Pertussis toxin inhibition of chemotactic factor-induced calcium mobilization and function in human polymorphonuclear leukocytes.

Chemotactic factors stimulate a rapid increase in the cytosolic concentration of intracellular calcium ions ([Ca2+]in) in human polymorphonuclear leukocytes (PMNL), which may be an event that is critical to the expression of chemotaxis and other PMNL functions. Treatment of PMNL with pertussis toxin catalyzes ADP-ribosylation of a protein similar or identical to the inhibiting regulatory protein of adenylate cyclase, Gi, and suppresses the increase in [Ca2+]in elicited by leukotriene B4(LTB4) and formyl-methionyl-leucyl-phenylalanine. Chemotactic migration and lysosomal enzyme release elicited by chemotactic factors were inhibited by pertussis toxin with a concentration-dependence similar to that for inhibition of the increase in [Ca2+]in, without an effect on lysosomal enzyme release induced by the ionophore A23187 and phorbol myristate acetate. Activated pertussis toxin catalyzed the [32P]ADP-ribosylation of a 41 kD protein in homogenates of PMNL. The extent of [32P]ADP-ribosylation of this protein was reduced 59% by pretreatment of intact PMNL with pertussis toxin. Pertussis toxin selectively decreased the number of high-affinity receptors for LTB4 on PMNL by 60% without altering the number or binding properties of the low-affinity subset of receptors. Pertussis toxin modification of a membrane protein of PMNL analogous to Gi thus simultaneously alters chemotactic receptors and attenuates the changes in cytosolic calcium concentration and PMNL function caused by chemotactic factors.

Function of NAD glycohydrolase in ADP-ribose uptake from NAD by human erythrocytes.

The function of the ectoenzyme NAD glycohydrolase (NADase) in ADP-ribose uptake from extracellular NAD was studied in human erythrocytes that express relatively high NADase activity (adult erythrocytes) and erythrocytes expressing very low activity (newborn erythrocytes). The rates of ADP-ribose uptake from NAD in human erythrocytes were correlated with their NADase activities. In contrast, there was no significant difference in the rates of ADP-ribose uptake among these cells when incubated with ADP-ribose. These results indicate that ecto-NADase may have a role as supplier of ADP-ribose for its uptake into the cells and that the cleavage of NAD by NADase is necessary for the ADP-ribose uptake by human erythrocytes. From ADP-ribose uptake studies at 37 degrees C a Km of 0.7 +/- 0.05 microM and a Vmax of 2.04 +/- 0.1 pmol/min per microliter cell water was found for the uptake of [3H]ADP-ribose. The thiol-reactive reagents p-chloromercuribenzene sulfonic acid and N-ethylmaleimide inhibited the uptake ADP-ribose with IC50 values of 50 +/- 4 and 750 +/- 25 mM, respectively. Since efflux of [3H]ADP-ribose was negligible, the ADP-ribose transport system appears to be unidirectional. The unidirectionality was supported by the evidence that transported ADP-ribose was rapidly degraded to AMP which is impermeable to the membrane.

Effect of pertussis toxin on the phosphodiesteratic cleavage of the polyphosphoinositides by guanosine 5'-O-thiotriphosphate and thrombin in permeabilized human platelets.

It has recently been shown in this laboratory that permeabilization of human platelets with 15-25 micrograms/ml saponin allows ADP-ribosylation by pertussis toxin of the alpha i-subunit of Gi (Ni), a guanine nucleotide-binding regulatory protein. The same assay conditions have been used to determine phospholipase C in permeabilized platelets. Guanosine 5'-O-thiotriphosphate- (GTP[gamma S]-) activated phospholipase C in permeabilized platelets whose inositol phospholipids were prelabeled with [3H]inositol. Phospholipase C activity was measured by [3H]polyphosphoinositide decreases and formation of [3H]inositol bisphosphate and [3H]inositol trisphosphate. Prostacyclin, cyclic AMP or pretreatment of permeabilized platelets with pertussis toxin did not alter this effect under conditions in which the alpha i-subunit was effectively ADP-ribosylated by pertussis toxin. This information indicated that ADP-ribosylation of Gi-protein was not directly related to activation or inhibition of platelet phospholipase C by GTP [gamma S]. Thrombin also activated phospholipase C in permeabilized platelets and, surprisingly, this action was enhanced by pertussis toxin pretreatment. This indicates that ADP-ribosylation of Gi-protein facilitates the action of thrombin on phospholipase C.

The stimulatory and inhibitory guanine nucleotide-binding proteins of adenylate cyclase in erythrocytes from patients with pseudohypoparathyroidism type I.

Both the inhibitory and stimulatory guanine nucleotide-binding proteins of the adenylate cyclase complex were measured in erythrocyte membranes from patients with pseudohypoparathyroidism (PHP). The inhibitory guanine nucleotide-binding protein (Ni) of adenylate cyclase was measured by incorporation of [32P]ADP-ribose from [32P]NAD into the 39K subunit of Ni catalyzed by pertussis toxin. The ADP-ribosyltransferase activity of the toxin was expressed through incubation with dithiothreitol and erythrocyte membranes. Erythrocytes from 12 patients with PHP type I (PHP-I) had Ni values similar to those of 9 normal subjects and 2 patients with pseudopseudohypoparathyroidism. In 6 PHP-I patients, decreased activity of the stimulatory guanine nucleotide-binding protein (Ns) of adenylate cyclase, as determined by reconstitution of adenylate cyclase in the Ns-deficient membranes of cyc-S49 cells, corresponded with the reduced degree of ADP-ribosylation of the 42K subunit of Ns catalyzed by cholera toxin. These data suggest that the defect of Ns results in reduced stimulation of adenylate cyclase in some PHP-I patients, and that enhanced inhibition of the enzyme due to an increase in the 39K subunit of Ni does not account for the biochemical lesion in PHP-I patients.

Guanine nucleotide binding proteins in opioid-dependent patients.

Guanine nucleotide binding (G) protein levels and functioning in the platelets of 19 methadone-maintained patients were compared to age and sex matched, normal controls. We found that in the methadone patients, G alpha s-levels were significantly higher, while the levels of G alpha i 1/2 and pertussis toxin catalyzed [32P]ADP ribosylation were significantly lower compared to control subjects in platelet membranes. We have further found that when all three of these biochemical indicators were combined in a discriminant function analysis, 79% of the methadone patients were correctly classified and 83% of the controls were correctly classified.

Identification in human erythrocytes of mono(ADP-ribosyl) protein hydrolase that cleaves a mono(ADP-ribosyl) Gi linkage.

A novel enzymatic activity, the hydrolysis of linkages between mono(ADP-ribose) and cysteine residues in Gi prepared by eukaryotic ADP-ribosyltransferase C [(1988) J. Biol. Chem. 263, 5485-5489] was found in the cytosol of human erythrocytes. The mono(ADP-ribosyl) Gi hydrolase, tentatively named ADP-ribosyl protein hydrolase C was partially purified by sequential chromatographies on DEAE-cellulose and Blue Sepharose. This enzyme catalyzes the release of ADP-ribose from mono(ADP-ribosyl) Gi. Its activity was enhanced by Ca2+ and inhibited by ADP-ribose. The presence of this enzyme in eukaryotic cells suggests that endogenous mono(ADP-ribosyl)ation of Gi is a reversible post-translational modification.

Alpha 2A-adrenergic receptors activate protein kinase C in human platelets via a pertussis toxin-sensitive G-protein.

4,4'-Diisothiocyanato-stilbene-2,2'-disulfonic acid (DIDS) stimulates human platelets via alpha 2A-adrenergic receptor-mediated activation of protein kinase C (PKC) independent of the phospholipase C pathway. Here we show, that in permeabilized platelets activation of PKC by DIDS (20 microM), measured as 32P incorporation in pleckstrin, is completely inhibited by guanosine 5'-(2-O-thio)diphosphate (200 microM), an inhibitor of heterotrimeric G-proteins. Also pertussin toxin (4 micrograms/ml), which ADP-ribosylates the alpha-subunits of Gi's and Go, prevents pleckstrin phosphorylation by DIDS. N-Ethylmaleimide (50 microM), which uncouples Gi from alpha 2A-adrenoceptors, inhibits pleckstrin phosphorylation by DIDS in intact platelets. Activation of PKC by 55 nM phorbol 12-myristate 13-acetate and 500 nM platelet-activating factor are not disturbed by NEM. DIDS inhibits by 40 +/- 5% (n = 4) the pertussis toxin-catalyzed [32P]ADP-ribosylation of a 41 kDa protein fraction previously shown to contain the alpha-subunits of Gi alpha-1, Gi alpha-2 and Gi alpha-3. Thus, the alpha 2A-adrenergic receptor activates PKC via a G-protein of the Gi-family.

Thrombin induces the association of cyclic ADP-ribose-synthesizing CD38 with the platelet cytoskeleton.

The effect of platelet stimulation on the subcellular localization of CD38, a membrane glycoprotein that catalyses the synthesis of cyclic ADP-ribose from beta-NAD+ was investigated. Treatment of human platelets with thrombin caused the association of about 40% of the total ADP-ribosyl cyclase activity with the cytoskeleton, through the translocation of the CD38 molecule from the Triton X-100-soluble to the insoluble fraction. The interaction of CD38 with the cytoskeleton was a specific and reversible process, mediated by the binding to the actin-rich filaments and was inhibited by treatment of platelets with cytochalasin D. This event was regulated by integrin alphaIIb beta3 and platelet aggregation as it was prevented by the inhibition of fibrinogen binding and was not observed in platelets from a patient affected by Glanzmann thrombasthenia. These results demonstrate that the subcellular localization of CD38 can be influenced by platelet stimulation with physiological agonists, and that membrane CD38 can interact with intracellular proteins.

G-protein level quantification in platelets and leukocytes from patients with panic disorder.

The objective of our study was to investigate if there are abnormalities in signal transducing G proteins in patients with panic disorder. We utilized selective antibodies to quantitate the levels of the G protein alpha subunits that regulate adenylyl cyclase activity (G alpha s and G alpha i2) and phosphoinositide turnover (G alpha q/11) in platelet membranes (and leukocyte membranes for G alpha s), and also carried out pertussis toxin (PT) catalyzed [32P]ADP-ribosylation in platelet membranes from a group of 13 untreated panic disorder patients, 10 untreated social phobia patients, and 12 healthy subjects. There were no significant differences among the three groups in the immunolabeling of G alpha s in platelets or leukocytes, or in the immunolabeling of G alpha i1/2, G alpha q/11, or PT-catalyzed [32P]ADP-ribosylation in platelets. Within the constraints imposed by using peripheral blood cells to reflect brain composition, our results do not provide support for G protein abnormalities in patients with panic disorder. These results contrast with those obtained using identical methodology in bipolar affective disorder, where elevated G alpha s in leukocytes has been reported (Manji et al. 1995).

Expression of nitric oxide synthase in human peripheral blood mononuclear cells and neutrophils.

It has been clearly demonstrated in rodents that nitric oxide (NO) plays an important role in host defense and immunity. However, evidence that human leukocytes express inducible nitric oxide synthase (iNOS) or its products has been inconclusive and a source of controversy. We report that iNOS could not be detected in human monocytes, HL-60 cells, neutrophils, and T cells by Western blotting analysis (< or = 10 pg) or by radiolabeled L-arginine-to-L-citrulline conversion (< or = 20 pmol L-citrulline) under conditions sufficient to induce iNOS in the rodent system and in human hepatocytes, which include activation with cytokines, endotoxins, and/or chemoattractants. However, sensitive methods such as RT-PCR and Northern blot analysis show "constitutively expressed" iNOS mRNA from human monocytes, neutrophils, Jurkat cells, and HL-60 cells. This iNOS mRNA is 4.4 kb and is similar to that seen in human hepatocytes and rodent macrophages. In spite of the constitutive expression of mRNA in neutrophils and the lack of detectable NOS activity (based on Western blotting and L-arginine-to-L-citrulline conversion assay), stimulation of human neutrophils unit FMLP in vitro induced the ADP-ribosylation of an intracellular NO target, glyceraldehyde-3-PO4 dehydrogenase (GAPDH), in a NO-dependent manner. These studies indicate that low levels of NOS protein are expressed in neutrophils (and perhaps T cells and monocytes) and produce NO following stimulation. The data indicate that, in addition to its phagocytic and tumoricidal activity. NO may also function as an autacoid signaling molecule within the cells.

NAD glycohydrolase activities and ADP-ribose uptake in erythrocytes from normal subjects and cancer patients.

Erythrocytes from cancer patients exhibited up to fivefold higher NAD glycohydrolase activities than control erythrocytes from normal subjects and also similarly increased [14C] ADP-ribose uptake values. When [adenosine-14C] NAD was used instead of free [14C] ADP-ribose, the uptake was dependent on ecto-NAD glycohydrolase activity. This was reflected in the inhibition of ADP-ribose uptake from [adenosine-14C] NAD by Cibacron Blue. ADP-ribose uptake in erythrocytes appeared to be complex: upon incubation with free [14C] ADP-ribose, the radiolabel associated with erythrocytes was located in nearly equal parts in cytoplasm and plasma membrane. Part of [14C] ADP-ribose binding to the membrane was covalent, as indicated by its resistance to trichloroacetic acid-treatment. A preincubation with unlabeled ADP-ribose depressed subsequent erythrocyte NAD glycohydrolase activity and binding of [14C] ADP-ribose to erythrocyte membrane; but it failed to inhibit the transfer of labeled ADP-ribose to erythrocyte cytoplasm. On the other hand, incubation with [adenosine-14C] NAD did not result in a similar covalent binding of radiolabel to erythrocyte membrane. In line with this finding, a preincubation with unlabeled NAD was not inhibitory on subsequent NAD glycohydrolase reaction and ADP-ribose binding. ADP-ribose binding and NAD glycohydrolase activities were found also in solubilized erythrocyte membrane proteins and, after size fractionation, mainly in a protein fraction of around 45kDa-molecular weight.

Ns and Ni, the stimulatory and inhibitory regulatory components of adenylyl cyclases. Purification of the human erythrocyte proteins without the use of activating regulatory ligands.

Methods were developed to adequately extract, separate and, without the use of NaF as stabilizing agent, purify to better than 90% purity human erythrocyte Ns and Ni, the stimulatory and inhibitory guanine nucleotide- and Mg-binding regulatory components of adenylyl cyclases, as well as a protein containing Mr = 35,000 subunits. On the basis of a functional assay for Ns, it was purified about 5,000-fold from starting washed erythrocyte membranes with a yield of about 10%. A typical purification yields from 60 units of outdated human blood, between 500 and 1,000 micrograms of pure Ns, and a similar amount of Ni. Pure Ns and Ni contain each at least one alpha and one beta subunit (Northup, J.K., Sternweis, P.C., Smigel, M.D., Schleifer, L.S., Ross, E.M., and Gilman, A.G. (1980) Proc. Natl. Acad. Sci. U.S.A. 74, 6516-6520; Codina, J., Hildebrandt, J.D., Iyengar, R., Birnbaumer, L., Sekura, R.D., and Manclark, C.R. (1983) Proc. Natl. Acad. Sci. U.S.A. 77, 4276-4280). Polyacrylamide gel electrophoresis in the presence of sodium dodecyl sulfate at varying acrylamide concentrations yielded Mr values of 42,000 and 40,000 for the alpha subunits of Ns and Ni, and of 35,000 for the beta subunits of Ns and Ni. Two-dimensional thin layer analysis of tryptic peptides obtained from digesting 125I-labeled subunits of Ns and Ni confirmed the finding of Manning, D., and Gilman, A.G. (1983) J. Biol. Chem. 258, 7059-7063) that while their alpha subunits are clearly different, their beta subunits are the same. Hydrodynamic analysis of the molecular weights of the nondenatured proteins showed behavior consistent with Mr = 95,500 for Ns, the same for Ni, and Mr = 40,000 for the protein containing the Mr = 35,000 beta subunit. Sedimentation coefficients and Stokes radii of the purified Ns were indistinguishable from those of Ns activity present in initial cholate extracts from human erythrocyte membranes. Further, the overall kinetics with which Ns activity in cholate extracts and Ns activity in the purified protein reconstituted the Ns-deficient adenylyl cyclase system of cyc- S49 cells was also indistinguishable. We conclude that we have purified the native unactivated form of Ns, and by serendipity the Ni, as well as a protein containing the 35 kDa beta subunit of Ns and Ni.(ABSTRACT TRUNCATED AT 400 WORDS)

Identification of a 42K phosphoprotein of platelets modulated by collagen: the alpha-subunit of pyruvate dehydrogenase.

Platelets exposed to collagen sufficient to stimulate the release reaction show an increase in labeling of two intracellular proteins with molecular weights of 20,000 and 42,000. The 20,000 Mr protein has already been identified as the light chain of myosin whereas the identity of the 42,000 Mr protein had not been established. By use of biochemical and immunological techniques, the identify of the 42,000 Mr component of prelabeled platelets found in the 100,000g supernatant of freeze-thawed or sonicated cells appears to be one of the subunits of pyruvate dehydrogenase complex which is translocated from the mitochondria to the 100,000g supernatant during the preparative procedure. Increased phosphorylation of the 42,000 Mr protein occurred after collagen stimulation and was accompanied by diminished pyruvate dehydrogenase activity.

NADPH: a stimulatory cofactor for nitric oxide-induced ADP-ribosylation reaction.

An endogenous ADP-ribosyltransferase is present in the cytosolic fraction of human platelets. Agents known to release nitric oxide activated this ADP-ribosylation reaction in a cGMP-independent fashion. This enzymatic activity was further enhanced by the addition of NADPH to the platelet cytosolic fraction. Interestingly, NADPH was unable to replace DTT, which has been described as an essential cofactor. Our results indicate that NADPH is a stimulatory factor of the endogenous ADP-ribosylation reaction. NADPH shifts the dose-response curve of NO to the left and possibly increases, in this way, the ADP-ribosylation reaction under physiological conditions.

Nitric oxide causes ADP-ribosylation and inhibition of glyceraldehyde-3-phosphate dehydrogenase.

Nitric oxide and nitric oxide-generating agents like 3-morpholinosydnonimine (SIN-1) stimulate the mono-ADP-ribosylation of a cytosolic, 39-kDa protein in various tissues. This protein was purified from human platelet cytosol by conventional and fast protein liquid chromatography techniques. N-terminal sequence analysis identified the isolated protein as the glycolytic enzyme glyceraldehyde-3-phosphate dehydrogenase (GAPDH). Nitric oxide stimulates the auto-ADP-ribosylation of GAPDH in a time and concentration-dependent manner with maximal effects after about 60 min. Associated with ADP-ribosylation is a loss of enzymatic activity. NAD(+)-free enzyme is not inhibited by SIN-1, indicating the absolute requirement of NAD+ as the substrate of the ADP-ribosylation reaction. Inhibition of the glycolytic enzyme GAPDH may be relevant as a cytotoxic effect of NO complementary to its inhibitory actions on iron-sulfur enzymes like aconitase and electron transport proteins of the respiratory chain.

Adenosine diphosphate ribulose in human erythrocytes: a new metabolite with membrane binding properties.

Incubation of ADPribose with yeast phosphoriboisomerase resulted in the formation of an adenylic nucleotide that was identified with ADPribulose by mass spectrometry. Synthesis of [32P]ADPribulose from [32P]NAD+ by the combined activities of commercial NAD+ glycohydrolase and phosphoriboisomerase allowed us to use it as a labeled internal standard throughout the procedure of purification from trichloroacetic acid extracts of human red blood cells. ADPribulose was purified by means of three sequential reverse phase HPLC separations and its concentration in human erythrocytes was estimated to be 0.11 +/- 0.1 microM. Unsealed erythrocyte ghosts did not transform ADPribulose, which bound to specific membrane proteins with a trichloroacetic and formic acid-resistant binding. The labeled proteins were identified as spectrin, bands 3, 4.1, 4.2 and Glyceraldehyde 3-phosphate dehydrogenase on the basis of their relative mobilities on SDS-PAGE.

Presence and turnover of adenosine diphosphate ribose in human erythrocytes.

ADP-ribose was detected in human red blood cells (RBC) at 0.45 +/- 0.1 microM concentrations. These levels could be estimated after purification of ADP-ribose by means of three sequential HPLC fractionations of RBC extracts. Extraction was performed by sonication of RBC either in trichloroacetic acid, followed by centrifugation, or in carbonate-bicarbonate buffer, pH 10.0, followed by rapid ultrafiltration. Neither procedure of extraction caused artefactual formation of ADP-ribose. Prolonged incubation of intact RBC in isotonic buffer containing labeled orthophosphate resulted in the slow incorporation of radioactivity into ADP-ribose. Identification of the labeled ADP-ribose was confirmed upon incubation of the purified metabolite with nucleotide pyrophosphatase, yielding radioactive 5'-AMP and ribose 5-phosphate, while its exposure to a nonspecific deaminase resulted in the quantitative formation of labeled inosine diphosphate ribose.

Botulinum ADP-ribosyltransferase C3. Purification of the enzyme and characterization of the ADP-ribosylation reaction in platelet membranes.

A novel ADP-ribosyltransferase C3 was purified to homogeneity from filtrates of certain strains of Clostridium botulinum type C by ammonium sulfate precipitation, gel filtration, ion-exchange chromatography and heat treatment. The molecular mass of botulinum ADP-ribosyltransferase C3 was found to be 25 kDa. In the presence of [32P]NAD but not with [carbonyl-14C]NAD, C3 labelled 21-24-kDa protein(s) in membranes of human platelets and other tissues. The Km value of the ADP-ribosylation reaction for NAD was about 2 microM. Labelling of the 21-24-kDa protein(s) by C3 was largely reduced by addition of nicotinamide. Snake venom phosphodiesterase cleaved the ADP-ribose attached to the 21-24-kDa protein(s) by C3 and released 5'AMP. C3 catalyzed hydrolysis of [carbonyl-14C]NAD and released [carbonyl-14C]nicotinamide. ADP-ribosylation of 21-24-kDa platelet membrane protein(s) was biphasically regulated by Mg2+, Mn2+ and Ca2+. In the absence of free divalent cations GTP, GTP[gamma S] and GDP but not GDP[beta S], GMP, ATP or ATP[gamma S] increased labelling by C3. In the presence of Mg2+, GTP[gamma S] was inhibitory. Guanine nucleotides prevented heat inactivation of the substrate protein(s) with the rank order GTP[gamma S] = GTP = GDP greater than GDP[beta S] greater than GMP much greater than ATP = GMP = ATP[gamma S]. The data support the view that the novel ADP-ribosyltransferase C3 modifies eukaryotic 21-24-kDa GTP-binding protein(s).

The inhibitory guanine nucleotide-binding regulatory component of adenylate cyclase. Subunit dissociation and guanine nucleotide-dependent hormonal inhibition.

The inhibitory and stimulatory guanine nucleotide-binding regulatory components (Gi and Gs) of adenylate cyclase both have an alpha X beta subunit structure, and the beta (35,000 Da) subunits are functionally indistinguishable. Gi and Gs both dissociate in the presence of guanine nucleotide analogs or Al3+, Mg2+, and F- in detergent-containing solutions. Several characteristics of Gi- and Gs-mediated regulation of adenylate cyclase activity have been studied in human platelet membranes. The nonhydrolyzable analog of GTP, guanosine-5'-(3-O-thio)triphosphate (GTP gamma S) mimics GTP-dependent hormonal inhibition or stimulation of adenylate cyclase under appropriate conditions. This inhibition or stimulation follows a lag period. The combined addition of epinephrine or prostaglandin E1 with GTP gamma S results in the immediate onset of steady state inhibition or activation. The effects of the GTP analog are essentially irreversible. Fluoride is also an effective inhibitor of prostaglandin E1-stimulated adenylate cyclase, while it markedly stimulates the basal activity of the enzyme. The addition of the resolved 35,000-Da subunit of Gi to membranes results in inhibition of adenylate cyclase, and the resolved 41,000-Da subunit has a stimulatory effect on enzymatic activity. The inhibitory action of the 35,000-Da subunit is almost completely abolished in membranes that have been irreversibly inhibited by GTP gamma S plus epinephrine; this irreversible inhibition is almost completely relieved by the 41,000-Da subunit. Detergent extracts of membranes that have been treated with GTP gamma S plus epinephrine contain free 35,000-Da subunit. The 41,000-Da subunit of Gi contained in such extracts has a reduced ability to be ADP-ribosylated by islet-activating protein (IAP), which implies that this subunit is in the GTP gamma S-bound form. The irreversible inhibition of adenylate cyclase caused by GTP gamma S (plus epinephrine) in membranes is highly correlated with the liberation of free 35,000-Da subunit activity and is inversely related to the 41,000-Da IAP substrate activity in detergent extracts prepared therefrom. The increase in free 35,000-Da subunit activity in extracts and the inhibition of adenylate cyclase activity in GTP gamma S (plus epinephrine)-treated membranes are both markedly inhibited by treatment with IAP.(ABSTRACT TRUNCATED AT 400 WORDS)