Proteolysis of spectrin by calpain accompanies theta-burst stimulation in cultured hippocampal slices

MOLECULAR BRAIN RESEARCH

'u ELSEVIER

Molecular Brain Research 32 (1995) 25-35

Research report

Proteolysis of spectrin by calpain accompanies theta-burst stimulation in cultured hippocampal slices Peter Vanderklish a,*, Takaomi C. Saido b, Christine Gall c, Amy Arai

a,

Gary Lynch

a

a Center for the Neurobiology of Learning and Memory, University of California at Irvine, lrvine, CA 92717, USA b Department of Molecular Biology, Tokyo Metropolitan Institute of Medical Science, 3-18-22 Honkomagome, Bunkyo-ku, Tokyo 113 c Department of Anatomy and Neurobiology, Unit,,ersity of California at Irvine, Irvine, CA 92717, USA Accepted 28 February 1995

Abstract

Tests were carried out to determine if repetitive bursts of afferent stimulation activate calpain, a calcium-dependent protease hypothesized to be involved in the production of long-term potentiation. Antibodies against a stable breakdown product that results from proteolysis of spectrin by calpain were used to identify sites of enzyme activation in cultured hippocampal slices. Slices in which theta-burst stimulation was applied to the Schaffer collateral fibers had pronounced accumulations of breakdown product that were restricted to field CA1, the zone innervated by the stimulated axons. Labelling occurred in the form of scattered puncta and was also present in dendritic processes. The extent of these effects was correlated (r = 0.73) with the amount of theta-burst stimulation delivered. Control slices or those receiving low frequency stimulation had variable, but uniformly lower, amounts of breakdown product and were clearly distinguishable from those given theta bursts. Statistical analyses using a six point rating scheme confirmed this point (P < 0.001). These results satisfy an essential prediction of the hypothesis that calpain plays an important role in the induction of long-term potentiation.

Keywords: Spectrin; Calpain; Long-term potentiation; Theta-burst stimulation; Synaptic cytoskeleton; Organotypic culture

I. Introduction

A diverse set of results indicates that the properties of AMPA-selective glutamate receptors respond to changes in the membrane/cytoskeletal environment [16,25,51]. Critical determinants of this environment are targets of Ca 2÷dependent enzymes presumably activated as part of the induction process for long-term potentiation (LTP). Given the evidence that LTP is expressed through modifications in the functional properties of A M P A receptors [19,27,49], the possibility exists that the 'potentiated state' of the receptor is imparted by its membrane/cytoskeletal context. This possibility is attractive from the perspective of LTP stability since receptor state, and thus potentiation, would be a secondary function of a more general structural arrangement - o n e that would likely persist for the same reasons that ultrastructural features remain constant in the

* Fax: (1) (714) 824-8481; E-maih [email protected] l)169-328X/95/$09.50 © 1995 Elsevier Science B.V. All rights reserved SSD1 01 6 9 - 3 2 8 X ( 9 5 ) 0 0 0 5 7 - 7

face of protein turnover. The enduring morphological correlates of LTP are in accord with the hypothesis that stable modifications to the membrane/cytoskeletal environment are associated with the potentiation effect. Installation of a new synaptic framework with LTP would necessitate destabilization of existing structures. Proteolysis of the synaptic cytoskeleton is one possible mechanism through which such changes could be achieved. Of particular relevance in this regard is the Ca2+-depen dent neutral cysteine protease, calpain [41]. An isozyme of calpain, calpain-1 or /z-calpain, is localized in dendrites [12,35] and is activated by Ca 2÷ at concentrations in the range of those achieved in dendritic processes with high frequency stimulation ( 1 - 3 /xM) [15,28]. Evidence for its participation in the induction of LTP is indirect, but compelling: (i) calpain-mediated digestion of synaptic membrane proteins mimics a correlate of LTP [24], (ii) stimulation of N M D A receptors in hippocampal slices leads to the production of calpain-specific spectrin breakdown products (BDPs) [6,44], and (iii) pharmacological inhibition of the enzyme blocks the development of LTP [7,8,32]. Impor-

26

P. Vanderklish et al. /Molecular Brain Research 32 (1995) 25-35

tantly, calpain has as a preferred substrate brain spectrin, a multifunctional, actin-binding structural protein that lines the inner surface of the plasma membrane and is a prominant postsynaptic density component [5,54]. In many cell types, spectrin contributes to the definition of cell shape [3,13,31,48], receptor distribution and mobility [9,21,30,36,38], and membrane lipid organization [20,22]; moreover, limited digestion of spectrin and other components of the submembranous cytoskeleton by calpain has been shown to be a possible route through which external stimuli can affect some of these features [2,10,11,53]. If, as suggested above, LTP is a receptor state imparted by an altered membrane/cytoskeletal environment, then the calpain-spectrin interaction would be an attractive candidate for a mechanism permitting changes to this environment. Lacking, however, is evidence that calpain becomes activated in response to physiologically relevant activity patterns used to induce LTP and that an ensuant cleavage of spectrin occurs in dendritic processes. These points were addressed in the present study using a BDP-specific antibody [41,42] and a cultured slice preparation largely devoid of the background proteolysis which normally accompanies the preparation of hippocampal slices.

2. Materials and methods 2.1. Hippocampal slice cultures

Organotypic cultures of hippocampus were prepared under sterile conditions from Sprague-Dawley rats of 10-12 days postnatal age. Rats were anesthetized with Metofane, decapitated and the brains rapidly removed and submersed in ice-cold artificial cerebrospinal fluid (ACSF; see below). Following dissection (and removal of ~ 1/8th the length of hippocampus on each end), transverse sections (400 /xm) were cut on a Brinkmann tissue chopper, collected in ACSF containing 4 mM MgC12 and 100 /xM adenosine, then randomly distributed among 6 MillicellCM inserts (30 mm; Millipore) in a 6-well culture plate; each plate contained explants from one animal only, with an average of 4 slices per well. Cultures were maintained for the entire culturing period at 37°C with 5% CO 2 in a humidified FORMA incubator. The media was exchanged the first day after explantation and then every other day with 1 ml per well: MEM w / H a n k ' s salts (Sigma), 5 mM NaHCO3, 25 mM HEPES, 25 mM dextrose, 3 mM glutamine, 2.5 mM MgC12, 2.0 mM CaCI2, 0.5 mM ascorbate, 1 mg/1 insulin, and 20% horse serum (Gemini Biotech) pH = 7.2 @ 37°C. Prior to any manipulations, cultures were allowed to 'mature' for a minimum of 10 clays as several physiological and biochemical markers of viability are stably expressed around this time point, these include: (i) the attainment of a high synaptic density [26], characterized by complex spine profiles [4] and the ability

to support large evoked field EPSPs and LTP [26,50,52]; (ii) the stable expression of pre-and post-synaptic terminal components such as synaptophysin and AMPA receptor subunits [18] (the latter exhibiting immunochemical and pharmacological properties indistinguishable from that observed in adult tissue); and, (iii) the conversion of NCAM isoforms from a developmental to an adult glycosylation pattern (Hoffman et al., unpublished observations). A total of 12 animals (not including those utilized in preliminary studies) were used in generating the data presented below. 2.2. Electrophysiology

For electrophysiological recordings and patterned stimulation, cultures were transferred to an interface recording chamber designed to accommodate the Millicell inserts on which the explants were grown. Within the recording chamber, slices were superfused with humidified carbogen (95% O2/5%CO2; ~ 1 l/min) and subfused with ACSF (20 ml/h): 124 mM NaC1, 3 mM KC1, 1.25 mM KH2PO4, 2.8 mM MgC12, 2 mM CaC12, 25 mM NaHCO3, 20 mM D-glucose, 5 mM HEPES, 0.5 mM ascorbate. The temperature was maintained at 35°C. In an effort to compensate for potential well-to-well variations in slice quality, all experimental manipulations were carried out on cultures of a common insert when possible. Thus, for the majority of cases (77%), theta-burst stimulation (TBS; bursts of 4 pulses at 100 Hz with 200 ms between bursts) and low frequency stimulation (LFS; 0.1 Hz) were administered to different cultures explanted to the same insert, side-by-side with 'untouched' controls. LFS and varying degrees of TBS were administered with twisted nichrome bipolar stimulation electrodes (50 /zm) positioned in the stratum radiatum of CA1. The stimulation duration was 0.1 ms and the intensity in each case was adjusted such that field EPSPs of half maximal amplitude were generated with test pulses; field EPSPs and the effects of TBS were monitored with pulled-glass micro-recording electrodes ( ~ 5 M O; filled with 2 M NaCI) positioned on the surface of the slice in the stratum radiatum. LFS controls were slices in which test pulses were collected for a 30 min period without any high frequency stimulation. In some instances, the stimulation electrodes were placed and left for an equivalent period without stimulation to assess the effects of electrode impact on spectrin breakdown. Theta-bursts were given in pairs separated by 10-15 s to maintain the maximal degree of facilitation across bursts and to avoid triggering epileptiform activity (seizure-like activity typically occurs only with long, continuous trains and was not a problem in the experiments presented here). A train usually involved 6 - 7 such pairs of bursts with the stimulus duration increased to between 0.12 ms and 0.14 ms; saturative LTP is typically obtained with 1-2 trains in cultured slices. To assess the relation between the amount of TBS and the degree of spectrin breakdown in situ, multiple trains were given at

P. Vanderklish et al. / Molecular Brain Research 32 (1995) 25-35

10 m i n intervals and the c u m u l a t i v e n u m b e r of bursts for a particular slice s u m m e d . Selection criteria were i m p o s e d based on the physiological behavior of the cultures. Specifically, slices disp l a y i n g s p o n t a n e o u s activity or not exhibiting m a x i m a l field E P S P amplitudes of at least - 2 m V were not in-

27

cluded. A s an additional precaution, slices not demonstrating paired-pulse facilitation (PPF) were omitted from the study since a deficit in transmitter mobilization w o u l d severely limit the postsynaptic effects of TBS. Slices failing to meet these criteria constituted a relatively small fraction of those prepared.

Fig. 1. Anti-spectrin BDP antibodies preabsorbed with the peptide immunogen do not label spectrin breakdown products in organotypic cultures. Photomicrographs show two sections from a cultured slice challenged with 200 /zM NMDA for 20 min. A: tissue processed with the primary antibody preincubated with the peptide immunogen (72 h @ 4°C) lacked staining of spectrin BDPs. B: a section processed with the same antibody dilution, without peptide pretreatment, shows dense labelling of somata and dendritic processes in CA1 (upper and lower arrows, respectively), confirming a robust induction of spectrLn BDPs with NMDA. Similar controls using this antibody on Western blots are published elsewhere [42]. Calibration bar = 250 g,m for panel A, 150 /xm for panel B.

P. Vanderklish et al. / Molecular Brain Research 32 (1995) 25-35

28

ro

.

.

.

.

•. . . . .

~

~F~.~7~

~

ii !~

P. Vanderklishet al. / Molecular Brain Research 32 (1995) 25-35 2.3. Polyclonal antibodies to spectrin breakdown products Polyclonal antibodies specific for the 150 kDa calpain cleavage product of human spectrin were developed using as an immunogen a 5 mer peptide (GMMPR) mimicking the 5 N-terminal amino acids of the calpain-proteolyzed a-subunit. Antibodies raised to this sequence recognize the 150 kDa fragment produced by calpain in Western blot and immunocytochemical assays [39,42], but do not react with intact a-spectrin or fragments produced by a variety of other mammalian proteases [39]. Specificity toward the calpain derived spectrin epitope is expected given the unique cleavage site specificity of calpain [17] and the use of a short peptide sequence as the immunogen. Procedures for peptide synthesis and conjugation, antibody development and purification are described elsewhere [40].

2.4. Immunohistochemistry Immediately following electrophysiological manipulations, all the cultures of a common well were processed in parallel by first washing the insert in an ice-cold buffer composed of 0.32 M sucrose, 20 m M Tris, 1 m M EDTA, 1 mM EGTA, 200 /xM leupeptin pH = 7.4. This buffer was aspirated and replaced with a 2.0% paraformaldehyde/0.1 M phosphate buffer (PB; pH = 7.2) fixation solution; fixation was allowed to continue overnight (typically 14 h) at 4°C. At least 1 h prior to sectioning, the paraformaldehyde was replaced with 20% s u c r o s e / P B . Twenty micron sections were then cut parallel to the broad base of the explant using a freezing microtome, collected into PB, mounted onto Vectabond-treated slides, and then allowed to air dry. Following rehydration, the tissue was rinsed through three 10 min washes in PB, incubated with 5% normal goat s e r u m / P B for 30 min at room temperature (RT), then rinsed for 5 min in PB. The anti-spectrin BDP primary antibody was diluted to a concentration of 1 ~ g / m l in PB

29

and applied to the surface of the sections for an overnight incubation at RT. The sections were then rinsed through three 10 rain washes in PB, incubated with biotinylated anti-rabbit IgG (1:200 in PB) for 2 h at RT and then rinsed in PB (3 washes; 10 min each). The distribution of biotin was localized using the HRP-Avidin-Biotin system of Vector Laboratories with diaminobenzidine as a chromogen. In some cases, the appearance of diaminobenzidine reaction product was intensified by the addition of cobalt chloride and nickel a m m o n i u m sulfate to the reaction solution. The slides were then coverslipped in Permount and analyzed for the distribution of immunostaining for spectrin breakdown products. To verify immunostaining specificity, alternate sections from some stimulated explants were processed through all steps as described above except that the primary antisera was either omitted from the overnight incubation or preincubated with the immunogen (72 h @ 4°C). In these instances, immunostaining was robust in tissue incubated with the antisera and entirely absent in tissue processed with the preabsorbed antisera (Fig. 1) or without anti-spectrin BDP exposure.

2.5. Scoring of spectrin breakdown product staining Sections from cultures meeting the above physiological criteria were compared for differences in spectrin BDP staining. A six point scoring system was used to reflect the nature and extent of spectrin breakdown; representative sections for scores 1, 3 and 5 are presented in Fig. 2. Staining was rated as follows: 0 = no staining; 1 = low numbers of stained somata or segments of dendritic branches (usually 2 - 3 ) ; 2 = elaborated dendritic trees, but few and discontinuous across CA1; 3 = elaborated dendritic trees and puncta within a continuous region of CA1; 4 = the same as 3, but extending over half the CA1 field; 5 = continuous dendritic staining with puncta across the entire CA1 field.

Fig. 2. Photomicrographsof sections representativeof three of the scoring catagories used. A: CA1 region of a section scored 1; this score indicateslow numbers of somata or segments of dendritic branches. B: a view of a whole section scored 3 which also documents the subfield specificity of the TBS-elicited spectrin proteolysis; the CA1 field is delimitedby the convergingarrows and the granule cell layer of the dentate gyrus is highlighted with black dots. Slices scored 3 are characterized by elaborated dendritictrees and puncta within a continuousregion of CA1. C: same section as presented in panel B showingdetails of CA1 staining and the orientationof the stratum radiatum(SR) and stratum oriens (SO). D: a section scored 5, presented in the same orientationas in C, exhibits continuousdendritic staining with puncta across the entire CA1 field. Note the extensivecell body staining which was present in many of the slices exhibiting this level of dendritic BDP induction. Sections scored 0 had no BDP staining, while sections scored 2 and 4 exhibited gradations in the continuityand extent of staining, respectively(see Materials and methods). Calibrationbar = 150 /zm for panels A, C and D; 400 /xm for panel B. Fig. 4. SpectrinBDP stainingin cultured slices varies with slice quality and electrophysiologicalmanipulation.A: an isolated focus of BDP stainingcan be seen in a control culture damaged by electrode impact. B: a section scored 2 demonstrates that a modest amount of BDP inductioncan be elicited with LFS. This particularsection represents the upper limit of BDP stainingobserved in LFS controls,which is characterized by sporadic staining of somata in the stratum pyramidal(SP) and fragments of dendriticarbors. C: a section scored as 4, illustratingthe effects of TBS on spectrin BDP immunoreactivityin cultured slices; as is typical of this treatment, stainingis continuousover all or part of the apical and basal dendriticregions of CA1. Spine-likepuncta can be seen juxtaposed with varicose appearingdendrites within the limits of the stained dendriticfields. D: a closer view suggests that the punctate nature of the BDP staining could reflect the depositionof reaction product in spine-like structures (indicated by arrows). Calibrationbar = 63 /xm for panel A, 70 p.m for panel B, 87 ~m for panel C, and 40 /zm for panel D.

30

P. Vanderklish et al. / Molecular Brain Research 32 (1995) 25-35

A

PPF

TBS

B 200-

8

t~ 150-

ra

g W ¢'1

q ff~ lO0UJ

ITBS 50

i

-10

-5

i

0

10 TIME (rain.)

15

20

25

Fig. 3. The electrophysiological properties of CA3-CA1 synapses in slice culture resemble those seen in acute and in vivo preparations. A: field EPSPs recorded in the stratum radiatum of CA1 are augmented via paired-pulse facilitation of release (i.p.i. = 75 ms). Facilitation is also seen across bursts (35 ms × 100 Hz) paired at the theta frequency (5 Hz). B: synaptic responses exhibit long-term potentiation in response to trains of such burst pairs (6-7 pairs separated by 10-15 s). These phenomena are useful as indices of cultured slice viability as their induction is dependent on several other factors, such as transmitter mobilization, Ca 2+ flux and a normal membrane potential. (Calibration: 10 ms by 0.5

Immunohistochemical analysis of control cultures yielded a staining pattern qualitatively distinct from that observed as a result of TBS. When cultures were transferred to the recording chamber and left untouched (n = 26), there was little or no baseline signal with the spectrin BDP-specific antibody; when it occurred, staining was mostly somatic and present in only a few cells. Moreover, this staining did not appear to be specific to any particular subfield. Similarly, if stimulation electrodes were placed on the slice, but pulses not delivered (n = 4), very few sections contained immunostaining. When present, staining in these cultures displayed a characteristic pattern. As shown in Fig. 4A, immunoreactive loci with a short radius of dendritic staining could be seen in the stratum radiatum surrounding the point of electrode impact. Staining of this type was not common and could be avoided by careful placement of electrodes. Overall, then, background staining for spectrin breakdown products was low in slices that had been maintained in culture for at least 10 days, even when subjected to the 'mechanics' of the recording process. As noted, slices exhibiting spontaneous activity were not used in the study. Sections through such slices often had extensive background staining distributed among all cell subfields of hippocampus (not shown). A similar pattern of BDP immunoreactivity was observed in slices maintained for less than one week in culture (though these were not included in the study); slices examined 5 - 6 days after explantation had widespread BDP immunoreactivity in dendrites and cell bodies (not shown). Low frequency stimulation sometimes resulted in increased spectrin breakdown. As with the unstimulated

mV.)

3. Results

The electrophysiological characteristics of hippocampal slice cultures resemble those of acute slices and in vivo preparations of hippocampus, with a few notable exceptions. Organotypic cultures tend to be more excitable, with steeper current-voltage relations, and to exhibit forms of synaptic plasticity (short-term potentiation, PPF, LTP) that, while qualitatively similar, are more variable than in preparations derived from adult tissue. As this intrinsic variability could impact on the results obtained, an attempt was made to reduce these and other problems associated with the culture paradigm by using slices within the period of optimal physiology (10-18 days in culture) and by screening explants within each well for spontaneous activity and the presence of PPF and LTP. Cultures exhibiting spontaneous activity or lacking PPF were omitted from the study; LTP was obtained with a frequency similar to that previously reported from this and other laboratories [26,52]. Typical effects elicited by paired-stimulation and TBS are presented in Fig. 3.

o .=_

/-"" //,

~=_(D

.. /,,"

0..

c~ Q3

<

• //z /. /

1~0

r = 073

2~0 3~0 410 510 60 7~0 8~0 Cumulative Number of Bursts Given

9J0

100

Fig. 5. Graph showing that scores of the extent of spectrin BDP formation in CA1 correlate with the number of theta-bursts delivered. Organotypic slices were transferred to an interface recording chamber where they were maintained in a humidified atmosphere (95% O ~ / 5 % CO z) with constant subfusion with ACSF. Theta-burst stimulation was delivered to axons traversing the stratum radiatum of CA1; slices received variable numbers of bursts according to the paradigm described in Materials and methods. Cultures meeting physiologic criteria were sectioned, processed with the BDP-specific antibody and scored for the extent of spectrin BDP staining in CA1. Note: some data points are overlapping. (r = 0.73; P = 0.001, n = 26)

P. Vanderklish et al. / Molecular Brain Research 32 (1995) 25-35

control slices, the pattern of staining in low frequency controls was not dense, involving scattered cell bodies and, occassionally, their processes (compare Figs. 2A and 4B). When scored as described in Materials and Methods, the mean rating for the low frequency group was 0.93 ___0.28 (mean, S.E.M., n = 14). This is significantly different from the mean score determined for the untouched group (0.128 ___0.53, n = 26; P = 0.034 Mann-Whitney test). The most extensive spectrin BDP staining, both at the cellular level and in terms of regional extent, was induced with TBS. Immunoreactivity resulting from TBS was, for the most part, restricted to CA1 and had defining characteristics. Staining was predominantly dendritic and usually involved more of the dendritic tree than observed for either of the control groups. Often, both the apical and basal dendrites were labelled over a continuous region of CA1 (Fig. 4C). In many cases, TBS also elicited spectrin breakdown in the distal aspects of dendrites in the upper blade of the dentate gyrus, just opposite the immunopositive CA1 region; labelled dendrites in the dentate gyrus were relatively sparse and restricted to a narrow band near the fissure. Individual dendritic processes in CA1 had a varicose appearance, resembling qualitatively the dendritic BDP immunoreactivity seen with a brief application of N M D A (Vanderklish, unpublished observations and Fig. 1B). BDP immunoreactivity was also manifest as a diffuse array of puncta distributed within the limits of stained dendritic arbors (Fig. 4D). Increasing the number of bursts by delivering more than one train resulted in more extensive spectrin breakdown in CA1; as can be seen in Fig. 5, these two variables (the total number of bursts delivered vs. BDP immunoreactivity score) were correlated (r = 0.73; P = 0.001). Taken as a group, the average score for cultures receiving TBS was 2.64 + 0.23 (n = 26); this is significantly different than the average score for cultures given low frequency stimulation ( P = 0 . 0 0 0 5 , M a n n Whitney test). Within-well comparisons were also made in the following way. First, the slices receiving TBS were subdivided into three groups based on the total number of bursts delivered: low, moderate, and high. Then, the immunoreactivity score for each slice was paired with the value of the highest scoring LFS control slice of the same well. All three theta burst groups had significantly higher

31

levels of BDP immunoreactivity than did their matched controls (see Table 1).

4. Discussion The results described here establish that the proteolysis of spectrin by calpain occurs in response to theta-burst stimulation. That calpain activation is triggered by a physiologically relevant stimulation pattern increases the likelihood of its involvement in LTP induction. While this has been inferred with the use of calpain inhibitors, the present study circumvents interpretive problems associated with inhibitor specificity and locus of action by assaying for a specific biochemical product of calpain activity. Inherent in the paradigm used, the results obtained, and the method of data analysis are the following interpretive problems: (i) the equivalence of processes engaged by bursts of afferent activity in cultured slices to those generated in acute slices or in vivo; (ii) the nature of dendritic staining; and, (iii) the validity of quantitative comparisons with the scoring system used. There are cleat' advantages in using cultured slices rather than acute or in vivo preparations for studying calpain activation by physiological stimulation. Established slice cultures lack the background degeneration found in conventional acute slices (i.e. that produced with the cutting process) that could easily obscure stimulus-induced BDP formation. The 3, also minimize the problems of electrode-induced damage, locating target synaptic populations, and rapid removal of tissue that complicate in vivo studies. A potential disadvantage exists regarding the relative maturity of cultured slices: explantation during the second postnatal week might preserve some mechanisms associated only with development, or hinder the proper expression of adult-like properties by removing the still immature hippocampus from the influence of afferent systems, vasculature, etc. Thus, an issue critical to the results presented here is whether processes set in motion by TBS in culture are equivalent to those resulting from TBS in adult preparations, where the majority of the physiological effects of TBS have been characterized. The presence of N M D A receptors, fast and slow IPSPs and a mature synaptic architecture (i.e. complex

Table 1 Paired comparisons between slices receiving TBS and the highest scoring LFS control slices of the same well for three burst total ranges with the value of the highest scoring LFS control slice of the same well Theta-burst bin Mean + S.E.M. (n) Mean _+S.E.M. (n) Two-tailed (number of bursts) (for TBS-treated) (for matched controls) P valuc 10-29 1.88 + 0.27 (11) 0.70 _+0.24 (11) 0.006 30-49 2.58 _+0.43 (6) 1.37 + 0.44 (6) 0.024 504.67 _+0.33 (3) 2.00 + 0.00 (3) 0.015 Within-well comparisons were made by subdividing slices receiving TBS into three groups based on the total number of bursts dclivered: low, moderate and high. The immunoreactivityscore for each slice was then paired.

32

P. Vanderklish et al. /Molecular Brain Research 32 (1995) 25-35

spines and high synaptic density [4]) would suggest that TBS operates upon the same substrate in the two contexts. Indeed, under optimal conditions the synaptic responses to burst stimulation appear to be equivalent in acute slices of adult hippocampus and cultured slices (unpublished observations). However, no systematic comparison of the effects of TBS or other stimulation paradigms on cultured versus acute slices has been presented. Another way to address this issue is to ask if the potentiation resulting from TBS in culture is comparable to that obtained in adult preparations. Three observations suggest that this is the case. First, potentiation in culture displays synapse specificity and a magnitude and temporal development similar to that seen with LTP in adult tissue [50,52]. Second, the potentiation obtained is reduced with AP5. Lastly, there is evidence that TBS-elicited potentiation is expressed by A M P A receptors in both cultured slices and acute slices [19,27,49,52]. In consideration of the above points, it appears likely that the observations made in the cultured slice paradigm can be extended to preparations of adult hippocampus. Spectrin BDP immunoreactivity elicited by TBS was predominantly dendritic and confined to the CA1 region as expected for stimulation of the Schaffer collaterals. However, certain aspects of the staining do not correspond to what one would expect from this manipulation. The synapse specificity property of LTP suggests that Ca 2÷ influxes high enough to activate calpain would be confined to dendritic spines in regions innervated by the stimulated afferents. In contrast, the present results indicate that calpain was activated, and spectrin degraded, through a large portion of the dendritic tree in both the apical and basal dendritic fields of CA1. This seemingly anomolous observation could be due to the presence of recurrent projections within CA1, resulting from a partial loss of their normal targets with explantation, which might allow Schaffer collateral stimulation to affect basal dendrites and depolarize the cell further. It is also possible that withinburst facilitation during theta pattern stimulation produced depolarization sufficient to trigger voltage-gated Ca 2÷ channels along the dendritic tree, leading to calpain activation at nonsynaptic sites. Imaging studies of conventional slices show that Ca 2+ increases can propagate throughout large portions of the dendritic tree during repetitive stimulation [1,29]. It should be noted that this latter possibility is consistent with the observation of a much steeper I - V relation at Schaffer collateral-CA1 synapses in cultured as compared to acute slices; this might also account for the greater variability of LTP magnitude attained in cultured slices, which, along with the use of a stimulation range that goes beyond that necessary to induce saturative LTP, precluded a correlation between LTP magnitude and spectrin breakdown in this study. Another feature of the spectrin BDP immunoreactivity unexpected from the paradigm used is the varicose appearance of dendritic processes. Varicose processes are usually considered to be a sign of

neuropathology and thus their occurrance in the present studies could indicate that theta-burst stimulation initiated pathogenesis. However, while it is possible that TBS had deleterious effects on dendrites, a more functional interpretation may be applicable. Calpain activation has been linked to changes in the plasma membranes and morphology of platelets and it is not unreasonable to speculate that cleavage of the spectrin network in dendritic processes might lead to similar membrane irregularities. Such a mechanism could have significance with regard to activity-dependent 'sculpting' of established dendritic processes [33,37]. Other factors not related to plasticity or pathology might also contribute to the varicose staining pattern, including: (i) irregularities in spectrin distribution within dendrites, and (ii) a non-homogeneous distribution of synaptic contacts along the dendritic tree. Finally, it is possible that dendritic varicosities are a normal feature of cultured slices, but emphasized in the present experiments because of high concentrations of activity-induced spectrin BDPs. In regard to this point, it should be noted that sporadic staining of dendritic branches in control cultures was not of a varicose nature. The scoring system employed did not rely on densitometric or image analysis techniques. These could, in principle at least, have provided a quantitative index of the extent of spectrin breakdown, but the lack of a linear standard, the nonhomogeneous and qualitatively variable nature of dendritic staining, and potential variability in the density of synapses across cultures made it difficult to apply these techniques in an unbiased manner. The system used here does not reflect the occurrence of a particular unit of immunoreactivity and this must be kept in mind in considerations of the statistical comparisons made. Nevertheless, the conclusion of a significant change with TBS is likely valid given that, in a sample size of 40 meeting the physiological criteria, none of the untouched or low frequency controls demonstrated continuous immunoreactivity profiles extending across a major portion of the CA1 subfield (i.e. none were ranked higher than 2; see Fig. 4B). Moreover, the induction of BDP formation with low numbers of theta bursts and the well correlated increases found with larger numbers of bursts together suggest that spectrin proteolysis is a process engaged in a reliable and graded fashion with LTP induction. These data might be open to a less specific interpretation in light of the correlation evident in Table 1 between the average scores for matched LFS control slices and the number of bursts delivered in the experimental group. In that LFS was applied uniformly to controls in all TBS catagories and untouched slices were not changed, this should not represent a difference in experimental manipulation between wells or an influence of burst trains on adjacent slices. A more tenable explanation considers that the increases in matched LFS control scores reflect a greater sensitivity of slices within a particular well to afferent stimulation, but this must be reconciled with a random distribution of slices during prepara-

P. Vanderklish et al. / Molecular Brain Research 32 (1995) 25-35

tion, identical culturing conditions for wells of a common plate, and low scores given to adjacent slices receiving TBS. It is also possible that the relationship between the score of LFS controls and burst number is coincidental (two of the slices in the 50 + burst group were paired with the same control, and a control scored 2 was distributed within the 10-29 burst group). In any event, qualitative differences between TBS-treated slices and their matched LFS controls were always found, and there were robust differences in the average scores irrespective of the burst range considered. Previous studies have established that brain spectrin is a preferred substrate of calpain [46], particularly in the presence of calmodulin [43], and that both are present in the dendrites and spines of hippocampal pyramidal neurons. Other work has shown that repetitive synaptic stimulation results in an elevation of Ca 2+ in spines sufficient to activate calpain. The experiments reported here follow from these results and confirm that physiologically relevant afferent activity triggers the formation of a breakdown product known to result from the proteolysis of spectrin by calpain. The consequences of calpain-mediated spectrin degradation are not addressed by the present study, but are likely to be numerous [41] and enduring. Based on the intermolecular associations of spectrin (i.e. actin filaments, membrane lipids and underlying cytoskeletal components, and transmembrane adhesion receptors) it can be assumed that its hydrolysis is part of a mechanism of cytoskeletal and morphological reorganization. Indeed, proteolysis and redistribution of spectrin have been shown to accompany rapid morphological/functional transformations in platelets [11] and T-lymphocytes [14]. In the present context, it is attractive to speculate that spectrin proteolysis leads to a 'relaxation' of determinants governing synaptic structure, thereby permitting the installation of a new synaptic framework with qualitatively different influences on AMPA receptors embedded in it. Exactly how an altered membrane/cytoskeletal environment could impact receptor function is not known, but a potential mechanism is offered in a recent report from Paoletti and Ascher [34]. They observed that NMDA currents are modulated by hydrostatic pressure and concluded that this ' mechanosensitivity' of the NMDA receptor could be a relevant factor in determining its function in spines and other sites undergoing changes in membrane tension. If adult hippocampal AMPA receptors are also mechanosensitive, then cytoskeletal proteolysis elicited by LTP-inducing stimulation could ultimately exert potentiation through changes in membrane tension. LTP would then be a secondary function of morphology and persist as long as the particular synaptic configuration imparting it. The postulated relationship between synaptic structure and LTP could also be formulated in biochemical terms, where the net effect of an altered membrane/cytoskeletal environment is to change the rate or quality of interactions between AMPA receptors and modulatory factors (such as

33

kinases and anchoring proteins). Alternatively, calpain activation brought on by intense synaptic activity could be disruptive in nature. However, while spectrin proteolysis is a very early event in many forms of pathology [23,45,47], the stimulation conditions used here do not elicit pathophysiological effects in acute slices, the hippocampus in vivo, or cultured slices. In all, then, the results support the conclusion that activation of calpain during LTP induction leads to non-degenerative changes in the membrane/cytoskeletal environment.

Acknowledgements We would like to thank Zhiquin Sun for excellent technical assistance, Eric Bednarski for help with blind data analysis, and Dr. Markus Kessler for helpful discussions. This work was supported by AFOSR Grant 92-J0307 and NIA Grant AG00538 to G.L. and C.G.

References [1] Alford, S., Frenguelli, B.G., Schofield, J.G. and Collingridge, G.L., Characterization of Ca 2+ signals induced in hippocampal neurones by the synaptic activation of NMDA receptors, J. Physiol., 469 (1993) 693-716. [2] Baldassare, J.J., Bakshian, S., Knipp, M.A. and Fisher, G.J., Inhibition of fibrinogen receptor expression and serotonin release by leupeptin and antipain, at. Biol. Chem., 260 (1985) 10531-10535. [3] Bennett, V. and Lambert, S., The spectrin skeleton: from red cells to brain, J. Clin. Invest., 87 (1991) 1483-1489. [4] Buchs, P.-A., Stoppini, L. and Muller, D., Structural modifications associated with synaptic development in area CA1 of rat hippocampal organotypic cultures, DeL,. Brain Res., 71 (1993)81-91. [5] Carlin, R.K., Bartelt, D.J. and Siekevitz, P., Identification of fodrin as a major calmodulin-binding protein in postsynaptic density preparations, J. Cell Biol., 96 (1983) 443-448. [6] del Cerro, S., Arai, A., Kessler, M., Bahr, B.A., Vanderklish, P., Rivera, S. and Lynch, G., Stimulation of NMDA receptors activates calpain in cultured hippocampal slices, Neurosci. Lett., 167 (1994) 149-152. [7] del Cerro, S., Larson, J., Oliver, M.W. and Lynch, G., Development of hippocampal long-term potentiation is reduced by recently introduced calpain inhibitors, Brain Res., 530 (1990) 91--95. [8] Denny, J.B., Polan-Curtain, J., Ghuman, A., Wayner, M.J. and Armstrong, D.L., Calpain inhibitors block long-term potentiation, Brain Res., 534 (1990) 317-320. [9] Dmytrenko, G.M. and Bloch, R.J., Evidence for transmembrane anchoring of extracellular matrix at acetylcholine receptor clusters, Exp. Cell Res., 206 (1993) 323-334. [10] Fox, J.E., Austin, C.D., Boyles, J.K. and Steffen, P.K., Role of the membrane skeleton in preventing the shedding of procoagulant rich microvesicles from the platelet plasma membrane, J. Cell Biol., 111 (1990) 483-493. [11] Fox, J.E., Reynolds, C.C., Morrow, J.S. and Phillips, D.R., Spectrin is associated with membrane-bound actin filaments in platelets and is hydrolyzed by the Ca2+-dependent protease during platelet activation, Blood, 69 (1987) 537-545. [12] Fukuda, T., Adachi, E., Kawashim/~, S., Yoshiya, 1. and Hashimoto, P.H., lmmunohistochemical distribution of calcium-activated neutral

34

[13] [14]

[15]

[16]

[17]

[18]

[19]

[20]

[21]

[22]

[23]

[24] [25]

[26]

[27]

[28]

[29]

[30]

[31]

P. Vanderklish et al. / Molecular Brain Research 32 (1995) 25-35

proteinases and endogenous CANP inhibitor in the rat hippocampus, J. Comp. Neurol., 302 (1990) 100-109. Goodman, S.R. and Zagon, I.S., The neural cell spectrin skeleton: a review, Am. J. Physiol., 250 (1986) C347-C360. Gregorio, C.C., Black, J.D. and Repasky, E.A., Dynamic aspects of cytoskeletal protein distribution in T lymphocytes: involvement of calcium in spectrin reorganization, Blood Cells, 19 (1993) 361-371. Guthrie, P.B., Segal, M. and Kater, S.B., Independent regulation of calcium revealed by imaging dendritic spines, Nature, 354 (1991) 76-79. Hall, R.A., Kessler, M. and Lynch, G., Evidence that high-and low-affinity DL-alpha-amino-3-hydroxy-5-methylisoxazole-4-propionic acid (AMPA) binding sites reflect membrane-dependent states of a single receptor, J. Neurochem., 59 (1992) 1997-2004. Harris, A.S., Croall, D.E. and Morrow, J.S., The calmodulin-binding site in a-fodrin is near the calcium-dependent protease-I cleavage site, J. Biol. Chem., 263 (1988) 15754-15761. Hoffman, K., Bahr, B.A., River, S., Vanderklish, P., Hall, R.A., Kessler, M., Arai, A., Gall, C. and Lynch, G., Stable maintenance of glutamate receptors and other synaptic components in organotypic hippocampal slice cultures, Soc. Neurosci. Abstr., 26 (1993) 1357. Kauer, J.A., Malenka, R.C. and Nicoll, R.A., A persistant postsynaptic modification mediates long-term potentiation in the hippocampus, Neuron, 1 (1988)911-917. Kumar, A., Gudi, S.R., Gokhale, S.M., Bhakuni, V. and Gupta, C.M., Heat-induced alterations in monkey erythrocyte phospholipid organization and skeletal protein structure and interactions, Biochim. Biophys. Acta, 1030 (1990) 269-278. Kwiatkowska, K., Khrebtukova, I.A., Gudkova, D.A., Pinaev, G.P. and Sobota, A., Actin-binding proteins involved in the capping of epidermal growth factor receptors in A-431 cells, Exp. Cell Res., 196 (1991) 255-263. Langner, M., Repasky, E.A. and Hui, S.W., Relationship between membrane lipid mobility and spectrin distribution in lymphocytes, Febs Lett., 305 (1992) 197-202. Lee, K.S., Frank, S., Vanderklish, P., Arai, A. and Lynch, G., Inhibition of proteolysis protects hippocampal neurons from ischemia, Proc. Natl. Acad. Sci. USA, 88 (1991) 7233-7237. Lynch, G. and Baudry, M., The biochemistry of memory: A new and specific hypothesis, Science, 224 (1984) 1057-1063. Massicotte, G., Vanderklish, P., Lynch, G. and Baudry, M., Modulation of DL-alpha-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid/quisqualate receptors by phospholipase A2: a necessary step in long-term potentiation?, Proc. Natl. Acad. Sci. USA, 88 (1991) 1893-1897. Muller, D., Buchs, P.-A. and Stoppini, L., Time course of synaptic development in hippocampal organotypic cultures, Det,. Brain Res., 71 (1993) 93-100. Muller, D., Joly, M. and Lynch, G., Contributions of quisqualate and NMDA receptors to the induction and expression of LTP, Science, 242 (1988) 1694-1697. Muller, W.A. and Connor, J.A., Dendritic spines as individual neuronal compartments for synaptic Ca 2+ responses, Nature, 354 (1991) 73-76. Muller, W.A. and Connor, J.A., Ca 2÷ signalling in postsynaptic dendrites and spines of mammalian neurons in brain slice, J. de Physiologie, 86 (1992) 57-66. Nelson, J.W., Colaco, C.A.L.S. and Lazarides, E., Involvement of spectrin in cell-surface receptor capping in lymphocytes, Proc. Natl. Acad. Sci. USA, 80 (1983) 1626-1630. Nelson, W.J. and Lazarides, E., Assembly and establishment of membrane-cytoskeletal domains during differentiation: spectrin as a model system. In E. Elson, W. Frazer and L. Glaser (Eds.), Cell membranes: Methods and Reviews, Vol. 2, Plenum Publishing Corp., New York, 1984, pp. 219-246.

[32] Oliver, M.W., Baudry, M. and Lynch, G., The protease inhibitor leupeptin interferes with the development of LTP in hippocampal slices, Brain Res., 505 (1989) 233-238. [33] Oshima, M., Koizumi, S., Fujita, K. and Guroff, G., Nerve growth factor-induced decrease in the calpain activity of PC12 cells, J. Biol. Chem., 264 (1989) 20811-20816. [34] Paoletti, P. and Ascher, P., Mechanosensitivity of NMDA receptors in cultured mouse central neurons, Neuron, 13 (1994) 645-655. [35] Perlmutter, L.S., Gall, C., Baudry, M. and Lynch, G., Distribution of calcium-activated protease calpain in rat brain, J. Comp. Neurol., 296 (1990) 269-276. [36] Pestonjamasp, K.N., Gokhale, S.M. and Mehta, N.G., The role of the membrane skeleton in concanavalin-A-mediated agglutination of human erythrocytes, Biotech. Appl. Biochem., 12 (1990) 544-549. [37] Pinter, M., Aszodi, A., Friedrich, P. and Ginzburg, I., Calpeptin, a calpain inhibitor, promotes neurite elongation in differentiating PCI 2 cells, Neurosci. Lett., 170 (1994) 91-93. [38] Pollerberg, E., Burridge, K., Krebs, K., Goodman, S. and Schachner, M., The 180 kD component of the neural cell adhesion molecule N-CAM is involved in cell-cell contacts and cytoskeleton-membrane interactions, Cell Tissue Res., 250 (1987) 227-236. [39] Roberts-Lewis, J.M., Savage, M.J., Marcy, V.R., Pinsker, L.R. and Siman, R., Immunolocalization of calpain I-mediated spectrin degradation to vulnerable neurons in the ischemic gerbil brain, J. Neurosci., 14 (1994) 3934-3944. [40] Saido, T.C., Nagao, S., Shiramine, M., Tsukaguchi, M., Sorimachi, H., Murofushi, H., Tsuchiya, T., Ito, H. and Suzuki, K., Autolytic transition of mu-calpain upon activation as resolved by antibodies distinguishing between the pre-and post-autolysis forms, J. Biochem. (Tokyo), 111 (1992) 81-86. [41] Saido, T.C., Sorimachi, H. and Suzuki, K., Calpain: new perspectives in molecular diversity and physiological-pathological involvement, FASEB J., 8 (1994) 814-822. [42] Saido, T.C., Yokota, M., Nagao, S., Yamaura, I., Tani, E., Tsuchiya, T., Suzuki, K. and Kawashima, S., Spatial resolution of fodrin proteolysis in postischemic brain, J. Biol. Chem., 268 (1993) 25239-25243. [43] Seubert, P., Baudry, M., Dudek, S. and Lynch, G., Calmodulin stimulates the degradation of brain spectrin by calpain, Synapse, 1 (1987) 20-24. [44] Seubert, P., Larson, J., Oliver, M., Jung, M.W., Baudry, M. and Lynch, G., Stimulation of NMDA receptors induces proteolysis of spectrin in hippocampus, Brain Res., 460 (1988) 189-194. [45] Seubert, P., Nakagawa, Y., Ivy, G., Vanderklish, P., Baudry, M. and Lynch, G., lntrahippocampal colchicine injection results in spectrin proteolysis, Neuroscience, 31 (1989) 195-202. [46] Siman, R., Baudry, M. and Lynch, G., Brain fodrin: substrate for the endogenous calcium-activated protease, calpain 1, Proc. Natl. Acad. Sci. USA, 81 (1984) 3276-3280. [47] Siman, R., Noszek, J.C. and Kegerise, C., Calpain 1 activation is specifically related to excitatory amino acid induction of hippocampal damage, J. Neurosci., 9 (1989) 1579-1590. [48] Sobue, K., Okabe, T., Itoh, K., Kadowaki, K., Tanaka, T., Kanda, K., Ueki, N. and Fujio, Y., The Ca2+-dependent regulation of the submembranous cytoskeleton in nonerythroid cells: the possible involvement of calspectin (nonerythroid spectrin or fodrin) and its interacting proteins, Biomed. Res., 8 (1987) 13-22. [49] Staubli, U., Kessler, M. and Lynch, G., Aniracetam has proportionately smaller effects on synapses expressing long-term potentiation: evidence that receptor changes subserve LTP, Psychobiology, 18 (1990) 377-381. [50] Stoppini, L., Buchs, P.-A. and Muller, D., A simple method for organotypic cultures of nervous tissue, J. Neurosci. Methods, 37 (1991) 173-182. [51] Terramani, T., Kessler, M., Lynch, G. and Baudry, M., Effects of

P. Vanderklish et al. / Molecular Brain Research 32 (1995) 25-35

thiol-reagents on [3H]alpha-amino-3-hydroxy-5-methylisoxazole-4propionic acid binding to rat telencephalic membranes, Mol. Pharm., 34 (1988) 117-123. [52] Vanderklish, P., Neve, R., Bahr, B.A., Arai, A., Hennegriff, M., Larson, J. and Lynch, G., Translational suppression of a glutamate receptor subunit impairs long-term potentiation, Synopse, 12 (1992) 333-337. [53] Verhallen, P.F., Bevers, E., Comfurius, P. and Zwaal, R.F., Correla-

35

tion between cytoskeletal degradation and expression of platelet procoagulent activity. A role for the platelet membrane-skeleton in the regulation of membrane lipid assymetry?, Biochim. Biophys. Acta, 903 (1987) 206-217. [54] Wu, K. and Siekevitz, P., Neurochemical characteristics of a postsynaptic density fraction isolated from adult canine hippocampus, Brain Res.. 457 (1988) 98-112.