Divergence of ANF analogs in smooth muscle cell cGMP response and aorta vasorelaxation: evidence for receptor subtypes

ANF analog potencies in stimulating smooth muscle cell cGMP were compared with the ability to relax histamine-constricted rabbit aorta in vitro. ANF[1-28], [5-28], [5-27] and Lys-11[5-28] elevated cGMP and were potent vasorelaxants. ANF[7-23] and Lys-11[7-23] were potent cGMP stimulators but 1000-fold weaker relaxants. Tyr-8[5-27] did not stimulate cGMP synthesis or antagonize the response of the other peptides, yet was a potent vasorelaxant. Crosslinking with 125I-ANF identified bands at 150 and 65 KD by SDS-PAGE. ANF[1-28], Lys-11[7-23] and Tyr-8[5-27] blocked crosslinking at low concentration despite disparate activities. These data support the existence of ANF receptor subtypes and suggest that cGMP elevation alone is not sufficient to promote atrial peptide-induced vasorelaxation.     

Atrial peptides induce mast cell histamine release.

Human atrial natriuretic peptide [ANF(1-28)] contains five arginine residues and carries an overall positive change of four. It was hypothesized that atrial peptides may induce mast cell histamine release. In vitro, three atrial peptides [ANF(1-28), (3-28) and (5-28)] were demonstrated to induce dose-dependent histamine release from isolated rat peritoneal mast cells. In vivo, ANF(3-28) produced a dose-dependent increase in rat skin permeability which was blocked by antagonists of histamine and serotonin. The results indicate atrial peptides are capable of inducing mast cell degranulation in a manner similar to that described for other positively charged peptides.     

Studies of murine erythroid cell development. Synthesis of heme and hemoglobin

Techniques of cell separation were used to isolate murine erythroid precursors at different states of maturation. Cells were studied before and after short-term incubation in the presence or absence of erythropoietin. Complementary results were obtained by direct examination of the cell fractions and by the short-term culture experiments. Indices of heme synthesis, including incorporation of 59Fe or [2-14C]glycine into heme and activity of delta-aminolevulinic acid synthetase, were already well developed in the least mature cells, chiefly pronormoblasts. Activity then rose moderately in the cell fractions consisting primarily of basophilic and polychromatophilic normoblasts, and fell off with further increases in cell maturity. On short-term culture in the presence of erythropoietin, activity declined with increasing cell maturation except in the least mature fraction where the original level of activity was maintained. By contrast, synthesis of labeled hemoglobin ([3H]leucine) was very low in the least mature cell fractions and rose progressively with increasing cell maturity. The rate of hemoglobin synthesis increase in cells at all stages of maturation when cultured in the presence of erythropoietin. Despite the different patterns observed for heme synthesis and hemoglobin synthesis, both synthetic activities were consistently higher in cells cultured with erythropoietin as compared to controls. These findings suggest that erythropoietin stimulates biochemical differentiation of erythroid precursors at various stages of maturation. They also demonstrate an asynchronism between heme synthesis and hemoglobin syhthesis; heme synthesis is already well developed in the least mature erythroid cells and begins to diminish as the capacity for hemoglobin synthesis continues to rise.     

High salt intake-induced changes in atrial natriuretic factor kinetics are mediated by clearance receptors.

We have reported a paradoxical plasma atrial natriuretic factor (ANF) decline following prolonged high salt intake that was attributed to an increased tissue uptake of circulating ANF, leading to its augmented distribution volume (Vas) and metabolic clearance rate (MCR) as compared with control rats on a standard diet. To explore this phenomenon further, we evaluated possible chronic salt-loading-induced changes in ANF clearance (C-ANF) receptors, which appear to play a major role in ANF removal from the circulation. We studied changes in plasma [125I]ANF(1-28) and its pharmacokinetics after preoccupation of C-ANF receptors by its specific ligand, C-ANF(4-23), in high-salt-treated rats and their controls. Following C-ANF(4-23) administration, we detected significantly higher circulating [125I]ANF levels throughout the study period (8 min) in high-salt-fed rats compared with the controls (280-470% vs 100-215% increase of basal values, P less than 0.05). C-ANF(4-23) infusion caused a significantly greater decrease of the metabolic clearance rate and distribution volume of [125I]ANF in high-salt-fed rats than in control animals (74 +/- 6% vs 41 +/- 6% and 75 +/- 4% vs 50 +/- 5% of basal values, respectively; P less than 0.05). These data suggest that a prolonged high salt diet may increase the availability of C-ANF receptors and, through this mechanism, may negatively modulate plasma ANF concentrations. C-ANF receptors may thus fulfill a regulatory function on circulating ANF during prolonged salt loading in rats.     

The isolation and characterisation of human atrial natriuretic factor produced as a fusion protein in Escherichia coli

Human atrial natriuretic factor [ANF(1-28)] has been isolated from a fusion protein produced in Escherichia coli. ANF(1-28) was linked to a naturally occurring E. coli protein, chloramphenicol acetyltransferase, via unique cleavage sequences susceptible to either human thrombin digestion, or the chemical action of 2-(2-nitrophenylsulphenyl)-3-methyl-3'-bromoindolenine (BNPS-skatole). The linker sequences were Gly-Val-Arg-Gly-Pro-Arg and Trp respectively. The liberated ANF was purified by reversed-phase HPLC. Optimised cleavage conditions released 5-10% (by mass) of the maximal yield of ANF(1-28) from the fusion protein with the thrombin-susceptible linker, whilst a 2-5% (by mass) yield was observed from the fusion protein with the tryptophan linker after BNPS-skatole treatment. The purified cleavage products were biologically active and shown to comprise intact ANF(1-28). Fast-atom-bombardment mass spectrometry confirmed [MH]+ of 3079 m/z, consistent with ANF(1-28).     

Purification and characterization of the erythropoietin-sensitive membrane phosphoprotein, pp43.

We have shown previously that purified human erythropoietin rapidly alters the phosphorylation of an integral erythroid membrane protein, pp43 (Choi, H.-S., Wojchowski, D. M., and Sytkowski, A. J. (1987) J. Biol. Chem. 262, 2933-2936). We have now purified pp43 to apparent homogeneity and have prepared antibodies to it. After sodium dodecyl sulfate-polyacrylamide gel electrophoresis and electrophoretic transfer of membrane proteins to nitrocellulose, the antibodies identified pp43 and a series of higher molecular weight antigenically related proteins, up to 50 kDa, in erythropoietin-responsive Rauscher murine erythroleukemia cells and in normal murine erythroid cells. Examination of purified subcellular fractions confirmed the localization of pp43 and the related proteins to the plasma membrane. Phosphorylation with [gamma-32P]ATP demonstrated that, in contrast to pp43, these higher molecular weight proteins were not phosphorylated. Marked differences in both the abundance of pp43 and related proteins and the degree of erythropoietin-sensitive pp43 phosphorylation were found between the plasma membranes of Rauscher cells and those of "non-responsive" Friend murine erythroleukemia cells. In addition only trace amounts of a 50-kDa antigenically related protein and no phosphorylated pp43 were detected in the plasma membranes of two erythropoietin-insensitive human erythroid cells lines, K562 and HEL. The results suggest that the abundance and degree of phosphorylation of pp43 and the antigenically related proteins is strongly correlated with the erythropoietin responsiveness of the particular erythroid cell types.     

Atrial natriuretic factor (ANF) binding sites in frog kidney and adrenal.

Atrial natriuretic factor (ANF) binding sites were localized and quantified in kidney and adrenal of the frog Rana temporaria by quantitative in vitro autoradiography. [125I]-rat ANF(99-126) binding was present in kidney glomeruli and in the outer layer of interrenal tissue in the adrenal gland. ANF binding exhibited positive cooperativity with a half-maximal binding concentration (EC50) of 102 +/- 16 pM in glomeruli and 93 +/- 19 pM in interrenal tissue (n = 8). The corresponding maximal binding capacities (Bmax) were 1.33 +/- 0.16 and 1.21 +/- 0.36 fmol/mm2. [125I]-Rat ANF(99-126) binding was competitively displaced by unlabeled ANF analogues with an intact disulfide bridge showing a lower affinity than the iodinated ligand. The presence of ANF binding in glomeruli and steroidogenic interrenal cells suggests physiological functions of ANF for the osmomineral regulation in the frog by influencing glomerular filtration rate and adrenal steroid secretion.     

Hemoglobin switching in sheep and goats. V. Effect of erythropoietin concentration on in vitro erythroid colony growth and globin synthesis

Erythroid colonies were generated in response to erythropoietin in plasma clot cultures of sheep and goat bone marrow cells. At low concentration erythropoietin only hemoglobin A (betaA globin) was synthesized in goat cultures, but at high concentrations 50% of the hemoglobin synthesized was hemoglobin C (betaC globin). This effect of erythropoietin on the expression of a specific beta globin gene was manifested only after 72 h in vitro and followed the development of erythroid colonies. Sheep colonies behaved differently from those of goat in that little or no betaC globin synthesis occurred even at high erythropoietin concentration. To investigate this difference, sheep marrow cells were fractionated by unit gravity sedimentation. The erythroid colony-forming cells sedimented more rapidly (3.5-6mm/h) than the hemoglobinized eththroid precursors (1-3.5 mm/h), suggesting that the colonies were formed from an early erythroid precursor, However, the colonies formed from the sheep marrow fractions synthesized only betaA globin even at concentrations of erythropoietin sufficient to stimulate betaC globin synthesis in goat colonies. Morphologically, the goat colonies were larger and more mature than those of the sheep. By 96 h in vitro three-fourths of the goat colonies contained enucleated red cells compared to only 3% of the sheep colonies. Thus, erythropoietin had an equivalent effect in stimulating erythroid colony growth from the marrow of both species although there were both biochemical and morphological differences between the colonies. Hemoglobin switching appeared to require exposure of an early precursor to high erythropoietin concentration, but the results with sheep marrow suggested that the rate of colony growth and cellular maturation might also be important.     

The role of posture on the changes in plasma atrial natriuretic factor and arginine vasopressin levels during immersion

In seven healthy male volunteers we investigated changes in plasma atrial natriuretic factor [( ANF]), arginine vasopressin [( AVP]) and plasma volume (PV) during supine immersion. Twenty minutes head-out water immersion in a supine position in a thermo-neutral water bath attenuated the increase in PV induced by 20 min in a supine position in air, but increased the mean plasma [ANF] from 32.0 pg.ml-1, SEM 5.1 to 53.3 pg.ml-1, SEM 3.6 and decreased the mean plasma [AVP] from 1.4 pg.ml-1, SEM 0.1 to 0.9 pg.ml-1, SEM 0.04. Simultaneously, diuresis and natriuresis increased markedly. During a 20-min control period in the supine posture without immersion, PV, plasma [ANF] and [AVP] remained unaffected while diuresis and natriuresis did not increase to the same extent. These data suggest that an increase in the central blood volume induced by a weak external hydrostatic pressure during supine immersion triggered the changes in plasma [ANF] and [AVP] and that the increase was probably due to a shift of blood volume from peripheral to central vessels. The changes in plasma [ANF] contributed to the changes in natriuresis.     

Characterization of specific receptors for atrial natriuretic factor in bovine adrenal zona glomerulosa

We have recently shown that synthetic rat atrial natriuretic factor (ANF) directly inhibits mineralocorticoid and glucocorticoid secretion in cultured bovine adrenal cells with a potency of 100 pM. [125I]iodo-ANF was used in the present study to characterize potential receptor sites in bovine zona glomerulosa membranes. ANF binds to a class of high affinity binding sites with a pK of 10.2 and a density of 1.3 pmol/mg protein. Detailed competition curves with ANF document a class of high affinity sites with a pK of 10.2 and also a second class of lower affinity sites with a pK of 8.5. Nonspecific binding amounts to less than 10% of [125I]iodo-ANF binding at concentrations less than 100 pM. High affinity binding of [125I]iodo-ANF is reversible with a half-time of association of 15 minutes at 25 pM and a half-time of dissociation of 140 minutes. Monovalent cations Na, Li and K equipotently enhance [125I]iodo-ANF specific binding. Divalent cations Mg, Ca and Mn also increase [125I]iodo-ANF specific binding, with Mn being the most active cation. No effect of guanine nucleotide could be detected on ANF binding. The binding of [125I]iodo-ANF is very specific and is not inhibited by 1 microM angiotensin II, ACTH, VIP, somatostatin, Leu-enkephalin, dynorphin or by the N-terminal of POMC. The N-terminal fragment ANF-(1-16) is also completely inactive. Reduction of the disulfide bridge of ANF inactivates the peptide. This enabled the development of a highly specific radio-receptor assay for ANF with a minimum detectable dose of 2 femtomoles. The results document the specific receptor involved in the potent inhibitory effect of ANF on adrenal steroidogenesis and indicate that bovine adrenal zonal glomerulosa provide a highly sensitive system for studying the recently discovered atrial natriuretic factor.     

Atrial natriuretic factor stimulates the frog urinary bladder osmotic permeability in presence of a cyclic nucleotide phosphodiesterase inhibitor

The role of atrial natriuretic factor (ANF) in regulation of osmotic water permeability was studied in isolated frog Rana temporaria L. urinary bladder. It was found that ANF (rANF, 1-28) added to the serosal solution at concentrations 5 x 10(-8) M and higher dosedependently stimulated the arginine-vasotocin (AVT)-induced increase of osmotic water permeability. The effect of ANF was revealed only in presence of 3-isobuthyl-1-methylxantine (180 microM) and was accompanied by significant elevation of cGMP level in urinary bladder homogenate and isolated mucosal epithelial cells. C-ANF (des[Gln18, Ser19, Gly20, Leu21, Gly22]-ANF-(4-23)-NH2), a specific agonist of NPR-C receptor, exerted no effect on osmotic water permeability. ANF induced a significant increase of cAMP in urinary bladder homogenates (AVT, 5 x 10(-11) M: 52.3 +/- 10.6; AVT + ANF, 10(-7) M: 114.2 +/- 26.9 pmol/mg protein, n = 5, p < 0.05). The activity of adenylate cyclase in crude plasmatic membrane fraction was not changed. Milrinone, a specific inhibitor of phosphodiesterase 3, at concentrations from 25 to 80 microM, enhanced both the hydroosmotic response to AVT and AVT-stimulated cAMP production. Altogether these data demonstrate that, in the frog urinary bladder, ANF stimulates the AVT-induced increase of osmotic water permeability acting probably through NPR-A receptor-coupled mobilization of cGMP and cGMP-dependent inhibition of phosphodiesterase 3.     

Stimulation of protein phosphorylation by atrial natriuretic factor in plasma membranes of bovine adrenal cortical cells

Atrial natriuretic factor (ANF) rapidly enhanced phosphorylation of plasma membrane proteins of bovine adrenal cortical cells. Pretreatment of the membranes with ANF (1 x 10(-8)M to 1 x 10(-7)M) resulted two- to four-fold in an incorporation of 32p-radioactivity from [gamma -32p]ATP as compared to the controls. The guanosine 3', 5' monophosphate (cGMP) which has been considered a second messenger of ANF also enhanced the phosphorylation of several proteins which were stimulated by ANF. However, the phosphorylation of certain proteins was stimulated differentially only by either ANF or cGMP. These results suggest that ANF-induced protein phosphorylation may play a role in transmembrane signalling pathway involving other second messenger(s) in addition to cGMP during the biological action of ANF.     

Studies of the penetration of the blood brain barrier by atrial natriuretic factor

The atrial natriuretic factors (ANF) have been detected in various areas of the brain. To determine whether circulating blood borne ANF could contribute to the ANF content in the central nervous systems we examined the ability of ANF-99-126 or ANF-102-126 to penetrate the blood brain barrier. Carotid artery injections of [3H] inulin with [125I] ANF in anesthetized rabbits resulted in a comparably minimal brain uptake index (BUI) for each labeled substance as measured in cerebral cortex extracts. Injection of [3H] HOH and [125I] ANF resulted in a mean BUI in cortex of 4.9 +/- .6 (SEM)% for ANF relative to triated water; this low uptake was not significantly saturable. The BUI ratio for ANF/HOH in olfactory bulb was somewhat higher though still low, at 7.0 +/- 9%, possibly reflecting the high density of ANF receptors in this structure. Infusion of [125I] ANF into the carotid artery of anesthetized rabbits resulted in little radioactivity being detected in the cerebrospinal fluid. Infusion of unlabeled ANF, which raised plasma levels as high as 26.3 ng/ml, resulted in little change in CSF levels. Our results demonstrate that the uptake of ANF into the brain is minimal and supports the idea that local synthesis of ANF predominantly accounts for the brain pool of this peptide.     

Comparative hemodynamic actions of ANF-related peptides in conscious sheep

The rapid hemodynamic effects of several N- and C-terminal deleted fragments of ANF, a potent ANF analogue and the recently characterised brain natriuretic peptide (BNP) were investigated in conscious sheep, and compared to the rapid hemodynamic actions of ANF 1-28. The hypotensive potency of all peptides studied was as follows: ANF 1-28 = PLO58 greater than 5-27 = ANF 5-28 = BNP greater than ANF 7-28 greater than ANF 13-28 = ANF 5-25. All peptides reduced blood pressure via a decrease in total peripheral resistance, excluding ANF 5-25 and 13-28 which were without effect on any parameter measured. These changes were associated with reflex increases in both heart rate and cardiac output, and a slight reduction in stroke volume. The duration of hypotensive/vasodilator action of ANF 1-28, 5-27, 5-28, 7-28 and BNP was approximately 3-4 minutes, whereas that of PLO58 was 7-8 minutes. In conclusion, amino acid deletions from the C- and N-terminal of the ANF molecule reduced the hypotensive/vasodilator potency of the peptide in conscious sheep. BNP produced similar rapid hemodynamic changes to ANF 1-28, suggesting that the two peptides may co-regulate blood pressure and possibly body fluids to promote fluid and cardiovascular homeostasis.     

HS1 interacts with Lyn and is critical for erythropoietin-induced differentiation of erythroid cells

Erythroid cells terminally differentiate in response to erythropoietin binding its cognate receptor. Previously we have shown that the tyrosine kinase Lyn associates with the erythropoietin receptor and is essential for hemoglobin synthesis in three erythroleukemic cell lines. To understand Lyn signaling events in erythroid cells, the yeast two-hybrid system was used to analyze interactions with other proteins. Here we show that the hemopoietic-specific protein HS1 interacted directly with the SH3 domain of Lyn, via its proline-rich region. A truncated HS1, bearing the Lyn-binding domain, was introduced into J2E erythroleukemic cells to determine the impact upon responsiveness to erythropoietin. Truncated HS1 had a striking effect on the phenotype of the J2E line-the cells were smaller, more basophilic than the parental proerythoblastoid cells and had fewer surface erythropoietin receptors. Moreover, basal and erythropoietin-induced proliferation and differentiation were markedly suppressed. The inability of cells containing the truncated HS1 to differentiate may be a consequence of markedly reduced levels of Lyn and GATA-1. In addition, erythropoietin stimulation of these cells resulted in rapid, endosome-mediated degradation of endogenous HS1. The truncated HS1 also suppressed the development of erythroid colonies from fetal liver cells. These data show that disrupting HS1 has profoundly influenced the ability of erythroid cells to terminally differentiate.     

Truncated atrial natriuretic factor analogs retain full agonist activity.

The synthesis, receptor binding, and agonist activity of a series of truncated atrial natriuretic analogs (ANF) are described. These analogs incorporate two portions of the native 28 amino peptide, the eight amino acids C-terminal to Cys7, and two amino acids from the C-terminus (phenylalanine and arginine), into disulfide-bonded cyclic peptides. The inclusion of the C-terminal amino acids converted the ANF analogs from receptor ligands to full agonists, as measured by several methods, including the stimulation of cGMP biosynthesis in endothelial cells, inhibition of aldosterone biosynthesis in rat adrenal cells, and natriuretic-hypotensive activity in vivo. The most potent analogs have cyclohexylalanine (Cha) at position 8. The lead compound (Arg6,Cha8 ANF 6-15 Phe-Arg-Cys-NH2) is a tridecapeptide that integrates the C-terminal amino acids inside the disulfide ring. This peptide, designated as A-68828, has a binding affinity of IC50 = 120 nM, approximately 1/400 of ANF 1-28. However, this analog, in vivo, is only slightly less natriuretic (1/20-1/50) than ANF 1-28. Unlike the native peptide, A-68828 is only mildly hypotensive and at the highest concentration tested reduced blood pressure less than 15 mmHg (1 mmHg = 133.322 Pa). A-68828 inhibited ACTH-induced aldosterone release to a greater extent than ANF 1-28: 100 vs. 50%. The selective natriuretic activity of A-66828, relative to ANF, suggests clinical utility for the treatment of acute renal failure.     

Localization of binding sites for ANF in the rat mammary gland

The localization of ANF binding sites in the mammary gland has been performed in virgin and lactating rats using an in vitro autoradiographic procedure performed on slide-mounted frozen sections. The radioligand used was rat [125I]ANF (99-126). It was demonstrated that specific ANF binding sites were present in the secretory cells and also to a lesser extent in acinar lumen. These results suggest that ANF could play some role in the production of fluid during lactation.     

Hormonal modulation of atrial natriuretic factor receptors and effects on adrenal glomerulosa cells of female rats

This study was done to determine if a decrease in the aldosterone-suppressant effect of atrial natriuretic factor (ANF) by progesterone and an increase by estrogen was caused by modulation of adrenal zona glomerulosa ANF receptors. Freshly dispersed glomerulosa cells from virgin, 13–15 day pregnant, ovariectomized (OVX) estradiol-17β-treated and OVX progesterone-treated rats were used. Competitive displacement of specifically bound [¹²⁵I]ANF_1–8 with unlabelled ANF_1–28 yielded concentrations of guanylate cyclase-linked ANF-R1 plus ANF-R2 (clearance) receptors and the displacement with unlabelled ANF_4–23 yielded ANF-R2 receptors; the difference between the two was treated as the concentration of ANF-R1 receptors. Pregnancy and progesterone decreased and estrogen increased the number of glomerulosa ANF-R1 receptors. ANF produced a significantly greater suppression of potassium-induced aldosterone secretion in cells from OVX estradiol-treated rats than in cells from OVX progesterone-treated animals. These data suggest that the inhibition of the aldosterone-suppressant activity of ANF by progesterone is the result of a downregulation of ANF-R1 receptors.     

Growth factor receptor expression during in vitro differentiation of partially purified populations containing murine stem cells

We have investigated, by semiquantitative RT-PCR, the kinetics of activation of hematopoietic receptors and differentiation markers in partially purified murine hematopoietic stem cells (HSC) induced to differentiate in serum-free culture with combinations of growth factor (GF). The combinations of GF used sustained either multilineage [stem cell factor (SCF) + interleukin 3 (IL-3)], or erythroid [SCF + IL-3 + erythropoietin (Epo)] or myeloid [SCF + IL-3 + granulocyte colony-stimulating factor (G-CSF)] differentiation. The GF receptor genes investigated were the α and β subunits of the IL-3 and granulocyte-macrophage colony-stimulating factor (GM-CSF) receptor, the erythropoietin receptor, the G-CSF receptor, and c-Fms, the receptor for macrophage colony-stimulating factor (M-CSF). The expression of Gata1 and α- and β-globin was investigated at the same time as a marker of erythroid differentiation. HSC were purified according to standard protocols, which include partitioning of lineage-negative bone marrow cells with the mitochondrial dye Rhodamine 123 (Rho) into Rho-dull (≥17% of which reconstitute long-term hematopoiesis in recipient mice) and into Rho-bright (which are as capable as Rho-dull of multilineage differentiation but do not permanently reconstitute the host). The following pattern of expression was observed: the α subunit of the IL-3 receptor clearly was expressed in both Rho-bright and Rho-dull cells at the outset, and its expression did not change over time in culture. The β subunits of the IL-3 and GM-CSF receptor, the α subunit of the GM-CSF receptor, the Epo and G-CSF receptors and Fms barely were expressed in purified Rho-bright and Rho-dull cells, but their expression increased in cells cultured both in erythroid and in myeloid GF combinations. Gata1 was expressed maximally in Rho-bright cells but was below the level of detection in Rho-dull cells. Rho-dull cells expressed Gata1 when cultured both in erythroid and in myeloid GF combinations. In contrast, α- and β-globin, which also were not expressed in the purified cells, were induced only in cells stimulated with Epo. These results indicate that the genes for all the GF receptors investigated (with the exception of the α subunit of the IL-3 receptor) are expressed at low levels, if any, in purified Rho-bright or Rho-dull cells, but are expressed in their progeny cultured either in erythroid or myeloid GF combinations. The expression of the Epo receptor,in particular, is activated both in erythroid (α- and β-globin positive) and in myeloid (α- and β-globin negative) cells. Therefore, activation of the expression of the Epo receptor gene and activation of the erythroid differentiation program are two independent events in normal hematopoiesis. J. Cell. Physiol. 171:343–356, 1997. © 1997 Wiley-Liss, Inc.