The Journal of Neuroscience, March 1989, g(3): 990-995 Localization of the Growth-Associated Phosphoprotein GAP-43 (B-50, Fl) in the Human Cerebral Cortex Larry I. Benowitz,is4 Nora I. Perrone-Bizzozero,’ Seth P. Finklestein,2 and Edward D. Bird3 Departments of ‘Psychiatry, *Neurology, and 3Neuropathology, and 4Program in Neuroscience, Harvard Medical School, Mailman Research Center, McLean Hospital, Belmont, Massachusetts 02178 The growth-associated phosphoprotein GAP-43 is a com- One PKC substrate that appears to play a role in synaptic ponent of the presynaptic membrane that has been linked plasticity is the neuron-specific,growth-associated phosphopro- to the development and functional modulation of neuronal tein GAP-43 (Fl, B-50, pp 46) (Akers and Routtenberg, 1985; connections. A monospecific antibody raised against rat GAP- Lovinger et al., 1985; Snipeset al., 1987). When neuronal con- 43 was used here to study the distribution of the protein in nections are first being established,this protein is synthesized cortical and subcortical areas of the human brain. On West- at high levels and transported to growth cones and immature ern blots, the antibody recognized a synaptosomal plasma synapses(Skene and Willard, 198la, b; Benowitz and Lewis, membrane protein that had an apparent molecular weight 1983; Katz et al., 1985; Meiri et al., 1986; Perrone-Bizzozero and isoelectric point similar to GAP-43 of other species. In et al., 1986; Skene et al., 1986). Although most neurons cease brain tissue reacted with the antibody, the heaviest immu- to expresshigh levels of GAP-43 after establishingmature syn- noreactivity was found in associative areas of the neocortex, aptic relationships (Skene and Willard, 198la, b; Benowitz and particularly within layers 1 and 6, in the molecular layer of Lewis, 1983; Katz et al., 1985; Jacobson et al., 1986; Meiri et the dentate gyrus, the caudate putamen, and the amygdala. al., 1986; Skeneet al., 1986; Baizer and Fishman, 1987), certain In contrast, primary sensory or motor regions of the cortex, subsetsof nerve cells continue to expressthe gene (Neve et al., portions of dorsal thalamus, and cerebellum showed only 1987, 1988) and convey the protein to their presynaptic ter- light staining. Staining was generally confined to the neu- minals (Gispen et al., 1985; Kristjansson et al., 1986; Benowitz ropil, which showed punctate labeling, whereas most neu- et al., 1988; McGuire et al., 1988) throughout life. In at least ronal somata and fiber bundles were unreactive. The pro- one site where the protein persistsin the mature CNS, the rat nounced variations in GAP-43 immunostaining among various hippocampus, changesin its phosphorylation state have been areas of the human brain may reflect different potentials for found to correlate highly with the intensity and duration of LTP functional and/or structural remodeling. (Akers and Routtenberg, 1985; Lovinger et al., 1985; Snipeset al., 1987). These observations have led to the suggestionthat phosphorylation of GAP-43 may be associatedwith alterations Functional plasticity in the mammalian brain appearsto involve in membrane activity or structure similar to those that take both pre- and post-synaptic mechanisms.In the rat hippocam- place during development (Nelson and Routtenberg, 1985; pus, long-lasting changesin synaptic efficacy that result from Routtenberg, 1985; Jacobson et al., 1986; Pfenninger, 1986; high-frequency stimulation of the perforant pathway require Benowitz and Routtenberg, 1987). activation of postsynaptic N-methyl-D-aspartate (NMDA) glu- In view of the possiblerole of GAP-43 in synaptic plasticity, tamate receptors (Collingridge et al., 1983) and are also asso- we have used a monospecificantibody to identify regions of the ciated with persistent increasesin the level of excitatory neu- adult human brain that contain high levels of this protein. Our rotransmitter released from presynaptic endings (Bliss et al., results demonstrate striking regional variations in the distri- 1986). The presynaptic changesmay involve the phosphoryla- butional pattern of GAP-43, with high concentrations in parts tion of one or more protein kinase C (PKC) substratessince of the associativeneocortex and the hippocampusbut low levels manipulations of PKC with phorbol esters can either mimic, in primary sensoryor motor areasof cortex and in much of the enhance, or prevent long-term potentiation (LTP) (Akers et al., brain stem. 1986; Malenka et al., 1986). Materials and Methods Antibodies. Antibodies against purified GAP-43 were raised in sheep as Received Apr. 7, 1988; revised July 20, 1988; accepted July 25, 1988. described (Neve et al., 1987; Benowitz et al., 1988). Cross-reactivity of the affinity-purified IgG fraction with human GAP-43 is described be- We gratefully acknowledge the support of the National Eye Institute (EY05690 low. to L.B.), the National Institute of Neurological, Communicative Diseases and Tissue. Brain tissue was dissected from 3 individuals, a 64-year-old Stroke and National Institute ofMenta.1 Health (NS25830 to L.B. and MH/NS3 1862 female (postmortem interval, PM1 = 9 hr, B1042), a 56-year-old male to E.D.B.), the American Heart Association (to S.P.F.), and the Hereditary Disease (PM1 = 12.5 hr. B1047). and a 52-vear-old male (PM1 = 2 hr. B1154). Foundation (to E.D.B.). We wish to thank David Weiner, Jonathan Winickoff, All subjects were free of any known neurological defect. Tissue was Paul Apostolides, and William Rodriguez for help in cutting and reacting sections, obtained with familial consent under the protocols of the McLean Hos- Dr. Ann McKee (Massachusetts General Hospital) for assistance in obtaining tissue, and Martha Shea for secretarial help. We also wish to acknowledge Paul pital Brain Tissue Resource Center. Cortical areas dissected included Apostolides for pointing out the similarities in distributional patterns between the primary somatosensory cortex (Brodmann area 1, Al), motor cortex GAP-43 and NMDA receptors in rat brain. (A4), frontal cortex (A10 and/or Al l), striate cortex (A17), peristriate Correspondence should be addressed to Dr. Larry I. Benowitz, McLean Hos- cortex (A 18 or A 19), inferior temporal cortex (A20), cingulate gyrus pital, 115 Mill Street, Belmont, MA 02178. (A24), parahippocampal gyrus (A28), supramarginal gyrus (A40), in- Copyright 0 1989 Society for Neuroscience 0270-6474/89/030990-06$02.00/O ferior precentral region (A44), and hippocampal region; also included The Journal of Neuroscience, March 1989, 9(3) 991 were the caudate putamen, amygdala, dorsal thalamus, hypothalamus, slightly higher molecular size of the human protein relative to and cerebellum. Tissue blocks, with volumes of approximately 0.5-l GAP-43 of rodents is at least partially explained by the primary cm3, were immersion-fixed in 4% paraformaldehyde buffered with 0.1 sequencedata (Kosik et al., 1988; Ng et al., 1988). On 1-D gel M phosphate, pH 7.4 (PBS), at 0°C overnight, cryoprotected with buff- ered 30% sucrose (0°C for 2 d), and stored at -70°C prior to sectioning. Western blots of total brain protein or SPM fraction, this 51 Unfixed tissue was also taken from several cortical regions and frozen kDa protein wasthe only band recognized by the antibody (data immediately to - 70°C for gel electrophoretic analysis of proteins. not shown). In brain tissuereacted with the anti-GAP-43 IgG, Immunocytochemistry. Sections were cut at 40 pm intervals on the a granular pattern of staining was seenin the neuropil (Fig. 1b), freezing stage of a sliding microtome (n = 2) or on a cryostat (-23°C) and stored for up to 2 weeks in PBS containing 0.1% NaN,. Cortical which varied greatly in intensity between different regions and sections were taken to include all laminae plus some underlying white laminae (Fig. lc). Most cell bodies and fiber tracts were un- matter. In the principal series to be analyzed, sections from all brain reactive, although stained somata were occasionally observed regions were processed together in the same 50 ml tubes. Free-floating in some regions (Fig. lc). In cortex, these were usually deep- sections were reacted for 30 min in CH,OH containing 1% H,O, to inhibit endogenous peroxidase activity, washed 3x in PBS, blocked lying pyramidal cells. CaseB 1154, however, showedno cellular with 20% normal rabbit serum (NRS) 30 min, and then incubated for staining anywhere. Conceivably, the presence of positively l-2 d with either neutral sheep serum, l/2000, as a control, or sheep staining somata in some casesbut not others may parallel the anti-GAP-43 serum (affinity-purified IgG fraction) at l/2000. Sections aging-relatedaccumulations of the protein visualized in the aging were rinsed 3 times, including once overnight, in PBS containing excess rat brain (Oestreicheret al., 1987). Control sectionsreacted with NRS (10%) NaCl(1.8%), and Triton X-100 (0.7%) in order to optimize elution of nonspecifically bound primary antibodies. Sections were then preimmune whole sheep serum instead of the affinity-purified incubated with a biotinylated secondary antibody to sheep IgG made anti-GAP-43 IgG showed no immunostaining at all (Fig. Id). in rabbit (l/250, 1 hr), washed 3x in PBS (+2% BSA, 0.3% Triton As another control, parallel experiments carried out in the rat X-100, 1% NRS), and then reacted with avidin-biotin complex con- showed that preabsorption of the anti-GAP-43 antibody with jugated to HRP according to the manufacturer’s specifications (Vector Labs, Burlingame, CA). HRP was visualized using 0.5% diaminoben- a 1OO-fold excessof purified GAP-43 eliminated almost all im- zidine and 0.01% H20, in 50 mM Tris-HCl (pH 7.5) for 5 min. Sections munoreactivity (Benowitz et al., 1988). were mounted onto chrom-alum-subbed slides, dehydrated, and cov- In the cerebral cortex, the intensity of GAP-43 immuno- ered. staining was generally low in primary sensoryand motor areas, Two-dimensional gel electrophoresis of synaptosomal plasma mem- but quite intense in associative regions.