SYNAPSE 8:119-136 (1991)

Subsurface Cisterns in a-Motoneurons of the Rat and Cat: Immunohistochemical Detection With Antibodies Against Connexin32 T. YAMAMOTO, E.L. HERTZBERG, AND J.I. NAGY Department of Physiology, Faculty of Medicine, University of Manitoba, Winnipeg, Manitoba, Canada R3E OW3 (T.Y., J.I.N.) and Departments of Neuroscience, Anatomy, and Structural Biology, Albert Einstein College of Medicine, Bronx, New York 10461 (E.L.H.)

KEY WORDS

C-terminals, Calcium, Ryanodine, Cholinergic synapse, Connexin, Gap junction

ABSTRACT A monoclonal antibody against amino acids 224-234 of the gap junction protein connexin32 was found by immunohistochemistry to label subsurface cisterns (SSCs) in alpha-motoneurons of the rat (Yamamoto et al., 1990) and was used here to document by light (LM) and electron microscopy (EM)the appearance of immunoreactive SSCs in motoneurons of the rat and cat. This antibody and a polyclonal antibody against connexin32 labelled gap junctions in rat liver as well as SSCs in facial motoneurons. By LM, SSCs were seen as labelled puncta on motoneuronal perikarya and proximal dendrites. In the rat, they appeared to be present on all motoneurons at cranial and spinal levels, but varied considerably in size and number among motor nuclei. Labelled SSCs were the smallest and most sparse in motoneurons of the dorsal vagal motor nucleus, moderate in size and most numerous in the trochlear, oculomotor, and trigeminal motor nuclei, and largest though less densely distributed in spinal motoneurons. Dendrites were seen to contain SSCs for distances of up to 230 pm from their soma1 origm. Labelling within individual SSCs seen en face consisted of either numerous small puncta or linear arrays of immunoreactivity. By EM, labelled SSCs in the rat facial nucleus were always seen beneath a cluster of C-terminals. Immunolabelling was most dense in the space between the plasma membrane and SSC,which we define as the subsurface cisternal cleft. The SSCs were usually intermittently labelled along their length and exhibited a narrow luminal space ranging from 2 to 5 nm. On the basis of structural analogies between SSCs in neurons and the sacroplasmic reticulum terminal cistern/”-tubule complex in muscle, SSCs have previously been suggested to be important sites of calcium mobilization. The constant association of C-terminal with SSCs in motoneurons may represent a useful model in which to study SSC function as well as to investigate the possible presence of a connexinlike protein at regions of SSCs that form a narrow lumen similar to that at gap junctions. INTRODUCTION Classification of the various types of axon terminals that contact alpha-motoneurons has been the subject of numerous studies. According to generally accepted nomenclature and notwithstanding differences among species and motor nuclei among cranial and spinal levels, the well-established types of terminals include S-boutons (spherical vesicles), T-boutons (postsynaptic Taxi body), F-boutons (flat vesicles), M-boutons (monosynaptic), P-boutons (presynaptic) and C-boutons (cistern). The C-terminals, as described by various authors 0 1991 WILEY-LISS, INC.

(Bak and Choi, 1974; Bodian, 1966; Boone and Aldes, 1984; Charlton and Gray, 1966; Conradi, 1969a-d; Barbas-Henry and Wouterlood, 1988; Conradi et al., 1979; Goshgarian and Rafols, 1984; Hamos and King, 1980; Johnson, 1986; Kellereth et al., 1979; Lagerback et al., 1986; McLaughlin, 1972a; Momoh and Mayhew, 1982; Pullen, 1988; Sakamoto et al., 1985; Schroder, Received October 8,1990; accepted in revised form December 13,1990 Address reprint requests to Dr. J.I. Nagy, Department of Physiology, Faculty of Medicine, University of Manitoba, 770 Bannatyne Avenue, Wmnipeg, Manltoba, Canada R3E OW3.

120

T. YAMAMOTO ET AL

1979;Voss et al., 1980), are found almost exclusively on somata and proximal dendrites, are not present on other than alpha-motoneurons, exhibit a wide range of sizes (2-10 pm) but are commonly the largest of boutons on motoneurons, and represent anywhere from 1to 14%of all synaptic terminals on motoneuronal somata. These terminals have been reported on motoneurons in monkey oculomotor nucleus, cat thoracic cord, trochlear motor nucleus, and motor nucleus of Onuf, rat cervical cord, phrenic motor nucleus and hypoglossal nucleus, opossum trigeminal motor nucleus, chicken cervical cord, frog lumbar cord, and lizard abducens motor nucleus. Although early investigators questioned their synaptic nature, it appears that C-terminals exhibit typical paramembranous specializations, presynaptic dense projections, and vesicle aggregations at what can be classified as “active” zones. The C-terminal is so designated because it is always accompanied postsynaptically by a subsurface cistern (SSC). Occasionally, several of these boutons are seen closely apposed to one another and where this occurs, a single postsynaptic SSC extends beneath the entire cluster. Conversely, SSCs in mammalian motoneurons have never been reported at locations other than in apposition to Cboutons, at least under nonexperimental conditions. Engstrom (1958) and subsequently Smith and Sjostrand (1961) were the first to describe what they called double or accessory membranes closely apposed to the plasma membrane in the outer hair cells of the organ of Corti. Rosenbluth (1962a,b)found similar structures in neurons of the rat cerebral cortex and spinal cord and termed them subsurface cisterns. He described SSCs as consisting of flattened, sheetlike arrangements of endoplasmic reticulum (ER) separated from the plasma membrane by 5 to 8 nm. Their luminal depth varied from 4 to 60 nm and their length from 1 to 4 pm. They were usually located in neuronal perikarya and proximal large calibre processes but were absent in glial cells. SSCs were later found in neurons in a number of other brain reGons as well as various peripheral cell types (Adinolfi, 1969; Anzil et al., 1971; Bodian, 1972; Herndon, 1964; Herves and Lafarge, 1979;Bunge et al., 1965; Gulley and Wood, 1981; Le Beux, 1972; Pappas and Waxman, 1972; Peters et al., 1988; Spoerri and Glees, 1977; Siegesmund, 1968; Takahashi and Wood, 1970; Watanabe and Burnstock, 1976; Weis, 1968). In nigral neurons of the rat and cat, Le Beux (1972)observed that one or more SSCs were stacked against the plasmalemma and that mitochondria were frequently associated with and occasionally closed applied to SSCs. He drew an analogy between the narrow luminal space of SSCs and the 2-nm extracellular gap of gap junctions, noting that such a gap may also occur between the cisternal membranes. Subsequently, Henkart et al. (1976) also described the inner and outer cisternal membranes as being separated by a gap of 2 to 5 nm. Although the function of SSCs is unknown, it has been

speculated that they may be important sites governing cellular calcium mobilization (McBurney and Neering, 1987). We previously reported that some antibodies against the gap junctin protein connexin32 recognize several distinct proteins on Western blots of brain homogenates and immunohistochemically label SSCs in motoneurons of the rat (Yamamoto et al., 1989a, 1990). Whereas the nature of the cross-reacting material, its possible relationship to connexins, and the significance of the structural similarities between gap junctions and SSCs have yet to be determined, it appears that the immunoreaction observed at SSCs may be useful in anatomical studies ofthese structures as well as, by definition, their associated C-terminals. Here, we provide an account of the appearance of SSCs immunohistochemically labelled by these antibodies in cranial and spinal motoneurons of the rat and cat. MATERIALS AND METHODS Immunohistochemistry Twenty-six male Sprague-Dawley rats, weighing 250-350 g, were deeply anesthetized with chloral hydrate and perfused transcardially with various combinations of prefixative and fixative solutions. For light microscopy (LM), the prefixative solution consisted of 70 ml of cold (4°C)50 mM sodium phosphate buffer (PB), pH 7.4,0.9% saline, 0.1% sodium nitrite, and heparin (1 unit/ml). The fixatives tested included: 400 ml of 0.1 M PB, pH 6.9, containing 4% paraformaldehyde and 0.2% picric acid; 400 ml of 0.1 M PB, pH 6.9, containing 4% paraformaldehyde, 0.2% picric acid, and 0.2% glutaraldehyde; 400 ml of periodate-lysine-paraformaldehyde (PLP) prepared as previously described (McLean and Nakane, 1974);and 200 ml of 0.1 M PB, pH 7.4, containing 4% paraformaldehyde followed by 200 ml of 50 mM sodium borate buffer, pH 9.0, containing 4% paraformaldehyde. The brain, spinal cord, and liver were removed and postfixed for 2 to 12 hr in fresh fixative solutions identical to those used for perfusion. For studies involving cats, four animals (2.0-2.5 kg) were deeply anesthetized with pentobarbital and perfused transcardially with 500 ml of 0.9% saline followed by 2 L of 0.1 M PB (pH 7.4) containing 4% paraformaldehyde and 0.2%picric acid. The lumbar spinal cord was removed and postfixed in the same fixative for 12 hr. The tissues were then cryoprotected in 50 mM PB, pH 7.4, containing 25% sucrose and 10% glycerol. Frozen sections cut at a thickness of 20 pm on a sliding microtome were collected in PB containing 0.9% saline (PBS), washed at 4°C for 12 to 18 hr in 0.1 M PB, containing 0.9% saline and 0.3% Triton X-100 (PBST), and then incubated at 4°C for 48 hr with a mouse ascites fluid monoclonal antibody diluted 15,000 in PBST. This antibody, designated 92B, was raised against a BSAconjugated synthetic peptide corresponding to amino acids 224-234 of connexin32 (Paul, 1986)and was char-

SUBSURFACE CISTERNS IN MOTONEURONS

acterized as previously reported (Hertzberg et al., 1988; Yamamoto et al., 1990). Sections of brain, spinal cord, and liver processed by the peroxidase antiperoxidase (PAP) method were then rinsed in PBST for 1 hr, incubated for 2 hr at room temperature with rabbit antimouse IgG (SternbergerMeyer) diluted 1:20 in PBST, washed for 1 hr in PBST, and then incubated for 2 hr at room temperature with mouse Clono PAP (Sternberger-Meyer)diluted 1 : l O O in PBST. The sections were then washed for 30 min in 50 mM Tris-HC1buffer, pH 7.4, and reacted for 5-10 min in the same buffer containing 0.02% 3,3’-diaminobenzidine (DAB)and 0.005% hydrogen peroxide. The sections then received a further wash in Tris-HC1 buffer and were mounted onto slides from gelatidalcohol, dehydrated in alcohol, and coverslipped with Lipshaw mounting medium. Some sections were counterstained with thionin. Control procedures included sections processed with omission of primary antibody or with this antibody preabsorbed with the synthetic peptide against which it was directed. For the preabsorption the antibody was diluted 1500 in PBS and a volume of 1 ml was incubated for 2 hr at room temperature with 3 ~1of a 10 pM solution ofpeptide (MW 1244).The preabsorbed antibody was used at a final dilution of 1:5,000. In order to compare labelling patterns produced by antibody 92B with that of a previously characterized polyclonal anticonnexin32 antibody of sheep origin (Hertzberg, 1984,1985; Hertzberg and Skibbens, 1984; Nagy et al., 1988;Yamamoto et al., 1989b), some tissues were processed for double immunofluorescence by a sequential labelling method. Sections of liver and brainstem were incubated for 48 h r at 4°C with antibody 92B diluted 1:5,000 in PBST, then washed in PBS for 1hr at room temperature, and incubated for 2 hr with fluorescein-conjugated rabbit antimouse IgG (Sigma).After a wash in PBST for 1hr, the sections were incubated for a further 48 hr at 4°C with the polyclonal antibody diluted 1:2,000 in PBST, washed in PBST for 1 hr, incubated with biotin-conjugated donkey antisheep antibody diluted 1 : l O O (Amersham), washed in PBST for 1hr, and incubated with avidin-conjugated Texas Red (Amersham). After a final wash in PBST, the sections were mounted onto slides and coverslipped using anti-fade medium (Valnes and Brandtzaeg, 1985). As control procedures, some sections were processed with omission of either antibody 92B or the polyclonal antibody in the first or second sequence reaction steps, respectively. Subsequent incubation of these sections with inappropriate secondary antibody produced no immunofluorescence labelling, and incubation with both secondary antibodies produced only single labelling corresponding to the primary antibody included. Sections were examined with a Leitz Dialux 20 under brightfield, darkfield, and differential interference contrast optics for PAP stained sections, and with I2 (excitation 450490 nm; long pass 515 nm) and N2 (excita-

121

tion 530-560 nm; long pass 580 nm) Ploemopak filter cubes for FITC and Texas Red fluorescence, respectively. Electron microscopy Six rats were used for EM observations. The animals were perfused with 70 ml of 0.1 M PB, pH 7.4, followed by two fixation procedures that were tested. The first involved perfusion with 400 ml of fixative consisting of 0.1 M PB, pH 6.9, 4% paraformaldehyde, and 0.2% glutaraldehyde followed by a 2-hr postfixation in the same fixation solution but without glutaraldehyde. The second involved perfusion with 200 ml of 4% paraformaldehyde in PB (pH 7.5) followed by 200 ml of 4% paraformaldehyde in 50 mM borate buffer, pH 9.0, and an overnight postfixation in the same fixative. Vibratome sections of brain, spinal cord, and liver, cut at a thickness of 20-30 km, were processed by the PAP method as described above except that Photo-Flo 200 (Kodak) instead of Triton X-100 was used as detergent in all incubation and wash steps. The sections were then further fixed for 2 hr in a solution of 0.1 M PB, pH 7.4, containing 2% osmium tetroxide, dehydrated in alcohol, and flat-embedded in Jembed. Sections were examined by LM and desired areas were trimmed and glued onto resin blocks. Ultrathin sections on mesh grids were counterstained with lead citrate for 0.5-1 min. The glutaraldehyde-containing fixation protocol was the tissue preparative procedure of choice for EM, but both of the fixatives tested gave the same qualitative patterns of staining as seen in LM preparations. EM specimens were examined with a Philips 201 electron microscope. RESULTS General observations In addition to previously published data on the specificity of the polyclonal antibody and the monoclonal antibody 92B used here (see Materials and Methods), both antibodies were found to label rat liver gap junctions by EM (Fig. lA,B). Immunolabelling was seen at both surfaces of the plasma membrane along regions where these formed gap junctions. In weakly stained gap junctions, a 2-nm extracellular junctional space was visible and where this space was occluded by DAB deposition the width of the deposition was also about 2 nm. The appearance of profiles labelled with antibody 92B (Fig. 1B) and the polyclonal antibody (Fig. 1A) was similar, except that occlusion of the extracellular junctional space occurred less frequently with the former. No immunolabelling was observed on nonjunctional cytoplasmic membranes or at tight junctions located near either the bile canaliculi or the space of Disse. By LM, structures immunolabelled in motoneurons with antibody 92B were categorized into two types on the basis of their size and cellular location. The first type appeared as large puncta with lengths or diameters greater than 0.5 pm. These structures were relatively

122

T. YAMAMOTO ET AL.

well labelled under all of the fixation conditions tested, including those containing glutaraldehyde where overall staining intensity was only slightly reduced. Based upon EM observations (see below), this large punctate staining was localized to SSCs. The second type consisted of small puncta with diameters of less than 0.5 pm. These were dispersed throughout neuronal somata and were most evident with the PLP fixative and exhibited slightly reduced staining intensity with the paraformaldehydetpicric acid fixative. Staining of these elements was largely but not totally suppressed with either the pH-change or glutaraldehyde-containing fixatives. Based on EM observations of motoneurons in the rat, this small punctate staining appeared to be localized to multivesicular bodies. The present LM analyses involved primarily tissues of rats fixed by the pHchange protocol, which was optimal for visualization of SSCs. Tissues from cats were fixed only by the paraformaldehydelpicric acid method and other fixatives were not tested. Preabsorption of antibody 92B (Fig. 1D) with synthetic peptide eliminated the characteristic large punctate immunoreactivity in motoneurons of the rat (Fig. 1C) and the cat (not shown), and no

immunolabelling was seen in sections processed with omission of this antibody (not shown). Similarly, labelling of small (