The herpesvirus protease: mechanistic studies and discovery of inhibitors of the human cytomegalovirus protease.

The herpesvirus protease is a recently identified enzyme which is essential for viral replication. It is found in all herpesviruses and offers a new molecular target for therapeutic intervention. Its genomic structure has recently been described and consists of a large open reading frame which encodes a fusion protein containing an amino-terminal protease domain in-frame with a carboxyl-terminal "assembly protein-like" domain. Auto-processing releases the amino-terminal protease as a maturational enzyme. The herpesvirus protease has been characterized as a novel serine protease. Four surface accessible sulfhydryl groups have been identified in the human cytomegalovirus (HCMV) protease. Utilizing a fluorogenic DABCYL-EDANS substrate assay, directed screening has identified a class of sulfhydryl-modifying benzimidazolylmethyl sulfoxides which inhibits recombinant HCMV protease. Site-directed mutagenesis studies suggest oxidative modification of surface-accessible HCMV protease Cys138 (and possibly Cys161) by this class of inhibitors. The benzimidazolylmethyl sulfoxide 1 inhibits HCMV protease (IC50 = 1.9 microM), exhibits selectivity vs. mammalian serine proteases, and exhibits antiviral activity in an HCMV infected cell culture assay.

Three-dimensional structure of human cytomegalovirus protease.

Herpesviruses encode a serine protease that specifically cleaves assembly protein. This protease is critical for replication, and represents a new target for antiviral drug design. Here we report the three-dimensional structure of the protease from human cytomegalovirus (hCMV) at 2.27 angstroms resolution. The structure reveals a unique fold and new catalytic strategy for cleavage. The monomer fold of the enzyme, a seven-stranded beta-barrel encircled by a chain of helices that form the carboxy terminus of the molecule, is unrelated to those observed in classic serine proteases such as chymotrypsin and subtilisin. The serine nucleophile at position 132 is activated by two juxtaposed histidine residues at positions 63 and 157. Dimerization, which seems to be necessary for activity, is observed in the crystals. Correlations of the structure with the sequences of herpesvirus proteases suggest that dimerization may confer specificity and recognition in substrate binding.

Intraocular production of a cytokine (CINC) responsible for neutrophil infiltration in endotoxin induced uveitis.

AIMS/BACKGROUND: The subcutaneous injection of bacterial endotoxin in Lewis rats produces an acute intraocular inflammation evolving over a 24 hour period. This endotoxin induced uveitis (EIU) is characterised by a biphasic protein exudation and a cellular infiltrate composed of macrophages and polymorphonuclear neutrophils (PMNs). This model was used to study the mechanism of cellular infiltration in ocular inflammation. METHODS: EIU was induced by a subcutaneous injection of lipopolysaccharide (LPS) (S typhimurium) at 350 micrograms/kg. The levels of cytokine induced neutrophil chemoattractant (CINC) were measured every 2 hours in the serum and in the aqueous humour by ELISA. The intraocular inflammation was quantified by protein measurement and leucocyte counting. RESULTS: The kinetics of CINC production in the systemic circulation showed a rapid rise, peaking 2 hours after LPS injection, followed by a progressive decline over the next 8 hours. In the eye, the CINC levels increased above the serum levels 10 hours after EIU induction corresponding to the time of cellular infiltration. When leucocyte entry in the eye was inhibited by 56% and 64% with an antiadhesion molecule antibody, there was only a slight reduction in the aqueous humour CINC levels of 9% and 16%, respectively, indicating that CINC was produced by ocular tissue cells. The specific effect of CINC in the eye was confirmed when a direct intraocular injection of 250 ng of purified CINC was followed by significant PMN infiltration, in the absence of protein exudation. CONCLUSION: The data indicate that the production of the CINC chemotactic factor by ocular tissue participates in the inflammatory reaction in EIU.

Chemokine gene expression in anti-glomerular basement membrane antibody glomerulonephritis.

Chemokines may be important in the pathogenesis of glomerular leukocyte infiltration in antiglomerular basement membrane (GBM) antibody (Ab) glomerulonephritis (GN). We studied the expression of the C-C chemokines [macrophage inflammatory protein (MIP)-1 alpha, monocyte chemotactic protein (MCP)-1, and RANTES] and C-X-C chemokines [platelet factor 4 (PF4), interferon-inducible protein of 10 kDa (IP-10), MIP-2, and cytokine-induced neutrophil chemoattractant (CINC)] at 30 min, 3, 6, 9, 15, and 24 h after induction of heterologous-phase anti-GBM Ab GN in Lewis rats. There was a rapid induction of CINC, MIP-2, MCP-1, and MIP-1 alpha mRNAs in the glomeruli of nephritic rats 30 min after administration of the anti-GBM Ab, whereas increases in PF4 and IP-10 mRNAs were not seen until 3 h. The mRNA expression of PF4, MIP-1 alpha, MIP-2, and IP-10 peaked at 3 h, whereas CINC and MCP-1 peaked at 6 and 15 h, respectively. By comparison, the expression of RANTES mRNA in rats with anti-GBM Ab GN did not differ from those of control rats. These changes in chemokine gene expression were associated with glomerular polymorphonuclear leukocytes (PMN) and monocyte/macrophage infiltration which peaked at 3 h (20.8 +/- 11.1 PMN/glomerulus) and 24 h (8.2 +/- 1.0 ED-1 cells/glomerulus), respectively. The administration of dexamethasone suppressed glomerular chemokine mRNA expression (60-98%) at both 3 and 15 h, which was associated with a 50-100% reduction in glomerular PMN and monocyte/macrophage infiltration, respectively.(ABSTRACT TRUNCATED AT 250 WORDS)

Activity of two-chain recombinant human cytomegalovirus protease.

The human cytomegalovirus UL80 protease was expressed in Escherichia coli and purified by metal-chelate chromatography using a histidine tag engineered at the amino terminus. Cleavage of the 30-kDa protease at an internal site, VEA/A144, resulted in the recovery of 16- plus 14-kDa two-chain protease. The amino-terminal 16-kDa chain and the carboxyl-terminal 14-kDa chain remained associated as an active enzyme that was modified specifically at Ser132 on the 16-kDa chain by [3H]diisopropyl fluorophosphate. Disruption of the cleavage site by mutation from VEA/A to AEA/A facilitated the recovery of active 30-kDa one-chain enzyme that could be similarly modified at Ser132 by [3H]diisopropyl fluorophosphate. Both one- and two-chain enzymes cleaved recombinant assembly protein at the maturation site, VNA/S, and a peptide, GVVNASARL, mimicking this site. Internal processing does not inactivate the protease but forms a two-chain enzyme that retains activity.

Cytokine-induced neutrophil chemoattractant mediates neutrophil influx in immune complex glomerulonephritis in rat.

Chemokines are a family of cytokines whose participation in inflammation in vivo remains to be established. Using the rat model of anti-glomerular basement membrane (GBM) nephritis, we found that mRNA for the chemokine CINC (cytokine-induced neutrophil chemoattractant) was induced in the kidney, and the corresponding protein was elaborated by isolated inflamed glomeruli. Production of CINC by glomeruli was unaffected by complement- or leukocyte-depletion prior to disease induction. Cytokines which induce CINC expression in renal cells (TNF-alpha and IL-1 beta) were also expressed in the kidney during glomerular inflammation. TNF-alpha production, unlike CINC, was complement and leukocyte dependent. In vivo administration of anti-CINC, but not anti-human IL-8, IgG selectively attenuated the influx of PMNs into the glomerulus and commensurately diminished proteinuria. The participation of CINC was not tissue-specific: anti-CINC IgG also diminished the influx of PMNs in dermal immune complex inflammation. In sum, we propose that glomerular immune complex deposition/complement activation leads to cytokine production which results in CINC expression by endogenous glomerular cells. The CINC produced plays a contributory role in the influx of PMNs into the glomerulus in the context of the activation of other inflammatory mediators. These results suggest a potential role for CINC homologues, IL-8 and the GRO family of chemokines, in human immune complex-mediated disease.

Investigation of possible autocrine functions for rat GRO/CINC (cytokine-induced neutrophil chemoattractant).

Rat cytokine-induced neutrophil chemoattractant (CINC) is an eight kilodalton polypeptide originally purified from media conditioned by interleukin-1 beta stimulated 52E, an epithelioid clone derived from normal rat kidney (NRK) cells. Using a fibroblastic clone of the NRK cells, 49F, we found expression of the CINC gene to be induced by either serum or cytokines in growth-arrested cultures within 1 hour of stimulation. There was no observable CINC expression in exponentially growing cells in the absence of cytokine stimulation. CINC protein had no significant effect on 3H-thymidine incorporation or growth rate of NRK49F. We have observed that CINC is constitutively produced by some transformed NRK cells, clone RC20, suggesting an association with the expression of a transformed phenotype. Unlike the parent 49F, RC20 cells are capable of growth in soft agar and serum-free media and form highly metastatic tumors in nude mice. We have examined the possible autocrine functions of CINC and its possible links to the expression of the transformed phenotype by these cells. The use of a blocking CINC polyclonal antibody demonstrated that CINC did not function as an autocrine growth factor for RC20. Though CINC is a potent chemoattractant for neutrophils, it did not induce migration of either RC20 or 49F cells. CINC only moderately promoted adhesion of RC20 cells when used as a matrix protein. These data do not support the hypothesis that production of CINC by the RC20 cells provides an obvious advantage for the transformed cells constitutively producing it.

High-level expression of cytokine-induced neutrophil chemoattractant (CINC) by a metastatic rat cell line: purification and production of blocking antibodies.

Significant levels of cytokine-induced neutrophil chemoattractant (CINC) were found in serum-free medium conditioned by a highly metastatic rat cell line, RC20. To study CINC's role in inflammation and metastasis, CINC was purified from this source for use in in vitro assays and for antibody production in goats and rabbits. CINC was a potent chemoattractant for rat neutrophils (EC-50 0.5 nM). A fusion protein of glutathione-S-transferase and CINC (GST-CINC) was produced in E. coli. Anti-CINC polyclonal IgG was purified from immune goat and rabbit sera by protein A and GST-CINC affinity chromatography. Both goat and rabbit anti-CINC antibody preparations at 4 micrograms/mL (an 11-fold molar excess) were found to completely block the activity of 2.5 nM CINC in a rat neutrophil chemotaxis assay. These antibodies have been used to develop a sensitive immunoassay for CINC. The availability of large amounts of affinity-purified blocking anti-CINC antibody will allow investigations into the role played by CINC in rodent inflammation models and in the metastasis of RC20 cells.

Oligosaccharides at each glycosylation site make structure-dependent contributions to biological properties of human tissue plasminogen activator.

The effect of altering oligosaccharide structures at sites 184 and 448 of tissue plasminogen activator (tPA) has been examined. Alteration to high-mannose forms at sites 184 and 448 was accomplished by the growth of cells in the presence of deoxymannojirimycin (dMM). Modification to neutral, unsialylated forms at these sites was achieved by neuraminidase treatment of control preparations of tPA. Oligosaccharides at site 117 were not markedly affected by either treatment because structures at this site are high-mannose and not sialylated in untreated preparations. The effect on enzymatic activity and on a related property, lysine affinity, was determined. dMM treatment was found to increase both the lysine affinity and catalytic activity of tPA. Neuraminidase treatment increased enzyme activity, but was without effect on affinity for lysine. To evaluate the effects of alterations at site 184 and site 448, the catalytic activity and lysine affinity of type I and type II tPA were monitored individually. In the dMM-treated sample, type I tPA (with sugars at sites 117, 184 and 448) was found to have 2- to 3-fold increased catalytic activity and an affinity for lysine which was greater than that of type I from untreated preparations, but less than that of control type II tPA (containing sugar only at sites 117 and 448). In neuraminidase-treated type I, catalytic activity was also enhanced but lysine affinity remained unchanged. Type II from dMM- and neuraminidase-treated preparations had catalytic activity that was increased approximately 1.5-fold compared to untreated controls, whereas affinity for lysine was unchanged.(ABSTRACT TRUNCATED AT 250 WORDS)

Effect of peptides on the inactivation of tissue plasminogen activator by plasminogen activator inhibitor-1 and on the binding of tissue plasminogen activator to endothelial cells.

The effectiveness of tissue plasminogen activator (tPA) in thrombolytic therapy is dependent upon the rate at which therapeutically administered tPA reaches the clot site and the proportion of that tPA which is enzymatically active. Interactions between tPA and its main plasma inhibitor (PAI-1) and between tPA and the endothelial cells lining blood vessels are two factors which may limit efficacy. In an attempt to identify the regions of the tPA molecule involved in these interactions, we have examined a series of synthetic peptides with amino acid sequences corresponding to different regions of the tPA molecule for their ability to protect tPA from inactivation by PAI-1 and for their ability to reduce the binding of tPA to endothelial cells. Three peptides were identified which were especially effective at maintaining tPA activity in the presence of PAI-1 and three others were found which had a lesser effect. These same peptides were also found to inhibit the binding of tPA to endothelial cells. This suggests that the same regions of the tPA molecule are involved in both processes. None of the peptides inhibited the binding of tPA to fibrin. These peptides may serve as models for the development of agents for enhancing the activity of both endogenous tPA and of tPA administered in thrombolytic therapy.

Secretion of active kringle-2-serine protease in Escherichia coli.

Active human tissue plasminogen activator variant kringle-2-serine protease (K2 + SP domains; referred to as MB1004) was synthesized as a secreted protein in Escherichia coli, isolated, and characterized. MB1004 is a relatively large and complex protein, approximately 38 kDa in size and containing nine disulfide bonds. MB1004 without a pro region was secreted into the periplasm of E. coli by fusing the protein to the PhoA leader peptide expressed from the tac promoter. Approximately 1% (20 micrograms/L broth) of the secreted MB1004 was purified from E. coli homogenates as a soluble, active enzyme by using a combination of lysine and Erythrina inhibitor affinity chromatography. Purified MB1004 was monomeric and single-chain, and the N-terminus was identical with the predicted amino acid sequence. The specific activity of purified MB1004 from E. coli was compared against the equivalent recombinant material purified from mammalian cells that was naturally glycosylated (MB1004G) or deglycosylated after treatment with N-glycanase (MB1004N). Results from four different in vitro assays showed that MB1004 and MB1004N had similar activities. Both exhibited 4-12-fold higher specific activity than MB1004G in plasminogen activation assays. These results suggest that an inaccurate picture of specific activity can be obtained if the effects of glycosylation are not considered. By utilization of secretion in E. coli, nonglycosylated MB1004 was purified without in vitro refolding and was shown to be suitable for structure-function studies.

Binding of tissue plasminogen activator to human aortic endothelial cells.

The experiments described in this paper were designed to examine the specific binding of tissue plasminogen activator (tPA) to cultured human aortic endothelial (HAE) cells. When 125I-labelled tPA was incubated with the cells at 4 degrees C, binding was found to plateau within 90 min after incubations were begun. Binding was saturable and the bound enzyme dissociated from the sites with a half-time of approx. 48 min. Scatchard analyses were performed using tPA molecules isolated from human melanoma and colon cells as well as from C127 and Chinese hamster ovary cells that had been transfected with the human tPA gene. These enzymes showed very similar binding characteristics in spite of the fact that they differ substantially in the types of sugars which comprise their side chains. Neither the chainedness of the molecules (one-chain or two-chain) nor the sites at which they are glycosylated (type I or type II) appear to affect their ability to interact with binding sites. The tPA molecules were found to have an average equilibrium dissociation constant of (1.15 +/- 0.10) x 10(-9) M and HAE cells appeared to have a single, homogeneous population of independent binding sites present at a concentration of (1.57 +/- 0.13) x 10(6) sites per cell. Lowering the pH of the binding buffer from 7.4 to 6.5 resulted in a reversible increase in specific binding of between 2-fold and 7-fold depending upon the particular preparation of cells. Preincubation of tPA with plasminogen activator inhibitor 1 (PAI-1) was found to have little effect on binding, suggesting that tPA interacts at sites distinct from surface-bound PAI-1. No evidence for either internalization or degradation of tPA was observed in assays run at 37 degrees C. This suggests that, like urokinase, tPA remains on cell surfaces for an extended period of time.

Glycosylation at Asn-184 inhibits the conversion of single-chain to two-chain tissue-type plasminogen activator by plasmin.

Tissue-type plasminogen activator (tPA) is a glycosylated serine protease which is an effective thrombolytic agent. Native single-chain tPA (sc-tPA) is converted to two-chain tPA (tc-tPA) by plasmin, the product of the reaction of plasminogen with tPA. Native sc-tPA occurs as two glycoforms. Type I sc-tPA is fully glycosylated, while type II lacks glycosylation at Asn-184. The rates at which type I and type II human melanoma sc-tPA were converted to type I and type II tc-tPA by plasmin were determined by two different methods. In each case, the second-order rate constant (kcat/Km) for type II sc-tPA (approximately 8 microM-1 s-1) was about twice that for type I sc-tPA (approximately 4 microM-1 s-1). These results indicate that glycosylation at Asn-184 hinders the conversion of sc-tPA to tc-tPA and suggest that under physiological conditions type I sc-tPA may persist in the single-chain form longer than type II sc-tPA. Previous studies have shown that type I tc-tPA has a lower activity than type II tc-tPA and that sc-tPA has a lower activity and susceptibility to inhibition when compared to tc-tPA. The present work provides further evidence that tPA glycosylation serves to modulate activity. The two major glycoforms may represent more persistent but slow acting (type I) and less persistent but faster acting (type II) variants of tPA.

Cell-type-specific and site-specific N-glycosylation of type I and type II human tissue plasminogen activator.

Tissue plasminogen activator (t-PA) is an important initiator of fibrinolysis. The t-PA polypeptide has four potential N-glycosylation sites of which three are occupied in type I (Asn-117, -184, and -448) and two in type II (Asn-117 and -448). In an effort to elucidate the factors controlling the expression of N-linked oligosaccharides on this polypeptide, we have used a combination of sequential exoglycosidase digestion, methylation analysis, and controlled acetolysis to determine the oligosaccharide structures at each of the N-glycosylation sites of type I and type II t-PA when isolated from a human colon fibroblast cell strain and from a Bowes melanoma cell line. Our results suggest the following: (i) type I and type II t-PA are N-glycosylated in an identical way at Asn-117 and Asn-448, when isolated from the same cell line; (ii) Asn-117 is predominantly associated with oligomannose-type structures in all cases; (iii) Asn-184 and Asn-448 are predominantly associated with complex-type structures when t-PA is isolated from fibroblast cells, but with both complex- and oligomannose-type structures when isolated from melanoma cells; (iv) fibroblast cell derived t-PA is associated with both neutral and sialylated oligosaccharides, while melanoma cell derived t-PA is also associated with sulfated oligosaccharides, which are located exclusively at Asn-448 of type II t-PA; (v) no complex-type structures occur in common between t-PA from the two cell lines. These results indicate that the t-PA glycoprotein is secreted by each cell line as a set of glycoforms, each glycoform being unique with respect to the nature and disposition of oligosaccharides on a common polypeptide. Further, the two cell lines express no glycoform in common, despite expressing the same t-PA polypeptide. The implications of these results for both the control of oligosaccharide processing in different cell lines and the genetic engineering of mammalian glycoproteins are discussed.

Effects of N-glycosylation on in vitro activity of Bowes melanoma and human colon fibroblast derived tissue plasminogen activator.

Tissue-type plasminogen activator (t-PA), when isolated from human colon fibroblast (hcf) cells, is N-glycosylated differently than when isolated from the Bowes melanoma (m) cell line (Parekh et al., 1988). Both hcf- and m-t-PA can be separated into type I t-PA (with three occupied N-glycosylation sequons, at Asn-117, -184, and -448) and type II t-PA (with two occupied sequons, at Asn-117 and -448). Oligosaccharide analysis of each of these types of t-PA indicates that hcf-t-PA and m-t-PA have no glycoforms in common, despite having the same primary amino acid sequence. We have therefore compared in vitro the enzymatic activities and fibrin binding of type I and type II hcf- and m-t-PA with those of aglycosyl t-PA isolated from tunicamycin-treated cells. Plasminogen activation kinetics were determined by using an indirect amidolytic assay with Glu-plasminogen and a chromogenic plasmin substrate. In the absence of stimulator, there was little difference in activity between type I and type II t-PA, but the activity of aglycosyl t-PA was 2-4-fold higher than that of the corresponding glycosylated t-PA. In the presence of a fibrinogen fragment stimulator, the Kcat value of type II t-PA was approximately 5-fold that of type I t-PA from the same cell line, while the Km values for activation of Glu-plasminogen were similar (0.13-0.18 microM). The stimulated activity of glycosyl t-PA was similar to that of type II t-PA.(ABSTRACT TRUNCATED AT 250 WORDS)

Selenium-containing tRNA(Glu) and tRNA(Lys) from Escherichia coli: purification, codon specificity and translational activity.

In response to low (approximately 1 microM) levels of selenium, Escherichia coli synthesizes tRNA(Glu) and tRNA(Lys) species that contain 5-methylaminomethyl-2-selenouridine (mnm5Se2U) instead of 5-methylaminomethyl-2-thiouridine (mnm5S2U). Purified glutamate- and lysine-accepting tRNAs containing either mnm5Se2U (tRNA(SeGlu), tRNA(SeLys] or mnm5S2U (tRNA(SGlu), tRNA(SLys] were prepared by RPC-5 reversed-phase chromatography, affinity chromatography using anti-AMP antibodies and DEAE-5PW ion-exchange HPLC. Since mnm5Se2U, like mnm5S2U, appears to occupy the wobble position of the anticodon, the recognition of glutamate codons (GAA and GAG) and lysine codons (AAA and AAG) was studied. While tRNA(SGlu) greatly preferred GAA over GAG, tRNA(SeGlu) showed less preference. Similarly, tRNA(SGlu) preferred AAA over AAG, while tRNA(SeLys) did not. In a wheat germ extract--rabbit globin mRNA translation system, incorporation of lysine and glutamate into protein was generally greater when added as aminoacylated tRNA(Se) than as aminoacylated tRNA(S). In globin mRNA the glutamate and lysine codons GAG and AAG are more numerous than GAA and AAA, thus a more efficient translation of globin message with tRNA(Se) might be expected because of facilitated recognition of codons ending in G.

Purification and characterization of a tissue plasminogen activator-inhibitor complex from human umbilical vein endothelial cell conditioned medium.

Tissue plasminogen activator-inhibitor complexes were purified from the conditioned medium of human umbilical vein endothelial cells by affinity chromatography followed by gel filtration. It was found that a single complex was isolated which can exist in two distinct interconvertible conformations. These may be separated by electrophoresis into a form with a 105,000 apparent molecular weight and a form with an 88,000 apparent molecular weight. The particular conformation which predominates may be altered by changing the pH at which preparations are incubated or by including dithiothreitol in incubation buffers. Plasminogen activator enzymatic activity may be partially recovered from purified complexes by incubation in the presence of fibrin. Incubation in 1.5 M NH4OH results in the dissociation of the complex into two major polypeptides of 67 and 40 kilodaltons (kDa). The 40-kDa protein was isolated by gel filtration high-pressure liquid chromatography. N-Terminal amino acid analysis of this protein revealed three distinct sequences. Two of these were nearly identical and matched the N-terminal sequence recently reported for the native plasminogen activator inhibitor from endothelial cells. The third sequence exactly matched an internal portion of the same protein. The results suggest that the internal sequence is located at the site where the inhibitor is cleaved by tissue plasminogen activator.

Biosynthesis of 5-methylaminomethyl-2-selenouridine, a naturally occurring nucleoside in Escherichia coli tRNA.

A selenium-containing nucleoside, 5-methylaminomethyl-2-selenouridine (mnm5se2U), is present in lysine- and glutamate-isoaccepting tRNA species of Escherichia coli. The synthesis of mnm5se2U is optimum (4 mol/100 mol tRNA) when selenium is present at about 1 microM concentration and is neither decreased by a high (8 mM) level of sulfur in the medium nor increased by excessive (10 or 100 microM) levels of selenium. Lysine- and glutamate-isoaccepting tRNA species that contain 5-methylaminomethyl-2-thiouridine (mnm5s2U) coexist with the seleno-tRNAs in E. coli cells and a reciprocal relationship between the mnm5se2U- and the mnm5s2U-containing species is maintained under a variety of growth conditions. The complete 5-methylaminomethyl side chain is not a prerequisite for introduction of selenium at the 2-position. In E. coli mutants deficient in the ability to synthesize the 5-methylaminomethyl substituent, both the 2-thiouridine and the corresponding 2-selenouridine derivatives of intermediate forms are accumulated. Broken cell preparations of E. coli synthesize mnm5se2U in tRNAs by an ATP-dependent process that appears to involve the replacement of sulfur in mnm5s2U with selenium.

Determination of the number of endothelial cells in culture using an acid phosphatase assay.

A simple assay is described in which small numbers of endothelial cells in culture can be determined by measuring acid phosphatase activity. After removal of the growth medium from cells grown in 96-well culture plates, the cells are lysed in buffer containing the detergent Triton X-100 and the phosphatase substrate p-nitrophenyl phosphate. After 2 h at 37 degrees C, the reaction is stopped with sodium hydroxide, and color development is determined using a rapid multiwell plate reader. The assay detects 100 to 10,000 cells per well. The assay has been used to determine growth curves for endothelial cells in the presence and absence of endothelial cell growth factor from bovine hypothalamus and to monitor fractions during purification of the growth factor. Minor modifications in the assay allow it to be fully automated.