The Src family kinase, Lyn, suppresses osteoclastogenesis in vitro and in vivo

Abstract

c-Src kinase is a rate-limiting activator of osteoclast (OC) function and Src inhibitors are therefore candidate antiosteoporosis drugs. By affecting αvβ3 and macrophage-colony stimulating factor (M-CSF)-induced signaling, c-Src is central to osteoclast activity, but not differentiation. We find Lyn, another member of Src family kinases (SFK) is, in contrast, a negative regulator of osteoclastic bone resorption. The absence of Lyn enhances receptor activator of NF-κB ligand (RANKL)-mediated differentiation of osteoclast precursors without affecting proliferation and survival, while its overexpression decreases osteoclast formation. In further contrast to c-Src, Lyn deficiency does not impact the activity of the mature cell. Reflecting increased osteoclast development in vitro, Lyn−/− mice undergo accelerated osteoclastogenesis and bone loss, in vivo, in response to RANKL. Mechanistically, Lyn forms a complex with receptor activator of NF-κB (RANK), the tyrosine phosphatase, SHP-1, and the adapter protein, Grb2-associated binder 2 (Gab2). Upon RANKL exposure, Gab2 phosphorylation, JNK, and NF-κB activation are enhanced in Lyn−/− osteoclasts, all critical events in osteoclast development. We therefore establish that Lyn regulates osteoclast formation and does it in a manner antithetical to that of c-Src. The most pragmatic aspect of our findings is that successful therapeutic inhibition of c-Src, in the context of the osteoclast, will require its stringent targeting.

Bone resorption by differentiated osteoclasts (OCs) depends upon the cell's capacity to polarize and organize its cytoskeleton, events initiated by activated integrins and cytokine receptors (1, 2). These cell surface-residing proteins, in turn, recruit a series of intracellular signaling molecules that prompt formation of an isolated extracellular microenvironment in which bone matrix is degraded. These signaling moieties, therefore, present as potential therapeutic targets.

Src family kinases (SFKs) are nonreceptor tyrosine kinases that are promiscuous in their impact on events such as growth, differentiation, cytoskeletal organization, and survival (3). In the OC, however, c-Src is the only SFK known to be functionally important, interacting with the intracellular domains of the αvβ3 integrin and the cytokine receptors, receptor activator of NF-κB (RANK) and c-Fms, all of which regulate the resorptive activity of the mature cell (4, 5). While the absence of c-Src does not reduce their number, OCs lacking the SFK fail to polarize and to resorb bone (6). Thus, c-Src−/− mice are severely osteopetrotic in the face of an abundance of dysfunctional OCs (6).

c-Src is specifically recruited to the terminal 3 aa of the β3 subunit of the αvβ3 integrin (7) and to Y559 of the macrophage-colony stimulating factor (M-CSF) receptor (8), c-Fms, to modulate cytoskeletal changes. It also binds to activated RANK (9), thereby recruiting TRAF6 and Grb2-associated binder 2 (Gab2) followed by phosphorylation of IκBα and JNK (10), which ultimately lead to respective activation of the transcription factors NF-κB and AP-1 (2, 11, 12).

While only c-Src has been implicated in OC function, the cell expresses other SFKs, including Lyn, which is a positive and negative regulator of signals emanating from immune receptors in lymphocytes, mast cells, and macrophages (13–15). In some circumstances, Lyn positively impacts B-cell receptor (BCR) and CD40 signaling (16, 17), yet Lyn^−/− B cells are overactive in response to BCR triggering (18, 19). In addition, Lyn^−/− mast cells hyperproliferate in response to growth factors (20, 21). Importantly, Lyn deficiency enhances granulocyte-macrophage-colony stimulating factor (GM-CSF) and M-CSF sensitivity of macrophages (15, 22).

Upon binding to immune receptor tyrosine-based inhibitory motif (ITIM)-containing receptors, Lyn phosphorylates, and thus downregulates, the protein tyrosine phosphatases SHP-1, SHP-2, and SHIP (15, 23, 24). SHP-1 and SHIP are highly expressed in OCs and both regulate OC formation, as shown by their respective null animals that develop osteoporosis characterized by numerous, activated resorbing cells (25–27). SHIP is linked to a signaling pathway activated by M-CSF (27), while the mechanism by which SHP-1 regulates osteoclastogenesis is unclear.

This study confirms Lyn as the second known SFK to impact the OC. In contrast to c-Src however, Lyn does not regulate the resorptive function of individual OCs but modulates differentiation of their precursors. Specifically, Lyn blunts osteoclastogenesis by suppressing RANKL-mediated Gab2 phosphorylation via activation of a SHP-1-dependent inhibitory signaling pathway. Our data establish the diversity of function of SFKs in OCs and underscores the necessity of specific targeting of individual members in designing antiresorptive agents.

Results

Lyn Retards OC Differentiation in Vitro.

c-Src is a marker of commitment to the OC phenotype as it is absent in naïve bone marrow macrophages (BMMs) but induced by receptor activator of NF-κB ligand (RANKL) and M-CSF. To determine whether the same is true regarding Lyn, WT BMMs were exposed to the 2 osteoclastogenic cytokines for up to 4 days. c-Src, as expected, appears with time while Lyn is equally expressed throughout osteoclastogenesis (Fig. 1A).

Fig. 1.

Lyn deficiency enhances osteoclastogenesis in vitro. (A) WT BMMs were cultured in M-CSF (10 ng/mL) and RANKL (100 ng/mL). Each day cells were lysed; equal amounts of protein were immunoblotted for Lyn or c-Src. Lyn−/− BMMs serve as negative control for antibody specificity, while β-actin serves as loading control. (B) WT and Lyn−/− BMMs were cultured in the presence of M-CSF and RANKL and stained for TRAP activity at days 4, 6, and 8. Stars mark dead cells. (C) Number of OCs obtained in coculture of WT or Lyn−/− BMMs with WT (*, P < 0.01) or Lyn−/− stromal cells (ST) (*, P < 0.05). (D) DNA fragmentation assay to determine presence of apoptotic cells in WT and Lyn−/− osteoclastic cultures (*, P < 0.01). (E) Real-time qPCR analysis of OC differentiation marker genes at days 0, 1, 3, and 5 in WT and Lyn−/− cells (*, P < 0.01 in TRAP and cSrc, P < 0.05 in CathK). (F) WT and Lyn−/− BMMs transduced with empty vector (pMX), flag-cSrc and flag-Lyn cultured in M-CSF (10 ng/mL) and RANKL (100 ng/mL) for 5 days to generate OCs. (G) Western blot analysis to detect expression levels of flag-Src and -Lyn and total Lyn in samples from F.

To determine whether Lyn participates in the osteoclastogenic process, we cultured WT and Lyn−/− BMMs in M-CSF and RANKL, staining the cells for tartrate-resistant acid phosphatase (TRAP) activity on days 4, 6, and 8. In fact, the loss of Lyn increases OC number as early as day 4 (WT OCs 75 ± 15, Lyn−/− OCs 180 ± 27, P < 0.005, and Fig. 1B). Similarly, cocultures of Lyn−/− BMMs with WT or Lyn−/− stromal cells form about 2-fold more OCs than WT BMMs under the same conditions (Fig. 1C). While we did not observe differences in apoptosis between WT and Lyn−/− cells during the first 4 days in culture with RANKL and M-CSF, by day 6 WT OCs undergo apoptosis yielding cell “ghosts” [Fig. 1 B (Upper panel, asterisks) and D]. In contrast, the absence of Lyn promotes the resorptive cell's viability [Fig. 1 B (Lower panel) and D]. The positive effect of Lyn deletion on osteoclastogenesis is confirmed by progressive upregulation of osteoclastogenic markers [Fig. 1E and supporting information (SI) Fig. S1A]. On the other hand, expression of c-Fos, NFAT2, and RANK by Lyn−/− and WT cells is indistinguishable (Fig. 1E and Fig. S1A). To further confirm the negative effect of Lyn on osteoclastogenesis, WT BMMs were transduced with Flag-c-Src, Flag-Lyn, or empty vector pMX as control and differentiated into OCs with RANKL and M-CSF. While overexpression of c-Src does not alter the capacity of WT cells to become OCs, Lyn overexpressing cells virtually lack TRAP, are smaller, and few contain more than 3 nuclei (Fig. 1 F and G). When c-Src and Lyn were transduced in Lyn−/− BMMs, only Lyn, not c-Src, reduces the ability of Lyn−/− cultures to form OCs [Fig. 1 F (Bottom panel) and G]. Hence, Lyn, in contrast to c-Src, is a negative regulator of osteoclastogenesis.

Lyn Does Not Regulate the Mature OC Activity.

c-Src modulates bone resorption by enabling the mature OC to function by organizing its cytoskeleton. To determine whether the same obtains in the context of Lyn, we generated WT and Lyn−/− OCs on dentin in the presence of M-CSF and RANKL. After 4 days the cells were stained with FITC-phalloidin to visualize actin rings or we removed them to evaluate resorption pits. These lacunae are more abundant in Lyn−/− cultures as is release of type 1 collagen fragments (Ctx-1) into medium. The number of actin rings is also increased in Lyn−/− cultures establishing that, unlike c-Src, the SFK does not participate in organizing the cell's cytoskeleton. (Fig. 2A−D).

Fig. 2.

Lyn does not modulate bone resorption of mature OCs. WT and Lyn−/− BMMs were cultured on dentin with M-CSF (10 ng/mL) and RANKL (100 ng/mL) for 4 days. (A) Cells were fixed and stained with FITC-phalloidin to visualize actin rings. (B) Statistical analysis of the number of WT and Lyn−/− actin rings/dentin slice on day 4. (*, P < 0.005). (C) Resorption lacuna formation was examined on day 4 by toluidine blue staining after removing OCs (arrows). (D) Resorptive activity was determined by collagen type 1 fragment (CrossLaps) ELISA of culture media on day 5. (*, P < 0.05). (E) Collagen type 1 fragment release from WT and Lyn−/− pre-OCs, seeded in equal number on dentin for 24 h.

These data are consistent with enhanced resorption in Lyn−/− cultures consequent to accelerated osteoclastogenesis. To determine whether Lyn deficiency impacts the activity of the mature resorptive cell, we plated the same number of OC precursors (cells that have been in culture with RANKL and M-CSF for 3 days) on dentin for 24 h. In this circumstance, in which an equal number of TRAP positive cells is present on each dentin slice (data not shown), there is no difference in collagen fragments mobilized by Lyn−/− and WT differentiated OCs (Fig. 2E). Taken with intact cytoskeletal organization, the absence of Lyn enhances OC differentiation but does not alter the resorptive capacity of mature OCs.

Lyn Negatively Regulates RANKL Signaling in OC Precursors.

To understand the mechanism by which Lyn retards OC differentiation, we analyzed M-CSF- and RANKL-induced signaling cascades. M-CSF promotes proliferation of OC precursors but BrdU incorporation into these cells is unaltered by the absence of Lyn (Fig. S1 B and C). Consistent with these findings, M-CSF-induced ERK and Akt phosphorylation are comparable in WT and Lyn−/− cells (Fig. 3A). The absence of Lyn, therefore, appears not to affect M-CSF signaling.

Fig. 3.

Lyn modulates RANK signaling. (A) WT and Lyn−/− pre-OCs, cytokine starved for 3 h and stimulated with M-CSF (50 ng/mL). Phosphorylation of Akt and ERK was determined by Western blotting. β-actin levels serve as loading control. (B) WT and Lyn−/− pre-OCs were treated with RANKL (100 ng/mL) for the indicated times. Total cell lysates were immunoblotted with phospho-p38, phospho-JNK, and phospho-IκBα antibodies. β-actin and JNK in total cell lysates serve as loading controls. (C) Phospho-c-Jun and NF-κB subunit levels in nuclear extracts of cells described in B were determined by immunoblot. Nucleophosmin serves as loading control and nuclear marker. Fold induction of normalized, phosphorylated proteins vs. time 0 are shown.

On the other hand, Lyn deficiency modifies RANKL signaling in pre-OCs. While RANKL-stimulated phosphorylation of p38 MAPK is not affected by the absence of the SFK, that of JNK and IκBα are (Fig. 3B). Establishing functional significance, these events are associated with increased nuclear translocation of activated c-Jun and the p65 NF-κB subunit in Lyn−/− pre-OCs following RANKL exposure (Fig. 3C). Hence, Lyn appears to negatively regulate OC differentiation by inhibiting RANKL signaling.

Lyn Is Recruited to RANK and Controls Gab2 Phosphorylation.

The fact that Lyn specifically modulates RANKL signaling raised the possibility that the SFK recognizes RANK. As shown in Fig. 4A, Lyn and RANK associate in pre-OCs in a RANKL-dependent manner.

Fig. 4.

RANKL-induced Gab2 phosphorylation is enhanced in Lyn−/− pre-OCs. (A and B) Pre-OCs from WT mice were stimulated with RANKL (100 ng/mL) for the indicated times. Lyn (A) or Gab2 (B) was immunoprecipitated and probed with the indicated antibodies. (C−E) WT and Lyn−/− pre-OCs were incubated in the presence of RANKL (C and D) or M-CSF (E) for the indicated times. (C) Lysates were immunoprecipitated with anti-phosphotyrosine Ab and immunoblotted with anti-Gab2 Ab. Gab2 content in total cell lysates serves as loading control. (D and E) Lysates were immunoblotted with anti-phospho-Gab2 antibody and reprobed with anti-Gab2 antibody. Fold induction of normalized, phosphorylated proteins vs. time 0 is shown.

Gab2 is an adapter protein required for OC formation (10). It is recruited to the RANK signaling complex in OCs wherein it modulates RANKL-induced activation of JNK and NF-κB (10). Taken with the observation that Gab2 is hyperphosphorylated in Lyn-deficient mast cells (28), these data suggest Gab2 may be involved in RANK signaling downstream of Lyn. Supporting this contention, reciprocal immunoprecipitation of Lyn and Gab2 from pre-OC lysates confirm that the 2 molecules associate in response to RANKL (Fig. 4 A and B). In keeping with the enhanced activation of JNK and NF-κB in Lyn−/− cells (Fig. 3), RANKL-induced Gab2 phosphorylation is enhanced in Lyn−/− OCs (Fig. 4 C and D). In contrast, Gab2 phosphorylation in response to M-CSF is not affected by the loss of Lyn (Fig. 4E). Lyn is therefore recruited to RANK in response to RANKL and negatively modulates RANKL-evoked Gab2 phosphorylation, a critical event in OC development.

Lyn Modulates Phosphorylation of SHP-1 in Response to RANKL.

Lyn-diminished Gab2 phosphorylation suggests protein phosphatase activity. Because Lyn activates SHIP, SHP-1, and SHP-2 in other cells, we asked whether the same obtains in OCs. We find phosphorylation of SHIP and SHP-2 indistinguishable in WT and Lyn−/− pre-OCs and unaltered by RANKL (Fig. S2). On the other hand, RANKL-stimulated phosphorylation of SHP-1, a phosphatase regulating OC differentiation and function (29), is impaired in Lyn null cells (Fig. 5A). Furthermore, while Lyn constitutively associates with SHP-1, the complex increases in the context of the osteoclastogenic cytokine (Fig. 5B).

Fig. 5.

SHP-1 phosphorylation and association with Gab2 is impaired in Lyn−/− pre-OCs. Cytokine starved WT or Lyn−/− pre-OCs were stimulated with RANKL for the indicated times. Lysates were immunoprecipitated with SHP-1(A and C), Lyn (B), and Gab2 (D) Abs. The immunoprecipitates were immunoblotted for phosphotyrosines, SHP-1, Lyn, Gab2, or SHP-2, as indicated. TCL, total cell lysates. Fold induction of normalized, phosphorylated proteins vs. time 0 are shown.

To determine whether SHP-1 modulates Gab2 phosphorylation, we asked whether the 2 molecules associate in response to RANKL. While such is the case in WT cells (Fig. 5 C and D), SHP-1/Gab2 recognition is impaired in Lyn−/− pre-OCs (Fig. 5C). Lyn therefore specifically activates the SHP-1 protein phosphatase in OCs, which forms a complex with Gab2 thereby mediating its dephosphorylation.

SHP-1 (me^v/me^v) Mutant Pre-OCs Display Enhanced RANKL Signaling.

To further explore the role of SHP-1 in RANKL signaling, we turned to me^v/me^v mice, which have diminished activity of the phosphatase (26). In fact, Gab2 phosphorylation in me^v/me^v cells is enhanced in response to the osteoclastogenic cytokine (Fig. 6A) as is activation of the Gab2 targets JNK and IκBα (Fig. 6B). Consistent with our findings in Lyn null cells (Fig. 3), phosphorylation of p38 MAPK was not altered by SHP-1 mutation (Fig. 6C). Thus, Lyn activates SHP-1 and promotes its association with, and dephosphorylation of, Gab2, which in turn suppresses RANKL-induced osteoclastogenic signals.

Fig. 6.

Gab2 phosphorylation is enhanced in SHP-1 mutant (me^v/me^v) pre-OCs. (A) Pre-OCs from SHP-1 mutant and WT were stimulated with RANKL. Lysates were immunoprecipitated with anti-phophotyrosine Ab and the immunoprecipitate was immunoblotted with anti-Gab2 Ab. Gab2, SHP-1, and β-actin in total cell lysates serve as loading controls. Cytokine-starved (B) BMMs or (C) pre-Ocs were treated with RANKL (100 ng/mL) for the indicated times. Lysates were immunoblotted with phospho-JNK, phospho-IκBα, and phospho-p38 Abs. β-actin and total JNK serve as loading controls. Fold induction of normalized, phosphorylated proteins vs. time 0 are shown.

Lyn Retards RANKL-Stimulated Bone Resorption in Vivo.

Finally, we addressed the biological relevance of Lyn-modulated osteoclastogenesis by determining whether our in vitro observations obtain in vivo. Bone histomorphometric analysis of long bones from same-sex, age-matched, unmanipulated WT and Lyn−/− mice reveals similar bone volume and numbers of OCs and osteoblasts (OBs) (Fig. S3 A−D). Mineral apposition and bone formation rates are also equivalent in the 2 strains (Fig. S3 E and F). Furthermore, WT and Lyn−/− animals express similar levels of RANKL and osteoprotegerin (OPG) in the bone marrow in basal conditions (Fig. S3G). Because elevated RANKL levels are commonly found in most if not all osteolytic diseases, including postmenopausal osteoporosis (30), we tested the clinical implications of Lyn suppression in vivo, by injecting RANKL daily, for 6 days, in WT and Lyn−/− mice. Confirming our in vitro observations, supracalvarial RANKL injections double the percentage of OC surface in the calvaria of mutant as compared to WT mice (Fig. 7 A and B). The increased osteoclastogenesis in RANKL-treated Lyn−/− mice is further confirmed by a 3.7-fold increase in serum tartrate-resistant acid phosphatase 5b (TRACP5b), a marker of osteoclastic recruitment, and a 2.5-fold increase in collagen type 1 fragments (CTX-1), a marker of global bone resorption, (Fig. 7 C and D). Thus, Lyn is a negative modulator of RANKL-stimulated osteoclastogenesis and bone resorption in vitro and in vivo.

Fig. 7.

Lyn deficiency enhances RANKL-stimulated bone loss in vivo. WT and Lyn−/− mice were injected, supracalvarially, with PBS or RANKL (100 μg) daily, for 6 days. (A) TRAP-stained histological sections of calvariae. (B) Percentage of OC surface per bone volume. (C) Serum TRACP 5b levels at time of killing. (D) Serum collagen type 1 fragment levels at time of killing. (*, P < 0.01 compared to WT.)

Discussion

c-Src is central to bone remodeling as its deletion yields a profound osteopetrotic phenotype because of defects in OC activity but not differentiation. The discovery that c-Src is critical for OC activity, therefore, positioned it as a potential antibone resorptive target (6, 31) and in fact, Src-inhibiting small molecules have been developed for this purpose (32).

The presumption that c-Src is the unique OC-regulating SFK suggests that stringency of its targeting, relative to other members of the kinase family, may not be essential. However, we establish that Lyn also regulates the OC and does it in a manner antithetical to that of c-Src. Thus, the most pragmatic aspect of our findings is that successful therapeutic inhibition of c-Src, in the context of the OC, will require its stringent targeting.

In contradistinction to the proresorptive properties of c-Src, Lyn is a negative modulator of OC differentiation as Lyn−/− BMMs have an enhanced propensity to develop into mature resorptive polykaryons while its overexpression dampens the osteoclastogenic capacity of WT cells. Confirming morphology, the osteoclastogenic markers TRAP, cathepsin K, and c-Src are increased in Lyn null cells treated with RANKL and M-CSF. This increase in osteoclastogenesis is also observed in cocultures of Lyn−/− BMMs with WT or Lyn−/− stromal cells. Importantly, RANK, RANKL, and OPG levels of expression are equivalent in BMMs and stromal cells from WT and Lyn−/− mice, further implicating a direct role for the SFK in OC formation. Lyn deletion also promotes longevity of the mature OC, another mechanism that controls OC number. Again in contrast to c-Src, Lyn does not impact the activity of the individual OC but negatively modulates resorption by decreasing cell number, in vitro and in vivo. Underscoring the importance of specific therapeutic targeting, our findings indicate that inhibition of Lyn could aggravate, rather than ameliorate, pathological bone loss.

Previous studies find that M-CSF-induced Akt phosphorylation is augmented in Lyn−/− macrophages, prompting us to postulate that hyperresponsibility to the cytokine promotes their enhanced propensity to OC differentiation (22). Lyn binds the M-CSF receptor, c-Fms, (data not shown), but M-CSF-mediated Akt and ERK phosphorylation are normal in Lyn-deficient cells committed to the OC phenotype. Furthermore, the absence of Lyn does not enhance pre-OC proliferation.

In contrast to its failure to impact the biological effects of M-CSF in osteoclastic cells, Lyn impairs RANKL signaling. Lyn deficiency enhances JNK and NF-κB activation, the latter manifested by increased phosphorylation of IκBα and p65 nuclear translocation. Reflecting this increased activation of major RANKL-mediated proosteoclastogenic signals, TRAP, c-Src, and cathepsin K expression is enhanced in Lyn−/− cultures. It is not clear why Lyn behaves differently in RANKL-, as compared to M-CSF-stimulated cells. It may relate, however, to the fact that Lyn is constitutively expressed in BMMs and OCs while c-Src appears only with exposure to RANKL. In that scenario, Lyn would be free to occupy the Src binding site, c-FmsY559, in BMMs, but would be in competition with c-Src in the differentiated resorptive cell. This hypothesis is consistent with the impaired M-CSF signaling extent in c-Src−/− OCs but not BMMs (3, 33).

The capacity of Lyn to negatively regulate RANK signaling may relate to the SFK's activation of receptors containing an ITIM. In B cells, for example, Lyn dampens signals emanating from CD22 (19), FcγRIIB (34), and PIR-B (35) receptors by phosphorylating their ITIM tyrosine residues, thereby recruiting protein tyrosine phosphatases SHIP, SHP-1, or SHP-2. Consequently, B-cell receptor-mediated signaling is exuberant in the absence of Lyn (36). Lyn also regulates nonimmunoreceptor signaling as mice lacking the SFK robustly expand their myeloid cell compartment in response to growth factors such as GM-CSF and G-CSF (15). This event correlates with reduced phosphorylation of the ITIM receptors, PIR-B and SIRP-1α, in resting and cytokine-stimulated macrophages from Lyn-null mice (15). While the role of immunoreceptor tyrosine-based activation motif (ITAM) receptors in OC differentiation and function is established (37), there is presently no evidence of participation by those bearing ITIM motifs. On the other hand, macrophages express PIR-B and SIRP-α raising the possibility that ITIM proteins are active in OCs (38, 39).

Lyn activates protein tyrosine phosphatases in B and mast cells (15, 23, 24). Mice lacking SHP-1 or SHIP have increased numbers of OCs and low bone mass (25, 27, 40). While the mechanisms underlying the effect of SHIP on OC formation and survival downstream of M-CSF are in hand (27), it is still unclear how SHP-1 exerts its resorptive effects. Our data indicate that Lyn promotes phosphorylation of SHP-1 but not SHP-2 or SHIP in OCs.

me^v/me^v mice bear a natural inactivating SHP-1 mutation (26). They also contain abundant OCs because of increased recruitment of TRAF6 to RANK in response to RANKL (41), indicating that SHP-1 negatively regulates RANK signaling. We find SHP-1 phosphorylation is induced by Lyn downstream of RANKL. SHP-1 and Lyn also form an inhibitory complex with the adaptor protein, Gab2, which activates NF-κB and JNK in the context of osteoclastogenesis. The reduction of Gab2 phosphorylation, by SHP-1 and Lyn, provides a novel link between the phosphatase and Gab2's osteoclastogenic properties.

Our data support a new, Lyn-based negative feedback mechanism for regulating osteoclastic bone resorption by affecting the RANK/RANKL axis. Enhanced RANKL activity is a common feature of many osteoporotic diseases including that following menopause or attending inflammatory arthritis. Thus, RANKL inhibition is among the most promising antiresorptive therapeutic strategies (30). Lyn-null mice stimulated, in vivo, with RANKL exhibit marked OC recruitment and associated bone resorption underscoring the importance of Lyn as negative regulator of RANKL signaling. Importantly, basal levels of Lyn, in normal OCs, do not prevent their differentiation nor block RANKL activation; however, Lyn, but not c-Src overexpression, dampens the osteoclastogenic capacity of WT cells. It is likely, therefore, that in basal conditions Lyn extinguishes RANK/RANKL signaling after a desired osteoclastogenic response is achieved. In this regard, the absence of Lyn does not impact osteoclastogenesis in naïve mice, but only in those that are RANKL stimulated. In fact, detailed histomorphometric analysis shows that the absence of Lyn does not modulate bone morphology or RANKL and OPG expression in unmanipulated mice. A similar scenario holds in the context of other proteins such as FHL2, which dampens OC formation in response to stress situations (42). Similarly, the absence of NF-kB molecules including NIK, p65, or RelB modulate in vivo osteoclastogenesis exclusively in the presence of elevated RANKL or TNF (43–45). Our data, therefore, establish Lyn as the second functionally significant SFK in the OC, exerting its effects in a manner antithetically different from c-Src.

Materials and Methods

Mice.

Lyn-deficient mice have been described (15). me^v/me^v mice were obtained from Jackson Laboratories. All mice were 6–10 weeks old, on a C57BL/6 background, and maintained at the Animal Facility of Washington University School of Medicine. Experiments were approved by the Animal Ethics Committee of Washington University.

OC Cultures. BMMs were prepared from whole bone marrow of 6- to 10-week-old mice as described (46) and cultured in the presence of GST-RANKL (100 ng/mL) and M-CSF (10 ng/mL) or cocultured with WT or Lyn−/− stromal cells. OCs were visualized by TRAP staining. In parallel experiments, BMMs were transduced with empty vector pMX, Flag-c-Src, or Flag-Lyn, selected with blasticidin, and then allowed to form OCs in the presence of RANKL and M-CSF for 5 days. To determine actin ring formation, OCs cultured on dentin with RANKL and M-CSF for 5 days, were fixed and stained with FITC-phalloidin. For bone resorption, OCs cultured on dentin were removed by brief treatment with 0.2 N NaOH and pits stained with toluidine blue. Alternatively, BMMs from WT and Lyn−/− were cultured on plastic for 4 days with M-CSF and RANKL, then lifted and replated in equal numbers on dentin in the presence of osteoclastogenic cytokines for 24 h. Bone resorption was analyzed by measuring the release of collagen type 1 in the media.

Real-Time qPCR.

Total RNA (1 μg) extracted from cultured cells during osteoclastogenesis was used as a template for cDNA synthesis. For OPG and RANKL detection, bone marrow cells were flushed from long bones and remaining bone adherent stromal cells were lysed by injecting lysis buffer directly in the bone marrow cavity. Primers were synthesized on the basis of the reported mouse cDNA sequence. The following primers were used: OPG, 5′-AGCTGCTGAAGCTGTGGAA-3′ and 5′-GGTTCGAGTGGCCGAGAT-3′; RANKL, 5′-CAC CAT CAG CTG AAG ATA GT-3′ and 5′-CCA AGA TCT CTA ACA TGA CG-3′; cSrc, 5′-GAACCCGAGAGGGACCTTC-3′ and 5′-GAGGCAGTAGGCACCTTTTGT-3′; c-Fos, 5′-CAAGCGGAGACAGATCAACTTG-3′ and 5′-TTTCCTTCTCTTTCAGCAGATTGG-3′; RANK, 5′-CTGCTCCTCTTCATCTCTGTG -3′ and 5′-CTTCTGGAACCATCTTCTCCTC-3′; and as indicated in ref. 47. Real-time qPCR was performed as previously described (47).

Western Blotting and Antibodies.

BMMs cultured with MCSF alone or with RANKL for 3 days (pre-OCs) were stimulated with RANKL (100 ng/mL) or M-CSF (50 ng/mL) before lysis as previously described (47). Monoclonal antibodies to p-JNK, p-Akt, p-IκBα, and p-ERK or polyclonal antibodies to JNK, Akt, ERK, p-P38, p-Gab2, p-SHIP, Lyn, and nucleophosmin were from Cell Signaling. Monoclonal antibodies to SHIP and p65 and polyclonal antibodies to Lyn, SHP-1, SHP-2, and p-c-Jun were from Santa Cruz Biotechnology. Monoclonal antibodies to Lyn and RANK were from Abcam. The monoclonal antibody to phosphotyrosines (4G10) was from Upstate. The polyclonal antibody to Gab2 was from Upstate.

Nuclear Cell Extraction.

RANKL-stimulated pre-OCs were washed with water and lysed with hypotonic buffer (10 mM Hepes, 1.5 mM MgCl2, 1 mM KCl, 1 mM DTT, and protease and phosphatase inhibitors), followed by the addition of 0.1% Nonidet P-40. Cells were centrifuged and washed with hypotonic buffer. Nuclear pellets were resuspended in high-salt buffer (hypotonic buffer plus 400 mM NaCl). Protein concentrations were determined using a modified Coomassie method (Pierce).

Serum TRAP Activity and Collagen Type 1 Fragment Assay.

Serum was collected in mice injected with RANKL (100 μg/day) or PBS on day 6. The levels of serum TRACP 5b and collagen type 1 fragment (Ctx-1) activity were measured by ELISA (Immuno Diagnostic Systems and Nordic Bioscience Diagnostics).

Histological Analysis.

Decalcified histological sections of calvariae were prepared and OC number, bone volume/total volume (BV/TV), and OB numbers were determined using Osteomeasure (Osteometrics).

Statistics.

All data are presented as mean ± SD. Statistical significance was determined by 2-tailed Student's t test.