In vitro selection and characterization of Bcl-X(L)-binding proteins from a mix of tissue-specific mRNA display libraries.

The covalent coupling of an mRNA to the protein that it encodes (mRNA display) provides a powerful tool for analysis of protein function in the post-genomic era. This coupling allows the selective enrichment of individual members from libraries of displayed proteins and the subsequent regeneration of an enriched library using the RNA moiety. Tissue-specific libraries from poly(A)(+) mRNA were prepared by priming first and second strand cDNA synthesis with oligonucleotides containing nine random 3' nucleotides, the fixed regions of which encoded the requisite sequences for formation of mRNA display constructs and a library-specific sequence tag. Starting with a pool of uniquely tagged libraries from different tissues, an iterative selection was performed for binding partners of the anti-apoptotic protein Bcl-X(L). After four rounds of selection, the pool was deconvoluted by polymerase chain reaction amplification with library-specific primers. Subsequent clonal sequence analysis revealed the selection of three members of the Bcl-2 family known to bind to Bcl-X(L). In addition, several proteins not previously demonstrated to interact with Bcl-X(L) were identified. The relative binding affinities of individual selected peptides were determined, as was their susceptibility to competition with a BH3 domain peptide. Based on these data, a putative BH3 domain was identified in most peptides.

PROfusion: genetically tagged proteins for functional proteomics and beyond.

As a new enabling technology in functional genomics, Phylos, Inc. has pioneered PROfusiontrade mark technology, whereby proteins are covalently tagged with their own genetic information (mRNA). Using this technology, both synthetic and natural libraries (representing the repertoire of proteins naturally expressed in a given cell type or tissue source) have been constructed. Due to the in vitro nature of library construction, Phylos libraries are the largest described to date, up to 10(14) in size. From a library of such molecules, one can select for a protein function of choice with the benefit that the genetic material is linked to the protein for subsequent PCR amplification, enrichment, and, ultimately, identification.

Dendritic cells and MPIF-1: chemotactic activity and inhibition of endogenous chemokine production by IFN-gamma and CD40 ligation.

We have examined the biological activity of the CC chemokine myeloid progenitor inhibitory factor 1 (MPIF-1) on human dendritic cells. MPIF-1 has chemotactic activity on dendritic cells derived from either peripheral blood monocytes or cord blood CD34+ progenitors. However, chemokine treatment did not induce further cell activation or maturation. In addition, MPIF-1 is constitutively released by monocyte-derived dendritic cells but not macrophages or monocytes (resting or stimulated). The proinflammatory stimuli lipopolysaccharide and tumor necrosis factor alpha, which induced the release of monocyte chemotactic protein-1, macrophage inflammatory protein-1alpha (MIP-1alpha), MIP-1beta, and interleukin-8, did not affect MPIF-1 release. In contrast, CD40 ligation and interferon-gamma treatment, while stimulating the production of the other chemokines, caused a pronounced reduction of MPIF-1 transcript and protein release. Thus, in dendritic cells the regulation of the production and release of MPIF-1 is distinct in comparison to other CC and CXC chemokines.

STRL22 is a receptor for the CC chemokine MIP-3alpha.

STRL22 is a human seven transmembrane domain orphan receptor related to known chemokine receptors and expressed in peripheral blood lymphocytes, tumor infiltrating lymphocytes and lymphoid tissues. MIP-3alpha/LARC/Exodus is a CC chemokine that is chemotactic for lymphocytes and that is expressed in activated cells, including monocytes, T cells, endothelial cells, and fibroblasts, and in liver, lung, and some lymphoid tissues. We report here that STRL22-transfected human embryonic kidney 293 cells demonstrated specific binding for MIP-3alpha and that MIP-3alpha, but no other chemokines, produced a calcium flux in the STRL22-transfected cells. We show that MIP-3alpha, unlike other chemokines, produced a calcium flux in freshly-isolated peripheral blood lymphocytes and we show that MIP-3alpha also produced a signal in tumor infiltrating lymphocytes that express STRL22. Since STRL22 is the sixth functional CC chemokine receptor identified, it should be re-named CCR6.

Molecular and functional characterization of two novel human C-C chemokines as inhibitors of two distinct classes of myeloid progenitors.

Two novel human beta-chemokines, Ck beta-8 or myeloid progenitor inhibitory factor 1 (MPIF-1), and Ck beta-6 or MPIF-2, were discovered as part of a large scale cDNA sequencing effort. The MPIF-1 and MPIF-2 cDNAs were isolated from aortic endothelium and activated monocyte libraries, respectively. Both of the cDNAs were cloned into a baculovirus vector and expressed in insect cells. The mature recombinant MPIF-1 protein consists of 99 amino acids and is most homologous to macrophage inflammatory protein (MIP)-1alpha, showing 51% identity. It displays chemotactic activity on resting T lymphocytes and monocytes, a minimal but significant activity on neutrophils, and is negative on activated T lymphocytes. MPIF-1 is also a potent suppressor of bone marrow low proliferative potential colony-forming cells, a committed progenitor that gives rise to granulocyte and monocyte lineages. The mature recombinant MPIF-2 has 93 amino acid residues and shows 39 and 42% identity with monocyte chemoattractant protein (MCP)-3 and MIP-1alpha, respectively. It displays chemotactic activity on resting T lymphocytes, a minimal activity on neutrophils, and is negative on monocytes and activated T lymphocytes. On eosinophils, MPIF-2 produces a transient rise of cytosolic Ca2+ and uses the receptor for eotaxin and MCP-4. In hematopoietic assays, MPIF-2 strongly suppressed the colony formation by the high proliferative potential colony-forming cell (HPP-CFC), which represents a multipotential hematopoietic progenitor.

The Evi-1 zinc finger myeloid transforming protein binds to genomic fragments containing (GATA)n sequences.

The EVI1 gene is activated by chromosomal translocations and inversions in approximately 5% of human acute myeloid leukemia (AML) and by retroviral insertion in approximately 20% of murine myeloid leukemias. EVI1 encodes a nuclear DNA-binding protein having 10 zinc finger motifs in two noncontiguous domains consisting of an amino-terminal domain of seven fingers and a carboxyl domain containing three fingers. To evaluate the sequence specificity of Evi-1 binding and potentially identify genomic targets, whole-genome PCR was utilized to isolate multiple Sau3A fragments which specifically bind to the amino-terminal zinc finger domain. The majority of these clones represented single copy sequences and virtually all contained variable numbers of repeats of the GATA motif, the target sequence for the erythroid-specific transcription factor GATA-1. GST/Evi-1 fusion proteins containing the amino-terminal domain of zinc fingers bound the GATA motif in these clones as well as to those present in the human gamma-globin promoter, similar to the binding of purified GATA-1 protein. By obtaining corresponding large genomic clones for eight of these fragments, transcription units were found associated with two. One corresponded to the glyceraldehyde-3-phosphate dehydrogenase gene and its expression was not affected by Evi-1. The second is a novel gene whose expression is repressed in murine myeloid cell lines that express Evi-1.

Interleukin-3 signals through multiple isoforms of Stat5.

The interleukin (IL)-3 family of cytokines mediates its numerous effects on myeloid growth and maturation by binding a family of related receptors. It has been shown recently that IL-3 induces the activation of two distinct cytoplasmic signal transducing factors (STFs) that are likely to mediate the induction of immediate early genes. In immature myeloid cells, IL-3 activates STF-IL-3a, which comprises two tyrosine-phosphorylated DNA binding proteins of 77 and 80 kDa. In mature myeloid cells, IL-3 and granulocyte-macrophage colony-stimulating factor activate STF-IL-3b, which consists of a 94 and 96 kDa tyrosine-phosphorylated DNA binding protein. Peptide sequence data obtained from the purified 77 and 80 kDa proteins (p77 and p80) indicate that they are closely related but are encoded by distinct genes. Both peptide and nucleotide sequence data demonstrate that these two proteins are the murine homologs of ovine mammary gland factor (MGF)/Stat5. The peptide data also indicate that p77 and p80 are phosphorylated on tyrosine 699, a position analogous to the tyrosine that is phosphorylated in Stat1 and Stat2 in response to interferon. Additionally, antiserum raised against bacterially expressed p77/p80 recognizes the 94 and 96 kDa protein components of STF-IL-3b, suggesting that these may be additional isoforms of Stat5. These studies indicate that the IL-3 family of ligands is able to activate multiple isoforms of the signal transducing protein Stat5.

Stat4, a novel gamma interferon activation site-binding protein expressed in early myeloid differentiation.

Interferon regulation of gene expression is dependent on the tyrosine phosphorylation and activation of the DNA-binding activity of two related proteins of 91 kDa (STAT1) and/or 113 kDa (STAT2). Recent studies have suggested that these proteins are substrates of Janus kinases and that proteins related in STAT1 are involved in a number of signalling pathways, including those activated in myeloid cells by erythropoietin and interleukin-3 (IL-3). To clone STAT-related proteins from myeloid cells, degenerate oligonucleotides were used in PCRs to identify novel family members expressed in myeloid cells. This approach allowed the identification and cloning of the Stat4 gene, which is 52% identical to STAT1. Unlike STAT1, Stat4 expression is restricted but includes myeloid cells and spermatogonia. In the erythroid lineage, Stat4 expression is differentially regulated during differentiation. Functionally, Stat4 has the properties of other STAT family genes. In particular, cotransfection of expression constructs for Stat4 and Jak1 and Jak2 results in the tyrosine phosphorylation of Stat4 and the acquisition of the ability to bind to the gamma interferon (IFN-gamma)-activated sequence of the interferon regulatory factor 1 (IRF-1) gene. Stat4 is located on mouse chromosome 1 and is tightly linked to the Stat1 gene, suggesting that the genes arose by gene duplication. Unlike Stat1, neither IFN-alpha nor IFN-gamma activates Stat4. Nor is Stat4 activated in myeloid cells by a number of cytokines, including erythropoietin, IL-3, granulocyte colony-stimulating factor, stem cell factor, colon-stimulating factor 1, hepatocyte growth factor, IL-2, IL-4, and IL-6.

The carboxyl domain of zinc fingers of the Evi-1 myeloid transforming gene binds a consensus sequence of GAAGATGAG.

Inappropriate expression of the Evi-1 zinc finger gene is associated with myeloid leukemia and myelodysplastic syndromes in mice and humans and has been hypothesized to contribute to pathology by blocking myeloid differentiation. Evi-1 contains two domains of zinc fingers, an amino-terminal domain of seven fingers and a carboxyl domain of three fingers. The first domain binds a consensus sequence of GA(C/T)AAGATAAGATAA in binding and amplication reactions or GATA repeat containing regions of genomic DNA. The experiments described here, establish a consensus sequence for the carboxyl domain of zinc fingers consisting of GAAGATGAG. Unlike the first domain, the consensus sequence established for the carboxyl domain is identical to that which would be predicted by the current rules relating to C2H2 zinc fingers and DNA recognition. Substitution of sequences in finger 8 with those in finger 9, demonstrate that the individual fingers bind the predicted region of the consensus sequence. In an attempt to engineer binding of constructs containing the carboxyl domain, a variety of mutations were made in the middle finger that would be predicted to change the consensus sequence in specific ways. Remarkably, most of the mutations were deleterious and destroyed specific DNA binding. Although Evi-1 contains potential transcriptional activation domains, it was not able to activate gene transcription from CAT constructs containing the consensus sequence.

Loss of erythropoietin responsiveness in erythroid progenitors due to expression of the Evi-1 myeloid-transforming gene.

Inappropriate expression of the Evi-1 zinc-finger gene in hematopoietic cells has been associated with acute myelogenous leukemia and myelodysplastic syndromes in murine models and in humans. Consistent with this, previous studies have shown that aberrant expression of the Evi-1 gene in a myeloid progenitor cell line blocks granulocytic differentiation. Here we demonstrate that the aberrant expression of the Evi-1 gene impairs the normal response of erythroid cells or bone-marrow progenitors to erythropoietin. Erythroid differentiation has been shown to require the GATA-1 transcription factor that binds to a sequence contained within the consensus binding sequence identified for Evi-1. In the studies presented here we also show that Evi-1 can repress GATA-1-dependent transactivation in transient chloramphenicol acetyltransferase assays. Together the data support the hypothesis that inappropriate expression of the Evi-1 gene blocks erythropoiesis by repressing the transcription of a subset of GATA-1 target genes.

Four of the seven zinc fingers of the Evi-1 myeloid-transforming gene are required for sequence-specific binding to GA(C/T)AAGA(T/C)AAGATAA.

Expression of the Evi-1 gene is activated in murine myeloid leukemias by retroviral insertions and in human acute myelogenous leukemia by translocations and inversions involving chromosome band 3q26 where the gene resides. Aberrant expression of the Evi-1 gene has been shown to interfere with myeloid differentiation, which is proposed to be the basis for its role in leukemias. The Evi-1 gene encodes a 145-kDa DNA-binding protein containing two domains of seven and three Cys2-His2 zinc fingers. Previous studies identified a portion of the consensus DNA-binding sequence for the first domain of zinc fingers. The experiments presented here extend these studies and demonstrate that the first domain recognizes a consensus of 15 nucleotides consisting of GA(C/T)AAGA(T/C)AAGATAA. The first three fingers of the first domain do not detectably bind DNA but contribute to the binding by conferring a relative specificity for GACAA verses GATAA in the first position. The first three fingers also contribute to optimal binding of the 15-nucleotide consensus sequence.

Phenotypically diverse mouse thymic stromal cell lines which induce proliferation and differentiation of hematopoietic cells.

The heterogeneity of the thymic stroma has made careful characterization of particular thymic stromal cell types difficult. To this end, we have derived a panel of cloned thymic stromal cell lines from simian virus 40 T antigen (SV40-T antigen) transgenic mice. Based on their analysis with monoclonal antibodies that distinguish among subsets of thymic stroma cells, and on the morphology and ultrastructural features of the different clones, we suggest that our panel includes representatives of the thymic subcapsular cortex or thymic nurse cells (427.1), the deep cortex or cortical reticular cells (1308.1) and the medulla including medullary interdigitating (IDC)-like cells (6.1.1) and medullary epithelial cells (6.1.7). A fifth cell type of undesignated but apparent medullary origin (6.1.11) was also isolated. All of the cell lines constitutively express the SV40 T antigen transgene and the class I antigens of the major histocompatibility complex (MHC), and they can be induced to express MHC class II antigens upon stimulation with recombinant interferon-gamma (IFN-gamma). These cell lines elaborate a factor(s) that induces the proliferation of cells from the fetal liver and bone marrow, but not from the neonatal thymus. A factor(s) elaborated by the 1308.1 cell line also induces the proliferation of fetal thymocytes in the absence of mitogens, phorbol esters or calcium ionophore which is augmented with the addition of recombinant interleukin-2 (IL-2). Analysis by reverse transcription polymerase chain reaction with primers for some mouse cytokines reveals that each of these cell lines contain granulocyte-macrophage colony-stimulating factor (GM-CSF) transcripts and that 1308.1, 6.1.1 and 6.1.7 produce IL-6 mRNA. Cell lines 1308.1 and 6.1.1 also produce IL-7; 6.1.1 produces IL-1 beta and tumor necrosis factor (TNF)-alpha while the 427.1 cell line produces IL-5 and IFN-gamma mRNA. None of the cell lines tested express the IL-2 receptor, IL-2, IL-3, IL-4, TNF-beta or macrophage inflammatory proteins mRNA. Conditioned medium (CM) from 1308.1 and 6.1.11 induced differentiation of cells purified from the mouse fetal liver into granulocytes; 1308.1 CM also induced differentiation of the mouse hematopoietic stem cell line 32DCl3(G) suggesting that the CM contains granulocyte (G)-CSF activity. Each cell line produces GM-CSF but the greatest activity is associated with 1308.1 and 6.1.11 CM. The availability of these well-characterized, functional, cloned thymic stromal cells will allow a more detailed analysis of the role of each cell type in both myeloid and T cell development.

Inhibition of myeloid differentiation by the helix-loop-helix protein Id.

Id is a helix-loop-helix (HLH) protein that represses activity of several basic helix-loop-helix (bHLH) proteins involved in cell type--specific transcription and cell lineage commitment. The myeloid precursor cell line 32DC13(G) expressed Id messenger RNA, which was transiently decreased when cells were induced to terminally differentiate with granulocyte--colony-stimulating factor. Concomitant with the decrease of Id messenger RNA was the appearance in nuclear extracts of DNA binding proteins that recognized a canonical E-box motif, a DNA binding site for some bHLH proteins. Constitutive expression of an Id complementary DNA in 32DC13(G) cells blocked their ability to differentiate and to induce E-box-binding activity. These results suggest that Id and, hence, bHLH proteins function in the process of myeloid differentiation.

The immediate early gene response to a differentiative stimulus is disrupted by the v-abl and v-ras oncogenes.

The immediate early gene activation in response to a differentiative stimulus was investigated in the hematopoietic progenitor cell line 32DC13(G). A transient and coordinated increase in mRNA levels for c-fos, c-jun, jun-B, TIS-7/PC-4, TIS-8/egr-1, and TIS-11 occurred during the first 2 h of treatment with granulocyte colony stimulating factor (G-CSF), which ultimately induces the 32DC13(G) cells to terminally differentiate into neutrophilic granulocytes. This pattern of mRNA induction was disrupted by v-abl and v-ras, two oncogenes known to interfere with G-CSF-induced differentiation of 32DC13(G) cells. Induction of the mRNAs for c-jun and TIS-7/PC-4 was blocked by the presence of v-abl, whereas v-ras caused constitutive expression of c-fos mRNA and blocked the c-jun, jun-B and TIS-7/PC-4 mRNA response. Release of the differentiation block in the ras-transformed 32DC13(G) cells by co-treatment with retinoic acid and G-CSF partially restored the normal c-fos and c-jun mRNA induction pattern, suggesting that the proper activation of these genes may be important for myeloid differentiation.

Structure of the murine lactotransferrin gene is similar to the structure of other transferrin-encoding genes and shares a putative regulatory region with the murine myeloperoxidase gene.

The structure and nucleotide sequence of the murine lactotransferrin-encoding gene (LTF) deduced partly by direct sequencing of genomic clones in the lambda phage vector and partly by enzymatic amplification of genomic DNA segments primed with the oligodeoxyribonucleotide primers homologous to the cDNA sequence. The lambda phage clones contained the 5' half of the gene corresponding to the first eight exons and an incomplete ninth exon interrupted by eight introns. Genomic clones corresponding to the 3' half of the LTF gene could not be obtained on repeated attempts from two different mouse genomic libraries, suggesting the possible presence of unclonable sequences in this part of the gene. Hence, PCR was used to clone the rest of the gene. Four out of the presumed eight remaining introns were cloned along with the flanking exons using PCR. Comparison of the structure of the LTF gene with those of the two other known transferrin-encoding genes, human serum transferrin-encoding gene and chicken ovotransferrin-encoding gene reveals that all three genes have a very similar intron-exon distribution pattern. The hypothesis that the present-day transferrin-encoding genes have originated from duplication of a common ancestral gene is confirmed here at the gene level. An interesting finding is the identification of a region of shared nucleotides between the 5' flanking regions of the murine LTF and myeloperoxidase-encoding genes, the two genes expressed specifically in neutrophilic granulocytes.

A neutralizing monoclonal antibody binds to an epitope near the amino terminus of murine granulocyte-macrophage colony-stimulating factor.

A rat anti-murine granulocyte-macrophage colony-stimulating factor (mGM-CSF) monoclonal antibody, A2, that neutralizes bioactivity in vitro was isolated. The binding epitope recognized by this antibody was identified using human-murine hybrid GM-CSF proteins. A2 was unable to immunoprecipitate a hybrid (hm7) protein containing the human GM-CSF sequence for the first 11 amino terminal amino acids, and the mGM-CSF sequence for amino acids 12-124. In contrast, A2 did recognize a hybrid which substitutes human GM-CSF amino acids 23-36 in the murine sequence. These data suggest that this neutralizing antibody recognizes an epitope at the amino terminus of mGM-CSF. Because hm7 did maintain in vitro bioactivity, it is probable that the epitope recognized by the neutralizing antibody is not itself part of the receptor-binding domain of mGM-CSF; rather, it is likely that neutralization occurs as a result of antibody binding near the receptor-binding site, with steric inhibition of mGM-CSF binding to its receptor. Interestingly, monoclonal antibody A2 does not recognize mGM-CSF glycosylation species corresponding to predicted maximal O-glycosylation variants. The presence of O-glycosylation sites within the antibody-binding epitope was confirmed using site-directed mutagenesis. Potential O-glycosylation sites in native mGM-CSF were removed by introducing conservative amino acid substitutions, and expected molecular weight reductions were obtained. These findings are consistent with previous reports that suggest the importance of the integrity of residues near the amino terminus to GM-CSF bioactivity.

Growth factor-dependent differentiation along the myeloid and lymphoid lineages in an immature acute T lymphocytic leukemia.

Bone marrow cells from a child with an immature (CD2+, CD5+, CD7+) acute T lymphocytic leukemia (T-ALL) were cultured in the presence and absence of human rIL-2, IL-3, or granulocyte-macrophage (GM)-CSF. Cells cultured without growth factors failed to divide and those initiated in the presence of IL-2 or GM-CSF underwent maturation and terminal T lymphoid or myelomonocytic differentiation, respectively. In contrast, a permanent growth factor-dependent cell line, designated TALL-103/3, was established upon culture in IL-3. The TALL-103/3 cells gradually lost the T cell-specific markers and acquired a myeloid phenotype (CD15+, CD33+). Switching of the IL-3-dependent cells at an early passage to medium containing only human rIL-2 resulted in the establishment of a subline, named TALL-103/2, with a T lymphoid phenotype (CD3+, CD8+, TCR-gamma delta +, CD7+). The TALL-103/2 cells strictly require IL-2 for growth, are irreversibly committed to the lymphoid lineage, and cannot survive in the presence of any other hemopoietic growth factor tested so far. In contrast, the IL-3-dependent TALL-103/3 cells could be adapted to grow in synthetic (serum-free) medium also in the presence of either GM-CSF or IL-5, in which they retain a myeloid phenotype. Interestingly, after 18 mo in culture in IL-3, the TALL-103/3 cells can still be phenotypically converted to the lymphoid lineage upon addition of IL-2, thus maintaining its bipotentiality. Despite the marked phenotypic differences, the TALL-103/2 and TALL-103/3 cell lines show the same karyotypes with multiple abnormalities present in the primary malignant clone and have identical rearrangements of the TCR-gamma and -delta loci, thus confirming their derivation from a common precursor cell. Together, these findings indicate that the phenotype of immature T-ALL cells can be drastically modified by the presence of specific hemopoietic growth factors in the environment, leading to either lymphoid or myeloid lineage commitment while leaving their karyotype and genotype intact.

Induction of the granulocyte-macrophage colony-stimulating factor (CSF) receptor by granulocyte CSF increases the differentiative options of a murine hematopoietic progenitor cell.

32DC13(G) is an interleukin-3-dependent murine hematopoietic precursor cell line which differentiates into neutrophilic granulocytes upon exposure to granulocyte colony-stimulating factor (G-CSF) but ceases to proliferate and dies when exposed to granulocyte-macrophage (GM)-CSF. Surface receptors for GM-CSF are undetectable on 32DC13(G) cells but can be induced by priming the cells with G-CSF. Exposure of the G-CSF-primed cells to GM-CSF then results in the generation of monocytes as well as granulocytes. The acquired competence to respond to GM-CSF remains irreversibly encoded in the primed cells, although the GM-CSF receptor can be down regulated by interleukin-3. This phenomenon suggests a mechanism by which hematopoietic precursors may obtain additional receptors, thereby increasing their differentiative potential.

A member of the ras gene superfamily is expressed specifically in T, B and myeloid hemopoietic cells.

A 1.26 kb murine cDNA having 31% homology with human ras and 55% homology with human rho proteins was isolated using an oligonucleotide probe homologous to catalytic subdomain of a tyrosine kinase subfamily. Northern blot analysis indicates that the expression of the murine gene is restricted to the cells of hemopoietic lineages and the mRNA levels increase with the terminal differentiation of hemopoietic cells into granulocytes.

The parathyroid hormone-related protein stimulates human osteoblast-like cells to secrete a 9,000 dalton bone-resorbing protein.

The mechanism by which parathyroid hormone-related protein (PTH-RP) stimulates bone resorption is not known. Like certain other resorbing agents it may act to release bone-resorbing cytokines from the osteoblast. To examine this hypothesis, we used serum-free conditioned media (CM) from SAOS II cells incubated with 10(-8) M h(1-74) PTH-RP for 48 h. Treated CM contained substantially more bone-resorbing activity (BRA) in the fetal-rat long-bone assay than CM from untreated cells (2.17 +/- 0.21 vs 1.38 +/- 0.16 fold stimulation over basal [f]; p less than 0.05]. After centrifugation and dialysis, 1 liter of treated CM contained a total BRA of 7102 ngeq b(1-34) PTH with a specific activity (SA) of 447 ngeq b(1-34) PTH/mg protein. Treated CM did not stimulate the ROS assay and the cytokines PGE2, TGF-alpha, EGF, GM-CSF and IL-1 were present in low concentrations. The BRA was heat sensitive. Ultrafiltration revealed that 97% of the BRA was in a 3-30 kD fraction. Further purification was achieved by sequential reverse phase HPLC and size exclusion-HPLC (SE-HPLC). A single fraction containing BRA from SE-HPLC was purified 277-fold to a SA of 123,810 ngeq b(1-34) PTH/mg protein and had an apparent MW of 9 kD. SDS-PAGE revealed 4 bands in this SE-HPLC fraction with 1 band at 9 kD unique to that fraction. PTH-RP may cause bone resorption in part by stimulating the release of a 9 kD protein from osteoblasts which is responsible for activating osteoclasts.