Neuroscience Letters, 153 (1993)80-84 © 1993 ElsevierScientificPublishers Ireland Ltd. All rights reserved0304-3940/93/$06.00
Distribution of calcium binding protein mRNAs in rat cerebellar cortex K o z o K a d o w a k i " , Eileen M c G o w a n a, G r a h a m M o c k b, S t e p h e n C h a n d l e r b a n d Piers C. E m s o n a aMRC Molecular Neuroscience Group, Department of Neurobiology AFRC, Institute of Animal Phsyiology and Genetics Research, Babraham, Cambridge ( UK) and bBritish Biotechnology Products Ltd., Abington, Oxon ( UK)
(Received30 September 1992;Revisedversion received12 January 1993;Accepted 20 January 1993) Key words: Calciumbinding protein; In situ hybridization;Calbindin D28k; Parvalbumin; Calretinin; Cerebellarcortex
The distribution of three calciumbinding protein mRNAs in the rat cerebellarcortex was investigatedusing alkaline phosphatase labelled specific antisense oligodeoxynucleotideprobes. Calbindin D28k mRNA was detected in the Purkinje cells, parvalbumin mRNA was located in the Purkinje cells and also in basket/stellate cells of the molecularlayer. Calretinin in contrast was found only in the granule cell layer. Use of multiple alkaline phosphatase (AP)-labelledoligodeoxynucleotidesresultedin an increasein signal strength and reduceddetectiontime with no increasein background staining indicatingthe utility of these enzymelabelledprobes for non-isotopicin situ.
Calcium ions play a key role in cellular signal transduction in the nervous system and their effects are mediated via interactions with a variety of intracellular calcium binding proteins (CaBPs). Some of these CaBPs such as calmodulin are ubiquitously distributed. Others such as the three CaBPs considered here, calbindin D28k, parvalbumin and calretinin, are present in high concentrations in largely separate sub-sets of neurones [ 2 4 , 13, 15] and their physiological functions are unclear. The high concentrations of these CaBPs in particular neurones, suggest that they may be calcium buffers protecting the neurone against elevated, potentially neurotoxic calcium levels [1, 2, 7, 9] and indeed parvalbumin seems to fulfill this role in fast-firing interneurones in the hippocampus . However the strong sequence conservation of calbindin D28k and calretinin [10-12, 14] suggests that they may interact functionally with other proteins, as does calmodulin, and indeed calretinin has recently been shown to influence the phosphorylation of a synaptic membrane protein from cerebral cortex . It also seems likely that the study of the distribution of these three proteins may provide clues as to their functions as it does for example with the localization of parvalbumin in fast-firing hippocampal interneurones . This type of distribution study using immunohistochemical methods could be problematic because of the potenCorrespondence: RC. Emson, MRC Molecular Neuroscience Group, Department of Neurobiology,AFRC, Institute of Animal Physiology and Genetics Research, Babraham, Cambridge CB2 4AT, UK.
tial cross-reactivity of the polyclonal or monoclonal antibodies used in immunohistochemistry for the conserved amino acid sequence present in many of these CaBPs. In order to avoid these potential problems and to unambiguously map the sites of expression of these three proteins we have designed sequence specific antisense enzyme labelled oligodeoxynucleotide probes for calbindin D28k, parvalbumin and calretinin. Here we show that these probes recognize distinct populations of neurones in the cerebellum. Alkaline phosphatase labelled antisense oligodeoxynucleotide probes were produced by coupling alkaline phosphatase (AP) either to the 5' end via AminoLink 2 (Applied Biosystems, Warrington, Cheshire), or to an amino group attached to a modified thymine base within the sequence, essentially as described by Jablonski et al. . For calbindin Dz8k and parvalbumin several AP-labelled probes were designed to different parts of the c D N A sequence so that their combined use could be assessed. The probes corresponded for calbindin Dzsk to amino acids 3 1 4 1 , 71-81,186-196, 241-250  and for parvalbumin amino acids 24-34, 54-63 and 79-88 . The calretinin sequence corresponds to bases 239-271 of the calretinin c D N A  and has no significant homology with calbindin D28k- The probes were purified by ion-exchange chromotagraphy using a Mono-Q column (Pharmacia), as shown in Fig. 1. In all experiments adult male Wistar rats (175-200 g) were used. In situ hybridization was performed as described previously . Briefly, after decapitation, rat
z .o_ i
When colour development was complete, the sections were washed in Buffer III (0.01 M Tris-HC1 pH 7.4, 10 mM EDTA, 0.15 M NaC1) for 1-2 h to terminate the reaction. Finally the sections were coverslipped with glycerin jelly and stored at 4°C in the dark. To demonstrate the specificity of hybridization, sections were hybridized with the AP-labelled probe in the presence of an excess of unlabelled antisense oligonucleotide (400 fmol/#l). This treatment abolished all the signal (see Fig. 3f for example). All the AP-labelled probes showed each CaBP had a
Fig. 1. A typical FPLC separation of an alkaline phosphatase (AP)labelled calbindin D28k oligonucleotide (B) from unconjugated AP (A) and unlabelled amino-linked oligonucleotide (C). This type of purification is essential for preparation of AP-labelled oligonucleotides, for use in in situ applications, failure to remove either AP or unlabelled aminolinked oligonucleotide results in high background and absence of specific hybridization signal.
brains were snap frozen on dry ice and stored at -80°C until required. Fresh frozen coronal cryostat sections (15 #m) of rat cerebellum were cut and thaw mounted onto gelatin coated glass slides, quickly dried, and stored at -80°C until used. The sections were fixed with 4% neutral buffered paraformaldehyde (pH 7.4) for 30 rain at room temperature and then rinsed twice in 0.1 M phosphate-buffered saline (PBS) followed by incubation in 0.25% acetic anhydride and 0.1 M triethanolamine/0.9% NaC1 for 10 min at room temperature. Sections were then dehydrated through graded ethanol (70, 80, 90, 95, 100%, 5-10 each), delipidated in chloroform (10 min) and partially rehydrated (95% ethanol). The slides were allowed to air dry thoroughly. All sections were then incubated in hybridization buffer (4 x SSC, 50% formamide, 1 x Denhardt's solution, 250 #g/ml sheared salmon testes DNA, and 10% dextran sulphate) containing the AP-labeUed antisense probes, at a final concentration of approximately 2 fmol//A. Hybridizations were performed overnight at 37°C in a moist-chamber. After hybridization the sections were rinsed in 1 x SSC and sequentially washed in 1 x SSC (3 x 30 min) at 55°C; 1 x SSC at room temperature for 1 h; Buffer I (0.1 M Tris-HCl, pH 7.4, 0.15 M NaC1) at room temperature for 30 min; and Buffer II (0.1 M Tris-HC1, pH 9.5; 0.1 M NaC1, 0.05 M MgC12) at room temperature for 10 min. The sections were then incubated in AP substrate solution (Buffer lI containing Nitroblue tetrazolium (340 #g/ ml) and 5-bromo-4-chloro-3-indolyl phosphate (170 #g/ ml)), in a dark humidified chamber for 12-24 h; AP substrates were obtained from Boehringer Mannheim.
A t '¢
Fig. 2. Distribution of mRNAs of calbindin-Dzsk, parvalbumin and calretinin in rat cerebellum. A: calbindin-D28k mRNA; B: parvalbumin mRNA; C: calretinin mRNA. The arrow in C points to some strongly calretinin mRNA positive cells in the motor trigeminal nucleus. Bar = 1 ram.
g ~ ~! i!: :i:¸ ~i¸ - /
~i!' i~(:i¸i!~! ii
ii! ¸ i;i -!
~i~':~, ¸ ~ ~ii
n~ ~!i~i'¸, ~ ~ :~~i~~i~~i~!~ii~i~ilill
i! i~i!~ ~ ,::~
ii ¸ ~ •::~
Fig. 3. Distribution of CaBPs in the rat cerebellum, a,b: calbindin D28k (CAB) m R N A here the target m R N A is localized only in the Purkinje cells (p). c,d: parvalbumin (PaV) m R N A , note the PAV m R N A in the Purkinje cells and basket/stellate cell types (b) in the molecular layer, e: calretinin (CAR) m R N A , note the calretinin signals is only present in the granule cell layer, f: a control section hybridized with ttie calretinin AP-labelled oligonucleotide and an excess of cold oligonucleotide. Note the absence of specific CaR signal, m: molecular cell layer, g: granule cell layer, w: white matter. Bars = 100/.tm.
83 distinct and different expression pattern in the cerebellum (Fig. 2). Calbindin D28k m R N A was concentrated exclusively in the Purkinje cells (Fig. 2a); no staining of any other cerebellar neurone types was seen, which is in good agreement with earlier immunohistochemical studies [2, 16]. The signal was uniformly distributed throughout the cytoplasm (excluding the nucleus). The combined use of AP-labelled calbindin probes (up to four different AP probes used together) resulted in a significant increase in signal strength and shortened the time required to develop the signal, from 24 to 12 h. Despite the relative increase in signal strength no other intracellular signal was visualized and there was no increase in background staining. In contrast to calbindin D28k mRNA, which was present only in Purkinje cells, parvalbumin m R N A signal was found in both Purkinje cells and in cells of the molecular layer (presumably basket or stellate cell types) 2 (Fig. 2b). As with calbindin D28k the use of multiple oligonucleotides for parvalbumin increased the signal intensity without increasing background or nonspecific cell labelling. Note that the amount of AP reaction product in the cells of the molecular layer is distinctly lower compared with the high intensity o f labelling in the Purkinje cells (Fig. 3c,d). The basket/stellate cell signal was cytoplasmic (i.e. non-nuclear). Again, as for calbindin D28k the distribution of parvalbumin m R N A was in good agreement with previous immunohistochemical studies [2, 3, 15]. The distribution of the calretinin m R N A was in marked contrast to that of calbindin D2ak and parvalbumin mRNAs. This signal was localized only in the granule cell layer and no calretinin m R N A was detected in the Purkinje cells (Figs. 2c, 3e). The presence of calretinin immunoreactivity in the granule cells has been reported by R6sibois and Rogers [13, 15]. These data confirm their observations and indicate that the majority of granule cells express the mRNA. The results presented here demonstrate the quality and clarity of the m R N A signals obtained with AP-labelled oligonucleotides, namely low background and a high signal to noise ratio. The ability to substantially increase the 'signal' strength by using multiple AP-labelled probes shortens development time and further increases their usefulness. The main advantages of enzyme linked oligonucleotides are: (1) clear single cell localization of an m R N A transcript; (2) no cross-reaction between homologous mRNAs; (3) detection of mRNAs with a lower cellular expression by using multiple oligonucleotides and (4) detection of the chosen m R N A species within 24 h, a considerably shorter time than that required for conventional 35S or 32p hybridization techniques. The specificity and ease of use of the calbindin D28k, parvalbumin and calretinin probes described here will
hopefully enable us to determine if their expression is influenced by different physiological and pharmacological treatments. The possible physiological functions of these calcium binding proteins remain to be established but it is striking that the cerebellar Purkinje cells have two high affinity calcium binding proteins calbindin D28k and parvalbumin whereas the granule cells and basket cells have only one each (respectively, calretinin and parvalbumin). This perhaps implies that the Purkinje cells need to have much more precise, or regional control over calcium signalling in their dendrites and it may be that the two proteins are required to provide different spatial or temporal aspects of calcium handling but this remains a speculation. Certainly all the cerebellar cells can potentially buffer calcium but why each cell has its own unique expression pattern remains to be established. We thank Mrs Barbara Oakley for typing the manuscript, the M R C and Bayer (FRG) for financial support. 1 Baimbridge,K.G., Miller, J.J. and Parkes, C.O., Calcium-binding protein distributionin the rat brain, Brain Res., 239 (1982)519-525. 2 Celio, M.R., Calbindin D-28K and parvalbumin in the rat nervous system, Neuroscience,35 (1990)375-475. 3 Celio,M.R. and Heizmann, C.W., Calcium-bindingprotein parvalbumin as a neuronal marker, Nature, 293 (1981) 300 302. 4 Celio, M.R., Parvalbumin in most ~'-aminobutyricacid-containing neurons of the rat cerebral cortex, Science,231 (1986)995 997. 5 Epstein, R, Means, A.R. and Berchtold, M.W., Isolation of a rat parvalbumin gene and full length cDNA, J. Biol. Chem., 261 (1986) 5886-5891. 6 Jablonski, E., Moomaw, E.W., Tullis, R.H. and Ruth, J.L., Preparation of oligodeoxynucleotide-alkalinephosphatase conjugates and their use as hybridizationprobes, NucleicAcid Res., 14 (1986) 6115-6128. 7 Kawaguchi,Y., Katsumaru, H., Kosaka, T., Heizmann, C.W. and Hama, K., Fast spiking cells in rat hippocampus (CA1region) contain the calcium-binding protein parvalbumin, Brain Res., 416 (1987) 36%374. 8 Kiyama, H., Emson, RC., Ruth, J.L. and Morgan, C.Y., Sensitive non-radioisotopicin situ hybridizationhistochemistry;demonstration of tyrosinegeneexpressionin rat brain and adrenal, Mol. Brain Res., 7 (1990)213-219. 9 Mattson, M.R, Rychlik, B., Chu, C. and Christakos, S., Evidence for calcium-reducing and exito-protectiveroles for the calciumbinding protein calbindin-D28Kin cultured hippocampal neurons, Neuron, 6 (1991)41 51. 10 Nordquist, D.T., Kozak, C.A. and Orr, H.T., cDNA cloning and characterizationof three genes uniquelyexpressedin cerebellumby Purkinje neurons, J. Neurosci., 8 (1988)4780-4789. 11 Parmentier,M., Lawson, D.E.M. and Vassart, G., Human 27-kDa calbindin complementaryDNA sequence; evolutionary and functional implications, Eur. J. Biochem., 170 (1987)207-215. 12 Parmentier, M. and Lefort, A., Structure of the human brain calcium-binding protein calretinin and its expression in bacteria, Eur. J. Biochem., 196 (1991)76-83. 13 R6sibois, A. and Rogers, J.H., Calretinin in rat brain: an immunohistochemicalstudy, Neuroscience,46 (1992) 101-134.
84 14 Rogers, J.H., Calretinin: A gene for a novel calcium-bindingprotein expressed principally in neurons, J. Cell Biol., 105 (1987) 13431353. 15 Rogers, J.H,, Immunoreactivity for calretinin and other calciumbinding proteins in cerebellum, Neuroscience, 31 (1989) 711-721. 16 Sequier, J.M., Hunziker, W., Andressen, C. and Celio, M.R., Cal-
binding D-28k protein and mRNA localization in the rat brain, Eur. J. Neurosci., 2 (1990) 1118-1126. 17 Yamaguchi, T., Winsky, L. and Jacobowitz, D.M., Calretinin, a neuronal calcium binding protein, inhibits phosphorylation of a 39k Da synaptic membrane protein from rat brain cerebral cortex, Neurosci. Lett., 131 (1991) 79-82.