Brain Research, 126 (1977) 397--425
397
© Elsevier/North-Holland Biomedical Press, Amsterdam - Printed in The Netherlands
Research Reports
RAT H I P P O C A M P A L N E U R O N S IN DISPERSED CELL C U L T U R E
GARY A. BANKER* and W. MAXWELL COWAN Department of Anatomy and Neurobiology, Washington University School of Medicine, St. Louis, Mo. 63110 (U.S.A.)
(Accepted September 9th, 1976)
SUMMARY An in vitro system has been developed for the study of isolated hippocampal neurons from 18- or 19-day rat fetuses. Following trypsinization the cells are plated out at low density on polylysine-treated coverslips in an enriched medium. The isolated neurons rapidly attach to the substrate and initiate process extension. Little reaggregation occurs and the number of non-neuronal cells present is minimal. Unless co-cultured with tissue explants the neurons survive for only a few days; in the presence of hippocampal explants the initial growth of the isolated cells is improved and their survival in culture is extended to about two weeks. Some of the cells in such cultures develop a characteristic branching pattern closely resembling that of maturing hippocampal pyramidal cells in vivo. There is a clear relationship between the stage of the cells' development and their growth in culture. Cells which had completed D N A synthesis about 48 h before dissociation, and which were in the process of migration to the cortical plate, survived best in our cultures. Early post-mitotic cells which were still within the ventricular zone and cells which had already reached the cortical plate grew poorly. This system should permit the study not only of process formation by these cells, but also of their capacity to form specific synapses in vitro and of the biochemical constituents of their surfaces.
INTRODUCTION Among the many unresolved problems in neurobiology two have come to assume particular prominence in recent years. The first concerns the morphogenetic * Present address: Department of Anatomy, Albany Medical College of Union University, Albany, N.Y. 12208, U.S.A.
398
Ent, Ent.-
=
Ss:lmcOl ..
Sep.
s, rad.
om.
s.ori.
I ' ~ ' ~ , - , - - Ent Com.
•
alveus
CA3
CA l
Fig. 1. A camera lucida drawing of two Golgi-impre~aated pyramidal cells, from fields CAz and C~s of the rat hippocampus to show some of the differences in the morphology of the cells from these two fields, and also the distribution of their principal afferent connections. Ent., entorhimtl affemnts; Com., commissural afferents; Sep., septal afferents; Den., afferents from the dentate gyrus; Bas., basket cell inputs; s. tool., s. lac., stratum lacunosum-moleculare; s. r a d , stratum radiatum; s. pyr., stratum pyramidale; S. ori., stratum oriens; Sch, Schaffer collaterals. (From Gottlieb and Cowan 13, with permission.)
factors responsible for the variety of complex shapes that neurons may have, while the second concerns the mechanisms that determine their distinctive connectivity. Both problems have been the subject of intensive study (reviewed in ref. 8 and 21), and a number of neuronal systems have been suggested as useful models for further analysis. One that may have considerable promise is the study of neurons grown in dispersed cell culture. To date this approach has been used most effectively for peripheral neurons, especially sympathetic ganglion cells which survive well in the presence of the nerve growth factor~,g, 19. On the whole, central neurons have proved to be more difficult to maintain when isolated from other nerve cells or associated glia, but recently some success has been obtained with this approachZl,l~, t4. As part of a continuing series of studies on the mammalian hippocampal formation, we have been interested to see if isolated pyramidal cells from the superior and inferior regions of the hippocampus of fetal rats could be isolated and maintained in culture, since in vivo they display a number of features of morphogenetic interest. First, the overwhelming majority of the cells from any one field are of one type - namely, pyramidal cells with distinctive apical dendrites, several basal dendrites and a single axon emerging from the base of the perikaryon. Second, the pyramidal cells from the superior and inferior regions of the hippocampus (corresponding more or less to fields CAt and CA3 of Lorente de N616) are sufficiently different in size and appearance that they can be readily distinguished both in Nissl,stained and Golgi preparations. Third, both types of pyramidal cell receive a characteristic pattern of afferent connections, and each major class of afferents terminates preferentially upon one or another part of the cell surface (see Fig. 1).
399 The present experiments were undertaken in an attempt to answer 3 questions. (1) Can hippocampal pyramidal cells from fetal rats be maintained in dispersed culture under conditions suitable for studying the growth of their processes? (2) Is there an optimal period in the life of the developing hippocampal neurons when they best survive the isolation procedure? (3) To what extent can the cells reproduce their characteristic morphology in vitro and, in particular, is it possible to identify cells in dispersed culture which display some, or most, of the distinguishing morphological features of neurons from the superior and inferior regions of the hippocampus ? In a later paper we shall consider how successfully these cells can form synaptic connections and to what extent they retain their connectional specificity in vitro. For these purposes it was important to develop a culture system in which individual cells and their processes could be easily observed at all stages of their development in vitro. Most procedures which have so far been described for culturing central neurons employ high cell densities in order to enhance long-term cell survival. This inevitably leads to extensive cell aggregation, and the processes from individual cells are seldom distinguishable until the cell aggregates begin to thin out after several days in culture. To avoid this, we have concentrated our efforts on developing a system in which the neurons could be maintained at low cell densities. Furthermore, we considered it important that the attachment and growth of the cells and their processes take place directly on the culture substrate rather than on the surface of non-neuronal cells. Under the latter conditions, the active outgrowth of neural processes may be confounded by their passive stretching through attachment to migrating non-neuronal cells. In most CNS culture systems that have been described, the neurons either initially adhere to the surface of flattened non-neuronal cells or are rapidly "undergrown" by the proliferation of such cells. Prompted by the reports of Letourneau 15 and Yavin and Yavin 2a, we have found that polylysine-treated substrates offer an improved surface for cell adhesion and allow reproducibly-good neurite outgrowth from low density cultures. So far we have been successful in maintaining cultures for up to two weeks, which has proved adequate for the analysis of the growth of the pyramidal cell processes and of the overall morphology of the cells. MATERIAL AND METHODS Culture media The balanced salt solutions and culture medium used were designed to equilibrate with air. They were based on a Hank's balanced salt solution with reduced NaHCO8 concentration (1 mM) and buffered with 10 mM Hepes (N-2-hydroxyethylpiperazine-NZ-2-ethane sulfonic acid). During the dissection and dissociation procedures the tissues were kept in a Caand Mg-free Hank's balanced salt solution (prepared from Gibco concentrate No. 418) modified to contain 1 mM NaHCOs, 0.6~o glucose, 100 U/ml penicillin, 100/~g/ml streptomycin, and 10 mM Hepes, pH 7.3 (hereafter referred to as BSS). The base for the culture medium was Eagle's minimum essential medium with Hank's
400
salts (prepared from Gibco concentrate 158H) containing twice the standard amounts of essential and non-essential amino acids (Gibco No. 113 and 114) as well as the above additions. Eight parts of this solution were diluted with one part of H20 to give a solution of 295 mOsM (hereafter referred to as MEM). Complete culture medium contained 15 parts MEM, 5 parts human placental serum and 2 parts 9-day chick embryo extract. In some early experiments, when this medium was applied to polylysine-treated culture dishes and incubated at 36 °C, a dense precipitate formed which obscured the cells that were present. To avoid this, in all subsequent experiments, the complete medium was heated to 60 °C for 30 rain, cooled and filtered through a sterile 0.45 # m filter before use. Some batches of medium gave only poor cell growth or failed to support growth altogether. We attribute this to variations in the human placental serum available to us. Routinely, each new batch of medium was first tested with a few control cultures to be sure it gave satisfactory results.
Preparation of culture dishes Special culture dishes, similar to those developed by Bray 2 and Bunge and Wood 6, were used for all the experiments. These were prepared by machining holes 16 mm in diameter in the bottom of 12 mm × 50 mm polystyrene culture dishes (Falcon No. 1006). Acid-cleaned, 22 mm diameter, No. 1½ cover glasses (Clay-Adams "Gold Seal") were then attached to the outer surface of the dish, over the hole, with a paraffinvaseline (3:1) mixture. This formed a well about 1 mm deep by 16 mm in diameter ( ~ 2.0 sq.cm in area) over the coverslip which formed the culture surface. In some experiments a thin layer of carbon was evaporated onto the coverslip before attaching it to the dish in order to permit subsequent embedding for electron microscopy 22. To permit observations with differential interference contrast (Nomarski) optics, the lids of the culture dishes were also modified by making a 30 mm diameter hole in the plastic and sealing a No. 2 coverslip over this with silastic medical adhesive (Type A, Dow Coming). These dishes allowed excellent observation of the living cells under an inverted microscope with a long-working distance condenser and up to 100 × oil immersion objectives. Before use, the dishes were sterilized in 70 ~o ethanol and dried in a sterile hood. Then a 0.1% solution of poly-L-lysine hydrobromide (Sigma Type VIIB) in 0.1 M borate buffer, pH 8.4, was applied to the well for about 24 h 1~. The dishes were then rinsed twice with sterile water (over a 24-h period) and dried. Before use (generally within 3 days of preparation), they were again rinsed twice, this time with MEM, and 0.2 ml of the complete medium was added. Dishes containing medium were stored in an incubator at 35-36 °C for up to 3 days before the cells were applied.
Preparation of the cells Cells were taken from 18- or 19-day-old fetuses from our own colony of Holtzman rats. Timed pregnancies were obtained by daily checking vaginal washings for sperm, the day on which sperm were found being regarded as day 0. At the appropriate stage of gestation the pregnant rats were anesthetized with chloral hydrate and the
401
Fig. 2. An enlarged view of the medial aspect of the right cerebral hemisphere of a 19-day rat fetus. The large arrows mark the location of the hippocampal fissure, and the small arrows the developing fimbria. On the right is shown (at the same scale) the dissected hippocampus from which the meninges have been removed. Again, the large and small arrows indicate the hippocampal fissure and the fimbria, respectively. uterus removed to a sterile dish. The remainder of the cell preparation was performed in a sterile hood. The brains were removed from the fetuses with a pair of fine scissors, and the cerebral hemispheres separated from the brain stem. When the hemisphere of the 18or 19-day-old fetus is viewed in a dissecting microscope the hippocampus can be clearly seen on its medial surface as shown in Fig. 2 (left). The hippocampal fissure, usually marked by a conspicuous group of blood vessels, indicates the approximate junction between the hippocampus and the adjoining subicular and entorhinal cortex. The developing fimbria is seen as a white translucent band along the free margin of the hippocampus. Before separating the hippocampus from the hemisphere, the meninges and adherent choroid plexus were carefully pulled off with fine forceps. At this stage the full depth of the hippocampal fissure can be seen. Then with iridectomy scissors the hippocampus was separated from the adjoining cortex by a cut parallel to the hippocampal fissure, and by transverse cuts at its rostral and caudal ends. The appearance of the hippocampus, after dissection, is shown in Fig. 2 (right). The method of cell dissociation is based on the procedure described by Fischbach 10. Eight to 24 hippocampi were placed in 5.0 ml of Ca- and Mg-free BSS containing 0.1 ~ trypsin (Difco 1:250) for 15 min at 37 °C. The intact hippocampi were then rinsed 3 times (5 min each) in BSS, and the cells dissociated by repeated passings through a fire-polished Pasteur pipette until no visible lumps of tissue
402 remained. The cells were then counted in a hemocytometer and their viability assayed by their exclusion o f t r y p a n blue or erythrosine B. Except as otherwise noted, 30,00035,000 viable cells in 15-40 #1 of BSS were added to the center of previously prepared culture dishes containing about 0.2 ml of medium. They were then incubated at 35-36 °C, in air saturated with H20 vapor. In some experiments, dissociated pyramidal cells were co-cultured with explants from the hippocampus at the same stage of development. The hippocampal explants were prepared by cutting slices about 0.5 mm thick in a plane transverse to the long axis of the fetal hippocampi. About 6 slices were placed in a culture dish and incubated until they had become firmly attached to the surface (usually 3 or 4 days). At that time dissociated cells, prepared as described above, were added and the cultures were returned to the incubator.
Observations Living cultures were examined with an inverted microscope using either phase contrast or differential interference contrast optics. Fixed cultures were prepared by slowly adding warm 4% glutaraldehyde in 0.1 M phosphate buffer at pH 7.2. After 30-60 min, this fixative was removed and fresh gtutaratdehyde added for a further hour. Permanent mounts for phase contrast or differential interference contrast observation were prepared by removing the coverslips from the dishes with the cells still attached to them. After rinsing with BSS the coverslips were mounted on glass slides in buffered 50 % glycerol and ringed with Permount. For silver staining, the coverslips were removed, dehydrated in a graded series of alcohols and placed in xylene to remove any residual paraffin adhesive. Following rehydration, the cells were stained by a modified Bielschovsky method 24. For the quantitative analyses of cell survival, the neurons were counted in square areas 307 pm on a side from 35 to 40 representative regions of the dish. When counts were made on living cultures, cell attachment was checked by lightly tapping the microscope stage. In fixed cultures the unattached cells were removed by rinsing. Cells with neurites were counted only if they had at least one process over one cell diameter in length and tipped with a growth cone. There are quantitative differences in growth between one cell preparation and another, and between one batch of medium and another. For this reason all comparisons were made between experiments using the same preparation of cells and the same medium. In all cases the results illustrated are representative of a series of 2, 3 or 4 independent experiments.
Thymidine labeling In order to determine the relationship between the time of cell origin (i.e., the time when the cells become post-mitotic) and their survival in culture, each of a series of pregnant rats was given a single intraperitoneal injection of [3H]thymidine (5 #Ci/g body weight, specific activity 20 Ci/mmole, New England Nuclear) at intervals between day 15 and day 19 of gestation. All the animals were sacrificed on day 19 and the brains from the fetuses were used to prepare a series of hippocampal cell cultures. The cultures from each series were fixed shortly after cell attachment (1.5-3.5 h after
403 plating) or after 1, 2, or 3 days growth in culture. In addition, the brains of two fetuses from each pregnant rat were used to localize the labeled cells within the intact hippocampal formation. These brains were fixed in 70 ~o ethanol and 2 % acetic acid, embedded in paraffin, serially sectioned in the frontal or horizontal planes at 7 pm, and processed for autoradiography as described by Cowan et alP. After an exposure time of two weeks, the autoradiographs were developed in D19 at 17 °C, fixed in Ektaflo and stained through the emulsion with thionin. The fixed cultures were dehydrated, cleared in xylene, then subsequently rehydrated, dried, and mounted on slides, with the cells facing upwards. They were then coated with undiluted Kodak NTB-2 emulsion, exposed for two weeks at 4 °C, developed in DI9 at 17 °C, and mounted in 50% glycerol under a second coverslip. Each culture was examined to determine the percentage of labeled cells present, different regions of the coverslip being systematically sampled. Only the heavily labeled cells were considered for the analysis; lightly labeled cells (overlain by fewer than 10 grains) were included with the unlabeled cells in this analysis. For cultures fixed before the initiation of neurite outgrowth, large and small cells were counted separately. For cultures fixed after 1-3 days growth, only those cells were counted that had clearly developed neurites more than 30/zm in length and were spatially distinct from the processes of neighboring cells. RESULTS
(1) The appearance of the cells immediately after dissociation After brief trypsinization the hippocampus from 18- or 19-day rat fetuses can be almost completely dissociated into single cells. No visible fragments of tissue remain and cell suspensions contain only an occasional cluster of 2-3 cells. The hippocampus from each hemisphere yields about 500,000 cells*. Of these, 85-90~o are viable, as judged by dye exclusion. Fig. 3 shows the appearance of the cells about 2 h after dissociation, before they have begun to form processes. At this stage about 80 % of the cells are small and spherical (with a mean diameter of about 8 pm). The remainder are distinctly larger and are typically pyramidal or ovoid in shape. The larger cells often retain one or more fairly lengthy processes, but the small cells are usually without processes immediately after dissociation. Some of the differences between the two classes of cells with regard to their time of origin and their survival in vitro will be considered below. But most of the findings to be described in this paper refer to the smaller cells which comprise the dominant population in these cultures. (2) Adhesion of the cells to the substrate After plating, the cells rapidly adhere to the surface of the polylysine-treated coverslips. As they adhere they often exhibit filiform or flattened membranous ex* This figuremay be comparedwith the figure of 450,000 found in a recent count of the number of pyramidalcells in the mature rat hippocampus(Schlessingerand Cowan,unpublishedobservations).
404
Fig. 3. The appearance of hippocampal cells shortly after dissociation. In this preparation, the cells were allowed to attach to a polylysine-treatedcoverglassand were photographed 2 h after dissociation. Two types of cells are present: a population of small, round cells, and a population of larger, more elliptical cells (arrows). The latter often show one or more short processes after dissociation. (Differential interference contrast micrograph of cells fixed in glutaraldehyde and mounted in 50 ~ glycerol.) tensions at one or more points along the margins of the cell perikarya which serve to increase the area of cell surface that is in contact with the substrate. This can be seen in some of the cells shown in Figs. 3 and 5. Under our usual conditions of culture, with an initial density of about 15,000 cells/sq.cm, the neurons adhere directly to the substrate, and there is essentially no adhesion of neurons to non-neuronal cells. Furthermore, because adhesion is rapid, neurons usually attach to the surface before any significant reaggregation can occur. This makes it possible to determine with reasonable accuracy the number of neurons that attach to the surface and, subsequently, the number of cells that develop neurites, Fig. 4 illustrates the extent of cell adhesion as a function of time after applying the cells to the culture dishes. It is evident from this that some of the cells become firmly attached within 10 rain and adhesion is essentially complete within 0.5-2 h. The number of cells that adhere to the surface varies from experiment to experiment but, when measured at 24 h, has been found to average about 60 ~ of the viable cells applied. These results are comparable to those of Yavin and Yavin 29 who examined the adhesion of cells from the cerebral hemispheres of rat embryos to polylysine-treated plastic petri dishes. They found that adhesion was complete within 0.5 h and that less than 25 O//oof the applied cells failed to adhere.
(3) Initial neurite outgrowth Many of the cells in our cultures show no indication of neurite outgrowth after adhering to the culture surface. On the other hand, some cells begin to extend processes with distinct growth cones within a few hours after plating. These cells with actively growing neurites can be distinguished from the larger cells which retain processes after dissociation since the processes of the latter lack growth cones and are seldom attached
I0
--
5O
I.
30
~ 2o -,
10
I
2
3
4
TIME IHRSJ
Fig. 4. The kinetics of cell adhesion to polylysine-treated coverglasses. At the times indicated culture dishes were rinsed with BSS and then fixed with glutaraldehyde. The numbei of attached cells was counted and expressed as a percentage of the number of viable cells originally applied. The results from two experiments are shown, each point being the mean of two or three cultures from one experiment. Cell attachment is half-maximal within 20 rain and is essentially complete within 1-2 h.
Fig. 5. The morphology of hippocampal cells after one day in culture. A: a low-power photomicrograph of a cell culture fixed and stained after 24 h in vitro with a reduced silver method. The isolated cells, which are attached directly to the substrate, have extended one or two unbranched neurites. Several cells which have not developed processes are also present. Scale- 50 pro. B, C: photomicrographs of typical isolated cells from such cultures. The cell in B was photographed in a living culture by phase contrast optics whereas C is taken from a fixed, silver-stained preparation. Note the appearance of the major and minor processes and the prominent growth cones at the tips of the major processes. Scale in B and C: 10 pm.
406
~,~ /
cells adhering 60
--
/
/
I
-
/I
1.8
assre~e/ size /
_
/
1.6
/
cells with
7"/ 1.2
I 7,500
I 15,000
PLATING OENSITY
I' 22;500
I
1.0
3O,OOO
(CELLS/CId 2)
Fig. 6. The effect of cell density on cell attachment, neurite formation and re,aggregation. Three cultures at each cell density were fixed after 24 h in vitro. The total number of cells attached to the surface (open circles) and the number of cells with neurites longer than twice the cell diameter (filled circles) were counted and expressed as a percentage of the number of viable ceUs initially applied. Cell density has little effecton cell attachment, but the percentage of cells that extend neurites increases markedly with increasing density. As a measure of the extent of cell reaggregation, mean aggregate size was determined using the following formula: mean aggregate size= [no. of single cells and no. of aggregates] ÷ total no. of cells present. If only single cells were present, the mean aggregate size was regarded as 1. Extensive cell aggregation (open squares and hroken line) occurs at higher cell densities. The cell densities shown on the abscissa refer to the number of viable cells/ sq.cm that were initially applied. to the culture surface at their tips. The initiation of neurite extension generally occurs within the first 24 h in culture. The appearance of several typical cells at this stage is shown in Fig. 5A which is a low-magnification photomicrograph of a culture fixedafter 24 h growth and stained by a reduced silver method. The cell perikarya are usually spherical or somewhat elliptical in shape and, under phase contrast, appear refractil¢. Usually one or two major processes can be seen extending from each perikaryon. These follow a rather meandering course and average about 70 ~m in length, but many are over 150/~m long. The processes are usually unbranched and have a characteristic growth cone at their tips (Fig. 5B, C). Typically, the growth cone takes the form of a-flattened membranous expansion which can be seen to undulate fairly rapidl3~. On occasion, microspikes can be seen extending from the expanded growth ctmes. Laterally extending fine branches, or microspikes, are also seen at intervals along the processes between the cell body and the tips of the process. A second class of process, which we shall term "minor processes", also commonly extends from the cell body (Fig. 5B, C). These are distinctly shorter than the major processes and seldom exceed 2 0 / a n in length. They vary in form from flattened membranous extensions, through thin
407 TABLE I Cell attachment and neurite growth
The extent of cell attachment and of neurite growth after 24 h in culture. The results shown represent 5 separate cell preparations grown in 7 lots of medium. Cell densities ranged from 15,000 to 17,500 cells/sq.cm. Exp. no.
Cells adhering ( % of those applied)
Cells with neurites ( % of those adhering)
Cells with neurites ( % of those applied)
1
2 3 4 5a 5b 5c
64 65 61 43 69 50 43
16 30 34 26 33 46 44
11 24 21 11 23 24 19
Median
61
33
21
elongated processes resembling microspikes, to stouter processes resembling short neurites. The significance of these minor processes, and their relationship to the major processes, is not clear. (4) Cell density and cell growth
Cell density has a major effect on the growth of neurons in culture. This is illustrated in Fig. 6, which shows the results from a typical experiment in which 2 or 3 cultures were grown at each of 4 different cell densities. All these cultures were fixed after 24 h o f growth and the number of cells with, and without, neurites in each culture determined. At all cell densities examined, the total number of cells that were present was a more or less constant percentage of the number of viable cells applied; in this experiment about 60 % of the cells had survived for 24 h. Cell adhesion thus appears to be independent of cell density, at least over the range of densities we have studied. However, neurite outgrowth is critically dependent on the number of cells present. When the initial cell density is below about 3500 cells/sq.cm (,-~ 35,000 cells/1.0 ml medium), less than 12% o f the applied cells extend neurites. With increasing density the number of neurites increases progressively so that at an initial density of 15,000 cells/sq.cm, 32 % of the cells have neurites. At higher densities this percentage increases only slightly. Unfortunately, although neurite growth increases with increasing cell density, cell aggregation increases concomitantly. This is also shown in Fig. 6 where the extent of cell aggregation is examined as the mean number of cells per aggregate .At densities of about 15,000 cells/sq.cm and lower, cell aggregates are infrequent, and the mean aggregate size is less than 1.3. Further increases in density greatly increase the number and size of aggregates. At 30,000 cells/sq.cm, mean aggregate size is about 1.9, and at higher densities aggregation is so extensive that accurate cell counts cannot be made. There is a rather narrow range o f cell densities, around 15,000-20,000 cells/sq.cm,
408 80
60
4o
==~
neurites
0
I 1
I 3 DAYS
I 5 IN
! 7
CULTURE
Fig. 7. Cell growth during one week in culture. Two cultures were examined daily, and the total number of cells attached to the substrate (open circles) and the number of cells with neurites (filled circles) were determined. Non-neuronal cells were excluded from the computations. The number of cells that have not developed neurites declines sharply during the second and third days in culture, whereas the number of cells with neurites declines only slightly during this period. The number of cells with neurites begins to fall on the third and fourth days and such cells have virtually disappeared by the seventh day. that permit reasonable neurite outgrowth without causing extensive cell aggregation. For this reason this density was chosen for most of the subsequent experiments. The extent of process growth obtained in a series of such experiments is shown in Table I. In 5 of the 7 experiments shown, between 19 % and 24 9/00of the viable cells developed neurites in culture. In two other experiments growth was distinctly worse, either because of poor cell adhesion (experiment 4) or because of poor fiber outgrowth from adherent cells (experiment 1). Such poor results are probably attributable to the particular batches of serum employed, and were later eliminated by discarding those serum lots that failed to support growth above the chosen criterion.
(5) Neuronal and non-neuronal cells When hippocampal cells are plated at a density of about 15,000/sq.cm, the resulting cultures are remarkably free of obviously non-neuronal cells. We do observe small numbers of flattened epithelial-like or fibroblast-like cells and occasionally a well-differentiated astrocyte. Such non-neuronal cells represent only about 2 ~o of the cells present in cultures after 24 h, This presumably reflects the low number of nonneuronal cells in the initial cell suspension. This number can be rougJhly estimated by determining the number o f cells that rapidly adhere to untreated tissueeulture dishes. We find that less than 1 9/00of the cells viable after dissociation attach to untreated Falcon tissue culture dishes within 2 h. Nearly all of these cells spread rapidly on the culture surface and assume an epithelial,like appearance.
409 ~
40
B
cells with
-~
30
-
s ,.~
20
-
10
-
Z
~'~
cells
2
4 DAYS
6 IN
i ///
i
8
14
CULTURE
Fig. 8. The effect of hippoeampal explants on the growth and survival of dissociated hippoeampal neurons. The number of dissociated neurons which had extended processes was determined on successive days in cultures containing dissociated cells alone (open circles) or dissociated cells together with the hippocampal explants (filledcircles). In the latter instance 6 hippocampal explants from 18-day rat fetuses were allowed to grow in culture for 4 days before adding the dissociated cells. Note that the co-culture with explants increases the percentage of cells that develop neurites and markedly enhances their survival in culture. Initially 47,000 viable cells were applied to each of 3 dishes in each condition.
In most of our cultures there are also a large number of cells which, although they neither extend processes like neurons nor spread out on the culture surface like epithelial cells, closely resemble neurons in their size, shape and affinity for silver stains (Fig. 5A). Most probably, these cells are neurons which for one reason or another are incapable of growth under our culture conditions. As we shall see later, many of these cells are generated earlier than those which extend processes (see Section 10).
(6) Long-term cell survival Most of these cultures have been studied for 5 or more days in vitro. The nerve cells continue to extend processes during the first 2 or 3 days in culture, but thereafter there is a progressive loss of cells which is essentially complete by 7 days. The survival of neurons in a typical experiment is shown in Fig. 7. In this experiment two cultures were examined at intervals, and the numbers o f cells with and without neurites counted. The number of cells with neurites remains fairly constant during the first 2 or 3 days in culture, but there is an obvious decrease in the number of cells which lack processes. Beyond the third day, virtually all o f the surviving cells have neurites. A similar selective loss of neurons that fail to form processes has been observed in cultures of mouse spinal ganglia 27. After the third day there is a rapid loss of neurons, and by day 7 less than 1% of the cells initially applied remain viable. The p o o r long-term survival in our cultures is probably related to the low cell densities used. One possible solution to this problem is to co-culture the dissociated cells with tissue explants. We have performed a number of experiments in which thin
410 ..........
..............
Fig. 9. A : a photomicrograph of hippocampal neurons after one week in culture. Note that the cell bodies have increased in size, and that the cells are interconnected by a rich fiber network. A single non-neuronal cell is also present (arrow). (Preparation fixed in glutaraldehyde, mounted in 50~o glycerol, and photographed under phase contrast.) Scale: 50 t~m. B: a hippocampal neuron maintained in culture for 11 days. Note the pyramidal-shaped cell body, the stout apical process which bifurcates near the cell body, and 3 slender basal processes. In these respects the cell resembles a typical young pyramidal cell from field CAz of the hippocampus. (Fixed in glutaraldehyde and stained with reduced silver.) Scale: 25 Mm.
slices of fetal rat hippocampi were co-cultured with dissociated hippocampal cells. In effect this procedure increases the total number of cells present in the cultures by about an order of magnitude, while retaining the advantage of having isolated cells for individual analysis. Except in the area immediately adjacent to the explants, free, dissociated nerve cells are readily recognized. Our experience with this method is limited, but in some cultures cell survival has been considerably improved. Fig. 8 illustrates the influence of hippocampal explants on the growth and survival of dissociated hippocampat neurons. Co-culture with explants both increases the percentage of cells that initially develop neurites and markedly prolongs the survival of the neurons in culture. Fig. 9A shows a small group of neurons cultured for 1 week under these conditions. Many of the cells have increased appreciably in size and have come to resemble maturing neurons (note especially the prominent nueleoli). Long processes extend from the cells and intermingle with the processes of other dissociated neurons to form a rather dense plexus. Neurites of cells within the explant and the non-neuronal cells do extend a short distance from the explants. However, in regions such as that illustrated that lie at some distance from the explants, the number of non-neuronal cells is quite low. Cell counts indicate that in most areas neurons still outnumber non-neuronal cells by about 4:1 and that less than 3 ~ of the culture surface is occupied by the latter,
411
k
o o Oog
oH
397
© Elsevier/North-Holland Biomedical Press, Amsterdam - Printed in The Netherlands
Research Reports
RAT H I P P O C A M P A L N E U R O N S IN DISPERSED CELL C U L T U R E
GARY A. BANKER* and W. MAXWELL COWAN Department of Anatomy and Neurobiology, Washington University School of Medicine, St. Louis, Mo. 63110 (U.S.A.)
(Accepted September 9th, 1976)
SUMMARY An in vitro system has been developed for the study of isolated hippocampal neurons from 18- or 19-day rat fetuses. Following trypsinization the cells are plated out at low density on polylysine-treated coverslips in an enriched medium. The isolated neurons rapidly attach to the substrate and initiate process extension. Little reaggregation occurs and the number of non-neuronal cells present is minimal. Unless co-cultured with tissue explants the neurons survive for only a few days; in the presence of hippocampal explants the initial growth of the isolated cells is improved and their survival in culture is extended to about two weeks. Some of the cells in such cultures develop a characteristic branching pattern closely resembling that of maturing hippocampal pyramidal cells in vivo. There is a clear relationship between the stage of the cells' development and their growth in culture. Cells which had completed D N A synthesis about 48 h before dissociation, and which were in the process of migration to the cortical plate, survived best in our cultures. Early post-mitotic cells which were still within the ventricular zone and cells which had already reached the cortical plate grew poorly. This system should permit the study not only of process formation by these cells, but also of their capacity to form specific synapses in vitro and of the biochemical constituents of their surfaces.
INTRODUCTION Among the many unresolved problems in neurobiology two have come to assume particular prominence in recent years. The first concerns the morphogenetic * Present address: Department of Anatomy, Albany Medical College of Union University, Albany, N.Y. 12208, U.S.A.
398
Ent, Ent.-
=
Ss:lmcOl ..
Sep.
s, rad.
om.
s.ori.
I ' ~ ' ~ , - , - - Ent Com.
•
alveus
CA3
CA l
Fig. 1. A camera lucida drawing of two Golgi-impre~aated pyramidal cells, from fields CAz and C~s of the rat hippocampus to show some of the differences in the morphology of the cells from these two fields, and also the distribution of their principal afferent connections. Ent., entorhimtl affemnts; Com., commissural afferents; Sep., septal afferents; Den., afferents from the dentate gyrus; Bas., basket cell inputs; s. tool., s. lac., stratum lacunosum-moleculare; s. r a d , stratum radiatum; s. pyr., stratum pyramidale; S. ori., stratum oriens; Sch, Schaffer collaterals. (From Gottlieb and Cowan 13, with permission.)
factors responsible for the variety of complex shapes that neurons may have, while the second concerns the mechanisms that determine their distinctive connectivity. Both problems have been the subject of intensive study (reviewed in ref. 8 and 21), and a number of neuronal systems have been suggested as useful models for further analysis. One that may have considerable promise is the study of neurons grown in dispersed cell culture. To date this approach has been used most effectively for peripheral neurons, especially sympathetic ganglion cells which survive well in the presence of the nerve growth factor~,g, 19. On the whole, central neurons have proved to be more difficult to maintain when isolated from other nerve cells or associated glia, but recently some success has been obtained with this approachZl,l~, t4. As part of a continuing series of studies on the mammalian hippocampal formation, we have been interested to see if isolated pyramidal cells from the superior and inferior regions of the hippocampus of fetal rats could be isolated and maintained in culture, since in vivo they display a number of features of morphogenetic interest. First, the overwhelming majority of the cells from any one field are of one type - namely, pyramidal cells with distinctive apical dendrites, several basal dendrites and a single axon emerging from the base of the perikaryon. Second, the pyramidal cells from the superior and inferior regions of the hippocampus (corresponding more or less to fields CAt and CA3 of Lorente de N616) are sufficiently different in size and appearance that they can be readily distinguished both in Nissl,stained and Golgi preparations. Third, both types of pyramidal cell receive a characteristic pattern of afferent connections, and each major class of afferents terminates preferentially upon one or another part of the cell surface (see Fig. 1).
399 The present experiments were undertaken in an attempt to answer 3 questions. (1) Can hippocampal pyramidal cells from fetal rats be maintained in dispersed culture under conditions suitable for studying the growth of their processes? (2) Is there an optimal period in the life of the developing hippocampal neurons when they best survive the isolation procedure? (3) To what extent can the cells reproduce their characteristic morphology in vitro and, in particular, is it possible to identify cells in dispersed culture which display some, or most, of the distinguishing morphological features of neurons from the superior and inferior regions of the hippocampus ? In a later paper we shall consider how successfully these cells can form synaptic connections and to what extent they retain their connectional specificity in vitro. For these purposes it was important to develop a culture system in which individual cells and their processes could be easily observed at all stages of their development in vitro. Most procedures which have so far been described for culturing central neurons employ high cell densities in order to enhance long-term cell survival. This inevitably leads to extensive cell aggregation, and the processes from individual cells are seldom distinguishable until the cell aggregates begin to thin out after several days in culture. To avoid this, we have concentrated our efforts on developing a system in which the neurons could be maintained at low cell densities. Furthermore, we considered it important that the attachment and growth of the cells and their processes take place directly on the culture substrate rather than on the surface of non-neuronal cells. Under the latter conditions, the active outgrowth of neural processes may be confounded by their passive stretching through attachment to migrating non-neuronal cells. In most CNS culture systems that have been described, the neurons either initially adhere to the surface of flattened non-neuronal cells or are rapidly "undergrown" by the proliferation of such cells. Prompted by the reports of Letourneau 15 and Yavin and Yavin 2a, we have found that polylysine-treated substrates offer an improved surface for cell adhesion and allow reproducibly-good neurite outgrowth from low density cultures. So far we have been successful in maintaining cultures for up to two weeks, which has proved adequate for the analysis of the growth of the pyramidal cell processes and of the overall morphology of the cells. MATERIAL AND METHODS Culture media The balanced salt solutions and culture medium used were designed to equilibrate with air. They were based on a Hank's balanced salt solution with reduced NaHCO8 concentration (1 mM) and buffered with 10 mM Hepes (N-2-hydroxyethylpiperazine-NZ-2-ethane sulfonic acid). During the dissection and dissociation procedures the tissues were kept in a Caand Mg-free Hank's balanced salt solution (prepared from Gibco concentrate No. 418) modified to contain 1 mM NaHCOs, 0.6~o glucose, 100 U/ml penicillin, 100/~g/ml streptomycin, and 10 mM Hepes, pH 7.3 (hereafter referred to as BSS). The base for the culture medium was Eagle's minimum essential medium with Hank's
400
salts (prepared from Gibco concentrate 158H) containing twice the standard amounts of essential and non-essential amino acids (Gibco No. 113 and 114) as well as the above additions. Eight parts of this solution were diluted with one part of H20 to give a solution of 295 mOsM (hereafter referred to as MEM). Complete culture medium contained 15 parts MEM, 5 parts human placental serum and 2 parts 9-day chick embryo extract. In some early experiments, when this medium was applied to polylysine-treated culture dishes and incubated at 36 °C, a dense precipitate formed which obscured the cells that were present. To avoid this, in all subsequent experiments, the complete medium was heated to 60 °C for 30 rain, cooled and filtered through a sterile 0.45 # m filter before use. Some batches of medium gave only poor cell growth or failed to support growth altogether. We attribute this to variations in the human placental serum available to us. Routinely, each new batch of medium was first tested with a few control cultures to be sure it gave satisfactory results.
Preparation of culture dishes Special culture dishes, similar to those developed by Bray 2 and Bunge and Wood 6, were used for all the experiments. These were prepared by machining holes 16 mm in diameter in the bottom of 12 mm × 50 mm polystyrene culture dishes (Falcon No. 1006). Acid-cleaned, 22 mm diameter, No. 1½ cover glasses (Clay-Adams "Gold Seal") were then attached to the outer surface of the dish, over the hole, with a paraffinvaseline (3:1) mixture. This formed a well about 1 mm deep by 16 mm in diameter ( ~ 2.0 sq.cm in area) over the coverslip which formed the culture surface. In some experiments a thin layer of carbon was evaporated onto the coverslip before attaching it to the dish in order to permit subsequent embedding for electron microscopy 22. To permit observations with differential interference contrast (Nomarski) optics, the lids of the culture dishes were also modified by making a 30 mm diameter hole in the plastic and sealing a No. 2 coverslip over this with silastic medical adhesive (Type A, Dow Coming). These dishes allowed excellent observation of the living cells under an inverted microscope with a long-working distance condenser and up to 100 × oil immersion objectives. Before use, the dishes were sterilized in 70 ~o ethanol and dried in a sterile hood. Then a 0.1% solution of poly-L-lysine hydrobromide (Sigma Type VIIB) in 0.1 M borate buffer, pH 8.4, was applied to the well for about 24 h 1~. The dishes were then rinsed twice with sterile water (over a 24-h period) and dried. Before use (generally within 3 days of preparation), they were again rinsed twice, this time with MEM, and 0.2 ml of the complete medium was added. Dishes containing medium were stored in an incubator at 35-36 °C for up to 3 days before the cells were applied.
Preparation of the cells Cells were taken from 18- or 19-day-old fetuses from our own colony of Holtzman rats. Timed pregnancies were obtained by daily checking vaginal washings for sperm, the day on which sperm were found being regarded as day 0. At the appropriate stage of gestation the pregnant rats were anesthetized with chloral hydrate and the
401
Fig. 2. An enlarged view of the medial aspect of the right cerebral hemisphere of a 19-day rat fetus. The large arrows mark the location of the hippocampal fissure, and the small arrows the developing fimbria. On the right is shown (at the same scale) the dissected hippocampus from which the meninges have been removed. Again, the large and small arrows indicate the hippocampal fissure and the fimbria, respectively. uterus removed to a sterile dish. The remainder of the cell preparation was performed in a sterile hood. The brains were removed from the fetuses with a pair of fine scissors, and the cerebral hemispheres separated from the brain stem. When the hemisphere of the 18or 19-day-old fetus is viewed in a dissecting microscope the hippocampus can be clearly seen on its medial surface as shown in Fig. 2 (left). The hippocampal fissure, usually marked by a conspicuous group of blood vessels, indicates the approximate junction between the hippocampus and the adjoining subicular and entorhinal cortex. The developing fimbria is seen as a white translucent band along the free margin of the hippocampus. Before separating the hippocampus from the hemisphere, the meninges and adherent choroid plexus were carefully pulled off with fine forceps. At this stage the full depth of the hippocampal fissure can be seen. Then with iridectomy scissors the hippocampus was separated from the adjoining cortex by a cut parallel to the hippocampal fissure, and by transverse cuts at its rostral and caudal ends. The appearance of the hippocampus, after dissection, is shown in Fig. 2 (right). The method of cell dissociation is based on the procedure described by Fischbach 10. Eight to 24 hippocampi were placed in 5.0 ml of Ca- and Mg-free BSS containing 0.1 ~ trypsin (Difco 1:250) for 15 min at 37 °C. The intact hippocampi were then rinsed 3 times (5 min each) in BSS, and the cells dissociated by repeated passings through a fire-polished Pasteur pipette until no visible lumps of tissue
402 remained. The cells were then counted in a hemocytometer and their viability assayed by their exclusion o f t r y p a n blue or erythrosine B. Except as otherwise noted, 30,00035,000 viable cells in 15-40 #1 of BSS were added to the center of previously prepared culture dishes containing about 0.2 ml of medium. They were then incubated at 35-36 °C, in air saturated with H20 vapor. In some experiments, dissociated pyramidal cells were co-cultured with explants from the hippocampus at the same stage of development. The hippocampal explants were prepared by cutting slices about 0.5 mm thick in a plane transverse to the long axis of the fetal hippocampi. About 6 slices were placed in a culture dish and incubated until they had become firmly attached to the surface (usually 3 or 4 days). At that time dissociated cells, prepared as described above, were added and the cultures were returned to the incubator.
Observations Living cultures were examined with an inverted microscope using either phase contrast or differential interference contrast optics. Fixed cultures were prepared by slowly adding warm 4% glutaraldehyde in 0.1 M phosphate buffer at pH 7.2. After 30-60 min, this fixative was removed and fresh gtutaratdehyde added for a further hour. Permanent mounts for phase contrast or differential interference contrast observation were prepared by removing the coverslips from the dishes with the cells still attached to them. After rinsing with BSS the coverslips were mounted on glass slides in buffered 50 % glycerol and ringed with Permount. For silver staining, the coverslips were removed, dehydrated in a graded series of alcohols and placed in xylene to remove any residual paraffin adhesive. Following rehydration, the cells were stained by a modified Bielschovsky method 24. For the quantitative analyses of cell survival, the neurons were counted in square areas 307 pm on a side from 35 to 40 representative regions of the dish. When counts were made on living cultures, cell attachment was checked by lightly tapping the microscope stage. In fixed cultures the unattached cells were removed by rinsing. Cells with neurites were counted only if they had at least one process over one cell diameter in length and tipped with a growth cone. There are quantitative differences in growth between one cell preparation and another, and between one batch of medium and another. For this reason all comparisons were made between experiments using the same preparation of cells and the same medium. In all cases the results illustrated are representative of a series of 2, 3 or 4 independent experiments.
Thymidine labeling In order to determine the relationship between the time of cell origin (i.e., the time when the cells become post-mitotic) and their survival in culture, each of a series of pregnant rats was given a single intraperitoneal injection of [3H]thymidine (5 #Ci/g body weight, specific activity 20 Ci/mmole, New England Nuclear) at intervals between day 15 and day 19 of gestation. All the animals were sacrificed on day 19 and the brains from the fetuses were used to prepare a series of hippocampal cell cultures. The cultures from each series were fixed shortly after cell attachment (1.5-3.5 h after
403 plating) or after 1, 2, or 3 days growth in culture. In addition, the brains of two fetuses from each pregnant rat were used to localize the labeled cells within the intact hippocampal formation. These brains were fixed in 70 ~o ethanol and 2 % acetic acid, embedded in paraffin, serially sectioned in the frontal or horizontal planes at 7 pm, and processed for autoradiography as described by Cowan et alP. After an exposure time of two weeks, the autoradiographs were developed in D19 at 17 °C, fixed in Ektaflo and stained through the emulsion with thionin. The fixed cultures were dehydrated, cleared in xylene, then subsequently rehydrated, dried, and mounted on slides, with the cells facing upwards. They were then coated with undiluted Kodak NTB-2 emulsion, exposed for two weeks at 4 °C, developed in DI9 at 17 °C, and mounted in 50% glycerol under a second coverslip. Each culture was examined to determine the percentage of labeled cells present, different regions of the coverslip being systematically sampled. Only the heavily labeled cells were considered for the analysis; lightly labeled cells (overlain by fewer than 10 grains) were included with the unlabeled cells in this analysis. For cultures fixed before the initiation of neurite outgrowth, large and small cells were counted separately. For cultures fixed after 1-3 days growth, only those cells were counted that had clearly developed neurites more than 30/zm in length and were spatially distinct from the processes of neighboring cells. RESULTS
(1) The appearance of the cells immediately after dissociation After brief trypsinization the hippocampus from 18- or 19-day rat fetuses can be almost completely dissociated into single cells. No visible fragments of tissue remain and cell suspensions contain only an occasional cluster of 2-3 cells. The hippocampus from each hemisphere yields about 500,000 cells*. Of these, 85-90~o are viable, as judged by dye exclusion. Fig. 3 shows the appearance of the cells about 2 h after dissociation, before they have begun to form processes. At this stage about 80 % of the cells are small and spherical (with a mean diameter of about 8 pm). The remainder are distinctly larger and are typically pyramidal or ovoid in shape. The larger cells often retain one or more fairly lengthy processes, but the small cells are usually without processes immediately after dissociation. Some of the differences between the two classes of cells with regard to their time of origin and their survival in vitro will be considered below. But most of the findings to be described in this paper refer to the smaller cells which comprise the dominant population in these cultures. (2) Adhesion of the cells to the substrate After plating, the cells rapidly adhere to the surface of the polylysine-treated coverslips. As they adhere they often exhibit filiform or flattened membranous ex* This figuremay be comparedwith the figure of 450,000 found in a recent count of the number of pyramidalcells in the mature rat hippocampus(Schlessingerand Cowan,unpublishedobservations).
404
Fig. 3. The appearance of hippocampal cells shortly after dissociation. In this preparation, the cells were allowed to attach to a polylysine-treatedcoverglassand were photographed 2 h after dissociation. Two types of cells are present: a population of small, round cells, and a population of larger, more elliptical cells (arrows). The latter often show one or more short processes after dissociation. (Differential interference contrast micrograph of cells fixed in glutaraldehyde and mounted in 50 ~ glycerol.) tensions at one or more points along the margins of the cell perikarya which serve to increase the area of cell surface that is in contact with the substrate. This can be seen in some of the cells shown in Figs. 3 and 5. Under our usual conditions of culture, with an initial density of about 15,000 cells/sq.cm, the neurons adhere directly to the substrate, and there is essentially no adhesion of neurons to non-neuronal cells. Furthermore, because adhesion is rapid, neurons usually attach to the surface before any significant reaggregation can occur. This makes it possible to determine with reasonable accuracy the number of neurons that attach to the surface and, subsequently, the number of cells that develop neurites, Fig. 4 illustrates the extent of cell adhesion as a function of time after applying the cells to the culture dishes. It is evident from this that some of the cells become firmly attached within 10 rain and adhesion is essentially complete within 0.5-2 h. The number of cells that adhere to the surface varies from experiment to experiment but, when measured at 24 h, has been found to average about 60 ~ of the viable cells applied. These results are comparable to those of Yavin and Yavin 29 who examined the adhesion of cells from the cerebral hemispheres of rat embryos to polylysine-treated plastic petri dishes. They found that adhesion was complete within 0.5 h and that less than 25 O//oof the applied cells failed to adhere.
(3) Initial neurite outgrowth Many of the cells in our cultures show no indication of neurite outgrowth after adhering to the culture surface. On the other hand, some cells begin to extend processes with distinct growth cones within a few hours after plating. These cells with actively growing neurites can be distinguished from the larger cells which retain processes after dissociation since the processes of the latter lack growth cones and are seldom attached
I0
--
5O
I.
30
~ 2o -,
10
I
2
3
4
TIME IHRSJ
Fig. 4. The kinetics of cell adhesion to polylysine-treated coverglasses. At the times indicated culture dishes were rinsed with BSS and then fixed with glutaraldehyde. The numbei of attached cells was counted and expressed as a percentage of the number of viable cells originally applied. The results from two experiments are shown, each point being the mean of two or three cultures from one experiment. Cell attachment is half-maximal within 20 rain and is essentially complete within 1-2 h.
Fig. 5. The morphology of hippocampal cells after one day in culture. A: a low-power photomicrograph of a cell culture fixed and stained after 24 h in vitro with a reduced silver method. The isolated cells, which are attached directly to the substrate, have extended one or two unbranched neurites. Several cells which have not developed processes are also present. Scale- 50 pro. B, C: photomicrographs of typical isolated cells from such cultures. The cell in B was photographed in a living culture by phase contrast optics whereas C is taken from a fixed, silver-stained preparation. Note the appearance of the major and minor processes and the prominent growth cones at the tips of the major processes. Scale in B and C: 10 pm.
406
~,~ /
cells adhering 60
--
/
/
I
-
/I
1.8
assre~e/ size /
_
/
1.6
/
cells with
7"/ 1.2
I 7,500
I 15,000
PLATING OENSITY
I' 22;500
I
1.0
3O,OOO
(CELLS/CId 2)
Fig. 6. The effect of cell density on cell attachment, neurite formation and re,aggregation. Three cultures at each cell density were fixed after 24 h in vitro. The total number of cells attached to the surface (open circles) and the number of cells with neurites longer than twice the cell diameter (filled circles) were counted and expressed as a percentage of the number of viable ceUs initially applied. Cell density has little effecton cell attachment, but the percentage of cells that extend neurites increases markedly with increasing density. As a measure of the extent of cell reaggregation, mean aggregate size was determined using the following formula: mean aggregate size= [no. of single cells and no. of aggregates] ÷ total no. of cells present. If only single cells were present, the mean aggregate size was regarded as 1. Extensive cell aggregation (open squares and hroken line) occurs at higher cell densities. The cell densities shown on the abscissa refer to the number of viable cells/ sq.cm that were initially applied. to the culture surface at their tips. The initiation of neurite extension generally occurs within the first 24 h in culture. The appearance of several typical cells at this stage is shown in Fig. 5A which is a low-magnification photomicrograph of a culture fixedafter 24 h growth and stained by a reduced silver method. The cell perikarya are usually spherical or somewhat elliptical in shape and, under phase contrast, appear refractil¢. Usually one or two major processes can be seen extending from each perikaryon. These follow a rather meandering course and average about 70 ~m in length, but many are over 150/~m long. The processes are usually unbranched and have a characteristic growth cone at their tips (Fig. 5B, C). Typically, the growth cone takes the form of a-flattened membranous expansion which can be seen to undulate fairly rapidl3~. On occasion, microspikes can be seen extending from the expanded growth ctmes. Laterally extending fine branches, or microspikes, are also seen at intervals along the processes between the cell body and the tips of the process. A second class of process, which we shall term "minor processes", also commonly extends from the cell body (Fig. 5B, C). These are distinctly shorter than the major processes and seldom exceed 2 0 / a n in length. They vary in form from flattened membranous extensions, through thin
407 TABLE I Cell attachment and neurite growth
The extent of cell attachment and of neurite growth after 24 h in culture. The results shown represent 5 separate cell preparations grown in 7 lots of medium. Cell densities ranged from 15,000 to 17,500 cells/sq.cm. Exp. no.
Cells adhering ( % of those applied)
Cells with neurites ( % of those adhering)
Cells with neurites ( % of those applied)
1
2 3 4 5a 5b 5c
64 65 61 43 69 50 43
16 30 34 26 33 46 44
11 24 21 11 23 24 19
Median
61
33
21
elongated processes resembling microspikes, to stouter processes resembling short neurites. The significance of these minor processes, and their relationship to the major processes, is not clear. (4) Cell density and cell growth
Cell density has a major effect on the growth of neurons in culture. This is illustrated in Fig. 6, which shows the results from a typical experiment in which 2 or 3 cultures were grown at each of 4 different cell densities. All these cultures were fixed after 24 h o f growth and the number of cells with, and without, neurites in each culture determined. At all cell densities examined, the total number of cells that were present was a more or less constant percentage of the number of viable cells applied; in this experiment about 60 % of the cells had survived for 24 h. Cell adhesion thus appears to be independent of cell density, at least over the range of densities we have studied. However, neurite outgrowth is critically dependent on the number of cells present. When the initial cell density is below about 3500 cells/sq.cm (,-~ 35,000 cells/1.0 ml medium), less than 12% o f the applied cells extend neurites. With increasing density the number of neurites increases progressively so that at an initial density of 15,000 cells/sq.cm, 32 % of the cells have neurites. At higher densities this percentage increases only slightly. Unfortunately, although neurite growth increases with increasing cell density, cell aggregation increases concomitantly. This is also shown in Fig. 6 where the extent of cell aggregation is examined as the mean number of cells per aggregate .At densities of about 15,000 cells/sq.cm and lower, cell aggregates are infrequent, and the mean aggregate size is less than 1.3. Further increases in density greatly increase the number and size of aggregates. At 30,000 cells/sq.cm, mean aggregate size is about 1.9, and at higher densities aggregation is so extensive that accurate cell counts cannot be made. There is a rather narrow range o f cell densities, around 15,000-20,000 cells/sq.cm,
408 80
60
4o
==~
neurites
0
I 1
I 3 DAYS
I 5 IN
! 7
CULTURE
Fig. 7. Cell growth during one week in culture. Two cultures were examined daily, and the total number of cells attached to the substrate (open circles) and the number of cells with neurites (filled circles) were determined. Non-neuronal cells were excluded from the computations. The number of cells that have not developed neurites declines sharply during the second and third days in culture, whereas the number of cells with neurites declines only slightly during this period. The number of cells with neurites begins to fall on the third and fourth days and such cells have virtually disappeared by the seventh day. that permit reasonable neurite outgrowth without causing extensive cell aggregation. For this reason this density was chosen for most of the subsequent experiments. The extent of process growth obtained in a series of such experiments is shown in Table I. In 5 of the 7 experiments shown, between 19 % and 24 9/00of the viable cells developed neurites in culture. In two other experiments growth was distinctly worse, either because of poor cell adhesion (experiment 4) or because of poor fiber outgrowth from adherent cells (experiment 1). Such poor results are probably attributable to the particular batches of serum employed, and were later eliminated by discarding those serum lots that failed to support growth above the chosen criterion.
(5) Neuronal and non-neuronal cells When hippocampal cells are plated at a density of about 15,000/sq.cm, the resulting cultures are remarkably free of obviously non-neuronal cells. We do observe small numbers of flattened epithelial-like or fibroblast-like cells and occasionally a well-differentiated astrocyte. Such non-neuronal cells represent only about 2 ~o of the cells present in cultures after 24 h, This presumably reflects the low number of nonneuronal cells in the initial cell suspension. This number can be rougJhly estimated by determining the number o f cells that rapidly adhere to untreated tissueeulture dishes. We find that less than 1 9/00of the cells viable after dissociation attach to untreated Falcon tissue culture dishes within 2 h. Nearly all of these cells spread rapidly on the culture surface and assume an epithelial,like appearance.
409 ~
40
B
cells with
-~
30
-
s ,.~
20
-
10
-
Z
~'~
cells
2
4 DAYS
6 IN
i ///
i
8
14
CULTURE
Fig. 8. The effect of hippoeampal explants on the growth and survival of dissociated hippoeampal neurons. The number of dissociated neurons which had extended processes was determined on successive days in cultures containing dissociated cells alone (open circles) or dissociated cells together with the hippocampal explants (filledcircles). In the latter instance 6 hippocampal explants from 18-day rat fetuses were allowed to grow in culture for 4 days before adding the dissociated cells. Note that the co-culture with explants increases the percentage of cells that develop neurites and markedly enhances their survival in culture. Initially 47,000 viable cells were applied to each of 3 dishes in each condition.
In most of our cultures there are also a large number of cells which, although they neither extend processes like neurons nor spread out on the culture surface like epithelial cells, closely resemble neurons in their size, shape and affinity for silver stains (Fig. 5A). Most probably, these cells are neurons which for one reason or another are incapable of growth under our culture conditions. As we shall see later, many of these cells are generated earlier than those which extend processes (see Section 10).
(6) Long-term cell survival Most of these cultures have been studied for 5 or more days in vitro. The nerve cells continue to extend processes during the first 2 or 3 days in culture, but thereafter there is a progressive loss of cells which is essentially complete by 7 days. The survival of neurons in a typical experiment is shown in Fig. 7. In this experiment two cultures were examined at intervals, and the numbers o f cells with and without neurites counted. The number of cells with neurites remains fairly constant during the first 2 or 3 days in culture, but there is an obvious decrease in the number of cells which lack processes. Beyond the third day, virtually all o f the surviving cells have neurites. A similar selective loss of neurons that fail to form processes has been observed in cultures of mouse spinal ganglia 27. After the third day there is a rapid loss of neurons, and by day 7 less than 1% of the cells initially applied remain viable. The p o o r long-term survival in our cultures is probably related to the low cell densities used. One possible solution to this problem is to co-culture the dissociated cells with tissue explants. We have performed a number of experiments in which thin
410 ..........
..............
Fig. 9. A : a photomicrograph of hippocampal neurons after one week in culture. Note that the cell bodies have increased in size, and that the cells are interconnected by a rich fiber network. A single non-neuronal cell is also present (arrow). (Preparation fixed in glutaraldehyde, mounted in 50~o glycerol, and photographed under phase contrast.) Scale: 50 t~m. B: a hippocampal neuron maintained in culture for 11 days. Note the pyramidal-shaped cell body, the stout apical process which bifurcates near the cell body, and 3 slender basal processes. In these respects the cell resembles a typical young pyramidal cell from field CAz of the hippocampus. (Fixed in glutaraldehyde and stained with reduced silver.) Scale: 25 Mm.
slices of fetal rat hippocampi were co-cultured with dissociated hippocampal cells. In effect this procedure increases the total number of cells present in the cultures by about an order of magnitude, while retaining the advantage of having isolated cells for individual analysis. Except in the area immediately adjacent to the explants, free, dissociated nerve cells are readily recognized. Our experience with this method is limited, but in some cultures cell survival has been considerably improved. Fig. 8 illustrates the influence of hippocampal explants on the growth and survival of dissociated hippocampat neurons. Co-culture with explants both increases the percentage of cells that initially develop neurites and markedly prolongs the survival of the neurons in culture. Fig. 9A shows a small group of neurons cultured for 1 week under these conditions. Many of the cells have increased appreciably in size and have come to resemble maturing neurons (note especially the prominent nueleoli). Long processes extend from the cells and intermingle with the processes of other dissociated neurons to form a rather dense plexus. Neurites of cells within the explant and the non-neuronal cells do extend a short distance from the explants. However, in regions such as that illustrated that lie at some distance from the explants, the number of non-neuronal cells is quite low. Cell counts indicate that in most areas neurons still outnumber non-neuronal cells by about 4:1 and that less than 3 ~ of the culture surface is occupied by the latter,
411
k
o o Oog
oH
