Leupaxin is a cytoskeleton adaptor protein that was first identified in human macrophages and was found to share homology with the focal adhesion protein, paxillin. Leupaxin possesses several protein-binding domains that have been implicated in targeting proteins such as focal adhesion kinase (pp125FAK) to focal adhesions. Leupaxin can be detected in monocytes and osteoclasts, both cells of hematopoietic origin. We have identified leupaxin to be a component of the osteoclast podosomal signaling complex. We have found that leupaxin in murine osteoclasts is associated with both PYK2 and pp125FAK in the osteoclast. Treatment of osteoclasts with TNF-α and soluble osteopontin were found to stimulate tyrosine phosphorylation of both leupaxin and leupaxin-associated PYK2. Leupaxin was found to co-immunoprecipitate with the protein tyrosine phosphatase PTP-PEST. The cellular distribution of leupaxin, PYK2, and protein tyrosine phosphorylation-PEST co-localized at or near the osteoclast podosomal complex. Leupaxin was also found to associate with the ARF-GTPase-activating protein, paxillin kinase linker p95PKL, thereby providing a link to regulators of cytoskeletal dynamics in the osteoclast. Overexpression of leupaxin by transduction into osteoclasts evoked numerous cytoplasmic projections at the leading edge of the cell, resembling a motile phenotype. Finally, in vitro inhibition of leupaxin expression in the osteoclast led to a decrease in resorptive capacity. Our data suggest that leupaxin may be a critical nucleating component of the osteoclast podosomal signaling complex.
OSTEOCLASTS ATTACH TO the mineralized bone matrix through integrin-dependent adhesion structures to form polarized membrane domains before initiation of bone resorption.(1) The osteoclast forms a tight sealing zone that effectively isolates the resorption apparatus of the cell, the ruffled membrane facing the resorption cavity.(2) In the osteoclast, adhesion structures called podosomes are the putative precursors of the sealing zone.(1,3) Compared with focal adhesions, podosomes are very dynamic structures because they move and change in size and appear and disappear continuously.(4–6)
Podosomes in osteoclasts form signaling complexes, similar to focal adhesions in other cell types, and form a structural link between the extracellular matrix and the cytoskeleton.(7,8) There are several protein-tyrosine kinases and phosphorylated proteins within the podosomes, associated with activation of the integrin-signaling cascade.(9) Osteopontin (OPN), an Arg-Gly-Asp (RGD)-containing autocrine factor for osteoclasts, has been shown to stimulate the αvβ3-associated signal-generating complex.(10,11) Integrin activation results in an increased tyrosine phosphorylation of PYK2 and induces its association with other signal transduction proteins such as c-src, PI 3-kinase, paxillin, and p130Cas.(9)
To date, the identification and characterization of the wide spectrum of signaling molecules contained within the podosomal complex is incomplete. We have now identified a new component of the osteoclast podosomal signaling complex. Leupaxin (LPXN) was originally identified as a cytoskeletal protein preferentially expressed in hematopoietic cells and is most homologous to the focal adhesion protein paxillin.(12) Although LPXN is a member of the paxillin superfamily, its tissue distribution is not ubiquitous like that of paxillin; in addition, LPXN and paxillin have distinct differences in their protein-binding motifs.(13,14) LPXN, a phosphotyrosine protein, was suggested to be a substrate for PYK2, by virtue of its association with the protein tyrosine kinase in lymphoid cells.(12)
In the current study, we have determined the cellular distribution of LPXN and its association with several other signaling proteins in the podosomal signaling complex of the osteoclast. In addition, we have determined a functional role for LPXN in osteoclast-mediated resorption. Our preliminary evidence suggests that LPXN is a “linker” protein present in osteoclast podosomes and associates with various other signaling proteins such as protein tyrosine kinases, a phosphatase, and structural proteins that are regulators of actin cytoskeletal dynamics. The characterization of LPXN as a novel component of the podosomal signaling complex adds further complexity to regulation of osteoclast function.
MATERIALS AND METHODS
The monoclonal antibodies (mAb) to leupaxin were provided by Dr Brian Lipsky (ICOS Corp., Bothell, WA, USA). The polyclonal antibody to protein tyrosine phosphorylation PTP-PEST (proline-glutamic acid-serine-threonine amino acid sequences) was provided by Dr Michel L Tremblay (Department of Biochemistry, McGill University, Montreal, Quebec, Canada). The polyclonal antibody (pAb) to PYK2 was obtained from Santa Cruz Biotechnology (Santa Cruz, CA, USA), and the anti-phosphotyrosine antibody PY20 was obtained from Upstate Biotechnology (Lake Placid, NY, USA). The antibody to paxillin kinase linker p95PKL was provided by Dr CE Turner (State University of New York, Syracuse, NY, USA). The polyclonal anti-PYK2 [pY402]-phosphospecific antibody was purchased from BIOSOURCE International (Camarillo, CA, USA). The Renaissance chemiluminescence kit was purchased from NEN Life Science Products (Boston, MA, USA). The pTAT-transduction vector was a gift of Dr SF Dowdy (Howard Hughes Institute, Washington University, St. Louis, MO, USA). The recombinant murine TNF-α (specific activity 1.1 × 108 units/mg) was purchased from R & D Systems (Minneapolis, MN, USA). All other chemicals and reagents were purchased from Sigma (St. Louis, MO, USA).
Cloning and sequencing of LPXN in the osteoclast
LPXN was cloned from a previously characterized rabbit osteoclast cDNA library,(15–17) using PCR-based amplification based on sequence information of the human LPXN cDNA.(12) The sequence for LPXN cDNA cloned from the rabbit osteoclast was determined and verified by automated sequencing of both strands of the cDNA clone using the Big Dye terminator kit (Perkin Elmer Life Sciences, Boston, MA, USA). The insert sequence was compared with database sequences using the NCBI BLAST program.
In vitro transcription/translation of cDNA
The cDNA for LPXN in pBluescript (SK-) was transcribed with T3 RNA polymerase and translated using the TNT cell-free reticulocyte lysate translation system (Promega, Madison, WI, USA), following the manufacturer's instructions. The reaction used [35S]methionine (1000 Ci/mmol) to label the translated product. The negative controls used were either the absence of any added template or the empty pBluescript (Sk-) vector. The translated products were separated on an 8% SDS-PAGE and visualized by autoradiography.
The murine macrophage cell line RAW 264.7 was obtained from American Type Culture Collection (Manassas, VA, USA) and was cultured as previously described.(18) Murine osteoclasts were generated as previously described.(18–21) Briefly, the tibias and femurs of 7-week-old mice were used to isolate bone cells. Bone marrow cells were suspended in α-MEM supplemented with 10% fetal bovine serum (α-10 MEM) and cultured at 37°C in a 5% CO2 incubator. After 24 h, the nonadhered cells were layered on Histopaque-1077 (Sigma) and centrifuged at 300g for 15 minutes at room temperature. The cell layer between the Histopaque and the medium was removed and washed with α-10 MEM and centrifuged at 2000 rpm for 7 minutes. M-CSF (R&D Systems) and recombinant osteoprotegerin ligand (OPGL) were added at 10 and 100 ng/ml, respectively. Multinucleated osteoclasts were seen to form and mature after day 4. Their viability was routinely assessed by trypan blue dye exclusion, and the percentage of TRACP+ cells was assessed to be ∼99%. All experiments with murine osteoclasts were performed after day 5 in culture. Human osteoclasts were derived from human blood monocytes and were prepared according to methods adapted from those previously described.(22,23) Briefly, human peripheral blood was obtained from healthy adult volunteers. Peripheral blood mononuclear cells were isolated by density centrifugation using Histopaque-1077; after collecting the cells from the interface, they were washed several times with Dulbecco's PBS supplemented with 2% FBS. The cells were counted and added at a density of 1 × 106 cells/well of a 24-well plate under serum-free conditions, because we have observed that human monocytes stick better under serum-free conditions. The cells were allowed to adhere for 2 h; thereafter, the nonadherent cells were washed off and resuspended in MEM supplemented with 15% heat inactivated FBS, 2 mM L-glutamine, 100 IU/ml penicillin, 100 μg/ml streptomycin, 25 ng/ml M-CSF, and 100 ng/ml OPGL. Four days later, the media was replaced. The cells were cultured at 37°C in 5% CO2/95% air and fed twice weekly. In approximately 5-6 days, osteoclasts were seen to form. These human osteoclast cultures were maintained for 14 days. Finally, a human osteoblast cell line (ATCC CRL 11372, h FOB 1.19) was cultured as a negative control for the reverse transcriptase-polymerase chain reaction (RT-PCR)-based detection of the transcript for LPXN.
Detection of LPXN mRNA in osteoclasts
For Northern blots, a random-labeled probe containing the entire LPXN coding region from the rabbit osteoclast was used to hybridize to ∼20 μg of total RNA obtained from murine osteoclasts. Total RNA isolated from murine osteoclasts and human osteoblasts were used for RT-PCR to detect the transcript for LPXN. The primers used for detection of LPXN by RT-PCR were as follows. Forward (F): 5′-GATGCCTTATTGGAGGAATTG-3′, corresponding to amino acids 6-12 (DALLEEL) of human and mouse LPXN; reverse (R): 5′-ATCCAACTGAGCAGCTGCTGA-3′, corresponding to amino acids 89-95 (SAAAQLD) of human and mouse LPXN. The conditions for the PCR reaction were as follows: hot start at 94°C for 4 minutes, 30 cycles of 94°C for 1 minutes, 50°C for 1 minute, 72°C for 2 minutes, and a final extension at 72°C for 10 minutes. The expected amplicon was 270 bp; the negative control used was the absence of RT.
Immunoprecipitation and Western blotting
All proteins of interest were immunoprecipitated from either murine osteoclasts or the mouse macrophage cell line RAW 264.7 (used as a positive control) that were differentiated to form osteoclast-like cells in the presence of M-CSF and OPGL. The cells were lysed in a Triton-containing lysis buffer (in mM): 50 Tris-HCl, pH 7.4, 75 NaCl, 50 NaF, 40 PPi, 1 Na3VnO4, 1 EDTA, 1% Triton X-100, 0.5% Na-DOC, and 0.1% SDS. For the co-immunoprecipitation of LPXN with p95PKL, the lysis buffer used was 10 mM Tris-HCl, pH 7.6, 50 mM NaCl, 1% NP-40, and 10% glycerol supplemented with protease inhibitors (Complete; Roche Molecular Biochemicals), as described previously.(24) In any given experiment, equal amounts of protein lysates were used for each immunoprecipitation (∼500-600 μg). Immunoprecipitations were performed as described previously.(18,25–27) For Western blotting, ∼50 μg of protein was loaded/lane for SDS-PAGE.
Preparation of GST-LPXN fusion protein
The cDNA for LPXN was directionally cloned into pGEX-5X-1 (Amersham Pharmacia) at the EcoR1 and XhoI sites. Positive clones were identified and confirmed by sequencing with pGEX-5′ end primer. Subsequently, these clones were transformed in BL-21 cells. Overnight cultures of the recombinant glutathione-S-transferase (GST)-LPXN in BL-21 cells were transferred into larger volumes of LB/Amp, grown for 2 h, until the OD600 was determined to be 1.0, and subsequently induced with isopropyl-B-D-thiogalactopyranoside (IPTG; 0.1 mM final concentration) for an additional 2 h at 37°C. Thereafter, the cultures were spun down at 2500 rpm for 20 minutes. The pellets were resuspended in 3.6 ml binding buffer (50 mM Tris-Cl, pH 8, 120 mM NaCl, 0.5% NP-40, 1 mM DTT, 1 mM PMSF, plus a protease inhibitor cocktail). Then, 0.4 ml of lysozyme (10 μg/ml) was added, and the mixture was allowed to stand at room temperature for 20 minutes. The bacteria were lysed by repeated freeze-thaw cycles. The lysates were centrifuged at 14,000 rpm for 20 minutes. The supernatants were collected (∼4 ml) and incubated with 200 μl of washed GST-Sepharose; these were rocked at room temperature for 1 h. The Sepharose beads were washed five times for 10 minutes each using 10 volumes of binding buffer. An aliquot of ∼10 μl of GST bound proteins was taken and separated on 8% SDS-PAGE. After confirming the extraction, a 50% suspension of this extract was prepared with binding buffer and stored at 4°C. GST-beads bound to LPXN or GST (as a control) were incubated with murine osteoclast lysates at 4°C for 2-4 h. The beads were pelleted at 500g for 2 minutes and washed with binding buffer four times. Finally, 1-2× volumes of SDS-PAGE sample buffer were added to the washed beads, boiled, and separated on SDS-PAGE; these GST-pull-downs were then transferred to PVDF membranes for Western analyses with the appropriate antibodies.
Transduction of LPXN into osteoclasts
The full-length cDNA for LPXN was cloned in-frame into a bacterial expression vector, pTAT-HA, to produce a pTAT fusion protein (pTAT-LPXN). The vector pTAT-HA has an N-terminal 6-histidine leader, followed by the 11-amino acid pTAT protein transduction domain flanked by glycine residues, a hemagglutinin (HA) tag, and a polylinker.(21,28–31) The pTAT-Herpes simplex virus thymidine kinase protein (pTAT-HSVTK) was used as a control (42 kDa). The purification protocol was adapted from the published procedure using a Ni-NTA column.(21,28–31) After cells were kept in serum-free α-MEM for 2 h, the pTAT proteins were added to cells to a final concentration of 100 nM in serum-free media. An optimal dose of 100 nM was used; maximal uptake and response has previously been seen to occur at concentrations between 100-150 nM.(21) Osteoclasts were assayed for temporal overexpression of the transduced pTAT-LPXN after 60 minutes of incubation; the transduced pTAT fusion proteins (either LPXN or HSVTK) were detected with a monoclonal antibody to the HA tag. The cells were viewed on a Nikon Eclipse 800 microscope, fitted with a SPOT camera (Diagnostic Instruments, Inc., Alexandria, VA, USA), and photomicrographs were obtained.
Briefly, human monocyte-derived osteoclasts or murine osteoclasts cultured on glass coverslips were fixed and permeabilized, as previously described.(25,26,32) The cellular distribution of LPXN in human and murine osteoclasts was probed with the mAbs 283G and 283P, respectively. Actin was visualized in human osteoclasts with Alexa 488-conjugated phalloidin (Molecular Probes, Eugene, OR, USA). Actin in murine osteoclasts was visualized using rhodamine phalloidin. The primary LPXN mAbs were detected with either Alexa-562-conjugated goat anti-mouse (for human osteoclasts) or Cy-2-conjugated goat anti-mouse (for murine osteoclasts) secondary antibodies (Jackson ImmunoResearch Laboratories, West Grove, PA, USA); the cells were visualized using either confocal laser microscopy (Bio-Rad MRC-1024; Hercules, CA, USA), or an Eclipse 800 Nikon epifluorescence microscope attached to a SPOT camera (Diagnostic Instruments Inc., Alexandria, VA, USA), as indicated in the individual figure legends.
Mature murine osteoclasts were plated on 1-μm-thick CaPO4-coated quartz discs (Osteologic discs; BD Biosciences, Lexington, KY, USA), as previously described.(26) These osteoclasts were treated with antisense (AAGGCATCCAATTCCTCCAT, which spans the ATG start site) and sense (ATGGAGGAATTGGATGCCTT) oligodinucleotides (ODNs) with phosphorothioate backbones, separately. The ODNs were added directly to the culture medium at a final concentration of 10 μM. After 24 h, the ODNs were added again; 48 h later, the effects on osteoclast-mediated resorption were assessed by rinsing the discs in distilled water to lyse the osteoclasts, as previously described.(26) The extent of resorption could be assessed by clear areas on the quartz discs where the CaPO4 film had disappeared.(26)
Cloning of LPXN from the osteoclast: nucleotide and amino acid sequence analysis
The cDNA for LPXN was cloned by PCR from a rabbit osteoclast (OC) cDNA library,(15,33) based on the corresponding sequence for LPXN originally cloned from the human macrophage.(12) The cDNA for LPXN in the osteoclast was isolated as a 1.9-kb clone, and analysis of the nucleotide sequence of LPXN revealed that the 1773-bp sequence contains a 5′ untranslated region of 72 bases, followed by 1161 bases encoding 386 amino acids and 540 bases of 3′ untranslated region. We have deposited this sequence in Genbank with the accession number AF 118146. The LPXN gene has been mapped to chromosome 11 (sts-W46419) by the International Radiation Hybrid Mapping Consortium. Henceforth, we will refer to LPXN cloned from the (rabbit) osteoclast as LPXN-OC.
Comparison of the amino acid sequences (aa) for LPXN originally cloned from the human macrophage (Accession no. NP_004802), the rabbit osteoclast (Accession no. AF 118146), and that cloned from the mouse spleen (Accession no. BAB40667) revealed ∼85% overall sequence identity between these different species (data not shown). LPXN-OC contains 10 tyrosine residues; however, an analysis of the aa sequence using PROSITE “pattern-search” program revealed that only tyrosine residues 9 and 10 match the tyrosine phosphorylation consensus sequence, the pattern R/K-X2 or 3-D/E-X2 or 3-Y taken as the proper motif.(34) Further analysis of the aa sequence of LPXN-OC revealed various protein kinase phosphorylation motifs, amidation and N-myristoylation sites. There is one N-glycosylation site, two cyclic adenosine monophosphate (cAMP)- and cGMP-dependent protein kinase phosphorylation sites, and seven PKC phosphorylation sites. In addition, LPXN-OC has five casein kinase II phosphorylation sites and two tyrosine phosphorylation sites. Finally, there is also one N-myristoylation site and one amidation site.(34)
LPXN, by virtue of its protein-protein interacting LIM (for Lin-11, Isl-1, Mec-3) and LD (leucine, aspartate) domains, is presumed to function as an “adaptor” protein, as previously reported.(12,35) LPXN-OC contains four LIM and four LD motifs. The discrete composition of the LD motifs in members of the paxillin superfamily suggest a specificity of function. The greatest divergence between the LD motifs of paxillin and LPXN is in the LD2 motif.(12) This divergence may explain the fact that LPXN does not necessarily bind the entire set of signaling molecules common to other members of the paxillin superfamily, such as paxillin and Hic-5.
Detection of LPXN transcript in the osteoclast
Figure 1 shows data on detection of the transcript for LPXN-OC using RT-PCR and Northern blot analysis and an in vitro translation of the cloned LPXN-OC cDNA. Primers for PCR amplification of LPXN were selected from regions of LPXN conserved between the mouse spleen and the human macrophage, and a 270-bp amplicon was obtained solely from the OC (Fig. 1A, lane 2); the LPXN transcript was absent in osteoblasts (OB, Fig. 1A, lane 3). There was no product obtained in the absence of RT (Fig. 1A, lane 4). A positive control for the PCR reaction shows amplification of the type II Na-dependent cotransporter (Npt2) from the osteoclast (Fig. 1A, lane 5) as previously reported.(26) Northern blot analysis (Fig. 1B) showed that the size of the LPXN-OC transcript is 2.4 kb, similar to that of LPXN from the human macrophage.(12) The cDNA for LPXN-OC in the plasmid pBluescript (SK-) was transcribed in vitro with T3 polymerase, translated, and labeled with35S-Methionine; as shown in Fig. 1C (lane 1), a single band at ∼50 kDa was obtained as a product. Also shown are the negative controls for the in vitro transcription/translation reactions performed in the absence of DNA (Fig. 1C, lane 2) or with the plasmid pBluescript alone (Fig. 1C, lane 3), respectively.
Detection of LPXN protein in osteoclasts
We have characterized several monoclonal antibodies (mAbs) that recognize LPXN in osteoclasts.(12) These data are represented in Fig. 2. The mAbs are as follows. (1) mAb 283P (Fig. 2A): the mAb 283P recognizes a single band for LPXN (∼51 kDa) both in the mouse macrophage RAW 264.7 cell line (Fig. 2A, lane 1) and in the murine osteoclast (Fig. 2A, lane 2). We have used this mAb 283P for all immunofluorescence studies in murine osteoclasts. (2) mAb 315G (Fig. 2B): first, when the cDNA for LPXN was translated in vitro and blotted with mAb 315G, a single band was seen at ∼50 kDa, as shown in Fig. 2B (lane 1). The mAb 315G also recognizes LPXN in murine osteoclast lysates (Fig. 2B, lane 2). The mAb 315G could also be used to immunoprecipitate and immunoblot LPXN from murine osteoclasts (Fig. 2B, lane 3); the control immunoprecipitation (with mouse IgG1 MOPC-21) is shown (Fig. 2B, lane 4). (3) mAbs 283C and 283G (Fig. 2C): the mAbs 283C (Fig. 2C, a) and 283G (Fig. 2C, b) recognized LPXN in cell lysates prepared from both human monocytes and macrophages. Neither of these mAbs detected LPXN in murine macrophages (B Lipsky and A Gupta, unpublished observations, 1999) and were considered specific for human LPXN. The mAb 283C was also effective in immunoprecipitation of LPXN from human monocytes (Fig. 2D, lane 2); the control immunoprecipitation was performed with the mouse IgG1 MOPC21 (Fig. 2D, lane 1).
In data presented below, the mAbs 283C and 283G were used for immunofluorescence studies in human osteoclasts. For studies involving immunoprecipitation and Western blotting of LPXN from murine osteoclasts, we have used the mAbs 315G and 283P, respectively; for immunofluorescence studies in murine osteoclasts, we have used the mAb 283P.
Cellular distribution of LPXN in human and murine osteoclasts
LPXN was previously reported to be diffusely distributed in the F-actin-rich areas of lymphoid cells and enhanced in the proximal regions of filopodial-like projections.(12) We have shown that both the mAbs 283C and 283G detected a single band in both human monocytes and macrophages (Fig. 2C, a-b). First, we examined the cellular distribution of LPXN in human osteoclasts using confocal microscopy. Osteoclasts were derived from human peripheral blood monocytes by the addition of mCSF-1 and OPGL.(36,37) These human osteoclasts were plated on glass and immunostained for both actin (green) and LPXN (red). Both LPXN and actin were found to be peripherally distributed, as shown in Fig. 3A (a-c). The distribution of LPXN (Fig. 3A, b) was found to decorate the contiguous areas around the podosomes, as delineated by the pattern for actin (Fig. 3A, a) in the adhesion zone; LPXN colocalized with actin (Fig. 3A, b) in a punctate pattern at the adhesion zone, as indicated by yellow (Fig. 3A, c). A similar peripheral distribution (as pointed to by the arrow) of LPXN alone (Fig. 3B, a) and LPXN and actin (Fig. 3B, b) was seen in murine osteoclasts. In murine osteoclast precursors, the peripheral pool of LPXN also colocalized with actin, as shown in yellow and pointed to by the arrows (Fig. 3C); an additional cytosolic pool of LPXN (green) was clearly evident in these murine osteoclast precursors.
Association of LPXN with PYK2
It has been previously reported that LPXN associates with PYK2 in cells of hematopoietic origin.(12) We have hypothesized that LPXN associates with PYK2 at specific subcellular sites such as the podosomes. Murine osteoclasts were used for immunoprecipitation of PYK2; the PYK2 immunoprecipitates were immunoblotted for LPXN, as shown in Fig. 4A. The Western blot signal for LPXN that co-immunoprecipitates with PYK2 can be detected at ∼52 kDa (Fig. 4A, lane 2); nonimmune serum (NIS) was used for the control immunoprecipitation (Fig. 4A, lane 1). Next, to determine what fraction of the total pool of PYK2 associates with LPXN, PYK2 was immunoprecipitated from murine osteoclast lysates and immunoblotted for PYK2 (Fig. 4B, lane 1). The PYK2 signal can be detected at ∼112 kDa, and it represents the total pool of immunoprecipitable PYK2 in the cell. LPXN was then immunoprecipitated from murine osteoclast lysates; the LPXN immunoprecipitate was immunoblotted for PYK2 (Fig. 4B, lane 2). As can be seen, only a small fraction of PYK2 can be detected in association with LPXN.
Given our co-immunoprecipitation data on PYK2 and LPXN, we examined whether the subcellular distribution of PYK2 colocalized with that of LPXN in the osteoclast. For this purpose, murine osteoclasts that were plated either on dentine slices (Fig. 4C) or on glass coverslips (Fig. 4D) were immunostained for both LPXN and PYK2. The cellular distribution of LPXN (green) and PYK2 (red) were strongly colocalized at or near the sealing zone in murine osteoclasts plated on dentine (Fig. 4C), whereas the two proteins colocalized weakly in murine osteoclasts that were plated on glass (Fig. 4D), as indicated in yellow. As is quite apparent, the subcellular colocalization of LPXN and PYK2 is small in osteoclasts that were plated on glass, in accordance with our previous co-immunoprecipitation data.
Tyrosine phosphorylation of LPXN-associated PYK2
LPXN is a phosphotyrosine protein.(12) Although LPXN-OC contains 10 tyrosine residues, only tyrosine residues 9 and 10 match the tyrosine phosphorylation consensus sequence. We have hypothesized that integrin-mediated stimulation of osteoclasts may promote the association of LPXN with protein tyrosine kinases and subsequent targeting to the podosomes. We examined the possibility whether osteoclast-activating cytokines could stimulate tyrosine phosphorylation of LPXN and LPXN-associated PYK2. TNF-α is a potent osteoclastogenic cytokine.(38,39) TNF-α has also been shown to increase the tyrosine phosphorylation of PYK2 and paxillin in neutrophils and fibroblasts.(40–42)Osteopontin (OPN) has been shown previously to stimulate integrin-mediated tyrosine phosphorylation of PYK2 in osteoclasts.(43) First, we examined tyrosine phosphorylation of PYK2 in response to both OPN and TNF-α. Murine osteoclasts were treated with recombinant OPN (25 μg/ml) or TNF-α (10 ng/ml) for 30 minutes. Next, LPXN was immunoprecipitated from murine osteoclast lysates; the LPXN-immunoprecipitates were immunoblotted with the PYK2-Y402 antibody, which detects tyrosine phosphorylation of PYK2 on tyrosine residue Y402. As shown in Fig. 5, both OPN and TNF-α stimulated tyrosine phosphorylation of PYK2 after 30 minutes of treatment (Fig. 5A, lanes 3-4) compared with control untreated osteoclasts (Fig. 5A, lane 2). The control immunoprecipitation with mouse IgG1 MOPC-21 is also shown (Fig. 5A, lane 1).
Next, we asked the question of whether tyrosine phosphorylation of LPXN itself could be stimulated with OPN and/or TNF-α. LPXN was immunoprecipitated from murine osteoclasts that were treated with either OPN or TNF-α. These LPXN-immunoprecipitates were immunoblotted with the phosphotyrosine antibody PY20. An increase in tyrosine phosphorylation of LPXN was apparent after treatment of murine osteoclasts with OPN (Fig. 5B, lane 2) and TNF-α (Fig. 5B, lane 3). The relatively modest increase in tyrosine phosphorylation of LPXN may be attributed to the presence of only two potential tyrosine phosphorylation sites, as previously indicated.
LPXN associates with PTP-PEST and pp125FAK
PTPs function to dephosphorylate tyrosine phosphorylated substrates.(44) PTP-PEST, a nonreceptor protein-tyrosine phosphatase, has been shown to regulate the tyrosine phosphorylation of focal adhesion-associated proteins such as FAK, paxillin, and p130Cas.(45,46) We hypothesized that, similar to paxillin, LPXN associates with PTP-PEST. We also hypothesized that if indeed LPXN is a substrate for PTP-PEST, the dephosphorylation of LPXN by the phosphatase may be to alter its association with other signaling molecules associated with the podosomes, thereby regulating turnover of the podosomal complex.
We examined whether LPXN is associated with PTP-PEST, similar to paxillin in other cell types.(47) As shown in Fig. 6, PTP-PEST (∼116 kDa) can be detected by Western blots of murine osteoclasts lysates (Fig. 6A, lane 1). Next, LPXN was immunoprecipitated from murine osteoclast lysates; the LPXN-immunoprecipitates were immunoblotted for PTP-PEST. PTP-PEST co-immunoprecipitated with LPXN (Fig. 6A, lane 2); the control immunoprecipitation was performed with the mouse IgG1 MOPC-21 (Fig. 6A, lane 3). Next, PTP-PEST was immunoprecipitated from murine osteoclast lysates; the PTP-PEST immunoprecipitates were then immunoblotted for LPXN. Our results indicate that LPXN also co-immunoprecipitated with PTP-PEST (Fig. 6B, lane 2); the results of the control immunoprecipitation with NIS are shown (Fig. 6B, lane 1).
It has been previously reported that other members of the paxillin superfamily associate with pp125FAK.(48) It is known that osteoclasts do express pp125FAK, albeit at very low levels compared with that of PYK2.(49,50) We asked the question of whether pp125FAK can associate with LPXN. First, LPXN was immunoprecipitated from murine osteoclast lysates; the LPXN immunoprecipitates were immunoblotted for pp125FAK (Fig. 6C, lane 2). The control immunoprecipitation was performed with the mouse IgG1 MOPC-21 (Fig. 6C, lane 1). Next, pp125FAK was immunoprecipitated from murine osteoclasts lysates; the pp125FAK immunoprecipitates were immunoblotted for LPXN (Fig. 6D, lane 1). A control immunoprecipitation was performed with the mouse IgG1 MOPC-21 (Fig. 6D, lane 2). Although given the low levels of pp125FAK known to be present in osteoclasts,(49) our data indicate that there is a fairly significant amount of LPXN that associates with pp125FAK (Fig. 6D, lane 1). After stripping the blots in Fig. 6D, the pp125FAK immunoprecipitates were reprobed with an antibody to PTP-PEST to show that PTP-PEST also co-immunoprecipitates with pp125FAK (Fig. 6E, lane 1).
Finally, we showed that the protein tyrosine kinase PYK2 itself co-immunoprecipitates with PTP-PEST. Murine osteoclast lysates were used to immunoprecipitate PTP-PEST; the PTP-PEST immunoprecipitates were immunoblotted for PYK2 (Fig. 6F, lane 1); the control immunoprecipitation with NIS is shown (Fig. 6F, lane 2). These findings suggest that PYK2 may also serve as a putative substrate for PTP-PEST in the osteoclast.
Changes in serine phosphorylation of LPXN-associated PTP-PEST in response to stimulation with OPN and TNF-α
We have shown that the tyrosine phosphorylation levels of LPXN are modestly altered in response to stimulation of osteoclasts with either OPN or TNF-α. It is known that increases in serine phosphorylation of PTP-PEST correlate with increased activity of the phosphatase.(51) Therefore, we surmised that if LPXN is a putative substrate for PTP-PEST, increased tyrosine phosphorylation of LPXN would be accompanied by corresponding changes in serine phosphorylation levels of PTP-PEST, as we have previously shown for gelsolin.(52) Lysates prepared from vehicle- (control), OPN-, and TNF-α-treated murine osteoclasts were used to immunoprecipitate LPXN. The LPXN immunoprecipitates were immunoblotted with an anti-phospho-serine (p-serine) antibody. One single band was recognized at ∼116 kDa, corresponding to the molecular weight of PTP-PEST, as shown in Fig. 7A (a, lanes 1-3). As can be seen, there was an increase in serine phosphorylation after treatments with both OPN (lane 2) and TNF-α (lane 3); however, the increase in serine phosphorylation was slightly higher in the OPN-treated osteoclasts. These blots were then stripped and reprobed with an anti-protein tyrosine phosphorylation-PEST antibody, which also recognized a single band, corresponding to the same band seen in the phospho-serine Western blot (Fig. 7A, b, lanes 1-3). In several other similar experiments, the levels of LPXN-associated PTP-PEST were noticeably higher in OPN- and TNF-α-treated osteoclasts than the untreated controls (lane 1). This association of PTP-PEST with LPXN through co-immunoprecipitation studies has also been corroborated by affinity precipitation of PTP-PEST (Fig. 7B, lane 1) with the fusion protein GST-LPXN (Fig. 7B, lane 3). The negative affinity precipitation was performed with GST alone (Fig. 7B, lane 2)
The cellular distribution of PTP-PEST in the murine osteoclast was examined in relation to gelsolin, which was previously characterized as an osteoclast podosomal protein.(20) The staining for PTP-PEST is mostly peripheral (Fig. 7C, a) and corresponds to the area defined by the podosomal distribution of gelsolin (Fig. 7C, b), as seen by colocalization of the two proteins, indicated in yellow (Fig. 7C, c). Therefore, the subcellular distribution of PTP-PEST is in an area contiguous to that of LPXN, within the sealing zone/podosomal complex of the osteoclast. Next, the cellular distribution of LPXN and PTP-PEST was examined in murine osteoclasts, as shown in Fig. 7D. Both proteins (LPXN and PTP-PEST) were found to colocalize near the adhesion zone of the osteoclast, as indicated in yellow (arrow). Although we have hypothesized that LPXN is a direct substrate for PTP-PEST, we currently do not have the appropriate reagents available to us to show that such is the case.
LPXN is associated with the ARF-GTPase-activating protein p95 paxillin kinase linker
The scaffolding proteins paxillin and Hic-5 have been shown to associate with a complex of proteins that are regulators of actin cytoskeletal dynamics.(53) Because there exists a conservation of LD function across the paxillin superfamily, we decided to examine whether LPXN is also associated with other focal adhesion proteins known to associate with paxillin. This complex consists of p21 actopaxin, the GTPase-activated kinase (PAK), and the ARF-GTPase-activating protein p95 paxillin kinase linker (p95PKL).(24,53–55) Actopaxin is a widely expressed focal adhesion structural protein that binds directly to both F-actin and paxillin LD1 and LD4 motifs.(24) The PAKs are a family of serine-threonine kinases, localized to focal adhesions, whose activity is regulated by binding avidly to the small GTPases, Cdc42, and Rac.(56)
p95PKL is a multidomain protein that include an N-terminal ARF-GAP domain and two paxillin-binding subdomains.(54) Paxillin has been hypothesized to bind directly to p95PKL, which in turn, links paxillin to a protein complex containing the PAKs, events that have been linked to disassembly of focal adhesions.(53) The potential capacity of LPXN to assemble a PAK-containing complex through the ARF-GAP protein p95PKL could provide an integration of ARF and Rho family signal transduction at the cytoskeleton and may determine the subcellular localization of LPXN and other podosomal components in the osteoclast. Therefore, the putative association of LPXN with p95PKL was examined in the osteoclast. Murine osteoclast lysates were immunoblotted for p95PKL; equal amounts of lysates prepared from rat fibroblasts were used as a positive control,(54) as shown in Fig. 8. The signal for p95PKL can be detected at ∼95 kDa in both fibroblasts (Fig. 8A, lane 1) and osteoclasts (Fig. 8A, lane 2); as is quite evident, the levels of p95PKL in murine osteoclasts are much lower compared with that present in rat fibroblasts. Next, LPXN was immunoprecipitated from murine osteoclast lysates; the LPXN-immunoprecipitates were immunoblotted for p95PKL (Fig. 8B, b, lane 2). The control immunoprecipitation was performed using the mouse IgG1, MOPC-21 (Fig. 8B, b, lane 1). Finally, the association of p95PKL with LPXN through co-immunoprecipitation studies has also been corroborated by affinity precipitation of p95PKL (Fig. 8C, lane 1) by incubation of murine osteoclast lysates with GST-LPXN (Fig. 8C, lane 3). The negative (control) affinity precipitation was performed by incubation of murine osteoclast lysates with GST protein alone (Fig. 8C, lane 2). We have evidence that LPXN also co-immunoprecipitates with both actopaxin and the PAKs, proteins that would provide links to both focal adhesions or podosomes in osteoclasts, and actin cytoskeleton reorganization through binding of activated Cdc42 and Rac, respectively (data not shown).
Transduction of pTAT-LPXN into osteoclasts and changes in osteoclast morphology
Previously, overexpression of paxillin in fibroblasts has been shown to cause disregulation of Rac activity during cell spreading with generation of multiple lamellipodia.(55) In the current study, we have analyzed the effects of LPXN overexpression on osteoclast morphology.
Recently, the method of pTAT-mediated transduction of proteins has been successfully used to overexpress exogenous proteins into osteoclasts.(21,28,31,57) The LPXN cDNA was cloned into the bacterial expression vector, pTAT-HA, to produce pTAT-LPXN fusion proteins. A control HSV-TK protein (pTAT-TK, 42 kDa) was also generated, as previously reported.(21,31,58) The transduced fraction can be readily detected by an antibody to the HA tag upstream of the protein. The transduced pTAT proteins are typically retained in the osteoclast for up to 3 days.(31) More than 90% of the cells were transduced with pTAT fusion proteins, as previously reported.(21) After transduction of pTAT proteins into murine osteoclasts for 1-2 h (100 nM; previously determined for maximal uptake of pTAT-fusion proteins),(21) murine osteoclasts were immunostained with a HA tag antibody to detect the transduced proteins.
As shown in Fig. 9, a diffuse intracellular staining was seen in pTAT-HSVTK-transduced cells (Fig. 9A, a). In murine osteoclasts that expressed the transduced pTAT-LPXN protein (Fig. 9A, b), LPXN was mainly distributed toward the cell periphery (as indicated by the arrow); numerous cytoplasmic projections, suggestive of lamellipodia, could be seen in these murine osteoclasts, as indicated by the arrows. There were numerous lamellipodia-like structures and enhanced membrane protrusive activity, reminiscent of transient activation of Rac in other cell types that overexpressed paxillin.(54) We have examined the temporal expression of transduced pTAT-LPXN in the osteoclast (Fig. 9B). In the current study, expression of the transduced pTAT-LPXN protein in osteoclasts was apparent after 60 minutes of incubation, reached maximal levels between 16-24 h, and was retained until at least 2 days in culture.
In vitro inhibition of osteoclast resorptive activity by reduction of LPXN protein expression
We asked the question of whether LPXN has a functional role in the resorptive capacity of the osteoclast, using antisense ODNs to decrease the levels of LPXN protein expression. If as hypothesized, LPXN is an essential component of the sealing zone and serves to recruit other podosomal proteins to the signaling complex, then attenuation of LPXN expression would be expected to result in inhibition of either cell adhesion or resorption. Murine osteoclasts were plated on CaPO4-coated (1.0 μm thick) quartz discs and allowed to resorb the matrix for 48 h, as previously reported.(26) A representative phase contrast image of a murine osteoclast resorbing such a matrix is shown in Fig. 10A. Both antisense and sense ODNs were added to the culture medium at a final concentration of 10 μM. The murine osteoclasts were allowed to resorb the CaPO4 matrix for 48 h; these osteologic discs were examined under phase-contrast to visualize the extent of resorption.(26) As can be seen, the control (Fig. 10B, a, untreated) murine osteoclasts showed high resorptive capacity, defined by the extensive clear areas formed on the quartz discs. A similar resorptive capacity was seen in murine osteoclasts treated with the LPXN-sense ODNs (Fig. 10B, b, sense-ODN). In contrast, there was a very significant reduction of resorptive capacity when murine osteoclasts were exposed to the LPXN-antisense ODNs (Fig. 10B, c, antisense-ODN).
We examined the morphology of osteoclasts for the possibility that there was loss of adhesion after two consecutive 24-h treatments with antisense ODNs (Fig. 11A). As shown, the antisense ODN-treated osteoclasts tended to spread more compared with the other groups (control untreated and sense-treated). There was no loss of osteoclast adhesion in any of the ODN-treated groups. We have also determined that the protein levels of LPXN were reduced after antisense (ODN) treatment of murine osteoclasts but remained unchanged with sense (ODN) treatment (Fig. 11B).
Complex signaling mechanisms regulate activation of osteoclasts, initiated by adhesion to bone surface, cellular polarization, and formation of the sealing zone.(1) The initial cytoskeletal reorganization and podosome formation is a critical aspect of osteoclast activation. Podosomes, precursors to formation of the sealing zone, comprise an essential element of the integrin-associated signaling cascade. In this study, we have identified yet another component of the osteoclast podosomal complex.
LPXN, preferentially expressed in hematopoietic cells, is a member of the paxillin superfamily. In the osteoclast, LPXN probably functions as a “scaffold” protein to regulate signaling at sites of adhesion, such as the podosomes. Cross-species conservation and discrete composition of the protein-protein interacting motifs (LIM and LD) of members of the paxillin superfamily indicate that these peptide sequences are a critical feature in the specificity of function.
We have shown that LPXN, a tyrosine phosphoprotein, is associated with the protein tyrosine kinases such as PYK2 and pp125FAK in the osteoclast. In this preliminary characterization of LPXN, we have not provided evidence of whether LPXN is indeed a substrate for either PYK2 or pp125FAK.
Our data suggest that LPXN may be one of several downstream effectors of the podosome-associated signaling pathway in osteoclasts. Both osteopontin and TNF-α are osteoclast-activating cytokines.(10,21,39,59) After OPN-mediated stimulation of osteoclast motility and resorption, a reorganization of the osteoclast podosomal complex has been seen to occur.(21) Although it has long been recognized that TNF-α is an osteoclastogenic cytokine, not much is known about its effects on cytoskeletal proteins in the osteoclast. Previously, it has been shown that TNF-α can induce alterations in the actin cytoskeleton in a variety of cell types, including the tyrosine phosphorylation of PYK2 and paxillin.(40,41,60,61) We have shown that both OPN and TNF-α elicited modest increases in tyrosine phosphorylation of PYK2. Stimulation of osteoclasts with chemotactic peptides may promote the release of LPXN and associated proteins from a perinuclear vesicle compartment, permitting targeting to podosomes and recruitment of the tyrosine kinases FAK and PYK2. Further studies are needed to determine whether the association of LPXN with other osteoclast podosomal proteins are constitutive or inducible on direct stimulation with cytokines or indirect after adhesion. Future studies will also address the question whether the association of LPXN with PYK2 is dependent on changes in intracellular calcium in the osteoclast after adhesion.
In the current study, we have also shown that PTP-PEST is associated with both PYK2 and LPXN in the osteoclast and is truly a podosomal protein by virtue of its association with gelsolin.(62) What is the mechanistic significance of the association of LPXN with protein tyrosine kinases and PTP-PEST? Although we have not presented direct evidence that LPXN is a substrate for either PYK2 or PTP-PEST, we hypothesize that the adaptor function(s) of LPXN may be regulated through phosphorylation and dephosphorylation mechanisms, thus contributing to turnover of podosomes in osteoclasts.
The functional significance of LPXN binding to p95PKL and possibly the PAK-protein complex in the osteoclast is that the putative shuttling of LPXN between a cytoplasmic pool and the podosomes can be regulated in response to cytoskeletal reorganization after cell adhesion, similar to what has been previously shown for paxillin in other cell types.(53)
The morphology of the osteoclasts that were transduced with the pTAT-LPXN-fusion protein is similar to what has been seen in fibroblasts that were transfected with paxillin and certain mutants of paxillin.(54) Overexpression of paxillin evoked numerous lamellipodia-like structures and enhanced membrane protrusive activity, phenomena that correlated with transient activation of Rac.(54,63) We propose that LPXN may facilitate the appropriate subcellular localization of Rac regulators by recruiting the p95PKL-PAK complex to the podosomes in osteoclasts. In future studies, we plan to examine whether increased Rho/Rac/cdc42 activity correlate with the altered morphology of osteoclasts transduced with the pTAT-LPXN fusion protein.
Our data suggest that LPXN may play a critical role as an adaptor protein in formation of the adhesion zone in osteoclasts and that apparent disruption of LPXN's association with the podosomal signaling complex could inhibit the ability of the osteoclast to migrate and resorb its matrix. Further studies will determine whether such inhibition of LPXN expression also affects its association with the other binding partners that we have identified. In conclusion, our preliminary results indicate that LPXN is likely to play a rate-limiting role in osteoclast activity through its primary function as a scaffold protein in maintenance of the podosomal signaling complex.
This work was supported in part by a Designated Research Initiative Fund, University of Maryland, Baltimore, MD, USA, and National Institutes of Health Grants AR 44706 (to AG), AR 46292 (to MAC), and AR 41677 (to KAH). Joshua Goldknopf was supported by a Summer Research Program at the University of Maryland. The authors acknowledge the expert technical help of Dongmei Wang with cloning of the LPXN gene in the osteoclast. The authors are indebted to Dr Brian P Lipsky, ICOS Corporation, Bothell, WA, USA, for providing the monoclonal antibodies to leupaxin. The authors thank Dr CE Turner, State University of New York, Syracuse, NY, USA, for providing invaluable reagents for this study. The authors also thank Dr Michel L Tremblay, Department of Biochemistry, McGill University, Montreal, Canada, for providing helpful comments and reagents.