Dihydroartemisinin, an anti-malaria drug, suppresses estrogen deficiency-

induced osteoporosis, osteoclast formation and RANKL induced signalling

pathways

Lin Zhou1, Qian Liu1, 2, Mingli Yang1, Tao Wang2, Jun Yao1,2, Jianwen Cheng1,2, Jinbo

Yuan1, Xixi Lin2, Jinmin Zhao2, Jennifer Tickner1, Jiake Xu1,2

1School of Pathology and Laboratory Medicine, The University of Western Australia, Perth,

Western Australia, 6009, Australia

2Research Centre for Regenerative Medicine and Guangxi Key Laboratory of Regenerative

Medicine, Guangxi Medical University, Guangxi, China, 530021

Correspondence to: Jiake Xu

E-mail: jiake.xu@uwa.edu.au

Running title: Dihydroartemisinininhibitsosteolysis

Keywords: Dihydroartemisinin, Osteoclast, RANKL, bone resorption, Osteolysis

Abstract Osteoporosis is an osteolytic disease which features enhanced osteoclast formation

and bone resorption. Identification of agents that can inhibit osteoclast formation and

function is important for the treatment of osteoporosis. Dihydroartemisinin is a natural

compound used to treat malaria but its role in osteoporosis in not known. Here, we found that

dihydroartemisinin can suppresses RANKL-induced osteoclastogenesis and bone resorption

in a dose dependent manner. Dihydroartemisin inhibited the expression of osteoclast marker

genes such as cathepsin K, calcitonin receptor, and TRAcP. Furthermore, Dihydroartemisinin

inhibited RANKL-induced NF-κB and NFAT activity. In addition, using an in vivo

ovariectomized mouse model, we show that dihydroartemisinin is able to reverse the bone

loss caused by ovariectomy. Together, this study shows that dihydroartemisinin attenuates

bone loss in ovariectomized mice through inhibiting RANKL-induced osteoclast formation

and function, indicating that dihydroartemisinin is a potential treatment option against

osteolytic bone diseases.

INTRODUCTION

Bone is a continuously renewing tissue which is formed by mineralization of an organic

matrix. Bone formation by osteoblasts and bone resorption by osteoclasts contributes to the

balance of bone remodelling. An imbalance between osteoclast and osteoblast formation and

function causes many osteopathic diseases, such as osteoporosis and Paget’s disease [1, 2].

Osteoporosis is a common age-related degenerative disease which often affects

postmenopausal women due to oestrogen deficiency. According to the 2004 US surgeon

General’s report, it was estimated that women over 50 years old have a 50% risk of fragility

fracture during the remainder of their lifetime, while men of same age have a 20% risk of

fragility fracture throughout the remainder of their life time[3]. Fractures caused by

osteoporosis are associated with significant morbidity and mortality, and place a large

economic burden on society [4].

The functional role of the osteoclast is to resorb bone through secreting acid and proteases to

dissolve the organic and mineral components of bone. Receptor activator of nuclear factor

kappa-B ligand (RANKL), also called osteoprotegerin ligand (OPGL), TNF-related

activation-induced cytokine (TRANCE), and osteoclast differentiation factor (ODF), is a

critical cytokine for the formation and activation of osteoclasts [5-7]. The interaction between

RANKL and its receptor RANK activates several transcription factors, such as NF-κB, AKT,

Activator Protein 1(AP-1), Mitogen Activated Protein kinase (MAPK) and Nuclear Factor of

Activated T-cell cytoplasmic 1(NFATc1). Activation of these downstream factors triggers the

expression of genes governing osteoclast differentiation and function, including TRACP,

Cathepsin K, Matrix Metalloproteinase 9 (MMP-9), and calcitonin receptor (CTR) ultimately

resulting in the production of mature multinucleated osteoclasts [8-10]. RANKL is now

recognised as an attractive target for the treatment of osteoporosis, as inhibition of RANKL

using antibodies, peptides and natural compounds could prevent osteoclast formation and

function. This study aims to determine the effect of several natural compounds on inhibiting

RANKL induced osteoclast formation and function.

Artemisinin is an effective drug for treating malaria. It belongs to the family of sesquiterpene

lactones, which are produced by an Artemisia annua plant[11]. Dihydroartemisinin (DHA) is

a water-soluble semi-synthetic derivative of artemisinin [12, 13] and it is commercially

combined with piperaquine as an effective therapy for malaria [14] with few side effects [15].

DHA has also been found to have the inhibitory effects on cancer cells mainly through

modulating the NF-κB pathway [16-19]. However, the effect of DHA on osteoclast formation

and osteoporosis is not known.

In this study, we examine the role of DHA in osteoporosis in vivo using an ovariectomized

(OVX) mouse model, and elucidate its cellular and molecular mechanism of action. We

demonstrate that DHA has a protective effect in an oestrogen deficiency-induced

osteoporosis mouse model. Furthermore, we found that DHA inhibits the formation and

function of osteoclasts via inhibition of the NF-κB and NFAT pathways. As DHA is already

used for treatment of malaria and has anti-tumour activity, our study further indicates that

DHA could be beneficial for the treatment of osteolytic bone conditions, such as cancer or

inflammation-induced osteolysis.

METHODS

Materials and Reagents

DHA was purchased from Chengdu Must Bio-Technology Co., Ltd (Chengdu, Sichuan

Province, China), and dissolved in dimethyl sulfoxide (DMSO). Alpha modified Eagles

Medium (α-MEM) and fetal bovine serum (FBS) were obtained from TRACE (Sydney,

Australia). Anti-P-ERK and anti-IκBα antibodies were purchased from Santa Cruz

Biotechnology (California, USA), anti-NFATc1 antibody was purchased from BD Bioscience

(America) and V-ATPase-d2 was generated as previously described [20]. Mouse anti-β-Actin

(catalog number: JLA20) was purchased from Developmental Studies Hybridoma Bank

(DSHB) (Iowa, IA, USA). The luciferase assay system and MTS assay kit were purchased

from Promega (Sydney, Australia). Recombinant GST-rRANKL protein was produced

according to a previous paper [21] and recombinant macrophage colony stimulating factor

(M-CSF) was used as previously described [22].

In vitro osteoclastogenesis assay

Freshly isolated bone marrow macrophages (BMM) from C57BL/6 mice were cultured in

T75 flasks in α-MEM containing M-CSF. When confluent, BMMs were seeded into 96 well

plates at 1x106cells per well with α-MEM and M-CSF overnight. The next day, BMMs were

stimulated with 50 ng/ml RANKL and in the presence of DHA for every two days until

osteoclasts formed. The plate was then fixed with 2.5% glutaraldehyde in phosphate-buffered

saline (PBS) for 10 minutes, and stained for tartrate-resistant acidic phosphatase activity

(TRAcP). TRAcP positive multinucleated cells with three or more nuclei were scored as

osteoclast-like (OCL) cells.

In vitro osteoblast differentiation and mineralization

For osteoblastogenesis assays, osteoblast precursors from adult calvaria or long bones were

obtained as outgrowth from collagenase-treated bone pieces as described previously [23]. The

cells were plated into culture dishes at a cell density of 1×106 cells/ml in Dulbecco's Modified

Eagle's Medium (DMEM) supplemented with 10% heat-inactivated FBS, 2mM L-glutamine,

100 units/ml penicillin, and 100 g/ml streptomycin (complete DMEM). When confluent

osteogenic media (complete DMEM, 10 nM dexamethasone, 10 mM β-glycerophosphate,

and 50 µg/ml ascorbate) was added and cells cultured for 7 days (alkaline phosphatase assay)

or 21 days (mineralisation). Alkaline phosphatase staining was performed using the

Leukocyte alkaline phosphatase staining kit (Sigma, Castle Hill, Australia). Mineralisation

was observed following fixation and staining with 1% alizarin red [24]. ImageJ software was

used to measure the mineralized area [25].

MTS assay for cell proliferation and viability

MTS assay was used to determine the effects of compounds on BMM cells using a

commercially available MTS assay kit (Promega, Sydney, Australia). BMMs were seeded on

96-well plates at 6x103 cells per well and incubated overnight. Varying concentrations of

DHA were then added to BMMs and the cells were incubated with the compounds for 48

hours. MTS/PMS mixture was then added to each well for two hours according to

manufacturer’s instructions. The absorbance of MTS was measured by spectrophotometric

absorbance at 490 nm using an ELISA plate reader (BMG, Germany).

Hydroxyapatite resorption assay

BMMs were seeded on to six-well collagen-coated culture plates at a density of 1x105 cells

per well. Cells were allowed to adhere overnight at 37°C and cells were then stimulated with

50ng/ml RANKL and M-CSF until osteoclasts began to form. The cells were gently

harvested using cell dissociation solution, counted and equal numbers of multinucleated cells

were seeded onto hydroxyapatite-coated plates (Corning, America). DHA was added to the

cells after they were seeded onto the hydroxyapatite plate. After 48 hours, half of the wells

were fixed and stained for TRAcP activity for osteoclast counting and the remainder of the

wells were bleached and dried for hydroxyapatite resorption visualisation using a Nikon

microscope and analysed using Image J software.

RNA extraction and analysis

Freshly isolated BMM cells from C57BL/6 mice were cultured in T75 flasks then seeded in

six-well plates at a density of 1x105 cells per well. Cells were then stimulated with RANKL

and M-CSF in the presence of DHA for 5 days. After osteoclasts were formed, cells were

lysed, and total RNA was extracted using Trizol® (Life Technologies, Mulgrave, Australia)

in accordance with the manufacturer’s protocol. For reverse transcription (RT)-PCR, single-

stranded cDNA was reverse transcribed from 1μg total RNA using reverse transcriptase with

an oligo-dT primer. For relative quantitative real-time polymerase chain reaction (PCR),

SYBR® Green PCR MasterMix was used. The cycling parameters for PCR were set as

follows: 94°C for 5 min, followed by 30 cycles of 94°C (40 sec), 60°C, (40 sec); 72°C (40

sec), followed by an elongation step of 5 min at 72°C. PCR reactions used specific primers

for detecting and quantifying the following genes: Cathepsin K (forward: 5’-

GGGAGAAAAACCTGAAGC-3’; reverse: 5’- ATTCTGGGGACTCAGAGC-3’), calcitonin

receptor (forward: 5’- TGGTTGAGGTTGTGCCCA-3’; reverse: 5’-

CTCGTGGGTTTGCCTCATC-3’), TRACP (forward: 5’-TGTGGCCATCTTTATGCT-3’;

reverse:5’-GTCATTTCTTTGGGGCTT-3’),, and 18s RNA (forward: 5’-

ACCATAGATGCCGACT- 3’; reverse: 5’-TGTCAATCCTGTCCGTGTC-3’. qPCR reaction

results were read on a ViiA™ 7Real-time PCR machine (Applied Biosystems, United

Kingdom).The comparative 2-ΔΔCT method was used to calculate the relative expression of

each target gene. The mean Ct value of target genes in the experimental group was

normalized to the Ct value of 18S to give a ΔCt value, which was further normalized to

control samples to obtain ΔΔCt. Three independent cultures were carried out, and all

experiments were performed in triplicate.

NF-κB and NFAT luciferase assay

For NF-κB and NFAT luciferase assay, RAW264.7 cells stably transfected with an NF-κB

luciferase reporter construct (3κB-Luc-SV40) (Wang et al., 2003) or an NFAT luciferase

reporter construct (van der Kraan et al., 2013) were seeded onto a 48-well plate at the density

of 1.5x105 or 5x104 cells per well. The following day, the cells were treated with DHA only

for one hour, followed by stimulation with RANKL for 6 hours (NF-κB luciferase reporter

gene assays) or 24 hours (NFAT luciferase reporter gene assays) inthe presence of DHA.

Cells were then lysed and centrifuged for 20 minutes at 14000g at 4°C. Luciferase activity

was then measured using the Promega luciferase kit and a BMG Polar Star Optima

luminescence reader (BMG, Germany).

Western blot assays

Freshly isolated BMM cells from C57BL/6mice were seeded into six-well plates, and

stimulated with RANKL and M-CSF in the presence of DHA. After 5 days, osteoclasts had

formed and cells were lysed with RIPA buffer for protein extraction. Protein was separated

by SDS-PAGE electrophoresis, and then transferred to a nitrocellulose membrane. The

membrane was blocked in a 5% skim milk for 1 hour then probed with specific antibodies,

including P-ERK, ERK, IKBα, NFATc1, V-ATPase-d2, and β-actin antibodies. After

washing three times with TBST, membranes were incubated with appropriate HRP-

conjugated secondary antibodies. Finally, the membrane was developed using Enhanced

Chemiluminescence (ECL) reagents (Amersham) according to manufacturer’s instructions

and imaged using an Image quant LAS 4000 (GE Healthcare, Australia).

Ovariectomized (OVX) mouse model

The animal experiments were approved by the Institutional Animal Ethics Committee of

Guangxi Medical University. Eighteen C57BL/6 mice aged 7 weeks old were randomly

divided into three groups, which were sham group, OVX group and OVX+DHA (1mg/kg)

group. Each group contained 6 mice. OVX group and OVX+DHA group were given an

ovariectomy operation, while the sham group were given a sham operation as a control. After

the procedure all mice had one week for postoperative recovery. Then the mice in

OVX+DHA group were given intraperitoneal injection of DHA at 1mg/kg for every two

days. The mice in the sham and OVX group were intraperitoneally injected with 4% DMSO

as a control. After 6 weeks, all the mice were sacrificed. Their tibiae were collected and fixed

in 10% neutral buffered formalin. Excess soft tissue was removed and the cleaned tibias were

prepared for microCT analysis by wrapping in tissue soaked in phosphate buffered saline

(PBS) and loading into a 2ml Eppendorf tube to maintain position, hydration, and to prevent

movement artefacts. Tubes were then immobilised using foam inserts in the bed of a Skyscan

1176 microCT instrument (Bruker). Each tibia was then imaged with the following

instrument settings: 50kV, 500μA, 0.5mm Al filter, 952ms exposure, pixel size 8.89 μm, 2

frame averaging, 0.4° rotation step through 180°. Images were then reconstructed using

NRecon software (Bruker) with a constant threshold value. A volume of interest containing

the proximal head of the tibia was generated and this volume was loaded into the analysis

program CTAn (Bruker). A refined volume of interest was then generated 0.5mm below the

growth plate and 1mm in height. The trabecular bone region of interest (ROI) within this

volume was manually defined and bone parameters within this ROI were determined using a

constant threshold for binarisation of the trabecular bone. Following microCT analysis bones

were decalcified and embedded into paraffin blocks for sectioning and staining. Sequential

5μm thick sections were prepared and stained using haematoxylin and eosin, or for TRACP

activity. Sections were scanned using Aperio Scanscope, and bone histomorphometric

analyses were performed using BIOQUANT OSTEO software (Nashville, USA).

Statistical analysis

All values are presented as the mean ± standard deviation (SD) of the values obtained from

three or more experiments. Statistical significance was determined by Student’s t test. A P

value of less than 0.05 was considered to be significant.

RESULTS

DHA inhibits RANKL induced osteoclastogenesis

In order to determine the effect of DHA on the formation of osteoclasts, osteoclastogenesis

assays were performed. BMMs were seeded in 96 well plate at 6x103 per well. After

overnight incubation, DHA was added to the plate every two days until osteoclasts were

formed. The results showed that DHA inhibited RANKL induced osteoclastogenesis in a

dose-dependent manner. DHA significantly inhibited osteoclast formation at doses greater

than 0.5 µM (Fig.1B. C). To determine any cytotoxic effects of DHA on BMM, a cell

survival assay was performed. Our results showed that concentrations of DHA that were

effective at inhibiting osteoclastogenesis (0.5µM and 1µM) did not have any cytotoxic effects

on BMM (Fig.1D). Similar results were also found when BMM were treated with

artemisinin; however, the effect of artemisinin was weaker than that of DHA. The minimum

dose of artemisinin which significantly inhibited osteoclastogenesis was 1µM (Supplementary

Fig. 1). We also assessed the effect of DHA on osteoblast formation and mineralization using

in vitro osteoblastogenesis assays. We found no effect of DHA on osteoblast alkaline

phosphatase staining, and no changes in mineralization in osteoblast cultures (Supplementary

Fig. 2). Thus, for the subsequent studies, DHA was chosen as a prototype to further examine

its mechanism of action in osteoclastogenesis and therapeutic potential in vivo.

DHA suppresses RANKL induced osteoclast function

Hydroxyapatite-coated plates were used to test the effect of 48 hour treatment with DHA on

mature osteoclast resorptive function. The percentage of resorbed area per osteoclast was

reduced in the presence of DHA. The number of osteoclasts was also reduced by DHA but to

a lesser degree relative to resorbing area (Fig.2). These results suggested that DHA restrains

the bone resorbing activity of osteoclasts.

DHA inhibits RANKL-induced gene expression

Real time PCR was used to assess the effect of DHA on RANKL-induced gene expression

levels during osteoclastogenesis. The results showed that the expression of the osteoclast

marker genes cathepsin K, calcitonin receptor and TRAcP were significantly and dose-

dependently inhibited by DHA at day 5 of culture. These results were consistent with

osteoclast formation and activity assays (Fig.3).

DHA suppresses RANKL-induced NF-κB activation and ERK phosphorylation

In order to further explore the mechanism of inhibition by DHA on osteoclast differentiation

and activity, NF-κB luciferase reporter assay and IκBα degradation analysis was performed.

RAW264.7 cells stably transfected with an NF-κB luciferase reporter construct were pre-

treated with different concentrations of DHA up to 5μM for 1 hour, then stimulated with

50ng/ml RANKL for another 6 hours. For Western blot assay, BMMs pre-treated for 1 hour

with 1μM DHA were stimulated with RANKL for 0, 5, 10, 20, 30, 60 minutes. The results

showed that DHA significantly inhibited NF-κB activation in a dose-dependent manner from

the concentration of 0.5 µM (Fig. 4A). The results of Western blot analysis showed that DHA

at 1 µM significantly inhibited IκBα degradation relative to no treatment at 10 minutes post-

stimulation (Fig. 4B. C), and also strongly inhibited the RANKL-induced phosphorylation of

ERK relative to total ERK (Fig. 4B. C).

DHA suppresses RANKL-induced NFAT activation

To examine the effect of DHA on NFAT activation, NFAT luciferase reporter assay was

performed. Similarly to the NF-κB luciferase reporter assay, RAW264.7 cells stably

transfected with an NFAT luciferase reporter construct were first pre-treated with different

concentrations of DHA up to 5 μM for 1 hour, followed by RANKL (50ng/ml) stimulation

for another 24 hours. Cells were then harvested for luciferase activity measurement. The

results showed that NFAT activation was significantly reduced by DHA at concentrations

greater than 0.5 µM. DHA produced a greater than ten-fold reduction in NFAT activation at

the concentrations of 2.5 and 5 µM (Fig. 5A).

NFATc1 is a key transcription factor regulating the differentiation of osteoclasts, and auto-

amplifies its own transcription during osteoclastogenesis [26]. To assess NFATc1 protein

levels, Western blot assay was performed with BMMs stimulated by RANKL (50ng/ml) for

0, 1, 3, and 5 days in the presence or absence of DHA. Protein levels of NFATc1 and V-

ATPase-d2 were significantly reduced by DHA treatment consistent with the observed

reduction in osteoclast formation (Fig. 5B. C). The maximum reduction of V-ATPase-d2 was

at 5 days while the maximum reduction of NFATc1 was at 3 days (Fig. 5B. C).

DHA inhibits ovariectomy-induced bone loss

To determine the effect of DHA on ovariectomy-induced bone loss, mice were either sham

operated, or ovariectomized (OVX) and were then treated with either DHA at a concentration

of 1mg/kg every 2 days, or DMSO only control, by intraperitoneal injection for 6 weeks post-

surgery. The tibias were then collected for micro-CT analysis. Our results showed that DHA

treatment protected the mice from bone loss associated with OVX, as shown by increased

BV/TV in treated mice relative to the OVX mice without treatment (Fig.6). Furthermore,

trabecular separation (Tb.Sp) was significantly reduced in the OVX + DHA group when

compared with OVX control group (Fig.6). These results indicated that DHA protected

against ovariectomy-induced bone loss.

To further confirm the effect of DHA on ovariectomy-induced bone loss, histomorphometric

analysis was also performed. Consistent with the microCT results, there was an increase in

BV/TV in the OVX + DHA group when compared with the OVX group (Fig.7). In addition,

osteoclast surface/bone surface (Oc.S/BS) and osteoclast number/bone surface (N.Oc/Bs)

were decreased in DHA treated OVX mice as compared with OVX group. Consistent with

our in vitro results the osteoblast number/bone surface (N.Oc/Bs) was not significantly

changed following DHA treatment versus the OVX group (Fig.7). This result suggested that

DHA protected against ovariectomy-induced bone loss through inhibiting osteoclast number

and activity.

DISCUSSION

Artemisinin (Qinghaosu) is a natural compound extracted from Artemisia annua, which has

been used to treat malaria in China for more than two thousand years. DHA is a semi-

synthetic derivative of artemisinin and is used in combination with piperaquine for the

treatment of malaria patients, which is an effective remedy for malaria. Recently, DHA was

also found to inhibit cancer cells, such as breast cancer cells [27, 28] and lung cancer cells

[29, 30]. Furthermore, DHA was able to inhibit tumor angiogenesis [31]. DHA was found to

exert these inhibitory effects on cancer cells through regulating the NF-κB pathway [16-19].

In this study, we have explored the effect of DHA on osteoclast formation and activity, and in

an OVX-induced bone loss mouse model, as we found that DHA is more effective than

artemisinin at inhibiting osteoclast formation.

Our results showed that DHA inhibited RANKL induced osteoclastogenesis at concentrations

of 0.5 µM and higher without affecting the viability of BMM; however, mature osteoclast

survival and function was affected at these doses, resulting in significant reductions in

osteoclast activity. We also found that osteoblast formation and function was not affected by

DHA treatment. These findings suggest that DHA might represent an efficient drug prototype

for the treatment of osteoclast-mediated osteolysis.

In concordance with its inhibitory effect on osteoclastogenesis, DHA was found to suppress

NF-κB in a luciferase reporter gene assay and significantly inhibit osteoclast marker gene

expression. The RANKL induced NF-κB pathway is the major signal pathway activated

during osteoclastogenesis [32]. The interaction between RANK and RANKL results in the

rapid degradation of IκB by the proteasome and subsequent release of NF-κB, which then

translocates from the cytoplasm to the nucleus. It then initiates transcription of osteoclast

specific genes necessary for osteoclast differentiation and function [33-36]. Our results have

demonstrated an inhibitory effect of DHA on NF-κB activation through modulating IκBα

protein levels, degradation, and downstream gene expression, indicating that DHA is

inhibiting the NF-κB signaling pathway similar to its known mechanism of action in other

cell types [16-19, 37].

DHA was also able to reduce NFAT activity in a luciferase reporter gene assay, and the

expression of NFAT by Western blot assay. RANKL-induced NFATc1 is another important

signaling pathway leading to osteoclast formation [38, 39]. The NFAT signalling cascade

activates phospholipase C γ (PLCγ) and leads to the release of intracellular Ca2+, resulting in

the activation of calcineurin and transcription and auto amplification of NFATc1 [40]. Our

studies are the first to identify NFATc1 regulation by DHA, although it is interesting to note

that the mechanism of DHA action in Plasmodium falciparum is to disrupt SERCA pumps

resulting in Ca2+ disturbances [41, 42]. Although it has been reported that artemisinin does

not inhibit mammalian SERCA isoforms, DHA has not been assessed. The importance of

Ca2+ signaling in osteoclast NFAT activation may explain the sensitivity of osteoclasts to

DHA. Similar inhibitory effects have been noted on osteoclast survival and differentiation

using other SERCA inhibitors such as thapsigargin [43].

NFAT signalling is also downstream of NF-κB, thus the suppression of NFATc1 by DHA

could potentially result from the suppression of NF-κB signaling [44]. Thus, the inhibitory

effect on NFATc1 by DHA could be due to indirect inhibition of NFAT through NF-κB. Our

results showed that downstream targets of NFAT activation [20], including V-ATPase-d2 that

is important for osteoclast fusion [45, 46], are inhibited by DHA suggesting that suppression

of NFATc1 is an important component of the mechanism of DHA inhibition of RANKL-

induced osteoclastogenesis.

In addition, our western blot results showed that DHA inhibits RANKL-induced ERK

phosphorylation. The binding of RANKL to RANK also activates ERK, which is a mitogen

activated protein kinase (MAPK) [47] that is downstream of tumor necrosis factor receptor-

associated factor 6 (TRAF6) [48]. Inhibition of ERK phosphorylation has been observed in

HUVEC cells treated with DHA [49], and ERK is crucial for the survival of osteoclasts [50].

Hence the observed reduction in ERK phosphorylation in the presence of DHA may account

for the reduced survival that we observed in mature osteoclast cultures. Thus, the effect of

DHA on osteoclasts is multifactorial via the inhibition of multiple signaling pathways

including NF-κB, ERK and NFATc1 signaling pathways.

Based on the in vitro results, an OVX animal model was used for the evaluation of the

function of DHA in osteoporosis. Our results showed that OVX mice were protected from

bone loss by DHA treatment via inhibiting osteoclast formation and function, with no toxicity

in mice. These preclinical experimental results are consistent with the in vitro results, and

suggest that in addition to its anti-malaria [14] and anti-cancer effects [16-19], DHA has a

new potential therapeutic effect against osteoporosis and other lytic bone diseases.

Acknowledgements

The authors acknowledge the facilities, and the scientific and technical assistance of the

National Imaging Facility at the Centre for Microscopy, Characterisation & Analysis, The

University of Western Australia, a facility funded by the University, State and

Commonwealth Governments. This study was supported in part by the National Health and

Medical Research Council of Australia. This study was also supported by a grant from the

National Natural Science Foundation of China (NSFC) (No: 8122801), Arthritis foundation

of Australia, The University of Western Australia (UWA) Research Collaboration Awards

(2014). It is supported by RAINE research fellowship to Dr. Jennifer Tickner and a PhD

scholarship from the Chinese Scholarship Council (Lin Zhou).This project is also sponsored

in part by Co-innovation Centre for Bio-Medicine, Guangxi Colleges and Universities Key

Laboratory of Regenerative Medicine and Innovation Team of Tissue Repair and

Reconstruction to Dr. Jianwen Cheng, Qian Liu, Jun Yao and Tao Wong who were Visiting

Scholars to the University of Western Australia (2014-2015). Prof. Jiake Xu and Dr. Lin

Zhou made mutual collaborative visits to Guangxi Medical University in 2014.

References

1. Goltzman, D., Osteolysis and cancer. J Clin Invest, 2001. 107(10): p. 1219-20. 2. Redlich, K., et al., Tumor necrosis factor alpha-mediated joint destruction is inhibited by

targeting osteoclasts with osteoprotegerin. Arthritis Rheum, 2002. 46(3): p. 785-92. 3. in Bone Health and Osteoporosis: A Report of the Surgeon General2004: Rockville (MD). 4. Harvey, N., E. Dennison, and C. Cooper, Osteoporosis: impact on health and economics. Nat

Rev Rheumatol, 2010. 6(2): p. 99-105. 5. Lacey, D.L., et al., Osteoprotegerin ligand is a cytokine that regulates osteoclast

differentiation and activation. Cell, 1998. 93(2): p. 165-76. 6. Kong, Y.Y., et al., OPGL is a key regulator of osteoclastogenesis, lymphocyte development and

lymph-node organogenesis. Nature, 1999. 397(6717): p. 315-23. 7. Yasuda, H., et al., Osteoclast differentiation factor is a ligand for

osteoprotegerin/osteoclastogenesis-inhibitory factor and is identical to TRANCE/RANKL. Proc Natl Acad Sci U S A, 1998. 95(7): p. 3597-602.

8. Takayanagi, H., Osteoimmunology and the effects of the immune system on bone. Nat Rev Rheumatol, 2009. 5(12): p. 667-76.

9. Zhu, L.L., et al., RANK-L induces the expression of NFATc1, but not of NFkappaB subunits during osteoclast formation. Biochem Biophys Res Commun, 2005. 326(1): p. 131-5.

10. Zou, W. and S.L. Teitelbaum, Integrins, growth factors, and the osteoclast cytoskeleton. Ann N Y Acad Sci, 2010. 1192: p. 27-31.

11. Klayman, D.L., Qinghaosu (artemisinin): an antimalarial drug from China. Science, 1985. 228(4703): p. 1049-55.

12. Lin, A.J., D.L. Klayman, and W.K. Milhous, Antimalarial activity of new water-soluble dihydroartemisinin derivatives. J Med Chem, 1987. 30(11): p. 2147-50.

13. Tu, Y., The development of new antimalarial drugs: qinghaosu and dihydro-qinghaosu. Chin Med J (Engl), 1999. 112(11): p. 976-7.

14. Ashley, E.A., et al., A randomized, controlled study of a simple, once-daily regimen of dihydroartemisinin-piperaquine for the treatment of uncomplicated, multidrug-resistant falciparum malaria. Clin Infect Dis, 2005. 41(4): p. 425-32.

15. Gordi, T. and E.I. Lepist, Artemisinin derivatives: toxic for laboratory animals, safe for humans? Toxicol Lett, 2004. 147(2): p. 99-107.

16. Ji, Y., et al., Anti-tumor effects of dihydroartemisinin on human osteosarcoma. Mol Cell Biochem, 2011. 351(1-2): p. 99-108.

17. Lee, J., et al., Growth inhibitory effect of dihydroartemisinin on Bcr/Abl+ chronic myeloid leukemia K562 cells involve AKT, ERK and NF-kappaB modulation. J Cancer Res Clin Oncol, 2012. 138(12): p. 2095-102.

18. Hwang, Y.P., et al., Suppression of PMA-induced tumor cell invasion by dihydroartemisinin via inhibition of PKCalpha/Raf/MAPKs and NF-kappaB/AP-1-dependent mechanisms. Biochem Pharmacol, 2010. 79(12): p. 1714-26.

19. Chen, H., et al., Growth inhibitory effects of dihydroartemisinin on pancreatic cancer cells: involvement of cell cycle arrest and inactivation of nuclear factor-kappaB. J Cancer Res Clin Oncol, 2010. 136(6): p. 897-903.

20. Feng, H., et al., Myocyte enhancer factor 2 and microphthalmia-associated transcription factor cooperate with NFATc1 to transactivate the V-ATPase d2 promoter during RANKL-induced osteoclastogenesis. J Biol Chem, 2009. 284(21): p. 14667-76.

21. Xu, J., et al., Cloning, sequencing, and functional characterization of the rat homologue of receptor activator of NF-kappaB ligand. J Bone Miner Res, 2000. 15(11): p. 2178-86.

22. Liu, Q., et al., SC-514, a selective inhibitor of IKKbeta attenuates RANKL-induced osteoclastogenesis and NF-kappaB activation. Biochem Pharmacol, 2013. 86(12): p. 1775-83.

23. Bakker, A.D. and J. Klein-Nulend, Osteoblast isolation from murine calvaria and long bones. Methods Mol Biol, 2012. 816: p. 19-29.

24. Chim, S.M., et al., EGFL6 promotes endothelial cell migration and angiogenesis through the activation of extracellular signal-regulated kinase. The Journal of biological chemistry, 2011. 286(25): p. 22035-46.

25. Schneider, C.A., W.S. Rasband, and K.W. Eliceiri, NIH Image to ImageJ: 25 years of image analysis. Nat Methods, 2012. 9(7): p. 671-5.

26. Takayanagi, H., The role of NFAT in osteoclast formation. Ann N Y Acad Sci, 2007. 1116: p. 227-37.

27. Ericsson, T., et al., Population pharmacokinetics of artesunate and dihydroartemisinin during long-term oral administration of artesunate to patients with metastatic breast cancer. Eur J Clin Pharmacol, 2014. 70(12): p. 1453-63.

28. Singh, N.P. and H. Lai, Selective toxicity of dihydroartemisinin and holotransferrin toward human breast cancer cells. Life Sci, 2001. 70(1): p. 49-56.

29. Liu, L.K., et al., Targeted efficacy of dihydroartemisinin for translationally controlled protein expression in a lung cancer model. Asian Pac J Cancer Prev, 2014. 15(6): p. 2511-5.

30. Zuo, Z.J., et al., [Effect of dihydroartemisinin combined irradiation on the apoptosis of human lung cancer GLC-82 cells and its mechanism study]. Zhongguo Zhong Xi Yi Jie He Za Zhi, 2014. 34(10): p. 1220-4.

31. Jung, M., et al., Antiangiogenic activity of deoxoartemisinin derivatives on chorioallantoic membrane. Bioorg Med Chem Lett, 2006. 16(5): p. 1227-30.

32. Xu, J., et al., NF-kappaB modulators in osteolytic bone diseases. Cytokine Growth Factor Rev, 2009. 20(1): p. 7-17.

33. Brown, K.D., E. Claudio, and U. Siebenlist, The roles of the classical and alternative nuclear factor-kappaB pathways: potential implications for autoimmunity and rheumatoid arthritis. Arthritis Res Ther, 2008. 10(4): p. 212.

34. Mercurio, F. and A.M. Manning, Multiple signals converging on NF-kappaB. Curr Opin Cell Biol, 1999. 11(2): p. 226-32.

35. Mercurio, F. and A.M. Manning, NF-kappaB as a primary regulator of the stress response. Oncogene, 1999. 18(45): p. 6163-71.

36. Dixit, V. and T.W. Mak, NF-kappaB signaling. Many roads lead to madrid. Cell, 2002. 111(5): p. 615-9.

37. Dong, F., et al., Dihydroartemisinin targets VEGFR2 via the NF-kappaB pathway in endothelial cells to inhibit angiogenesis. Cancer biology & therapy, 2014. 15(11): p. 1479-88.

38. Hirotani, H., et al., The calcineurin/nuclear factor of activated T cells signaling pathway regulates osteoclastogenesis in RAW264.7 cells. J Biol Chem, 2004. 279(14): p. 13984-92.

39. Matsuo, K., et al., Nuclear factor of activated T-cells (NFAT) rescues osteoclastogenesis in precursors lacking c-Fos. J Biol Chem, 2004. 279(25): p. 26475-80.

40. Kular, J., et al., An overview of the regulation of bone remodelling at the cellular level. Clin Biochem, 2012. 45(12): p. 863-73.

41. Krishna, S., et al., Artemisinins: their growing importance in medicine. Trends in pharmacological sciences, 2008. 29(10): p. 520-7.

42. Eckstein-Ludwig, U., et al., Artemisinins target the SERCA of Plasmodium falciparum. Nature, 2003. 424(6951): p. 957-61.

43. Yip, K.H., et al., Thapsigargin modulates osteoclastogenesis through the regulation of RANKL-induced signaling pathways and reactive oxygen species production. Journal of bone and mineral research : the official journal of the American Society for Bone and Mineral Research, 2005. 20(8): p. 1462-71.

44. Takatsuna, H., et al., Inhibition of RANKL-induced osteoclastogenesis by (-)-DHMEQ, a novel NF-kappaB inhibitor, through downregulation of NFATc1. J Bone Miner Res, 2005. 20(4): p. 653-62.

45. Kim, K., et al., NFATc1 induces osteoclast fusion via up-regulation of Atp6v0d2 and the dendritic cell-specific transmembrane protein (DC-STAMP). Mol Endocrinol, 2008. 22(1): p. 176-85.

46. Lee, S.H., et al., v-ATPase V0 subunit d2-deficient mice exhibit impaired osteoclast fusion and increased bone formation. Nat Med, 2006. 12(12): p. 1403-9.

47. Nakamura, H., et al., Role of osteoclast extracellular signal-regulated kinase (ERK) in cell survival and maintenance of cell polarity. J Bone Miner Res, 2003. 18(7): p. 1198-205.

48. Kobayashi, N., et al., Segregation of TRAF6-mediated signaling pathways clarifies its role in osteoclastogenesis. EMBO J, 2001. 20(6): p. 1271-80.

49. Dong, F., et al., Dihydroartemisinin inhibits endothelial cell proliferation through the suppression of the ERK signaling pathway. International journal of molecular medicine, 2015. 35(5): p. 1381-7.

50. Miyazaki, T., et al., Reciprocal role of ERK and NF-kappaB pathways in survival and activation of osteoclasts. J Cell Biol, 2000. 148(2): p. 333-42.

Figure legends

Fig.1. DHA inhibits RANKL-induced osteoclastogenesis. (A) The molecular structure of Dihydroartemisinin (DHA). (B) Representative images of osteoclast cultures treated with varying concentrations of DHA for 5 days. Mag = 100X, scale bar = 100μm (C) Quantification of TRAcP positive, multinucleated cells following treatment with varying concentrations of DHA. n=3 *p <0.05, **p<0.01, ***p<0.001 relative to DHA-untreated controls. (D) Proliferation of M-CSF stimulated BMM cells following incubation with DHA at different concentrations for 48 hours as measured by MTS assay. n=3. *p <0.05, **p<0.01, ***p<0.001 relative to DHA-untreated controls.

Fig.2. DHA suppresses hydroxyapatite resorption. (A) Representative images of TRAP staining and osteoclastic resorption on hydroxyapatite coated surfaces (scale bars = 500 µm). (B) Quantitative analysis of the number of TRAP positive multinucleated cells. n=3 *p<0.05, **p<0.01, ***p<0.001 relative to DHA-untreated controls. (C) Quantitative analysis of the percentage of the area of hydroxyapatite surface resorbed per osteoclast. n=3. *p<0.05, **p<0.01, ***p<0.001 relative to DHA-untreated controls.

Fig.3. DHA suppresses RANKL-induced gene expression during osteoclastogenesis. Real time-PCR analysis was performed on RNA extracted from cells stimulated for 5 days with RANKL and varying concentrations of DHA. Gene expression of osteoclast marker genes Cathepsin K, Calcitonin Receptor and TRAcP was normalized to 18S RNA and then compared to RANKL-only control samples to obtain the relative fold change. n=3. *p<0.05, **p<0.01, ***p<0.001 relative to RANKL-treated, DHA-untreated controls.

Fig.4. DHA inhibits RANKL-stimulated NF-κB activity and IκB-α degradation. (A) RAW264.7 cells stably transfected with an NF-κB luciferase reporter construct were first treated with DHA for 1 hour, then stimulated with RANKL for 6 hours. The luciferase activity was then measured using a BMG Polar Star Optima luminescence reader. n=3. *p<0.05, **p<0.01, ***p<0.001 relative to RANKL-treated, DHA-untreated controls. (B) BMM cells pretreated with DHA (1µM) for 1 h, were then stimulated with RANKL (100 ng/ml) for 0, 5, 10, 20, 30 and 60 mins. Protein was then extracted for Western Blot analysis using antibodies against IκB-α, β-actin, P-ERK1/2 and total ERK1/2. (C) The ratios of the density of IκB-α bands relative to β-actin bands and P-ERK1/2 bands relative to total ERK1/2 bands were then determined using Image J. n=3. *p<0.05, **p<0.01, ***p<0.001 relative to RANKL-treated, DHA-untreated controls.

Fig.5. DHA inhibits RANKL-stimulated NFAT activity and the expression of NFATc1 and V-ATPase-d2. (A) RAW264.7 cells stably transfected with an NFAT luciferase reporter

construct were first treated with DHA for 1 hour, then stimulated with RANKL for 24 hours. The luciferase activity was then measured by a BMG Polar Star Optima luminescence reader. n=3. *p<0.05, **p<0.01, ***p<0.001 relative to RANKL-treated, DHA-untreated controls. (B) BMM cells were pretreated with DHA (1µM) for 1 h, then stimulated with RANKL (100 ng/ml) for 0, 1, 3, 5 days. Protein was then extracted for Western Blot analysis using antibodies against NFATc1, V-ATPase-d2 and β-actin antibodies. (C) The ratio of the density of NFATc1 bands relative to β-actin bands, and V-ATPase-d2 bands relative to β-actin bands was then determined using Image J. n=3. *p<0.05 relative to RANKL-treated, DHA-untreated controls.

Fig.6. DHA reduces ovariectomy-induced bone loss. (A) Representative 3D reconstructions of trabecular bone from the tibia of sham, OVX and OVX treated with 1mg/kg DHA mice showing the protective effect of DHA treatment following OVX. (B) Quantitative analysis of bone volume/total volume (BV/TV), trabecular separation (Tb.Sp), trabecular number (Tb.N), and trabecular thickness (Tb.Th). n=8. *p<0.05, **p<0.01, ***p<0.001 relative to OVX untreated controls.

Fig.7. DHA protects against ovariectomy-induced bone loss via inhibiting osteoclast activity. (A) Representative images of decalcified bone stained with H&E and TRAP from sham, OVX and OVX treated with 1mg/kg DHA mice. Scale bar =500µM. (B) Quantitative analysis of bone volume/total volume (BV/TV), osteoclast surface/bone surface (Oc.S/BS), osteoclast number/bone surface (N.Oc/BS) and osteoblast number/bone surface (N.Ob/BS). n=3. *p<0.05 and **p<0.01 relative to OVX untreated controls.

Supplementary Fig.1. Artemisinin inhibits RANKL-induced osteoclastogenesis. (A) The molecular structure of artemisinin. (B) Representative images of osteoclast cultures treated with varying concentrations of artemisinin for 5 days. Mag = 100X, scale bar = 100μm (C) Quantification of the effect of artemisinin at different concentrations by counting TRAcP+ multinucleated cells (Nuclei ≥3) as osteoclasts. (C) Quantification of TRAcP positive, multinucleated cells following treatment with varying concentrations of artemisinin. n=3. **p<0.01, ***p<0.001 relative to artemisinin-untreated controls.

Supplementary Fig.2. DHA does not affect osteoblast mineralisation and alkaline phosphatase activity. Cells derived from calvarial bone outgrowth were cultured in osteogenic medium and stained for alkaline phosphatase (ALP) activity and mineralization. (A) Low power image showing ALP staining of calvarial cells. (B) Low power scanned image of tissue culture plate containing calvarial cells treated with DHA and stained with alizarin red. (C) High magnification image of mineralised nodules formed in the presence and

absence of DHA (Mag = 40X). (D) Quantitative analysis of area of ALP staining relative to untreated cells. (E) Quantitative analysis of mineralization area following treatment with DHA relative to untreated cells.