206 The Journal of Clinical Investigation http://www.jci.org Volume 114 Number 2 July 2004
The HIV protease inhibitor ritonavir
blocks osteoclastogenesis and function
by impairing RANKL-induced signaling
Michael W.-H. Wang,1 Shi Wei,1 Roberta Faccio,1 Sunao Takeshita,1 Pablo Tebas,2
William G. Powderly,2 Steven L. Teitelbaum,1 and F. Patrick Ross1
1Department of Pathology and Immunology and 2Department of Medicine, Washington University School of Medicine, St. Louis, Missouri, USA.
Highly active antiretroviral therapy (HAART), which includes HIV protease inhibitors (PIs), has been associ-
ated with bone demineralization. To determine if this complication reflects accelerated resorptive activity,
we studied the impact of two common HIV PIs, ritonavir and indinavir, on osteoclast formation and func-
tion. Surprisingly, we find that ritonavir, but not indinavir, inhibits osteoclast differentiation in a revers-
ible manner and also abrogates bone resorption by disrupting the osteoclast cytoskeleton, without affecting
cell number. Ritonavir given in vivo completely blunts parathyroid hormone–induced osteoclastogenesis in
mice, which confirms that the drug is bone sparing. In keeping with its antiresorptive properties, ritonavir
impairs receptor activator of nuclear factor κB ligand–induced (RANKL-induced) activation of NF-κB and
Akt signaling pathways, both critical to osteoclast formation and function. In particular, ritonavir is found
to inhibit RANKL-induced Akt signaling by disrupting the recruitment of TNF receptor–associated factor
6/c-Src complex to lipid rafts. Thus, ritonavir may represent a bone-sparing PI capable of preventing develop-
ment of osteopenia in patients currently on HAART.
Throughout life, bone remodeling occurs by a finely orches-
trated process of osteoclastic resorption and osteoblastic forma-
tion. When intact, this process ensures maintenance of skeletal
integrity and calcium homeostasis. In all circumstances bone loss
occurs when the activity of the resorptive cell exceeds that of its
anabolic counterpart. For example, postmenopausal osteoporosis
is caused by an absolute increase of osteoclasts and osteoblasts,
with the former activity outpacing that of the latter (1). Senile
(type II) osteoporosis, in contrast, is a low-turnover disease, but
once again bone resorptive activity exceeds that of matrix depo-
sition and calcification (1). Therefore, one approach to dissect-
ing the cause of bone loss in a specific clinical circumstance is
to examine the direct effects of various drugs on generation and
activity of osteoclasts and osteoblasts.
Osteoclasts are multinucleated cells generated by the fusion
of mononuclear progenitors of the monocyte/macrophage
family (2). The pathway involved in osteoclast differentiation
and activation requires two key elements: receptor activator of
nuclear factor κB (RANK), found in osteoclasts and their precur-
sors, and RANK ligand (RANKL), produced by osteoblasts and
stromal cells in the bone marrow (2, 3). In addition, M-CSF is
required for survival and proliferation of osteoclast precursors.
Ligation of RANKL to RANK on macrophages prompts selec-
tive intracellular signals that eventuate in the assumption of the
osteoclast phenotype (4).
Bone loss is a recently described clinical condition in HIV-
infected patients on highly active antiretroviral therapy
(HAART). Prior to the introduction of HAART, HIV-infected
adults generally exhibited normal bone mineral density, which
remained stable with time (5). While one component of HAART,
namely HIV protease inhibitors (PIs), are candidate osteopenic
agents (6–9), a firm link between this family of drugs and bone
loss remains to be established.
To address this issue we examined the effects of two PIs on
osteoblast and osteoclast function. In keeping with the loss of
bone experienced by HAART-treated patients on PI, indinavir
attenuates osteoblast recruitment and the capacity of these
cells to synthesize bone (10). Surprisingly, however, another
PI, ritonavir, while not impacting osteoblasts, exerts similar
repressive effects on the osteoclast. Formation and activation of
osteoclasts is mediated primarily by the activity of the unique
cytokine RANKL (11). We have demonstrated previously that
exposure of osteoclasts or their precursors to IL-4, a molecule
that inhibits osteoclastogenesis and function, blocks several
major RANKL-stimulated signaling pathways (12). Given these
observations, we examined the impact of ritonavir on these
events and find that the PI selectively inhibits NF-κB and Akt
signaling stimulated by the cytokine.
Ritonavir inhibits osteoclastogenesis in vitro. To determine the effects
of PIs on osteoclastogenesis, we turned to pure (>99%) popu-
lations of bone marrow macrophages that, in the presence of
M-CSF and RANKL, differentiate into multinucleated cells
phenotypically and functionally indistinguishable from authen-
tic osteoclasts. While addition of the osteoblast-inhibiting PI
indinavir (10) does not impact the osteoclastogenic process,
ritonavir dose dependently impairs osteoclast formation with an
IC50 of approximately 10 µg/ml (Figure 1, A and B). Reflecting
Nonstandard abbreviations used: constitutively active (CA); electrophoretic
mobility shift assay (EMSA); extracellular signal-related kinase (ERK); highly active
antiretroviral therapy (HAART); parathyroid hormone (PTH); protease inhibitor (PI);
RANK ligand (RANKL); receptor activator of nuclear factor κB (RANK); tartrate-resis-
tant acid phosphatase (TRAP); TNF receptor–associated factor 6 (TRAF6).
Conflict of interest: The authors have declared that no conflict of interest exists.
Citation for this article: J. Clin. Invest. 114:206–213 (2004).
The Journal of Clinical Investigation http://www.jci.org Volume 114 Number 2 July 2004
the drug’s inhibitory effect on morphological osteoclastogene-
sis, the PI blunts the expression of a range of osteoclast-defining
genes in a parallel fashion (Figure 1C), indicating that the drug
acts at an early stage of osteoclast differentiation.
Ritonavir inhibition of osteoclast formation is reversible. To exclude
the possibility that ritonavir exerts a toxic effect on osteoclast
precursors we asked if its effect on osteoclastogenesis is revers-
ible. Thus, the drug was added at the beginning of osteoclast-
generating cultures at a dose that completely suppresses
osteoclast formation (20 µg/ml) and was withdrawn on day 3,
4, or 5. Removal of the drug, up to 5 days after its addition, nor-
malizes osteoclast differentiation by day 7 (Figure 2, A and B).
Apoptosis assay of differentiating cells exposed to ritonavir (up
to 10 µg/ml) shows no alterations (data not shown). Cessation
of osteoclastogenesis by ritonavir therefore reflects arrested dif-
ferentiation, rather than toxicity.
Ritonavir inhibits bone resorption by mature osteoclasts. We next
turned to the effect of PIs on the resorptive activity of mature
osteoclasts. To this end, ritonavir (10 µg/ml) or indinavir (10 µg/
ml) was added to cultures after mature osteoclasts had been gen-
erated on whale dentine slices in the absence of drugs. Osteoclast
number and bone resorption pits were determined 2 days later.
As is evident in Figure 3A, ritonavir fails to decrease the num-
ber of mature osteoclasts as determined by tartrate-resistant
acid phosphatase (TRAP) cytochemistry of whale dentine slices
(mean osteoclast number per randomly selected field is 54 ± 7
and 62 ± 12 in vehicle and ritonavir-treated cultures, respective-
ly), reaffirming the nontoxic effect of the drug on osteoclasts
and their precursors. In contrast, pit number, area, and depth
are dramatically decreased with ritonavir treatment (Figure 3, A
and B). Unlike ritonavir, indinavir fails to alter either the num-
ber or functional capacity of mature osteoclasts.
Osteoclastogenesis is impaired by ritonavir but not indinavir. (A)
Osteoclasts were generated from bone marrow macrophages stimu-
lated with RANKL and M-CSF for 4 days in the presence of the indi-
cated doses of ritonavir or indinavir. TRAP solution assay quantitation
of osteoclast formation shows that the IC50 for ritonavir is near 10
µg/ml. In contrast, cultures exposed to indinavir show no inhibition
or enhancement of osteoclast formation. (B) Representative fields of
TRAP-stained osteoclasts in the presence of control medium, indinavir
(10 µg/ml), and ritonavir (10 µg/ml). Magnification, ×100. (C) Ritonavir
dose dependently suppresses osteoclast gene markers determined by
RT-PCR analysis of osteoclasts on day 4 culture.
Osteoclastogenic arrest by ritonavir is reversible. (A) Osteoclasts were
generated from bone marrow macrophages as in Figure 1 and quan-
titated by TRAP solution assay. The presence of ritonavir (20 µg/ml)
completely suppresses osteoclast formation, while normalization of
osteoclast number follows withdrawal of the PI on day 3, 4, and 5.
+ Ritonavir, continuous ritonavir exposure; – Ritonavir, ritonavir with-
drawal. (B) TRAP-stained osteoclasts on day 7 of culture following
persistent exposure to ritonavir (left panel) and withdrawal of ritonavir
on day 5 (right panel). Magnification, ×200.
208 The Journal of Clinical Investigation http://www.jci.org Volume 114 Number 2 July 2004
Ritonavir inhibits parathyroid hormone–induced osteoclastogenesis in vivo.
Having established that ritonavir arrests osteoclastogenesis in vitro,
we asked if the same is obtained in vivo. For these experiments we
turned to a murine model of acute osteoclast formation induced by
supracalvarial injection of the osteoclastogenic agent, parathyroid
hormone (PTH) (13, 14), with concomitant intraperitoneal admin-
istration of ritonavir. PTH stimulates osteoclastogenesis by target-
ing directly bone marrow stromal cells to increase the expression of
RANKL, thus providing an optimal cellular milieu for osteoclast
formation (3, 15). Similar to its in vitro effects, ritonavir totally
abrogates PTH-induced osteoclast formation when administered
systemically (Figure 4, A and B). PTH is also a known inducer of
stromal cell proliferation. The increase in stromal cells surround-
ing the bone matrix is not blocked by ritonavir, as seen in Figure
4B and by quantitation (percentage of stromal cell area per total
bone marrow in drug-exposed and untreated animals is 75 ± 12 and
78 ± 15, respectively), arguing that ritonavir inhibits osteoclast for-
mation by targeting the osteoclast precursors rather than impair-
ing PTH action on bone marrow stromal cells.
Ritonavir inhibits RANKL-induced signal transduction. RANKL is essen-
tial for both osteoclast differentiation and resorptive function (2, 3).
Thus, one potential mechanism by which ritonavir impairs osteoclast
Osteoclast function is impaired by ritonavir. (A) Osteoclasts, generated on whale dentine slices for 3 days by treatment with RANKL and M-CSF,
were exposed to control medium, ritonavir (10 µg/ml), or indinavir (10 µg/ml) for an additional 2 days. TRAP-stained dentine slices show no change
in osteoclast number with exposure to PIs (top panels). Following cell removal, Coomassie blue staining of dentine slices show decreased bone
pits with ritonavir treatment (bottom panels). Magnification, ×100. (B) Ritonavir, but not indinavir, decreases pit number (per 0.36 mm2), percentage
of pit area, and pit depth (in micrometers) by half (each P < 0.01).
Ritonavir blocks PTH-induced osteoclast formation in vivo. (A) Osteoclast number was determined from TRAP-stained histologic sections of
calvariae from mice stimulated with PTH or vehicle and intraperitoneally injected with ritonavir or vehicle. Ritonavir abrogates the osteoclast
increase stimulated by PTH (n = 3 mice per group; P < 0.05). Cal. inj., calvarial injection; i.p. inj., intraperitoneal injection; Veh, vehicle. (B) Rep-
resentative fields of TRAP-stained sections of calvariae show PTH injection fails to induce osteoclast formation in ritonavir-treated mice despite
a robust PTH-dependent stromal cell response. Magnification, ×250.
The Journal of Clinical Investigation http://www.jci.org Volume 114 Number 2 July 2004
formation and function involves the blockade of RANKL-induced
signal transduction. To test this possibility, we examined the effect
of ritonavir on the known major signaling pathways activated by
RANKL ligation of its receptor RANK. Specifically, we analyzed the
NF-κB pathway, the Akt axis, and all three MAPK members, namely
extracellular signal-related kinases (ERKs; p42/p44), stress activated
protein kinase (SAPK/JNK), and p38. Since complete suppression of
osteoclast formation is achieved with 20 µg/ml of ritonavir, this con-
centration was used for all subsequent cell-signaling studies.
NF-κB transcription factors are known to play a pivotal role
in osteoclast formation as manifested by p50/p52 NF-κB dou-
ble-deficient mice, which exhibit osteopetrosis due to a lack
of osteoclasts (16, 17). Examination of ritonavir’s impact on
RANKL-induced NF-κB activation reveals impaired degrada-
tion of IκBα (Figure 5A), the key NF-κB–binding protein, which
retains the transcription factor in the cytosol. In consequence,
the amount of NF-κB that translocates to the nucleus is also
decreased, as determined by electrophoretic mobility shift assay
(EMSA; Figure 5B). Of considerable interest, phosphorylation
of IκBα, a prerequisite for its subsequent degradation, is
unimpaired by ritonavir (Figure 5C), indicating that ritonavir
does not abolish proximal signaling events leading to IκBα
phosphorylation but suppresses IκBα degradation.
RANKL activates the Akt signal cascade through a signaling
complex containing c-Src and TNF receptor–associated factor 6
(TRAF6), which eventually results in the phosphorylation and
activation of Akt (18). To examine RANKL-induced Akt activation
in the absence of M-CSF, which is itself a strong Akt activator,
we used RAW 264.7 cells, an M-CSF–independent monocytic cell
line that can readily differentiate into osteoclasts with RANKL
alone. As determined by Western blot analysis, ritonavir impairs
RANKL-induced phosphorylation of Akt (Figure 6A). Confirm-
ing the functional consequences of this finding, phosphorylation
of the forkhead transcription factor (FKHR), a downstream
phosphorylation substrate of Akt (19), is also suppressed by the PI
(Figure 6A). The impairment of ritonavir-associated Akt activity
was similarly observed with primary bone marrow macrophages
stimulated with RANKL (data not shown). Given that M-CSF, a
cytokine essential for osteoclast survival, is also a potent activa-
tor of Akt, we asked whether the ability of ritonavir to block Akt
activation is a cytokine-specific event. Thus, we examined Akt sig-
naling in primary macrophage-derived osteoclasts that had been
starved of cytokines and serum and then stimulated with RANKL
or M-CSF separately. While ritonavir blunts RANKL-induced Akt
activation, it does not affect that stimulated by M-CSF (Figure 6B),
a result consistent with the in vitro observation that cell survival
is unaffected by ritonavir.
Finally, we find that ritonavir has no effect on the three major
MAPK signal cascades, ERKs, SAPK/JNK, and p38, activated by
RANKL in bone marrow macrophages, osteoclasts, and RAW cells
(data not shown). Thus, ritonavir selectively arrests RANKL-induced
activation of NF-κB and Akt, while sparing the MAPK pathways.
Constitutively active PI3K rescues ritonavir-induced inhibition of Akt
signaling and osteoclast function. To elucidate the molecular basis
of Akt pathway impairment by ritonavir, a constitutively active
Ritonavir inhibits RANKL-induced NF-κB activation. (A) Bone mar-
row macrophages pretreated with ritonavir (20 µg/ml) or vehicle for 1
hour were stimulated with RANKL for the indicated time points, and
protein lysates were prepared for IκBα immunoblot. Ritonavir pretreat-
ment impairs RANKL-induced degradation of IκBα. (B) Bone marrow
macrophages pretreated with ritonavir (20 µg/ml) for the indicated time
were stimulated with RANKL for 15 minutes and evaluated for NF-κB
activation. Ritonavir inhibits NF-κB activation as assessed by EMSA after
1 hour and after 18 hours of pretreatment (Pretx). Numbers below lanes
indicate relative band intensity. (C) Cell lysates prepared as in A and
immunoblotted for phospho-IκBα (p-IκBα) and IκBα reveal similar levels
of phospho-IκBα intensity, irrespective of ritonavir pretreatment, despite
the failure of ritonavir pretreatment to decrease total IκBα levels. β-Actin
immunoblots confirm similar amounts of cell extracts were analyzed.
Ritonavir inhibits RANKL-induced Akt activation. (A) RAW 264.7
cells pretreated with ritonavir (2 µg/ml) for 1 hour and stimulated with
RANKL (+RANKL) for the indicated time points. Phospho-Akt (p-Akt),
total Akt (Akt), and phospho-FKHR (p-FKHR) immunoblots were per-
formed on cell extracts. Immunoblots reveal impaired Akt and FKHR
phosphorylation with ritonavir pretreatment. (B) Osteoclasts were stim-
ulated with either RANKL or M-CSF for the indicated time points, and
cell extracts were immunoblotted for Akt activation. Ritonavir inhibits
only RANKL-induced Akt activation but not that stimulated by M-CSF.
Total Akt immunoblots confirm that similar amounts of cell extracts
were analyzed. +V, vehicle pretreatment; +R, ritonavir pretreatment.
210 The Journal of Clinical Investigation http://www.jci.org Volume 114 Number 2 July 2004
(CA) PI3K (PI3K-CA) containing retrovirus was constructed. The
activating mutation is generated by the addition of the avian src
myristoylation sequence at the N terminus of the p110α com-
ponent of the PI3K. Akt signaling by RANKL in preosteoclasts
containing PI3K-CA, but not vector control, are rescued from
inhibition by ritonavir (Figure 7A).
Osteoclast actin ring formation and bone resorption is known
to be inhibited by PI3K inhibitors such as Wortmannin. Given
that ritonavir inhibits osteoclast function (Figure 3, A and B), we
determine if the molecular rescue of RANKL-induced Akt sig-
nal through the introduction of PI3K-CA can be demonstrated
functionally. To this end, mature osteoclasts, transduced with
vector or PI3K-CA retroviruses, were exposed to ritonavir and
examined by immunofluorescence microscopy for β-actin after
2 hours of exposure. The reorganization and assembly of actin
ring reflects cell polarization and is essential for the resorptive
process. In vector control, ritonavir dose dependently disrupts
the characteristic actin ring at the osteoclast periphery (Figure
7, B and C). At the highest dose, complete actin ring dissolu-
tion is evident. In contrast, actin rings of PI3K-CA–transduced
osteoclasts are intact even at the highest dose. These results
indicate that ritonavir-induced inhibition of osteoclast function
is attributable to the disruption of the PI3K-signaling axis.
Ritonavir inhibits TRAF6 and c-Src recruitment to lipid rafts fol-
lowing RANKL stimulation. The activation of Akt pathway by
RANKL has been shown to require the recruitment of TRAF6-
Src-PI3K complex to the membrane surface by ligand-activated
oligomerization of RANK. The ability of PI3K to rescue Akt sig-
naling indicates that ritonavir inhibition of this signaling axis
is at or upstream of PI3K. To examine the membrane recruit-
ment of TRAF6 and c-Src, starved preosteoclasts pretreated with
ritonavir were exposed to RANKL for 5 minutes, followed by the
isolation of the lipid raft membrane fraction. The recruitment
of c-Src and TRAF6 by RANKL to the lipid raft was inhibited
by ritonavir treatment (Figure 8). Interestingly, TRAF2, another
TRAF protein known to be associated with RANK, continues to
be recruited to lipid rafts following RANKL simulation, indicat-
ing ritonavir specificity in TRAF6 inhibition.
Reports that HIV PI usage is associated with bone demineralization
(6–9) prompted us to investigate the direct effect of this class of drugs
on the two main cellular determinants of bone mass, osteoblasts and
osteoclasts. We noted that HIV PIs have differential effects on osteo-
blast and osteoclast maturation and function, despite targeting the
same catalytic site of the retroviral protease. We reported previously
that indinavir, but not ritonavir, is unique in inhibiting osteoblast
maturation (10). In contrast, this study demonstrates that ritonavir
selectively inhibits osteoclast maturation and function.
Introduction of PI3K-CA restores RANKL-induced
phosphorylation of Akt and osteoclast actin ring formation
in the presence of ritonavir. (A) Retroviral transduction of
either vector or PI3K-CA (p110�-CA) into bone marrow
macrophages was followed by 3 days of culture in selection
media, M-CSF, and RANKL. After starvation (3 hours) and
pretreatment (1 hour) with either vehicle or ritonavir, cells
were stimulated with RANKL for 15 minutes. Immunoblots
reveal restoration of RANKL-induced Akt phosphorylation
when PI3K-CA is introduced. As expected, total PI3K is
enhanced as a result of transduction (p110α blot). TRAF6
Western blots act as a loading control. (B) Percentage of
osteoclasts with intact actin rings after ritonavir exposure
is quantitated. (C) Osteoclasts, retrovirally transduced with
either vector or PI3K-CA, were generated on glass coverslips.
After 4 days, cells were exposed to various doses of ritonavir
for 2 hours, then processed for immunofluorescence micros-
copy for β-actin. Dose-dependent disruption of the charac-
teristic actin ring of the osteoclast cytoskeleton is observed
in vector but not PI3K-CA–transduced cells.
RANKL-induced recruitment of c-Src and TRAF6 to lipid raft com-
ponent is inhibited by ritonavir treatment. Preosteoclasts generated
after 3 days of culture with RANKL and M-CSF were starved for 3
hours, followed by pretreatment with either vehicle or ritonavir (1
hour). Cells were then stimulated with RANKL for 5 minutes, followed
by lipid raft isolation. Note the continued recruitment of TRAF2 to lipid
rafts with ritonavir treatment, indicating that the inhibition is specific
to TRAF6 and c-Src.
The Journal of Clinical Investigation http://www.jci.org Volume 114 Number 2 July 2004
Our in vitro findings reveal that ritonavir directly targets
osteoclast precursors, arresting their differentiation in a reversible
manner. The PI also directly impacts mature osteoclasts, attenuat-
ing their resorptive capacity. This arrest of bone degradation does
not reflect cytotoxicity, but rather disruption of the osteoclast
cytoskeleton, a complex required for proper polarization and
subsequent efficient removal of bone. The in vivo consequences
of the antiosteoclastic properties of ritonavir are manifest by the
ability of the drug to block PTH-induced osteoclastogenesis.
RANKL is a member of the TNF superfamily with the unique
capacity to directly induce osteoclast precursors to differenti-
ate (2). Reflecting its profound impact on osteoclast biology,
recent years have witnessed characterization of the major signal-
ing pathways emanating from its receptor, RANK (12, 18, 20,
21). We find that ritonavir-mediated inhibition of osteoclasto-
genesis induced by RANKL reflects blocking of specific com-
ponents of these intracellular signals. Interestingly, ritonavir
inhibits RANKL-induced NF-κB activation, but does so with-
out altering IκBα phosphorylation, a process that usually leads
to degradation of this NF-κB–binding protein. This failure of
NF-κB activation in the face of phosphorylated IκBα indicates
that the PI impacts NF-κB by preventing the degradation of its
major cytosolic-binding protein. In keeping with this conclu-
sion, ritonavir is an established proteasome inhibitor (22, 23).
We also find that ritonavir impairs RANKL-induced but not
M-CSF–induced Akt activation. Given that RANKL modulates
bone resorption, whereas M-CSF affects survival of mature
osteoclasts (2, 3), this differential signal inhibition is consis-
tent with our in vitro observation of impaired bone resorption
without cell death. Furthermore, in mature osteoclasts the ERK
signal is responsible for osteoclast survival, while stimulation of
the NF-κB pathway activates bone resorption (24). The observa-
tion that ritonavir inhibits NF-κB but not the MAPK pathways
is in line with the capacity of ritonavir to impair bone resorption
in vitro without altering osteoclast number.
We further delineated the specific pathway within the Akt-sig-
naling axis altered by ritonavir, namely the failure of TRAF6 and
c-Src to translocate to lipid rafts following RANKL stimulation.
The observation is consistent with the ability of a constitutively
active form of PI3K, a known downstream signal activated by
the TRAF6/c-Src complex, to rescue both Akt signaling and
osteoclast cytoskeletal disruption caused by ritonavir. The exact
mechanism by which ritonavir inhibits the membrane recruit-
ment of TRAF6 and c-Src remains to be determined, however.
Reports associating PIs with decreased bone mineral density
have assumed all members of this class of drugs are implicated.
The fact that ritonavir, but not indinavir, is antiosteoclastic
indicates such is not the case. In this regard, others have dem-
onstrated unique properties of these drugs in vitro (22, 25, 26).
Also, clinical studies examining therapy-induced adverse effects
have noted a stronger association of hypertriglyceridemia with
ritonavir compared with other PIs (27, 28).
Recently, Liang et al. proposed that PI-induced hyperlipidemia
may be caused by an impaired apoB degradation due to protea-
somal inhibition by ritonavir (29). Concerns have been raised
regarding the pharmacological relevance of the concentrations
used in these experiments (2 µg/ml as the minimal concentra-
tion of ritonavir to exert an effect) (30, 31). Our study showed a
50% inhibition of osteoclast formation at a substantially lower
ritonavir concentration (10 µg/ml). Moreover, ritonavir detect-
ably blunts osteoclastogenesis at levels as low as 2.5 µg/ml
(Figure 1A). Therefore, the plasma concentrations used in this
study are within the clinical maximum and minimum values for
ritonavir (11.2 ± 3.6 and 3.7 ± 2.6 µg/ml, respectively) and are
likely to be pharmacologically relevant.
It is likely that the problem of bone demineralization in
HIV-infected patients is complex and multifactorial in origin.
Features of HIV itself, especially the increased catabolic rate
and associated weight loss, as well as the wasting syndrome
seen in advanced AIDS, may contribute to an increased risk of
osteopenia in patients. The effect of antiretroviral treatment
may also have multiple components, including a direct effect of
decreased HIV activity in bone, indirectly because of changes in
the immune status or cytokine milieu, or as a direct toxic effect
of the drugs used to treat HIV infection. Although we have dem-
onstrated an effect of PIs on bone metabolism, patients receive
combination therapies, and the other antiretroviral agents may
also have effects that have yet to be elucidated.
Thus, despite the fact that ritonavir and indinavir are designed
to arrest the same antiretroviral target, ritonavir is unique in
dampening osteoclast function and may be a promising agent in
preventing bone loss in HIV-infected individuals. Furthermore,
our molecular dissection of signaling pathways specifically
impacted by ritonavir may aid in characterizing the mechanisms
by which individual PIs prompt idiosyncratic side effects.
Reagents. Ritonavir and indinavir were from Abbott Laboratories
(Abbott Park, Illinois, USA) and Merck & Co. Inc. (West Point,
Pennsylvania, USA), respectively. All immunoblotting Ab’s were
from Cell Signaling Technology (Beverly, Massachusetts, USA).
Protease inhibitor mixtures used for cell lysis were from Calbio-
chem-Novabiochem (San Diego, California, USA). The bicin-
choninic acid kit for protein determination and ECL kits were
obtained from Pierce Biotechnology Inc. (Rockford, Illinois,
USA). Recombinant murine M-CSF was from R&D Systems Inc.
(Minneapolis, Minnesota, USA). Murine RANKL was expressed
in our laboratory as described previously (32). All other chemicals
were obtained from Sigma-Aldrich (St. Louis, Missouri, USA).
Primary cells and cell line. Macrophages/osteoclast precursors and
osteoclasts were generated from bone marrow precursors as described
(12). RAW 264.7 cells, obtained from American Type Culture Collec-
tion (Manassas, Virginia, USA), were maintained as described (33).
Characterization of osteoclasts. Bone marrow macrophages were cul-
tured in 48- or 96-well cell culture dishes in the presence of M-CSF
(10 ng/ml), RANKL (100 ng/ml), and HIV PIs as appropriate, and
medium was changed on day 3. Osteoclast-like cells were charac-
terized by staining for TRAP activity. The number of osteoclasts
was assessed by TRAP solution assay as described previously (34).
RNA extraction and amplification by RT-PCR. RNA purification
and RT-PCR conditions were as described previously (12). The
oligonucleotide primers used are as follows: β3 integrin, 5′-
TTACCCCGTGGACATCTACTA-3′ and 3′-AGTCTTCCATC-
CAGGGCAATA-5′; calcitonin receptor, 5′-CATTCCTGTACTT-
GGTTGGC-3′ and 3′-AGCAATCGACAAGGAGTGAC-5′;
cathepsin K, 5′-GGAAGAAGACTCACCAGAAGC-3′ and 3′-
GTCATATAGCCGCCTCCACAG-5′; matrix metalloproteinase
9 (MMP9), 5′-CCTGTGTGTTCCCGTTCATCT-3′ and 3′-CGCT-
GGAATGATCTAAGCCCA-5′; GAPDH, 5′-ACTTTGTCAAGCT-
CATTTCC-3′ and 3′-TGCAGCGAACTTTATTGATG-5′.
212 The Journal of Clinical Investigation http://www.jci.org Volume 114 Number 2 July 2004
Bone resorption. Osteoclasts were generated on whale dentin slic-
es from bone marrow macrophages using conditions described
above. After 3 days of culture to generate osteoclasts, PIs were
added to the culture for 2 days. At the end of the experiment, cells
were TRAP stained and photographed to document cell number.
Cells were then removed from the dentin slices with 0.5 M ammo-
nium hydroxide and mechanical agitation. Maximum resorption
lacunae depth was measured using a confocal microscope (Micro-
radiance; Bio-Rad Laboratories Inc., Hercules, California, USA) as
described (35). For evaluation of pit number and resorbed area,
dentin slices were stained with Coomassie brilliant blue and ana-
lyzed with light microscopy using Osteomeasure software (Osteo-
metrics Inc., Decatur, Georgia, USA) for quantitation.
Apoptosis assay. Apoptotic cells were assessed using the cell
death detection ELISAPLUS kit (Roche Molecular Biochemicals,
Mannheim, Germany). In brief, equal amounts of protein from
osteoclastogenic cultures treated with vehicle or ritonavir were
incubated with a mixture of Ab’s to histones and DNA conjugated,
respectively, with biotin and peroxidase. The complex was captured
on streptavidin-coated wells, and DNA content was quantitated
colorimetrically using a peroxidase substrate.
Immunostaining. Osteoclasts transduced with vector or constitu-
tively active PI3K were generated on glass coverslips using meth-
ods described previously (35). After exposing cells to the indicated
doses of ritonavir or vehicle control for 2 hours, cells were fixed in
4% paraformaldehyde, permeabilized in 0.1% Triton X-100, rinsed
in PBS, and immunostained with Alexa 488-phalloidin (Molecular
Probes Inc., Eugene, Oregon, USA).
PTH-induced osteoclastogenesis in vivo. To measure responses
of murine calvarial bone to PTH, we followed the procedures
developed by Boyce and coworkers (13, 14, 36). Synthetic human
PTH(1-34) was obtained from Bachem California Inc. (Torrance,
California, USA) and dissolved in vehicle (1 mM HCl, 0.1% BSA).
PTH, at 10 µg in 25 µl or vehicle in the same volume, was inject-
ed subcutaneously four times daily for 3 days into the subcuta-
neous tissue overlying the calvariae, using a Hamilton syringe.
During this time, ritonavir (1 mg) or vehicle (50 µl of 25% etha-
nol) was injected intraperitoneally twice daily for 3 days. After
sacrifice by CO2 narcosis, calvariae were removed intact, soft tis-
sues were gently dissected, and the calvariae were fixed in 10%
phosphate-buffered formalin for 24 hours and further processed
as described (14). Samples were analyzed with light microscopy
using Osteomeasure software (Osteometrics Inc.). All experi-
ments in vivo were approved by the Washington University Ani-
mal Studies Committee.
Cell stimulation and immunoblotting. For NF-κB and MAPK signal-
ing experiments, in vitro–differentiated bone marrow macrophages
were stimulated by adding RANKL in the presence or absence
of ritonavir as indicated. For Akt activation, macrophages were
cultured in serum and M-CSF–free medium for 24 hours before
stimulation. All RANKL-induced signaling in RAW 264.7 cells
was performed in nonserum-starved conditions. Ritonavir was
added to cultures at 10 µg/ml 1 hour prior to stimulation unless
otherwise indicated. This dosage level was based on preliminary
data, reported here, that the drug does not cause toxicity to cells
in the osteoclast lineage. After stimulation, cells were subjected to
immunoblotting analysis as described (12).
EMSA. EMSA was performed as described previously (12).
Membrane lipid rafts isolation. Lipid rafts isolation was performed
using the detergent-free method of Smart et al. (37).
Statistical analysis. Statistical significance was analyzed using an
Excel spreadsheet program (Microsoft Corp., Redmond, Washing-
ton, USA) using an unpaired t test.
We thank Deborah Novack for reviewing the manuscript and
Paulette Shubert for expert secretarial assistance. This study was
supported by the following NIH grants: training grant AR-07033
and F32 fellowship AR-08648 to M.W.-H. Wang; AI-25903 to W.G.
Powderly and P. Tebas; AR-32788, AR-46523, and DE-05413 to S.L.
Teitelbaum; and AR-46852 and AR-48812 to F.P. Ross.
Received for publication April 25, 2002, and accepted in revised
form May 28, 2004.
Address correspondence to: F. Patrick Ross, Department of
Pathology and Immunology, Washington University School of
Medicine, Campus Box 8118, 660 South Euclid Avenue, St. Louis,
Missouri 63110, USA. Phone: (314) 454-8079; Fax: (314) 454-5505;
1. Manolagas, S.C. 2000. Birth and death of bone
cells: basic regulatory mechanisms and impli-
cations for the pathogenesis and treatment of
osteoporosis. Endocr. Rev. 21:115–137.
2. Teitelbaum, S.L. 2000. Bone resorption by
osteoclasts. Science. 289:1504–1508.
3. Suda, T., et al. 1999. Modulation of osteoclast dif-
ferentiation and function by the new members of
the tumor necrosis factor receptor and ligand
families. Endocr. Rev. 20:345–357.
4. Ross, F.P. 2000. RANKing the importance of
measles virus in Paget’s disease. J. Clin. Invest.
5. Paton, N.I.J., Macallan, D.C., Griffin, G.E., and
Pazianas, M. 1997. Bone mineral density in
patients with human immunodeficiency virus
infection. Calcif. Tissue Int. 61:30–32.
6. Tebas, P., et al. 2000. Accelerated bone mineral
loss in HIV-infected patients receiving potent
antiretroviral therapy. AIDS. 14:F63–F67.
7. Moore, A.L., et al. 2001. Reduced bone min-
eral density in HIV-positive individuals. AIDS.
8. Brown, T., et al. 2002. Bone loss associated with
abnormalities in glucose metabolism in HIV patients
on protease inhibitors. In Programs and Abstracts of the
Ninth Conference of Retroviruses and Opportunistic Infec-
tions. Seattle, Washington, USA. 314.
9. Arnsten, J.H., Freeman, R., Santoro, N., Kunins,
H., and Schoenbaum, E.E. 2002. Bone mineral
density and protease inhibitor use in older HIV-
infected women. In Programs and Abstracts of the
Ninth Conference on Retroviruses and Opportunistic
Infections. Seattle, Washington, USA. 314.
10. Wang, M.W., Wei, S., Teitelbaum, S.L., and Ross,
F.P. 2001. The HIV protease inhibitor indinavir
uniquely inhibits bone formation (abstract).
J. Bone Miner. Res. 16(Suppl.):S372.
11. Teitelbaum, S.L., and Ross, F.P. 2003. Genetic
regulation of osteoclast development and func-
tion. Nat. Rev. Genet. 4:638–649.
12. Wei, S., Wang, M.W., Teitelbaum, S.L., and Ross,
F.P. 2002. Interleukin-4 reversibly inhibits osteo-
clastogenesis via inhibition of NF-κB and MAP
kinase signaling. J. Biol. Chem. 21:6622–6630.
13. Yates, A.J.P., et al. 1988. Effects of a synthetic
peptide of a parathyroid hormone-related protein
on calcium homeostasis, renal tubular calcium
reabsorption, and bone metabolism in vivo and
in vitro in rodents. J. Clin. Invest. 81:932–938.
14. Zhao, W.G., Byrne, M.H., Boyce, B.F., and Krane,
S.M. 1999. Bone resorption induced by parathy-
roid hormone is strikingly diminished in col-
lagenase-resistant mutant mice. J. Clin. Invest.
15. Ma, Y.L., et al. 2001. Catabolic effects of continu-
ous human PTH (1-38) in vivo is associated with
sustained stimulation of RANKL and inhibition
of osteoprotegerin and gene-associated bone for-
mation. Endocrinology. 142:4047–4054.
16. Franzoso, G., et al. 1997. Requirement for NF-κB
in osteoclast and B-cell development. Genes Dev.
17. Iotsova, V., et al. 1997. Osteopetrosis in mice
lacking NF-kappaB1 and NF-kappaB2. Nat. Med.
18. Wong, B.R., et al. 1999. TRANCE, a TNF family
member, activates Akt/PKB through a signaling
complex involving TRAF6 and c-Src. Mol. Cell.
19. Brunet, A., et al. 1999. Akt promotes cell survival
by phosphorylating and inhibiting a Forkhead
transcription factor. Cell. 96:857–868.
20. Matsumoto, M., Sudo, T., Saito, T., Osada, A.,
and Tsujimoto, M. 2000. Involvement of p38
research article Download full-text
The Journal of Clinical Investigation http://www.jci.org Volume 114 Number 2 July 2004
mitogen-activated protein kinase signaling path-
way in osteoclastogenesis mediated by receptor
activator of NF-κB ligand (RANKL). J. Biol. Chem.
21. Kobayashi, N., et al. 2001. Segregation of TRAF6-
mediated signaling pathways clarifies its role in
osteoclastogenesis. EMBO J. 20:1271–1280.
22. Andre, P., et al. 1998. An inhibitor of HIV-1
protease modulates proteasome activity, antigen
presentation, and T cell responses. Proc. Natl.
Acad. Sci. U. S. A. 95:13120–13124.
23. Schmidtke, G., et al. 1999. How an inhibitor of the
HIV-I protease modulates proteasome activity.
J. Biol. Chem. 274:35734–35740.
24. Miyazaki, T., et al. 2000. Reciprocal role of ERK
and NF-κB pathways in survival and activation
of osteoclasts. J. Cell Biol. 148:333–342.
25. Lenhard, J.M., Weiel, J.E., Paulik, M.A., and Furf-
ine, E.S. 2000. Stimulation of vitamin A(1) acid
signaling by the HIV protease inhibitor indinavir.
Biochem. Pharmacol. 59:1063–1068.
26. Phenix, B.N., Lum, J.J., Nie, Z., Sanchez-Dardon,
J., and Badley, A.D. 2001. Antiapoptotic mecha-
nism of HIV protease inhibitors: preventing
mitochondrial transmembrane potential loss.
27. Tsiodras, S., Mantzoros, C., Hammer, S., and
Samore, M. 2000. Effects of protease inhibitors
on hyperglycemia, hyperlipidemia, and lipodys-
trophy: a 5-year cohort study. Arch. Intern. Med.
28. Fellay, J., et al. 2001. Prevalence of adverse events
associated with potent antiretroviral treatment:
Swiss HIV Cohort Study. Lancet. 358:1322–1327.
29. Liang, J.S., et al. 2001. HIV protease inhibitors
protect apolipoprotein B from degradation
by the proteasome: a potential mechanism for
protease inhibitor-induced hyperlipidemia. Nat.
30. Kelleher, A.D., Sewell, A.K., and Price, D.A. 2002.
Dyslipidemia due to retroviral protease inhibi-
tors. Nat. Med. 8:308.
31. Sturley, S.L., et al. 2002. Reply to “Dyslipidemia
due to retroviral protease inhibitors”. Nat. Med.
32. McHugh, K.P., et al. 2000. Mice lacking β3
integrins are osteosclerotic because of dysfunc-
tional osteoclasts. J. Clin. Invest. 105:433–440.
33. Wei, S., Wang, M.W.-H., Teitelbaum, S.L., and
Ross, F.P. 2001. RANK ligand activates nuclear
factor-κB in osteoclast precursors. Endocrinology.
34. Simonet, W.S., et al. 1997. Osteoprotegerin: a
novel secreted protein involved in the regulation
of bone density. Cell. 89:309–319.
35. Feng, X., et al. 2001. A Glanzmann’s mutation of
the β3 integrin gene specifically impairs osteoclast
function. J. Clin. Invest. 107:1137–1144.
36. Boyce, B.F., Aufdemorte, T.B., Garrett, I.R.,
Yates, A.J.P., and Mundy, G.R. 1989. Effects of
interleukin-1 on bone turnover in normal mice.
37. Smart, E., Ying, Y., Mineo, C., and Anderson, R.
1995. A detergent-free method for purifying
caveolae membrane from tissue culture cells.
Proc. Natl. Acad. Sci. U. S. A. 92:10104–10108.