Demonstration of exosite I-dependent interactions of thrombin with human factor V and factor Va involving the factor Va heavy chain: analysis by affinity chromatography employing a novel method for active-site-selective immobilization of serine proteinases

In an essential step of blood coagulation, factor V is proteolytically processed by thrombin to generate the activated protein cofactor, factor Va, and to release the activation fragments E and C1. For the identification and characterization of sites of thrombin binding to factor V and its activation products, a new method was developed for immobilizing thrombin and other serine proteinases specifically (>/=92%) through their active sites and used in affinity chromatography studies of the interactions. Interactions of factor V with exosite I of thrombin were shown to regulate the factor V activation pathway from the 93% +/- 12% inhibition of the rate of activation correlated with specific binding of hirudin54-65 to this exosite. Chromatography of factor V on active-site-immobilized thrombin showed only a weak interaction, while the factor Va heterodimer bound specifically and with apparently higher affinity, in an interaction that was prevented by hirudin54-65. The heavy chain of subunit-dissociated factor Va bound to immobilized thrombin, while the light-chain subunit and fragment E had no detectable affinity. These results demonstrate a previously undescribed, exosite I-dependent interaction of thrombin with factor Va that occurs through the factor Va heavy chain. They support the further conclusion that similar exosite I-dependent binding of thrombin to the heavy-chain region of factor V contributes to recognition of factor V as a specific thrombin substrate and thereby regulates proteolytic activation of the protein cofactor.     

Purification of transforming growth factor beta 1 from human platelets

Transforming growth factor-beta (TGF beta) is a growth and differentiation factor which can be released from many cell types. In previous studies, platelets were identified as a rich source of TGF beta. Here we present a rapid and convenient method for TGF beta purification from human platelets which includes acid-ethanol extraction and gelfiltration, cation exchange and reversed phase chromatography. All purification steps are performed under acidic conditions to prevent adsorption of TGF beta to the vial walls. In addition, volatile solvents and buffers were used which allowed easy removal of solvent and salt by lyophilization. Using this method pure TGF beta can be easily obtained in high yield (370 micrograms) from 20 units of platelet concentrate.     

Human factor Va1 and factor Va2: properties in the procoagulant and anticoagulant pathways

Human plasma factor V is heterogeneous and yields two forms of activated factor V that bind with low (factor Va1) and high affinity (factor Va2) to phospholipids. The properties of factor Va1 and factor Va2 in the anticoagulant and procoagulant pathways were evaluated by comparing their sensitivity for inactivation by APC and their ability to act as cofactor in prothrombin activation. At low phospholipid concentrations and on membranes containing low amounts of phosphatidylserine (PS), factor Va1 was inactivated by APC at 15-fold lower rates than factor Va2, both in the absence and in the presence of protein S. At high phospholipid concentrations and on membranes with more than 15 mol % PS, factor Va1 and factor Va2 were inactivated with equal efficiency. Differences between cofactor activities of factor Va1 and factor Va2 in prothrombin activation were only observed on membranes with less than 7.5 mol % PS. Due to the different phospholipid requirements of APC-catalyzed factor Va inactivation and of expression of factor Va cofactor activity in prothrombin activation, the thrombin-forming capacity of factor V1 was 7-fold higher than that of factor V2 in a reaction system containing factor Xa, prothrombin, APC, protein S, vesicles with a phospholipid composition resembling that of activated platelets, and traces of thrombin to initiate prothrombin activation. This shows that in the process of generation, expression, and down-regulation of factor Va cofactor activity on physiological membranes, the overall procoagulant activity of factor V1 can considerably exceed that of factor V2.     

Identification and partial characterization of three calcium- and zinc-independent gelatinases constitutively present in human circulation

Three constitutive gelatinases in human plasma were identified and characterized relative to known matrix metalloproteinase (MMP) gelatinases: MMP-2 (fibroblast 72-kDa) and MMP-9 (neutrophil 92-, 130-, and 225-kDa). Substrate gel electrophoresis (gelatin zymography) revealed an apparent Mw of 78-, 82-, and 89-kDa for these gelatinases. Densitometry revealed that MMP-9 and MMP-2 were highly calcium sensitive requiring 50-150 microM and 500 microM calcium for half-maximal activity, respectively. Of the new gelatinases, only the 89-kDa form demonstrated slight calcium activation. The three gelatinases were unaffected by known MMP inhibitors: EDTA (5 mM), 1,10-phenanthroline (2 mM), and pepstatin (18 microM). Serine and thiol protease inhibitors (leupeptin, aprotinin, PMSF, TLCK, TPCK, antichymostatin, antipain) were also ineffective. Solution-phase IEF revealed that the 78- and 82-kDa forms focused at neutral pI 6.72-7.95 whereas the 89-kDa focused at an acidic pI 4.89-5.18 (similar to neutrophil and fibroblast forms). The data indicate that these gelatinases are not MMPs or partially activated MMPs. Their role in normal and pathological conditions is not known.     

Identification and partial characterization of the high-affinity choline carrier from rat brain striatum

[3H]Choline mustard aziridinium ion binds irreversibly to the sodium-coupled high-affinity choline transport protein in a sodium-dependent and hemicholinium-sensitive manner, and thus is a useful affinity ligand. In rat striatal synaptosomal membranes, it radiolabels two polypeptides with apparent molecular masses of 58 and 35 kDa. Based upon the use of two different experimental approaches, it appears that neither of these polypeptides is glycosylated.     

Protein C activation and factor Va inactivation on human umbilical vein endothelial cells

The inactivation of factor Va was examined on primary cultures of human umbilical vein endothelial cells (HUVECs), either after addition of activated protein C (APC) or after addition of alpha-thrombin and protein C (PC) zymogen. Factor Va proteolysis was visualized by Western blot analysis using a monoclonal antibody (alpha HVaHC No. 17) to the factor Va heavy chain (HC), and cofactor activity was followed both in a clotting assay using factor V-deficient plasma and by quantitation of prothrombinase function. APC generation was monitored using the substrate 6-(D-VPR)amino-1-naphthalenebutylsulfonamide (D-VPR-ANSNHC4H9), which permits quantitation of APC at 10 pmol/L. Addition of APC (5 nmol/L) to an adherent HUVEC monolayer (3.5 x 10(5) cells per well) resulted in a 75% inactivation of factor Va (20 nmol/L) within 10 minutes, with complete loss of cofactor activity within 2 hours. Measurements of the rate of cleavage at Arg506 and Arg306 in the presence and absence of the HUVEC monolayer indicated that the APC-dependent cleavage of the factor Va HC at Arg506 was accelerated in the presence of HUVECs, while cleavage at Arg306 was dependent on the presence of the HUVEC surface. Factor Va inactivation proceeded with initial cleavage of the factor Va HC at Arg506, generating an M(r) 75,000 species. Further proteolysis at Arg306 generated an M(r) 30,000 product. When protein C (0.5 mumol/L), alpha-thrombin (1 nmol/L), and factor Va (20 nmol/L) were added to HUVECs an APC generation rate of 1.56 +/- 0.11 x 10(-14) mol/min per cell was observed. With APC generated in situ, cleavage at Arg506 on the HUVEC surface is followed by cleavage at Arg306, generating M(r) 75,000 and M(r) 30,000 fragments, respectively. In addition, the appearance of two novel products derived from the factor Va HC are observed when thrombin is present on the HUVEC surface: the HC is processed through limited thrombin proteolysis to generate an M(r) 97,000 fragment, which is further processed by APC to generate an M(r) 43,000 fragment. NH2-terminal sequence analysis of the M(r) 97,000 fragment revealed that the thrombin cleavage occurs in the COOH-terminus of the intact factor Va HC since both the intact HC as well as the M(r) 97,000 fragment have the same sequence. Our data demonstrate that the inactivation of factor Va on the HUVEC surface, initiated either by APC addition or PC activation, follows a mechanism whereby cleavage is observed first at Arg506 followed by a second cleavage at Arg306. The latter cleavage is dependent on the availability of the HUVEC surface. This mechanism of inactivation of factor Va is similar to that observed on synthetic phospholipid vesicles.     

Binding of chick skin collagen alpha 1 chain by isolated membranes from human platelets

Binding of a denaturated polypeptide chain derived from chick skin collagen, the alpha 1(I) chain, by isolated membranes of human platelets has been demonstrated. The process is reversible, and time- and protein concentration-dependent. The binding is specific, with an association constant of 1.88 X 10(-6) M. Prior treatment of the isolated membranes with trypsin, chymotrypsin, and pronase, resulted in significant inhibition of the 14C-labeled alpha 1 chain binding, but neuraminidase or collagenase treatment had no effect. Dissociation of the bound radioactivity and subsequent chromatographic analyses on carboxymethylcellulose and agarose A-1.5m revealed that the alpha 1 chain was unaltered. Scatchard plot analysis suggested that there are approximately 20,000 binding sites per platelet. The binding of the alpha 1 chain was inhibited by a glycopeptide derived from alpha 1, alpha 1-CB5 and by purified glucosylgalactosyl hydroxylysine, but was not affected by other cyanogen bromide peptides of alpha 1, namely alpha 1-CB3, -CB4, -CB7, and -CB8. Kinetic studies demonstrated that inhibition by the hydroxylysine glycoside is competitive. Dose-response curves of platelet aggregation induced by alpha 1 and the binding of alpha 1 by platelet membranes correlate closely. These results indicate that there are specific binding sites for collagen alpha 1 chain on platelet membranes, and that the carbohydrate moiety of the alpha 1 chain plays a role in the binding. The findings also support the hypothesis that the chick skin alpha 1 chain mediates platelet aggregation and the release reaction by acting on platelet membranes.     

Purification and characterization of angiogenic factor from human bladder cancer

An angiogenic factor from human transitional cell cancer of bladder was purified by protein extraction, cation exchange chromatography, gel filtration high-performance liquid chromatography (GE-HPLC) and reversed-phase high-performance liquid chromatography (RP-HPLC). The purified substance was named as bladder cancer angiogenic factor (BCAF). Biological activity of the BCAF was assessed by using the method of chick embryo chorioallantoic membrane (CAM) assay and 3H-TdR incorporation into DNA in Balb/c 3T3 cells. The BCAF displayed the potent activities of neovascularization in CAM and DNA synthesis in Balb/c 3T3 cells. The ultrastructural features of blood vessels induced by the BCAF were similar to the blood vessels in tumors. The BCAF contained a protein with an approximate molecular weight of 15,000 D, which was identified by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and silver staining. Amino acid compositions of the BCAF were also analysed by acid hydrolysis.     

Identification and characterization of a human protein kinase related to budding yeast Cdc7p

The Cdc7p protein kinase is essential for the G1/S transition and initiation of DNA replication during the cell division cycle in Saccharomyces cerevisiae. Cdc7p appears to be an evolutionarily conserved protein, since a homolog Hsk1 has been isolated from Schizosaccharomyces pombe. Here, we report the isolation of a human cDNA, HsCdc7, whose product is closely related in sequence to Cdc7p and Hsk1. The HsCdc7 cDNA encodes a protein of 574 amino acids with predicted size of 64 kDa. HsCdc7 contains the conserved subdomains common to all protein-serine/threonine kinases and three "kinase inserts" that are characteristic of Cdc7p and Hsk1. Immune complexes of HsCdc7 from cell lysates were able to phosphorylate histone H1 in vitro. Indirect immunofluorescence staining demonstrated that HsCdc7 protein was predominantly localized in the nucleus. Although the expression levels of HsCdc7 appeared to be constant throughout the cell cycle, the protein kinase activity of HsCdc7 increased during S phase of the cell cycle at approximately the same time as that of Cdk2. These results, together with the functions of Cdc7p in yeast, suggest that HsCdc7 may phosphorylate critical substrate(s) that regulate the G1/S phase transition and/or DNA replication in mammalian cells.     

Urate crystal-induced chemotactic factor: isolation and partial characterization

A factor with chemotactic properties for neutrophils and mononuclear cells was extracted from the lysosomal fraction of both human and rabbit neutrophils that had been allowed to phagocytose monosodium urate crystals. The chemotactic factor was found to be a glycoprotein with a mol wt of 8,400 daltons. The factor is heat labile and has chemotactic activity for human as well as rabbit cells. Preincubation of the cells with the urate induced chemotactic factor or with complement activated plasma prevents the cell from migrating chemotactically when challenged with either factor in the chemotactic chamber. The chemotactic factor induces release of lysosomal enzymes for cytochalasin B treated human neutrophils.     

Autophosphorylation-independent activation of Acanthamoeba myosin I heavy chain kinase by plasma membranes

The three isoforms of Acanthamoeba myosin I (non-filamentous myosin with only a single heavy chain) express actin-activated Mg(2+)-ATPase activity only when phosphorylated at a single site by myosin I heavy chain kinase. The kinase is activated by autophosphorylation that is greatly stimulated by acidic phospholipids. Substantial fractions of the three myosins I and the kinase are associated in situ with membranes, and all four enzymes bind to purified membranes in vitro. We now report that when kinase and myosin I are incubated together with phosphatidylserine vesicles not only does the kinase autophosphorylate more rapidly than soluble kinase in the absence of phosphatidylserine but that, probably as a result, the kinase phosphorylates myosin I more rapidly than soluble kinase phosphorylates soluble myosin I. Similarly, plasma membrane-bound kinase phosphorylates membrane-bound myosin I and activates its actin-activated Mg(2+)-ATPase activity more rapidly than soluble kinase phosphorylates and activates soluble myosin I in the absence of membranes. However, the enhanced activity of membrane-bound kinase (which is comparable to the activity of kinase in the presence of phosphatidylserine) is not due to autophosphorylation of the membrane-bound kinase, which is very much slower than for kinase activated by phosphatidylserine vesicles.     

Identification and partial characterization of Arkansas isolates of chicken anemia virus

BACKGROUND: The underlying mechanism of enhanced antinociceptive effects and increased susceptibility to local anesthetics during pregnancy is not known. Mechanical, hormonal, biochemical, and neural changes have been suggested. The authors measured the susceptibility of individual spinal root axons to bupivacaine during late pregnancy in rats and compared them with similar measurements in nonpregnant rats. METHODS: Lumbar dorsal and ventral roots were excised from anesthetized pregnant and nonpregnant rats. Single-fiber dissection and recording techniques were used to isolate activity in individual axons. Supramaximal constant voltage stimuli were delivered to the distal end of the root. During in vitro perfusion, each root was exposed to increasing concentrations of bupivacaine, and the minimum blocking concentration (Cm) and the concentration that increased conduction latency by 50% (EC50) were measured. RESULTS: Myelinated and unmyelinated dorsal and ventral root axons of pregnant rats appeared to be less sensitive to steady-state conduction block and to the latency-increasing effects of bupivacaine than were equivalent axons from nonpregnant rats. Although when comparing specific axon types, only the difference in C-fibers was significant (Cm = 29.8 microM for pregnant and Cm = 22.1 microM for nonpregnant rats, P < 0.05; EC50 = 19.9 microM and 13.6 microM, respectively). CONCLUSIONS: In contrast to clinical expectations, the susceptibility to bupivacaine conduction block in individual dorsal and ventral root axons during late pregnancy in rats was not greater in pregnant animals. Pregnancy-related changes in diffusion barriers and activation of endogenous analgesic systems without changes in the electrophysiologic properties of spinal root axons are suggested as possible explanations for the discrepancy between clinical and experimental observations.     

Identification of a microtubule-binding domain in a cytoplasmic dynein heavy chain

As a molecular motor, dynein must coordinate ATP hydrolysis with conformational changes that lead to processive interactions with a microtubule and generate force. To understand how these processes occur, we have begun to map functional domains of a dynein heavy chain from Dictyostelium. The carboxyl-terminal 10-kilobase region of the heavy chain encodes a 380-kDa polypeptide that approximates the globular head domain. Attempts to further truncate this region fail to produce polypeptides that either bind microtubules or UV-vanadate cleave, indicating that the entire 10-kilobase fragment is necessary to produce a properly folded functional dynein head. We have further identified a region just downstream from the fourth P-loop that appears to constitute at least part of the microtubule-binding domain (amino acids 3182-3818). When deleted, the resulting head domain polypeptide no longer binds microtubules; when the excised region is expressed in vitro, it cosediments with added tubulin polymer. This microtubule-binding domain falls within an area of the molecule predicted to form extended alpha-helices. At least four discrete sites appear to coordinate activities required to bind the tubulin polymer, indicating that the interaction of dynein with microtubules is complex.     

Proteolytic release and partial characterization of human sperm-surface glycopeptides

Sperm-surface glycopeptides were obtained from intact sperm membranes after proteolytic release by different enzymatic treatments such as autoproteolysis, trypsin, papain and pronase. Glycopeptides were isolated, their properties and composition were examined, and their monosaccharide and amino acid constituents were characterized. The monosaccharides identified were fucose, mannose, galactose, N-acetylglucosamine, and N-acetylgalactosamine, which form part of more than one type of oligosaccharide units. Autoproteolytic treatment mainly provided O-glycosidic type oligosaccharides, while a mixture of O- and N-glycosidic oligosaccharides was obtained in variable proportions when treated with trypsin, papain or pronase. The highest degree of peptide cleavage was obtained with pronase. Despite the higher yields reached with trypsin, these glycopeptides contain the lowest percentage of oligosaccharide chains. Proteolytic treatment provides a simple, rapid procedure for the isolation of glycopeptides from the sperm surface.     

Identification and characterization of pancreatic eukaryotic initiation factor 2 alpha-subunit kinase, PEK, involved in translational control

In response to various environmental stresses, eukaryotic cells down-regulate protein synthesis by phosphorylation of the alpha subunit of eukaryotic translation initiation factor 2 (eIF-2alpha). In mammals, the phosphorylation was shown to be carried out by eIF-2alpha kinases PKR and HRI. We report the identification and characterization of a cDNA from rat pancreatic islet cells that encodes a new related kinase, which we term pancreatic eIF-2alpha kinase, or PEK. In addition to a catalytic domain with sequence and structural features conserved among eIF-2alpha kinases, PEK contains a distinctive amino-terminal region 550 residues in length. Using recombinant PEK produced in Escherichia coli or Sf-9 insect cells, we demonstrate that PEK is autophosphorylated on both serine and threonine residues and that the recombinant enzyme can specifically phosphorylate eIF-2alpha on serine-51. Northern blot analyses indicate that PEK mRNA is expressed in all tissues examined, with highest levels in pancreas cells. Consistent with our mRNA assays, PEK activity was predominantly detected in pancreas and pancreatic islet cells. The regulatory role of PEK in protein synthesis was demonstrated both in vitro and in vivo. The addition of recombinant PEK to reticulocyte lysates caused a dose-dependent inhibition of translation. In the Saccharomyces model system, PEK functionally substituted for the endogenous yeast eIF-2alpha kinase, GCN2, by a process requiring the serine-51 phosphorylation site in eIF-2alpha. We also identified PEK homologs from both Caenorhabditis elegans and the puffer fish Fugu rubripes, suggesting that this eIF-2alpha kinase plays an important role in translational control from nematodes to mammals.     

Identification and partial characterization of a proteinase specific for insulin-like growth factor binding protein-3 in aqueous and vitreous humors

The IGFs (-I and -II) are normally found in serum and other extracellular fluids complexed to specific binding proteins (IGFBPs). While several IGFBPs have been identified in vitreous and aqueous humors, the major serum carrier of IGF, IGFBP-3, is notably absent from these fluids. To determine if this paucity could be due to an IGFBP-3 proteinase (IGFBP-3ase), samples of bovine vitreous or aqueous humor were mixed with serum and incubated at 37 degrees C for 4 h followed by western ligand blotting. In these experiments, a distinct loss of the 46 kDa band representing IGFBP-3 was observed while other bands present at 35, 28 and 25 kDa were unaltered. The IGFBP-3ase activity is temperature sensitive, has a pH optimum of about 8.0 and is inhibited by EDTA. Acid treatment of serum to remove endogenously bound IGF does not affect the specificity or activity of the IGFBP-3 proteinase. Size exclusion chromatography of bovine aqueous indicates an approximate molecular weight of 260 kDa. Incubation of recombinant IGFBP-3 or serum with partially-purified IGFBP-3ase results in the appearance of low molecular weight fragments of approximately 30 kDa. These fragments are undetectable by western ligand blotting but are readily visualized using an IGFBP-3 specific antibody. Comparison of normal and diabetic vitreous humor reveals the presence of an increased amount of IGFBP-3 proteolytic fragments in the diabetic as compared to control. These findings indicate the presence of a IGFBP-3 proteinase in aqueous and vitreous humors that may be important in regulating ocular homeostasis.     

Myosin heavy chain gene expression in human heart failure

Two isoforms of myosin heavy chain (MyHC), alpha and beta, exist in the mammalian ventricular myocardium, and their relative expression is correlated with the contractile velocity of cardiac muscle. Several pathologic stimuli can cause a shift in the MyHC composition of the rodent ventricle from alpha- to beta-MyHC. Given the potential physiological consequences of cardiac MyHC isoform shifts, we determined MyHC gene expression in human heart failure where cardiac contractility is impaired significantly. In this study, we quantitated the relative amounts of alpha- and beta-MyHC mRNA in the left ventricular free walls (LVs) of 14 heart donor candidates with no history of cardiovascular disease or structural cardiovascular abnormalities. This group consisted of seven patients with nonfailing (NF) hearts and seven patients with hearts that exhibited donor heart dysfunction (DHD). These were compared with 19 patients undergoing cardiac transplantation for chronic end-stage heart failure (F). The relative amounts of alpha-MyHC mRNA to total (i.e., alpha + beta) MyHC mRNA in the NF- and DHD-LVs were surprisingly high compared with previous reports (33.3+/-18.9 and 35.4+/-16.5%, respectively), and were significantly higher than those in the F-LVs, regardless of the cause of heart failure (2.2+/-3.5%, P < 0.0001). There was no significant difference in the ratios in NF- and DHD-LVs. Our results demonstrate that a considerable amount of alpha-MyHC mRNA is expressed in the normal heart, and is decreased significantly in chronic end-stage heart failure. If protein and enzymatic activity correlate with mRNA expression, this molecular alteration may be sufficient to explain systolic dysfunction in F-LVs, and therapeutics oriented towards increasing alpha-MyHC gene expression may be feasible.     

Identification and molecular characterization of the p24 dynactin light chain

Intracellular transport along microtubules uses the motor proteins cytoplasmic dynein and kinesin. Cytoplasmic dynein is responsible for movement to the minus ends of microtubules and the evidence indicates that dynein interacts with another protein complex, dynactin. In order to better understand how these proteins function, we have sought to identify and clone the subunit polypeptides of these two complexes, in particular their light chains. Dynactin is made up of eight subunits of approximately 24,000 to 160,000 Da. In order to clone the p24 subunit, the components of purified dynactin were resolved by SDS polyacrylamide gel electrophoresis. The amino acid sequence of a tryptic peptide from the 24,000-Mr region of the gel was obtained and a candidate polypeptide identified by a screen of the databases. This polypeptide has a predicted molecular weight of 20,822 Da. Using an antibody to a different region of this protein, we demonstrate that it copurifies with microtubules and elutes from the microtubule pellet with characteristics similar to those of the dynactin complex and distinct from those of cytoplasmic dynein. This polypeptide co-sediments with dynactin on sucrose density gradients and it also co-immunoprecipitates with dynactin, but not with kinesin or cytoplasmic dynein. Together these results demonstrate that this polypeptide is the p24 subunit of dynactin. Analysis of the predicted amino acid sequence of p24 shows that it is a unique protein that has no significant similarity to known enzymes or other proteins. Structural analysis indicates that most of this protein will form an alpha-helix and that portions of the molecule may participate in the formation of coiled-coils. Since stoichiometric analysis of dynactin indicates that there is one molecule of p24 per dynactin complex, these characteristics suggest that this polypeptide may be involved in protein-protein interactions, perhaps in the assembly of the dynactin complex.     

Purification and characterization of human sarcomeric mitochondrial creatine kinase

PURPOSE: A dose-escalation study of irinotecan hydrochloride (CPT-11) combined with fixed-dose cisplatin was conducted to determine the maximum-tolerated dose (MTD), dose-limiting toxicities, and objective response rate in patients with advanced gastric cancer. PATIENTS AND METHODS: Twenty-four patients with or without prior chemotherapy were enrolled. All patients were assessable for toxicities and response. On day 1, CPT-11 was administered as a 90-minute intravenous (I.V.) infusion, which was followed 2 hours later by a 120-minute I.V. infusion of cisplatin 80 mg/m2. CPT-11 alone at the same dose was administered again on day 15. The treatment was repeated every 4 weeks until disease progression was observed. The initial dose of CPT-11 was 60 mg/m2, and was escalated in increments of 10 mg/m2 until severe or life-threatening toxicity was observed. RESULTS: The MTD of this combination was CPT-11 80 mg/m2. At this dose level, 16.7% of patients (two of 12) had leukopenia of less than 1,000/microL, 66.7% (eight of 12) had neutropenia of less than 500/microL, and 16.7% (two of 12) had severe diarrhea of grade 4 during the first course. The dose-limiting toxicity was neutropenia. Ten patients achieved a partial response (PR), and the overall response rate was 41.7% among 24 patients (95% confidence interval, 21.9% to 61.4%). CONCLUSION: The recommended dose and schedule is CPT-11 70 mg/m2 on days 1 and 15 and cisplatin 80 mg/m2 on day 1 every 4 weeks. This combination of CPT-11 and cisplatin, considered to be active against advanced gastric cancer with acceptable toxicity, should be further assessed in a phase II study.