Distribution of calcium and potassium in presynaptic nerve terminals ...

Proc. Natl. Acad. Sci. USA Vol. 84, pp. 1713-1717, March 1987 Neurobiology

Distribution of calcium and potassium in presynaptic nerve terminals from cerebellar cortex (synaptic vesicles/electron probe analysis/elemental images)

S. BRIAN ANDREWS*, RICHARD D. LEAPMANt, DENNIS M. D. LANDISt, AND THOMAS S. REESE*§ *Laboratory of Neurobiology, National Institute of Neurological and Communicative Disorders and Stroke, and tBiomedical Engineering and Instrumentation Branch, Division of Research Services, National Institutes of Health, Bethesda, MD 20892; tDepartment of Developmental Genetics and Anatomy, Case Western Reserve University School of Medicine, Cleveland, OH 44106; and §Laboratory of Neurobiology, National Institute of Neurological and Communicative Disorders and Stroke at the Marine Biological Laboratory, Woods Hole, MA 02543

Communicated by Edwin J. Furshpan, November 24, 1986

taken on added importance with the recent indications that in neurons (14), as in a variety of other cell types (15), many regulatory cell functions may themselves be regulated by calcium release from internal stores. We have used electron probe elemental imaging (16) and electron probe microanalysis (17) to search for and to determine the elemental composition of calcium-rich organelles in resting and stimulated parallel fiber-Purkinje cell synapses in mouse cerebellar cortex. This tissue permits direct cryofixation of intact in situ synapses from brain slices in which neither the pre- and postsynaptic neurons nor the freezing surface have been cut (18). Elemental imaging of this preparation affords an exhaustive unbiased survey, covering large expanses of neuropil, of the distribution of important tissue constituents such as potassium, calcium, and phosphorus down to the level of individual terminals. Electron probe analysis complements elemental imaging with precise measurements of element concentrations within specific organelles at the submillimolar level. The results indicate that the synaptic vesicles of the presynaptic axoplasm contain 5'a~ lowi;

Proc. Natl. Acad. Sci. USA 84

(1987)

microscope by means of a Gatan Model 626 cryotransfer device and then freeze-dried in the microscope. Electron and Elemental Imaging. Transmission electron microscopy of freeze-dried cryosections was carried out in a JEOL 100-CX electron microscope at a nominal stage temperature of -1450C. For elemental imaging, a modified Hitachi H700H electron microscope was used; the unique hardware and software features of this digitally controlled analytical microscope have been described (16). In brief, the instrument generates a scanning image by collecting, at each pixel of a digitally defined raster, a variety of user-selected signals during a specified dwell time. The digitized signals are then processed and stored in a computer before or while moving to the next pixel. X-rays from four predefined energy windows can be acquired simultaneously, digitally-filtered (17), and the central lobe of the filtered peak stored as representative of the amount of element in the illuminated microvolume. Typical experimental parameters were 5 nA beam current into a -40 nm spot; live dwell time, 200 msec/pixel at 10% spectrometer deadtime; 128 x 128 raster at 36 nm/pixel, for a field of 22 ,m2; collection time, 1.5 hr. The analysis temperature ranged from -105'C to 250C; temperature was not a major factor for the data reported here, although drift of the microscope cold stage (which is a function of temperature) does influence the resolution of x-ray images. Under these conditions, the characteristic x-ray count rate for K (intracellular concentration on the order of 100 mmol/liter wet tissue) was typically 15 counts/ pixel, corresponding to a detectability of -20 mmol/liter in a structure the size of a single 50-nm synaptic vesicle, and 10 mmol/liter in a 100-nm smooth-membrane cistern. For structures >1 gm in diameter, the detectability is better than 1 mmol/liter. Sensitivity for Ca is slightly better, but it is offset by overlap between calcium Ka and potassium Kp x-rays. Full details of the applications and limitations of this elemental imaging approach to biological cryosections will be published elsewhere (26). Electron Probe Microanalysis. Both the Hitachi and JEOL microscopes (the former equipped with a Tracor TN-5500 x-ray analysis system and the latter with a Kevex 7000 interfaced to a DEC PDP 11/34 minicomputer) were used for electron probe analysis in the point mode. A small raster approximately equivalent to a 50-nm diameter focused probe was actually used for excitation; this was estimated to encompass one to four synaptic vesicles, depending on the specific geometry of a given cluster. The paradigm for quantitative microanalysis has been described, both in general (17, 27) and as it specifically was used here (25). Of particular note is the use of the erythrocytes frequently encountered in brain capillaries as internal standards of known composition to estimate elemental concentrations as amount of total element per unit of analyzed volume-that is, mmol/liter of wet tissue (C.) (25); mmol/liter is approximately equal to mmol/kg wet weight, which is a unit often used to express electron probe results. Values obtained this way depend on the following assumptions: (i) the standards are compositionally homogeneous and accurately known; (it) the hydrated section thickness does not vary significantly; and (iii) differential shrinkage between subcellular compartments during freeze-drying can be neglected. Quantitation by the Hall method (28) gives results in mmol/kg dry weight (D.), which is related to C. according to the equation C, = (dry mass fraction) x Dx.

varicosities of parallel fiber axons and the postsynaptic spines of Purkinje cell dendrites in rapidly frozen preparations. There are many parallel fiber synapses within the well-frozen zone extending -10 gm in from the uncut surface. These synapses are readily recognized in unstained freeze-dried cryosections cut en face from rapidly frozen slices of cerebellar cortex (Fig. 1). Elemental imaging of Na, P, K, and Ca in cryosections from cerebellar slices frozen between 20-30 sec of the cessation of blood flow provided direct evidence that the neuropil had been little disturbed by excision and freezing (Fig. 2). Specifically, the distributions of P and K coincided with cellular structures as imaged by transmission electron microscopy, the intracellular concentration of K was high and that of Na was low within neural processes, and the K concentration over the whole neuropil (88 mmol/liter wet tissue, including extracellular space) was similar to that measured previously for fresh brain (29). The average Ca concentration (1.0 mmol/liter) was also appropriate for normal tissue, although there were discrete Ca-rich foci with concentrations as high as 30 mmol/liter, some of which were nonmitochondrial. However, the Ca-rich sites did not appear within identified presynaptic terminals (Fig. 2), which, based on the image data, had a Ca concentration of 0.8 ± 0.4 mmol/liter (Table 1). The physiological activity of the synapses in rapidly excised cerebellar slices is unknown, although it is likely that many synapses had been stimulated by intrinsic neuronal activity, while others were depolarized as a result of surgical injury and spreading depression. To examine synapses under better-defined conditions of rest and depolarization, slices were rapidly excised and incubated in a physiological saline. When the cortex was depolarized with 55 mM K after 1 hr in vitro, the distribution of elements was essentially the same as that in fresh cerebellum, including the presence of Ca-rich areas. In contrast, incubated slices directly frozen in the resting state lacked the Ca-rich areas, although they were otherwise similar; the Ca concentration of specifically iden-

RESULTS Electron and Elemental Imaging. Electron microscopy of freeze-fractured and freeze-substituted cerebellar cortex (18) has been used extensively to establish a picture of the shape and distribution of the synapses between the presynaptic

view shows clusters of mitochondria within branches of Purkinje cell dendrites (D) and synaptic terminals of vesicle-containing parallel fiber axons (arrowheads) on dendritic spines. The nucleus of a stellate cell is at the lower left. Transmission electron micrograph. (x 13,000.) (Inset) Cross-section of a presynaptic terminal, which can be identified by its content of synaptic vesicles (arrows). (x50,000.)

t

't aktA b a20 mmol/liter) are not present in the presynaptic terminals

1716

Neurobiology: Andrews et al.

Table 1. Potassium and calcium distribution in presynaptic terminals of cerebellar cortex

Fresh slab Whole terminals* Vesicle clusters

Mitochondriat

X; ~P

Proc. Natl. Acad. Sci. USA 84

Concentration Ca

No. of terminals

K

10 17 10

95 ± 6 97 ± 4 99 ± 6

0.8 ± 0.4 0.4 ± 0.1 1.2 ± 0.2

In vitro slice, resting Whole terminals* Vesicle clusters

7 103 ± 9 1.4 ± 0.7 11 97 ± 5 0.7 ± 0.2 Mitochondriat 6 125 ± 4 0.9 ± 0.2 All concentrations are expressed in mmol/liter wet tissue ± SEM. Except as noted, wet weight values were obtained by converting dry weight concentrations [mmol/kg dry weight, calculated from peak/ continuum ratios (17, 28) of point-mode electron probe data] using known (62.5 g/100 g for erythrocytes, 65 g/100 g for mitochondria) or continuum-derived values for the water content of cellular compartments. To achieve the stated precision, enough x-ray counts were collected from specific sites within a single terminal (typically over 500 sec) so that counting errors were insignificant (K) or small (Ca) relative to terminal-to-terminal variations. Therefore, the number of data points is equal to the number of terminals, and errors in K concentration reflect only biological variability, while errors in Ca are combined values dominated by biological errors. *Values from image data. For fresh slabs, in situ erythrocytes were used as internal standards, in which case the K concentration for whole brain slabs was 88 mmol/liter wet tissue. For in vitro slices, image data were quantitated by assuming that the wet weight K concentration over large areas of neuropil was equal to that of similar areas in fresh slices. tMitochondrial concentrations from point-probe data are consistent with concentrations derived from elemental images. This indicates that systematic quantitation errors between images and point-probe spectra are relatively small. See also ref. 26.

of either resting or stimulated cerebellar synapses. Analysis of oxalate-stabilized freeze-substituted preparations giso showed that Ca (or externally applied Sr) was never observed within synaptic vesicles, although occasional deposits were found in axonal mitochondria in strongly stimulated slices. To determine preciselythe ionic content of organelles, quantitative elemental analysis of synaptic vesicles and mitochondria within identified terminals in cryosections was carried out using point probes. This established 0.7 ± 0.2 and 0.9 ± 0.2 mmol/liter, respectively, as the calcium concentrations for synaptic vesicles and mitochondria in resting synapses, and showed that these values do not change significantly upon stimulation. These results imply that neither synaptic vesicles nor mitochondria could function as effective calcium buffers, because the Ca content of these organelles remains low even after intense stimulation. In those few instances when presynaptic organelles did accumulate some Ca, it was either in smooth endoplasmic reticulum, whose Ca-accumulating capability is well-established in several other neuronal systems (8, 20, 24), or in mitochondria. Mitochondrial accumulation may have been due to damage to the synaptic plasmalemma (31) or to excessive stimulation in vitro. These results do not support the view that synaptic vesicles are primary Ca buffers, as has been proposed for other synapses (6) on the basis of evidence requiring biochemical fractionation (5, 6, 19) or chemical fixation (7). In the present study, the calcium concentrations were measured after cryofixation only and are presumably closer to in vivo conditions. However, it remains to be determined whether the differences between our findings and those of others depend on technique or reflect differences between different types of synapses. Although synaptic vesicles do not appear to buffer large amounts of Ca, it is nevertheless interesting to consider

V

(1987) v

1.

0W v

FiG 3 (A Paale fie syas wit adnrtcsie()na cerbela slc rapidl frze afte deplarztinnaSrctiig and untie sections* were prepard. Deailofsynaptic stutr m

ar evdn nea th orgnlcyscindsrae(rage)

hr

synaptic vesicles(arrow) nor, inthepresentexamplewas SrAindthe ..,

FIG. oS (A) Parallel fiber synapse with a dendritic spine (S) in a cerebellar slice rapidly frozen after depolarization in a Sr-containing medium. Following cryosectioning, the tissue was freeze-substituted and unstained sections were prepared. Details of synaptic structure are evident near the original cryosectioned surface (triangles). There was no evidence, from either inspection or x-ray analysis, for Ca or Sr in the synaptic vesicles (arrow) nor, in the present example, was Ca or Sr concentrated in the dendritic spine. (B) Another presynaptic terminal of a parallel fiber shows small deposits of Sr (circle) in a mitochonsr(14, demonstrating that the intracellular Sr must have been at least transiently high. Nevertheless, no deposits of Ca or Sr were detected in the cloud of synaptic vesicles at left.(xer,but.)

whether they might store Ca for release during synaptic activity. There is now much evidence that Ca release from the smooth endoplasmic reticulum is an important and general feature of Ca regulation in numerous cell types (31), including neurons (14). The present data show that the Ca content of resting synaptic vesicles is only 700 cmol/liter, but this would be more than enough to raise free cytoplasmic Ca to a few micromolar if it were released during synaptic activity. There are numerous uncertainties in determining the precise amount of releasable Ca in synaptic vesicles, as well as in specifying the level of intraorganelle Ca that would be judged functionally significant. Nevertheless, we note that the local Ca concentration in synaptic vesicle clusters (0.7 ± 0.2 mmol/liter) is no more than the overall concentration in terminals (1.4 ± 0.7 mrnol/liter), implying that the synaptic vesicles have no special capacity to accumulate Ca. Further-

more, the Ca concentration of vesicle clusters is below that of any known storage organelle; for example, the Ca con-

Neurobiology: Andrews et al.

4-)

CaKO

4

3

keV FIG. 4. Cumulative energy-dispersive x-ray spectrum obtained by summing 17 spectra from vesicle-rich regions of freshly frozen presynaptic terminals, including those in Fig. 2B. An expanded region centered on the K and Ca peaks is shown. Original summed spectrum (upper trace) illustrates the statistical quality of the spectrum and the overlap between the K K and Ca Ka peaks. The same spectrum after subtracting the calculated amount of K (lower trace) indicates the goodness of fit between the calculated and experimental spectra and the peak/continuum ratio for a Ca peak equivalent to 400 umol/liter.

centration within the rough endoplasmic reticulum of hepatocytes can be estimated at 3 mmol/liter (32). The present results show that Ca-handling in cerebellar presynaptic terminals is characterized by relatively subtle ion movements, with dramatic changes occurring in only a small number of buffering organelles. This paradigm has parallels in postsynaptic regions of the molecular layer, where there is at least one nonmitochondrial site that accumulates readily detectable amounts of extracellular Ca in response to membrane depolarization (33). This site also appears to be a smooth membrane organelle, the characterization of which will be described elsewhere. We thank M. O'Connell for excellent technical advice and assisphotographic work. We also thank C. Fiori and C. Swyt for valuable discussions and assistance on the application of the digital-imaging electron microscope. This work was supported in part by National Institutes of Health Grant NS 23641 (to D.M.D.L.).

tance, and J. Murphy for his

1. Katz, B. (1968) The Release of Neural Transmitter Substances (Thomas, Springfield, IL).

Proc. Natl. Acad. Sci. USA 84 (1987)

1717

2. Miledi, R. (1973) Proc. R. Soc. London Ser. B 183, 421-425. 3. Katz, B. & Miledi, R. (1967) J. Physiol. (London) 192, 407436. 4. Blaustein, M. P., McGraw, C. F., Somlyo, A. V. & Schweitzer, E. S. (1980) J. Physiol. (Paris) 76, 459-470. 5. Michaelson, D. M., Ophir, I. & Angel, I. (1980) J. Neurochem. 35, 116-124. 6. Israel, M., Manarache, R., Marsal, J., Meunier, F. M., Morel, N., Frachon, P. & Lesbats, B. (1980) J. Membr. Biol. 54, 115-126. 7. Pappas, G. D. & Rose, S. (1976) Brain Res. 103, 362-365. 8. McGraw, C. F., Somlyo, A. V. & Blaustein, M. P. (1980) J. Cell Biol. 85, 228-241. 9. Lazarewicz, J. W., Haljamae, H. & Hamburger, A. (1974) J. Neurochem. 22, 33-45. 10. Somlyo, A. V., Shuman, H. & Somlyo, A. P. (1977) J. Cell Biol. 74, 828-857. 11. Bond, M., Kitazawa, T., Somlyo, A. P. & Somlyo, A. V. (1984) J. Physiol. (London) 355, 677-695. 12. Bulenda, D. & Gratzl, M. (1985) Biochemistry 24, 7760-7765. 13. Clemente, F. & Meldolesi, J. (1975) J. Cell Biol. 65, 88-102. 14. Neering, I. R. & McBurney, R. N. (1984) Nature (London) 309, 158-160. 15. Berridge, M. J. (1984) Biochem. J. 220, 345-360. 16. Leapman, R. D., Fiori, C. E., Gorlen, K. E., Gibson, C. C. & Swyt, C. R. (1984) Ultramicroscopy 12, 281-292. 17. Shuman, H., Somlyo, A. V. & Somlyo, A. P. (1976) Ultramicroscopy 1, 317-339. 18. Landis, D. M. D. & Reese, T. S. (1983) J. Cell Biol. 97, 1169-1178. 19. Schmidt, R., Zimmermann, H. & Whittaker, V. P. (1980) Neuroscience 5, 625-639. 20. Henkart, M. P., Reese, T. S. & Brinley, F. J., Jr. (1978) Science 202, 1300-1303. 21. Nordmann, J. J. & Chevalier, J. (1980) Nature (London) 287, 54-56. 22. Heuser, J. E., Reese, T. S., Dennis, M. J., Jan, Y., Jan, L. & Evans, L. (1979) J. Cell Biol. 81, 275-300. 23. Van Harreveld, A. & Crowell, J. (1964) Anat. Rec. 149, 381-386. 24. Ornberg, R. L. & Reese, T. S. (1980) Fed. Proc. Fed. Am. Soc. Exp. Biol. 39, 2802-2808. 25. Andrews, S. B., Mazurkiewicz, J. E. & Kirk, R. G. (1983) J. Cell Biol. 96, 1389-1399. 26. Fiori, C. E., Leapman, R. D., Swyt, C. R. & Andrews, S. B. (1987) Ultramicroscopy, in press. 27. Kitazawa, T., Shuman, H. & Somlyo, A. P. (1983) Ultramicroscopy 11, 251-262. 28. Hall, T. A. (1979) in Microbeam Analysis in Biology, eds. Lechene, C. P. & Warner, R. R. (Academic, New York), pp. 185-208. 29. Pappius, H. (1970) Riv. Patol. Nerv. Ment. 91, 311-322. 30. Miledi, R. (1966) Nature (London) 212, 1233-1234. 31. Somlyo, A. P. (1984) Nature (London) 309, 516-517. 32. Somlyo, A. P., Bond, M. & Somlyo, A. V. (1985) Nature (London) 314, 622-625. 33. Andrews, S. B., Leapman, R. D., Landis, D. M. D., Fiori, D. E. & Reese, T. S. (1984) J. Cell Biol. 99, 424a (abstr.).