WO1993007271A1 - Osteoclast colony stimulating factor (o-csf) - Google Patents

Abstract

A substantially purified O-CSF protein capable of stimulating growth of TRAP-positive osteoclast progenitors in bone marrow cell cultures, and an isolated cDNA coding for O-CSF. Osteoclast progenitors grown in the presence of the O-CSF protein exhibit bone resorption activity. The O-CSF protein has an apparent molecular weight in the 20-25kD range, with a peak of biological activity at approximately 20kD, as determined by SDS-PAGE electrophoresis and biological assay. The O-CSF protein may be purified as disclosed herein from the culture medium of CESJ-3 cells (ATCC No. CRL 10886). The O-CSF protein substantially purified in this manner has an osteoclast progenitor stimulating activity of at least 660,000 units per milligram of protein.

Description

OSTEOCLAST COLONY STIMULATING FACTOR (O-CSF)

This is a continuation-in-part of prior application Serial No. 07/770,042, filed October 1, 1991, the benefit of the filing date of which is claimed under 35 U.S.C. § 120. This invention was made with government support under grants CA-38189,

AR-40820, AR-37318, AI-19565, and DK-38531 awarded by the National Institutes of Health and grant DE-FG06-86ER 60409 awarded by the Department of Energy. The United States government has certain rights in the invention.

Field of the Invention The invention relates to molecular biology, particularly to colony stimulating factors (CSF) that promote the growth and differentiation of hemopoietic stem cells and stimulate end-cell functional activity, and more particularly to an osteoclast colony stimulating factor (O-CSF) that induces osteoclast progenitors to differentiate and proliferate from bone marrow stem cells. Background of the Invention

Osteoclasts are bone resorbing cells that are distinguished from osteoblasts (bone forming cells) and osteocytes (resident cells in bone that are derived from osteoblasts). The opposing functions of osteoblasts and osteoclasts are necessary for normal bone growth and remodeling. Osteoblasts build bone by depositing calcium phosphate mineral (hydroxyapatite) within the collagenous organic framework of bone. Osteoclasts dissolve the hydroxyapatite, release calcium into the bloodstream, and also resorb the collagenous organic matrix. When the normal balance between these two activities is disrupted, various pathological conditions result.

Osteoclasts are derived from a hemopoietic stem cell population (see the appended Citations 1-3). Osteoclast-like cells have been identified by multinuclearity, calcitonin receptors, and tartrate-resistant acid phosphatase (TRAP), a cytochemical marker. TRAP-positive cells have been generated from bone marrow liquid cultures in vitro (4-7), and they have also been found in bone marrow populations enriched for hemopoietic stem cells (3,8). Hemopoietic cells and osteoclasts appear in parallel in the developing embryo and fetus (9). Although such prior studies have indicated that osteoclasts are derived from a cell that may be similar to the hemopoietic progenitor, the precise identity of osteoclast progenitors and the factors that regulate their development have been poorly understood. Similarly, it is not clear where the osteoclast lineage fits into the genealogical tree of the hemopoietic system. Systemic hormones such as vitamin D and parathyroid hormone (PTH) reportedly influence relatively mature mononuclear precursors to fuse and form multϊnucleated giant osteoclasts, and increase osteoclast function (10). Cytokines and growth factors such as transforming growth factor (TGF-α) (6), granulocyte- macrophage colony stimulating factor (GM-CSF) (5), and interleukin 3 (IL-3) (7-8) have been reported to promote the generation of osteoclast-like cells in bone marrow liquid cultures when added in conjunction with 1,25-dihydroxyvitamin D₃ (l,25(OH)₂-D₃₎ or parathyroid hormone (PTH) (5,7-8,11-12).

Changes noted in bone structure in disease and cancer may be related to bone metabolism including an increase in bone resorption. Osteoclasts have been implicated in bone disorders such as osteoporosis (increased numbers or activity) and osteopetrosis (decreased numbers or activity). Extensive marrow hyperplasia is also known clinically to result in skeletal abnormalities, and altered bone metabolism in thalassemia (13). Increased osteoclast activities have also been implicated in humoral hypercalcemia of malignancy, a serious complication associated with certain cancer patients. For example, hypercalcemia has been associated with exacerbation of leukemia (14). Certain tumors are also known to be abundant sources for growth factors and are known to influence functions of the host, such as causing severe hypercalcemia.

Possible tumor-derived humoral agents with osteolytic activity have included several factors: namely, PTH (15-17), PTΗ-related protein (18-20), prostaglandins

(21-23), and vitamin D metabolites (24). Tumor-derived humoral osteolytic factors, particularly PTH and PTΗ-related protein, have been shown to stimulate adenylate cyclase in vitro (18-20,25). Some of these factors were even recognized by PTH receptors and their in vitro actions, i.e., stimulating adenylate cyclase activity, could be blocked by a PTH antagonist (26). Ibbotson et al. (27) also recently provided evidence that a tumor-derived osteolytic factor might be a TGF. Certain growth factors, i.e., EGF, PDGF, and interleukin 1, have been shown to stimulate local synthesis of prostaglandin E₂ in vitro (28-31) and also to stimulate bone resorption activity in vitro (28,32). A number of other agents, i.e., complement (33), serum (34), phorbol ester (35), and melittin (35), have also been shown to stimulate bone resorption in vitro by a mechanism requiring local synthesis of prostaglandins in bones.

Transplantable CE mammary carcinoma has been known for years as a tumor that causes marked granulocytosis in mice (36-38). Lee and Baylink (39) demonstrated that this tumor also causes severe hypercalcemia secondary to osteoclastic bone resorption in mice. A cloned cell line (CESJ) from this tumor was established (40), and conditioned medium has been reported to have mitogenic activity for bone marrow granulocytic cells and embryonal bone cells in vitro (41). Recently, this cell line was reported to elaborate a colony stimulating factor selective for the putative osteoclast progenitor (42). Subsequent to the priority date of the present application, the inventors published their initial findings regarding the subject O-CSF (89).

In 1992 a review of cytokines and other local factors which affect osteoclast function was published by Mundy (90). This article emphasizes the likelihood of the existence of numerous factors which impact osteoclast development and activity.

Summary of the Invention We have isolated a specific colony stimulating factor (CSF) for the osteoclast lineage from conditioned medium of a hypercalcemia-inducing murine mammary tumor. This osteoclast CSF (O-CSF) appears to be a distinct molecule from other known CSFs of myeloid cells, as demonstrated by the biochemical and immunological properties of the protein and by its cDNA, which has been isolated, and which appears to be unique based upon comparison with sequences contained in genetic databases. The availability of this purified O-CSF and its cDNA provides new tools to investigate the physiology of osteoclast generation. Thus, the invention provides a substantially purified O-CSF protein capable of stimulating growth of TRAP-positive osteoclast progenitors in bone marrow cell cultures. The osteoclast progenitors grown in the presence of the O-CSF protein exhibit bone resorption activity as well. The subject O-CSF protein has an apparent molecular weight of approximately 15,000-25,000 daltons, with a peak of biological activity at approximately 20,000 daltons, as determined by SDS-PAGE electrophoresis and biological assay. The O-CSF protein may be purified as disclosed herein from the culture medium of CESJ-3 cells (ATCC No. CRL 10886). The O-CSF protein substantially purified in this manner has an osteoclast progenitor stimulating activity of at least 660,000 units per milligram of protein. Also provided are immunologic binding partner such as antibodies and their antigen binding fragments capable of binding the O-CSF protein. Reagent kits for diagnostic assays are provided, containing the O-CSF protein in a first container and an anti-O-CSF binding partner in a second container. The invention also provides a method of stimulating bone resorption by administering to a mammalian host the subject O-CSF protein. Brief Description of the Drawings

FIGURE 1 contrasts the effects of the detergents Tween 20 (FIGURE 1A) and CHAPS (FIGURE IB) on separation of O-CSF by gel filtration chromatography; FIGURE 2 presents photomicrographs which compare a red TRAP-positive osteoclast progenitor colony (FIGURE 2A) grown from bone marrow cells in CESJ- conditioned medium with a blue TRAP-negative colony, as discussed in Example 1;

FIGURE 3 presents photomicrographs which confirm the bone resorbing activity of osteoclast progenitor cells grown in CESJ-conditioned medium (FIGURES 3 A and 3B) and in the presence of purified O-CSF (FIGURE 3 C), as discussed in Examples 4 and 5; FIGURE 4 presents a photomicrograph of a TRAP-positive osteoclast grown in CESJ-conditioned medium showing indications of bone excavation, as discussed in Example 4;

FIGURE 5 presents results of the gel filtration chromatography purification protocol discussed in Example 5; FIGURE 6 presents results of the reverse phase HPLC purification protocol discussed in Example 5;

FIGURE 7 presents results of the SDS-PAGE protocol discussed in Example 5;

FIGURE 8 presents results of the reverse phase HPLC purification protocol discussed in Example 6;

FIGURES 9 A and 9B present results of the SDS-PAGE protocol discussed in Example 6;

FIGURE 10 presents photomicrographs showing in vivo effects of CESJ- conditioned medium on osteopetrotic mouse bones, as discussed in Example 8; FIGURE 11 A presents results of reverse phase HPLC of Sep-Pak™ eluted enriched hydrophobic proteins from CESJ-conditioned medium, and FIGURE 11B presents results of TRAPase-positive colony assays of C8 reverse phase HPLC fractions, as discussed in Example 9;

FIGURE 12 presents results of SDS-PAGE of C8 fraction aliquots, as discussed in Example 9; FIGURE 13 A presents results of analytical SDS-PAGE of pooled C8 fractions 17-19 fractionated further by a preparative SDS-PAGE transblot to PVDF membrane, and FIGURE 13B presents results of analysis of aliquots of the electroblotted proteins for O-CSF activity, as discussed in Example 9; FIGURE 14 presents results of mRNA hybridization analysis, as discussed in

Example 10; and,

FIGURE 15 presents the DNA sequence and the corresponding predicted protein sequence of the O-CSF cDNA clone, as discussed in Example 10.

Detailed Description of the Invention Cultures of a cell line derived from the CE murine mammary carcinoma that induces hypercalcemia were examined for soluble products that could induce osteoclasts to differentiate from murine bone marrow cells. The serum-free culture supernatant of the CESJ cell line stimulated growth of colonies from bone marrow cells which exhibited tartrate resistant acid phosphatase (TRAP) activity. These TRAP-positive cells demonstrated essential features of osteoclasts when cocultured with mineralized bone or dentin. The culture period required for colony development and the frequency of colony-forming cells indicated that relatively primitive marrow progenitors were stimulated by a tumor-derived factor(s) to form immature osteoclasts. Other colony stimulating factors (CSFs), including granulocyte CSF (G-CSF), macrophage CSF (M-CSF), granulocyte-macrophage CSF (GM-CSF), and interleukin 3 (IL-3), were ruled out as the source of the activity produced by the tumor cells. The biological activity was substantially purified by gel filtration chromatography and reverse phase HPLC. By SDS/PAGE, the activity was traced to a protein of approximately 20 kDa molecular weight. Functional and biochemical studies of the purified factor indicate that it is distinct from any known colony stimulating factors of myeloid cells. This protein appears to be a new colony stimulating factor for the osteoclast lineage, osteoclast CSF (O-CSF).

The term "osteoclast" as used herein means a mononuclear or multinuclear cell characteristically situated in close contact with calcified matrix showing signs of matrix resoφtion such as ruffled borders, clear zones, and bone excavation (43). The direct precursors of the osteoclasts are postmitotic cells which fuse and become osteoclasts under described conditions (43). Mature osteoclasts can be identified by such distinctive morphological characteristics. However, distinguishing progenitors of osteoclasts from other mononuclear cells in the bone marrow requires special techniques. TRAP staining is useful to identify colonies derived from putative osteoclast progenitors. Although TRAP-positive cells in the marrow are considered as putative osteoclasts by many investigators, the enzyme has also been demonstrated in other cell types (44-45). Thus, definitive identification of a cell as an osteoclast progenitor requires TRAP-positϊvity in combination with another indicator of osteoclast biological activity, e.g., ability to form osteoclasts in bone cultures, or bone resoφtion activity in an appropriate assay. For example, bone marrow cells are inoculated in culture flasks in defined liquid culture medium containing O-CSF. In 2 to 3 weeks, an adherent layer of bone marrow stromal cells and various hemopoietic cells including osteoclasts will develop in the bottom of the culture flask. Cells expressing calcitonin receptors in the adherent cell layer can be examined by specific binding of radioiodinated calcitonin and autoradiography of cells. Simultaneous staining of adherent cells for TRAP can be also performed (4,6,7,45).

The confirming test for identifying osteoclasts or osteoclast precursors is the biological activity of bone resoφtion, which can be demonstrated in vitro as bone excavation visualized by scanning electron microscopy (SEM) (6,45). In a representative example, described in greater detail below, dentin pieces are cocultured with putative osteoclasts, and then prepared for and examined by SEM. Dentin slices are more convenient than bone pieces for this puφose because the excavations made by the osteoclast and osteoclast precursor cells are readily demonstrated (and distinguished) by SEM (45-46). The number of resorption pits are estimated by screening the whole surface of the dentin slice. Demonstration of excavation on dentin slices cocultured with TRAP-positive cells in the presence of the O-CSF growth factor confirms that the cells are osteoclasts and/or osteoclast precursors. The size of bone pits formed by osteoclasts may vary depending on their maturational stages and functional status. Bone pits that are created by mature osteoclasts are larger than those created by osteoclast precursors: The former pits are generally in the range of 50-100 μm in diameter, or even larger, while the latter pits are generally in the range of 10-15 μm in diameter, the size of mononuclear cells in the marrow. Another assay method is to place a dentin slice directly in the culture dish in which bone marrow colonies are grown in agar in the presence of O-CSF. Osteoclast activity is confirmed by demonstrating that resultant bone excavations correspond with the locations of TRAP-positive colonies. Optimal concentrations of O-CSF induce nearly all the colonies to be TRAPase-positive. Controls for such experiments include M-CSF, either with or without any additional G- or GM-CSF, and with or without l,25(OH)₂-D₃. Osteoclast progenitors may also be conveniently demonstrated by simultaneous demonstration of calcitonin receptors by autoradiography (11) on TRAP-positive colonies which appear in response to O-CSF. Simultaneous expression of TRAP and calcitonin receptors, and comparison with control colonies stimulated by M-CSF, GM-CSF, or LL-3, allows identification of osteoclast progenitor cells.

Purification and characterization of the O-CSF for biochemical and molecular cloning techniques is reproducibly accomplished as follows. A clone of the deposited CESJ-3 cells (ATCC No. CRL 10886) is cultured in an 8 L stirred flask in serum-free, protein-defined HL-1™ medium (Ventrex) for 7 to 10 days to achieve confluent growth. CESJ cells grow well in this culture medium, but (as noted below) HL-1™ medium contains human transferrin that can contaminate O-CSF in the final RP-HPLC step of protein purification. Therefore, the cells growing in HL-1™ medium are preferably transferred into a serum-free, protein-free Medium 199, and then cultured for an additional 3 days prior to collecting the medium. Starting material for a representative purification protocol is about 10 liters of this CESJ medium in Medium 199. The medium is concentrated approximately 500-fold using Minitan™ and Amicon™ ultrafiltration systems with membranes having molecular cutoffs of 10 kDa, and stored at 4°C. O-CSF activity is stable for up to 6 months when stored in this manner.

Two different columns and elution systems may be used for further chromatographic purification (see below). A gel filtration column procedure clearly separates the O-CSF activity from the major protein peak. The major O-CSF activity is found in fractions corresponding to a molecular weight of about 30 kDa. The recovery of the O-CSF activity from a Sephacryl S-200 column is about 85%, and approximately 100-fold purification is achieved. Reverse phase HPLC is then used to obtain substantially purified O-CSF.

In a representative embodiment, the fractions with O-CSF activity (confirmed by TRAP and/or bone excavation) are pooled from several gel filtration runs, dialyzed and concentrated. 1-2 mg of protein is applied to the RP-HPLC column (e.g., Waters radial compression RCM-100 system and a radial PAK 5μ C18 cartridge). Suitable elution solvents are: A, 0.1% (v/v) trifluoroacetic acid (TFA)/H₂O, and B, 60% acetonitrile in 0.1% (v/v) TFA. A suitable elution gradient is described in Example 5. O-CSF is completely separated from G-CSF and M-CSF by this RP-HPLC procedure. The O-CSF at this stage is substantially purified by approximately 1300-fold, demonstrating specific activity of 660,000 U/mg; where 1 unit (U) of activity is defined as such that stimulates one colony or cluster from 10⁵ normal mouse bone marrow cells in 1 ml of semi-solid culture. A colony is defined as a group of 50 or more cells, and a cluster is defined as a group of more than 8 but less than 50 cells. In an alternative embodiment, C18 Sep-Pak™ cartridges (Waters, Division of Millipore, Millipore Coφ.) were used to enrich hydrophobic proteins, including O-CSF, in CESJ-culture medium prior to reverse phase HPLC and sequence analysis, as described in Example 9.

At each protein purification step, O-CSF activity is monitored by in vitro assays for detecting colony stimulating activities and TRAP staining of colonies. In order to facilitate analysis of many samples (e.g., fractions from column chromatography), the assay is modified to be performed in Linbro™ multi-well culture plates (0.5 ml/well) or in 96-well culture plates (0.2 ml well). M-CSF activity is detected by TRAP-negative macrophage colony growth. G-CSF activity is detected by factor-dependent proliferation of NFS-60 cells, using incoφoration of ³H-thymidine (47).

FIRST SERIES OF EXAMPLES In Examples 1-8 provided below, O-CSF is purified by molecular-sieve gel filtration chromatography in the presence of CHAPS (or similar zwitterionic detergent) using Sephacryl HR S-200 and then reverse phase HPLC (RP-HPLC). Use of a zwitterionic detergent was found to be critical to achieving an appropriately defined separation of O-CSF activity into resolvable peaks, such as those shown in FIGURE 5. A detergent, Tween 20, routinely used in protein separation, and in particular, cytokine purification (84), repeatedly failed to provide the resolution required to clearly identify a particular discrete set of column fractions which contained TRAPase-positive colony stimulation activity. FIGURE 1 contrasts the results of gel filtration chromatography of CESJ medium in the presence of 0.02% Tween 20 (FIGURE 1A) and 0.02% CHAPS (FIGURE IB). In both FIGURE 1A and IB the fraction number appears on the bottom x-axis, molecular weight markers are indicated by arrows across the upper x-axis, TRAPase-positive colony and cluster number/10⁵ bone marrow cells as indicated on the left y-axis (closed circles), and total protein, determined as optical density at 280 nm (dotted line) is indicated on the right y-axis. In the presence of Tween 20, O-CSF activity appeared to a varying degree in fractions collected throughout the column (FIGURE 1A). This was in contrast to G-CSF, which was clearly resolved into a single peak after elution with buffer containing 0.02% Tween 20 (data not shown). It was not until 0.02% CHAPS was used that a distinct peak of O-CSF activity was resolved (FIGURE IB). By these procedures O-CSF is well separated from M-CSF and G-CSF activities, and one skilled in the art can readily determine, in light of this disclosure, the presence of these extraneous factors either by assaying for their activities and/or by ELISA assay or immunoblotting with anti-M-CSF antibody or anti-G-CSF antibody on SDS-PAGE. EXAMPLE 1

Demonstration of TRAP-positive colony stimulating • activity in CESJ-culture medium When culture medium exposed to CESJ cells was concentrated by ultrafiltration and added to bone marrow cell cultures, distinct colonies composed predominantly of TRAP-positive cells consistently developed after 14 days of incubation. In control cultures, TRAP-positive colonies were almost absent. Representative results are shown in FIGURE 2, which presents color photomicrographs (magnification x250) of bone marrow colonies stimulated by CESJ-conditioned medium. A red TRAP-positive colony (FIGURE 2A) is contrasted with a blue TRAP-negative colony (FIGURE 2B). For additional experimental details, see the appended Materials and Methods.

As the concentration (v/v) of CESJ-conditioned medium increased in the bone marrow cultures, the number of TRAP-positive colonies as well as the intensity of TRAP expression increased. At optimal concentrations (7.5-10% v/v) of the conditioned medium in culture, 75-100 colonies were stimulated per 10⁵ bone marrow cells and over 90% of these stimulated colonies were strongly TRAP-positive. At a higher concentration (20% v/v) of the conditioned medium, the number of colonies shaφly decreased while all colonies were strongly positive for the TRAP activity. In contrast, similarly prepared culture medium of Bc66 or NFSA-c9 tumor cells stimulated many colonies but virtually all of them were composed of macrophages negative for TRAP activity (data not shown). In the presence of an optimal concentration (7.5% v/v) of CESJ medium, the number of TRAP-positive colonies increased linearly with increasing numbers of cells plated and the regression line intercepted zero, indicating each TRAP-positive colony was clonally derived. There was no apparent strain specificity, as indicated below. Effect of CESJ-cuIture medium (CM) on different strains of mice.

Data is for average of duplicative wells.

EXAMPLE 2

Elimination of known myeloid CSFs as TRAP-positive colony stimulating activity of CESJ-conditioned medium The effect of specific antisera to known CSFs was assessed against the TRAP-positive colony stimulating activity. Serial dilutions of antisera or purified antibodies were tested in the colony assays using a constant optimal dose of CESJ-conditioned medium. Anti-M-CSF, anti-GM-CSF, and anti-IL-3 had no effect on the TRAP colony stimulating activity of the conditioned medium, while the same antibodies completely neutralized the respective colony stimulating activities of M-CSF, GM-CSF, and LL-3. Since conditioned media of NFSA-c9 cells, which constitutively produce G-CSF and M-CSF, and conditioned medium of Bc66 cells, which produce M-CSF, do not stimulate TRAP-positive colonies, the ability of CESJ-conditioned medium to stimulate TRAP-positive colonies must involve factor(s) other than M-CSF or G-CSF. The lack of neutralization of TRAP-positive colony stimulating activity by antisera to GM-CSF or IL-3 also supports our previous findings that neither GM-CSF nor LL-3 are expressed in CESJ cells (47). Thus, G-CSF, M-CSF, GM-CSF, and LL-3 were excluded as factors responsible for the TRAP-positive colony stimulating activity.

EXAMPLE 3 Elimination of other factors Since TGF-β-like activity was demonstrated in the crude conditioned medium of the CESJ tumor (48), we examined the effect of recombinant TGF-βl on TRAP-positive colony formation at 0-500 pMoles/ml culture in the presence or absence of G-CSF (25-100 U/ml) and/or M-CSF (25-100 U/ml). None of these factor combinations stimulated TRAP-positive colonies. In fact, TGF-βl inhibited colony formation. Production of prostaglandins or vitamin D metabolites from the CE tumor has been excluded (48). PTH and PTH-related protein were ruled out by the failure of the conditioned culture medium to stimulate cAMP production from isolated kidney cells in culture (48). The CE tumor did not express mRNA for IL-1, and the tumor-conditioned medium contained no LL-1 activity (47). Furthermore, neither human nor murine LL-1 (10² to 10⁸ U/ml culture), with or without G-CSF (25- 100 U/ml) or M-CSF (25-100 U/ml), stimulated TRAP-positive colonies (data not shown). Tumor necrosis factor and platelet-derived growth factor were absent from the CESJ tumor products by L929 cell assay and radioreceptor assay, respectively (data not shown).

EXAMPLE 4 Osteoclastic function of TRAP-positive colony cells Dentin slices were placed at the bottom of Linbro™ wells and were overlaid with the typical bone marrow colony assay culture. On days 14, 21, and 28 of culture, colonies in the agar or on the dentin were evaluated, and the dentin pieces were processed for scanning electron microscopy. Corresponding to the site of TRAP-positive colonies, clusters of what appeared to be rudimentary resorption pits were observed on dentin pieces which were cocultured for 14 days. Representative results are shown in FIGURE 3.

FIGURE 3 presents scanning electron micrographs of the above-described dentin pieces. FIGURE 3 A is a low-power view (xl50) of dentin cocultured in agar with bone marrow cells for 14 days in the presence of crude CESJ-conditioned medium. Note the pitted appearance of the dentin surface in the lower field where a TRAP-positive colony was located. The smooth area in the left upper field is typical of dentin unaffected by TRAP-negative cells or colonies. FIGURE 3B is a higher magnification (x300) of the patchy area shown in FIGURE 3 A. Evident are several round craters suggestive of initiated resoφtion sites surrounded by the debris of cell processes (solid arrows), while many cells remain attached to the dentin surface (arrow heads). The numerous small light spots are dentinal tubules. Such resoφtion pits were not found in the areas where TRAP-negative colonies were present.

Histological sections of calvaria (bone pieces) cocultured with cells grown in the presence of CESJ-conditioned medium showed that 95-98% of the cells attached to the concave surfaces of the calvaria were TRAP-positive, some of which demonstrated multinuclearity and a ruffled border, both characteristic features of the functional osteoclast. FIGURE 4 is a photomicrograph (oil immersion, xl600) of such a calvarium section, showing TRAP-positive cells with typical features of osteoclasts. Note multinuclearity in the cell on the right and the ruffled border penetrating into the calcified matrix from the cellular extensions. In contrast, calvaria cocultured with cells grown in the presence of L-cell-conditioned medium revealed that more than 95% of cells attached were TRAP-negative (data not shown).

These studies provided strong evidence that TRAP-positive colony stimulating activity of the CESJ-conditioned medium indeed stimulates the growth of osteoclast progenitors and their terminal differentiation into functional osteoclasts. EXAMPLE 5

Biochemical isolation of O-CSF To isolate the osteoclast stimulating factor, medium exposed to CESJ tumor cells was concentrated and fractionated by gel filtration chromatography, followed by reverse phase HPLC. Initial isolation experiments using HL-1™ culture medium, which contained supplementary exogenous proteins, were not successful because these proteins tended to co-elute with the protein of interest. This problem was avoided by using protein-free Medium 199 to prepare tumor cell-conditioned medium which exhibited the same biological activities as that prepared in HL-1™ medium.

Gel filtration chromatography: Concentrated CESJ-conditioned medium containing 8-9 mg of total protein was applied to the gel filtration column, producing the elution profile shown in FIGURE 5. In FIGURE 5 (and FIGURE 6, discussed below), fraction numbers are arrayed on the lower x-axis; molecular weight markers on the upper x-axis (FIGURE 5, arrows); G-CSF activity (CPM x 10³) (dotted line) and optical density at 280 nm (solid line) on the left y-axis; and TRAPase-positive (solid line, closed circles) and -negative activity (dotted line, open circles) units (TRAPase-positive colonies and clusters/10⁵ bone marrow cells) are arrayed on the right y-axis. Colony stimulating activity appeared as two peaks clearly separated from the major protein peak (FIGURE 5, solid line, closed circles). Material from the first peak (corresponding to a molecular size of about 150 kDa) stimulated the formation of macrophage colonies that were TRAP-negative. A second distinct peak of CSF-activity was observed in fractions corresponding to an apparent molecular weight of about 30 kDa. In these latter fractions, almost all colonies and clusters were TRAP-positive and only a few TRAP-negative macrophage colonies were present. A peak of G-CSF activity also eluted in this region (FIGURE 5). Repeated gel filtration separations using batches of conditioned medium from different cultures consistently demonstrated similar protein and activity profiles. From these replicate separations, fractions 55-70 were pooled, concentrated, and applied to a C18 reverse phase column.

Reverse phase HPLC: Two mis of the pooled fractions from the gel filtration procedure containing 2-5 mg of protein were chromatographed. A single peak of TRAP-positive colony- and cluster-stimulating activity was repeatedly found in fractions 30-33 (FIGURE 6). In contrast to the gel filtration chromatography, G-CSF activity was completely separated from this peak (FIGURE 6). The HPLC fractions were tested individually for their ability to form TRAP-positive cells capable of excavating dentin. We found groups of resoφtion pits corresponding to the location of colonies containing TRAP-positive cells when dentin pieces were cocultured with marrow cells in the presence of material from fractions 30-33. FIGURE 3C (x600) shows groups of such resoφtion pits (arrows) formed on a dentin piece which was cocultured with bone marrow cells in a liquid culture for 21 days in the presence of material from the peak fraction of TRAP-positive activity on HPLC. These excavations (5-15 μm diameter) mark the locations of osteoclast progenitors. Such excavations were not found when other fractions were tested, or in areas on the dentin where TRAP-negative cells were present.

SDS/PAGE gel analysis: Reverse phase HPLC fractions containing TRAP-positive colony- and cluster-stimulating activity (fractions 28-37) were analyzed by SDS/PAGE without disulfide cleavage. When an SDS/PAGE gel was sectioned and eluted, the biological activity was found in slices corresponding with a Mr of 15-20 kDa.

Thus, FIGURE 7 presents representative results of this SDS-polyacrylamide gel electrophoresis of TRAP-positive colony stimulating activity from reverse phase HPLC. Protein in individual fractions (lanes 16-19) from the reverse phase separation containing the activity were run in separate lanes of the slab-gel. An unstained lane from fraction 19 was cut into 10 mm slices; protein was eluted and assayed for TRAP-positive colony stimulating activity ("+"). The results are shown to the left (++, +, and -). Activity was recovered in the 15-25 kDa range. Protein standards were run in the far right lane (14-200 kDa). The major protein components stained with Coomassie blue seen in lanes 17-19 are histones, which contaminate this particular preparation.

EXAMPLE 6 Further purification of O-CSF was achieved as follows Adsoφtion of hydrophobic protein to C18 Sep Pak™ column: Our approach is to utilize a hydrophobic property of O-CSF we have learned from our reverse phase HPLC. A large quantity (20-40 L) of CESJ tumor cell conditioned medium is passed through C18 Sep Pak™ columns. Hydrophobic protein including O-CSF is adsorbed to this column. Adsorbed protein is eluted by 70% acetonitrile in 0.1% trifluoroacetic acid, lyophilized, and reconstituted with 0.05 M phosphate buffer containing 0.02% CHAPS.

Gel filtration column chromatography: The sample (approximately 4-6 mg of protein) is applied to a Sephacryl HR 200 column and eluted with 0.05 M phosphate buffer containing 0.02% CHAPS, at a flow rate of 20 ml/hr, with 5 ml fractions collected. As shown before, O-CSF is eluted in fractions 55-70, and this elution profile is reproducible. Fractions 55-70 are pooled, concentrated using Amicon™ YM-10 membranes and exchanged with phosphate buffer without CHAPS.

Reverse phase HPLC: The sample is fractionated on a C8 reverse phase chromatography column using the solvents A = 0.1% trifluoroacetic acid (TFA) and B = 0.1% TFA/acetonitrile. The sample is eluted with a gradient from 0-15% B in 0-3 min, 30-70% B in 80 min, then 80% B, at a flow rate of 1 ml/min. Individual HPLC fractions are concentrated using Centricon™ 10 filtration units, and sterilized before O-CSF bioassay (TRAP-positive colony stimulating activity).

Results are shown in FIGURE 8. O-CSF activity contained in the pooled fractions from the gel filtration column (fractions 55-70) was concentrated, applied to the a C8 reserve phase HPLC column and eluted with an acetonitrile gradient beginning at 0.1% (v/v) trifluoroacetic acid in 20% acetonitrile for 3 min, then in 20-75% for 80 min, at 1 ml/min, and 1 ml fractions were collected. O-CSF activity was assayed in aliquots of the fractions (numbers on x-axis). A distinct peak of activity was observed at fractions 31-34 in this HPLC separation (FIGURE 8 solid line, closed circles, TRAPase-positive clusters/well, left y-axis). Absorbance (UV) is shown as the dotted line, right y-axis. SDS/PAGE: Individual fractions from the C8 reverse phase chromatography are run on 15% polyacrylamide slab gels, staining for protein with Coomassie blue, or silver staining. Results are shown in FIGURE 9 A and B. Protein in paired fractions (1-2, 3-4, etc.) from the C8 reverse phase HPLC (FIGURE 8) was run on 12.5% polyacrylamide slab-gel electrophoresis and stained with Coomassie blue (FIGURE 9A and B). Also run were portions of eluate from the C18 Sep Pak™ column (70% acetonitrile) that was applied to the Sephacryl HR 200 gel filtration column (lane A, FIGURE 9B) and of the pooled gel filtration fractions taken for reverse phase HPLC (lane B, FIGURE 9B). Only very faintly stained bands were observed in the 17 kDa region (fractions 31-34) where the O-CSF biological activity runs (FIGURE 9A and B), indicating that this purification procedure successfully eliminated contaminating proteins, and achieved several-fold further purification of O-CSF (> 1 million units per milligram). Molecular weight standards appear in the far right lane in FIGURE 9 A, and to the left of lane B in FIGURE 9B. Other combinations of molecular sieve, ion exchange and/or reverse phase chromatography on low pressure or HPLC column systems can be used to purify O-CSF. The novel bioassay procedure described herein is required to identify O-CSF and distinguish it from other CSFs in such preparations.

EXAMPLE 7 The effect of CESJ-conditioned medium on isolated monkey bone marrow progenitors The presence of high degrees of molecular and biological homologies between various murine and human colony stimulating factors has been demonstrated (49). We have tested the effects of CESJ tumor cell-conditioned medium on immature monkey bone marrow progenitors expressing CD34 antigen. The CD34 antigen has been shown to be expressed in human immature myeloid progenitors, as well as in osteoclast progenitors (50).

CD34-positive bone marrow cells were isolated from Macaca monkey bone marrow by an immunoadsoφtion column technique (51). Isolated cells were plated at 1000 cells per culture well containing various concentrations (% v/v) of CESJ tumor cell-conditioned medium with or without added l,25(OH)2~D3 in agar colony assays.

As shown in Table 1, CESJ tumor cell-conditioned medium was capable of stimulating TRAP-positive colonies from isolated CD34 positive monkey bone marrow cells in the absence of added vitamin D metabolites. These data indicate biological cross-reactivity of O-CSF with primate bone marrow cells, as well as direct action of O-CSF on bone marrow progenitors.

TABLE 1

The effect of CESJ conditioned medium on isolated monkey bone marrow progenitors.

Colony types and numbers/1000 CD34 positive cells

Values are average of duplicate wells.

EXAMPLE 8 In vivo effects of CESJ tumor cell-conditioned medium on osteopetrotic (op/op) mouse bones

FIGURE 10 presents photomicrographs of transverse sections of tibia from op/op mice injected with either CESJ tumor cell-conditioned medium (FIGURE 10F-J) or control medium (FIGURE 10A-E). Sex- and age-matched op/op littermates were injected daily with 0.1 ml of highly concentrated CESJ tumor cell-conditioned medium or control HL-1™ medium subcutaneously for 21 days. Five consecutive transverse ground bone sections were prepared, starting at the tibia-fibula junction and moving toward the proximal end of the tibia. The sections appear in FIGURE 10 in this consecutive sequence. Note larger marrow cavities with fewer bony trabeculae at all cross-sectional sites along the bone for mice that received tumor cell-conditioned medium compared with the control, indicating the significant bone-resorbing effects in vivo of the CESJ tumor cell-derived factor(s).

Little is known about the factors that regulate osteoclasts in their early stages of development. Although LL-3, GM-CSF, and M-CSF are reported to be involved in the formation of osteoclast-like cells in the presence of l,25(OH)₂-D3, it is unresolved as to which CSF has definitive effects on osteoclast generation (5,1 1-12,50). LL-3 and GM-CSF are absent in our CESJ tumor culture medium, and the O-CSF factor we have purified is capable of stimulating proliferation and differentiation of TRAP-positive osteoclast progenitors in the absence of added vitamin D metabolites in the culture.

Recently, Yoshida et al. (52) reported a mutation in the coding region of the M-CSF gene in an osteopetrotic (op) mouse, which mutation presumably caused the osteopetrosis. In our studies, M-CSF-producing tumors did not induce hypercalcemia, and treatment of mammary tumor-bearing mice with^' antiserum to M-CSF did not affect the tumor-induced hypercalcemia (47). Therefore, elevated levels of M-CSF in the circulation did not seem to affect osteoclast function. However, it is possible that M-CSF is needed for the local activation of osteoclasts, so that a complete lack of M-CSF in op mice may lead to osteopetrosis (53). Further characterization of O-CSF would elucidate the relationship of O-CSF to M-CSF as well as to the recently identified stem cell factor (54).

Materials and Methods Tumor cell cultures: Cloned cell lines of a hypercalcemia- and granulocytosis- inducing murine mammary adenocarcinoma (41,47), designated as CESJ-3, were used. Each CESJ clone produces bone modulating activity in addition to granulocyte CSF (G-CSF) and macrophage CSF (M-CSF) (41,47). Two other murine tumors were used as controls: a mammary carcinoma clone (Bc66), which does not induce neutrophilia or hypercalcemia in mice (55) and which produces M-CSF but not G-CSF (47); and a murine fibrosarcoma clone (NFSA-c9) (provided by Dr. M. Shikita, National Institute of Radiological Sciences, Chiba, Japan), which causes neutrophilia (56) but not hypercalcemia (Lee, unpublished) and is a known source of murine G-CSF and M-CSF (47). All tumor cells were cultured and maintained in serum-free, protein-defined HL-1™ medium (Ventrex, Bio Ventures Groups, Portland, ME), supplemented with 2 mM L-glutamine, 50 units/ml penicillin, 50 μg/ml streptomycin, and 0.125 μg/ml amphotericin B (Gibco).

The CESJ-3 cell line has been deposited as accession No. CRL 10886 at the American Type Culture Collection, 12301 Parklawn Drive, Rockville, Maryland 20852, U.S.A. Tumor-cell conditioned medium: The supernatant of culture medium in which tumor cells had been cultured for 7-8 days was concentrated approximately 500-fold by ultrafiltration (Amicon™ YM10 membranes, Amicon Corp.) and filtered (Millex™ GV, 0.22 μm, Millipore Coφ.). For the purification of the growth factor protein, CESJ cells were cultured in 5 L of HL-1™ medium in a stirred flask. Cells in the active growth phase were collected, washed twice, and suspended in 3 L of serum- free Medium 199 (Whittaker Bioproducts Inc.), supplemented with 40 μg/ml L-asparagine, 10 μM sodium pyruvate (Gibco), 50 units/ml penicillin, 50 μg/ml streptomycin, and 0.125 μg ml amphotericin B, and cultured for 3 days to obtain tumor cell-conditioned medium free of exogenous protein.

Bone marrow progenitor analysis: Bone marrow cells of young adult (Balb/c CE) Fl mice were cultured in 35 x 10 mm petri dishes or in 15 x 10 mm Linbro™ wells (Flow Laboratory) at 10⁵ cells per ml in supplemented Medium 199 containing 20% fetal calf serum (HyClone Laboratory), 0.3% Bacto agar (Difco Laboratories), and various concentrations of added tumor cell-conditioned medium or other test samples (9). Cultures were incubated at 37°C in a humidified atmosphere with 5% CO₂ for 7-21 days. Spleen cells were cultured similarly to bone marrow cells but using 5 x 10⁵ cells per ml. Colonies derived from putative osteoclast progenitors were identified by staining for tartrate resistant acid phosphatase (TRAP) activity. After the agar gel was transferred from the culture dish onto a glass slide and fixed (57), slides were stained for TRAP activity using hexazotized pararosaniline as a coupling dye and counterstained (58). Mononuclear cells containing TRAP activity were distinctively stained bright red. The colonies were examined under an inverted microscope and assigned to three categories based on their percentage of red (TRAP-positive) cells: positive (>90%), mixed (10-90%), and negative (<10%). Colonies were defined as groups of 50 or more cells, and clusters were defined as groups of more than 8 but less than 50 cells. All colonies and clusters in the agar plate were scored, and the results were expressed as colony or cluster numbers per 10⁵ bone marrow (BM) cells or per 5 x 10⁵ spleen cells.

Cocultivation of bone marrow cells with devitalized bone pieces: Bone marrow colonies were grown in the presence of an optimal concentration of tumor cell-conditioned medium as described above but using 0.8% methylcellulose instead of agar. On day 14 of the culture, colonies were collected, gently dispersed, and the cells were cocultured with devitalized mouse calvaria at 2 x 10⁵ cells per calvarium for 48 hours in 1.0 ml of Medium 199 containing 20% fetal calf serum and the optimal concentration of tumor cell-conditioned medium. Calvaria were fixed in formalin, decalcified, and 4-μm-thick sections were stained for TRAP activity.

Demonstration of resorption pits: Sterilized dentin slices (8 x 8 x 0.1 mm) were prepared from cow teeth and placed at the bottom of Linbro™ wells. 5 x 10⁴ bone marrow cells were cultured over the dentin slices in 0.5 ml medium containing 20% fetal calf serum, the optimal concentrations of tumor-conditioned medium, and 0.3% agar. In some instances, 10⁵ marrow cells were cultured in 0.5 ml liquid culture without agar using the same ingredients as above. After 14-28 days of incubation at 37°C in 5% CO₂, colonies that developed in the agar or on the dentin were stained for TRAP, and the locations of positive and negative colonies in relation to the dentin slice were recorded. The dentin pieces were subsequently immersed in distilled water or in sodium hypochlorite (50% v/v) for 30 min, washed, and dehydrated in ethanol (59). The specimens were sputter-coated with gold/palladium and viewed in an ETEC scanning electron microscope.

Growth factors and antibodies: Recombinant growth factors were from the following sources: murine G-CSF (Dr. S. Nagata, Osaka Bioscience Institute, Osaka, Japan); murine granulocyte-macrophage CSF (GM-CSF) (Genzyme, Boston, MA); murine interleukin 3 (LL-3) (Biogen, Geneva, Switzerland); human interleukin 1 (LL-1) (Hoffman LaRoche, Nutley NJ); murine 1L-1 (Genzyme, Boston, MA); and transforming growth factor-βl (TGF-βl) (Oncogen, Seattle, WA). Purified murine M-CSF (Dr. R. Shadduck, Montefiore Hospital, Pittsburgh PA) or L cell-conditioned medium was used as a source for murine M-CSF. Biological activities of these CSFs were tested by standard colony assays of bone marrow cells (57). G-CSF activity was also assessed by proliferation of factor-dependent NFS-60 cells (47). Rabbit antiserum raised against murine M-CSF was a kind gift from Dr. R. Shadduck. Anti-murine GM-CSF serum (60) and rat anti-mouse IL-3 monoclonal antibody (61) were kind gifts from Dr. D. Mochizuki (Immunex Corp., Seattle, WA) and Dr. J. Abrams (DNAX Research Institute, Palo Alto, CA), respectively.

Gel filtration column chromatography: A column (2.6 x 100 cm) of Sephacryl HR S-200 (Pharmacia Fine Chemicals) was equilibrated at 4°C in phosphate buffer (0.05 M NaH₂PO₄, 0.05 M NaCl, pH 7.2) containing 0.02% CHAPS detergent (Calbiochem Corp.). Two ml of 500 x concentrated CESJ tumor cell-conditioned medium was dialyzed against the above elution buffer, and the sample, containing 4-5 mg/ml of protein, was eluted at room temperature at a flow rate of 20 ml/hour collecting 5 ml fractions. The column effluent was monitored for protein absorbance at 280 nm. Fractions were sterilized by filtration (0.22μm) and stored at 4°C for biological assays. Reverse phase high performance liquid chromatography: Fractions containing

TRAP-positive colony stimulating activity were pooled from several gel filtration runs, concentrated by Amicon™ YM10 ultrafiltration, dialyzed against 0.05 M NaH₂PO₄ buffer (pH 7.2), and then against 0.1% trifluoroacetic acid (TFA). A 2 ml sample was applied to a C18 reverse phase radial pressure column (Waters, Millipore, MA). Elution solvents were: A, 0.1% (v/v) TFA/H₂O, and B, 0.1% TF A/60% acetonitrile. The sample was eluted with a complex gradient from 0 to 40% B (0 to 5 min), 40-72% B (5-40 min), then isocratically at 72% B (40-70 min), and finally 72-95% B (70-120 min), at a flow rate of 1 ml min, collecting 3 ml fractions in polypropylene tubes. Individual HPLC fractions were dialyzed against 0.05 M NaH₂PO4 buffer containing 0.02% (v/v) Tween 20, concentrated to 0.5 ml using Centricon™ 10 filtration units (Amicon, Danvers, MA), and sterilized by passage through 0.22 μm filters before CSF bioassay.

Sodium dodecyl sulfate/polyacrylamide gel electrophoresis (SDS/PAGE): The method of Laemmli (62) was used. Samples of individual fractions spanning the peak of TRAP-positive, colony stimulating activity from the reverse phase HPLC column were run on 15% polyacrylamide slab gels, staining for protein with Coomassie blue.

SECOND SERIES OF EXAMPLES In Examples 9-15 below, O-CSF is purified by reverse phase HPLC of C18 Sep-Pak™ enriched hydrophobic proteins from CESJ-culture medium, subjected to SDS-PAGE, and then subjected to amino-terminal sequence analysis. This alternative procedure was found to be necessary after it was determined that O-CSF, purified as discussed above by molecular-sieve gel filtration chromatography in the presence of CHAPS zwitterionic detergent, could not be prepared for sequencing by reverse phase HPLC due to the presence of an amino-terminal block. However, the CHAPS procedure allowed a band with the bone resoφtion activity to be identified which could then be isolated and sequenced from a more complex mixture of protein bands (see FIGURE 12).

In addition, in the following experiments the CESJ-culture medium was prepared in HL-1™ medium without transferring the cells into protein-free Medium 199 as described above. At this point in the experimentation, candidate protein bands had been identified and could be distinguished from exogenous protein on SDS-PAGE analysis. Furthermore, cell viability is enhanced in the HL-1™ medium as compared to Medium 199, and CESJ-culture medium prepared in is manner gave higher O-CSF activity and contained less histone than did CESJ-culture medium prepared in Medium 199. The level of histone in the medium decreases as cell viability increases, since a significant source of histone is lysis of the nuclear membrane of dead cells.

As noted above, once sufficient O-CSF had been purified by molecular-sieve gel filtration chromatography in the presence of CHAPS, reverse phase HPLC could be performed on a large enough quantity of material to resolve a faint, but distinct band on SDS-PAGE which possessed the O-CSF biological activity (see FIGURE 7, lane 18, band migrating at about 25 kDa). Unfortunately, this isolation protocol resulted in an N-terminally blocked protein. However, it did allow identification of a candidate protein band, and also allowed us to determine that O-CSF is a hydrophobic species. Furthermore, due to the hydrophobicity of the O-CSF protein, and/or to potential confoπnational changes which may occur in preparation of bioactive fractions for SDS-PAGE, our results indicate that O-CSF, at least under certain electrophoretic conditions, migrates anomalously on SDS-PAGE.

Therefore, in order to circumvent the amino-terminal block which was apparently caused by CHAPS or a contaminant in the CHAPS reagent, the C18 Sep-Pak™ column was used to enrich hydrophobic proteins from CESJ-culture medium. The hydrophobic species were then separated using reverse phase HPLC (see FIGURES 1 1 and 12 for elution profiles and analytical SDS-PAGE of individual fractions). Following this C18 Sep-Pak/reverse phase HPLC procedure, the pool of O-CSF activity was purified using SDS-PAGE, transblotting the proteins to PVDF membrane. This additional purification step was required due to a further complication caused by the co-elution of a particular viral core protein with O-CSF on reverse phase HPLC. After SDS-PAGE purification, a single protein band was identified that retained O-CSF biological activity on elution from a PVDF gel-blot and gave an N-terminal sequence on automated microsequence analysis. A synthetic oligonucleotide probe was constructed against a region of this sequence. The probe was used to isolate O-CSF mRNA and cDNA. The cDNA is capable of hybridizing to the sense or antisense strand of an O-CSF DNA element (as shown in FIGURE 15) under stringent conditions. The details of purification, sequencing and production of recombinant O-CSF are set forth below in Examples 9, 10 and 11, respectively. Particular examples set forth below also disclose production of recombinant O-CSF, receptor isolation and sequencing, animal model studies, production of anti-O-CSF immunological binding partners, and diagnostic and therapeutic applications.

The term "capable of hybridizing under stringent conditions" is used to mean annealing a first nucleic acid to a second nucleic acid under stringent conditions. For example, the first nucleic acid may be a test sample, and the second nucleic acid may be the sense or antisense strand of an O-CSF DNA element (as shown in FIGURE 15). Hybridization of the first and second nucleic acids is conducted under standard stringent conditions, e.g., high temperature and/or low salt content, which tend to disfavor hybridization of dissimilar nucleotide sequences. A suitable protocol (involving hybridization in 0.1 X SSC, at 68°C for 2 hours) is described in Maniatis, T., et al., Molecular Cloning: A Laboratory Manual, Cold Springs Harbor Laboratory, Cold Springs, NY, 1982, at pages 387-389. EXAMPLE 9 Sequencing the O-CSF protein Tumor cell-conditioned medium: A cloned cell line (CESJ3) of a hypercalcemia- and granulocytosis-inducing murine mammary adenocarcinoma (89) was cultured in serum-free, protein-defined HL-1™ medium (Ventrex, Bio Ventures Groups, Portland, ME), supplemented with 2 mM L-glutamine and 50 units of penicillin, 59 μg of streptomycin, and 0.125 μg of amphotericin B (GIBCO) per ml, in a 6 liter stirred flask. The supernatant of culture medium in which tumor cells were cultured for 8-10 days was collected, filtered through 0.22 μm filters (Millipore) and stored at 4°C as a starting material for purification and characterization of murine O-CSF.

Binding of hydrophobic proteins by C18 Sep-Pak: C18 Sep-Pak™ cartridges

(Waters, Division of Millipore, Millipore Corporation) were used to enrich hydrophobic proteins including O-CSF in the tumor-cell conditioned medium. Tumor-cell culture supernatant was supplemented with acetonitrile and trifluoroacetic acid (TFA) to a final concentration of 30% (v/v) and 0.1% (v/v), respectively, and passed through C18 Sep-Pak™ cartridges, 500 ml medium per cartridge with a flow rate of 40 ml/hr at 4°C. Upon completion of medium passages, each cartridge was washed by passing through 5 ml of 30% acetonitrile in 0.1% TFA at 1 ml/min using a 10 ml syringe. Bound protein was then eluted with 5 ml of 70% or 80% acetonitrile in 0.1% TFA at 1 ml/min. Eluates were diluted with an equal volume of distilled water and were lyophilized.

Reverse phase HPLC and bioactivity profile of O-CSF preparation: Sep-Pak eluted peptides were resolved by reverse phase HPLC (Brownlee C8 Aquapore™ RP-300) with a linear gradient of acetonitrile: 1-propanol (3:1, v/v) in aqueous 0.1% trifluoroacetic acid from 30-70% B (0.1% TFA/acetonitrile) over 80 minutes, preceded by a 3 minute hold at 15% B. 2 ml fractions were collected (x-axis) and monitored for UV absorbance at 220 nm (left y-axis), as shown in FIGURE 11 A.

C8 fractions were assayed for O-CSF activity in a bone marrow culture system that allows the detection of tartrate resistant acid phosphatase (TRAPase)-positive colony formation from marrow progenitors as described (89). In brief, bone marrow cells of young adult (Balb/c x CE)Fl, or C57Black 6 mice were cultured in 15 x 10 mm Linbro™ wells (Flow Laboratories) at 10⁵ cells per ml, in supplemented medium 199 containing 20% (v/v) fetal calf serum, 0.3% Bacto agar (Difco) and various concentrations of added test samples. After cultures were incubated at 37°C in a humidified atmosphere with 5% CO for 14 days, the agar gel was transferred from the culture dish onto a glass slide, stained for TRAPase, and stained colonies (>50 cells) and clusters (>8 but <50 cells) were scored for TRAPase-positive, mixed, and negative as defined previously (89). The O-CSF activity was expressed as the number of TRAPase-positive colonies and clusters developed from 10⁵ normal mouse bone marrow cells in response to added test samples. As shown in FIGURE 1 IB, fractions 15-19 (x-axis) were found to stimulate TRAPase-positive colony formation (right y-axis). G-CSF activity was assessed by proliferation of factor-dependent NFS-60 cells as previously described (89) and eluted as a single peak, with highest activity at fraction 35 (data not shown). C8 fraction aliquots and PVDF blot elution aliquots (see below) were prepared for bioassays by diluting with distilled water to a starting volume of 2 ml, concentrating 10-fold in Centricon™ 10 ultrafiltration units (Amicon), washing three times with 0.05 M phosphate buffer (pH 7.2) containing 0.001% Tween 20, 2 ml at each wash in the same filtration tubes, and finally equilibrating with medium 199 and concentrating to appropriate volumes for bioassays. Samples were sterilized by passage through 0.22 μm filters (GV4, Millipore) and stored at 4°C.

SDS-PAGE and sequence analysis of C8 fractions: C8 fraction aliquots were analyzed by 12.5% SDS-PAGE stained with Coomassie brilliant blue, as shown in FIGURE 12. Molecular weight standards were run in the right lane. In addition, peak O-CSF activity containing fractions were run on 10% SDS-PAGE and electroblotted to PVDF membrane (Westran, 0.45 μm) using a Milliblot™ semi-dry system (Millipore). Protein bands were stained with Coomassie brilliant blue, individually excised, and subjected to amino-terminal sequence analysis using a Porton 2090E gas-phase microsequencer equipped with on-line HPLC analysis of phenylthiohydantoin derivatives.

The labels TF, VCP, and H in FIGURE 12 correspond to proteins which were identified as human transferrin (a component of HLl™ medium), Moloney Leukemia viral core protein (at least two variant forms are present), and mouse histones, respectively, by sequencing the indicated bands in fractions 18 and 19 of FIGURE 12 after transblotting to PVDF. In addition, the faint 25-26kDa doublet in fraction 35 was subjected to N-terminal sequence analysis with results that identified it as mouse G-CSF. The two bands labeled O-CSF, located at approximately 25kDa and 28kDa on electrophoresis of fraction 18, both gave a novel N-terminal sequence, PG(G)LL(Y)GDEAPNFEAN. The sub-sequence GDEAPNFEAN was used to construct a synthetic oligonucleotide probe for isolating O-CSF mRNA and cDNA. Association of O-CSF activity with peptides eluted from PVDF Blot: C8 fractions 17-19 were pooled, run on 10% SDS-PAGE, and electroblotted to PVDF membrane but not stained. The PVDF membrane was then sliced horizontally into 5 mm sections, beginning at the buffer front and extending upward beyond the 92.5kDa molecular weight standard (Amersham Rainbow Markers™). Each section was submerged in 500 μl 70% acetonitrile, 1% trifluoroacetic acid, and incubated 48 hours at 4°C. Supernatant was removed and aliquots were either assayed for O-CSF activity as described above, or analyzed by 10% SDS-PAGE stained with Coomassie Blue, as shown in FIGURE 13 A. In the former system, as shown in FIGURE 13B, peak TRAPase-positive O-CSF activity (left y-axis, TRAPase-positive cells/well) was present in segments 4, 5, and 6 (x-axis). It can be seen from these results, when compared to those above, that the bioactive O-CSF protein fractions from conditioned medium may contain bioactive degradation products, or the bioactive species may in fact be heterogeneous due to proteolytic cleavage of active precursors into products possessing greater or lesser bioactivity. Nevertheless, the protein band at 25kDa (molecular weight standards are in FIGURE 13 A, far right lane) was obtained as a single component (segment 6) which exhibited O-CSF biological activity and contained the amino-terminal sequence used to prepare the oligonucleotide probe for cloning the cDNA. It is of particular interest to determine the active sites of O-CSF by specifically cleaving purified or recombinant O-CSF with particular proteolytic enzymes or chemical reagents. Such cleavage gives rise to predictable, identifiable fragments which are readily tested for bioactivity as described herein. For example, the reagent cyanogen bromide is used to cleave the O-CSF adjacent to methionine residues, and the peptide cleavage products are then resolved and analyzed for bioactivity.

EXAMPLE 10

Molecular cloning of O-CSF cDNA

RNA analysis: Total RNA was harvested from CESJ-3 (O-CSF +) cells and

Bc66 (O-CSF -) control cells, and 30 μg of RNA from each cell type was fractionated on a 1.2% agarose/formaldehyde gel. The RNA was then transferred to Zeta Bond™ nylon membrane (Biorad) and probed with a 30mer P³² end-labeled oligonucleotide derived from the N-terminal sequence of purified O-CSF protein (67). A specific hybridization signal was observed at approximately 1.4 kb in the CES J-3 positive lane, as shown in FIGURE 14 (SJ-3 lane, arrow). This signal was absent from the O-CSF negative control cells, as is also demonstrated in FIGURE 14. Purified poly A⁺ mRNA from CESJ-3 O-CSF producing cells was used to construct a cDNA library, and colony hybridization using the end-labeled oligonucleotide probe identified positive clones for investigation of nucleotide size and sequence.

Polymerase Chain Reaction (PCR) amplification of O-CSF mRNA: Total RNA from O-CSF positive and negative cells was reverse transcribed using M-MLV reverse transcriptase (Gibco-BRL) in the presence of dT-tailed universal primer. After reverse transcription the reaction mixture was treated with RNAase H and then subjected to PCR. PCR reaction was performed with Taq Polymerase (Perkin-Elmer) using a specific primer at the 5' end (based on the amino acid sequence) and a Universal primer at the 3' end. The amplification reaction involved denaturation at 94°C for 1 min, annealing at 55°C for 2 min and extension at 72°C for 1 min. A specific amplified product of approximately 800 bp was observed in the O-CSF cells which was subsequently cloned into a T- Vector generated using pBSK plasmid. Double stranded dideoxy sequencing was performed using the Sequenase kit (USB). The coding sequence of O-CSF cDNA and the amino acid sequence of the protein coded for by that cDNA are presented in FIGURE 15. It should be noted that to date no significant homology to any known DNA sequences has been detected in searches of genetic information databases such as Gene Bank and the Swiss Protein database.

The underlined sequence in FIGURE 15 shows the GDEAPNFEAN sub¬ sequence used to construct the oligonucleotide probe. The entire coding sequence to the stop codon (TAG, marked with * in FIGURE 15) predicts the 250 amino acid residues shown, which, together with the six additional residues at the amino terminus, those being PG(G)LL(Y), from protein sequence analysis indicates a mature protein (less any signal peptide or processed propeptides) of 256 amino acid residues, and a molecular weight of 28 kDa. This amino acid sequence predicts a hydrophobic protein with potential glycosylation sites which could result in aberrant behavior on SDS-PAGE, and the possibility of more than one subspecies. It is important to note that one skilled in the art of protein analytical techniques might expect a hydrophobic species to run faster on SDS-PAGE (i.e., to give an apparent molecular weight on SDS-PAGE lower than predicted based upon amino acid composition). The O-CSF cDNA is conveniently tested for expression of bioactive O-CSF by insertion into a retroviral vector and transfection into NIH 3T3 cells or COS cells. Such vectors based on the Molony murine leukemia virus express human purine nucleoside phosphorylase (67), adenosine deaminase (68), and canine G-CSF. A particularly strong promoter is provided by the retroviral long terminal repeat, LTR (67-68), and this type of available vector is preferred to express O-CSF. The vector includes the selectable neo gene which confers resistance to the antibiotic G-418. The plasmid construct is designated LOSN, indicating that O-CSF is expressed from the viral LTR and that Neo expression is controlled by the SV40 promoter (67-68). Following transfection with this construct of NIH 3T3 or monkey COS cells, randomly pooled clones are selected in G-418 antibiotic. These selected cells are used to provide conditioned medium to assay for O-CSF using the bone marrow-TRAP colony assay. If necessary or desired, the conditioned medium may be first concentrated and/or HPLC purified. Furthermore, other available vectors (e.g., plasmids) and host cells (e.g., bacteria, yeast, etc.) may alternatively be employed to produce recombinant O-CSF. For example, we are using the Invitrogen Xpress™ system (pTrcHis) to obtain rapid purification of O-CSF protein (Invitrogen publications, Nos. 120521sa and 12050sa). The pTrcHis vectors are pUC-derived expression vectors designed for efficient protein expression and purification from cloned genes in E. coli. This system provides high level production of the recombinant protein fused at its N-terminus to a tract of six histidine residues that function as a metal binding domain and an enterokinase cleavage recognition sequence. The metal binding domain of the fusion protein allows simple one-step purification of recombinant protein by immobilized metal affinity chromatography. The enterokinase recognition site between the metal binding domain and the recombinant protein enables cleavage of the fusion peptide from the purified recombinant protein which can men be isolated free of the fusion peptide by passage through a metal affinity column.

Murine O-CSF cDNA is used to isolate human O-CSF cDNA which enables purification of recombinant human O-CSF protein. Murine O-CSF cDNA is used to screen human cDNA libraries derived from hematopoietic or marrow stromal cells. Because we have shown cross-reactivity of murine O-CSF with primate bone marrow cells (Example 7), and the previously demonstrated high degrees of molecular homology between murine and human colony stimulating factors (49), we expect to readily identify clones encoding human O-CSF cDNA by high stringency screening using isolated radiolabeled murine O-CSF cDNA as a hybridization probe.

O-CSF cDNA is also useful to examine expression of O-CSF in various tissues of normal and osteopetrotic mice by Northern blot analysis. For example, messenger

RNAs isolated from cultured stromal cells of various mouse tissues and organs are size-fractionated by electrophoresis, blotted to membranes, and hybridized with ³²P-labeled O-CSF cDNA as described (64). All blots can be probed for beta-actin transcript as a control for quantitative RNA expression (47). Different levels of O-CSF expression in marrow stromal cells of certain osteopetrotic mouse strains versus controls would suggest an important role of marrow stromal O-CSF in the pathogenesis of subtypes of osteopetrosis. Northern analysis can also be used to show the presence of O-CSF mRNA in CESJ cells but not control cells.

Southern analysis (64) of genomic DNA from CESJ clones and control cells is performed as previously described (69). Southern analysis can also be used to demonstrate the presence of a single locus or more complex gene structure for O-CSF. In situ hybridization with RNA probes: Synthesis and purification of a complementary RNA probe (cRNA) is achieved as described (70). Anti-sense RNA transcripts (cRNA) complementary to the corresponding O-CSF mRNA are prepared by in vitro transcription of O-CSF cDNA in the cloning vector plasmid Gemini. The plasmid is linearized and following phenol/chloroform extraction is oriented such that transcription from the SP6 RNA polymerase promoter yields anti-sense transcripts complementary to the corresponding region of O-CSF mRNA. For labeling, the linearized template is transcribed in a reaction mixture containing ¹²⁵I-CTP, and the RNA product is separated from the template by electrophoresis on a low-gelling-temperature agarose gel and excised from the gel. In situ localization of O-CSF mRNA in tissues: For /^'// si/u hybridization, mice are perfused transcardially with 2% paraformaldehyde and 15% picric acid in 0.1 M phosphate buffer. Cryostat sections are prepared from various tissues. The radiolabeled probe is applied to the tissue sections, which are covered with siliconized coverslips and incubated. The sections are then exposed to RNAase A (70-71). Hybridization of the probe to O-CSF mRNA is detected by autoradiography (70-71). Localization of grains in various cell types and tissues provides significant information about the sites of O-CSF production in vivo.

EXAMPLE 1 1 Production of recombinant O-CSF Recombinant O-CSF is produced, in a representative embodiment, using

C127I cells engineered with an O-CSF cDNA that is shown to direct the synthesis of bioactive protein.

Mouse C127I cells are transformed by bovine papilloma virus (BPV), and the BPV DNA is maintained extrachromosomally as multicopy circular DNA (72-73). This system has been previously used to synthesize human interferons (73) and human G-CSF (72). Recombinant murine G-CSF has also been produced using this cell system and the BPV construct. Here, recombinant protein is produced by mouse C127I cells transformed with a bovine papilloma virus expression vector containing a hybrid gene in which mouse O-CSF cDNA is expressed from the SV 40 early promoter (72). O-CSF cDNA is inserted in a plasmid vector (pd CR) immediately downstream of the SV40 early promoter and upstream of the rabbit β-globin intron splice sites and the SV40 polyadenylation site. The hybrid gene can be joined with BPV vector pdBPV-1. Mouse C127I cells are transfected with the recombinant DNA by calcium phosphate coprecipitation (73). After 12 hours at 37°C, cells are trypsinized, replated, and, two weeks later, moφhologically transformed foci are individually picked and grown and maintained in Dulbecco's medium with 10% FCS. DNA of transformed cells is analyzed by Southern blot analysis. The transformed cells secrete a constitutively high level of O-CSF.

Secreted O-CSF is purified to homogeneity by gel filtration and RP-HPLC. The purified protein is electrophoresed on SDS-PAGE, and demonstration of a single band at the expected molecular weight corresponding with the biological activity indicates purification of the homogeneous protein. The biological activity of the recombinant molecule can be compared with the activity of the natural O-CSF protein.

EXAMPLE 12 Animal model studies

There appears to be no mouse-strain specificity in O-CSF in vitro activity (see above). However, isolated recombinant O-CSF is conveniently tested by daily injections into normal mice; alternatively, a mini-osmotic pump (Alzet pump) may be employed. Young adult mice are divided into two groups, experimental and control groups, 8-10 mice in each. The isolated factor or an appropriate control medium is injected daily at two or three different concentrations into these animals for 3 weeks.

Serum calcium is checked at several day intervals from a drop of tail blood.

After 3 weeks of injections, the mice are sacrificed, and bone specimens are prepared from the femur and tibia. The number of endosteal osteoclasts is evaluated on TRAP-stained histological sections of the femur as described (74). The bone area measurements are obtained by moφhometric analysis of standardized ground bone sections of the tibia-fibula junction, as described (39). These bone parameters and the observed serum calcium values are compared between the O-CSF- and control medium-treated mice, and the statistical significance is analyzed by t-test. Increased serum calcium values indicate the hypercalcemic effects of the O-CSF factor. O-CSF can be used to induce an increase in marrow cavity area, and osteoclast numbers in mice treated with O-CSF in such assays are helpful for establishing that O-CSF is stimulatory for osteoclast cells in vivo.

Osteopetrosis is a condition characterized by generalized skeletal sclerosis due to reduced bone resoφtion, with almost complete filling of the long-bone marrow cavities with bone. Although some forms of osteopetrosis can be cured by bone marrow transplantation at an early stage (75), there is no known treatment for reversing the osteopetrosis once the process is advanced, and administration of hormones such as PTH or Vitamin D have been unsuccessful (75). There are four distinct mutations in the mouse that can individually result in osteopetrosis: osteopetrotic (op), microphthalmic (mi), grey-lethal (gl), and osteosclerotic (os) (75-76). The primary lesion in mi/mi is believed to be in the stem cells, which fail to give rise to osteoclasts. This condition can be cured by transplantation of normal bone marrow cells (75-76). It is believed that lack of M- CSF in the bone marrow microenvironment fails to support osteoclast development in op/op mice. This condition cannot be corrected by bone marrow transplantation (75- 76). The potential therapeutic effects of O-CSF in treating osteopetrosis can be investigated in these various mouse strains. The mi/mi osteopetrotic mouse is a potential candidate for correction by O-CSF, but op/op is also a useful strain for comparing the corrective activities of O-CSF and M-CSF. Murine O-CSF cDNA is used to obtain O-CSF genomic DNA which enables, by targeted mutation (91, 92), production of transgenic mice lacking O-CSF gene product. Study of mice carrying a mutation of the O-CSF gene provides information of the developmental and functional role of O-CSF. In addition, progeny of such transgenic products of homologous recombination can be rescued by administration of recombinant O-CSF or its active moieties and by gene therapy employing retroviral vectors to obtain normal O-CSF gene expression in vivo.

EXAMPLE 13 Production of anti-O-CSF immunologic binding partners Polyclonal antibodies have been successfully developed in rabbits for murine GM-CSF (99), murine M-CSF (77), and murine G-CSF (S. Nagata, personal communication); we expect O-CSF also to be antigenic in rabbits. Standard immunization protocol is followed. Rabbits are immunized by reverse phase HPLC-purified recombinant O-CSF (5-10 μg protein) showing a single band on silver-stained SDS-PAGE, by multiple intradermal injections with Freund's adjuvant (100-101). The antibody production is measured by ELISA screening assay on microtiter plates or, if neutralizing antibodies are required, by selective neutralization of O-CSF-mediated TRAP-positive colony formation in agar cultures in vitro. When there is a high titer of antibody, the rabbit is bled, and IgG antibody is separated by a protein A column (78). The specificity of the antibody to O-CSF is tested by its ability to immunoprecipitate radioiodinated O-CSF but not other cytokines such as M-CSF, GM-CSF, and G-CSF (60). Radioiodination of CSF is achieved by the chloramine-T method, which has been used to iodinate M-CSF, GM-CSF, and G-CSF by other investigators (60,78). Radioimmunoprecipitation of antibody-radioiodinated CSF complex is evaluated on PAGE followed by autoradiography (60).

Monoclonal antibodies to O-CSF are, for example, prepared in the mouse. Suitable immunization protocols for the mouse have been described (78). Spleen cells from the immunized mouse are harvested, homogenized, and thereafter fused with myeloma cells in the presence of polyethylene glycol to produce a fused cell hybrid which produces monoclonal antibodies specific to O-CSF. Screening protocol may involve the use of a particular target cell line that proliferates in response to O-CSF, for example, if a neutralizing antibody is required. Various factor-dependent cell lines have been developed by other investigators, and have been used in similar situations for other hemopoietic growth factors (79-80).

In this manner, immunologic binding partners that bind to O-CSF are raised and selected. Of course, it is understood that O-CSF-specific immunologic binding partners include antigen-binding fragments (e.g., Fab) of such immunoglobulins, whether raised in animals, produced by hybridoma techniques (81), or produced by recombinant DNA techniques involving expression of a genetic construct that encodes an expression product that binds O-CSF. In particular, several authors have recently described a commercial system available from Stratacyte, La Jolla, California, which enables the production of immunologic binding partners through recombinant techniques (82). The subject immunologic binding partners preferably exhibit O-CSF binding constants higher than 10-⁸, and more preferably 10~¹⁰ or higher.

EXAMPLE 14

Diagnostic use of O-CSF and anti-O-CSF reagents Antibodies produced by the above procedures, or by equivalent procedures known in the art, are employed in various immunological assays to quantitate the concentration of O-CSF in various tissue fluids in normal and pathological conditions.

These assays utilize an antibody coupled to a detectable marker. Examples of suitable detectable markers include: enzymes, coenzymes, enzyme inhibitors, chromophores, fluorophores, radionuclides, etc. Examples of standard immunometric methods suitable for quantitating the O-CSF growth factor include radioϊmmunoassay (RIA) and enzyme-linked immunosorbent assay (ELISA). For example, RIA for murine M-CSF has been well established (83-84).

These and other immunometric methods known in the art are used to determine the concentration of O-CSF in serum, plasma, or urine to observe physiological levels of O-CSF in normal conditions. Elevation or reduction of O-CSF levels may be observed in pathological conditions where the normal, rate of bone resoφtion is altered, or in congenital defects such as osteopetrosis.

The immunological reagents can also be used to detect which cells or tissues secrete O-CSF protein. Standard immunocytochemical technique (85) can be applied to bone marrow samples, or other tissues samples. For example, frozen sections are prepared from various tissues of normal and op/op mice. The frozen sections of bone marrow are prepared by carefully scooping the bone marrow core from longitudinally split femurs and placing it in gelatin capsules containing freezing solution (OTC). Frozen sections are air dried and reacted with rabbit antiserum to O-CSF followed by secondary labeling with fluorescein-conjugated goat anti-rabbit IgG (H+L) obtained from a commercial source (Pierce). Appropriate controls are prepared using normal rabbit serum. Stained sections are evaluated under a fluorescence microscope (Leitz, Ortholux II).

O-CSF and anti-O-CSF reagents can also be used to detect and isolate O-CSF receptors. Receptors for various hemopoietic growth factors such as erythropoietin, M-CSF, and G-CSF (86) have been isolated in this manner. This information provides the basis for designing and/or screening for reagents to block O-CSF receptor binding and O-CSF recruitment. Such reagents have potential value in prevention and treatment of osteoporosis and other osteolytic diseases. For example, expression cloning of the O-CSF receptor(s) is accomplished using radiolabeled O-CSF protein to identify clones expressing the O-CSF receptor (87). mRNA from bone marrow cells (enriched for osteoclast progenitors) is used to construct a cDNA library (>10⁵ clones) in an expression vector (87). Pooled clones are screened following transfection into COS 7 cells by autoradiography using ¹²⁸I-labeled O-CSF. Positive pools of clones are partitioned and screened by retransfection to identify single clone(s) expressing O-CSF receptor. The isolated cDNA is sequenced and the amino acid sequence of the O-CSF receptor determined.

An alternative strategy for isolating the O-CSF receptor involves affinity chromatography. Purified O-CSF is cross-linked to an affinity matrix and used in column chromatography to isolate the O-CSF receptor from solubilized bone marrow cell membrane proteins. Nonspecifically bound proteins are washed off the column and O-CSF receptor protein eluted with a chaotropic agent.

Immunological binding partners are then raised against the isolated O-CSF receptor by available techniques. Such binding partners are screened for the ability to block O-CSF binding to its receptor on cell surfaces.

EXAMPLE 15 Therapeutic use of O-CSF, anti-O-CSF, and agents that interfere with O-CSF/receptor interactions Certain forms of human osteopetrosis may be treatable by administration of human recombinant O-CSF. This very rare, heritable condition is believed to have different underlying causes, which may include the failed differentiation of osteoclast precursors, progenitors, or functional osteoclasts. Treatment progress is monitored by assaying patient serum or urine for indicator markers of bone degradation, such as increased serum levels of calcium or urine and serum levels of collagen cross-linked peptides (88).

Perhaps the most significant application to emerge from understanding O-CSF properties and biological activities (including the cloning and characterization of its receptor(s) on marrow stem cells) would be to design chemical and biological agents that can block or inhibit O-CSF's action. In this way, a fundamental approach to the prevention and treatment of osteoporosis would be possible by regulating the numbers of recruited osteoclasts and their bone-resorbing activity. Increased osteoclast recruitment is believed to be a significant factor in postmenopausal osteoporosis.

In addition, measurement of circulating levels of O-CSF (serum O-CSF concentration) by immunochemical or other means could have diagnostic value in the clinical assessment of metabolic bone disease and risk of such disease.

Citations

1. Walker, D. (1975) J. Exp. Med. 142, 651-663.

2. Ash, P., Loutit, J.F. & Townsend, K.M.S. (1980) Mature 283, 669-670.

3. Scheven, B.A.A., et al., (1986) Nature 321, 79-81. 4. Roodman, G.D., et al., l,25-dihydroxyvitamin D₃ causes formation of multinucleated cells with several osteoclast characteristics in cultures of primate marrow. Proc Nat Acad Sci 82, 8213 -8217, 1985. 5. MacDonald, B.R., et al., Effects of human recombinant CSF-GM and highly purified CSF-1 on the formation of the multinucleated cells with osteoclast characteristics in long-term bone marrow cultures. J Bone Mineral Res 1,

227-233, 1986. 6. Takahashi, N., et al., Recombinant human transforming growth factor-alpha stimulates the formation of osteoclast-like cells in long-term human marrow cultures. J Gin Invest 78, 894-898, 1986.

7. Barton, B.E., and Mayer, R., IL-3 induces differentiation of bone marrow precursor cells to osteoclast-like cells, J Immunology 143, 321 1-3216, 1989.

8. Kurihara, N., et al., Generation of osteoclastics from isolated hemopoietic progenitor cells. Blood 1295-1302, 1989.

9. Thesingh, C.W., Formation sites and distribution of osteoclast progenitor cells during the ontogeny of the mouse. Developmental Biology 1 17, 127-134, 1986.

10. Chambers, T.J., The regulation of osteoclastic development and function, In Cell and Molecular Biology of Vertebrate Hard Tissues. Wiley, Chichester (Ciba Foundation Symposium 136), p. 92-107, 1988.

11. Hattersley, G. and Chambers., T.J., Effects of interleukin 3 and of granulocyte-macrophage and macrophage colony stimulating factors on osteoclast differentiation from mouse hemopoietic tissues. J Cell Physiol 142, 201-209, 1990.

12. Shinar, D.M., et al., The effect of hemopoietic growth factors on the generation of osteoclast-like cells in mouse bone marrow cultures. Endocrinology 126, 1728- 1735, 1990.

13. Pootrakul P., et al., Relation between erythropoiesis and bone metabolism in thalassemia. NEnglJ Me 304, 1470-1473, 1981.

14. Haskell CM., et al., Hypercalcemia in chronic granulocytic leukemia. Cancer 27, 872-879, 1971. 15. Benson, R.C., et al., (1974) Am. J. Med. 56, 821-826.

16. Mavligit, G.M., et al., ( 1971 ) N. Engl. J. Med. 285, 154- 156.

17. Knill-Jones, et al., (1970) N Engl. J. Med. 282, 704-708.

18. Hirshorn, J.E., et al., ( 1979) J. Gin. Endocrinol. Metab. 48, 217-221.

19. Neiman, R.S., and Li, H.C. (1972) Cancer 30, 942-944. 20. Tashijian, A.H., Jr., et al., (1964) J. Exp. Med. 1 19, 467-484.

21. Stewart, A.F., et al., (1980) N. Engl. J. Med. 303, 1377-1383.

22. Robertson, R.P., et al., (1976) J. Gin. Endocrinol. Metab. 43, 1330-1335.

23. Seyberth, H.W., et al., (1975) N. Engl. J. Med. 293, 1278-1281.

24. Gordan, G.S., et al., (1966) Science 151, 1226-1228. 25. Rodan, S.B., et al., J. Gin. Invest. 72, 151 1-1515.

26. Strewler, G.J., et al., ( 1983) J. Gin. Invest 71 , 769-774. 27. Ibbotson, K.J., et al., (1983) Science 221, 1292-1294.

28. Raisz, L.G., et al., (1980) Endocrinology 107, 270-273.

29. Rosengurt, E., et al., (1983) Cell 34, 265-272.

30. Coughlin, S.R., et al., (1980) Nature 306, 378-380. 31. Dayer, J.M., et al., (1981) FEBS Lett. 124, 253-256.

32. Gowen, M., et al, (1983) Nature 306, 378-380.

33. Hausmann, E., et al., (1972) Calcif. Tissue Res. 9, 272-277.

34. Raisz, L.G., et al., (1974) Science 185, 789-791.

35. Tashjian, A.H., Jr., et al., (1978) Prostaglandins 16, 221-223. 36. Delmonte, L. and R.A. Liebelt, Granulocytosis-promoting extract of mouse tumor tissue: partial purification, Science 148:521-523, 1965.

37. Boggs, D.R., et al., Kinetic studies of a tumor-induced leukemoid reaction in mice, J. Lab. Gin. Med. 89:80-92, 1977.

38. Lee, M., et al., Kinetics of tumor-induced murine neutrophilia, Blood 53(4):619-632, 1979.

39. Lee, M.Y., and D.J. Baylink, Hypercalcemia, excessive bone resorption, and neutrophilia in mice bearing a mammary carcinoma (41582), Proceedings of the Society for Experimental Biology and Medicine, 172:424-429, 1983.

40. Washington Research Foundation Non-Confidential Summary: Clones of Murine Mammary Carcinoma Cells - CESJ, Washington Resource

Foundation, Seattle, WA, 1988.

41. Lee, M.Y., et al., Production of granulocyte-stimulating and bone cell- modulating activities from a neutrophilia hypercalcemia-inducing murine mammary cancer cell line, Cancer Research 47:4059-4065, 1987. 42. Lee et al., Experimental Hematology 18 :608 (A228), July 1990.

43. Chambers, T.J, The pathobiology of the osteoclast (Review article). J Gin PathollS, 241-252, 1985.

44. Kuehler, D., et al., Comparison of tartrate-resistant acid phosphatase in normal mouse tissues to that in the reticulum cells of leukemic reticuloendotheliosis. Cancer 47, 710-714, 1981.

45. Hattersley, G., and Chambers, T.J., Generation of osteoclastic function in mouse bone marrow cultures; multinuclearity and tartrate-resistant acid phosphatase are unreliable markers for osteoclastic differentiation. Endocrinology 124, 1689-1696, 1989. 46. Takahasi, N, et al., Osteoclast-like cell formation and its regulation by osteotropic hormones in mouse bone marrow cultures. Endocrinology 122, 1373-1382, 1988.

47. Lee, M.Y., et al., (1989) Blood 74, 1 15-122. 48. Lau, K.H.W., et al., (1985) Bioche . Biophy. Act. 840, 56-68.

49. Clark, S.C, and Kamen R., The human hemopoietic colony-stimulating factors. Science 236:1229, 1987.

50. Kurihara, N., et al., Identification of committed mononuclear precursors for osteoclast-like cells formed in long term human marrow cultures. Endocrinology 126:2733, 1990.

51. Berenson, R.J., et al., Antigen CD34⁺ marrow cells engraft lethally irradiated baboons. J Gin Invest 81 :951, 1988.

52. Yoshida, H, et al., (1990) Nature 345, 442- 443.

53. Wiktor-Jedrzejczak, W., et al., (1990) Proc. Nad. Acad. Sci. USA. 87, 4828- 4832.

54. Zsebo, KM., et al., (1990) Cell 63, 195-201.

55. Thomas, E., et al., (1985) Cancer Res. 45, 5840-5844.

56. Sakai, N., et al., (1984) Gann 75, 355-361.

57. Lee, MY., and Lottsfeldt, J.L., (1984) Blood 64, 499-506. 58. Liu, C, et al., (1987) Histochemistry 86, 559-565.

59. Hattersley, G, & Chambers, T.J., (1989) J. Cell. Physiol. 140, 478-482.

60. Mochizuki, D.Y., et al., (1986) J. Immunol. 136, 3706- 3709.

61. Abrams, J.S., and Pearce, M.K., (1988) J. Immunol. 140, 131-137.

62. Laemmli, U.K. (1970) Nature 277, 680-685. 63. Huynh, T.V., et al., Constructing and screening cDNA libraries in lambda gtlO and lambda gtl l. In DNA cloning: A practical approach (ed. D.M.

Glover) vol. 1, pp. 49, IRL Press, Oxford. 64. Sambrook, J., et al., Molecular Cloning: A laboratory manual. Cold spring

Harbor Laboratory, Cold Spring Harbor, N.Y., 2nd edition, 1989. 65. Nagata, S., et al., Molecular cloning and expression of cDNA for human granulocyte colony-stimulating factor. Nature 319, 415-418, 1986.

66. Metcalf, D., The Molecular Control of Blood Cells, Harvard University Press, Cambridge, Massachusetts, 1989.

67. Osborne, W.R.A., and Miller, A.D., Design of vectors for efficient expression of human purine nucleoside phosphorylase in skin fibroblasts from enzyme-deficient humans. Proc Nat I Acad Sci 85, 6851-6855, 1988. 68. Hock, R.A., et al., Expression of human adenosine deaminase from various strong promoters after gene transfer into human hemopoietic cell lines. Blood 74, 876-881, 1989.

69. Kaleko, M., et al., Expression of human adenosine deaminase in mice after transplantation of genetically-modified bone marrow. Blood 75, 1733-1744,

1990.

70. Valentino, K., et al, (eds) 1987. In situ hybridization in neurobiology. Application to Neurobiology. Oxford University Press.

71. Koury, S.T., et al., Localization of erythropoietin synthesizing cells in murin kidneys by in situ hybridization. Blood 71 , 524-527, 1988.

72. Tsuchiya, M., et al., Characterization of recombinant human granulocyte- colony-stimulation factor produced in mouse cells. EMBO 6, 611-616, 1987.

73. Fukunaga, R., et al., Constitutive production of human interferons by mouse cells with bovine papillomavirus as a vector. Proc Nail Acad Sci 81, 5086- 5090, 1984.

74. McCracken, CH., et al., Concurrent activation of granulocytes and osteoclasts in busulfan-suppressed bone marrow in response to transplantation of a mammary carcinoma in mice. Exp Hemalol 16, 285-289, 1988.

75. Marks, S.C, Congenital osteopetrotic mutations as probes of the origin, structure, and function of osteoclasts. Gin Orthoped Related Research, 189,

239-263, 1984.

76. Wiktor-Jedrzejczak, W„ et al., Hematological characterization of congenital osteopetrosis in op/op mouse. Possible mechanism for abnormal macrophage differentiation. ExpMed 156, 1516-1527, 1982. 77. Shadduck, R.K., and Metcalf, D., Preparation and neutralization characteristics of an anti-CSF antibody. JCellPhysiol 86:247-252, 1975.

78. Goding, J.W., Monoclonal Antibodies: In Principle and Practice. Academic Press, pp. 250-258, 1983.

79. Ihle, J.N., et al., Phenotypic characteristics of cell lines requiring interleukin 3 for growth. J. Immunology, 129:1377, 1982.

80. Dexter, T.M., et al., Growth of factor-dependent hemopoietic precursor cell lines. J. Exp. Med., 152:1036, 1980.

81. Antibodies: A Laboratory Manual, Harlow and Lane (eds.), Cold Spring Harbor Laboratory Press, 1988. 82. William D. Huse et al., "Generation of a Large Combinational Library of the Immunoglobulin Repertoire in Phage Lambda," Science 246: 1275-1281, December 1989; see also L. Sastry et al., "Cloning of the Immunological Repertoire in Escherichia coli for Generation of Monoclonal Catalytic Antibodies: Construction of a Heavy Chain Variable Region-Specific cDNA Library," Proc. Natl. Acad. Sci. USA 86:5728-5732, August 1989; see also Michelle Alting-Mees et al., Monoclonal Antibody Expression Libraries: A rapid Alternative to Hybridomas," Strategies in Molecular Biology 3: 1-9, January 1990.

83. Stanley, E.R., Proc. Natl. Acad. Sci., 76:2969, 1979.

84. Stanley, E.R., Methods in En∑ymology, 116: 1985. 85. Harlow, E., and Lane, D., Antibodies, a laboratory manual, Cold Spring Harbor Laboratory, 359, 1988.

86. Fukunaga, R., et al., Expression cloning of a receptor for murine granulocyte colony-stimulating factor. Cell 67:341, 1990.

87. Holmes, W.E., et al., Structure and functional expression of a human interleukin-8 receptor. Science 253: 1278-1280, 1991.

88. U.S. Patent No. 4,973,666.

89. Lee, M.Y., et al., Isolation of a murine osteoclast colony-stimulating factor. Proc. Natl. Acad. Sci. USA 88:8500-8509, October, 1991.

90. Mundy, G.R., Cytokines and local factors which affect osteoclast function. Intl. J. Cell Cloning 10:215-222, 1992.

91. Tybulewicz, V.L.J., et al., Animal model of Gaucher's disease from targeted disruption of the mouse glucocerebrosidase gene. Nature 357:407-410, 1992.

92. te Riele, H., et al., Highly efficient gene targeting in embryonic stem cells through homologus recombination with isogenic DNA constructs. Proc. Natl. Acad. Sci. 89:5128-5132, 1992.

While the preferred embodiment of the invention has been illustrated and described, it will be appreciated that various changes can be made therein without departing from the spirit and scope of the invention.

Claims

The embodiments of the invention in which an exclusive property or privilege is claimed are defined as follows:

1. A substantially purified O-CSF protein having an apparent molecular weight of approximately 15,000-25,000 daltons as determined by SDS-PAGE, capable of stimulating growth of TRAP-positive osteoclast progenitors in bone marrow cell cultures, and which is not neutralized by anti-IL-3 or anti-GM-CSF antibodies.

2. The O-CSF protein of claim 1, having an apparent molecular weight of approximately 20,000 daltons as determined by SDS-PAGE electrophoresis.

3. The O-CSF protein of claim 1, produced by CESJ-3 cells (ATCC No. CRL 10886).

4. The O-CSF protein of claim 1, capable of stimulating growth of TRAP-positive osteoclast progenitors exhibiting bone resorption activity.

5. The O-CSF protein of claim 1, having an osteoclast progenitor stimulating activity of at least 660,000 units per milligram of protein.

6. An immunologic binding partner capable of binding the O-CSF protein of claim 1.

7. A reagent kit comprising an O-CSF protein in a first container and the immunologic binding partner of claim 7 in a second container.

8. A method of stimulating bone resoφtion, comprising administering to a mammalian host the O-CSF protein of claim 1.

9. An isolated DNA molecule capable of hybridizing under stringent conditions to the nucleotide sequence shown in FIGURE 15.

10. A recombinant expression vector comprising at least one strand of the DNA molecule of claim 9 operably linked to suitable control sequences.

11. A cell transfected with the recombinant expression vector of claim 10.

12. The expression product of the cell of claim 11, capable of stimulating growth of TRAP-positive osteoclast progenitors in bone marrow cell cultures.

13. The expression product of claim 12 consisting essentially of the amino acid sequence shown in FIGURE 15.