Antiserum to activity-dependent neurotrophic factor produces neuronal cell death in CNS cultures: immunological and biological specificity

Developmental Brain Research 99 Ž1997. 167–175

Research report

Antiserum to activity-dependent neurotrophic factor produces neuronal cell death in CNS cultures: immunological and biological specificity Illana Gozes a , Ariane Davidson a , Yehoshua Gozes b, Richard Mascolo c , Rolf Barth c , Dale Warren c , Janet Hauser c , Douglas E. Brenneman c,) a

c

Department of Clinical Biochemistry, Sackler School of Medicine, Tel AÕiÕ UniÕersity, Tel AÕiÕ, Israel b The Israel Institute for Biological Research, Ness Ziona, Israel Section on DeÕelopmental and Molecular Pharmacology, Laboratory of DeÕelopmental Pharmacology, National Institute for Child Health and Human DeÕelopment, National Institutes of Health, Bethesda, MD 20892, USA Accepted 26 November 1996

Abstract Activity-dependent neurotrophic factor ŽADNF. is a glia-derived protein that is neuroprotective at femtomolar concentrations. ADNF is released from astroglia after treatment with 0.1 nM vasoactive intestinal peptide ŽVIP.. To further assess the biological role of ADNF, antiserum was produced following sequential injections of purified ADNF into mice. Anti-ADNF ascites fluid Ž1:10,000. decreased neuronal survival by 45–55% in comparison to untreated cultures or those treated with control ascites. The neuronal death after anti-ADNF treatment was observed in cultures derived from the spinal cord, hippocampus or cerebral cortex at similar IC 50’s. Using a terminal deoxynucleotidyl transferase in situ assay to estimate apoptosis in cerebral cortical cultures, anti-ADNF was shown to produce a 70% increase in the number of labeled cells in comparison to controls. In spinal cord cultures, anti-ADNF treatment produced a 20% decrease in choline acetyltransferase activity in comparison to controls. Neuronal cell death produced by the antiserum to ADNF was prevented in cultures co-treated with purified ADNF or ADNF-15, an active peptide derived from the parent ADNF. In vitro binding between the anti-ADNF and ADNF-15 was demonstrated with size exclusion chromatography. Comparative studies with other growth factors Žinsulin-like growth factor-1, platelet-derived growth factor, nerve growth factor, epidermal growth factor, ciliary neurotrophic growth factor, and neurotrophin-3. demonstrated that only ADNF prevented neuronal cell death associated with electrical blockade. These investigations indicated that an ADNF-like substance was present in cultures derived from multiple locations in the central nervous system and that ADNF-15 exhibited both neuroprotection and immunogenicity. ADNF appears to be both a regulator of activity-dependent neuronal survival and a neuroprotectant. q 1997 Elsevier Science B.V. All rights reserved. Keywords: Apoptosis; Astroglia; Vasoactive intestinal peptide; Neurotrophic factor neurotrophin; Cerebral cortex

1. Introduction An increasing number of diverse neuronal growth factors are being discovered. Included in this group of regulatory molecules are trophic factors such as nerve growth factor ŽNGF. w29,30x, ciliary neurotrophic factor ŽCNTF. w31x, fibroblast growth factor ŽFGF. w15,40x, insulin-like growth factors 1 and 2 ŽIGF 1 and 2 w27,41x. brain-derived neurotrophic factor ŽBDNF. w28x, neurotrophin-3 and neurotrophin-4r5 ŽNT3 w16x and NT4, w25x. and glial-derived neurotrophic factor w32x. Furthermore, cytokines also have neurotrophic properties w12,34x. This expanding class of

)

Corresponding author. Fax: q1 Ž301. 496-9939.

substances includes the various interleukins w12,13,34,36x and leukemia inhibitory factor w35x. Although many of the classic growth factors were first recognized to play important trophic roles in neuronrtarget cell interactions, it is now clear that glial cells in the central nervous system ŽCNS. express most of these growth factorsrcytokines, and that these glial cells have significant roles during development and nerve repair. A neuroprotective, glia-derived neurotrophic protein Ž14,000 Da and pl 8.3 " 0.25., was recently isolated by sequential chromatographic methods w14x. The protein was named activity dependent neurotrophic factor ŽADNF., as it protected neurons from death associated with electrical blockade. The strategy used in isolating ADNF entailed measuring changes in neuronal survival after treatment of

0165-3806r97r$17.00 Copyright q 1997 Elsevier Science B.V. All rights reserved. PII S 0 1 6 5 - 3 8 0 6 Ž 9 6 . 0 0 2 1 5 - 5

168

I. Gozes et al.r DeÕelopmental Brain Research 99 (1997) 167–175

developing spinal cord cultures with tetrodotoxin, an agent that blocks synaptic activity. The tetrodotoxin provided both a tool to delineate neurons that were dependent on ongoing electrical activity for their survival and a means of producing neuronal cell death by a recognized neurotoxin, thus providing a paradigm to assess neuroprotection. The end result was the isolation of a molecule that both regulates activity-dependent neurodevelopment and provides a wide spectrum of neuroprotection at femtomolar concentrations w14x. The discovery of ADNF originated from investigations on trophic support for activity-dependent neurons, a class of cells that die during electrical blockade w2,4x. Previous studies indicated that activity-dependent neurons also require glutamate for their survival, specifically through receptors characterized by their stimulation with Nmethyl-D-aspartate, NMDA w10,11x. Neuronal cell death produced by blockade of NMDA receptors or tetrodotoxin can be prevented by the neuropeptide vasoactive intestinal peptide, VIP w11x. Electrical blockade after treatment with tetrodotoxin has been demonstrated to inhibit the synthesis and release of trophic materials, including VIP w1,5,6x. VIP has been shown to prevent neuronal cell death associated with the envelope protein from the human immunodeficiency virus Žglycoprotein 120, w8x., and the beta amyloid peptide Žputative cytotoxin in Alzheimer’s disease w24x.. Previous experiments have indicated that the neuroprotective actions of VIP are contingent on the presence of glial cells w7,9x expressing high-affinity VIP binding sites w23x. Therefore, we searched for novel survival-promoting proteins released from glial cells stimulated by VIP, leading to the isolation of ADNF w14x. During the course of studies directed on the structural characteristics of ADNF, an active peptide fragment was discovered: ADNF-14. This peptide had strong homology, but not identity, to an intracellular stress protein: heat shock protein 60 Žhsp60.. ANDF-14, like ADNF, has been shown to exhibit neuroprotection from a wide variety of neurotoxic substances including the envelope protein from the human immunodeficiency virus, N-methyl-D-aspartate Žexcitotoxicity., beta amyloid peptide and tetrodotoxin w14x. Classical studies on the biological effects of nerve growth factors, the foremost being NGF, relied on neutralizing antisera w30x. In addition to effects on neuronal system development, neutralizing anti-NGF antibodies were recently utilized for the identification of NGF as an autocrine factor for memory B lymphocytes w39x. Similarly, neutralizing antibodies to BDNF reduced axon arborization and complexity w17x and neutralizing antisera to FGF, platelet-derived growth factor ŽPDGF. and epidermal growth factor ŽEGF. inhibited mitogenic activity released from mechanically injured vascular smooth muscle w18x. The current report describes the production of neutralizing antibodies to ADNF, permitting the demonstration that CNS cultures contain an endogenous ADNF-like neuronal survival factor. Furthermore, these investigations have

identified an active neuroprotective site for ADNF, ADNF15, as a prominent immunogenic epitope.

2. Materials and methods 2.1. Antibody preparation Antibodies to ADNF were produced following sequential injections of purified ADNF Ž0.5–1 m grinjection. into Balbrc mice. The first injection was into the foot pad in complete Freund’s adjuvant. Subsequent injections were subcutaneous in incomplete Freund’s adjuvant. Ten days after the last Žfourth. injection, Krebs cells were injected and ascites fluid was then collected. 2.2. Cell culture The source of the ADNF was rat cortical astrocytes, a superior source for astroglia because of rapid growth characteristics and established cellular composition w23,33x. For measurements of neuroprotective actions, cerebral cortical cultures derived from new born rats w24,26x were used. In this system, post-mitotic neurons were maintained on a confluent layer of cortical astrocytes. Neuronal cell counts were conducted after fixation with glutaraldehyde. Dishes were coded and counted without knowledge of the treatment group. Neuronal identity was established with sister cultures immunocytochemically stained with antiserum against neuronal specific enolase w38x. For electrical blockade, treatment with 1 m M tetrodotoxin ŽTTX. for five days was utilized. Dissociated spinal cord cells from E12 mice were also used w5,37x, as well as rat hippocampal cultures prepared as previously described w8x. All statistical comparisons were done by analysis of variance with the StudentNewman-Keuls multiple comparison of means test. 2.3. Purification of actiÕity dependent neurotrophic factor (ADNF) The purification of ADNF was recently described w14x. In short, two-week-old astroglial w23,33x cultures Žconfluent 75 cm2 flasks. were washed three times with phosphate-buffered saline ŽPBS. and conditioned medium was collected Ž10 ml PBSrflask. during a 3 h incubation with 0.1 nM VIP. The medium was centrifuged Ž3000 = g for 10 min. and dialyzed Ž3.5 kDa cutoff. against 50 mM sodium phosphate buffer, pH 7.0, 48C. The purification scheme for ADNF began with DEAE-Sephacel chromatography of VIP-stimulated astroglia-conditioned medium in 50 mM sodium pyrophosphate buffer, pH 7.0. The column was washed sequentially with 50 mM sodium pyrophosphate buffer ŽpH 7.0. supplemented with increasing concentrations of NaCl: 0.1 M, 0.26 M, 0.5 M, 1.0 M, 2 M. ADNF was collected in the 2 M fraction. The second purification step consisted of size separation of the active

I. Gozes et al.r DeÕelopmental Brain Research 99 (1997) 167–175

DEAE fraction Ž2 M NaCl eluate. on fast performance liquid chromatography ŽFPLC system, Pharmacia, Superose e 12 column, pre-packed HR 10r30. in 50 mM sodium phosphate ŽpH 7.3. containing 0.15 M NaCl. The third purification step of the low molecular weight neuroprotective activity Žabout 14–16,000 Daltons eluted from the FPLC sizing column. included hydrophobic interaction FPLC ŽAlkyl Superose e HR5r5, Pharmacia.. The column was washed with 0.1 M phosphate buffer ŽpH 7.0. and then equilibrated with 0.1 M phosphate buffer ŽpH 7.0. containing 2.0 M ŽNH 4 . 2 SO4 . The sample was loaded in 0.1 M sodium phosphate buffer, pH 7.0 containing 1.43 M ŽNH 4 . 2 SO4 . Elution was performed with a linear gradient of salt removal Ž2.0–0 M. initiated 10 min. after injection and lasting 50 min. The purification was monitored by tests of biological activity in electrically blocked spinal cord cultures. After hydrophobic interaction chromatography, the amount of protein in the active fraction was determined by total amino acid analysis on a Beckman Model 7300 instrument, following hydrolysis Ž24 hr1108C. in 6 N HCl containing 0.2% phenol. 2.4. Enzyme-linked immunoadsorbance assay for ADNF ADNF Ž5 ngrwell. was attached to 96-well plates ŽNunc. in 0.5 M carbonate buffer, pH 9.6. The plate was blocked with 2% bovine serum albumin ŽBSA. and 0.05% tween 20 in PBS. Increasing dilutions of the antibody were added Žin the above mentioned buffer. and reactions were incubated for 12 h. Rabbit anti-mouse antibodies conjugated to biotin were used for detection of antibody–antigen reaction with avidin, biotinylated horseradish peroxidase and chromogen ŽSigma.. Absorbance was monitored a wavelength of 414 nm. 2.5. Peptide synthesis For peptide synthesis, the solid phase strategy employing optimum side chain protection was utilized w14,24x. Products were purified on Sephadex G-25 and reversed phase HPLC. Peptides showed the desired molar ratios of constitutive amino acids and were ) 99% pure ŽPeptide Technologies, Gaithersburg, MD.. The amino sequence of ADNF-15 is VLGGGSALLRSIPAL, a carboxy leucine extension of the previossly described ADNF-14 w14x.

169

cm Superdex 75 precision column Žlinear fractionation range 3–70 kDa.. Detection was by m Peak Monitor ŽPharmacia. at 214 nm. 2.7. In situ apoptosis Identification of apoptotic neurons was estimated with a TACS e 2 terminal deoxynucleotidyl transferase in situ apoptosis kit ŽTrevigen, Gaithersburg, MD.. Cerebral cortical cells Ž125,000 cellsrwell. were seeded onto 4-chambered glass slides ŽLab-Tek, Naperville, IL. containing a confluent layer of astroglia. Cultures were maintained in 5% horse serumrMEM nutrient growth medium w14x. Treatment of the chambered slides was from day 6 to day 11 after seeding of the neurons. With the following modifications, the assay was conducted as described by the manufacturer. Cultures were fixed with 2.5% glutaraldehyde in 0.15 cacodylate buffer for 1 h. After removal of the fixative, the cells were rinsed three times with 0.15 M cacodylic acid and stored in this solution. Prior to the labeling reaction, the cells were treated with Cytopore for 30 min. Endogenous peroxidase was removed with a 5 min treatment with hydrogen peroxide. The labeling reaction was conducted as described in the kit, utilizing the MnCl 2 option. TACS Blue label was used for detection. Cultures were then stained with Red Counterstain B for 2 min. Stained neurons in 20 fields were counted at 400 = magnification. 2.8. Choline acetyltransferase assay Choline acetyltransferase activity ŽChAT. was assayed as previously described w22x. Spinal cord cells were plated in 24 well plates at a density of 125,000 cellsrwell. Culture conditions were as described previously w3x. Briefly, cells were rinsed three times with phosphate buffered saline and then disrupted in situ using a 30 min incubation Ž378C. with a buffer that consisted of the following: 50 mM NaPO4 ŽpH 7.4., 1 mM disodiumEDTA, 0.25% Triton-X 100 and 0.2 M NaCl. ChAT was measured in a 1 h assay with a reaction mixture w22x that contained 300,000 CPM of 3 H-acetyl CoA Ž200 m M.. Reaction mixture without choline substrate was used as a blank for the assay.

2.6. Peptider antibody interaction

3. Results

Evidence for an association between anti-ADNF and ADNF-15 was obtained by chromatographic analysis of antibodyrpeptide complex after incubation in vitro. For these studies, the SMART chromatographic system ŽPharmacia Biotech. was used because of its high recovery of small quantities of injected protein and its high sensitivity detection system. Separation was accomplished by gel filtration liquid chromatography using a 3.2 mm i.d.= 30

3.1. Anti-ADNF serum produces neuronal cell death: Protection by ADNF Increasing concentrations of ascites fluid directed against ADNF were incubated for five days with rat cerebral cortical neurons. At a dilution of 1:10,000, antiADNF ascites fluid decreased neuronal cell counts to 54% control Ž P - 0.001., while control ascites fluid had no

170

I. Gozes et al.r DeÕelopmental Brain Research 99 (1997) 167–175

effect ŽFig. 1A.. Similar decreases in neuronal survival were observed with anti-ADNF treatment of either rat hippocampal cultures Ž47% of control. or dissociated spinal cord cultures Ž57% of control. derived from fetal mice Ždata not shown.. The IC 50 for antibody dilution was 1:60,000 for all three CNS preparations. Administration of increasing concentrations of purified ADNF exhibited a dose-dependent prevention of the neuronal death associated with the antiserum treatment of cerebral cortical cultures Žantibody dilution of 1:10,000; Fig. 1B, closed circles.. The most effective concentration of ADNF was 10y1 3 M Ž P - 0.001., with an EC 50 of 10y1 4 M. Treatment with ADNF alone significantly Ž P - 0.05. increased neuronal cell counts of cerebral cortical neurons at 1 fM ŽFig. 1B.. To test if the neuronal cell death produced by anti-ADNF involved apoptosis, an in situ assay to detect double strand breaks in genomic DNA was conducted on cerebral cortical cultures. As shown in Fig. 2, a five day treatment produced a 70% increase in the number of labeled cells in comparison to controls. This increase in apparent apoptosis produced by the antiserum was prevented by co-treatment with purified ADNF. Addition of ADNF alone to the cultures significantly decreased the number of labeled cells in comparison to controls. Treatment with control ascites

Fig. 1. An endogenous ADNF-like immunoreactivity provided neuroprotection: A: ADNF antiserum produced neuronal cell death in dissociated cerebral cortical cultures: Increasing concentrations of antiserum Žascites fluid. were added to one week-old dissociated cerebral cortical cultures. Cells were treated with the antiserum for five days Žopen circles, control ascites; closed circles anti-ADNF ascites fluid.. Statistical comparisons ŽANOVA, SNK. revealed significant decreases from controls in cultures treated with antibody dilutions from 10y5 to 0.3=10y2 Ž P - 0.001.. No significant changes in neuronal cell counts from cultures treated with control ascites were observed in comparison to those of control cultures. Each value is the mean of 3–4 determinations"the standard error. B: ADNF prevented neuronal cell death associated with antiserum to ADNF in dissociated rat cerebral cortical cultures: Cerebral cortical cells Žas in A. were incubated with increasing concentrations of ADNF in the absence Žopen circles. or in the presence Žclosed circles. of anti-ADNF serum Ž1:10,000., for five days. In cultures treated with anti-ADNF, significant increases in neuronal cell counts were observed in cultures co-treated with ADNF from 10y1 5 to 10y12 M Ž P - 0.01. as compared to cultures treated with anti-ADNF alone. In cultures treated with ADNF alone, 10y1 5 M ADNF produced a significant increase in cell counts in comparison to controls Ž P - 0.05.. Each value is the mean of 3–4 determinations"standard error.

Fig. 2. Anti-ADNF increased the number of labeled cells in a terminal deoxynucleotidyl transferase apoptosis assay. Cerebral cortical cultures were treated with anti-ADNF ŽAb., purified ADNF or control ascites for 5 days. The anti-ADNF and the control ascites was tested at 1:5000. The concentration of the purified ADNF was 10y1 3 M, as determined from a bioassay Žw14x.. A TACS in situ apoptosis detection kit was used to assess the number of apoptotic neurons. The number of labeled cells was counted in 20 fields at 400= magnification. Control counts were 18.2" 0.9 cellsr20 fields. Significant increases Ž P - 0.001. in the number of labeled cells were observed in cultures treated with anti-ADNF in comparison to controls. Significant decreases Ž P - 0.001. in the number of labeled cells were observed in cultures treated with ADNF alone or ADNF plus anti-ADNF in comparison to control cultures. Each value is the mean of 4–6 determinations. The error bar is the S.E.M.

I. Gozes et al.r DeÕelopmental Brain Research 99 (1997) 167–175

171

treated for five days with the antiserum exhibited a decrease in ChAT activity from 292 " 10 in controls to 240 " 10 pmolrhrwell in antiserum-treated cultures. 3.2. ADNF-15: An actiÕe site and immunogenic epitope Administration of femtomolar concentration of ADNF15 ŽVLGGGSALLRSIPAL., a one amino acid extension

Fig. 3. ADNF-15 prevented anti-ADNF-associated cell death in cerebral cortical cultures. Cultures were treated as in Fig. 1B, substituting ADNF15 for ADNF. In comparison to cultures treated with anti-ADNF alone, co-treatment with ADNF-15 between 10y1 5 and 10y1 3 M significantly increased Ž P - 0.01. neuronal cell counts Žclosed circles.. A significant increase Ž P - 0.001. in cell counts was observed in cultures treated with 10y1 5 M ADNF-15 alone as compared to controls Žopen circles.. Each value is the mean of 3–4 determinations"S.E.M.

produced no significant change in the number of labeled cells from control cultures. Previous studies indicated that a subpopulation of cholinergic neurons in spinal cord cultures were among the cells that were dependent on electrical activity for their survival w2x. Therefore, the effect of anti-ADNF Ž1:5000. on choline acetyltransferase ŽChAT. activity was tested on one-month-old dissociated spinal cord cultures. Cultures

Fig. 4. Anti-ADNF ascites interacts with ADNF in an ELISA assay. Five ng of purified ADNF was attached to 96-well plates. Dilutions of the antiserum were incubated with the coated wells for 12 h. A representative experiment of two trials is shown. Each value is the mean of two determinations.

Fig. 5. Anti-ADNF ascites fluid interacts with a core peptide of 15-amino acid in ADNF, identification by size separation: A: Separation of VLGGGSALLRSIPAL ŽADNF-15. was by gel filtration liquid chromatography: Separation was achieved using a 3.2 mm i.d.=30 cm Superdex 75 precision column Žlinear fractionation range 3–70 kDa., on SMART System ŽPharmacia Biotech.. ADNF-15 Ž20 m g. was applied to the column in 50 m l phosphate buffered saline ŽPBS.. The column flow rate was 50 m lrmin. B: Separation of anti-ADNF ascites fluid using the SMART system: Separation conditions were as described in ŽA.. The column was loaded with 50 m l of 1:1000 dilution of the antibody, in PBS. q Separation of anti-ADNFqANDF-15 on the SMART system: One microgram ADNF-15 was incubated by itself Žsolid line. or with 50 m l of 1:1000 dilution of the ascites fluid, for 2 hours, at 378C Ždotted line.. At least 60% of the peak eluting at 1.67 ml, corresponding to ADNF-15 was not eluted in this position in the presence of the antibody.

172

I. Gozes et al.r DeÕelopmental Brain Research 99 (1997) 167–175

of the previously described ADNF-14 w14x., prevented neuronal cell death associated with anti-ADNF ascites fluid Ž1:10,000. in rat cerebral cortical cell cultures ŽFig. 3, closed circles.. The potency ADNF-15 ŽEC 50 : 10y1 5 M. was about 10 times greater than intact ADNF Žsee Fig. 1B. in preventing neuronal cell death associated with antiADNF. Treatment with ADNF-15 alone significantly Ž P 0.05. increased the survival of cerebral neurons over a narrow range of concentrations ŽFig. 3, open circles..

1.06 ml ŽFig. 5B.. In the mixture of antibodies and peptide following a 2 h incubation, two peaks were obtained in corresponding positions ŽFig. 5C.. However, a marked decrease Ž3–4-fold. in the amount of the ADNF-15 was observed with the anti-serum as compared to ADNF-15 incubated without the serum. After 2 h incubation of ADNF-15 alone, a small proportion of lower molecular weight peaks were observed. These lower molecular weight peaks were not increased in the presence of the antiserum.

3.3. ADNFr antibody interactions

3.4. Comparison of growth factors: Protection from tetrodotoxin-induced neurotoxicity

To investigate the association between ADNF and antiADNF, an ELISA assay was performed. As shown in Fig. 4, antibody binding was detected at 1:200 dilution when 50 m l of the diluted antibody was exposed to 5 ng of purified ADNF. At 1:1000, no detectable association was observed. These data indicated that approximately 1 m g of ADNF is neutralized by 50 m l of undiluted antiserum in this assay system. Attempts to use the anti-ADNF for Western blot analysis of conditioned medium or of purified ADNF were not successful. To further investigate the association between ADNF-15 and the antisera, size exclusion chromatography was employed. A peak of ADNF-15 was apparent at an elution volume of 1.67 ml ŽFig. 5A., while the antibodies eluted at

ADNF Ž10y1 5 M. exhibited neuroprotection against TTX-associated neuronal cell death in cerebral cortical cultures ŽFig. 6.. In the same experiment, several known growth factors ŽNT-3, CNTF, Fig. 6. did not increase neuronal survival. In addition, IGF-1, PDGF, FGF and NGF were tested in TTX-treated spinal cord cultures at concentrations Ž10y8 –10y14 M. known to be biologically active in other systems. In these experiments, no detectable survival-promoting activity was detected, whereas purified ADNF had an EC 50 of 5 fM Ždata not shown.. These results indicate specificity for ADNF’s neuroprotective effects in both cerebral cortical and spinal cord cultures.

4. Discussion

Fig. 6. Comparison of growth factors for protection against neuronal cell death associated with tetrodotoxin. Increasing doses of ADNF Žclosed circles., NT-3 Žclosed inverted triangles. and CNTF Žopen circles. were added to cerebral cortical cultures together with 1 m M tetrodotoxin and incubated for 5 days. Neuronal cell counts were performed as in Fig. 1. Growth factors were purchased from Collaborative Research Laboratories. Significant increases Ž P - 0.01. in cell counts were observed in cultures treated with 10y1 6 to 10y1 5 M ADNF in comparison to cultures treated with tetrodotoxin alone. Neuronal cell counts after treatment with NT-3 or CNTF did not produce significant changes from that of controls. Each value is the mean of 3 determinations"the standard error.

The present study has shown that interference with ADNF by a neutralizing antiserum produced neuronal cell death in cultures derived from three regions of the CNS: spinal cord, hippocampus and cerebral cortex. The cell death produced by the antiserum was characterized by the following observations: Ž1. In all regions, only a portion of the total neurons were affected by the killing actions of the antiserum; Ž2. the relative number of cells affected by the antiserum in these various cultures was similar, 40–60%; Ž3. the cell death produced by the antiserum could be prevented by purified ADNF; Ž4. the neuronal death produced by the antibody was apparently apoptotic and involved some cholinergic neurons; and Ž5. an important epitope of antibody recognition was localized to ADNF-15. The phenotype of the cellular targets for ADNF and anti-ADNF has yet to be characterized. In the present study, the 20% decrease in choline acetyltransferase after anti-ADNF treatment suggested that a small population of cholinergic neurons in the spinal cord were among the cells that were dependent on ADNF. The apoptosis assays also suggested that neuronal cell death associated with blocking ADNF was due to apoptosis. Previous studies of the effects of electrical blockade may provide clues to other neuronal phenotypes that may be influenced by ADNF w2,10,11,14x. Inhibition of electrical activity has been shown to produce cholinergic deficits in spinal cord cultures without apparent effects on GABAergic neurons

I. Gozes et al.r DeÕelopmental Brain Research 99 (1997) 167–175

Fig. 7. The structure of the active site of ADNF-15 in comparison to other known hsp60-like molecules w20x: VLGGGSALLRSIPAL, ADNF15; VLGGGCALLRCIPAL, Human hsp60, hsp60 y 15; VLGGGCALLRCIPAL, Mouse hsp60, hsp60 y 15; VAGGGVALIRVASKL, E. coli (gram-negatiÕe bacteria); VAGGGVALLRARAAL, M. gonorrhoea (gram-negatiÕe bacteria); VPGGGVALLRSSVKI, Rh. meliloti (gramnegatiÕe bacteria); LPGGGTALIRCIPTL, C. trachomatis (of the chlamydia and spirochetes); VSGGGSALVHAVKVL, S. albus (grampositiÕe bacteria); differences are outlined by bold letters. The three consecutive glycines of ADNF-15 were deduced amino acid sequence taken from HSP60 since these positions were not determinable in the sequencing analysis of ADNF-derived peptides w14x.

w2x. In addition, the neuronal cell death produced by electrical blockade with TTX was very similar to that produced by NMDA antagonists w10,11x. These data imply that a subpopulation of neurons that was dependent on NMDA currents for their survival during development were among the cells that require ADNF. Current studies are focused on examining this possibility directly. Incubation of ADNF-15 with the antiserum resulted in a diminution of the chromatographic peak of ADNF-15. Our explanation of this effect is the association of the peptide with the antiserum. The observations do not support the hypothesis that the decrease in ADNF-15 peak size was due to proteolytic digestion, as there were no apparent increases in lower molecular peaks. However, a broadening of the ADNF-15 peak occurred in both the absence or presence of antiserum after a 2 h incubation period. The peak broadening did not increase in the presence of the antiserum. The size exclusion data, the ELISA results, and the biological activity in cell culture strongly indicate an association between the antiserum and the ADNF-15 epitope. ADNF-15 exhibited a remarkable structural homology to a portion of heat shock protein 60 Žhsp60.; this amino acid sequence in hsp60 is one of the most conserved stretch of amino acids from bacteria to man ŽFig. 7, w20x.. However, in contrast to ADNF, recombinant hsp60 did not protect neurons from death associated with electrical blockade w14x. ADNF-15 homolog in HSP 60 has been demonstrated before as the immunogenic site, as well as the site that can induce resistance and protection against the later developments of spontaneous diabetes w19x. Our studies now identify ADNF-15 as an important immunogenic site in ADNF. ADNF and ADNF peptides exhibit unusual pharmaco-

173

logical characteristics in regard to potency, efficacy and specificity. In Figs. 3 and 6, it is apparent that the maximum potency of both ADNF and ADNF-15 are evident at 1–10 femtomolar. This potency is 100–1000 times greater than that observed for other recognized growth factors. The striking potency indicates that very few molecules of ADNF are required to elicit a biological response and implies a mechanism that is non-conventional. The fact that very low dilutions of the anti-ADNF are effective in producing neuronal cell death also supports the conclusion that very few ADNF molecules are present in the culture medium, and they appear to be obligatory for neuronal survival for a subpopulation of neurons. Another unusual aspect of ADNF pharmacology is the attenuation of the biological response with increasing concentration. With both the intact protein and ADNF-15, virtually no biological activity is observed at concentrations greater than 1 pM, an amount that is often optimal for the activity of the neurotrophins. The explanation for the marked diminution of the biological response at such low concentrations is not yet clear. However, other peptides, cytokines and growth factors also exhibit attenuations as a function of increasing concentration w5,12,21x. Several alternatives are offered as potential explanations for these data: Ž1. receptor downregulation with increasing agonist concentration, a classical pharmacological phenomena; Ž2. stimulation of a receptor with an opposite pharmacological response; i.e., receptor cross-talk; Ž3. mixed agonistrantagonist properties of the compound itself, and Ž4. self association of the ligand, thereby decreasing the free concentration. Thus, ADNF exhibits a response similar to other growth-regulating substances, although the concentrations are generally much lower for both activation and attenuation. Finally, the biological specificity of ADNF and ADNF peptides is apparent from the current studies. Our studies also indicate that 40–60% of neurons in the CNS cultures do not require ADNF, thereby suggesting that their survival is regulated by other classes of growth factors. The comparative investigations with many of the recognized growth factors indicated no apparent neuroprotective actions from neurotoxicity associated with tetrodotoxin treatment. This is unusual in that many of the growth factors exhibit overlapping biological activities in neuroprotection and survivalpromoting actions in other systems w15,16,41x. The results suggest that ADNF action is not mediated by receptors of recognized growth factors. This strongly implies a unique and important role for ADNF in neuroprotection.

Acknowledgements This manuscript is dedicated to the memory of Mr. Dale Warren and Maxine Schaffer. Supported in part by the US–Israel Binational Science Foundation. We are grateful to Mr. Dan Bolling for his generous help with the SMART chromatographic system. This manuscript was written when

174

I. Gozes et al.r DeÕelopmental Brain Research 99 (1997) 167–175

Prof. Illana Gozes was a Fogarty-Scholar-in-Residence at the National Institutes of Health, Bethesda, MD, USA.

w20x w21x

References w1x Agoston, D.V., L.E. Eiden, D.E., Brenneman and I. Gozes. Spontaneous electrical activity regulates vasoactive intestinal peptide expression in dissociated spinal cord cell cultures. Mol. Brain Res., 10 Ž1991. 235–240. w2x Brenneman, D.E., Neale, E.A., Habig, W.H., Bowers, L.M. and Nelson, P.G. Developmental and neurochemical specificity of neuronal deficits produced by electrical impulse blockade in dissociated spinal cord cultures. DeÕ. Brain Res., 9 Ž1983. 13–27. w3x Brenneman, D.E. and Warren, D. Induction of cholinergic expression in developing spinal cord cultures. J. Neurochem., 41 Ž1983., 1349–1356. w4x Brenneman, D.E., Fitzgerald, S. and Nelson, P.G. Interaction between trophic action and electrical activity in spinal cord cultures. DeÕ. Brain Res., 15 Ž1984. 211–217. w5x Brenneman, D.E., Eiden, L.E. and Siegel, R.E. Neurotrophic action of VIP on spinal cord cultures. Peptides 6 ŽSuppl. 2. Ž1985. 35–39. w6x Brenneman, D.E. and Eiden, L.E. Vasoactive intestinal peptide and electrical activity influence neuronal survival. Proc. Natl. Acad. Sci. USA, 83 Ž1986. 1159–1162. w7x Brenneman, D.E., Neale, E.A., Foster, G.A., d’Autremont, S. and Westbrook, G.L. Nonneuronal cells mediate neurotrophic action of vasoactive intestinal peptide. J. Cell Biol., 104 Ž1987. 1603–1610. w8x Brenneman, D.E., Westbrook, G.L., Fitzgerald, S., Ennist, D.L., Elkins, K.L., Ruff, M.R. and Pert, C.B. Neuronal cell killing by the envelope protein of HIV and its prevention by vasoactive intestinal peptide. Nature 335 Ž1988. 639–642. w9x Brenneman, D.E., Nicol, T., Warren, D. and Bowers, L.M. Vasoactive intestinal peptide: a neurotropic releasing agent and an astroglial mitogen. J. Neurosci. Res. 25 Ž1990. 386–394. w10x Brenneman, D.E., Forsyth, I.D., Nicol, T. and Nelson, P.G. Nmethyl-D-aspartate receptors influence neuronal survival in developing spinal cord cultures. DeÕ. Brain Res. 51 Ž1990a., 63–68. w11x Brenneman, D.E., Yu, C. and Nelson, P.G. Multi-determinate regulation of neuronal survival: neuropeptides, excitatory amino acids and bioelectric activity. Int. J. DeÕ. Neurosci. 8 Ž1990b. 371–378. w12x Brenneman, D.E., Schultzberg, M., Bartfai, T. and Gozes, I. Cytokine regulation of neuronal survival. J. Neurochem. 58 Ž1992. 454–460. w13x Brenneman, D.E., Hill, J.M., Glazner, G.W., Gozes, I. and Phillips, T.W. Interleukin-1 alpha and vasoactive intestinal peptide: enigmatic regulation of neuronal survival. Int. J. DeÕ. Neurosci. 13 Ž1995. 187–200. w14x Brenneman, D.E. and Gozes, I. A femtomolar-acting neuroprotective peptide. J. Clin. InÕest. 97 Ž1996. 2299–2307. w15x Cheng, B. and Mattson, M.P. NGF and bFGF protect rat hippocampal and human cortical neurons against hypoglycemic damage by stabilizing calcium homeostasis. Neuron 7 Ž1991. 1031–1041. w16x Cheng, B. and Mattson, M.P. NT-3 and BDNF protect CNS neurons against metabolicrexcitotoxic insults. Brain Res. 640 Ž1994. 56–67. w17x Cohen-Cory, S. and Fraser, S.E. Effects of brain-derived neurotrophic factor on optic axon branching and remodeling in vivo. Nature 378 Ž1995. 192–196. w18x Crowley, S.T., Ray, C.J., Nawaz, D., Majack, R.A. and Horwitz, L.D. Multiple growth factors are released from mechanically injured vascular smooth muscle cells. Am. J. Physiol. 269 Ž1995. H1641– H1647. w19x Elias, D., Marcus, H., Reshef, T., Ablamunits, V. and Cohen, I.R. Induction of diabetes in standard mice by immunization with p277

w22x w23x

w24x

w25x

w26x

w27x

w28x

w29x w30x

w31x

w32x

w33x

w34x

w35x

w36x

w37x

w38x

peptide of a 60-kDa heat shock protein. Eur. J. Immunol. 25 Ž1995. 2851–2857. Ellis, R.J. Žed.., The Chaperonins. Cell Biology: A Series of Monographs. Academic Press, San Diego, 1996. Festoff, B.E., Nelson, P.G. and Brenneman, D.E. Prevention of activity-dependent neuronal death: vasoactive intestinal polypeptide stimulates astrocytes to secrete the thrombin-inhibiting, neurotrophic serpin, protease nexin I. J. Neurobiol. 30 Ž1996. 255–266. Fonnum, F. A rapid radiochemical method for determination of choline acetyltransferase. J. Neurochem. 24 Ž1975. 407–409. Gozes, I., McCune, S.K., Jacobson, L., Warren, D., Moody, T.W., Fridkin, M. and Brenneman, D.E. An antagonist to vasoactive intestinal peptide: effects on cellular functions in the central nervous system. J. Pharmacol. Exp. Ther. 257 Ž1991. 959–966. Gozes, I., Bardea, A., Reshef, A., Zamostiatno, R., Zhukovsky, S., Rubinraut, S, Fridkin, M. and Brenneman, D.E. Novel Neuroprotective strategy for Alzheimer’s disease: inhalation of a fatty neuropeptide. Proc. Natl. Acad. Sci. USA 93 Ž1996. 427–432. Henderson, C.E., Camu, W., Mettling, C., Go in, A., Pulsion, K., Karihaloo, M., Rullamas, J., Evans, T., McMahon, S.B., Aramanini, M.P., Berkemeier, L., Phillips, H.S. and Rosenthal, A. Neurotrophins promote motor neuron survival and are present in embryonic limb bud. Nature 363 Ž1993. 266–269. Hill, J.M., Mervis, R.F., Avidor, R., Moody, T.W. and Brenneman, D.E. HIV envelope protein-induced neuronal damage and retardation of behavioral development in rat neonates. Brain Res. 603 Ž1993. 222–233. Ishii, D.N., Glazner, G. and Pu, S.F. Role of insulin-like growth factors in peripheral nerve regeneration. Pharmacol. Therap. 62 Ž1994. 125–144. Leibrock, J., Lottspeich, F., Hohn, A., Hofer, M., Hengerer, B., Masiakowski, P., Thoenen, H. and Barde, Y.A. Molecular cloning and expression of brain-derived neurotrophic factor. Nature 341 Ž1989. 149–152. Levi-Montalcini, R. Recent studies on the NGF-target cells interaction. Differentiation 13 Ž1979. 51–53. Levi-Montalcini, R., Caramia, F. and Angeletti, P.U. Alteration in the fine structure of nucleoli in sympathetic neurons following NGF-antiserum treatment. Brain Res. 12 Ž1969. 54–73. Lin, L.F., Mismer, D., Lile, J.D., Armes, L.G., Butler 3rd, E.T., Vannice, J.L. and Collins, F. Purification, cloning and expression of ciliary neurotrophic factor ŽCNTF.. Science 24 Ž1989. 1023–1025. Lin, L.F., Doherty, D.H., Lile, J.D., Bektesh, S. and Collins, F. GDNF: a glial cell line-derived neurotrophic factor for midbrain dopaminergic neurons. Science 260 Ž1993. 1130–1132. McCarthy, K.D. and Partlow, L.M. Preparation of pure neuronal and non-neuronal cultures from embryonic chick sympathetic ganglia: a new method based on both differential cell adhesiveness and the formation of homotypic neuronal aggregates. Brain Res. 114 Ž1976. 391–414. Mehler, M.F., Rozental, R., Dougherty, M., Spray, D.C. and Kessler, J.A. Cytokine regulation of neuronal differentiation of hippocampal progenitor cells. Nature 362 Ž1993. 62–65. Nobes, C.D. and Tolkovsky, A.M. Neutralizing anti-p21ras Fabs suppress rat sympathetic neuron survival induced by NGF, LIF, CNTF and cAMP. Eur. J. Neurosci. 7 Ž1995. 344–350. Patterson, P.H. The emerging neuropoietic cytokine family: first CDFrLIF, CNTF and IL6; next ONC, MGF, GCSF? Curr. Opin. Neurobiol. 2 Ž1992. 94–97. Romijn, H.J., Habets, A.M.M.C., Mud, M.T. and Wolters, P.S. Nerve outgrowth, synaptogenesis and bioelectric activity in fetal rat cerebral cortex tissue cultured in serum-free, chemically defined medium. DeÕ. Brain Res. 2 Ž1982. 583–589. Schmechel, D., Marangos, P.J., Zis, A.P., Brightman, M. and Goodwin, F.K. Brain enolases as specific markers of neuronal and glial cells. Science 199 Ž1978. 313–315.

I. Gozes et al.r DeÕelopmental Brain Research 99 (1997) 167–175 w39x Torcia, M., Bracci-Laudiero, L., Lucibello, M., Nencioni, L., Labardi, D., Rubartelli, A., Cozzolino, F., Aloe, L. And Garaci, E. Nerve growth factor is an autocrine survival factor for memory B lymphocytes. Cell 85 Ž1996. 345–356. w40x Wallicke, P.A. and Baird, A. Internalization and processing of basic fibroblast growth factor by neurons and astrocytes. J. Neurosci. 11 Ž1991. 2249–2258.

175

w41x Zhang, Y., Tatsuno, T., Carney, J.M. and Mattson, M.P. Basic FGF, NGF and IGFs protect hippocampal and cortical neurons against iron-induced degeneration. J. Cereb. Blood Flow Metab. Ž1993. 378–388.