cisternal stacks, a specialized endoplasmic reticulum subcompartment rich in inositol 1,4,5-trisphosphate receptors. D. A. RUSAKOV 1'3, P. PODINI 2, A. VILLA ...
Journal of Neurocytology 22, 273-282 (1993)
Tridimensional organization of Purkinje neuron cisternal stacks, a specialized endoplasmic reticulum subcompartment rich in inositol 1,4,5-trisphosphate receptors D . A . R U S A K O V 1'3, P. P O D I N I 2, A . V I L L A 2 a n d J. M E L D O L E S I 2.
1 State University, Dniepropetrovsk; and Neurocytology, Bogomoletz Institute of Physiology, Academy of Science, Kiev 252601 Ukraine 2 Department of Pharmacology, CNR Cytopharmacology and B. Ceccarelli Centers, S. Raffaele Scientific Institute, Via Olgettina 60, 20132 Milano, Italy. 3 Histophysics and Cytophysics, Institute of Theoretical Biology, Angers 49000, France Received 22 June 1992; revised 14 October and 19 November 1992; accepted 23 November 1992
Summary Stacks of regularly spaced, flat, smooth-surfaced endoplasmic reticulum cisternae, frequently observed in both the cell body and dendrites of cerebellar Purkinje neurons, were previously shown by immunocytochemistry to be highly enriched in receptors for the second messenger, inositol 1,4,5-trisphosphate. Morphometric analyses have been carried out on randomly selected thin section images of rat Purkinje neurons to reveal the tridimensional organization of these structures. Individual stacked cisternae (on the average ~ 3.5 per stack) were shown to be separated from each other by a 23.5 nm space occupied by perpendicular bridges, ~ 20 nm in diameter, most probably composed by two apposed receptor homotetramer molecules, inserted into the parallel membranes in their hydrophobic domains. In the stacked membranes the density of the bridges was 500 ~m -2, corresponding to ~ 15% of the surface area. The lateral distribution of bridges was not random, but revealed regular distances that might correspond to unoccupied receptor slots. In each stack, the external cisternae were often in direct lumenal continuity with conventional elements of the endoplasmic reticulum, whereas the internal cisternae were not. Since continuities between stacked cisternae were never observed, the results indicate that the internal cisternae are at least transitorily discrete, i.e. they are not in permanent lumenal continuity with the rest of the endoplasmic reticulum. To our knowledge this is the first demonstration of a physical subcompartmentalization of the latter endomembrane system in a non-mitotic cell. A model for the biogenesis of cisternal stacks, based on the head-to-head binding and lateral interaction of the inositol 1,4,5-trisphosphate receptor molecules in the plane of the interacting membranes, is proposed and critically discussed.
Introduction Cisternal stacks (CS) of Purkinje n e u r o n s are peculiar structures c o m p o s e d of piles of regularly spaced, flat, s m o o t h surfaced cisternae, some of which are in direct m e m b r a n e continuity with r o u g h and smoothsurfaced cisternae of the endoplasmic reticulum (ER), initially described in the late sixties. These structures bear some resemblance to, but are probably distinct from, the stacks that constitute the spine apparatus of a variety of n e u r o n s (Gray, 1959). For quite sometime, the Purkinje n e u r o n CS were considered to result from the artifactual extraction of lipids during fixation ( H e r n d o n , 1964; Karlsson & Schultz, 1966; Van Nimw e g e n & Sheldon, 1966; H a n s s o n , 1981; see h o w e v e r * To whom correspondence should be addressed. 0300--4864/93 $03.00 +.12
9 1993 Chapman and Hall Ltd
Le Beux, 1972), a n d therefore interest about t h e m declined. Recently, h o w e v e r , high resolution imm u n o c y t o c h e m i c a l w o r k from three laboratories has d e m o n s t r a t e d the m e m b r a n e s of CS to be extraordinarily rich in the receptor for inositol 1,4,5-trisphosphate (Ins-P3) (Berridge & Irvine, 1984; Otsu et al., 1990; Satoh et al., 1990; Volpe et aI., 1990; Villa et al., 1991; Takei et al., 1992; Yamamoto et al., 1992), a second m e s s e n g e r g e n e r a t e d following the activation of m a n y receptors which is k n o w n to induce the release of Ca 2+ from intracellular stores. The existence of CS correlated well with the u n u s u a l l y high expression of the receptor ( ~ 50 fold higher than in most
274 other cells) that had been previously observed in Purkinje n e u r o n s (Mignery et al., 1989; Ross et al., 1989). Taken together, these considerations ruled out the possibility that CS are simple artifacts and raised a n u m b e r of i m p o r t a n t questions about their physiological significance, architecture and relationship with the ER. Recent immunocytochemical studies, carried out in chicken Purkinje neurons, have raised questions about the Ca 2+ storage function of these structures. In fact, no labelling of the CS was observed for Ca 2+ ATPase while the concentration of the Ca 2§ storage protein, calsequestrin, was not higher than in the rest of the ER (Villa et al., 1991, 1992; Takei et al., 1992). The possibilities that CS m a y serve as stores not of Ca 2+, but of the Ins-P3 receptor itself, or that they m a y operate to buffer transient elevations in Ins-P3 have therefore b e e n suggested (Villa et al., 1991, 1992; Takei et al., 1992). O n the other hand, the CS of both the rat and the chicken Purkinje n e u r o n s were s h o w n to contain the so called immunoglobulin binding protein (BiP), a typical marker of the ER l u m e n (Villa et al., 1991, 1992; Takei et al., 1992). This finding excludes the possibility that these structures belong to an e n d o m e m b r a n e system distinct from the ER. So far, however, the mechanisms by which the Ins-P3 receptor concentrates in the membranes, and the stacked cisternae are assembled, have not been clarified. In the case of other peculiar structures, such as the sarcoplasmic reticulum of skeletal muscle fibres and the active zones of the presynaptic m e m b r a n e (see Heuser, 1976; Franzini-Armstrong et al., 1987), detailed k n o w l e d g e of architecture has eventually proved of importance for the u n d e r s t a n d i n g of function. This consideration has p r o m p t e d us to carry out a m o r p h o m e t r i c analysis of CS in the Purkinje n e u r o n of the rat, the species that so far has been most profoundly investigated. In these studies, attention was given not only to the cisternae and their a r r a n g e m e n t but also to the bridges, discrete units interposed between adjacent cisternae that are probably comp o s e d by pairs of Ins-P3 receptor tetramers arranged in register (Otsu et al., 1990; Satoh et al., 1990; Takei et al., 1992; Yamamoto et al., 1992). Materials and Methods ELECTRON MICROSCOPY Male rats, 100-150 g in body weight, were anaesthetized with thiopental and perfused through the heart, first with 200 ml phosphate-buffered saline (PBS) at 4 ~C and then with a mixture of 4% formaldehyde (freshly prepared from paraformaldehyde) and 0.2% glutaraldehyde in 125mM phosphate buffer (pH 7.4) (500 ml at 4~C). The cerebellum was rapidly removed and ~ 1 mm thick frontal slices were prepared. The slices were immersed into the fixative mixture and then sectioned into small cubes. Those including the boundary between the molecular and granular layers of the
RUSAKOV, PODINI, VILLA and MELDOLESI cortex were further fixed in the same mixture for two additional hours, then washed extensively with the phosphate buffer, postfixed with 1% OsO4 in 125 mM cacodylate buffer, dehydrated in ethanol, block-stained with uranyl acetate, and embedded in Epon. The thin sections obtained from these samples were examined in a Hitachi H-7000 electron microscope. MORPHOMETRIC ANALYSES
Primary measurements Electron micrographs displaying whole CS profiles in Purkinje neurons were selected randomly and collected in two samples, corresponding to cell bodies and dendrites (155 and 171 CS profiles, respectively). The pictures were analyzed using a plane image analyzer. Volume and shape of CS Because of their irregular shape, CS geometry was analyzed not by the sphere or disk (De Hoff, 1962; Santalo, 1976; Ambartsumyan et al., 1989), but by the 'crushed' ellipsoid approximation paradigm (Rusakov, 1992). In brief, the frequency distribution (FD)fR(j) of the three-dimensional CS radius was derived from fr(i) of the profile radius by solving the equation for fR(j): I I a-lfi = j=i ~ R(i,j)fR(j) + -~ Ti fR(z). j=l ~ fR(j)j 1; i = 1,j where T is the (ultrathin) section thickness; a is a normalizing coefficient; c~(i,j)is defined to De Hoff (1962) and h is the class width. The CS volume distribution can then be derived using Monte Carlo simulation (Rusakov, 1992). To describe the shape of CS, we used the profile elliptical coefficient, i.e. the ratio of the smaller and larger principal radii of the ellipse approximating the profile. Most often, the I larger principal radius was found to be parallel to the membranes of the stacked cisternae.
Distances between apposing membranes The FD of the actual (spatial) distance between the membranes, W, can be derived from the FD of the distances w, visible in the micrographs, through Monte Carlo simulation of a random variable W = w cos~ + T sinf~ where T is the ultrathin section thickness, f3 an angle between the section plane and the plane of CS cisternal membranes and w is a random variable with the FD estimated from direct observations of CS membrane thickness. Based on the study of the analyzed images, the angle was found to be 79~. Lateral arrangement of intercisternal bridges To quantify and reconstruct the surface topography of bridges interposed between CS membrane surfaces, a numerical approach described elsewhere was employed (Rusakov et al., 1992). In brief, the lateral arrangement of the bridges was considered to be a stationary isotropic (twodimensional, 2D) random point process (Santalo, 1976; Ambartsumyan et al., 1989). This suggested the micrographs to represent random sections made in an infinite surface covered with points (bridges). In the micrographs, FD gq of distances q between the centres of all, not only neighbouring, bridges along membrane profiles was obtained to
ER cisternal stacks in Purkinje n e u r o n s simulate FD gQ of lateral distances Q between the bridges on the membrane surface (Rusakov et al., 1992). Using the FD gQ obtained, we could simulate a typical arrangement of bridges on the 'unfolded' membrane area (Rusakov et al., 1992). To avoid edge effects, the width of the simulated windows was selected to be two-fold wider than maximal q (300 and 150 nm, respectively). The local, at least 150 nm wide surface pattern of the bridges can thus be considered valid in the reconstructed pictures. Structural continuity between stacked and conventional ER cisternae
As previously observed, some of CS profiles in Purkinje neurons reveal direct continuities with ordinary ER (Otsu et al., 1990; Satoh et al., 1990; Villa et al., 1991, 1992; Takei et al., 1992; Yamamoto et al., 1992). To quantify spatial relationships between CS and the continuities, we considered the problem from a simple probabilistic point of view. The chance that a cisternal profile reveals an existing continuity depends on whether the latter falls in the sectioning (microtome) plane. As the microtome intersects the periphery of each cistern randomly, the probability of appearance of one continuity in an individual cisternal profile should be equal to 1/L = Po where L is the total length of the tridimensional lateral cisternal periphery, and I is the part of L occupied by continuities (if any). Similarly, two continuities will appear in an individual cisternal profile with the probability p~. For the case of a cisternal stack, k continuities should appear, according to combinatories, with the probability prkl = C~p~(1 - po) n-k,
where Cf =
k = O,n
n! k!(n - k)!,
n = maximal number of continuities per tridimensional stack, and the values of probability pckJ can be easily estimated from an adequate sample of micrographs. Preliminary analyses showed that the maximal number of continuities visible in individual CS profiles is four. Most often, these continuities belonged to either one of external (marginal) cisternae. Thus, accepting n = 4, five four-order equations for p0 and, therefore, five independent estimates of p0 can be calculated. Since observations with k = 3, 4 are seldom, i.e. the estimate can be unstable, we used three estimates with k = 0, 1, 2. Similarly, to evaluate the degree of continuity of a single cisterna located deep in a CS, we used n = 2, k = 0,1. The equations were solved graphically. The estimated p0 illustrates the percentage of the cisternal periphery (edge, perimeter) in direct continuity with ER, on average.
Results General architecture of CS
Figure 1A and B illustrates representative examples of CS observed in the cell b o d y and dendrites of rat Purkinje n e u r o n s . In most of these structures the general organization was ordered, with parallel cisternae very similar to one a n o t h e r except for the continuity of a single (usually external) cistern with
275 the rest of the ER (Fig. 1A). In a few cases h o w e v e r , CS profiles exhibited more complex architecture, due to either multiple continuities (Fig. 1C) or to the a p p a r e n t twisting of the structure (Fig. 1D). Bridges were visible s p a n n i n g the space b e t w e e n adjacent m e m b r a n e s . In a few cases, because of the a p p a r e n t d e t a c h m e n t (or sectioning away) of the adjacent m e m b r a n e , the bridges a p p e a r e d as small particles sticking out of a single m e m b r a n e (Fig. 1D). In a few favourable sections they w e r e visible also in front view, appearing as discrete dots, ~ 20 n m in diameter (Fig. 1D). In the CS p o p u l a t i o n investigated, the n u m b e r of cisternae per CS was on average 3.38_+ 0.98 and 3.55 _+ 1.49 for the cell b o d y and dendrites, respectively ( n u m b e r of analyzed CS profiles was 135 and 160 respectively). Figure 2 provides information on the size and shape of these structures as revealed by the statistical analysis. O n average, both the profile area (A,B) and the estimated v o l u m e (C,D) of CS were higher (60-70%) in dendrites t h a n in the cell body, as recently r e p o r t e d also b y others (Yamamoto et al., 1992). F r e q u e n c y analyses of the two parameters suggested the existence of at least two s u b p o p u lations, one smaller, the other larger (profile areas: 0.02 - 0.10 a n d 0.14 - 0.20 b~m 2, respectively), tn contrast, the CS elliptical coefficient (i.e. the ratio b e t w e e n the two principal radii, w i d t h and length, of the sectioned CS profiles) was ~ 19% lower in dendrites than in the cell b o d y (Fig. 2E, F) indicating that CS are o n the average thinner in the first than in the second cell area. The distance b e t w e e n adjacent cisternae in the analyzed CSs is r e p o r t e d in Fig. 3. After stereological analysis, the estimated average values ( ~ 23.5 nm) were similar in the cell b o d y and dendrites and their spectra a p p r o a c h e d Gaussian distribution. Distribution of intercisternal bridges and surface attached ribosomes
The bridges i n t e r p o s e d b e t w e e n adjoined CS cisternae were a n a l y z e d stereologically. Figure 4A and B illustrates typical examples of their lateral distribution, reconstructed statistically based on a set of experimental data in the cell b o d y and dendrites, respectively. As can be seen, the various bridges (drawn as dots) 20 n m in diameter, a p p e a r e d either closely a p p o s e d to one a n o t h e r in the lateral plane or separated by variable spaces. The spectra of the lateral distances b e t w e e n bridges revealed not a uniformily r a n d o m distribution but f r e q u e n c y modes, appreciable in both the cell b o d y a n d dendrites, c o r r e s p o n d i n g to 30-35, 60-70 and 90-95 nm. Bridges a p p e a r therefore to be regularly a r r a n g e d in the lateral plane, according to a general layout revealed by the topographic images s h o w n in Fig. 4A and B. Quantitative data on the distribution of the bridges in the CS of cell b o d y and dendrites are given in Table 1. Notice that in both areas the calculated lateral
276
RUSAKOV, PODINI, VILLA and MELDOLESI
Fig. 1. Ultrastructural images of CS. Representative examples of the CS observed in the cell body and dendrites are shown in (A) and (B) respectively. Notice the numerous flat, smooth surfaced parallel cisternae, separated by a regular space occupied by perpendicular bridges. In (A) the continuity of an external stacked cistern with a conventional, rough-surfaced cistern of the ER is marked by a large arrow, and the ribosomes attached to the cytosolic surfaces of the CS by small arrows. In (B) the CS is located adjacent to mitochondria (M). A dendritic spine is marked SP. (C) shows a stack in which the external cisternae are both continuous with rough-surfaced ER cisternae (large arrows); (D) a highly unusual, irregularly organized twisted CS that shows multiple continuities with the ER (large arrows). Bridges are visible in both lateral and front (circle) view. (A) x 78 000; (B-D) x 64 000.
density was of ~ 5 0 0 ~ m 2, corresponding to an occupied area of ~ 15%. In the lower part of the table, data on the distribution of ribosomes attached to the cytosolic surface of the external cisternae in the cell b o d y CS are c o m p a r e d with those c o m p u t e d for near-by conventional rough-surfaced ER cisternae. As can be seen, no major difference was observed bet w e e n ribosome densities over the two analyzed structures.
CS-ER continuities Continuities b e t w e e n the edges of CS and conventional ER cisternae have been observed previously (Otsu et al., 1990; Satoh et al., 1990; Volpe et al., 1990; Takei et al., 1992; Yamamoto et al., 1992). Most often, these continuities involved the external cisternae of the stacks. These continuities are particularly impressive in the cell body, w h e r e the ER continuous with the CS is mostly r o u g h surfaced, but are easily appreciable
ER cisternaI stacks in Purkinje n e u r o n s
0.5
i
277
t
A
0.5
i
0.3
0.3
~D
0.2
0.2
0.1
0.1 I
0.1
0
0.2
0.3
M-O. 109
0.4
t-
0
i
B
M=0.068
0.4
i
0.4
0
~-3
0
1-7
i
0.1
0.2
0.3
!
i
i
0.4
Stack profile area (lam2)
0,5
i
C
t_
D
M=O.OIO
0.4 e-, ~D
0.5
!
0.3
0.3
0.2
0.2
0.1
0.1
0
0.01
0
0.02
0.03
M=0.017
0.4
0.04
0
0.01
0
0.02
0.03
0.04
Estimated stack volume (iam3)
0.5
0.5
i
F
F
M=0.496
0.4 =
0.3
0.3
I,,-,
0.2
0.2
0.1
0.1
0
i
i
0.5
i
]
1
I
1.5
M=0.405
0.4
2
0
t
0
0.5
I
1
1
1.5
2
Elliptical coefficient of stack profile Fig. 2. Size and shape of CS. The panels to the left refer to cell body CS, those to the right to dendritic CS. M is the average in the analyzed sample.
278
RUSAKOV, P O D I N I , VILLA and MELDOLESI
A
B 0.5
0.5 M 1 =22.1
0.4
M2=23.6
M2=23.8
L)
0.3
0.3
~D
0.2
0.2
k-.,
M 1 =21.7
0.4
s
0.1 0 10
--n
I
0.1
i i
20
30
40
50
I
0 10
20
I L. . . . -40 50
30
Distance between cistemae (nm) Fig. 3. Intercisternal distances in cell body (A) and dendritic (B) CS. Solid lines illustrate the values directly measured on the micrographs, dotted lines those estimated by the statistical analysis (M1 and M2 = averages, respectively).
A
9 9 ' 9
300 250
O 9
tOo0 O
200 .E
B
0
100
~
9
9
,::,"
O
A d
Oooo0
Oo o
h
o= . 200
100
250
o:...
000 ~
200
O
i
0
300
q
~ 9
150
9 9
9
9
9 O
150
0 9 D 0
100
O0
9
OO
9
50 A
w 300
O9
0 0
I
1
100
200
300
Lateral abscissa (nm) 0.15
0.15
0.1
0.1
0.05
0.05
0
F0
50
100
L
0
0
Lateral interval between "bridges" (nm)
/
I
50
100
ER cisternal stacks in P u r k i n j e n e u r o n s
0.6
279
0.6
+
B
A :~ r
0.4
0.4
0.2
0.2
0
I 0
I
2
4
0
2
4
r
r
Number of continuities 1
1
--'--'w
1------
D 0.8
0.8
r.-
0.6
0.6
G~ r+
0.4
0.4
0.2
0.2
0
0
0
2
4
0
_XZZEL___~ 2
4
Number of continuities Fig. 5. Cisternal stacks-ER cisternal continuities. Histograms in (A) and (B) show the number of continuities observed in all cisternae of the analyzed CS population of the cell body and dendrites, respectively. The corresponding values, concerning the internal cisternae only, are given in (C) and (D).
Table 1. Principal parameters of the lateral topography of intercisternal bridges and ribosomes on the membrane surfaces.
Electron dense particles
Bridges Ribosomes
Cell structures
Measured profile density ( p+m-1 )
Estimated lateral density ( p+m- 2)
Estimated occupied area (percentage)
CS cell body CS dendrites CS, ext. cisternae ER cisternae
32.0 31.6 24.1 22.5
500 475 325 315
14.2 14.5 15.1 13.7
+ + + +
5.2* 7.8* 5.5 ~ 4.2 t
*~: values in the pairs of data labelled with the same mark do not differ significantly from each other (p > 0.25).
Fig. 4. Distribution of intercisternal bridges in cell body (A) and dendritic (B) CS. The statistically reconstructed lateral topography of the bridges in a 300 • 300 nm CS membrane surface is shown in the upper panels. The graphs in the lower panels illustrate the initial parts of the frequency spectra of the reconstructed lateral distances between all bridges on the membrane surface. Modes are marked by arrows.
RUSAKOV, PODINI, VILLA and MELDOLESI
280 Table 2. Estimated percentage of the CS free cisternal edges that possess continuities with ordinary rough ER.
Whole CS Single deep cisternae
Cell b o d y
Dendrites
12.9 + 2.42 3.19 + 0.43
9.9 _+2.87 2.9 + 0.71
Percentage means were estimated based on the experimental data shown in Fig. 4. also with the smooth-surfaced cisternae of dendrites (Fig. 1). Stereological analyses were carried out in order to establish the degree of direct interaction between the CS and ER membrane types. Figure 5A and B, shows that on the average approximately half of the cell body CS sectioned profiles failed to show continuities, and that this value increased to above 50% in the dendrites. A single cistern in continuity with the ER was observed in approximately 30% of the CS, while continuities of two, three and even four cisternae were less frequently observed. Statistical analysis of these data (Table 2) revealed that, when the cisternae are analyzed together, continuities occupy approximately 13% and 10% of their free edges in the cell body and dendrites. Calculated separately for the external and deep cisternae, however, these values changed markedly (almost 20% and about 3%, respectively, Fig. 5C, D and Table 2). Discussion
Cisternal stacks have emerged from our morphometric and statistical analyses as ordered structures characterized by the presence of bridges running in the space interposed between adjacent cisternae, perpendicularly to the major axis. Based on this general architecture we conclude that the CS of Purkinje neurons should be classified together with another structure formed by the specific surface interaction of two membranes specialized for Ca 2+ transport, i.e. the triads of skeletal muscle fibres (FranziniArmstrong et al., 1987; Fleischer & Inui, 1989). The organization of muscle triads has been thoroughly investigated both from the morphological and the molecular points of view. The bridges interposed between the two membranes (feet) are now known to be composed by an intracellular channel, the ryanodine receptor, a high Mr homotetramer (Inui et al., 1987; Anderson et al., 1989; Fleischer & Inui, 1989) concentrated in the junctional membrane of the sarcoplasmic reticulum terminal cisternae and coupled in a 1:1 ratio to molecules of the L type Ca 2+ channel. These last molecules are located in the plasmalemmal transverse tubules and work as voltage sensors (Rios & Gonzales, 1991). The reported size of the feet (~ 20 nm diameter: Inui et al., 1987; Anderson et al.,
1989) correlates well with that observed in the present and previous studies for the CS bridges (Maeda et al., 1989; Satoh et al., 1990; Takei et al. 1992; Yamamoto et al., 1992). This was not unexpected because the ryanodine (Inui et al., 1987; Anderson et al., 1989; Fleischer & Inui, 1989) and the Ins-P3 receptors (Maeda et al., 1989; Mignery et al., 1989) are known to resemble one another in general organization (a large cytosolic domain superimposed over the transmembrane channel). Since, however, the cytosolic domain is approximately two-fold larger in the ryanodine receptor (Inui et al., 1987; Anderson et al., 1989; Fleischer & Inui, 1989) than in the Ins-P3 receptor (Maeda et al., 1989), the foot-bridge size similarity suggests the latter structures to be composed of two interacting Ins-P3 receptor molecules inserted in adjacent membranes. Two of the results we have obtained appear worthy of particular attention. The first concerns the distribution of bridges. Recent immunocytochemical results (Villa et al., 1992; Yamamoto et al., 1992) have already demonstrated the specialization of CS membranes to be far from absolute. In fact, the antigens recognized by two anti-ER membrane antibodies were revealed in the CS at concentrations similar to those of the adjacent conventional ER cisternae. The present demonstration that bridges occupy only 15% of the CS membrane surface appears in agreement with the previous conclusion. Interestingly, when the bridges were not laterally apposed, the spaces remaining in between appeared not random but corresponded to values moderately larger than a single bridge diameter or multiples of it. This suggests that the CS membranes are not saturated by the bridges btIt exhibit free slqts where additional bridges may be inserted. This, admittedly hypothetical, interpretation of our data implies the existence in the CS membrane of an ordered scaffold where bridges, or alternatively other membrane components, can be distributed. That the bridge density in CS is not maximal is suggested also by the analogy with the muscle triads, where foot density is distinctly higher (Franzini-Armstrong et al., 1987; Fleischer & Inui, 1989). The second interesting result we have obtained concerns the continuities of stacked with conventional ER cisternae. Quantitatively, the data we have obtained by our statistical analysis should be considered with some caution because, in order for continuities to be recognized, their plane needs to be not too far from perpendicular with respect to the section plane. As a whole, however, the data appear quite clear. The external cisternae of the CS were found to be continuous for a large fraction (almost 20%) of their edges. Direct membrane continuity with the rest of the ER therefore represents a property common to many, if not all, external cisternae. In this respect it is worth emphasizing that the latter resemble conventional ER cisternae, and differ therefore from the cisternae more
ER cisternal stacks in Purkinje neurons deeply located in the CS, also because their cytosolic membrane is not specialized. In both cell body and dendrites the latter shows, in fact, an Ins-P3 receptor density similar to the rest of the ER, not to the deep CS membranes (Satoh et aI., 1990). Moreover, in the cell body the cytosolic membrane is often covered with ribosomes, at concentrations similar to the conventional rough-surfaced ER. Continuities with the ER, similar in diameter to those of the external cisternae, were observed also with the deep cisternae, however only rarely (~ 3% of the total edge surface). This, coupled to the fact that continuities between adjacent cisternae were never seen in a CS, indicates that deep cisternae are most often discrete, i.e. correspond to separate compartments. This conclusion is important because the whole ER is widely believed to consist of a single, structurally continuous membrane system, and the possibility of at least transient fragmentation has never been considered except during mitosis. A final, so far open question, concerns the biogenesis of CS. Recent developmental studies in striated muscle fibres have suggested the triad organization derives from the apposition of discrete vesicles, rich in ryanodine receptor, with tubules rich in Ca 2+ channels, followed by the fusion of these structures with the sarcoplasmic reticulum and the plasma membrane, respectively (Yuan et al., 1991). A similar mechanism concerning however only Ins-P3 rich vesicles and ER cisternae could also account for CS formation. At the moment, however, we favour the possibility that the direct interaction of Ins-P3 molecules inserted in two ER membranes could induce their alignment and thus favour the successive interaction of additional receptor molecules, with ensuing concentration in the stacked membranes. Successive alignment of other cisternae could then result in multicisternal CS, and the compact, probably rigid geometry of the structure thus assembled could ultimately favour the detachment of part of the stacked cisternae from the conventional of the ER. The latter is in fact a highly dynamic membrane system capable of continuous oscillatory movements (Dabora & Sheetz,
281 1988; Lee & Chen, 1988; Terasaki & Jaffe, 1991) which could pull a CS in various directions and thus provide the energy for membrane discontinuity. This interpretation could also explain the frequently observed association of stacks with the cytosolic surface of both the plasmalemma and mitochondria (Herndon, 1964; Satoh et al., 1990; Takei et al., 1992), i.e. at sites where oscillations of the ER might be partially blunted. In any event, the observations of a few continuities and the persistence of lumenal ER markers at levels comparable to those of the rest of the ER (Villa et aI., 1991, 1992; Takei et al., 1992) suggest that even the deep stacked cisternae can fuse from time to time with the ER in order to equilibrate their pools of specific components. The hypotheses that we have discussed may also explain w h y typical CS are observed only in Purkinje neurons. In other cells the concentration of Ins-P3 receptors is much lower; therefore the probability of multiple interactions capable of inducing the coupling of two adjacent membranes may be reduced accordingly. In addition, in at least some of these other cells, the Ins-P3 receptors might be of additional types, molecularly distinct from the cerebellar (or type I) Ins-P3 (see Meldolesi, 1992 for a recent review). However, transient interactions of a few receptor couples, difficult to recognize by a simple morphological analysis, cannot be excluded even in other cells, where they could play a role in receptor function.
Acknowledgements This work was supported in part by grants from the CNR Target Project on Biotechnology and Bioinstruments and Special Project Ca 2+ homeostasis. The secretarial assistance of L. Di Giorgio and B. Penati is gratefully acknowledged. DAR was supported by the Ecole Pratique des Hautes Etudes, Paris. DAR thanks the International Centre for Theoretical Physics, College on Neurophysics, Trieste, for interest and support.
References (1989) Introduction to Stochastic Geometry. Moscow/Berlin: Nauka/ WTB.
AMBARTSUMYAN, R. V., MECKE, J. & STOYAN, D.
ANDERSON, K., LAI, F. A., LIN, Q. Y., ROUSSEAU, E., ERICKSON, H. P. & MEISSNER, G. (1989) Structural a n d
functional characterization of the purified cardiac ryanodine receptor-Ca R+release channel complex. Journal of Biological Chemistry 264, 1329-35. BERRIDGE, M. J. & IRVINE, R. (1984) Inositol t r i s p h o s p h a t e , a
novel second messenger in cellular signal transduction. Nature 312, 315--21.
DABORA, S. L. & SHEETZ, M. P. (1988) The microtubuledependent formation of a tubulovesicular network with characteristics of the ER from cultured cell extracts. Cell 54, 27-35. DE HOFF, R. T. (1962) The determination of the size distribution of ellipsoidal particles from measurement made in random plane sections. Transactions of the Metallurgical Society of AIME 224, 474-7. FLEISCHER, S. & INUI, M. (1989) Biochemistry and biophysics of excitation-contraction coupling. Annual Review of Biophysics and Biophysical Chemistry 18, 333-64.
282 F R A N Z I N I - A R M S T R O N G , C., KENNEY, L. J. & V A R R I A N O -
MARSTON, E. (1987) The structure of calsequestrin in triads of vertebrate skeletal muscle: a deep-etch study, Journal of Cell Biology 105, 49-56. GRAY, E. G. (1959) Electron microscopy of synaptic contacts on dendrite spines of the cerebral cortex. Nature 183, 1592-3. H A N S S O N , H.-A. (1981) LameUar bodies in Purkinje nerve cells experimentally induced by electric field. Brain Research 216, 187-91. HERNDON, R. M. (1964) Lamellar bodies, an unusual arrangement of the granular endoplasmic reticulum. Journal of Cell Biology 20, 338-42. HEUSER, J. E. (1976) Morphology of synaptic vesicle discharge and reformation at the frog neuromuscular junction. In Motor Innervation of Muscle (edited by THESLEFF, S.) 51-115. London: Academic Press. INUI, M., SAITO, A. & FLEISCHER, S. (1987) Isolation of the ryanodine receptor from cardiac sarcoplasmic reticulum and identity with the feet structures. Journal of Biological Chemistry 262, 15637-42. KARLSSON, U. & S C H U L T Z , R. L. (1966) Fixation of the central nervous system for electron microscopy by aldehyde perfusion. III. Structural changes after exsanguination and delayed perfusion. Journal of Ultrastructure Research 14, 47-63. LE BEUX, J. (1972) Subsurface cisterns and lamellar bodie~: particular forms of the endoplasmic reticulum in the neurons. Zeitschrift fidr Zellforschung 133, 327-52. LEE, C. & CHEN, L. B. (1988) Dynamic behavior of endoplasmic reticulum in living cells. Cell 54, 37-46. MAEDA, N., NIINOBE, M., INOUE, Y. & MIKOSHIBA, K.
(1989) Developmental expression and intracellular location of P400protein characteristic of Purkinje cells in the mouse cerebellum. Developmental Biology 133, 67-76. MELDOLES1, J. (1992) Multivarious IP3 receptors. Current Biology 2, 393-4. MIGNERY, G . A . , S U D H O F , T . C . , T A K E I , K . & D E C A M I L L I , P.
(1989) Putative inositol 1,4,5-trisphosphate receptor similar to ryanodine receptor. Nature 342, 192-5. OTSU, H., Y A M A M O T O , A., MAEDA, N., MIKOSHIBA, K. &
TASHIRO, Y. (1990) Immunogold localization of inositol 1,4,5-trisphosphate (Ins-P3) receptor in mouse cerebellar Purkinje cells using three monoclonal antibodies. Cell Structure and Function 15, 163-73. RIOS, E., MA, J. & G O N Z A L E S , A. (1991) The mechanical hypothesis of excitation-contraction (EC) coupling in skeletal muscle. Journal Muscle Research and Cell Motility 12, 127-35. ROSS, C. A., MELDOLESI, J., MILNER, T. A., S A T O H , T., S U P A T T A P O N E , S. & SNYDER, S. (1989) Inositol 1,4,5-
trisphosphate receptor localized to endoplasmic reticulum in cerebellar Purkinje neurons. Nature 339, 468-70. RUSAKOV, D. A. (1992) Estimation of the size distribution of
RUSAKOV, PODINI, VILLA and MELDOLESI closed cell elements from analysis of their plane random sections. Biometrics, in press. RUSAKOV, D. A., BEREZOVSKAYA, O. L. & SKIBO, G. G.
(1993) Cytoskeleton-mediated, age-dependent lateral topography of lectin-gold labelled molecules on the plasma membrane of cultured neurons: a statistical view. Neuroscience 52, 369-79. SANTALO, L. A. (1976) Integral geometry and geometric probability. In Encyclopedia of Mathematics and its Applications, vol. 1, pp. 1-311, Reading, MA: Addison-Wesley. S A T O H , T., ROSS, C. A., VILLA, A., S U P A T T A P O N E , S., P O Z Z A N , T., SNYDER, S. & MELDOLESI, J. (1990) T h e
inositol 1,4,5-trisphosphate receptor in cerebellar Purkinje cells: quantitative immunogold labeling reveals concentration in an ER subcompartment. Journal of Cell Biology 111, 615-24. TAKEI, K., STUKENBROK, H . , METCALF, A., MIGNERY, G., S U D H O F , T., VOLPE, P. & DE CAMILLI, P. (1992) C a 2+
stores in Purkinje neurons: endoplasmic reticulum subcompartments demonstrated by the heterogeneous distribution of the Ins-P3 receptor, Caa+-ATPase and calsequestrin. Journal of Neuroscience 12, 489-505. TERASAKI, M. & JAFFE, L. A. (1991) Organization of the sea urchin egg endoplasmic reticulum and its reorganization at fertilization. Journal of Cell Biology 114, 929-40. V A N t N I M W E G E N , D. & S H E L D O N , H. (1966) Early postmortem changes in cerebellar neurons of the rat. Journal of Ultrastructure Research 14, 36--46. VILLA, A., P O N D I N I ,
P., CLEGG, D. O., P O Z Z A N , T. &
MELDOLESI, J. (1991) Ca 2+ stores in chicken Purkinje neurons: differential distribution of the low affinity-high capacity Ca 2+ binding protein, calsequestrin, of Ca 2+ ATPase and of ER lumenal protein, BiP. Journal of Cell Biology 113, 779-91. VILLA, A., SHARP, A. H., RACCHETTI, G., PODINI, P., BOLE, D. G., D U N N , W. A., P O Z Z A N , T., SNYDER, S. H. & ~v~ELDOLESI, J. (1992) The endoplasmic reticulum of
Purkinje neuron body and dendrites: molecular identity and specializations for Ca 2+ transport. Neuroscience 49, 467-77. VOLPE, P., A L D E R S O N - L A N G , B. H . , M A D E D D U , L., D A M I A N I , E., COLLINS, J. H. & M A R G R E T H , A. (1990)
Calsequestrin, a component of the inositol 1,4,5-trisphosphate-sensitive Ca 2+ store of chicken cerebellum. Neuron 5, 713-21. Y A M A M O T O , A., OTSU, H., YOSHIMURI, T., MAEDA, N., MIKOSHIBA, K. & TASHIRO, Y. (1992) S t a c k s o f f l a t t e n e d
smooth endoplasmic reticulum enriched in inositol 1,4,5trisphosphate (Ins-P3) receptor in mouse cerebellar Purkinje cells. Journal of Cell Science, in press. YUAN, S., A R N O L D , W. & JORGENSEN, A. O. (1991) Biogenesis of transverse tubules and triads: immunolocalization of the 1,4-dihydropyridine receptor, TS28, and the ryanodine receptor in rabbit skeletal muscle developing in situ. Journal of Cell Biology 112, 289-301.