THE JOURNAL OF COMPARATIVE NEUROLOGY 314~750-762 (1991)
Calmodulin and Calbindin Localization in Retina From Six Vertebrate Species ROLAND POCHET, BRIGITTE PASTEELS, AKIKO SETO-OHSHIMA, ENRICO BASTIANELLI, SATOKO KITAJIMA, AND LINDA J. VAN ELDIK Laboratoire d'Histologie, Facult6 de MQdecine,Universitk Libre de Bruxelles, B- 1070 Bruxelles, Belgium (R.P.,B.P.,E.B.); Laboratoire de Biologie Cellulaire, Facult6 des Sciences, Universite de Poitiers, 86022 Poitiers, France (R.P.); Institute for Developmental Research, Aichi Prefectural Colony, Aichi 480-03, Japan (AS.-O.,S.K.);Departments of Pharmacology and Cell Biology, Vanderbilt University, Nashville, Tennessee 37232-6600 (L.J.V.E.)
ABSTRACT Calmodulin is abundant in the central nervous system, including the retina. However, the localization of calmodulin in the retina has not been described in detail. We therefore decided to investigate calmodulin localization in retinae from six vertebrate species, by using immunohistochemical labeling with four different rabbit polyclonal antibodies against calmodulin. The localization of calbindin-D28k, another calcium-binding protein already well described in retina, was compared. We found that calmodulin distribution is more highly conserved among species, contrasting with calbindin variability. The most striking result emerging is that calmodulin could not be detected in photoreceptors although other layers are intensely calmodulin-immunoreactive, casting doubt about a direct role of calmodulin in phototransduction. Horizontal cells are weakly calmodulin-immunoreactive, bipolar cells are calmodulinimmunoreactive except in turtle retina, numerous amacrine and ganglion cells are labeled in all species, and the fiber layer is always labeled. These data demonstrate that, while the calmodulin distribution in retina is similar among vertebrate species, selective differences in localization can be detected not only among the same cell types in different species but also among different cell types in the same species. The results showing differences in calmodulin immunoreactivity among cell types also provide further evidence that calmodulin expression in eukaryotes is not constitutive, in the sense that not every cell expresses similar levels of calmodulin. Key words: calcium-bindingproteins, immunohistochemistry,photoreceptors
Calmodulin and calbindin D-28k are two calcium binding proteins having common Ca2+-bindingmotifs. Calmodulin is an intracellular Ca" receptor and is highly conserved at the primary amino acid sequence level. In addition, calmodulin serves as the obligatory Ca'+-dependent activator of a variety of enzymes and is associated with several intracellular structural proteins (for reviews see Cohen and Klee, '88; Means, '88; Means and Dedman, '80; Van Eldik and Roberts, '88; Van Eldik et al., '82). Although it is generally accepted that calmodulin is found in all eukaryotes, its levels and localization can vary in different cell types and species. In retina, calmodulin has been investigated biochemically (Kohnken et al., '81; Nagao et al., '871, but to date there has not been a detailed description of calmodulin-containing neurons, although their presence is mentioned by Wood et al. ('80b), studying immunocytochemically the distribution of calmodulin in different rat brain regions; by SetoOhshima et al. ('871, studying mouse embryos; and by Wakakura and Yamamoto ('87), mentioning the presence of 0 1991
WILEY-LISS. INC.
calmodulin in feline rod outer segments. The two biochemical manuscripts focused their studies on rod outer segments while the three morphological papers examined rat, mouse, and cat retina layers. In rat retina, Wood et al. ('80b) showed calmodulin immunoreactivity in the cytoplasm of ganglion cells as well as in fibers from amacrine and ganglion cells of the inner plexiform layer. No photoreceptors were positive for calmodulin. In adult mouse (SetoOhshima et al., '87), the ganglion cell layer, some somata in the inner nuclear layer, and fibers in the inner plexiform layer were calmodulin-immunoreactive (IR). Again no photoreceptors were positive for calmodulin. In feline retina, Wakakura and Yamamoto ('87) reported that in addition to the nerve-fiber layer, the ganglion cell layer, and the inner plexiform layer, rod inner- and outer-segments were also calmodulin-IR. Some discrepancies thus seem to exist about the presence of calmodulin in photoreceptors.
Accepted September 20, 1991.
751
CALMODULIN AND CALBINDIN IN RETINA
Fig. 1. Section of the same mouse retina treated with four anti-calmodulin antibodies: antibody #465 in a, #860 in b, #009 in c and #449 in d. #009 was used at 0.4 pgiml; #465, #449, and #860 at 112,000 dilution. Scale bar = 25 pm.
We therefore decided to investigate in detail and compare in different species the distribution of calmodulin-IR neurons in the retina. We selected retinae from six vertebrate species that are widely studied in many laboratoriesnamely, goldfish (fish), turtle (reptile), frog (amphibian), pigeon (avian),mouse and monkey (mammalian).CalbindinD28k localization in retinae from those species has been included in this work for two purposes: 1)to extend studies performed on human (Verstappen et al., '861, rabbit (Schreiner et al., '85), rat (Rabi6et al., '851, pigeon (Pasteels
Abbreviations
b CaBP CAM For FL GC h
INL IPL IR OLM ONL OPL 1-5 Ph RF ROS
YF
bipolar cells calbindin calmodulin fiber layer ganglion cell layer horizontal cells inner nuclear layer inner plexiform layer immunoreactive outer limitingmembrane outer nuclear layer outer plexiform layer laminae of the inner plexiform layer photoreceptor layer red field rod outer segment yellow field
Fig. 2. Section of frog retina treated with anti-rhodopsin (A01 at l/1,000 dilution. Photoreceptor outer segments (as) are AO-IR, whereas they are CAM-negative (see Fig. 4). is: photoreceptor inner segments. Scale bar = 25 pm.
3 9
s v E
ldl
Z L
lNI 1dO 1NO
cld
J
9
ldl
1NI 1dO 1NO
4d
*WJ 3 J3H30d *?I
753
CALMODULIN AND CALBINDIN IN RETINA
Fig. 5. Sections of goldfish retina treated with anti-calmodulin (CAM) at 0.4 Fg/ml. The section in b was situated in a more peripheral part of the retina than that in a. Photoreceptors are clearly CAMnegative. CAM-positive horizontal and bipolar cells are seen well in b,
whereas CAM-positiveamacrine cells, ganglion cells and three immunoreactive bands in the IPL appear more clearly in a. A large bistratified amacrine cell showing intense labeling is indicated in a by an arrow. Scale bars = 25 um.
and named #449, #465, #860. Antibodies #449 and #465 were produced against performic acid-oxidized, vertebrate calmodulin, and #860 was made against a synthetic peptide of the COOH terminus (residues 134-148) of vertebrate calmodulin. Purified IgG fractions were routinely used at two dilutions ( l / l , O O O and 1/2,000). The fourth antibody used was produced against bovine brain calmodulin as described earlier (Kitajima et al., '83; Sano et al., '87) and named #009. This antibody was purified by calmodulin affinity chromatography and used at 0.4 kg IgGiml. Controls included antibody preincubated with 6 pg/ml calmodulin (Sigma, St-Louis) for 24 hours before staining. Under these conditions, no labeling was observed (see Fig. 14). Anti-rhodopsin. Preparation of the rat antiserum against purified bovine opsin was described elsewhere (Szel et al., '86). Briefly, opsin was purified from bovine retinae and rats were immunized with a mixture of 1 mg/ml opsin solution and complete Freund adjuvant. Rat anti-rhodopsin serum was used a t l/l,OOO dilution. Anti-calbindin. Rabbit antiserum raised against chick duodenal calbindin-D28k was prepared as described (Spencer et a]., '76), and routinely used at 1/6,000 dilution. Tissue preparation. Retinae from goldfish (Carassius auratus), frogs (Rana esculenta), and turtle (Pseudemis scripta elegans) were dissected after deep anesthesia in cold water containing 1 g/L tricaine (MS-222, Sandoz, Basle). Pigeon (Columba liuia donestica) and mouse (Mus musculus) were deeply anesthetized with ether and monkeys (Cercopithecus aethiops sabaeus) with pentobarbital. After removal of cornea and lens, the eyecups were fixed for 24 hours in Sublimate Bouin Holland containing 4 g cupric
acetate, 4 g picric acid, 10 ml formalin (40% W/V solution), 1g trichloroacetic acid, and 1ml mercuric chloride aqueous saturated solution, in 100 ml distilled water. The specimens were then dehydrated and embedded in paraffin wax. Five micrometer sections were cut for immunohistochemical analyses. Peroxidase immunohistochemistry. Immunoperoxidase staining was done as previously published (Vaccaet al., '80) and modified by us (Pasteels et al., '87a,b; Pochet et al., '87). In brief, routinely dewaxed and hydrated sections were processed as follows: (1) rinse in Coons Verona1 Buffer saline (CVBS); (2) preincubation in 5% normal sheep serum; (3) incubation with rabbit antibodies or rat antirhodopsin for 48 hours at 4°C in a moist chamber; (4) 15-minute incubation with goat anti-rabbit IgG serum (Bio-Yeda, Israel) at 1:80 dilution for anti-calmodulin and anti-calbindin, or biotinylated anti-rat IgG (Vector Laboratories, Burlingame, CA, USA) at 20 pg/ml for antirhodopsin; (5) incubation with soluble rabbit PAP complex 1:300dilution (DAKO, Denmark) or horseradish peroxidaseavidin D at 25 pg/ml (Vector Laboratories, Burlingame, CA, USA); (6) stainingin a citrate phosphate buffer, pH 6.2, containing 0.5 mg/ml 3,3'-diaminobenzidine HC1 (DAB) and 0.1% H,O,.
RESULTS The four different anti-calmodulin antibodies gave similar staining patterns and no discrepancies in the neuronal and fiber labeling could be detected. Minor differences were seen at the background level (see Fig. 1).The calmodulin
754
R. POCHET ET AL.
Fig. 6. Sections of frog retina treated with anti-calmodulin (CAM) at 0.4 pg/ml. The section in a was situated in a more peripheral part of the retina than that in b. In the photoreceptor layer, only some cell
bodies are CAM-IR (a). Horizontal (h),bipolar (b),numerous amacrine and ganglion cells are labeled, as well as the fiber layer and bands in the IPL. Scale bars = 25 km.
affinity purified IgG fraction (#009) prepared by Kitajima gave the lowest background. Therefore, all subsequent figures are photomicrographs from sections immunohistochemically labeled with this antibody. Nomenclature for the laminae of the inner plexiform layer (IPL) was adopted from Brecha ('83),who divided the IPL into five laminae of equal heights. Laminae numbers are written on each figure in the middle of each lamina. Rod outer segment labeling with anti-rhodopsin was performed on the different species tested and all gave positive results. As an example, frog outer segment labeling is shown in Figure 2. The frequency of positive cells for calmodulin and calbindin-D28k is summarized in Table I. Schematic diagrams summarizing the calmodulin and calbindin immunoreactivities in retinae of the different species are presented in Figures 3 and 4. Detailed results from each species are described below.
Goldfish (8 retinae tested) (see Fig. 5). The photoreceptor layer is negative even for rod outer segments. In the inner nuclear layer (INL), horizontal cells and some bipolar cells are calmodulin-IR (Fig. 5b). Numerous amacrine cells are calmodulin-IR (Fig. 5a), and the large pyriform, bistratified amacrine cells localized in the most vitreal part of the INL are the most intensely stained (Fig. 5a). Dendritic ramifications of those cells extend into laminae 1and 3 of the IPL. In addition to the staining seen in laminae 1 and 3, lamina 2 appears diffusely labeled. Numerous ganglion cells are calmodulin-IR (Fig. 5a), and the fiber layer is also calmodulin-positive. As summarized in Figure 4, calbindin immunoreactivity in the goldfish retina is much less intense than calmodulin immunoreactivity. However, some amacrine and ganglion cells are calbindin positive and there is only one weak positive band in the middle of the IPL (data not shown).
CALMODULIN AND CALBINDIN IN RETINA
755
through laminae 3 and 4. The labeling is more intense in the middle part of this broad band. Numerous ganglion cells and the fiber layer are intensely calmodulin-IR (Fig. 6b). A comparative analysis of the calbindin-D28k immunoreactivity in the peripheral part of frog retina is shown in Figure 7. Many photoreceptors are faintly calbindin-IR. This labeling is not restricted to the nucleus or the cell body but is present in all the length of the photoreceptor, suggesting that the labeled photoreceptors are probably cones. Horizontal cells are weakly calbindin-IR. A comparison of the calbindin (Fig. 7) and calmodulin (Fig. 6b) staining shows that intensely labeled bipolar cells are more numerous with calbindin antibody than with calmodulin antibody, while amacrine cells, ganglion cells and fibers are less immunoreactive for calbindin than for calmodulin. The pattern of IPL labeling is also different between the two antibodies: with calbindin-D28k, labeling appears more diffuse with no clearly defined immunoreactive bands in the IPL (Fig. 7).
Turtle (5 retinae tested) (see Fig. 8)
Fig. 7. Section of frog retina (peripheral part) treated with anticalbindin (CaBP) at 116,000 dilution. Compared with a n t i - c m labeling (Fig. 61, labeling of photoreceptor cell bodies and horizontal cells appears the same, but more bipolar cells and fewer amacrine and ganglion cells are CaBP-IR. The fiber layer is less CaBP-IR and the IPL is diffusely CaBP-IR, without well-delineated bands, Scale bar = 25 Fm.
Frog (4 retinae tested) (see Figs. 6,7) In the photoreceptor layer, some nuclei and a few cell bodies are weakly calmodulin-IR (Fig. 6a). This labeling disappeared when the antibody was preincubated with purified calmodulin (not shown). No positive outer segments could be detected. Most of the labeled nuclei belong to photoreceptors; displaced bipolar cells are rare in frog outer nuclear layer (ONL). A few horizontal and some bipolar cells are calmodulin-IR, with the bipolar cell labeling being more intense at the junction with photoreceptors (Fig. 6b). Calmodulin-positive amacrine cells are numerous and intensely labeled (Fig. 6a,b), and they ramify in two bands of immunoreactivity in the IPL. One band is located within lamina 1and the second band is broader and extends
Figure 8 shows staining of turtle retinae with anticalmodulin (Fig. 8a) or anti-calbindin (Fig. 8b). Weak calmodulin immunoreactivity in some nuclei appears on both sides of the OPL (Fig. 8a). The location of the nuclei makes them good candidates for bipolar and displaced bipolar cells known to be numerous in turtle ONL. Their weak staining is close to background, and therefore photoreceptor and bipolar cells are probably calmodulin negative. The weak labeling observed in the outer limiting membrane (OLM) is probably also due to background. The outer segments of rods are calmodulin-negative. Horizontal cells are weakly calmodulin-IR. Numerous amacrine cells are calmodulin-IR, and they ramify in two immunoreactive bands located in laminae 1 and 3 of the IPL. Numerous ganglion cells and the fiber layer are strongly calmodulinIR. A comparison of the calbindin staining (Fig. 8b) shows that in the photoreceptor layer, cones are calbindimD28k positive in some parts of the retina, although exhaustive mapping has not been done. Labeling of horizontal cells is more intense with calbindin than with calmodulin, and some bipolar cells are calbindin-IR. On the other hand, Calbindin-IR amacrine, displaced amacrine, and ganglion although fewer than c a l m o d u l i n - ~ cells ~ cells are (Fig. 8b). For calbindin (Fig. 8b), the IPL stratification is somewhat more diffusethan the two calmodulin-IR bands seen in Figure 8a.
Mouse (6 retinae tested) (see Fig. 9) Photoreceptors, rod outer segments included, are calmodulin-negative. Horizontal cells are also calmodulin-negative or so weakly labeled that they are masked by the intense calmodulin labeling of bipolar cells (Fig. 9a). As with the previous species, numerous amacrine, ganglion cells, and the fiber layer are calmodulin-IR. Amacrine cells ramify in two immunoreactive bands within laminae 2 and 3 of the IPL, whereas a dotted calmodulin staining pattern is present within laminae 4 and 5 (Fig. 9a). By comparison, photoreceptors are also calbindin-negative, but with this antibody, horizontal cells are intensely labeled and bipolar cells are negative (Fig. 9b). Amacrine and ganglion calbindin-IR cells are less numerous than calmodulin-IR ones. Fibers are less strongly labeled with
R. POCHET ET AL.
756
Fig. 8. Sections of turtle retina treated with anti-calmodulin (CAM) at 0.4 Kgiml (a)and anti-calbindin (CaBP) at 116,000 dilution (b).Only some ONL cell bodies appear weakly stained for calmodulin, whereas cones are clearly CaBP-IR. Horizontal cells (h) are more strongly labeled with anti-CaBP than with anti-CAM, and bipolar cells appear
only CaBP-IR. CAM-IR amacrine, displaced amacrine and ganglion cells are more numerous than CaBP-IR ones. CAM-IR bands in the IPL are more well-defined and fibers are more strongly labeled for CAM than for CaBP. Scale bars = 25 Km.
calbindin than with calmodulin, and stratification of immunoreactive bands in the IPL is different.
glion cells are also calmodulin-IR, and the IPL and fiber layer are intensely and diffusely labeled. A comparison of calbindin-D28k immunoreactivity in monkey retinae (Fig. 11) demonstrates that cones are calbindin-positive (except in the fovea). Horizontal cells are more strongly labeled with calbindin than with calmodulin, whereas calbindin-IR bipolar and amacrine cells are less numerous than calmodulin-IR ones. The majority of ganglion cells and the fiber layer are intensely labeled with both antibodies. CalbindimD28k labeling of the IPL is less homogenous than the calmodulin labeling.
Monkey ( 6 retinae tested) (see Figs. 1 0 , l l ) Photoreceptors, rod outer segments included, are calmodulin-negative. Horizontal cells are weakly calmodulin-IR, and bipolar cells are calmodulin-IR (Fig. 10). In the OPL, calmodulin labeling is intense at the junction between photoreceptors and bipolar cells (Fig. 10). Because photoreceptors are calmodulin-negative and many bipolar cells (pericarya and dendrites) are calmodulin-positive, we assume that labeled junctions belong to bipolar cells and are postsynaptic, although they resemble photoreceptor synaptic terminals. A definitive answer must await analysis at the electron microscopic level. Amacrine and numerous gan-
Pigeon (7 retinae tested) (see Figs. 12,13) In pigeon retina, the microvillous processes from retinal pigment epithelium are weakly stained for calmodulin (Fig.
CALMODULIN AND CALBINDIN IN RETINA
757
Fig. 9. Sections of mouse retina treated with anti-calmodulin (CAM) at 0.4 pgiml (a)and anti-calbindin (CaBP) at 1/6,000 dilution (b).Photoreceptors are negative with both antibodies, horizontal cells (h) appear to be labeled only with anti-CaBP, and bipolar cells appear
only CAM-IR. CAM-IR amacrine and ganglion cells are more numerous than CaBP-IR cells. The stratification patterns of the immunoreactive bands in the IPL are also different. Scale bars = 25 pm.
12). This may be due to some calmodulin immunoreactivity or to pigment granules. Horizontal, bipolar, and amacrine cells are calmodulin-IR (Fig. 12). The IPL shows diffuse calmodulin staining, with two more intensely labeled bands of immunoreactivity within laminae 1 and 2, and one broader band in laminae 3 and 4. Ganglion cells and the fiber layer are again calmodulin-IR. The same structures appear labeled in both yellow and red fields (Fig. 121, but the intensity of labeling is weaker in red field than in yellow field in all the layers, especially for amacrine and ganglion cells which are the most strongly labeled cells. Pigeon retina calbindin-D28k labeling was described in detail in a previous publication (Pasteels et al., '87b). However, for comparison, calbindin staining in pigeon retina is shown in Figure 13. While photoreceptors are calmodulin-negative (Fig. 12), some yellow field cones are calbindin-IR (Fig. 13). In yellow field, horizontal cells are labeled more strongly with calbindin-D28k than with calmodulin. Calbindin-IR bipolar cells are less numerous but labeled more intensely than calmodulin-IR bipolar cells. The number of calmodulin-IR and calbindin-IR amacrine cells appears to be similar, while ganglion cells and the fiber
layer are more numerous and more strongly labeled for calmodulin than for calbindin-D28k. The staining patterns of immunoreactive bands in the IPL are also different for the two antibodies (compare Figs. 12 and 13). Finally, as described in Methods, our experiments included control sections treated with antibody that had been pre-incubated with antigen. Under these conditions and as exemplified in Fig. 14 for staining of goldfish and pigeon retinae with pre-absorbed calmodulin antibody, no specific labeling was observed.
CONCLUSIONS Our results show that calmodulin distribution in adult retina is highly conserved among six different species belonging to five phylogenetic classes. They also show that not every cell expresses calmodulin, as measured by immunohistochemistry. Organization of the retina in layers containing different cell types offered a good model allowing the demonstration not only of selective expression of calmodulin among cell types, but also differential expression among the same cell types in different species.
758
Fig. 10. Section of monkey retina treated with anti-calmodulin (CAM) at 0.4 pgiml. Photoreceptors are CAM-negative. Horizontal, bipolar (b),amacrine, and ganglion cells are CAM-IR. The IPL and fiber layer show a diffuse, intense CAM imrnunoreactivity. Scale bar = 25 +m.
In photoreceptors, rod outer segments are negative for calmodulin. This was observed in all the species investigated, even in cat retina (data not shown). Calmodulin could only be detected weakly in some nuclei and in a few cell bodies from frog and turtle photoreceptors. In bipolar cell layers, two modulations of calmodulin expression are clearly seen. First, bipolar cells from all species, except possibly turtles, are calmodulin-IR. Second, the ratio of calmodulin-IR cells versus total bipolar cells is variable from one species to the other. For example, in turtle retina bipolar cells do not appear to be calmodulin-IR, in frog and goldfish a few bipolar cells are calmodulin-IR, in monkey many bipolar cells are calmodulin-IR, and in mouse many if
R. POCHET ET AL.
Fig. 11. Section of monkey retina treated with anti-calbindin (CaBP). Compared with anti-CAM labeling (Fig. lo), the cones in the photoreceptors are only calbindin-positive, whereas horizontal, bipolar, arnacrine, and ganglion cells are positive for both CAM and CaBP. CaBP immunoreactivity in the IPL is more scattered and less intense than the CAM staining pattern. Scale bar = 25 pm.
not all bipolar cells are calmodulin-IR. In bipolar cells from frog and monkey, calmodulin could clearly be detected in dendrites and accumulated at the level of photoreceptor junctions. Calmodulin accumulation at postsynaptic densities was already described by Lin et al. ('801, Wood et al. ('80a) and Caceres et al. ('83). In contrast with some calmodulin variability observed in bipolar cells throughout species, the calmodulin-IR amacrine and ganglion cells are remarkably constant, and numerous cells of both types are intensely labeled. The fiber layer was also always calmodulinIR. Stratifications of immunoreactivity in the IPL are variable, but in most species there are two immunoreactive bands within laminae 1and 3. The present results on the morphological localization of calmodulin in retinae are in general agreement with earlier studies on mouse (Seto-Ohshimaet al., '87) and rat retinae (Wood et al., '80b). However, our results differ from those of
759
CALMODULIN AND CALBINDIN IN RETINA
Fig. 12. Yellow field (YF) and red field (RF) section of pigeon retina treated with anti-calmodulin (CAM) at 0.4 Fgiml. Photoreceptors are CAM-negative. Horizontal, bipolar, amacrine, and ganglion cells are labeled, as well as the fiber layer and immunoreactive bands in the IPL. In red field, labeling is less intense in the IPL and GC. Scale bars = 25 bm.
Wakakura and Yamamoto (’87) who described calmodulin immunoreactivity in cat rod outer and inner segments. We have no definitive explanation for this discrepancy. However, in that work no controls were presented showing that rod outer and inner segment labeling could be abolished when anti-calmodulin antibody was preincubated with purified calmodulin. Therefore, it cannot be excluded that the labeling of rod outer segments (ROS) was a non-specific effect (“effet de bord”). Because of our negative results in rod outer segments, we tested whether some positive ROS immunoreactivity could indeed be seen. As expected, antirhodopsin antibody labeled ROS in all investigated species. An additional control was added in this work: we tested four different anti-calmodulin antibodies; one was prepared against a synthetic peptide of the COOH terminus of calmodulin (Van Eldik et al., ’83), two were produced against performic acid-oxidized calmodulin (Van Eldik and Watterson, ’81),and the fourth against bovine brain calmodulin (Kitajima et al., ’83).Thus, these polyclonal antibod-
ies probably recognize a variety of epitopes and would detect calmodulin molecules even if some epitopes were masked or unavailable to antibody. Our results also differ somewhat from two previous papers (Kohnken et al., ’81; Nagao et al., ’87) reporting that calmodulin is present in ROS, as determined by radioimmunoassay. Those investigators measured calmodulin levels in purified toad and bovine ROS and found approximately 450 ng CaM/mg rod protein. This amount of calmodulin is low, representingonly 0.05% of the rod protein. Therefore, it is possible that these low levels of calmodulin could represent a contaminant copurified with the ROS preparation, or that the sensitivity of the immunohistochemical technique used here is not sufficient to allow detection of these low amounts of calmodulin. Because four different anti-calmodulin antibodies gave similar results on retinae from six different species, we are confident that the immunohistochemical localization we detect does represent specific calmodulin distribution. We
-
760
R. POCHET ET AL.
Fig. 13. Yellow field (YF) and red field (RF) section of pigeon retina treated with anti-calbindin. Double straight cones are positive only in yellow field. Compared with anti-CAM labeling (Fig. 12) in yellow field, horizontal cells are more strongly labeled with anti-calbindin. Ganglion
cells and the fiber layer are more strongly labeled for calmodulin. The stratification patterns in the IPL are also different for the two antibodies. Scale bars = 25 bm.
thus conclude that calmodulin is probably not involved directly in the mechanism of phototransduction. Besides the data discussed above on photoreceptors and calmodulin, few reports have appeared on calmodulin localization or activity in retina. A review of the recent literature indicated that among retinal cells and fiber layers, only ganglion cells have been investigated somewhat. For instance, in mouse retinal ganglion cells, calcium and calmodulin added to neurofilament preparations in the presence of heparin modestly stimulates phosphorylation of two neurofilaments subunits (Sihag and Nixon, '89). However, no direct evidence is provided for an in vivo mode of action of calmodulin in those cells. One interesting feature emerging from our study is the presence of calmodulin in many IPL fibers, where neuromodulin (p57, GAP-43, F1, B-50), a major neural-specific calmodulin binding protein, is almost exclusively located (J.J.Norden, personal communication). Neuromodulin has been postulated to play a role in regulation of free calmodulin levels in neurons by binding and concentrating calmodulin at specific sites in neurons, where protein kinase C-catalyzed phosphorylation of neuromodulin could result in a
local release of calmodulin and subsequent activation of calmodulin regulated enzymes (Alexander et al., '87; Estep et al., '90). Whether calmodulin and neuromodulin interact with each other in the IPL of the retina remains to be demonstrated. However, our findings are clearly consistent with this possibility. Calbindin-D28k, which belongs t o the same EF-hand superfamily of proteins as calmodulin and whose function is still conjectural, has been well studied for its distribution (for a review see Christakos et al., '89) especially in brain (Baimbridge et al., '82; Celio, '90; Feldman and Christakos, '83; Garcia-Segura et al., '84; Jande et al., '81; Sloviter, '89) and retina (Pasteels et al., '87b, '90; Rabie et al., '85; Rohrenbeck et al., '87, '89; Schreiner et al., '85; Verstappen et al., '86). In contrast to calmodulin, calbindin distribution in retina is more variable among species. For instance, calbindin is present in cones only in some well-defined subtypes of pigeon cones (Pasteels et al., '87b), whereas in monkey all cones except those of the fovea are calbindin positive (Pasteels et al., '90). A high variability in the pattern of calbindin labeling is also seen in the IPL from the investigated species. However, some aspects of calbindin
761
CALMODULIN AND CALBINDIN IN RETINA
Fig. 14. Immunohistochemical controls used with goldfish retina section (a)and pigeon retina section
(b).Anti-calmodulin (0.4 Wgiml) was preincubated for 24 hours with 6 pgiml calmodulin before immunoperoxidase staining. No significant labeling could he seen. By comparison, note the contrast obtained with pigment epithelium. Scale bars = 25 Wm.
TABLE 1. Frequency of Positive Cells'
Frog
Goldfish
CAM
CaBP
CAM
~~
Photoreceptor Horizontal cell Bipolar cell Arnacrine cell Ganglion cell
~
+++ ++
-
++2
Mouse
Turtle CAM
CaBP
+++
+
CaBP
Pigeon
Monkey
CAM
CaBP
CAM
-
-
-
CaBP
CAM
CaBP
~~
-
+ ++++ ++++
+++3
+
+++R
-
+++.I
++ +++ +++ ++ +++ ++ +++ +++ +++ ++ +++ ++ 'For each species, this table gives a n estimate of the No. of positive cells according to peroxidase staining for calrnodulin (CAM) and calbindin (CaBP),as follows: + (few), + + (some), + + + (many), + + + + (nearly all). See text and Figures 3 and 4 for relative intensities of staining.
+ ++ +++ +++
+++ ++ +++
++ ++ +++ +++
-/+
+++ +++ ++++
+++ ++ +++
+++ +++ ++++
++ ++ ++ ++++
'Only nuclei and a few cell bodies. 'Only cones and only in some parts of the retina.
distribution are conserved among species, allowing comparison with calmodulin. For example, horizontal cells are always more strongly labeled for calbindin than for calmodulin, whereas amacrine and ganglion cells are more strongly labeled and often more numerous for calmodulin than for calbindin. In conclusion, in addition to the high conservation of calmodulin sequence throughout evolution another level of conservation is observed, which is the conservation of calmodulin distribution among different cell types. However, calmodulin expression is not constitutive in the sense that not every cell expresses calmodulin to the same extent.
We are still far from a precise knowledge of why calmodulin is expressed within one particular neuron or group of neurons and not within its neighbor. However, we believe that detailed and systematic immunohistochemical and biochemical observations are leading to substantial progress in our understanding of the functions of calmodulin and calbindin-D28k in retinal information processing.
ACKNOWLEDGMENTS We thank Georgette Pattyn and Leon Surardt for excellent technical assistance, Paulette Miroir for typing and
R. POCHET ET AL.
762
Mrs. Lotteau for drawings. This work was supported by FRSM (grant No. 3.4511.88)and by funds from the Cystic Fibrosis Foundation.
LITERATURE CITED Alexander, K.A., B.M. Cimler, K.E. Meier, and D.R. Storm (1987) Regulation of calmodulin binding to P-57. A neurospecific calmodulin binding protein. J. Biol. Chem. 262:6108-6113. Baimbridge, K.G., J.J. Miller, and C.O. Parkes (1982) Calcium-binding protein distribution in the rat brain. Brain Res. 239:519-525. Brecha, N. (1983) Retinal neurotransmitters: histochemical and biochemical studies. In P.C. Emson (ed): Chemical Neuroanatomy. New York: Raven Press, pp. 85-129. Caceres, A., P. Bender, L. Smavely, L.J. Rebhun, and 0. Steward (1983) Distribution and subcellular localization of calmodulin in adult and developing brain tissue. Neuroscience 10:449-461. Celio, M.R. (1990) Calbindin D-28k and parvalbumin in the rat nervous system. Neuroscience 35:373-475. Christakos, S., C. Gabrielides, and W.B. Rhoten (1989)Vitamin D-dependent Calcium binding proteins: Chemistry, distribution, functional considerations and molecular biology. Endocr. Rev. 10:3-26. Cohen, P., and C.B. B e e (1988) Calmodulin. Amsterdam: Elsevier. Estep, R.P., K.A. Alexander, and D.R. Storm (1990) Regulation of free calmodulin levels in neurons by neuromodulin: relationship to neuronal growth and regeneration. Curr. Top. Cell Regul. 31r161-180. Feldman, S.C., and Christakos, S. (1983) Vitamin-D-dependent calciumbinding protein in rat brain: biochemical and immunocytochemical characterization. Endocrinology 112:290-302. Garcia-Segura, L.M., D. Baetens, J. Roth, A.W. Norman, and L. Orci (1984) Immunohistochemical mapping of calcium-binding protein immunoreactivity in the rat central nervous system. Brain Res. 296r75-86. Jande, S.S., L. Maler, and D.E.M. Lawson (1981) Immunohistochemical mapping of vitamin D-dependent calcium-binding protein in brain. Nature 294:765-767. Kitajima, S., A. Seto-Ohshima, M. Sano, and K. Kato (1983) Production of antibodies to calmodulin in rabbits and enzyme immunoassays for calmodulin and anti-calmodulin. J. Biochem. (Tokyo) 94:559-564. Kohnken, R.E., J.G. Chafouleas, D.M. Eadie, A.R. Means, and D.G. Me Connell(1981) Calmodulin in bovine rod outer segments. J. Biol. Chem. 256:12517-12522. Lin, C.T., J.R. Dedman, B.R. Brinkley, and A.R. Means (1980) Localization of calmodulin in rat cerebellum by immunoelectron microscopy. J. Cell Biol. 85t473-480. Means, A.R. (1988) Molecular mechanisms of action of calmodulin. Recent Prog. Horm. Res. 44:223-262. Means, A.R., and J.R. Dedman (1980)Calmodulin: an intracellular calcium receptor. Nature 28573-77. Nagao, S.,A. Yamazaki, and M.W. Bitensky (1987) Calmodulin and calmodulin binding proteins in amphibian rod outer segments. Biochemistry 26:1659-1665. Pasteels, B., M. Miki, S. Hatakenaka, and R. Pochet (1987a) Immunohistochemical cross-reactivity and electrophoretic comigration between calbiiidin-D27kDa and visinin. Brain Res. 422:107-113. Pasteels, B., M. Parmentier, D.E.M. Lawson, A. Verstappen, and R. Pochet (198713) Calcium binding protein immunoreactivity in pigeon retina. Invest. Ophthalmol. Vis. Sci. 28:658-664. Pasteels, B., J. Rogers, F. Blachier, and R. Pochet (1990) Calbindin and calretinin localization in retina from different species. Vis. Neurosci. 5:l-16. Pochet, R., G.D. Pipeleers, and W.J. Malaise (1987) Calbindin-D27kDa:
preferential localization in non-p-islet cells of the rat pancreas. Biol. Cell 61:155-161. Rabie, A,, M. Thomasset, C.O. Parkes, and M.C. Clavel (1985) Immunocytochemical detection of 28000-MW calcium-binding protein in horizontal cells of the rat retina. Cell Tissue Res. 240:493-496. Rohrenbeck, J., H. Wassle, and C. Heizmann (1987) Immunocytochemical laheling of horizontal cells in mammalian retina using antibodies against calcium-binding proteins. Neurosci. Lett. 77:255-260. Rohrenbeck, J., H. Wassle, and B.B. Boycott (1989) Horizontal cells in the monkey retina: Immunocytochemical staining with antibodies against calcium binding proteins. Eur. J. Neurosci. lt407-420. Sano, M., A. Seto-Ohshima, S. Kitajima, and K. Kato (1987) Immunoperoxidase staining of S-100 and calrnodulin in tissue sections and cultured cells. In A.R. Means, and P.M. Conn (eds):Methods in Enzymology, Vol. 139. New York Academic Press, pp. 495-504. Schreiner, D.S., S.S. Jande, and D.E.M. Lawson (1985) Target cells of vitamin D in the retina. Acta Anat. (Basel) 121:153-162. Seto-Ohshima, A., Y. Yamazaki, N. Kawarnura, M. Sano, S. Kitajima, and A. Mizutani (1987) The early expression of immunoreactivity for calmodulin in the nervous system of mouse embryos. Histochemistry 86:337343. Sihag, R.K., and R.A. Nixon (1989) In vivo phosphorylation of distinct domains of the 70-kilo-dalton neurofilament subunit involves different protein kinases. J. Biol. Chem. 2641457-464. Sloviter, R.S. (1989)Calcium-binding protein (Calbindin-D28K)and parvalbumin immunocytochemistry: Localization in the rat hippocampus with specific reference to the selective vulnerability of hyppocampal neurons to seizure activity. J. Comp. Neurol. 280:183-196. Spencer, R., M. Charman, J.S. Emtage, and D.E.M. Lawson (1976) Production and properties of vitamin-D-induced mRNA for chick calciumbinding protein. Eur. J. Biochem. 71:399-409. Szi.1, A., L. Takacs, E. Monostori, T. Diamantstein, I. Vigh-Teichmann, and P. Rohlich (1986)Monoclonal antibody recognizing cone visual pigment. Exp. Eye Res. 43:871-883. Vacca, L.L., S.J. Abrahams, and N.E. Naftchi (1980) A modified peroxidaseantiperoxidase procedure for improved localization of substance P in rat spinal cord. J. Histochem. Cytochem. 28997-304. Van Eldik, L.J., and D.M. Watterson (1981) Reproducible production of antiserum against vertebrate calrnodulin and determination of the immunoreactive site. J. Biol. Chem. 256t4205-4210. Van Eldik, L.J., and D.M. Roberts (1988) Calcium modulated proteins in pathophysiology. In M.P. Thompson (ed): Calcium Binding Proteins. Boca Raton, FL: CRC Press, pp. 59-76. Van Eldik, L.J., J.G. Zendegui, D.R. Marshak, and D.M. Watterson (1982) Calcium-binding proteins and the molecular basis of calcium action. Int. Rev. Cytol. 77:l-61. Van Eldik, L.J., K.-F. Fok, B.W. Erickson, and D.M. Watterson (1983) Engineering of site-directed anrisera against vertebrate calmodulin by using synthetic peptide imrnunogens containing an irnmunoreactive site. Proc. Natl. Acad. Sci. U.S.A. 80:6775-6779. Verstappen, A,, M. Parmentier, M. Chirnoaga, D.E.M. Lawson, J.L. Pasteels, and R. Pochet (1986) Vitamin D-dependent calcium binding protein immunoreactivity in human retina. Ophthalmic Res. 18:209-214. Wakakura, M., and N. Yamamoto (1987) Immunological localization of calmodulin in feline rod outer segments. Exp. Eye Res. 44:451-458. Wood, J.G., R.W. Wallace, J.N. Whitaker, and W.Y. Cheung (1980a) Immunocytochemical localization of calmodulin and a heat-labile calmodulinbinding protein (CaM-BP80)in basal ganglia of mouse brain. J. Cell Biol. 84.66-76. Wood, J.G., R.W. Wallace, J.N. Wbitaker, and W.Y. Cheung (1980b) Immunocytochemical localization of calmodulin in reaons of rodent brain. Ann. N.Y. Acad. Sci. 356:75-82.
Calmodulin and Calbindin Localization in Retina From Six Vertebrate Species ROLAND POCHET, BRIGITTE PASTEELS, AKIKO SETO-OHSHIMA, ENRICO BASTIANELLI, SATOKO KITAJIMA, AND LINDA J. VAN ELDIK Laboratoire d'Histologie, Facult6 de MQdecine,Universitk Libre de Bruxelles, B- 1070 Bruxelles, Belgium (R.P.,B.P.,E.B.); Laboratoire de Biologie Cellulaire, Facult6 des Sciences, Universite de Poitiers, 86022 Poitiers, France (R.P.); Institute for Developmental Research, Aichi Prefectural Colony, Aichi 480-03, Japan (AS.-O.,S.K.);Departments of Pharmacology and Cell Biology, Vanderbilt University, Nashville, Tennessee 37232-6600 (L.J.V.E.)
ABSTRACT Calmodulin is abundant in the central nervous system, including the retina. However, the localization of calmodulin in the retina has not been described in detail. We therefore decided to investigate calmodulin localization in retinae from six vertebrate species, by using immunohistochemical labeling with four different rabbit polyclonal antibodies against calmodulin. The localization of calbindin-D28k, another calcium-binding protein already well described in retina, was compared. We found that calmodulin distribution is more highly conserved among species, contrasting with calbindin variability. The most striking result emerging is that calmodulin could not be detected in photoreceptors although other layers are intensely calmodulin-immunoreactive, casting doubt about a direct role of calmodulin in phototransduction. Horizontal cells are weakly calmodulin-immunoreactive, bipolar cells are calmodulinimmunoreactive except in turtle retina, numerous amacrine and ganglion cells are labeled in all species, and the fiber layer is always labeled. These data demonstrate that, while the calmodulin distribution in retina is similar among vertebrate species, selective differences in localization can be detected not only among the same cell types in different species but also among different cell types in the same species. The results showing differences in calmodulin immunoreactivity among cell types also provide further evidence that calmodulin expression in eukaryotes is not constitutive, in the sense that not every cell expresses similar levels of calmodulin. Key words: calcium-bindingproteins, immunohistochemistry,photoreceptors
Calmodulin and calbindin D-28k are two calcium binding proteins having common Ca2+-bindingmotifs. Calmodulin is an intracellular Ca" receptor and is highly conserved at the primary amino acid sequence level. In addition, calmodulin serves as the obligatory Ca'+-dependent activator of a variety of enzymes and is associated with several intracellular structural proteins (for reviews see Cohen and Klee, '88; Means, '88; Means and Dedman, '80; Van Eldik and Roberts, '88; Van Eldik et al., '82). Although it is generally accepted that calmodulin is found in all eukaryotes, its levels and localization can vary in different cell types and species. In retina, calmodulin has been investigated biochemically (Kohnken et al., '81; Nagao et al., '871, but to date there has not been a detailed description of calmodulin-containing neurons, although their presence is mentioned by Wood et al. ('80b), studying immunocytochemically the distribution of calmodulin in different rat brain regions; by SetoOhshima et al. ('871, studying mouse embryos; and by Wakakura and Yamamoto ('87), mentioning the presence of 0 1991
WILEY-LISS. INC.
calmodulin in feline rod outer segments. The two biochemical manuscripts focused their studies on rod outer segments while the three morphological papers examined rat, mouse, and cat retina layers. In rat retina, Wood et al. ('80b) showed calmodulin immunoreactivity in the cytoplasm of ganglion cells as well as in fibers from amacrine and ganglion cells of the inner plexiform layer. No photoreceptors were positive for calmodulin. In adult mouse (SetoOhshima et al., '87), the ganglion cell layer, some somata in the inner nuclear layer, and fibers in the inner plexiform layer were calmodulin-immunoreactive (IR). Again no photoreceptors were positive for calmodulin. In feline retina, Wakakura and Yamamoto ('87) reported that in addition to the nerve-fiber layer, the ganglion cell layer, and the inner plexiform layer, rod inner- and outer-segments were also calmodulin-IR. Some discrepancies thus seem to exist about the presence of calmodulin in photoreceptors.
Accepted September 20, 1991.
751
CALMODULIN AND CALBINDIN IN RETINA
Fig. 1. Section of the same mouse retina treated with four anti-calmodulin antibodies: antibody #465 in a, #860 in b, #009 in c and #449 in d. #009 was used at 0.4 pgiml; #465, #449, and #860 at 112,000 dilution. Scale bar = 25 pm.
We therefore decided to investigate in detail and compare in different species the distribution of calmodulin-IR neurons in the retina. We selected retinae from six vertebrate species that are widely studied in many laboratoriesnamely, goldfish (fish), turtle (reptile), frog (amphibian), pigeon (avian),mouse and monkey (mammalian).CalbindinD28k localization in retinae from those species has been included in this work for two purposes: 1)to extend studies performed on human (Verstappen et al., '861, rabbit (Schreiner et al., '85), rat (Rabi6et al., '851, pigeon (Pasteels
Abbreviations
b CaBP CAM For FL GC h
INL IPL IR OLM ONL OPL 1-5 Ph RF ROS
YF
bipolar cells calbindin calmodulin fiber layer ganglion cell layer horizontal cells inner nuclear layer inner plexiform layer immunoreactive outer limitingmembrane outer nuclear layer outer plexiform layer laminae of the inner plexiform layer photoreceptor layer red field rod outer segment yellow field
Fig. 2. Section of frog retina treated with anti-rhodopsin (A01 at l/1,000 dilution. Photoreceptor outer segments (as) are AO-IR, whereas they are CAM-negative (see Fig. 4). is: photoreceptor inner segments. Scale bar = 25 pm.
3 9
s v E
ldl
Z L
lNI 1dO 1NO
cld
J
9
ldl
1NI 1dO 1NO
4d
*WJ 3 J3H30d *?I
753
CALMODULIN AND CALBINDIN IN RETINA
Fig. 5. Sections of goldfish retina treated with anti-calmodulin (CAM) at 0.4 Fg/ml. The section in b was situated in a more peripheral part of the retina than that in a. Photoreceptors are clearly CAMnegative. CAM-positive horizontal and bipolar cells are seen well in b,
whereas CAM-positiveamacrine cells, ganglion cells and three immunoreactive bands in the IPL appear more clearly in a. A large bistratified amacrine cell showing intense labeling is indicated in a by an arrow. Scale bars = 25 um.
and named #449, #465, #860. Antibodies #449 and #465 were produced against performic acid-oxidized, vertebrate calmodulin, and #860 was made against a synthetic peptide of the COOH terminus (residues 134-148) of vertebrate calmodulin. Purified IgG fractions were routinely used at two dilutions ( l / l , O O O and 1/2,000). The fourth antibody used was produced against bovine brain calmodulin as described earlier (Kitajima et al., '83; Sano et al., '87) and named #009. This antibody was purified by calmodulin affinity chromatography and used at 0.4 kg IgGiml. Controls included antibody preincubated with 6 pg/ml calmodulin (Sigma, St-Louis) for 24 hours before staining. Under these conditions, no labeling was observed (see Fig. 14). Anti-rhodopsin. Preparation of the rat antiserum against purified bovine opsin was described elsewhere (Szel et al., '86). Briefly, opsin was purified from bovine retinae and rats were immunized with a mixture of 1 mg/ml opsin solution and complete Freund adjuvant. Rat anti-rhodopsin serum was used a t l/l,OOO dilution. Anti-calbindin. Rabbit antiserum raised against chick duodenal calbindin-D28k was prepared as described (Spencer et a]., '76), and routinely used at 1/6,000 dilution. Tissue preparation. Retinae from goldfish (Carassius auratus), frogs (Rana esculenta), and turtle (Pseudemis scripta elegans) were dissected after deep anesthesia in cold water containing 1 g/L tricaine (MS-222, Sandoz, Basle). Pigeon (Columba liuia donestica) and mouse (Mus musculus) were deeply anesthetized with ether and monkeys (Cercopithecus aethiops sabaeus) with pentobarbital. After removal of cornea and lens, the eyecups were fixed for 24 hours in Sublimate Bouin Holland containing 4 g cupric
acetate, 4 g picric acid, 10 ml formalin (40% W/V solution), 1g trichloroacetic acid, and 1ml mercuric chloride aqueous saturated solution, in 100 ml distilled water. The specimens were then dehydrated and embedded in paraffin wax. Five micrometer sections were cut for immunohistochemical analyses. Peroxidase immunohistochemistry. Immunoperoxidase staining was done as previously published (Vaccaet al., '80) and modified by us (Pasteels et al., '87a,b; Pochet et al., '87). In brief, routinely dewaxed and hydrated sections were processed as follows: (1) rinse in Coons Verona1 Buffer saline (CVBS); (2) preincubation in 5% normal sheep serum; (3) incubation with rabbit antibodies or rat antirhodopsin for 48 hours at 4°C in a moist chamber; (4) 15-minute incubation with goat anti-rabbit IgG serum (Bio-Yeda, Israel) at 1:80 dilution for anti-calmodulin and anti-calbindin, or biotinylated anti-rat IgG (Vector Laboratories, Burlingame, CA, USA) at 20 pg/ml for antirhodopsin; (5) incubation with soluble rabbit PAP complex 1:300dilution (DAKO, Denmark) or horseradish peroxidaseavidin D at 25 pg/ml (Vector Laboratories, Burlingame, CA, USA); (6) stainingin a citrate phosphate buffer, pH 6.2, containing 0.5 mg/ml 3,3'-diaminobenzidine HC1 (DAB) and 0.1% H,O,.
RESULTS The four different anti-calmodulin antibodies gave similar staining patterns and no discrepancies in the neuronal and fiber labeling could be detected. Minor differences were seen at the background level (see Fig. 1).The calmodulin
754
R. POCHET ET AL.
Fig. 6. Sections of frog retina treated with anti-calmodulin (CAM) at 0.4 pg/ml. The section in a was situated in a more peripheral part of the retina than that in b. In the photoreceptor layer, only some cell
bodies are CAM-IR (a). Horizontal (h),bipolar (b),numerous amacrine and ganglion cells are labeled, as well as the fiber layer and bands in the IPL. Scale bars = 25 km.
affinity purified IgG fraction (#009) prepared by Kitajima gave the lowest background. Therefore, all subsequent figures are photomicrographs from sections immunohistochemically labeled with this antibody. Nomenclature for the laminae of the inner plexiform layer (IPL) was adopted from Brecha ('83),who divided the IPL into five laminae of equal heights. Laminae numbers are written on each figure in the middle of each lamina. Rod outer segment labeling with anti-rhodopsin was performed on the different species tested and all gave positive results. As an example, frog outer segment labeling is shown in Figure 2. The frequency of positive cells for calmodulin and calbindin-D28k is summarized in Table I. Schematic diagrams summarizing the calmodulin and calbindin immunoreactivities in retinae of the different species are presented in Figures 3 and 4. Detailed results from each species are described below.
Goldfish (8 retinae tested) (see Fig. 5). The photoreceptor layer is negative even for rod outer segments. In the inner nuclear layer (INL), horizontal cells and some bipolar cells are calmodulin-IR (Fig. 5b). Numerous amacrine cells are calmodulin-IR (Fig. 5a), and the large pyriform, bistratified amacrine cells localized in the most vitreal part of the INL are the most intensely stained (Fig. 5a). Dendritic ramifications of those cells extend into laminae 1and 3 of the IPL. In addition to the staining seen in laminae 1 and 3, lamina 2 appears diffusely labeled. Numerous ganglion cells are calmodulin-IR (Fig. 5a), and the fiber layer is also calmodulin-positive. As summarized in Figure 4, calbindin immunoreactivity in the goldfish retina is much less intense than calmodulin immunoreactivity. However, some amacrine and ganglion cells are calbindin positive and there is only one weak positive band in the middle of the IPL (data not shown).
CALMODULIN AND CALBINDIN IN RETINA
755
through laminae 3 and 4. The labeling is more intense in the middle part of this broad band. Numerous ganglion cells and the fiber layer are intensely calmodulin-IR (Fig. 6b). A comparative analysis of the calbindin-D28k immunoreactivity in the peripheral part of frog retina is shown in Figure 7. Many photoreceptors are faintly calbindin-IR. This labeling is not restricted to the nucleus or the cell body but is present in all the length of the photoreceptor, suggesting that the labeled photoreceptors are probably cones. Horizontal cells are weakly calbindin-IR. A comparison of the calbindin (Fig. 7) and calmodulin (Fig. 6b) staining shows that intensely labeled bipolar cells are more numerous with calbindin antibody than with calmodulin antibody, while amacrine cells, ganglion cells and fibers are less immunoreactive for calbindin than for calmodulin. The pattern of IPL labeling is also different between the two antibodies: with calbindin-D28k, labeling appears more diffuse with no clearly defined immunoreactive bands in the IPL (Fig. 7).
Turtle (5 retinae tested) (see Fig. 8)
Fig. 7. Section of frog retina (peripheral part) treated with anticalbindin (CaBP) at 116,000 dilution. Compared with a n t i - c m labeling (Fig. 61, labeling of photoreceptor cell bodies and horizontal cells appears the same, but more bipolar cells and fewer amacrine and ganglion cells are CaBP-IR. The fiber layer is less CaBP-IR and the IPL is diffusely CaBP-IR, without well-delineated bands, Scale bar = 25 Fm.
Frog (4 retinae tested) (see Figs. 6,7) In the photoreceptor layer, some nuclei and a few cell bodies are weakly calmodulin-IR (Fig. 6a). This labeling disappeared when the antibody was preincubated with purified calmodulin (not shown). No positive outer segments could be detected. Most of the labeled nuclei belong to photoreceptors; displaced bipolar cells are rare in frog outer nuclear layer (ONL). A few horizontal and some bipolar cells are calmodulin-IR, with the bipolar cell labeling being more intense at the junction with photoreceptors (Fig. 6b). Calmodulin-positive amacrine cells are numerous and intensely labeled (Fig. 6a,b), and they ramify in two bands of immunoreactivity in the IPL. One band is located within lamina 1and the second band is broader and extends
Figure 8 shows staining of turtle retinae with anticalmodulin (Fig. 8a) or anti-calbindin (Fig. 8b). Weak calmodulin immunoreactivity in some nuclei appears on both sides of the OPL (Fig. 8a). The location of the nuclei makes them good candidates for bipolar and displaced bipolar cells known to be numerous in turtle ONL. Their weak staining is close to background, and therefore photoreceptor and bipolar cells are probably calmodulin negative. The weak labeling observed in the outer limiting membrane (OLM) is probably also due to background. The outer segments of rods are calmodulin-negative. Horizontal cells are weakly calmodulin-IR. Numerous amacrine cells are calmodulin-IR, and they ramify in two immunoreactive bands located in laminae 1 and 3 of the IPL. Numerous ganglion cells and the fiber layer are strongly calmodulinIR. A comparison of the calbindin staining (Fig. 8b) shows that in the photoreceptor layer, cones are calbindimD28k positive in some parts of the retina, although exhaustive mapping has not been done. Labeling of horizontal cells is more intense with calbindin than with calmodulin, and some bipolar cells are calbindin-IR. On the other hand, Calbindin-IR amacrine, displaced amacrine, and ganglion although fewer than c a l m o d u l i n - ~ cells ~ cells are (Fig. 8b). For calbindin (Fig. 8b), the IPL stratification is somewhat more diffusethan the two calmodulin-IR bands seen in Figure 8a.
Mouse (6 retinae tested) (see Fig. 9) Photoreceptors, rod outer segments included, are calmodulin-negative. Horizontal cells are also calmodulin-negative or so weakly labeled that they are masked by the intense calmodulin labeling of bipolar cells (Fig. 9a). As with the previous species, numerous amacrine, ganglion cells, and the fiber layer are calmodulin-IR. Amacrine cells ramify in two immunoreactive bands within laminae 2 and 3 of the IPL, whereas a dotted calmodulin staining pattern is present within laminae 4 and 5 (Fig. 9a). By comparison, photoreceptors are also calbindin-negative, but with this antibody, horizontal cells are intensely labeled and bipolar cells are negative (Fig. 9b). Amacrine and ganglion calbindin-IR cells are less numerous than calmodulin-IR ones. Fibers are less strongly labeled with
R. POCHET ET AL.
756
Fig. 8. Sections of turtle retina treated with anti-calmodulin (CAM) at 0.4 Kgiml (a)and anti-calbindin (CaBP) at 116,000 dilution (b).Only some ONL cell bodies appear weakly stained for calmodulin, whereas cones are clearly CaBP-IR. Horizontal cells (h) are more strongly labeled with anti-CaBP than with anti-CAM, and bipolar cells appear
only CaBP-IR. CAM-IR amacrine, displaced amacrine and ganglion cells are more numerous than CaBP-IR ones. CAM-IR bands in the IPL are more well-defined and fibers are more strongly labeled for CAM than for CaBP. Scale bars = 25 Km.
calbindin than with calmodulin, and stratification of immunoreactive bands in the IPL is different.
glion cells are also calmodulin-IR, and the IPL and fiber layer are intensely and diffusely labeled. A comparison of calbindin-D28k immunoreactivity in monkey retinae (Fig. 11) demonstrates that cones are calbindin-positive (except in the fovea). Horizontal cells are more strongly labeled with calbindin than with calmodulin, whereas calbindin-IR bipolar and amacrine cells are less numerous than calmodulin-IR ones. The majority of ganglion cells and the fiber layer are intensely labeled with both antibodies. CalbindimD28k labeling of the IPL is less homogenous than the calmodulin labeling.
Monkey ( 6 retinae tested) (see Figs. 1 0 , l l ) Photoreceptors, rod outer segments included, are calmodulin-negative. Horizontal cells are weakly calmodulin-IR, and bipolar cells are calmodulin-IR (Fig. 10). In the OPL, calmodulin labeling is intense at the junction between photoreceptors and bipolar cells (Fig. 10). Because photoreceptors are calmodulin-negative and many bipolar cells (pericarya and dendrites) are calmodulin-positive, we assume that labeled junctions belong to bipolar cells and are postsynaptic, although they resemble photoreceptor synaptic terminals. A definitive answer must await analysis at the electron microscopic level. Amacrine and numerous gan-
Pigeon (7 retinae tested) (see Figs. 12,13) In pigeon retina, the microvillous processes from retinal pigment epithelium are weakly stained for calmodulin (Fig.
CALMODULIN AND CALBINDIN IN RETINA
757
Fig. 9. Sections of mouse retina treated with anti-calmodulin (CAM) at 0.4 pgiml (a)and anti-calbindin (CaBP) at 1/6,000 dilution (b).Photoreceptors are negative with both antibodies, horizontal cells (h) appear to be labeled only with anti-CaBP, and bipolar cells appear
only CAM-IR. CAM-IR amacrine and ganglion cells are more numerous than CaBP-IR cells. The stratification patterns of the immunoreactive bands in the IPL are also different. Scale bars = 25 pm.
12). This may be due to some calmodulin immunoreactivity or to pigment granules. Horizontal, bipolar, and amacrine cells are calmodulin-IR (Fig. 12). The IPL shows diffuse calmodulin staining, with two more intensely labeled bands of immunoreactivity within laminae 1 and 2, and one broader band in laminae 3 and 4. Ganglion cells and the fiber layer are again calmodulin-IR. The same structures appear labeled in both yellow and red fields (Fig. 121, but the intensity of labeling is weaker in red field than in yellow field in all the layers, especially for amacrine and ganglion cells which are the most strongly labeled cells. Pigeon retina calbindin-D28k labeling was described in detail in a previous publication (Pasteels et al., '87b). However, for comparison, calbindin staining in pigeon retina is shown in Figure 13. While photoreceptors are calmodulin-negative (Fig. 12), some yellow field cones are calbindin-IR (Fig. 13). In yellow field, horizontal cells are labeled more strongly with calbindin-D28k than with calmodulin. Calbindin-IR bipolar cells are less numerous but labeled more intensely than calmodulin-IR bipolar cells. The number of calmodulin-IR and calbindin-IR amacrine cells appears to be similar, while ganglion cells and the fiber
layer are more numerous and more strongly labeled for calmodulin than for calbindin-D28k. The staining patterns of immunoreactive bands in the IPL are also different for the two antibodies (compare Figs. 12 and 13). Finally, as described in Methods, our experiments included control sections treated with antibody that had been pre-incubated with antigen. Under these conditions and as exemplified in Fig. 14 for staining of goldfish and pigeon retinae with pre-absorbed calmodulin antibody, no specific labeling was observed.
CONCLUSIONS Our results show that calmodulin distribution in adult retina is highly conserved among six different species belonging to five phylogenetic classes. They also show that not every cell expresses calmodulin, as measured by immunohistochemistry. Organization of the retina in layers containing different cell types offered a good model allowing the demonstration not only of selective expression of calmodulin among cell types, but also differential expression among the same cell types in different species.
758
Fig. 10. Section of monkey retina treated with anti-calmodulin (CAM) at 0.4 pgiml. Photoreceptors are CAM-negative. Horizontal, bipolar (b),amacrine, and ganglion cells are CAM-IR. The IPL and fiber layer show a diffuse, intense CAM imrnunoreactivity. Scale bar = 25 +m.
In photoreceptors, rod outer segments are negative for calmodulin. This was observed in all the species investigated, even in cat retina (data not shown). Calmodulin could only be detected weakly in some nuclei and in a few cell bodies from frog and turtle photoreceptors. In bipolar cell layers, two modulations of calmodulin expression are clearly seen. First, bipolar cells from all species, except possibly turtles, are calmodulin-IR. Second, the ratio of calmodulin-IR cells versus total bipolar cells is variable from one species to the other. For example, in turtle retina bipolar cells do not appear to be calmodulin-IR, in frog and goldfish a few bipolar cells are calmodulin-IR, in monkey many bipolar cells are calmodulin-IR, and in mouse many if
R. POCHET ET AL.
Fig. 11. Section of monkey retina treated with anti-calbindin (CaBP). Compared with anti-CAM labeling (Fig. lo), the cones in the photoreceptors are only calbindin-positive, whereas horizontal, bipolar, arnacrine, and ganglion cells are positive for both CAM and CaBP. CaBP immunoreactivity in the IPL is more scattered and less intense than the CAM staining pattern. Scale bar = 25 pm.
not all bipolar cells are calmodulin-IR. In bipolar cells from frog and monkey, calmodulin could clearly be detected in dendrites and accumulated at the level of photoreceptor junctions. Calmodulin accumulation at postsynaptic densities was already described by Lin et al. ('801, Wood et al. ('80a) and Caceres et al. ('83). In contrast with some calmodulin variability observed in bipolar cells throughout species, the calmodulin-IR amacrine and ganglion cells are remarkably constant, and numerous cells of both types are intensely labeled. The fiber layer was also always calmodulinIR. Stratifications of immunoreactivity in the IPL are variable, but in most species there are two immunoreactive bands within laminae 1and 3. The present results on the morphological localization of calmodulin in retinae are in general agreement with earlier studies on mouse (Seto-Ohshimaet al., '87) and rat retinae (Wood et al., '80b). However, our results differ from those of
759
CALMODULIN AND CALBINDIN IN RETINA
Fig. 12. Yellow field (YF) and red field (RF) section of pigeon retina treated with anti-calmodulin (CAM) at 0.4 Fgiml. Photoreceptors are CAM-negative. Horizontal, bipolar, amacrine, and ganglion cells are labeled, as well as the fiber layer and immunoreactive bands in the IPL. In red field, labeling is less intense in the IPL and GC. Scale bars = 25 bm.
Wakakura and Yamamoto (’87) who described calmodulin immunoreactivity in cat rod outer and inner segments. We have no definitive explanation for this discrepancy. However, in that work no controls were presented showing that rod outer and inner segment labeling could be abolished when anti-calmodulin antibody was preincubated with purified calmodulin. Therefore, it cannot be excluded that the labeling of rod outer segments (ROS) was a non-specific effect (“effet de bord”). Because of our negative results in rod outer segments, we tested whether some positive ROS immunoreactivity could indeed be seen. As expected, antirhodopsin antibody labeled ROS in all investigated species. An additional control was added in this work: we tested four different anti-calmodulin antibodies; one was prepared against a synthetic peptide of the COOH terminus of calmodulin (Van Eldik et al., ’83), two were produced against performic acid-oxidized calmodulin (Van Eldik and Watterson, ’81),and the fourth against bovine brain calmodulin (Kitajima et al., ’83).Thus, these polyclonal antibod-
ies probably recognize a variety of epitopes and would detect calmodulin molecules even if some epitopes were masked or unavailable to antibody. Our results also differ somewhat from two previous papers (Kohnken et al., ’81; Nagao et al., ’87) reporting that calmodulin is present in ROS, as determined by radioimmunoassay. Those investigators measured calmodulin levels in purified toad and bovine ROS and found approximately 450 ng CaM/mg rod protein. This amount of calmodulin is low, representingonly 0.05% of the rod protein. Therefore, it is possible that these low levels of calmodulin could represent a contaminant copurified with the ROS preparation, or that the sensitivity of the immunohistochemical technique used here is not sufficient to allow detection of these low amounts of calmodulin. Because four different anti-calmodulin antibodies gave similar results on retinae from six different species, we are confident that the immunohistochemical localization we detect does represent specific calmodulin distribution. We
-
760
R. POCHET ET AL.
Fig. 13. Yellow field (YF) and red field (RF) section of pigeon retina treated with anti-calbindin. Double straight cones are positive only in yellow field. Compared with anti-CAM labeling (Fig. 12) in yellow field, horizontal cells are more strongly labeled with anti-calbindin. Ganglion
cells and the fiber layer are more strongly labeled for calmodulin. The stratification patterns in the IPL are also different for the two antibodies. Scale bars = 25 bm.
thus conclude that calmodulin is probably not involved directly in the mechanism of phototransduction. Besides the data discussed above on photoreceptors and calmodulin, few reports have appeared on calmodulin localization or activity in retina. A review of the recent literature indicated that among retinal cells and fiber layers, only ganglion cells have been investigated somewhat. For instance, in mouse retinal ganglion cells, calcium and calmodulin added to neurofilament preparations in the presence of heparin modestly stimulates phosphorylation of two neurofilaments subunits (Sihag and Nixon, '89). However, no direct evidence is provided for an in vivo mode of action of calmodulin in those cells. One interesting feature emerging from our study is the presence of calmodulin in many IPL fibers, where neuromodulin (p57, GAP-43, F1, B-50), a major neural-specific calmodulin binding protein, is almost exclusively located (J.J.Norden, personal communication). Neuromodulin has been postulated to play a role in regulation of free calmodulin levels in neurons by binding and concentrating calmodulin at specific sites in neurons, where protein kinase C-catalyzed phosphorylation of neuromodulin could result in a
local release of calmodulin and subsequent activation of calmodulin regulated enzymes (Alexander et al., '87; Estep et al., '90). Whether calmodulin and neuromodulin interact with each other in the IPL of the retina remains to be demonstrated. However, our findings are clearly consistent with this possibility. Calbindin-D28k, which belongs t o the same EF-hand superfamily of proteins as calmodulin and whose function is still conjectural, has been well studied for its distribution (for a review see Christakos et al., '89) especially in brain (Baimbridge et al., '82; Celio, '90; Feldman and Christakos, '83; Garcia-Segura et al., '84; Jande et al., '81; Sloviter, '89) and retina (Pasteels et al., '87b, '90; Rabie et al., '85; Rohrenbeck et al., '87, '89; Schreiner et al., '85; Verstappen et al., '86). In contrast to calmodulin, calbindin distribution in retina is more variable among species. For instance, calbindin is present in cones only in some well-defined subtypes of pigeon cones (Pasteels et al., '87b), whereas in monkey all cones except those of the fovea are calbindin positive (Pasteels et al., '90). A high variability in the pattern of calbindin labeling is also seen in the IPL from the investigated species. However, some aspects of calbindin
761
CALMODULIN AND CALBINDIN IN RETINA
Fig. 14. Immunohistochemical controls used with goldfish retina section (a)and pigeon retina section
(b).Anti-calmodulin (0.4 Wgiml) was preincubated for 24 hours with 6 pgiml calmodulin before immunoperoxidase staining. No significant labeling could he seen. By comparison, note the contrast obtained with pigment epithelium. Scale bars = 25 Wm.
TABLE 1. Frequency of Positive Cells'
Frog
Goldfish
CAM
CaBP
CAM
~~
Photoreceptor Horizontal cell Bipolar cell Arnacrine cell Ganglion cell
~
+++ ++
-
++2
Mouse
Turtle CAM
CaBP
+++
+
CaBP
Pigeon
Monkey
CAM
CaBP
CAM
-
-
-
CaBP
CAM
CaBP
~~
-
+ ++++ ++++
+++3
+
+++R
-
+++.I
++ +++ +++ ++ +++ ++ +++ +++ +++ ++ +++ ++ 'For each species, this table gives a n estimate of the No. of positive cells according to peroxidase staining for calrnodulin (CAM) and calbindin (CaBP),as follows: + (few), + + (some), + + + (many), + + + + (nearly all). See text and Figures 3 and 4 for relative intensities of staining.
+ ++ +++ +++
+++ ++ +++
++ ++ +++ +++
-/+
+++ +++ ++++
+++ ++ +++
+++ +++ ++++
++ ++ ++ ++++
'Only nuclei and a few cell bodies. 'Only cones and only in some parts of the retina.
distribution are conserved among species, allowing comparison with calmodulin. For example, horizontal cells are always more strongly labeled for calbindin than for calmodulin, whereas amacrine and ganglion cells are more strongly labeled and often more numerous for calmodulin than for calbindin. In conclusion, in addition to the high conservation of calmodulin sequence throughout evolution another level of conservation is observed, which is the conservation of calmodulin distribution among different cell types. However, calmodulin expression is not constitutive in the sense that not every cell expresses calmodulin to the same extent.
We are still far from a precise knowledge of why calmodulin is expressed within one particular neuron or group of neurons and not within its neighbor. However, we believe that detailed and systematic immunohistochemical and biochemical observations are leading to substantial progress in our understanding of the functions of calmodulin and calbindin-D28k in retinal information processing.
ACKNOWLEDGMENTS We thank Georgette Pattyn and Leon Surardt for excellent technical assistance, Paulette Miroir for typing and
R. POCHET ET AL.
762
Mrs. Lotteau for drawings. This work was supported by FRSM (grant No. 3.4511.88)and by funds from the Cystic Fibrosis Foundation.
LITERATURE CITED Alexander, K.A., B.M. Cimler, K.E. Meier, and D.R. Storm (1987) Regulation of calmodulin binding to P-57. A neurospecific calmodulin binding protein. J. Biol. Chem. 262:6108-6113. Baimbridge, K.G., J.J. Miller, and C.O. Parkes (1982) Calcium-binding protein distribution in the rat brain. Brain Res. 239:519-525. Brecha, N. (1983) Retinal neurotransmitters: histochemical and biochemical studies. In P.C. Emson (ed): Chemical Neuroanatomy. New York: Raven Press, pp. 85-129. Caceres, A., P. Bender, L. Smavely, L.J. Rebhun, and 0. Steward (1983) Distribution and subcellular localization of calmodulin in adult and developing brain tissue. Neuroscience 10:449-461. Celio, M.R. (1990) Calbindin D-28k and parvalbumin in the rat nervous system. Neuroscience 35:373-475. Christakos, S., C. Gabrielides, and W.B. Rhoten (1989)Vitamin D-dependent Calcium binding proteins: Chemistry, distribution, functional considerations and molecular biology. Endocr. Rev. 10:3-26. Cohen, P., and C.B. B e e (1988) Calmodulin. Amsterdam: Elsevier. Estep, R.P., K.A. Alexander, and D.R. Storm (1990) Regulation of free calmodulin levels in neurons by neuromodulin: relationship to neuronal growth and regeneration. Curr. Top. Cell Regul. 31r161-180. Feldman, S.C., and Christakos, S. (1983) Vitamin-D-dependent calciumbinding protein in rat brain: biochemical and immunocytochemical characterization. Endocrinology 112:290-302. Garcia-Segura, L.M., D. Baetens, J. Roth, A.W. Norman, and L. Orci (1984) Immunohistochemical mapping of calcium-binding protein immunoreactivity in the rat central nervous system. Brain Res. 296r75-86. Jande, S.S., L. Maler, and D.E.M. Lawson (1981) Immunohistochemical mapping of vitamin D-dependent calcium-binding protein in brain. Nature 294:765-767. Kitajima, S., A. Seto-Ohshima, M. Sano, and K. Kato (1983) Production of antibodies to calmodulin in rabbits and enzyme immunoassays for calmodulin and anti-calmodulin. J. Biochem. (Tokyo) 94:559-564. Kohnken, R.E., J.G. Chafouleas, D.M. Eadie, A.R. Means, and D.G. Me Connell(1981) Calmodulin in bovine rod outer segments. J. Biol. Chem. 256:12517-12522. Lin, C.T., J.R. Dedman, B.R. Brinkley, and A.R. Means (1980) Localization of calmodulin in rat cerebellum by immunoelectron microscopy. J. Cell Biol. 85t473-480. Means, A.R. (1988) Molecular mechanisms of action of calmodulin. Recent Prog. Horm. Res. 44:223-262. Means, A.R., and J.R. Dedman (1980)Calmodulin: an intracellular calcium receptor. Nature 28573-77. Nagao, S.,A. Yamazaki, and M.W. Bitensky (1987) Calmodulin and calmodulin binding proteins in amphibian rod outer segments. Biochemistry 26:1659-1665. Pasteels, B., M. Miki, S. Hatakenaka, and R. Pochet (1987a) Immunohistochemical cross-reactivity and electrophoretic comigration between calbiiidin-D27kDa and visinin. Brain Res. 422:107-113. Pasteels, B., M. Parmentier, D.E.M. Lawson, A. Verstappen, and R. Pochet (198713) Calcium binding protein immunoreactivity in pigeon retina. Invest. Ophthalmol. Vis. Sci. 28:658-664. Pasteels, B., J. Rogers, F. Blachier, and R. Pochet (1990) Calbindin and calretinin localization in retina from different species. Vis. Neurosci. 5:l-16. Pochet, R., G.D. Pipeleers, and W.J. Malaise (1987) Calbindin-D27kDa:
preferential localization in non-p-islet cells of the rat pancreas. Biol. Cell 61:155-161. Rabie, A,, M. Thomasset, C.O. Parkes, and M.C. Clavel (1985) Immunocytochemical detection of 28000-MW calcium-binding protein in horizontal cells of the rat retina. Cell Tissue Res. 240:493-496. Rohrenbeck, J., H. Wassle, and C. Heizmann (1987) Immunocytochemical laheling of horizontal cells in mammalian retina using antibodies against calcium-binding proteins. Neurosci. Lett. 77:255-260. Rohrenbeck, J., H. Wassle, and B.B. Boycott (1989) Horizontal cells in the monkey retina: Immunocytochemical staining with antibodies against calcium binding proteins. Eur. J. Neurosci. lt407-420. Sano, M., A. Seto-Ohshima, S. Kitajima, and K. Kato (1987) Immunoperoxidase staining of S-100 and calrnodulin in tissue sections and cultured cells. In A.R. Means, and P.M. Conn (eds):Methods in Enzymology, Vol. 139. New York Academic Press, pp. 495-504. Schreiner, D.S., S.S. Jande, and D.E.M. Lawson (1985) Target cells of vitamin D in the retina. Acta Anat. (Basel) 121:153-162. Seto-Ohshima, A., Y. Yamazaki, N. Kawarnura, M. Sano, S. Kitajima, and A. Mizutani (1987) The early expression of immunoreactivity for calmodulin in the nervous system of mouse embryos. Histochemistry 86:337343. Sihag, R.K., and R.A. Nixon (1989) In vivo phosphorylation of distinct domains of the 70-kilo-dalton neurofilament subunit involves different protein kinases. J. Biol. Chem. 2641457-464. Sloviter, R.S. (1989)Calcium-binding protein (Calbindin-D28K)and parvalbumin immunocytochemistry: Localization in the rat hippocampus with specific reference to the selective vulnerability of hyppocampal neurons to seizure activity. J. Comp. Neurol. 280:183-196. Spencer, R., M. Charman, J.S. Emtage, and D.E.M. Lawson (1976) Production and properties of vitamin-D-induced mRNA for chick calciumbinding protein. Eur. J. Biochem. 71:399-409. Szi.1, A., L. Takacs, E. Monostori, T. Diamantstein, I. Vigh-Teichmann, and P. Rohlich (1986)Monoclonal antibody recognizing cone visual pigment. Exp. Eye Res. 43:871-883. Vacca, L.L., S.J. Abrahams, and N.E. Naftchi (1980) A modified peroxidaseantiperoxidase procedure for improved localization of substance P in rat spinal cord. J. Histochem. Cytochem. 28997-304. Van Eldik, L.J., and D.M. Watterson (1981) Reproducible production of antiserum against vertebrate calrnodulin and determination of the immunoreactive site. J. Biol. Chem. 256t4205-4210. Van Eldik, L.J., and D.M. Roberts (1988) Calcium modulated proteins in pathophysiology. In M.P. Thompson (ed): Calcium Binding Proteins. Boca Raton, FL: CRC Press, pp. 59-76. Van Eldik, L.J., J.G. Zendegui, D.R. Marshak, and D.M. Watterson (1982) Calcium-binding proteins and the molecular basis of calcium action. Int. Rev. Cytol. 77:l-61. Van Eldik, L.J., K.-F. Fok, B.W. Erickson, and D.M. Watterson (1983) Engineering of site-directed anrisera against vertebrate calmodulin by using synthetic peptide imrnunogens containing an irnmunoreactive site. Proc. Natl. Acad. Sci. U.S.A. 80:6775-6779. Verstappen, A,, M. Parmentier, M. Chirnoaga, D.E.M. Lawson, J.L. Pasteels, and R. Pochet (1986) Vitamin D-dependent calcium binding protein immunoreactivity in human retina. Ophthalmic Res. 18:209-214. Wakakura, M., and N. Yamamoto (1987) Immunological localization of calmodulin in feline rod outer segments. Exp. Eye Res. 44:451-458. Wood, J.G., R.W. Wallace, J.N. Whitaker, and W.Y. Cheung (1980a) Immunocytochemical localization of calmodulin and a heat-labile calmodulinbinding protein (CaM-BP80)in basal ganglia of mouse brain. J. Cell Biol. 84.66-76. Wood, J.G., R.W. Wallace, J.N. Wbitaker, and W.Y. Cheung (1980b) Immunocytochemical localization of calmodulin in reaons of rodent brain. Ann. N.Y. Acad. Sci. 356:75-82.
