MOLECULAR BF~UN RESEARCH ELSEVIER
Molccular Brain Rescarch 33 (1995) 277-287
R e s e a r c h report
Expression of agfin mRNA is altered following seizures in adult rat brain Lawrence T. O'Connor 1, Julie C. Lauterborn, Mar6n A. Smith *, Christine M. Gall Departmew: of Anatomy and Neurobiolog3,, University of California at lrt'ine, lrvine, CA 92717-1275, USA
Accepted 7 June 1995
Abstract Agrin mRNA is broadly distributed throughout the adult rat brain, consistent with its proposed role in synaptogenesis and the organization of synaptic proteins in the central nervous system. The present study examined the effect of neuronal activity on agrin mRNA expression in adult rat forebrain using the hilus lesion paradigm for seizure induction and in situ hybridization and polymerase chain reaction techniques for quantification and characterization of agrin mRNA content. Seizures induced rapid, prolonged, and region-specific changes in agrin mRNA expression with the most prominent alterations occurring in hippocampal and cortical neurons. However, there were no detectable perturbations in the relative abundance of alternatively spliced agrin transcripts in affected brain regions. Activity-dependent changes in agrin expression suggest a role for this protein in modifications of synoptic structure associated with functional synaptic plasticity. Keywords: Agrin; Synapse; S enaptic plasticity; Learning and memory; Seizure; Epilepsy; Gone expression
The formation and maintenance of chemical synapses is coordinated by signals exchanged between neurons and their targets. Much of what is known about this process comes from studies of the neuromuscular junction, where differentiation of the postsynaptic apparatas in skeletal muscle fibers during development and regeneration is induced by molecules supplied by motor neurons (reviewed in ). One such signalling molecule is agrin, a neurally derived extracellular matrix protein that directs the organization of high density aggregates of acetylcholine receptors (AChR) in the muscle fiber at the site of nerve-muscle contact (reviewed in [5,14]). In addition to ;.ts effects on AChR, agrin influences the distribution of other postsynaptic neuromuscular components, including acetylcholinesterase, 43 kDa AChR-associated protein, and laminin [45,62], suggesting that agrin plays a general role as a synaptic organizer at the neuromuscular junction.
"Corresponding author. Fax: (1) 1,714) 824-2932; Email: [email protected]
Present address: Department of Medical Science, University of Wisconsin, Madison. WI 53706-1102, USA. 0169-328X/95/$09.50 © 1995 Elsevier Science B.V. All rights reserved SSDI 0169-328X(95)00147-6
The possibility that agrin might also play a role in the formation and maintenance of synapses elsewheie in the nervous system was first suggested when agrin-like molecuies were identified in extracts of frog brain . Subsequently, using molecular probes derived from agrin cDNAs, it was shown that agrin mRNA is expressed in the brains of embryonic chick and rat [38,52] and developing peripheral ganglia [38,52,61]. Consistent with a general role for agrin as a synaptogenic protein, levels of agnn mRNA in developing peripheral ganglia  and brain  are highest during periods of synapse formation. Moieover, the recent finding that dystroglycan acts as an agrin receptor [4,8,26,57], together with earlier observations that dystroglycan is expressed in brain as well as skeletai muscle [25,30], suggests that a common signal transduction pathway for agrin exists throughout the nervous systeni. Activity-dependent alterations in gone expression are believed to contribute to long-term changes in synaptic physiology that underlie the acquisition and consolidation of memory [1,18,53]. ,Mthough many neuronal populations in adult brain contain significant levels of agrin mRNA, the highest levels of agrin gene expression are found among neurons that maintain a high degree of synaptic plasticity in the mature animal, including those in olfactory
L. 7". O'Connor et al. / Molecular Brain i~serlrch 33 (1995) 2 77-287
L. "L 0 'Connor etal. / Molecular Brain Researcli 3 ~ (1995) 277-2,~7
bulb, hippocampus, and neocortex . These observations sugge.~t that in addition to a role in synapse formation du.mo dew~lopment, agrin may serve as a regulator of long-term, activity-dependent alteration~ in synaptic function. As a first step towards testing this hypothesis, we examined the possibility that agdn gene expression might be aif,z-"~ed by neurona! activity. Experimem,d~y induced seizures have proven useful in studying the in vivo reg--'!~tion of many genes in mammalian brain. To this end, we have ~rudied the effects of lesion-induced recurrent limbic seizures. In t~-'_'s0aradigm, iron deposition at a small electrolytic lesion of the cz:~t~te gyms hilus induces behavioral and e',ectrographic limbic seizures which recur intermittently over a period of several hours without secondary neuronal degeneration or the potentially confounding effects of convulsant drugs [2,21,47]. We have examined the effect of hilus legion (HL)-induced seizures on agrin mRNA levels in the adult rat brain. Our results show that seizure activity alters agrin gene expression in specific neuronal populations, including cells in neocortex and hippocampus. These effects are rapid and long-lasting, and are thereby consistent with a role for agrin in synaptic plasticity in the adult CNS.
no surgery was performed. Additional control rats were lesioned in the hihxs with platinum-~ridium wire: platinum wire lesions produce a similar size ablation field as a stainless steel electrode but do not induce seizure activity [9,47]. At various intervals after HL placement, rats were killed either by an overdose of sodium pentobarbhal and then intracardially perfused with 4% paraformaldehyde in 0.1 M phosphate buffer (pH 7.2) for in situ hybridization analysis or by decapitation under fluorothane anesthesia for RNA isolation. For in situ hybridization, tissue was processed from 22 anesthetic and 4 platinum lesion control rats, and from HL rats killed at 3 h (3), 6 h (5), 8 h (4), 10 ("~), 12 h (6), 18 h (3), 24 h (8), 30 h (2), 48 h (5), and 96 h (4) post lesion (numbers of rats per time point indicated in parentheses). Due to the lack of sufficient numbers of sections for densitometric mca=vrement, fewer rats were used for the quantitative analysis of some brain areas; the numbers of rats used for each region are given -~n the appropriate figure legend. For the PCR analysis, tissue samples were collected from untreated control rats and from rats killed at 3, 6, 12, 24, 48, and 96 h after a contralateral HL (n -- 2 rats/time point). 2.2. In situ hybridization
2. Materials and methods 2.1. Hilus lesion Adult male Sprague-Dawley rats (Simonson Labs, 250300 g) were used. Experimental rats (n = 56 total) were anesthetized with 50 m g / k g of ketamine and 10 m g / k g of xylazine and a focal electrolytic lesion was placed in the hilus of the right dentate gyrus (stereotaxic coordinates: 3.8 mm posterior and 2.5 mm lateral to bregma; 3 mm ventral to brain surface) with a stainless steel electrode using anodal current (0.8 mA for 7 s). Lesions of this type induce recurrent and bilateral electrographic seizures within hippocampus and behavioral seizures of the limbic-kindling type. Seizures begin approximately 2 h after lesion placement ar.d recur intermittently for 8-10 h thereafter, with over 80% of seizure discharges occurring within the first 3 h of seizure onset [22,47]. All experimental I-IL rats exhibited at least two stage 4 (rearing with forelimb clonus) or stage 5 (rearing with clonus and falling) iimbic seizures  within 5 h of lesion placement. Paired anesthetic control rats were anesthetized with ketamine/xylazine but
Perfusion-fixed brains were postfixe4 in the perfusate overnight at 4°C, cryoprotected with 20% sucrose in 4% paraformaldehyde for 48 h, and then frozen on dry ice and either processed immediately or stored at -70°C. Brains were sectioned at a thickness of 25/xm in a coronal plane on a freezing microtome. Free-floating tissue sections were then processed for in situ hybridization to localize agrin mRNA using a riboprobe derived from a 3.2 kb agrin eDNA described previously  correspoading to nucleotides 4070-7286 of rat agrin . Sense and anfisense riboprobes were transcribed from linearized plasmids using either T7 or T3 RNA polymerase in the presenc: of [35S]UTP. Hybridization and post-hybridization steps including RNase A digestion and washing conditions were identical to those previously described [33,46]. Tissue from experimental and paired control rats was processed in parallel in order to minimize variation that might occur between experiments. After in situ hybridization, sections were mounted onto gelatin-coated slides, air dried, and exposed to /J-Max Hyperfilm (Amersham, Arlington Heights, IL) for 1-3 days at room temperature. The tissue was subsequently defatted with chloroform, dipped in Ko-
Fig. 1. Seizure activity induces changes in agrin mRNA expression in the adult rat forebrain. Darkfield low magnification pho;omicrographs showing in situ hybridization of antisense (A-D, F) and sense (E) aSs-labeled agrin cRNA probes in coronal hemi-sections from control (A, C), experimental seizure (B, D), and platinum wire lesion (F) rats killed 24 h after a contralateral hilus lesion. Compared to control, hybridization is greater in experimental rats (B, D) in hippocampal stratum granulosum (sg), lateral septum (LS), olfactory tubercle (Tu), and caudate putamen (CPu) but markedly lower in the superficial cell layers of piriform cortex (PC) and superficial (s) and deep (d) cell layers of neocortex (NC). In contrast, no significant difference in labeling density is apparent between the unlesioned control [C) rat and the rat lesioned with a platinum electrode that does not induce seizures (F). Only background levels of labeling are observed in experimental rats hybridized with a sense probe (E). Note no labeling is evident in the internal capsule tic) or any other major fiber tract of control or experimental seizure rats. VP, ventral posterior thalamus; Hy, hypothalamus; sp, hippocampal stratum pyramidale. Scale bar: 1500 #m.
L.T. 0 'Connor et al. / Molecular Brain Research 33 (; 995) 2 77-28 7
dak NTB2 emulsion (1:1 with H_,O; Eastman-Kodak, Rochester, NY) and exposed for 3-6 weeks at 4°C. Following autoradiographic development, sections were stained with cresyl violet and coverslipp_ed in Permount (Fisher Scientific, "lustin, CA). Adjacent brain ~ections from some animals were mounted onto slides and stained with cresyl violet alone in order to view the cytoarchitecture and lamination of neocortex.
2.3. Quantitation of in situ hybridization Levels of hybridization were quantified by calibrated densitometric analysis of film autoradiograms using a MicroComputer Image Device (imaging Research Inc., Ontario, Canada). Film density was calibrated and linearized relative to sections of ~4C-labeled brain paste standards which were exposed to film with tissue sections as described in detail elsewhere . Individual animal means were calculated from multiple meamrements of at least five different tissue sections from the following areas: dentate gyrus stratum granulosum (dorsal leaf); CA1 and CA3 stratum pyramidale of rostral hippocampus; layer II of the piriform and entorhinal cortices; layers ll/III, V and VI of parietal neocortex; the superficial cell layer of olfactory tubercle; lateral and medial septum; caudate putamen; the ventral posterolateral, ventral posteromedial, and central medial nuclei of thalamus, and dorsomedial nuclei of hypothalamus. Major fiber tracts (e.g. corpus callosum, internal capsule, or fimbrae fornix) exhibited background levels of autoradiographic labeling ~¥~th the antisense agrin cRNA probe; therefore for each r~,t, measures of these fiber tracts were used to ~:alcu!ate~ a mean background density which was then subtracted [rom values obtained from regions of neuropil detailed above. To avoid possible direct effects of the HL (independent of seizure activity), all measures were made on tissues contralateral to the HL. In order to compare effect:~ on different brain areas on the same scale, individual animal means for experimental HL rats are presented graphically as a percent of the corresponding region measured in paired control rats. In all cases the significance of the effect of treatment (i.e., HL seizures) was evaluated by one way analysis of variance (ANOVA) conducted using i~,~dividual animal means (i.e., not normalized to control values) and the Fisher PLSD test for post hoc comparisons.
2.4. PCR Adult Sprague-Dawley rats were decapitated at various times following the hilus lesion. Brains were then removed, dissected on ice, quickofrozen on dry ice, and stored at -70°C. RNA was i~,~!ated from ~r-,~zea tissue by acid guanidinium thiocyanate~phen~.!-c~!~roform extrac~ tion . Alternatively spliced agrin mI~NAs were idc~.'i~ fled by RT-PCR amplification of RNA isolated from specific brain regions using nested primers flanking the z
locus of the rat agrin cDNA [15,16] as previously described . In some experiments, different forward and reverse ,~rimers were used to selectivel) amplify mRNAs encoding agrin proteins with high AChR aggregating activity. cDNAs representing agrin8 and -19 mRNAs were amplified using a forward primer corresponding to nucleotides 5528~5549 of rat agrin who,;e 3'-end is embedded within exon 32, togeti~er with a reverse primer complementary to nucleotides 5610-5631 which is common to all agrin mRNAs. Similarly, agrin11 and -19 were selectively amplified using a pan-specific forward primer corresponding to nucleotides 5462-5485 and reverse primer complementary to nucleotides 5562-5582 present within exon 33. PCR products were separated by electrophoresis on an 8% polyacrylamide gel and visualized by autoradiography. The relative abundance of each agrin mRNA was determined by quantitative analysis on a phosphorimager (Molecular Dynamics, Sunnyvale, CA).
3.1. Seizure activity alters agrin mRNA expression Seizures were induced in adult rats by the placement of an electrolytic lesion in the hilus of the dentate gyrus and the distribution and levels of agrin mRNA were evaluated by in situ hybridization to a pan-specific 35S-rat agrin cRNA probe  that recognizes all of the alternatively spliced agrin mRNAs. When compared to the anesthetic controls, HL-induced seizures were associated with changes in agrin mRNA expression that were broadly distributed throughout the forebrain (Fig. 1). However, both the magnitude ano direction of change in agrin mRNA expression appeared to be region-specific. For example, elevated levels of hybridization were observed in dentate gyrus stratum granulosum whereas hybridization in stratum pyramidale ~f hippocampal fields CA1 and CA3 was comparable in HL and control rats (Fig. 1D). Moderate increases in hybridization were observed in lateral septum (Fig. 1B), caudate putamen (Fig. 1B), and olfactory tubercle (Fig. 1B) in HL rats. In contrast, levels of hybridization of the agrin cRNA in experimental rats were below control values in se~,~ral areas including the piriform (Fig. IB) and entorhinal cortices (data not shown) and superficial and deep layers of neocortex (Fig. 1D). Finally, in many brain regions labeling appeared unaffected by seizures (e.g., thalamic and hypothalamic nuclei shown in Fig. 1D). Consistent with previous descriptions of agrin mRNA expression in adult rat brain , labeling with the antisense probe along major fiber tracts in control rats was similar to that obtained with the sense probe and remained unchanged after seizures (Fig. 1E). "lw,~ ca.n_trori groups were included in this study to 6c:,:~-~e whe:.her chan~es,~ in agrin cRNA hybridization were specitlc,~) .....~-"!~ted with HL:~duced seizure ac-
L.T. 0 'Com~or et al. / Molecular Brain Reseats, .:_:' :: "i!,?¢i 977-287
t~ ,ity. In the first group, animals received ~,nesthesia but l~o lesion. To control for the effects of the lesion per se, a second group of rats received an electrolytic hilar lesion using a platinam-iridium electrode. Lesions generated with a platinum-iridium electrode proouce a similar size ablation field as that made by stainless steel electrodes but do not induce seizures [9,47]. lr~ all regions examined, levels of hybridization to sections taken from both control groups were not significantly different from comparable section,s from untreated age-matched control rats (Fig. IF). Thus, changes in agrin mRNA levels described above occur in association v,,it~ HI.-induced seizures and are not produced by the anesthetic or by hiia~ tissue d:~mage. 3.2. Temporal changes it: agrin mRNA levels The results described above demonstrate that HL-induced seizures influence agrin mRNA levels in adult rat brain. To investigate this phenomenon in more detail, we evaluated the thne course of changes in agrin mRNA expression following HL placelnent. As shown in Fig. 2, HL-induced seizures result in a biphasic change in agrin cRNA hybridization in the dentate stratum granulosum. Agrin mRNA content decreased during the period coincident with seizure activity (approximately 2-12 h post lesion; Fig. 2B) and then increased above control levels by 24 h and remained elevated up to 4 days post lesion (Fig. 2C,D). In contrast, in stratum pyramidale, hybridization was uniformly high in regions CA1 and CA3, and remained low in region CA2, at all time points examined. Time-dependent changes in agrin mRNA levels in hippoeampus were confirmed by densitometric analysis of film autoradiograms (Fig. 3). ANOVA demonstrated a significant effect of seizures on agrin cRNA labeling in stratum granulosum (P<0.0001). By 6 h post tesion, hybridization in the granule cell layer had declined by about 50% and was maintained at this level through 12 h post lesion, At later time points, levels of hybridization in,:~,eased such that by 24 h through 48 h post lesion they were approximately twice control levels. Densitometric analysis of stratam pyramidale revealed a small reduction in hybridization in both region CA! and region CA3 of HL rats but this effect was not statistically significant. The time course, direction and magnitude of seizure-induced changes in agrin mRNA levels in entorhinai and pirifonn cortices were essentially equivalent. Its shown in Fig. 4, in both of these cortical regions labeling density in layer I! was decreased by about 25% during 'the period of seizures, followed by a further decline to about 40% control values by 24 h post Lesion. In contrast, hybridization in the olfactory tubercle was increased at the 24 h time point although this effect was not statistically significant. In neocortex, the results of our initial studies saggested that HL-indueed seizures elicited changes in agrin mRNA expression that were associated with specific layers. As
Fig. 2. Seizures induce biphasic changes in agrin mRNA expression in specific cell populalions of the hippocampus. Darkfield photomicrographs showing the autoradiographic localization of agrin antisense [35S]cRNA hybridization in coronal sections through rostral hippocampus of control (A) and experimental seizure (B-D) rats killed 12 (B), 24 (C), and 96 I1 (D) after HI. placement. In stratum granulosum (sg) of HL rats, hybridization densities are markedly reduced at 12 h in comparison to the paired control, but are then elevated above control levels at 24 and 96 h. In contrast, labeling within stratum pyramidale (sp) remains comparable to control levels. Scale bar: 500/xm.
L.T. 0 'Connor et ai./ Molecular Brain Research 33 (1995) 277-287
1so I  CAI @ CA3
< z <> OT .i
i Oi 0
. o i
. 1"2 1'8 2~4 Time Post-lesion (Hours)
, ;I 30
,~.~. 48 72 96
Fig. 3. HL seizures induce r,,~,J ,~,::: prolonged changes in expression of agrin mRNA in the hippocampus. Graph showing quantitative changes in agrin cRNA hybridization densities in rostral hippocampus. In this and subsequent graphs, measures from each experimental rat are expressed as a percent of hybridization densities from the same field in sections from paired control rats, and the bar over the X axis indicates the period of seizure recurrence. Note that hybridization in stratum granulosum (SG) is decreased during recurrent seizure activity; this is followed by a much longer period of elevated hybridization. In contrast, seizures had no significant effect on labeling densities in CA1 and CA3 stratum pyramidale. Values at each time point represent group means+SEM (n > 3 for each time point except 30 h where n = 2) ( P < 0.0001 for effect of treatment on SG, ANOVA; " P < 9,05, " ° * P < 0.001 for comparison with control values, Fisher PLSD~
shown in Fig. 5, the most prominent change in agrin mRNA content was associated with layers II and III. When compared to controls, hybridization appeared maximally
D PC @ EC 6
48 72 96
Time Post-lesion (Hours)
Fig. 4. Time course of seizure-induced changes in the density of agrin cRNA hybridization in olfactory tubercle (OT) and layer II of piriform cortex (PC) and entorhinal cortex (EC). For the piriform and entorhinal cortices, the decline in agrin cP,NA hybridization begins during the period of seizures, is maximal at 24 h post lesion and then returns toward control levels. Bar indicates period of recurrent seizure activity. Measures at each time point are group mean values_+ SEM with n > 3 for each time point. (For overall effect of treatment, P < 0.001 for EC, P < 0.0001 for PC, ANOVA; " P < 0.05, " * P < 0.0i, " " " P < 0.001 for PC in comparison with values from unlesioned control rats, Fisher PI.SJD. Similar paired comparison values were obtained for EC).
decreased at 24 h (Fig. 5C) but returned to near normal levels by 48 h post lesion JFig. 5D). A similar, but generally less marked decline, was evident in layers V and VI. In contrast~ little ot no change was observed in layer IV. Consistent with our earlier report , hybridization in
Fig. 5. Seizures result in changes in agrin mRNA expression in neocortex. Darkfield photomicrographs ( B , D ) showing the autoradiographic localization of ["SS]cRNA hybridization in coronal sections through parietal neocortex from control (B)and experimental rats (C-D) killed at 24 (C) and 48 h (D) post lesion. L.~minae were defined by the cytoarchitecture of the parietal neocortex as illustrated in the Nissl-stained section in panel A. In control tissue, neurons labeled with the antisense agrin cRNA probe appear broadly distributed across superficial (layers II/III and IV) and deep neocortical layers (layer VI). Hybridization densities over these same cell layers are markedly lower at 24 h post lesion but return to levels similar to control by 48 h. Note that labeling is less affected by treatment in cortical layer IV. The boundry between layers I and II (arrowhead) and the position of layer IV are indicated in each panel. Scale bar: 300/.tin
L.T. O'Cmmor et al./ Molecular Brm, Research 33 (1995) 277-287 150
beltum, was not different from control values at 24 h post lesion and remained constant throughout the time course examined. In addition, although qualitative increases in hybridization were apparent in caudate putamen of some rats killed 24 h post lesion, there was considmable interanimal variability in this ,.egion and statistical analysis revealed no significant change in hybridization ~,evels
Ne,~c,~rtex Laver Ililll
T %. j. ~..
=- 100~ L~ i
r- 50q ._= <
3.3. Ahernatit'e splicing of agrin mRNA I
~-- 18~ '--- 2~4 ' 30~ //-----48 Time Post-lesion (Hours)
Alternative splicing of exons 32 and 33 at the z locus gives ri~e to four agrin isoforms (agrin0, -8, - i l , and -19) that differ in AChR aggregating activity [i5,16]. We have shown that each of these alternatively spliced agrin mRNAs is expressed in adult rat forebrain . To examine the possibility that HL-induced seizures might cause changes in the pattern of alternative agrin RNA splicing, RNA iscdated from neocortex and hippocampus at different times following hilus lesion was subjected to m,o rounds of amplification by PCR using nested primers flanking the alternatively spliced z locus as previously described , At each' time point examined, both the pattern and relative abundance of agrin mRNAs present in RNA isolated from either neocortex or hippocampus of lesioned animals was mdiatinguishable from their anesthetized but unlesioned controls (Fig. 7A). Similar results were obtained in a second experiment using RNA isolated from the same regions in different animals (data not shown). The results of the in situ hybridization studies demonstrated that within a particular brain area the major changes in agrin mRNA levels following hilus lesion were usually confined to a specific lamina or :sub-region. We were concerned therefore that our inability to detect any change in the pattern of alternative splicing of agrin mRNA might be due to heterogeneity in our RNA isolates. Accordingly,
Fig. 6. Time course of HL-induced changes in the hybridization densities in neocortical layers II/llI. Bar indicates period of recurrent seizure activity. Values at each time point are group mean values+SEM with n > 3 per time point except for 6 h and 30 h ~,vhere ,i = I ( P = 0.0002 for overall effect of treatment, ANOVA: " " P < 0.01. " " " P < 0.(,X)I for comparison with control values, Fisher PLSD).
layer I of neocortex was low and indistinguishable from background at all time points examined. As shown in Fig. 6, densitometric analysis demonstrated a significant and time-dependent effect of seizure v neocortex ( P = on the agrin mRNA content of ,,,.,,.;,,,ol 0.0002, ANOVA). In comparison to control values, labeling of cortical layers l I / I l I declined almost five-fold by 24 h post lesion. A similar trend was evident in layers V and V1 (data not shown), but for these laminae we were unable to detect significant differences from control values based on the present sample size. Quantitative analysis of neocortex gave the impression of two phases in the decrease in agrin mRNA expression, with a modest decline coincident with the period of seizures and a greater decrease during the 12 h following cessation of seizure activity. Hybridization of the agrin cRNA in several other brain regions including dorsal thalamus, hypothalamus and cere. . . . . . .
Fig. 7. Alternative splicing of agrin m R N A is unchanged following HL-induced seizures in adult rat torebrain. First strand cDNA synthesized from RNA isolated from different regions of the adult rat forebrain was sub~ect to two rounds of amplification using nested primers flanking the alternatively :,pliced z locus of rat agrin and fractionated by polyacrylamide gel e!ectrophoresis. Film autoradiograms from typical experiments show the relative levels of each of the four agrin isoforms expressed in (A) the neocortex and hippocampus or (B) the dentate gyms from an unlesioned control rat (u) or experimemal seizure rats killed at the indicated number of hours post lesion. The control lane (c) represents a reaction in which RNA was omitted from first strand cDNA synthesis. Small differences in band intensities (e.g. hippocampus, 6 h post lesion) were not consistently observed, and are likely to represent animal to animal variation. No significant difference in the relative abundance of each agrin mRNA was detected when data from two experiments using independent RNA isolates were analyzed quantitatively using a phosphorimager.
L.T. O'Connor et ai. / Molecular Brain Re.search 33 ¢t 995) 2 77-28 7
the agrin mRNA profile for RNA isolated from pieces of tissue dissected from hippocampus, including the dentate gyrus and a small portion of the overlying region CA1, was analyzed at 12 and 24 h after HL placement (Fig. 7B). Despite enrichment for granule cell mRNA afforded by the dissection, we again found no evidence for a change in the relative abundance of alternatively spliced agrin mRNAs even when measured at times when levels of agrin mRNA expression by in situ hybridization are at their minimum or maximum values. Moreover, similar results were oi~tained in experiments in which we examined the relative abundance of agrin mRNAs with high AChR ~gg!egating activity (agrin8, -11, and -19) by replacing the forward or reverse primers usea in the second routed PCR with either forward or reverse primers located withi, the sequence of exons 32 or 33, respectively, precluding amplification of PCR products for agrin0, the most abundant isoform. Even under these conditions: the relative abundance of agrin8 versus agrinl9, or agrinll versus agrinl9, was unchanged in RNA isolated from the dentate gyrus following seizures (data not shown).
We have previously shown that agrin mRNA is expressed by many neuronal populations in the adult rat brain, particularly in regions associated with high levels of synaptic plasticity . In the present study, we demonstrate that HL-induced seizures cause changes in agrin gone expression in several brain structures. Changes in agrin mRNA levels were restricted to neurons, had rapid onset, and persisted for periods of 24 h or more after the seizures stopped. These results, demonstrating that agrin gone expression in the adult rat brain is regulated by the intense neuronal activity of seizure, are consistent with the hypothesis that agrin plays a role in activity-dependent changes in synaptic structure and neuronal architecture found to occur following seizures [7,51,58] and possibly in association with lower, more normal, levels of physiological activity [6,10,35]. These results further raise the possibility that agrin may be important for cellular mechanisms which mediate long-term enhancement of synaptic function in the CNS. Numerous studies support the hypotllesis that changes in gone expression following seizures are dependent on neuronal activity [21,44,53] and several lines of evidence indicate that the selective changes in agrin mRNA levels observed in the present study are similarly activity-induced. First, areas of the brain not directly involved in propagation of seizures, such as thalamus and hypothalamus, showed no change in agrin mRNA content. Moreover, agrin expression was unaltered in regions of the brain containing few neurons (e.g., cortical layer I and raajor fiber tracts), indicating that changes in expression are confined to neurons and not associated with gila or other
non-excitable cells. Second, the results i~dicate that the changes in agrin gone expression are not a consequence of damage, deafferentation, or indirect effects of the lesion per so. The lesion is small, has not been found to cause secondary neuronal degeneration in the co~-~tralateral hemisphere , and does not deafferent many of the fields showing major HL-induced changes in agrin mRNA content (e.g., neocortex and piriform cortex). In addition, platinum wire hilar lesior~s, which ablate the hilus but do not induce seizures, were not found to perturb agrin mRNA content in any field measured. Therefore, we conclude that changes in agrin gone expression are directly related to the intense neuronal activity produced by seizures. It will be interesting to determine if agrin gone expression can be modulated by lower levels of physiological activity such as those which induce long-term potentiation. As indicated above, the HL had little if any effect on a~r;n mRNA expression in several brain regions. Whereas some of these areas are likely to be outside the spread of seizure activity, other regions such as the hippocampus and neocortex are not. However, even within these two regions, distinct laminar changes were observed. For example, changes in agrin mRNA expression in hippocampus were restricted to the granule cells. Based on circuitry, all neurons throughout the hippocampus would be subject to increased activity during seizures, a prediction supported by the observation that HL seizures increase the expression of c-los and other immediate early genes (lEG) throughout hippocampus (O'Connor, unpublished observations; ). Thus, agrin gone expression in some neurons, including those that normally express high levels of agr~n mRNA, appears to be less sensitive or insensitive to regulation by neuronal activity. Changes in agrin mRNA content in neocortex were also cell-specific. I.e.~,-els of agrin mRNA in neocortical layers II-IV are normally high . However, following seizure there was a pronounced decline in agrin mRNA expression in layers II/III whereas layer IV was not affected. This pattern may reflect lesser zctivatien of layer IV neurons during the seizure episode. Consistent with this possibility, we have observed that levels of c-los mRNA following HL-induced seizures are also highest in layers lI/III compared to other layers (t2"Connor and Lauterborn, unpublished observations). Alternatively, seizure-induced regulation of agrin gene expression may correlate with differences in plasticity between cortical layers. This is suggested by the observation that modification of receptive fields within somatosensory cortex following a change in sensory experience occurs rapidly in superficial and deep cortical layers but is considerably delayed in layer IV . Activation of NMDA receptors has been implicated in neuronal plasticity. In cortex~ NMDA receptor density is high in superficial layers but low in layer IV . Increases in the expression of IEG and several neuron-specific genes following synaptic activation can be blocked by NMDA antagonists [40,63]. It will be interesting to learn
L.T. 0 'Com~or et al. / Molecular Brain Research 33 (1995~ 277-287
whether changes in agrin gene expression exhibit a s~milar pharmacology. Experiments in cell culture that al.!ow direct depolarization or activation of specific neurotransmitter receptors could be used to resolve these issues. Agrin mRNA content varied with a complex time course after HL-induced seizures. The initial response occurred during the seizure period and wa~ marked by a decline in agrin mRNA levels. Subsequently, mRNA content decreased %rther or increased to above -mrmal levels depending on the region. As described abo~e, it seems unlikely tha~; declines in agrin expression represent a response to injury. Rather, compicx changes in mRNA content over time could reflect the recruitment of specific populations of neurons that respond to elevated levels of neuronal activity with different latencies. However, in stratum gran:dosum, which contains a relatively homogeneous population of neurons, changes in agrin mRNA levels were clearly biphasic, indicating that even for a single cell type, agrin expression is controlled by signals that vary with time after stimulus onset. Numerous studies have demonstrated that seizures induce a rapid, transient induction of lEG expression and that many of the lEG products are transcriptional regulatory factors that influence the expression of families of late response genes [44,53]. The time of onset of chan, zes in agrin mRNA expression is consistent with the possit, ility that agrin is a member of the late response gene group The various IEG mRNAs and products induced by activity are knovn to have markedly different periods of elevation folle~ing seizure [23,54,56,60] and, through participation in different DNA binding dimers, to both facilitar.e and suppress the expression of different target genes [24,32,56]. Thus, the early and late phases in the agrin mRNA response could reflect the relative balance of lEG p~t, ducL~ in the cell at different post-stimulus latencies. Alternatwely, the late changes in agrin expression could reflect trophic interactions. Hilus lesion seizures induce the expression of nerve growth factor (NGF) [20,34], brain-derived neurotrophic factor [3I], and basic fibroblast growth factor  in hippocampus and other brain areas with time courses that precede the late phase of activity-dependent changes in agrin gene expression. While it remains to be determined whether neurotrophic factors affect agrin expression in CNS neurons, the observation that NGF increases the agrin mRNA content of PC12 cells (Smith, unpublished observations) is consistent with this possibility. Alternative splicing of agrin pre-mRNA gives rise to four major protein isoforms with different ACnR-aggregating activities. Several studies have shown in both PNS and CNS the pattern of alternative splicing of agrin mRNA is regulated during development [28,39,61]. A striking observation in the present study is that dest;ite changes in the levels of agrin mRNA, paaerns of a~z,zmative splicing were unaffected, indicating that activity-dependent changes in agrin gene expression involve coordinated expression of all four agrin transcripts. Thus, signals regulating agrin
gene expressi,~, i~voS.e6 by activity may be different from those present during de~,ek~pmen~, H,,~wever, we do not know if activity-dep,mdent changes in agrin ge-e expression reflect the resp,nse of all neurons or subsets of cells within a given region. Analysis of alternative splicing by single cell RT-PCR has shown that small but significant differences exist in t~,e agrin mRNA profiles of twopopulations of choline~gic neurons in the chick ciliary genglion . Applying this technique in the rat CNS s!':o~!~ help resolve whether seizures induce cell-specific changes in alternative splicing of agrin RNA. Previous studies have shown that experimentally induced seizures and chronic human epilepsy lead to enduring changes in synaptic organization and fiber architecture within hippocampus [7,29,50,59]. If, as appears to be the case at the neuromuscular junction, agrin plays a role as an anterograde signal regulating the formation and maintenance of neuronal synapses, a sustained increase in expression in stratum granulosum would be consistent with an increased requirement for agrin to build and maintain a seizure-induced increase in granule cell synaptic fields. Apparently at odds with this interpretation is the fact that the initial and, in many instances, the second phase of the seizure response, were characterized by a decline in agrin mRNA lew;ls. Perhaps remodelling of the affected synapses requires a lower level of agrin expression. However, there is no reason a priori to a~sumc that activity-induced changes in synoptic circuitry involve strengthening or addition of excitatery synapses rather than weakening or loss of inhibitory contacts. Indeed, the decline in agrin expression observed in some populations of neurons is consistent with the results of recent studies showing that synoptic modifications associated with epileptiform activity may include the loss of functional GABA receptors . The observation that agrin supplied by muscle is also enriched at neuromuscular synapses [13,36,37] suggests that activitydependent regulation of agrin gene expression in postsynoptic neurons will also be an important factor for consideration. Wh~e ~,~her roles for agrivJ: such as local regulation of proteinase activity , remain to be examined, regulation of agrin gene expression by neuronal activity is likely to have important functional consequences in the brain.
Acknowledgements The authors would like to thank Dr. D.K. O'Dowd for her critical reading of earlier versions of the manuscript. This work was supported by NS33213, Committee' of 1000 Young Investigator Award, and IRU in Molecular Neurobiology to M.A.S., AG00533 and RSDA MH00974 to C.M.G., NIH Postdoctoral Training Fellowship NS07351 to L T O . , and Amerh, an Epilepsy Society Fellowship EFA1800 to J.C.L.
L.T. 0 'Connor etal. / Molecular Brain Res~,'arch 33 (1995) 2 77-287
References  Agranoff, B.W., Learning and memory: biochemical approaches. In G.J. Siegel et ai. (Eds.), Basic Neurochemistry, Little Brown, Boston, MA, 1981, pp. 801-820.  Baudry, M., Lynch, G. and Gall, C., h~duc:ion of ornithine decarboxylase as a possible mediator of seizure-elicited changes in genomic expression in rat hippocampus, J. N~urosci, 6 (1986) 34303435.  Bimc, S.L., Payan, D.G. anti Fisher, J.M., Isoforms of agrin are widely expressed in the developing rat and may function as protease inhibitors, Dev. Brain Res., 75 (1993) 119-129.  Bowe, M.A., Deyst, K.A., Leszyk, J.D. and Fallon, J.R., Identification and purification of an agrin receptor from Torpedo postsynaptic membranes: a heteromeric complex related to the dystroglycans, Neuron, 12 (1994) 1173-1180.  Bowe, M.A. and Fallon, J.F., Tile role of agrin in synapse formation, Annu. Retd. Neurosci., 18 (1995) 443-462.  Buchs, A., Stoppinn,.', L. and Muller, D., Identification of stimulated synapses reveals major morphological changes following LTP induction, Soc. Neurosci. Abstr., 20 (1994) 266.  Bundman, M.C., Pico, R.M. and Gall, C.M., Uitrastroctural plasticity of the dentate gyros granule cells following recurrent limbic seizures: !. Increase in somatic spines, titppocampus, 4 (1994) 601-610.  Campanelli, J.T., Roberds, S.L., Campbell, K.P. and Schel!er, RH., A role for dyst~ophin-assoeiated glycoproteins and utrophin in agrin-induced AChR clustering, Cell, 77 (1994) 663-674.  Campbell, K.A., Bank, B. and Milgram, N.W., Epileptogenic effects of electrolytic lesions in the hippocampus: Role of iron deposits, Exp. Neurol., 85 (1984) 506-514.  Chang, F.-L.F. and G,'eenough, W.T., Transient and enduring morphnlogical correlates of synaptic activity and efficacy change in the rat hippocampal slice, Brain Res., 309 (1984) 34-46.  Chomczynski, P. and Sacchi, N. Single-step method of RNA isolation by acid guanidinium thiocyanate-phenol-chlorolorm extraction, Anal. Biochem., 162 (1987) 156-159.  Diamond, M.E., Huang, W. and Ebner, F.F., Laminar comparison of somatosensory cortical plasticity, Science, 265 (1994) 1885-1888.  Fallon, J.R. and Gelfman, C.E., Agrin-related molecules are concentrated at acetylcholi~,c receptor clasters in normal and aneural developing muscle, J. Cell Biol., 108 (1989) 1527-1535.  Fallon, J.R. and Hall, Z.W., Building synapses: agrin and dystroglycan ~tick together, Trends Neurosci., 17 (1994) 469-473.  Ferns, M., Hoch, W., Campineili, J.T., Rupp, F., Hall, Z.W. and Scheller, R.H., RNA splicing regulates agrin-mediated acetylcholine receptor clustering activity on cultured myotubes, Neuron, 8 (1992) 1079-1086.  Ferns, M.J., Campanelli, J.T., Hoch, W., Scheller, R.H. and Hall, Z., The ability of agrin to cluster AChRs depends on alternative splicing and on cell surface proteoglycans, Neuron, 11 (1093) 491-502.  Gall, C. and Lauterborn, J., Activity-dependent gene expression in hippocampus and cortex: implications for memory storage mechanisms. In L.R. Squire et al. (Eds.), Memory Organization and Locus of Change, Oxford University Press, New York, 1992, pp. 301-329.  Gall, C., Lauterborno,J., Bundman, M., Murray, K. and Isackson, P., Seizures and the regulation of neurotrophic factor and neuropeptide gone expression in brain. In V.E. Anderson et al. (Eds.), Genetic Strategies in Epilepsy Research, Elsevier, Amsterdam, 1991, pp. 219-239.  Gall, C.M., Berschauer, R. and Isackson, P.J., Seizures increase basic fibroblast growth factor mRNA in adult rat forebrain neurons and glia, MoL Brain Res., 21 (1994) 190-205.  Gall, C.M. and lsackson, P.J., Limbic seizures increase neuronal production of messenger RNA for nerve growth factor, Science, 245 (1989) 758-761.  Gall, C.M. and Lauterbomo, J., The dentate gyros: a model system
for studies of neurotrophin regulation. In C.E. Ribak, C.M. Gall and L. Mody (Eds.), The Dentate Gyrus and its Rote in Seizures, Elsevier, Amsterdam, 1992, pp. 171-185.  Gall, C.M., Pico, R.M. and Lauterborn, J.C., Focal hippocampal lesions indt~ce seizures and long-lasting changes in moss~, fiber enkephalin and CCK immunoreactivity. Peptides, 9 (1988) 7~-84.  Gass, P., Herdegen, T., Bravo, R. and Kiessling, M., Induction of !~ .~:,cdiate early gene encoded proteins in the rat i~ippocampus after bicucuiline-induced .-,eizures: differential expression of KROX-24, FOS and JUN proteins, Neuroscience, 48 (1992) 315-324.  Gass, P., Herdegen, T., Bravo, R. and Kiessling, M., Spatiotemporal induction of immediate early genes in the rat brain after iimbic seizures: effects of NMDA receptor antagonist MK-801, Eu. J. Neurosci., 5 (1993) 933-943.  Gee, S.H., Blacher, R.W., Douville, P.J., Provost, P.R., Yurchenco, P.D. and Carbonetto, S., Laminin-binding protein 120 from brain is closely related to the dystrophin-associated glycoprotein, dystroglycan, and binds with high affinity to the major heparin binding domain of laminin, d. Biol. Chem., 268 (1993) 14972-14980.  Gee, S.H., Montanaro, F., Lindenbaum, M.H. and Carbonetto, S., Dystroglycan-a, a dystrophin-associated glycoprotein, is a functional agrin receptor, Cell, 77 (1994)675-686.  Hall, Z.W. and Sanes, J.R., Synaptic structure and development: the neuromuscular junction, Neuron, 10 (Suppl.) (!993) 99-121.  Hoch, W., Ferns, M., Campenelli, J.T., Hall, Z.W. and Scheller, R.H., Developmental ~egulation of highly active alternatively spliced forms of agrin, Neuron, 11 (1993) 479-490.  Houser, C.R., Morpliologicai changes in the dentate gyros in human temporal lobe epilepsy. In C.E. Ribak, C.M. Gall and L. Mody (Eds.), The Dentate Gyrus and its Role in Seizures, Elsevier, Amsterdam, 1992, pp. 223-234.  Ibraghimov-Beskrovnaya, O., Ervasti, J.M., Leveille, C.J., Slaughter, C.A., Sernett, S.W. and Campbell, K.P., Primary structure of dystrophin-associated glycoproteins linking dystrophin to the extracellular matrix, Nature, 355 (1992) 696-702.  Isackson, P.J., Huntsman, M., Murray, K.D. and Gall, C.M., BDNF mRNA expression is increased in adult rat forebrain after limbic seizures: temporal patterns of iaduction distinct from NGF, Neuron, 6 (1991) 937-948.  Kiessling, M. and Gass, P., Immediate early gone expression in expe-imental epilepsy, Brain Pathol., 3 ¢~093) 381-393.  Lauterbom, J.C., lsackson, P.J. and Gall., C.M., Nerve growth factor mRNA-containing cells are distributed within regions of cholinergic neurons in the rat basal forebrain, J. Comp. Neurol., 306 (1991) 439-446.  Lauterborn, J.C., Isackson, P.J. and Gall, C.M., Seizure-induced increases in NGF mRNA exhibit di~,erent time courses across forebrain regions and are biphasic in hippo~.ampus, Exp. NeuroL, 125 (1994) 22-40.  Lee, K.C., Scho:;ler, F., Oliver, M. and Lynch, G., Brief bursts of high-frequency stimulation produce two types of structural change in rat hippocampus, J. NeurophysioL, 44 (1980) 247-258.  Lieth, E., Cardasis, C.A. and Fallon, J.R., Muscle-derived agrin in cultured myotubes; Expression in the basal lamina and at induced acetylcholine receptor clusters, Det'. Biol., 149 (1992) 41-54.  Lieth, E. and Fallon, J.R., Muscle agrin: Neural regulation and localization at nerve-induced acetylcholine receptor clusters, J. Neurosci., 13 (1993) 2509-2514.  Ma, E., Morgan, R. and Godfrey, E.W., Distribation of agrin mRNAs in the chick embryo nervous system, J. Neurosci., 14 (1994) 2943-2952.  Ma, E., Morgan, R. and Godfrey, E.W., Agrin mRNA variants are differentially regulated in developing chick embryo spinal cord and sensory ganglia, J. Neurobiol., (1995) in press.  Mackler, S.A., Brooks, B.P. and Eberwine, J.H., Stimulus-induced coordinate changes in mRNA abundance in single postsynaptic hippocampal CA1 neurons, Neuron, 9 (1992) 539-548.
L. 1". O'Connor et al. / Molecular Brain Researe~ 33 (1995) 277-287
 Magill-Solc, C. and McMahan, U.3., Motor neurons contain agrinlike molecules, d. Cell Biol., t07 (1988) 1825-1833.  Merlin, L.R. and Wong, R.K.S,, Synaptic modifications accompanying epileptogenesis in vitro: long-term depression of GABA-mediated inhibition, Brain Re.s., 627 (1993) 330-340.  Monaghan, D.T. and Cotman, C.W., Dlstributior~ of ¥~rnethybDasparta~e-sensitive L-[3H]glutamate-binding sites in r~:t brain, .L Neuro.~,ci., 5 (19,~:5,~ 29!~,-29|9.  Morgan, 3.I. and Curran, T., Proto-onc,,~gene transcrip:5on fact~rs and epilepsy, Trends Pharmacol. Sci., 12 (1991) 343-349.  Nitkin, R.M. and Rothschild, T.V., Agrin-induced reorganization of extrace!lular matrix components on cultured m3otubes: relationship to AChR aggregation, d. Cell Biol., l 11 (1990) 1161- ! t 7(}.  O'Connor, L.T., Lauterborn, J.C., Gall, C.M. and Smith, M.A., Localization and alternative splicing of agrin mRNA in adult rat brain: transcripts encoding isoforms that aggregate acetylcholine receptors are not restricted to cholincrgic regions, J. Neurosci., 14 (1994) 1141-1152.  Pico, M.R. and Gall, C.M., Hippocampal epileptogencsis produced by electrolytic iron deposition in rat dentate gyrus, Epilepsy Res., 19 (1994) 27-36.  Pico, R.M and Gall, C., Continuities between outer nuclear membrane arid the rough endoplasmic reticulum increase in hippocampal neurons during seizure-induced protein synthesis, Brain Res., .~07 (1989) 387-392.  Racine, R,, Modulation of seizure activity by electrical stimulation. II. Motor seizure, Electroencepi~. Clin. NeurophysioL, 32 (1972) 281-294.  Represa, A. and Ben-Ari, Y., Long-term potentiation a~.i sTrouting of mossy fibers produced by brief episodes of hyperactivity. In C.E. Ribak, C.M. Gall and L. Mody (Eds.), The Dentate Gyrus and its Role in Seizures, Elsevier, Amsterdam, 1992, pp. 261-269.  Represa, A., Salle, G. and Ben-Ari, Y. Hippocampal plasticity in the kindling model of epilepsy in rats, Neurosci. Left., 99 (1989} 345-35O.  Rupp, F., Payan, D.G., Magill-Solc, C., Cowan. DM. and Scheller, R.H., Structure and expression of a rat agrin, Neuron, 6 (199!) 811-823.
 Sheng, M. and Greenberg, M.E., The: ~c~uladon and funcuon of c-los and o~her immediate carry genes in the nervous system, Neuron, 4 (1990) 477-485.  Simonato, M., Hosford, D.A., Labiner~ DM., Shin, C, Mansbach, H.H. and McNamara, J.O., Differential expression of immediatc early gents in the hippocampus in the kindling model of epilepsy, Mol. Brai.l Res., t I ( ]99l ) ] 15-124.  Smith, M.A. and O'Dowd, D.K., Cell-specific regulation of agrin RNA splicing in ~he chic~ ciUia,"y g,,ng[ion, ~~ur~;n, 12 (1994) 795 -804.  Sonv.enbcrg, J.L., Mitchelmore, C., Macgregor-Leon, P.F., Hempstead, J., Morgan, J.I. and Ct~rran, T., Glutamate recept,or agonists increase the e xpre~i,Jn of Fos, Fro, and AP-1 DNA binding activity in ~ammalian brain, J. Neurosci. Res., 24 (1989) 72-80.  Sugiyama, J., Bowen, DC. and Hal!, Z.W., Dystroglycan binds nerve and mu'~cle agrin, Neuron, 13 (1994) 103-115.  Sutula, T., Xiao-Xian, H., Cavazos, J. and Scott, G., Synoptic reorganization in the hippocampus induced by abnormal functional activity, Science, 239 (1988)1 ~47-1150. [5 o] Sutula, T.P., Golarai, G. and Cavazos, J., Assessing the functional significance of mossy fiber sprouting. In C.E. Ribak, C.M. Ga!! and L. Mody (Eds.), The Dentate Gyrus and its Role in Se:zures, Elsevier, Amsterdam, 1902, pp. 251-259,  Szekely, A.M., Costa, E. and Graysol, D,R., Transcriptional program coordination by N-me~.nyl-D-aspartate-se'asitive glutamate receptor stimulation in primary cultures of cerebellar neurons, Mol. Pharmacol., 38 (1989) 624-632,.  "Ihomas, W.S., O'Dow& D.K. and Smith, M.A., Developmental expression and alternative splicing of chick agrin RNA, Dec. Biol., 158 (1993) 523-535.  Wallace, B.G., Agrin-induced specializations contain cytoplasmic, membrane, and extracellular matrix-associated components of the postsynaptic apparatus. J. Neurosci., 9 (1989) 1294-1302. [(o3i Woriey, P.F., Bhat, R.V., Baraban, J.M., Erickson, C.A., McNaughton, B.L. and Barnes, C.A., Thresholds for synaptic activation of transcription factors i:~ hippocampus: correlation with long-term enhancement, J. Neurosci., 13 (1993) 4776-4786.