Pharmacological inhibition of PAR2 with the pepducin P2pal-18S protects mice against acute experimental biliary pancreatitis.

Pancreatic acinar cells express proteinase-activated receptor-2 (PAR2) that is activated by trypsin-like serine proteases and has been shown to exert model-specific effects on the severity of experimental pancreatitis, i.e., PAR2(-/-) mice are protected from experimental acute biliary pancreatitis but develop more severe secretagogue-induced pancreatitis. P2pal-18S is a novel pepducin lipopeptide that targets and inhibits PAR2. In studies monitoring PAR2-stimulated intracellular Ca(2+) concentration changes, we show that P2pal-18S is a full PAR2 inhibitor in acinar cells. Our in vivo studies show that P2pal-18S significantly reduces the severity of experimental biliary pancreatitis induced by retrograde intraductal bile acid infusion, which mimics injury induced by endoscopic retrograde cholangiopancreatography (ERCP). This reduction in pancreatitis severity is observed when the pepducin is given before or 2 h after bile acid infusion but not when it is given 5 h after bile acid infusion. Conversely, P2pal-18S increases the severity of secretagogue-induced pancreatitis. In vitro studies indicate that P2pal-18S protects acinar cells against bile acid-induced injury/death, but it does not alter bile acid-induced intracellular zymogen activation. These studies are the first to report the effects of an effective PAR2 pharmacological inhibitor on pancreatic acinar cells and on the severity of experimental pancreatitis. They raise the possibility that a pepducin such as P2pal-18S might prove useful in the clinical management of patients at risk for developing severe biliary pancreatitis such as occurs following ERCP.

A mouse model of ethanol dependent pancreatic fibrosis.

BACKGROUND AND AIMS: The majority of patients with chronic pancreatitis are alcoholics. Our goal was to develop a mouse model of alcohol dependent chronic pancreatitis. METHODS: Mice were fed either the non-alcohol containing Lieber-DeCarli diet or the Lieber-DeCarli diet containing 24% of calories as ethanol. After eight weeks and while on their respective diets, mice were subjected to repeated episodes of acute pancreatitis elicited by administration of caerulein. They were sacrificed 1, 3, and 5 weeks after the last dose of caerulein. Pancreatic morphology and collagen deposition were evaluated in samples stained with haematoxylin-eosin and Sirius red. Collagen content was quantitated by measuring OH-proline. Gene expression was determined by quantitative polymerase chain reaction. RESULTS: Both groups of mice gained weight at the same rate. Those receiving the alcohol containing diet had serum alcohol levels of approximately 100 mM. No histological or gene expression differences were found in mice that were not subjected to acute pancreatitis, regardless of their diet. Necrosis, Sirius red staining, OH-proline content, and expression of alpha-1 collagen I, alpha-smooth muscle actin, transforming growth factor beta1, and tissue inhibitor of metalloproteinase 1 were all increased in mice fed the alcohol containing diet and given caerulein compared with those fed the control diet and given caerulein. Matrix metalloproteinase 9 expression was transiently decreased in mice fed ethanol and given caerulein compared with the group given caerulein but not fed ethanol. CONCLUSION: We have developed a mouse model of alcohol dependent chronic pancreatic fibrosis. This mouse model may be useful in studies examining the effects of genetic manipulation on chronic pancreatitis.

Secretin differentially sensitizes rat pancreatic acini to the effects of supramaximal stimulation with caerulein.

Supramaximal stimulation of the rat pancreas with CCK, or its analog caerulein, triggers acute pancreatitis and a number of pancreatitis-associated acinar cell changes including intracellular activation of digestive enzyme zymogens and acinar cell injury. It is generally believed that some of these various acinar cell responses to supramaximal secretagogue stimulation are interrelated and interdependent. In a recent report, Lu et al. showed that secretin, by causing generation of cAMP and activation of PKA, sensitizes acinar cells to secretagogue-induced zymogen activation, and, as a result, submaximally stimulating concentrations of caerulein can, in the presence of secretin, trigger intracellular zymogen activation. We found that secretin also sensitizes acinar cells to secretagogue-induced cell injury and to subapical F-actin redistribution but that it did not alter the caerulein concentration dependence of other pancreatitis-associated changes such as the induction of a peak plateau intracellular [Ca(2+)] rise, inhibition of secretion, activation of ERK1/2, and activation of NF-kappaB. The finding that secretin sensitizes acinar cells to both intracellular zymogen activation and cell injury is consistent with the concept that these two early events in pancreatitis are closely interrelated and, possibly, interdependent. On the other hand, the finding that, in the presence of secretin, caerulein can trigger subapical F-actin redistribution without inhibiting secretion challenges the concept that disruption of the subapical F-actin web is causally related to high-dose secretagogue-induced inhibition of secretion in pancreatic acinar cells.

Selective gene expression in brain microglia mediated via adeno-associated virus type 2 and type 5 vectors.

Microglia represent a crucial cell population in the central nervous system, participating in the regulation and surveillance of physiological processes as well as playing key roles in the etiologies of several major brain disorders. The ability to target gene transfer vehicles selectively to microglia would provide a powerful new approach to investigations of mechanisms regulating brain pathologies, as well as enable the development of novel therapeutic strategies. In this study, we evaluate the feasibility of specifically and efficiently targeting microglia relative to other brain cells, using vectors based on two different serotypes of adeno-associated virus (AAV) carrying cell-type-specific transcriptional elements to regulate gene expression. Among a set of promoter choices examined, an element derived from the gene for the murine macrophage marker F4/80 was the most discriminating for microglia. Gene expression from vectors controlled by this element was highly selective for microglia, both in vitro and in vivo. To our knowledge, this is the first demonstration of selective expression of transferred genes in microglia using AAV-derived vectors, as well as the first utilization of recombinant AAV-5 vectors in any macrophage lineage. These results provide strong encouragement for the application of these vectors and this approach for delivering therapeutic and other genes selectively to microglia.

Lipid-associated sialoprotein in the cerebrospinal fluid: association with brain malignancies.

BACKGROUND: Changes in the glycosylation process by tumor cells result in larger amounts of sialoproteins on their surface compared with normal cells. Sialoproteins then are released into the surrounding environment primarily by shedding or cell lysis. In the current study, the authors attempted to evaluate whether lipid-associated sialoprotein (LSP) in the cerebrospinal fluid (CSF) can distinguish patients with primary and metastatic brain tumors from those without brain tumors as well as determine response to treatment. METHODS: CSF samples were obtained from a tissue bank. The concentration of LSP was determined after chloroform:methanol extraction followed by protein precipitation. One-way analysis of variance and Scheffe pairwise comparisons were used for statistical analysis. RESULTS: The CSF of neurologically normal controls, patients with a normal leukocyte count (< or = 5/microl), and patients with various neurologic disorders or systemic tumors without central nervous system (CNS) malignancies contained similar levels of LSP. The CSF from patients with a normal leukocyte count and newly diagnosed primary or metastatic brain tumors contained on average 3.7-fold higher levels of LSP compared with CSF from patients without CNS tumors (P = 0.0001). The CSF from patients with brain tumors with progressive disease not responding to treatment contained high levels of LSP comparable to the levels found in newly diagnosed patients. The CSF from treatment-responsive patients contained decreased levels of LSP similar to that found in control patients. CONCLUSIONS: The LSP in CSF may be a useful marker with which to determine the presence of intracranial malignancies and assess response to treatment.

Host metalloproteinases in Lyme arthritis.

OBJECTIVE: To assess the role of matrix metalloproteinases (MMPs) in cartilage and bone erosions in Lyme arthritis METHODS: We examined synovial fluid from 10 patients with Lyme arthritis for the presence of MMP-2, MMP-3, MMP-9, and "aggrecanase" activity using gelatinolytic zymography and immunoblot analysis. We developed an in vitro model of Lyme arthritis using cartilage explants and observed changes in cartilage degradation in the presence of Borrelia burgdorferi and/or various protease inhibitors. RESULTS: Synovial fluid from patients with Lyme arthritis was found to contain at least 3 MMPs: gelatinase A (MMP-2), stromelysin (MMP-3), and gelatinase B (MMP-9). In addition, there was evidence in 2 patients of "aggrecanase" activity not accounted for by the above enzymes. Infection of cartilage explants with B. burgdorferi resulted in induction of MMP-3, MMP-9, and "aggrecanase" activity. Increased induction of these enzymes by B. burgdorferi alone was not sufficient to cause cartilage destruction in the explants as measured by glycosaminoglycan (GAG) and hydroxyproline release. However, addition of plasminogen, which can act as an MMP activator, to cultures resulted in significant GAG and hydroxyproline release in the presence of B. burgdorferi. The MMP inhibitor batimastat significantly reduced the GAG release and completely inhibited the collagen degradation. CONCLUSION: MMPs are found in synovial fluids from patients with Lyme arthritis and are induced from cartilage tissue by the presence of B. burgdorferi. Inhibition of MMP activity prevents B. burgdorferi-induced cartilage degradation in vitro.

Cardiac conduction abnormalities in a mouse model of Lyme borreliosis.

BACKGROUND: Borrelia Burgdorferi (BB) induces cardiac conduction abnormalities in infected humans. Mice models of Lyme disease have been developed, however their electrophysiologic (EP) properties of conduction are unknown. METHODS: Seventy-six C3H/J mice (BB infected and age- and gender-matched controls) underwent blinded in vivo EP studies. In a first phase of the study, 40 male C3H/J mice were divided into 2 groups: Group (A) mice were infected at age 3 (weeks) and studied at 5, and Group (B) mice were infected at 9 and studied at 11. In a second phase, 36 female mice were divided into 2 groups: Group (C) mice were infected at 3 weeks and studied at 5, and Group (D) mice were infected at 3 and studied at 11. RESULTS: Infected mice of group (A) and (C) had wider QRS complexes (21.0+/-1.6 versus 17.3+/-1.3ms, p< or =0.0001 and 20.3+/-2.1 versus 18.5+/-1.7, p = 0.05, respectively) compared to the healthy controls (HC). Infected mice of group (B) and group (D) were similar to the HC. In all groups, the presence of conduction abnormalities correlated very closely with the amount of inflammation on pathology. CONCLUSION: This study describes the first EP mouse model of Lyme carditis. C3H/J mice exhibit conduction abnormalities that are reversible 8 weeks after inoculation, closely paralleling the resolution of inflammation on pathology. This model can be a valuable tool in the developing and testing of new modalities for the prevention and treatment of Lyme carditis.

Borrelia burgdorferi induces matrix metalloproteinases by neural cultures.

Matrix metalloproteinases (MMPs) are associated with chronic neurologic diseases such as multiple sclerosis and senile dementia. Lyme disease is a multisystemic infection involving the nervous system, skin, joints, and heart. Neurologic manifestations of chronic Lyme disease include encephalopathy and cranial and peripheral neuropathy. Borrelia burgdorferi, the spirochaete causing Lyme disease, has been cultured from the cerebrospinal fluid (CSF), and B. burgdorferi DNA is frequently detected in the CSF of patients with Lyme neuroborreliosis. We used cerebral and cerebellar primary cultures to determine whether B. burgdorferi induces the production of MMPs by primary neural cultures. B. burgdorferi in a dose- and time-dependent manner induced the expression of MMP-9 by primary neural cultures but had no effect on the expression of MMP-2. Human and rat type I astrocytes expressed MMP-9 when incubated with B. burgdorferi in the same manner as primary neural cultures. This response may play a role in the symptomatology and the pathogenesis of Lyme neuroborreliosis.

Specific matrix metalloproteinase profiles in the cerebrospinal fluid correlated with the presence of malignant astrocytomas, brain metastases, and carcinomatous meningitis.

BACKGROUND: Detection in tumor tissue of specific matrix metalloproteinases (MMPs), particularly gelatinases A and B, correlates with the grade and aggressiveness of primary and metastatic brain tumors. The ability to detect these enzymes in the cerebrospinal fluid (CSF) would be a minimally invasive method of evaluating brain tumors. METHODS: CSF from 66 patients with white blood cell counts of < or = 5 microL were analyzed for the presence of gelatinolytic activity by zymography. Twenty-nine patients had malignant astrocytomas, 10 had brain metastases from systemic malignancies, 4 had systemic cancer not involving the central nervous system, 4 had nonmalignant neurologic diseases, and 19 were healthy controls. Fifteen CSF samples had positive cytologies. The zymographic results were retrospectively correlated with clinical information and CSF cytologic data. RESULTS: CSF from all patients with malignant astrocytomas or brain metastases contained precursor gelatinase A (pMMP2) and precursor gelatinase B (pMMP9), whereas control CSF contained only pMMP2. All patients with positive CSF cytologies had activated MMP2. A similar correlation was observed between the presence of activated MMP9 and positive CSF cytology. CONCLUSIONS: The precursor and activated forms of gelatinases A and B can be detected in the CSF of patients with primary and metastatic brain tumors. The distribution of gelatinase activity in CSF distinguishes patients with malignant gliomas or brain metastases from those without brain tumors, and distinguishes patients with meningeal carcinomatosis from those without CSF spread of tumor, regardless of their brain tumor status. Analysis of MMPs in the CSF may be a sensitive technique for diagnosing CNS tumors and provide an early indication of tumor recurrence. This technique may also provide longitudinal information that would be useful in evaluating ongoing treatment and predicting tumor behavior.

Matrix metalloproteinases in the cerebrospinal fluid of patients with Lyme neuroborreliosis.

Neurologic manifestations of Lyme disease include meningitis, encephalopathy, and cranial and peripheral neuropathy. There are no sensitive markers for neuroborreliosis, and diagnosis is often based on clinical presentation and cerebrospinal fluid (CSF) abnormalities, including intrathecal antibody production. Matrix metalloproteinase (MMP) activity in CSF was compared in patients with neuroborreliosis, patients with diverse neurologic disorders, and healthy controls. The CSF of 17 of 18 healthy subjects and 33 of 37 patients with neurologic symptoms and normal CSF and imaging studies contained only MMP2. The CSF of several patients with neurologic disorders contained MMP2, MMP9, and gelatinolytic activity at 130 and 250 kDa. The 130-kDa MMP was found without the 92-kDa MMP9 in the CSF of 11 (79%) of 14 patients with neuroborreliosis and only 7 (6%) of 118 control patients (P < .001). This pattern of CSF gelatinase activity may be a useful marker for neuroborreliosis.

Isolation, cloning, and expression of a 70-kilodalton plasminogen binding protein of Borrelia burgdorferi.

Surface receptors for plasminogen are expressed by many gram-positive and gram-negative bacteria and may play a role in the dissemination of organisms by binding plasminogen, which upon conversion to plasmin can digest extracellular matrix proteins. Two plasminogen binding proteins have been identified for Borrelia burgdorferi, outer surface protein A and a 70-kDa protein (BPBP). We purified BPBP by plasminogen affinity chromatography and obtained its amino acid sequence by Edman degradation of a tryptic digest. The gene coding for BPBP was isolated from a lambda-ZAP II genomic library with probes developed from sequenced portions of the protein. This gene was expressed in Escherichia coli; the recombinant product was seen by antibody raised against native BPBP and also bound 125I-labeled plasminogen. The experimentally derived amino acid sequences corresponded to the predicted sequence encoded by the BPBP gene. The deduced amino acid sequence for BPBP revealed significant similarity to p30, a 30-kDa protein of B. burgdorferi (54% identity and 65% similarity), to a 60-kDa protein in Borrelia coriaceae (66% identity and 80% similarity), to oligopeptide binding protein A of E. coli (34% identity and 57% similarity), and, more generally, to the periplasmic oligopeptide binding family of proteins.

Alcohol inhibits cell-cell adhesion mediated by human L1.

Mental retardation, hydrocephalus, and agenesis of the corpus callosum are observed both in fetal alcohol syndrome (FAS) and in children with mutations in the gene for the cell adhesion molecule L1. We studied the effects of ethanol on cell-cell adhesion in mouse fibroblasts transfected with human L1. L1-transfected fibroblasts exhibited increased cell-cell adhesion compared with wild-type or vector-transfected controls. Ethanol potently and completely inhibited L1-mediated adhesion both in transfected L cells and NIH/3T3 cells. Half-maximal inhibition was observed at 7 mM ethanol, a concentration achieved in blood and brain after ingesting one alcoholic beverage. In contrast, ethanol did not inhibit the adhesion of fibroblasts transfected with vector alone or with N-CAM-140. L1-mediated cell-cell adhesion was inhibited with increasing potency by n-propanol and n-butanol, but was not inhibited at all by n-alcohols of 5 to 8 carbons, acetaldehyde, or acetate, suggesting that ethanol interacts directly with a small hydrophobic pocket within L1. Phenylalanine, teratogenic anticonvulsants, and high concentrations of glucose did not inhibit L1-mediated cell-cell adhesion. Ethanol also inhibited potently the heterotypic adhesion of rat cerebellar granule cells to a monolayer of L1-transfected NIH/3T3 cells, but had no effect on their adhesion to N-CAM-140 or vector-transfected NIH/3T3 cells. Because L1 plays a role in both neural development and learning, ethanol inhibition of L1-mediated cell-cell interactions could contribute to FAS and ethanol-associated memory disorders.

Inhibition of Borrelia burgdorferi-bound fibrinolytic enzymes by alpha2-antiplasmin, PAI-1 and PAI-2.

Lyme disease is caused by Borrelia burgdorferi. Human plasminogen and urokinase-type plasminogen activator bind to the surface of the spirochete where plasmin is generated. We have suggested that bound urokinase and plasminogen are utilized by the organism to disseminate. We tested whether the physiological inhibitors of urokinase, plasminogen activator inhibitor-1 and -2 (PAI-1, PAI-2), could regulate the activity of spirochete-bound urokinase. The k(ass) of PAI-1 and PAI-2 for bound urokinase were 1.3 x 10(6) M(-1)s(-1) and 6.9 x 10(4)M(-1)s(-1), respectively, whereas the k(ass) for free urokinase were 7.2 x 10(6) M(-1)s(-1) and 5.3 x 10(5) M(-1)s(-1), respectively. Plasmin associated with the spirochete was not inhibited by alpha2-antiplasmin. These results suggest that PAI-1, PAI-2 and alpha2 antiplasmin would not be efficient regulators of fibrinolytic protease activity on the Borrelial surface and would not pose a barrier to utilization of these enzymes for dissemination in the human host.

Glial hyaluronate-binding protein: a product of metalloproteinase digestion of versican?

Glial hyaluronate-binding protein (GHAP) is a 60 kDa glycoprotein with an amino acid sequence identical to that of the hyaluronate-binding region of versican, a large fibroblast aggregating proteoglycan found in the brain. Both GHAP and versican were identified by immunoblot in bovine brain extracts prepared only minutes after death. Human recombinant collagenase, stromelysin, mouse gelatinase and gelatinases isolated from human brain by affinity chromatography digest versican and give rise to a polypeptide with electrophoretic mobility identical to GHAP. Immunoblot analysis, peptide mapping and C-terminal amino acid sequencing indicate that the polypeptide generated by digestion with human brain gelatinases is identical to GHAP. We suggest that GHAP is a naturally occurring versican degradation product.

Binding of human plasminogen to Borrelia burgdorferi.

We studied the binding of plasminogen to Borrelia burgdorferi, a spirochete which causes Lyme disease and produces no endogenous proteases which digest extracellular matrix proteins. Using 125I-labeled plasminogen, we demonstrated that B. burgdorferi bound human plasminogen and that this binding was inhibitable with unlabeled plasminogen. 125I-labeled plasminogen binding by B. burgdorferi was also inhibited by the lysine analog epsilon-aminocaproic acid. There was no significant difference in the binding of Glu- or Lys-plasminogen to B. burgdorferi. Binding of plasminogen was similar in low-passage (infectious) and high-passage (noninfectious) isolates of B. burgdorferi. Plasminogen bound to the surface of B. burgdorferi could be converted into plasmin by a human urokinase-type plasminogen activator. 125I-labeled plasminogen ligand blots of borrelial membrane proteins demonstrated two prominent binding proteins at approximately 70 and approximately 30 kDa. By Western blot (immunoblot), the 30-kDa protein was found to be outer surface protein A (Osp A) of B. burgdorferi. 125I-labeled plasminogen binding to both the 70-kDa protein and Osp A was inhibited by approximately 90% with a 1,000-fold excess of unlabeled plasminogen. By scanning densitometry, the 70-kDa band bound > 10 time more 125I-labeled plasminogen than did Osp A. An Osp A-deficient mutant of B. burgdorferi and wild-type B. burgdorferi bound equal amounts of 125I-labeled plasminogen. Ligand blots of membrane proteins from an Osp A-deficient mutant showed association of 125I-labeled plasminogen at only the 70-kDa protein. Two-dimensional gel electrophoresis showed that the 70-kDa protein had a pI of approximately 5.3, clearly separable from Osp A. The association of host plasmin(ogen) with borrelial surface proteins provides a mechanism by which B. burgdorferi can digest extracellular matrix and disseminate.

Neuroprotective effect of human osteogenic protein-1 in a rat model of cerebral hypoxia/ischemia.

Possible neuroprotective actions of osteogenic protein-1 (OP-1) were evaluated in a rat model of cerebral hypoxia/ischemia. Intraperitoneal injection of 50 micrograms of OP-1 prior to bilateral carotid ligation and transient hypoxia in 12-day-old rats reduced cerebral infarct area from 44.8 +/- 3.3% in vehicle-injected controls to 29 +/- 4.9%. Treatment of 14-day-old rats with 20 micrograms of OP-1 1 h after hypoxia reduced mortality from 45% to 13%. OP-1 may represent a novel class of neuroprotective agents.

Ethanol inhibits neural cell-cell adhesion.

Gestational exposure to ethanol causes defects in neuronal migration, fasciculation, and synaptogenesis, developmental events that depend on the patterned expression and function of cell adhesion molecules (CAMs). Recombinant human osteogenic protein-1 (hOP-1) increases cell-cell adhesion and promotes cell clustering in proliferating neuroblastoma x glioma hybrid NG108-15 cells by strongly inducing N-CAM and L1. Here we show that concentrations of ethanol achieved during social drinking inhibit hOP-1-induced cell clustering without affecting cell proliferation, the induction and cell surface expression of N-CAM and L1, or the alternative splicing and sialylation of N-CAM. This inhibition was reproduced by other alcohols in proportion to their chain length, but not by teratogenic anticonvulsants or phenylalanine. Ethanol inhibition of hOP-1 morphogenesis was inversely proportional to the concentration of hOP-1 and, hence, to the levels of N-CAM and L1. Low concentrations of ethanol (IC50 5-10 mM) inhibited cell-cell adhesion in hOP-1-treated cells, and this action too was reproduced more potently by propanol and butanol. Ethanol may perturb brain and skeletal development by inhibiting CAM-mediated cell-cell interactions.

A role for extracellular matrix degradation and matrix metalloproteinases in senile dementia?

In brain as in cartilage, the extracellular matrix contains aggregates formed by hyaluronic acid (HA) and proteoglycans. In osteoarthritic cartilage, release of the proteoglycans from the aggregates by cleavage of the HA-binding region results in the accumulation of the HA-binding region and in the fragmentation of the released proteoglycans. Stromelysin, a matrix neutral metalloproteinase, is one of the enzymes responsible for the cleavage of the HA-binding region. We suggest that a similar process also occurs in senile dementia. The brain proteoglycan contains sequences identical to those of aggrecan, which are recognized and cleaved by stromelysin, and is, in fact, susceptible to stromelysin digestion. Monoclonal antibodies reacting with glial HA-binding protein, but not with the parent protein, stained several senile plaques as defined by their reactivity with antibodies to the amyloid-beta protein in double-labeling experiments.

Regulation of neural cell adhesion molecule and L1 by the transforming growth factor-beta superfamily. Selective effects of the bone morphogenetic proteins.

The transforming growth factor-beta (TGF-beta) superfamily plays a role in embryogenesis and regeneration. We have reported that osteogenic protein-1 (OP-1) promotes cell aggregation and induces the expression of the neural cell adhesion molecules N-CAM and L1 in proliferating neuroblastoma x glioma hybrid NG108-15 cells (Perides, G., Safran, R. M., Rueger, D. C., and Charness, M. E. (1992) Proc. Natl. Acad. Sci. U.S.A. 89, 10326-10330; Perides, G., Hu, G., Rueger, D. C., and Charness, M. E. (1993) J. Biol. Chem. 268, 25197-25205). Here we show that the structurally homologous bone morphogenetic proteins (BMP) BMP-2 and BMP-4 are 10-50-fold more potent in these actions than the subfamily comprising BMP-5, BMP-6, and OP-1 (BMP-7). In contrast, members of the TGF-beta subfamily, activin-A, inhibin-A, and 29 additional growth factors and cytokines did not induce N-CAM. The addition of serum to cells growing in serum-free medium caused a concentration-dependent increase in N-CAM and L1 expression; however, serum did not potentiate the induction of N-CAM and L1 by 40 ng/ml OP-1. These findings suggest the presence in NG108-15 cells of a BMP-2/BMP-4 receptor that discriminates subtle differences in structure among homologous members of the TGF-beta superfamily. An endogenous ligand for this receptor may be present in serum.