A chromogranin peptide is co-stored with insulin in the human pancreatic islet B-cell granules

Summary The immunoreaction of a rabbit chromogranin A and B antiserum was studied in normal human pancreatic islets. By examination of consecutive light microscopical sections, it was revealed that, at high antiserum concentrations (1:2000 or less), the whole islet area was heavily labelled, although the peripheral glucagon (A)-cells were the most intense in their immunoreaction. At low antiserum concentrations (1:4000 or more) the A-cells still showed the same intense labelling reaction, but the central B-cells were weakly labelled. Electron microscopically, reactivity towards the chromogranin A and B antiserum and the monoclonal insulin antibodies was present in the same central electron-dense core of the B-cell secretory granules, as demonstrated after application of the immunogold technique at different antibody dilutions. In the A-cells, the chromogranin immunoreactivity was concentrated at the peripheral mantle of the secretory granules. The D-cell granules showed a weak immunolabelling. Examination of human islets with the monoclonal chromogranin A antibody LK2H10 revealed immunogold labelling only in the peripheal mantle of the A-cell granules, while the B-cell granules were unlabelled.The present results show that a chromogranin peptide is co-stored with insulin the in normal human B-cell secretory granules. Although the exact composition of this B-cell chromogranin is unknown, it is not identical to that of the chromogranin A present in the A-cell granules.     

Expression of a male accessory gland peptide of Leptinotarsa decemlineata in insect cells infected with a recombinant baculovirus

The male accessory glands (MAGs) of Leptinotarsa decemlineata produce an 8kDa peptide, designated Led-MAGP, that is recognized by monoclonal antibody MAC-18. The site of synthesis, amino acid sequence and the gene encoding this peptide have been documented ([Smid and Schooneveld, 1992][Smid et al., 1997]). The primary structure is homologous to the N-terminal hexa-repeat section of the chicken prion protein ([Harris et al., 1991]). The biological function of the Led-MAGP has yet to be determined. For further research, large amounts of Led-MAGP is required, both for the production of a more specific antiserum, as well as for application in bio-assays. This paper describes the expression of Led-MAGP in insect cells infected with recombinant baculovirus, and the production of a polyclonal antibody against this recombinant peptide. The peptide was expressed under the control of the polyhedrin promotor. The resulting product was HPLC-purified, and analysis on Western blots immuno-labelled with MAC-18 confirmed that the correct peptide was produced. Purified recombinant peptide was also analyzed by Edman degradation and mass spectrometry; this indicated that it was N-terminally blocked and that the methionine residue at position 7 was oxidized. Large scale production resulted in the formation of aggregations of Led-MAGP, nevertheless a substantial proportion remained in a soluble state and could be harvested. A polyclonal antiserum encoded #87 was produced against recombinant Led-MAGP and its specificity was tested on Western blots of authentic peptide and on LM and EM sections of MAGs. All labelling results were equal to those obtained after MAC-18 labelling. However, antiserum #87 proved to be superior compared to MAC-18, since it recognizes the MAG peptide in normally fed, sexually active males, whereas MAC-18 labelling can only be accomplished after 7 days of starvation of the males. Therefore, the new antiserum #87 enables us to study the transfer dynamics of the Led-MAGP on histological sections.     

Glucagon-like immunoreactivity in hypothalamic neurons of the rat

Summary Antisera specific for three different regions of pancreatic proglucagon were used to examine the distribution of such immunoreactivity in rat hypothalamus. Neurons in the supraoptic and paraventricular nuclei were immunoreactive with an antiserum against glucagon, but not with antisera directed towards the aminoterminal region of proglucagon (glicentin) or the glucagon-like peptide I sequence in the carboxyl-terminal region of proglucagon. These findings confirm a previous report of glucagon-like immunoreactivity in the supraoptic and paraventricular nuclei, but indicate that, while this material is immunochemically related to glucagon, it is not derived from a proglucagon-like precursor.     

Visualisation of specific binding sites for atrial natriuretic peptide on non-neuronal cells of cultured rat sympathetic ganglia

Summary The distribution of atrial natriuretic peptide binding sites on cells in dissociated culture preparations of neonatal rat superior cervical ganglia and in explant cultures of rat thoracic sympathetic chain ganglia has been studied. The autoradiographic visualisation of atrial natriuretic peptide binding sites has been combined with the use of specific immunocytochemical markers for glial cells (antiserum to S-100 protein), fibroblasts (antiserum to fibronectin) and neurones (antiserum to protein gene product 9.5) in order to achieve unambiguous identification of the cell types in culture. Specific binding sites for rat¹²⁵I-atrial natriuretic peptide_(1–28) were observed over subpopulations of fibronectin-like-immunoreactive fibroblasts and S-100-like-immunoreactive glia in the dissociated superior cervical ganglion cultures. However, only a subpopulation of fibronectin-like-immunoreactive fibroblasts possessed atrial natriuretic peptide binding sites in the explant culture preparations. No atrial natriuretic peptide-like-immunoreactive cells were present in either culture. The distribution of autoradiographic grains over individual cell surfaces in culture was uniform, but there were distinct differences in the density of labelling of single cells of the same type. This apparent variation in the number of binding sites on glial cells and fibroblasts in culture did not seem to be related to the morphology of the cells or the surrounding cell types. No sympathetic neurones were labelled with autoradiographic grains in either the dissociated or explant culture preparations. However, the presence of atrial natriuretic peptide binding sites on non-neuronal cells of sympathetic ganglia in culture may be linked to the relationship between atrial natriuretic peptide and the sympathetic nervous system.     

Neurotensin-like immunoreactivity in the nervous system of hydra

Summary Neurotensin-like immunoreactivity is found in nerve fibers present in all body regions of hydra. The nerve fibers are especially numerous in the ectoderm at the bases of the tentacles and in the ectoderm at a site just above the foot. Radioimmunoassays of acetic-acid extracts of hydra, using various region-specific antisera towards mammalian neurotensin, show the presence of multiple neurotensin-related peptides. The amounts of these peptides vary between 1 and 350 pmol per gram wet weight. Gel filtration on Sephadex G-25 reveals a fraction of neurotensin-like peptides that crossreacts equally well with an antiserum directed against sequence 1–8 and an antiserum directed against sequence 6–13 of neurotensin. This fraction elutes also at the position of neurotensin and might closely resemble the mammalian peptide. A fraction eluting with the void volume crossreacts preferentially with antisera directed against sequences 1–8 and 10–13 of neurotensin. Several components of apparent lower molecular weight than neurotensin crossreact preferentially with an antiserum against sequence 10–13. These last peptides represent the major portion of the neurotensin-like peptides in hydra.     

Neurochemical identification of enteric neurons expressing P2X(3) receptors in the guinea-pig ileum

It was hypothesised that P2X(3) receptors, predominantly labelling spinal and cranial sensory ganglionic neurons, are also expressed in intrinsic sensory enteric neurons, although direct evidence is lacking. The aim of this study was to localise P2X(3) receptors in the enteric nervous system of the guinea-pig ileum, and to neurochemically identify the P2X(3)-expressing neurons. In the submucous plexus, cholinergic neurons expressing calretinin (CRT), were immunostained for P2X(3). These neurons made up about 12% of the submucous neurons. In the myenteric plexus, approximately 36% of the neurons expressed P2X(3). Half of the latter neurons were immunoreactive for CRT, whereas about 20% were immunoreactive for nitric oxide synthase (NOS). Based on earlier neurochemical analysis of enteric neurons in the guinea-pig, the myenteric neurons exhibiting P2X(3)/CRT immunoreactivity were identified as longitudinal muscle motor neurons, and those expressing P2X(3)/NOS immunoreactivity as short inhibitory circular muscle motor neurons. In both plexuses, no colocalisation was observed between P2X(3) and calbindin, a marker for intrinsic sensory neurons. Multiple staining with antisera raised against somatostatin, neuropeptide Y, substance P or neurofilament protein did not reveal any costaining. It can be concluded that in the guinea-pig ileum, intrinsic sensory neurons do not express P2X(3) receptors. However, this does not negate the possibility that extrinsic sensory nerves expressing P2X(3) are involved in a purinergic mechanosensory transduction pathway as demonstrated in other organs.     

Glial cells revealed by GFAP immunoreactivity in fish gut

Glial fibrillary acidic protein (GFAP) is a commonly used marker to identify enteric glia in the mammalian gut. Little is however known about enteric glia in other vertebrates. The aim of the present study was to examine the distribution of GFAP immunoreactivity in adult and developing fish. In adult shorthorn sculpin (Myoxocephalus scorpius) and zebrafish (Danio rerio), GFAP immunoreactivity was seen in the myenteric plexus in all regions of the gut. Co-staining for the neuronal markers Hu C/D and acetylated tubulin showed that GFAP immunoreactivity was not associated with nerves. GFAP immunoreactivity was predominantly seen in processes with few glial cell bodies being demonstrated in adult fish. GFAP immunoreactivity was also found in the gut in larval zebrafish from 3 days post-fertilisation, i.e. at approximately the same time that differentiated enteric nerve cells first occur. Immunoreactivity was most prominent in areas with no or a low density of Hu-immunoreactive nerve cell bodies, indicating that the developing glia follows a different pattern from that of enteric neurons. The results suggest that GFAP can be used as a marker for enteric glia in fish, as in birds and mammals. The distribution of GFAP immunoreactivity implies that enteric glia are widespread in the fish gastrointestinal tract. Glia and neurons diverge early during development of the gastrointestinal tract.     

Monoclonal antibodies against Nereis virens brain homogenates as probes for isolation and in situ detection of new neurosecretions.

The brain of Nereis contains 26 ganglionic nuclei which produce numerous neurosecretions. Only a few of them have been characterized. The production of monoclonal antibodies was adopted as an approach to discover unknown neurosecretions. Monoclonal antibodies produced against Nereis virens brain homogenates were selected using stepwise ELISA tests first with brain homogenates, then with brain neurosecretions. Eight antibodies specific for Nereis neurosecretions were selected. The results are illustrated with one of these monoclonal antibodies which was directed against a major peak after HPLC purification of brain neurosecretions. This antibody was subsequently used for the in situ detection of recognized epitope(s) in the brain and ventral nerve cord cells.     

Immunocytochemistry of 5-bromo-2'-deoxyuridine labelled oligonucleotide probes. A novel technique for in situ hybridization

A synthetic oligonucleotide probe, complementary to oxytocin m-RNA was labelled enzymatically with 5-bromo-2'-deoxyuridine (5-BrdU) and with [gamma-32P]-ATP. The labelled probes were used for in situ hybridization of histological sections of the mouse hypothalamus. A monoclonal antibody to 5-BrdU and the streptavidine-peroxidase technique were used in order to visualize hybridization with the 5-BrdU labelled probe. In situ hybridization with [32P] labelling was detected autoradiographically. With both methods hybridized neurons were visible in the magnocellular hypothalamic nuclei. While immunostaining and radio-labelling provided similar localization of oxytocin m-RNA, only the immunocytochemical technique showed clear cellular resolution of the reaction product. In situ hybridization with 5-BrdU labelled probes followed by 5-brdU immunocytochemistry seems to be a powerful alternative to common autoradiographic techniques.     

Immunochemical analysis and immunocytochemical localization of crustacean hyperglycemic hormone from the eyestalk of Macrobrachium rosenbergii

Heptapeptide (YANAVQV-NH2 = T-) and octapeptide (YANAVQTV-NH2 = T+), the putative C-terminus of crustacean hyperglycemic hormone (CHH) from the eyestalk of the giant freshwater prawn Macrobrachium rosenbergii, was synthesized by solid phase peptide synthesis and conjugated to bovine serum albumin, then used for immunization in swiss mice. Specificity of the antisera against both peptides was determined by indirect immunoperoxidase ELISA. The best response of antiserum against each peptide was used to determine the presence of the natural CHH in the eyestalk extract after separation by one step of RP-HPLC using dot-ELISA. The peptide immunoreactive substances were found in fraction 30 using anti-T- antiserum and in fraction 38 using anti-T+ antiserum. However, the CHH activity was found only in fractions 37-39. Immunocytochemical localization of peptide immunoreactive substances in the eyestalk of M. rosenbergii using the anti-T- antiserum did not show any specific staining. In contrast, the anti-T+ antiserum revealed specific staining on a group of 24 +/- 5 neurons in medulla terminalis ganglionic x-organ and their processes through the sinus gland. Similar results were also obtained using the eyestalk of another species, the giant tiger prawn Penaeus monodon, in which 34 +/- 4 neuronal cells were recognized. These results strongly indicate that the anti-T+ antibody can bind to the natural CHH while the anti-T- antibody can not; therefore, this isoform of CHH in M. rosenbergii should consist of 72 residues and threonine is predicted to be present at position 71.     

Control of peripheral glial cell proliferation: enteric neurons exert an inhibitory influence on Schwann cell and enteric glial cell DNA synthesis in culture

Neuronal membranes from rat dorsal root ganglia provide a mitogenic signal to cultured Schwann cells and it has been suggested this is an important factor in regulating Schwann cell numbers during development. In this study, the influence of enteric neurons on the DNA synthesis of both Schwann cells and enteric glia has been investigated as well as the effect of axonal membrane fractions (axolemma) on enteric glia. The proliferation rate of rat Schwann cells and enteric glia was assessed in culture using [3H]thymidine uptake and autoradiography in combination with immunolabelling to identify cell types. When purified rat Schwann cells were co-cultured with guinea pig enteric neurons, their DNA synthesis rate was reduced compared with control cultures of pure Schwann cells or Schwann cells not close to neurites or neuronal cell bodies. Nevertheless, in accordance with previous findings that sensory neurons stimulate Schwann cell division, these Schwann cells increased their DNA synthesis rate when in contact with neurites from purified guinea pig or adult rat dorsal root ganglion neurons and on exposure to bovine axolemmal fractions. The enteric neurons also suppressed the DNA synthesis of enteric glia in co-cultures of purified enteric neurons and enteric glia, while bovine axolemma stimulated their DNA synthesis. These results indicate that a mitotic inhibitory signal is associated with enteric neurons and can exert its effect on both Schwann cells and enteric glia, and that enteric glia, like Schwann cells, are stimulated to divide by axolemmal fractions. It thus seems possible that during development glial cell numbers in the peripheral nervous system may be controlled by both positive and negative regulators of cell growth.     

MIT-23: A mitochondrial marker for terminal neuronal differentiation defined by a monoclonal antibody

We describe a monoclonal antibody directed against a neuron-specific mitochondrial protein from rat brain. On protein blots the antibody recognizes a single polypeptide of apparent molecular weight 23,000. By solid-phase immunoassay the antigen was detected in all brain regions tested but was not detected in nonneural tissues. Within neurons, the antibody stains cytoplasmic granules that immunoelectron microscopy shows are mitochondria, hence the designation MIT-23. Immunocytochemical staining of the cerebellar cortex showed that MIT-23 occurs in all the neuronal types but is absent from glial and other nonneuronal cells. During neonatal development of the cerebellum, MIT-23 appears in neurons after their final cell division or migration is completed, suggesting that specific proteins associated with mitochondria participate in neuronal maturation.     

Immunolocalization of a candidate pheromone receptor in the antenna of the male moth, Heliothis virescens

Pheromone recognition in insects is thought to involve distinct receptor proteins in the dendritic membrane of antennal sensory neurons. We have generated antibodies directed against a peptide derived from the sequence of the candidate pheromone receptor HR13 from Heliothis virescens. The antibodies specifically labelled the cell bodies of a distinct neuron population housed in male-specific pheromone-sensitive sensilla. Combining antibody staining with in situ hybridization the reactive cells were found to express the HR13 gene. In addition, dendrites projecting into sensilla hairs as well as the axonal processes of immunoreactive cells were labelled. Labelling of axons has allowed visualization of their fasciculation within antennal segments and permits tracking of axons as they merge into the antennal nerve. The HR13 protein was first detected 1 day before eclosion. Thus, the distribution of HR13 protein in the antennal neurons of the male moth strongly suggests a role of the HR13 receptor in recognition of pheromones.     

Regulation of vasoactive intestinal peptide expression in sympathetic neurons in culture and after axotomy: The role of cholinergic differentiation factor/leukemia inhibitory factor

Vasoactive intestinal peptide (VIP) expression increases in sympathetic neurons when they are grown in dissociated cell or explant cultures and when they are axotomized in vivo. In dissociated cell culture, the magnitude of the VIP increase was reduced when nonneuronal cells were removed and medium conditioned by ganglionic nonneuronal cells increased VIP in neuron-enriched cultures. Antiserum Against cholinergic differentiation factor (also leukemia inhibitory factor; CDF/LIF), but not against ciliary neurotrophic factor, immunoprecipitated this activity. Medium conditioned by sympathetic ganglion explants also contained a VIP-stimulatory molecule that was immunoprecipitated by CDF/LIF antiserum, and CDF/LIF antiserum partially blocked VIP induction in explants. CDF/LIF mRNA was increased in dissociated cell cultures, in ganglion explants and in vivo after axotomy. Our results suggest that CDF/LIF released from ganglionic nonneuronal cells plays an important role in regulating VIP after axotomy. 1994 John Wiley & Sons, Inc.     

Characterisation of neurons expressing calbindin immunoreactivity in the ileum of the unweaned and mature sheep

We have identified the enteric neuron types expressing immunoreactivity for the calcium-binding protein calbindin D28k (CALB) in cryostat sections and whole-mount preparations of myenteric (MP) and submucosal (SMP) plexuses of sheep ileum. We wished to determine whether CALB-IR in the sheep enteric nervous system was expressed in Dogiel type II cells, as in guinea-pig and rat ileum, and could therefore be used as a marker for intrinsic primary afferent neurons. The neurochemical coding of CALB-containing myenteric and submucosal neurons in ileum of unweaned lamb and mature sheep and its co-localisation with various neural markers was studied immunohistochemically. An antiserum against neuronal nuclear protein (NeuN) failed to detect the entire neuronal population; it was expressed only in 48% of neuron-specific enolase (NSE)-immunoreactive (NSE-IR) neurons. Human neuronal protein appeared to occur in the large majority or all neurons. Almost all CALB-IR neurons were: (1) radially multidendritic; (2) eccentric multidendritic; (3) Dogiel type II. CALB-IR occurred in 20–25% of myenteric and 65–75% of submucosal neurons in lamb and mature sheep, with higher values in mature sheep. Nearly all CALB-IR neurons were common choline acetyltransferase (cChAT)-IR, whereas only about 20% of cChAT-IR somata were CALB-IR. In lamb and mature sheep, 90% of MP CALB-IR neurons were peripheral choline acetyltransferase (pChAT)-IR. In lamb SMP, 80±13% of CALB-IR cells were also pChAT-IR, whereas all those in mature SMP were pChAT-IR. Fewer myenteric CALB-IR neurons exhibited tachykinin (TK) in mature sheep (49%) than in lamb (88%). This was also the case for submucosal ganglia (mature sheep, 63%; lamb, 89%). In lamb MP, 77±7% of CALB-IR cells were NeuN-positive. In mature sheep, 73±10% of CALB-IR somata were NeuN-IR, but NeuN failed to stain SMP neurons. In the MP of suckling and mature sheep, Dogiel type II CALB-IR neurons were calcitonin gene-related peptide (CGRP)-IR. In the SMP at both stages, Dogiel type II CALB-IR somata (about 50% of CALB-IR neurons) were also CGRP-IR. Only small proportions of CALB-IR neurons showed immunoreactivity for calretinin or nitric oxide synthase (NOS), although large populations of CALB and NOS neurons occurred in the ganglia. Thus, CALB is a marker of most Dogiel type II neurons in the sheep but is not confined to Dogiel II neurons. CGRP is a more selective marker of Dogiel type II neurons, being only found in this neuron type.This work was supported by a grant from the Ministero dellIstruzione, dellUniversità e della Ricerca (MIUR)     

Antibodies Raised Against Synthetic Peptides React with Choline Acetyltransferase in Various Immunoassays and in Immunohistochemistry

Abstract: Antisera were raised in rabbits against five synthetic peptides. These peptides have been identified as potentially antigenic epitopes from the sequence of porcine choline acetyltransferase (ChAT) using primary and secondary structure analysis. All five antisera recognized immunoaffinity-purified antigen from porcine brain in an ELISA and on western blots. Four antisera recognized ChAT on dot blots, and another four antisera reacted with native and degraded enzyme in a sandwich ELISA using monoclonal antibodies as the capture antibody. One peptide antiserum was of similar avidity in this sandwich ELISA as a polyclonal antibody raised against immunoaffinity-purified ChAT. The same antiserum reacted with the enzyme from human placenta in an ELISA and on western and dot blots and recognized ChAT in rat, primate, and human neurons. Thus, a single peptide (amino acids 168- 189) provides the means for easy, reliable, and reproducible generation of antibodies against ChAT suitable for replacing conventional polyclonal and monoclonal antibodies.     

NGIWYamide-induced contraction of tube feet and distribution of NGIWYamide-like immunoreactivity in nerves of the starfish Asterina pectinifera

NGIWYamide, a neuropeptide recently isolated from sea cucumbers, was tested on tube feet of the starfish Asterina pectinifera. NGIWYamide (10(-6)-10(-4) M) caused contraction of isolated tube feet. NGIWYamide-like immunoreactivity (NGIWYa-LI) was investigated with an antiserum against NGIWYamide. NGIWYa-LI was found in the radial nerve cord (RNC), the marginal nerve, and the tube feet. Both ectoneural and hyponeural parts of the RNC showed NGIWYa-LI. Immunoreactive cell bodies were found in both parts of RNC. Extensive labeling in the basal region of the ectoneural part suggests that a substantial proportion of axons in this part contains NGIWYamide or a similar substance. In tube feet, NGIWYa-LI was found in the sub-epithelial nerve plexus and in the basal nerve ring. Double labeling along with 1E11, a neuron-specific monoclonal antibody developed from A. pectinifera, indicated that the structures with NGIWYa-LI are neurons. These results suggest that NGIWYamide or an NGIWYamide-like peptide exists in starfish and functions as a neurotransmitter or a neuromodulator.     

Localization of connexins in neurons and glia cells of the Helix aspersa suboesophageal brain ganglia by immunocytochemistry

The aim of the present study was to examine the distribution of cells expressing connexin 26 (Cx26) in the suboesophageal visceral, left and right parietal and left and right pleural ganglia of the snail Helix aspersa by immunocytochemistry. Altogether we have found approximately 452 immunoreactive neurons which represent the 4.7% of the total neurons counted. The stained large neurons (measured diameter 55-140 microm) occurred mostly on the peripheral surface of the ganglia while the small immunostained cells (5-25 microm diameter) were observed in groups near the neuropil. The number of large neurons giving positive Cx26-like immunostaining was small in comparison with that for medium (30-50 microm diameter) and small sized cells. The expression of Cx26 was also observed in the processes of glia cells localized among neurons somata and in the neuropil showing that the antiserum recognized epitopes in both protoplasmic and fibrous glia cells of Helix aspersa. The neuropils of all ganglia showed fibers densely immunostained. While we have observed a good specificity for Cx26-antiserum in neurons, a lack of reaction for Cx43 antiserum was observed in neurons and glia cells. The reaction for enolase antiserum in neurons was light and non-specific and a lack of reaction in glia cells and processes for GFAP antiserum was observed. Although the percentage of positive neurons for Cx26 antiserum was low is suggested that in normal physiological conditions or under stimulation the expression of connexin could be increased. The observed results can be considered of interest in the interpretation of Helix aspersa elemental two neuron networks synchronizing activity, observed under applied extremely low frequency magnetic fields.     

Sources of thalamic afferents projecting to the suprasylvian gyrus of the black sea porpoise

Neurons sending fibers to different loci of the suprasylvian gyrus (SSG) of the porpoise(Phocaena phocaena) cortex were located in the thalamus by retrograde horseradish peroxidase transport and fluorescent tracing techniques. Horseradish peroxidase injection into the anterior section of the suprasylvian gyrus led to retrograde labelling of neurons in the lateral portion of the ventrobasal complex of nuclei and the ventroposteroinferior nucleus. A group of labelled cells was found in the ventral section of the main medial geniculate nucleus. Injecting bisbenzimide into different loci of the medial suprasylvian gyrus also led to retrograde labelling of neurons belonging to the ventral division of the main medial geniculate nucleus. Somewhat lower numbers of labelled cells were found in the inferior nucleus of the pulvinar. Small groups of labelled neurons were also found in the lateral nucleus of the pulvinar, the medioventral nucleus of the medial geniculate body, and the posterior complex of nuclei. A similar distribution of labelled cells was also observed after injecting bisbenzimide into the more caudal portion of the gyrus, although the location of labelled cells in the ventral division of the main medial geniculate nucleus and the lower pulvinar nucleus were shifted in a lateral direction.A. N. Severtsov Institute of Animal Evolutionary Moprhology and Ecology, Academy of Sciences of the USSR, Moscow. National University, Singapore. Translated from Neirofiziologiya, Vol. 21, No. 4, pp. 529–539, July–August, 1989.