.

[7] Epstein AC, Gleadle JM, McNeill LA, Hewitson KS, O'Rourke J, Mole DR, et al.C. elegans EGL-9 and mammalian homologs define a family of dioxygenases thatregulate HIF by prolyl hydroxylation. Cell 2001;107:43–54.

[8] Riddle RC, Khatri R, Schipani E, Clemens TL. Role of hypoxia-inducible factor-1alphain angiogenic–osteogenic coupling. J Mol Med 2009;87:583–90.

[9] Glowacki J. Angiogenesis in fracture repair. Clin Orthop Relat Res Oct 1998:S82–9.[10] Wan C, Shao J, Gilbert SR, Riddle RC, Long F, Johnson RS, et al. Role of HIF-1alpha in

skeletal development. Ann N Y Acad Sci 2010;1192:322–6.[11] Shomento SH, Wan C, Cao X, Faugere MC, Bouxsein ML, Clemens TL, et al. Hypoxia-

inducible factors 1alpha and 2alpha exert both distinct and overlapping functions inlong bone development. J Cell Biochem 2010;109:196–204.

[12] Ferguson C, Alpern E, Miclau T, Helms JA. Does adult fracture repair recapitulateembryonic skeletal formation? Mech Dev 1999;87:57–66.

[13] Liu L, Marti GP,Wei X, Zhang X, Zhang H, Liu YV, et al. Age-dependent impairment ofHIF-1alpha expression in diabetic mice: correction with electroporation-facilitatedgene therapy increases wound healing, angiogenesis, and circulating angiogeniccells. J Cell Physiol 2008;217:319–27.

[14] Ceradini DJ, Kulkarni AR, Callaghan MJ, Tepper OM, Bastidas N, Kleinman ME, et al.Progenitor cell trafficking is regulated by hypoxic gradients through HIF-1 inductionof SDF-1. Nat Med 2004;10:858–64.

[15] Le AX, Miclau T, Hu D, Helms JA. Molecular aspects of healing in stabilized and non-stabilized fractures. J Orthop Res 2001;19:78–84.

[16] Colnot CI, Helms JA. A molecular analysis of matrix remodeling and angiogenesisduring long bone development. Mech Dev 2001;100:245–50.

[17] Carlevaro MF, Cermelli S, Cancedda R, Descalzi Cancedda F. Vascular endothelialgrowth factor (VEGF) in cartilage neovascularization and chondrocyte differentia-tion: auto-paracrine role during endochondral bone formation. J Cell Sci 2000;113(Pt 1):59–69.

[18] Wang Y, Wan C, Deng L, Liu X, Cao X, Gilbert SR, et al. The hypoxia-inducible factoralpha pathway couples angiogenesis to osteogenesis during skeletal development.J Clin Invest 2007;117:1616–26.

[19] Wan C, Gilbert SR, Wang Y, Cao X, Shen X, Ramaswamy G, et al. Activation of thehypoxia-inducible factor-1alpha pathway accelerates bone regeneration. Proc NatlAcad Sci U S A 2008;105:686–91.

[20] Jacobsen KA, Al-Aql ZS, Wan C, Fitch JL, Stapleton SN, Mason ZD, et al. Bone forma-tion during distraction osteogenesis is dependent on both VEGFR1 and VEGFR2signaling. J Bone Miner Res 2008;23:596–609.

104 R.E. Tomlinson, M.J. Silva / Bone 73 (2015) 98–104

[21] Riddle RC, Leslie JM, Gross TS, Clemens TL. Hypoxia-inducible factor-1alpha proteinnegatively regulates load-induced bone formation. J Biol Chem 2011;286:44449–56.

[22] MartinezMD, Schmid GJ, McKenzie JA, Ornitz DM, SilvaMJ. Healing of non-displacedfractures produced by fatigue loading of the mouse ulna. Bone 2010;46:1604–12.

[23] Wohl GR, Towler DA, Silva MJ. Stress fracture healing: fatigue loading of the rat ulnainduces upregulation in expression of osteogenic and angiogenic genes that mimicthe intramembranous portion of fracture repair. Bone 2009;44:320–30.

[24] McKenzie JA, Bixby EC, Silva MJ. Differential gene expression from microarray anal-ysis distinguishes woven and lamellar bone formation in the rat ulna following me-chanical loading. PLoS One 2011;6:e29328.

[25] McKenzie JA, Silva MJ. Comparing histological, vascular and molecular responsesassociated with woven and lamellar bone formation induced by mechanical loadingin the rat ulna. Bone 2011;48:250–8.

[26] Matsuzaki H, Wohl GR, Novack DV, Lynch JA, Silva MJ. Damaging fatigue loadingstimulates increases in periosteal vascularity at sites of bone formation in the ratulna. Calcif Tissue Int 2007;80:391–9.

[27] Tomlinson RE, McKenzie JA, Schmieder AH, Wohl GR, Lanza GM, Silva MJ. Angiogen-esis is required for stress fracture healing in rats. Bone 2013;52:212–9.

[28] Tomlinson RE, Schmieder AH, Quirk JD, Lanza GM, Silva MJ. Antagonizing the alphavbeta3 integrin inhibits angiogenesis and impairs woven but not lamellar bone for-mation induced by mechanical loading. J Bone Miner Res 2014;29:1970–80.

[29] Ryan HE, Poloni M, McNulty W, Elson D, Gassmann M, Arbeit JM, et al. Hypoxia-inducible factor-1alpha is a positive factor in solid tumor growth. Cancer Res2000;60:4010–5.

[30] Zhang M, Xuan S, Bouxsein ML, von Stechow D, Akeno N, Faugere MC, et al.Osteoblast-specific knockout of the insulin-like growth factor (IGF) receptor genereveals an essential role of IGF signaling in bone matrix mineralization. J BiolChem 2002;277:44005–12.

[31] Ferguson VL, Ayers RA, Bateman TA, Simske SJ. Bone development and age-relatedbone loss in male C57BL/6 J mice. Bone 2003;33:387–98.

[32] Brodt MD, Ellis CB, Silva MJ. Growing C57Bl/6 mice increase whole bone mechanicalproperties by increasing geometric and material properties. J Bone Miner Res 1999;14:2159–66.

[33] Lee KC, Maxwell A, Lanyon LE. Validation of a technique for studying functional adap-tation of the mouse ulna in response to mechanical loading. Bone 2002;31:407–12.

[34] Uthgenannt BA, Kramer MH, Hwu JA, Wopenka B, Silva MJ. Skeletal self-repair:stress fracture healing by rapid formation and densification of woven bone. J BoneMiner Res 2007;22:1548–56.

[35] Liedert A, Mattausch L, Rontgen V, Blakytny R, Vogele D, Pahl M, et al. Midkine-deficiency increases the anabolic response of cortical bone to mechanical loading.Bone 2011;48:945–51.

[36] Xiao Z, DallasM,QiuN,NicolellaD, Cao L, JohnsonM, et al. Conditional deletion of Pkd1in osteocytes disrupts skeletal mechanosensing in mice. FASEB J 2011;25:2418–32.

[37] Sugiyama T, Saxon LK, Zaman G, Moustafa A, Sunters A, Price JS, et al. Mechanicalloading enhances the anabolic effects of intermittent parathyroid hormone (1–34)on trabecular and cortical bone in mice. Bone 2008;43:238–48.

[38] Robling AG, Turner CH. Mechanotransduction in bone: genetic effects onmechanosensitivity in mice. Bone 2002;31:562–9.

[39] Schindelin J, Arganda-Carreras I, Frise E, Kaynig V, Longair M, Pietzsch T, et al.Fiji: an open-source platform for biological-image analysis. Nat Methods 2012;9:676–82.

[40] Dempster DW, Compston JE, Drezner MK, Glorieux FH, Kanis JA, Malluche H, et al.Standardized nomenclature, symbols, and units for bone histomorphometry: a 2012update of the report of the ASBMR Histomorphometry Nomenclature Committee.J Bone Miner Res 2013;28:2–17.

[41] Stroka DM, Burkhardt T, Desbaillets I, Wenger RH, Neil DA, Bauer C, et al. HIF-1 isexpressed in normoxic tissue and displays an organ-specific regulation under sys-temic hypoxia. FASEB J 2001;15:2445–53.

[42] Milkiewicz M, Pugh CW, Egginton S. Inhibition of endogenous HIF inactivation in-duces angiogenesis in ischaemic skeletal muscles of mice. J Physiol 2004;560:21–6.

[43] Chappell JC, Wiley DM, Bautch VL. How blood vessel networks are made and mea-sured. Cells Tissues Organs 2012;195:94–107.

[44] Carvalho RS, Einhorn TA, Lehmann W, Edgar C, Al-Yamani A, Apazidis A, et al. Therole of angiogenesis in a murine tibial model of distraction osteogenesis. Bone2004;34:849–61.

[45] Soltanoff CS, Yang S, ChenW, Li YP. Signaling networks that control the lineage com-mitment and differentiation of bone cells. Crit Rev Eukaryot Gene Expr 2009;19:1–46.

[46] Kaidi A, Williams AC, Paraskeva C. Interaction between beta-catenin and HIF-1 pro-motes cellular adaptation to hypoxia. Nat Cell Biol 2007;9:210–7.

[47] Robinson JA, Chatterjee-KishoreM, Yaworsky PJ, Cullen DM, ZhaoW, Li C, et al.Wnt/beta-catenin signaling is a normal physiological response to mechanical loading inbone. J Biol Chem 2006;281:31720–8.

[48] Robling AG, Niziolek PJ, Baldridge LA, Condon KW, Allen MR, Alam I, et al. Mechan-ical stimulation of bone in vivo reduces osteocyte expression of Sost/sclerostin. J BiolChem 2008;283:5866–75.

[49] Li X, Ominsky MS, Niu QT, Sun N, Daugherty B, D'Agostin D, et al. Targeted deletionof the sclerostin gene inmice results in increased bone formation and bone strength.J Bone Miner Res 2008;23:860–9.

[50] Babij P, Zhao W, Small C, Kharode Y, Yaworsky PJ, Bouxsein ML, et al. High bonemass in mice expressing a mutant LRP5 gene. J Bone Miner Res 2003;18:960–74.

[51] Krishnan V, Bryant HU, Macdougald OA. Regulation of bone mass by Wnt signaling.J Clin Invest 2006;116:1202–9.

[52] Javaheri B, Stern AR, Lara N, Dallas M, Zhao H, Liu Y, Bonewald LF, Johnson ML. De-letion of a single beta-catenin allele in osteocytes abolishes the bone anabolic re-sponse to loading. J Bone Miner Res 2014;29:705–15.