ABSTRACT
In culture, hippocampal neurons initially establish several short, apparently identical processes; of these, only one acquires axonal characteristics, the remainder becoming dendrites. We examined the organization of cytoplasmic constituents that might influence which of the initial processes becomes the axon. The Golgi complex was visualized using either fluorescent wheat germ agglutinin or a specific antibody. Presumptive microtubule-organizing centers were identified by depolymerizing microtubules with nocodazole, then allowing them to repolymerize for brief periods. As judged by light microscopy, hippocampal neurons contained a single Golgi region and a single microtubule-organizing center, which were frequently localized together adjacent to a shallow indentation in the nucleus. In cells fixed shortly after the axons had emerged, there was no correlation between the position of the Golgi complex or the microtubule organizing center and the site of origin of the axon. Based on nuclear shape, the position of the Golgi complex and microtubuleorganizing center could also be inferred in living cells. When axonal outgrowth was followed in individual cells by time-lapse microscopy, so that the location of the Golgi complex and microtubuleorganizing center could be determined at the exact moment when the axon emerged, no correlation was apparent. Antibodies that recognize specific post-translational modifications of a-tubulin-acetylation and de-tyrosination-were used to assess the distribution of arrays of stable microtubules. Stable microtubules were present in all processes, both before and after the emergence of the axon. They were not confined to the axon. Thus the localization of these cellular constituents does not play a major role in determining which of the processes initially extended by hippocampal neurons becomes the definitive axon.
INTRODUCTION
In culture, hippocampal neurons become polarized according to a defined sequence of morphological changes (Dotti et al. 1988). Initially they appear to be unpolarized; they extend several, short neurites, but all of these appear identical in morphology and in pattern of growth (developmental stage 2). Polarity is first expressed when one of these neurites begins to grow rapidly, becoming significantly longer than the others (developmental stage 3) (Dotti et al. 1988; Goslin and Banker, 1989). This process is the cell’s axon, and at this stage it can be distinguished from the remaining processes by its growth properties (Dotti et al. 1988), by its molecular composition (Goslin et al. 1990) and by the complement of organelles it contains (J. Deitch and G. Banker, unpublished observations).
Because the expression of polarity occurs in the apparently homogeneous culture environment, it seems unlikely that the identity of the axon is determined by an extracellular signal. What then determines which of the initial processes becomes the axon? One possibility is that the identity of the axon is specified by some aspect of the intrinsic organization of the cell. Alternatively, which process becomes the axon might be a matter of chance, but an intrinsic program of development may prevent the formation of multiple axons. If the axon is transected near the cell body, a new axon may form from a different process (Dotti and Banker, 1987). This observation is certainly consistent with the second hypothesis, but it is equally compatible with the first if axotomy scrambles the cytoplasmic organization within the cell body.
Several intrinsic determinants of cell polarity have been identified in non-neuronal cells. For example, in migrating fibroblasts and leukocytes (Malech et al. 1977; Gotlieb et al. 1981; Kupfer et al. 1982), the centriolar region, from which most microtubules arise, is consistently oriented toward the leading edge of the cell. Microtubules contribute to the stabilization of asymmetric form and serve as the substrate for the transport of vesicles toward the leading edge of the cell. The Golgi complex, which produces the membrane constituents that are added at the leading edge, also faces toward this pole of the cell (Kupfer et al. 1982; see review by Singer and Kupfer, 1986). Moreover, signals that induce oriented migration of fibroblasts or leukocytes cause a repositioning of the Golgi complex and the centrosome (Kupfer et al. 1982; Gotlieb et al. 1983; Nemere et al. 1985). In neurons, the positions of the Golgi complex and centrosome change during development, (Hinds and Hinds, 1974; Shoukimas and Hinds, 1978), but the relationship of these organelles to the site of axonal outgrowth is unclear (see Discussion).
It has also been hypothesised that the development and spatial localization of a stable population of microtubules plays a role in the generation of cell asymmetry (Kirschner and Mitchison, 1986; Bulinski et al. 1988). In unpolarized cells, most microtubules are dynamically unstable, having life-times of only a few minutes before they undergo rapid depolymerization. Stable microtubules undergo spontaneous depolymerization much less frequently (Schultze and Kirschner, 1986). Independently of their behavior, stable microtubules can often be distinguished because of specific post-translational modifications of a-tubulin, i.e. acetylation of the e-amino group on one lysine residue (L’Hernault and Rosenbaum, 1985) and removal of the tyrosine at the carboxy terminus (Argarana et al. 1978). These enzymatic modifications do not themselves influence microtubule stability (Khawaja et al. 1988), but because they occur preferentially on polymerized tubulin (Kumar and Flavin, 1981; Gunderson et al. 1984) they serve to mark stable microtubules in cells (Schultze et al. 1987; Kreis, 1987; Bulinski et al. 1988; Khawaja et al. 1988). The spatial organization of post-translationally modified microtubules can be analyzed by the use of the antibodies which specifically recognize de-tyrosinated (stable), tyrosinated (unstable) (Gunderson et al. 1984, 1987; Kreis, 1987), or acetylated a-tubulin (Piperno and Fuller, 1985).
In non-neuronal cells the generation of a stable array of microtubules is an early event in differentiation, and the localization of stable microtubules may contribute to the development of cell polarity (Gunderson and Bulinski, 1988; Gunderson et al. 1988; Houliston and Maro, 1989). Neurons also contain a population of acetylated, de-tyrosinated microtubules that have distinct stability properties (Black et al. 1989; Baas and Black, 1990). Studies of neurons in culture suggest that post-translationally modified tubulin appears relatively early in neuronal development (Robson and Burgoyne, 1988; Ferreira and Caceres, 1989), consistent with a role in neurite outgrowth, but its involvement in the generation of neuronal polarity is uncertain.
Here we report studies concerning the localization of the Golgi complex and microtubule organizing centers, and the distribution of acetylated and detyrosinated microtubules, in cultured hippocampal neurons during the expression of polarity. We found no indication of a preferential positioning of these structures in relation to the site from which the axon emerged, nor did we observe consistent changes in cytoplasmic organization that preceded or accompanied the outgrowth of the axon.
Experimental procedures
Hippocampal cell culture
Hippocampal cultures were prepared from the brains of 18 day old fetal rats as described previously (Banker and Cowan, 1977; Barlett and Banker, 1984; Goslin and Banker, 1991). Briefly, hippocampi were treated with 0.25% trypsin for 15 min at 37’, dissociated by repeated passage through a fire-polished Pasteur pipet, and plated onto glass coverslips that had been treated with poly-L-lysine. After the cells had attached, the coverslips were transferred to a dish containing a confluent monolayer of astroglial cells, and maintained in a serum-free medium.
Visualization of the Golgi complex, microtubuleorganizing centers, and post-translationally modified tubulin
The Golgi complex was detected using rhodamine-conjugated wheat germ agglutinin (WGA) (Vector Laboratories), as originally described by Virtanen et al. (1980). After 1 or 2 days in culture, cells were fixed in paraformaldehyde (4 % in PBS) for 20 min, permeabilized by dehydration in ethanol, and incubated in 10 % BSA for 30 min to reduce non-specific binding. They were then incubated for 30 min with rhodamine-WGA (15 μg ml−1 in PBS containing 1% BSA). After several rinses in PBS, coverslips were mounted in PBS:glycerol (1:1) containing 0.1% p-phenylenediamine (Johnson et al. 1981) and analyzed by epifluorescence microscopy. Alternatively, the Golgi complex was localized using a rabbit polyclonal antibody (TGN-38) raised against a 38 amino acid synthetic peptide corresponding to a portion of the deduced sequence of a 100 kD integral membrane protein of the trans-Golgi network (Luzio et al. 1990).
Microtubule-organizing centers were visualized after microtubule depolymerization (see Brinkley, 1985). Noca-dazole (1μgml−1) was added to the culture medium for 90 min to depolymerize microtubules. Some cultures were then extracted for 1 min with Triton X-100 (0.1 %) in a microtubule stabilizing buffer (130DIM Hepes, 2mM MgC12, 10 mM EGTA, pH 6.9), rinsed in the same buffer, and fixed in 4% paraformaldehyde. In other cultures nocodazole was removed and the cells were allowed to recover for periods of 10-30 min before extraction and fixation. Cells were then reacted with anti-tubulin antibody (1:100, Southern Biotechnology, Birmingham AL) followed by fluorescein-conjugated second antibody.
Tyrosinated, detyrosinated and acetylated microtubules were localized after fixation of cells in methanol (10 min at-20°) using the following antibodies: a monoclonal antibody that recognizes acetylated a-tubulin (clone 6-11B, Pipemo and Fuller, 1985) at 1:10 dilution; an affinity purified antibody against the detyrosinated form of a-tubulin (Kreis, 1987) at 1:1000 dilution; a monoclonal antibody against tyrosinated n-tubulin (clone 1 A2; Kreis,
1987) at 1:100 dilution. Cells were rinsed in PBS (3 times, 5 min each) and incubated for 30 min with an appropriate fluorochrome-conjugated secondary antibody. The coverslips were rinsed and mounted in PBS:glycerol, as described above.
The position of the Golgi complex relative to the axon
Following staining with rhodamine-WGA, stage 3 neurons were selected at random, photographed, and the negatives printed. On each photograph lines were drawn from the center of the nucleus through the base of the axon, and from the center of the nucleus through the center of the Golgi complex. Then the angle between these lines was measured. In all, 97 cells were analyzed.
Changes in the position of the Golgi complex
To examine possible changes in the orientation of the Golgi complex, living cells were studied by video-enhanced contrast, differential-interference contrast microscopy using a computer-controlled microscope developed by Stephen Smith (Stanford University) that permitted unattended time-lapse recording. Coverslips were removed from culture dishes 12-24 h after plating and transferred to sealed chambers designed for long-term microscopic observations (Forscher and Smith, 1988). Several cells in stage 2 of development were chosen and their position stored. Recordings were made from each cell in turn at 30-60 min intervals and stored on an optical disk recorder (Panasonic 2028). Focus was maintained using an autofocus algorithm, and a through focal series was recorded for each cell at each time point. The approximate position of the Golgi complex was inferred from the shape and position of the nucleus (see Results), and a series of drawings was prepared illustrating its position before, during, and after the outgrowth of the axon.
RESULTS
Localization of the Golgi complex and microtubule organizing centers during the development of polarity In a first series of experiments we analyzed the morphological appearance and the spatial localization of the Golgi complex in hippocampal neurons before (stage 2 cells) and after (stage 3 cells) morphological polarization. For this purpose, rhodamine-conjugated wheat germ agglutinin (WGA), a specific marker for this organelle (Virtanen et al. 1980), was utilized. In neurons fixed after 1-2 days in culture the Golgi complex appeared as a small, intensely labeled spot located eccentrically in the cells’ cytoplasm and juxtaposed to the nuclear envelope (Fig. 1B,D). This was true both for cells at stage 2 of development, before the outgrowth of the axon, and at stage 3. Because the nucleus was eccentrically located, the bulk of the cytoplasm, including the site of the Golgi complex, was restricted to one pole of the cell. In most cells the nucleus itself was slightly concave or flattened on the aspect facing the cytoplasmic pole, and the Golgi complex was invariably positioned at this nuclear indentation. An association of the Golgi complex with the nuclear indentation has also been observed by electron microscopy (J. Deitch and G. Banker, unpublished observations). To verify that the organelle labeled with WGA corresponded to the Golgi complex we performed double labeling experiments with TGN-38, an antibody that recognizes an integral membrane protein of the trans-Golgi network (Luzio et al. 1990). As shown in Fig. 1 (E,F), both markers appeared to label the same region.
The existence and localization of microtubule organizing centers (MTOCs) in stage 2 and stage 3 cells was analyzed next. MTOCs were identified by a microtubule depolymerization-repolymerization procedure, followed by immunofluorescence staining with anti-tubulin antibodies. When cells were exposed to nocadazole for 90 min and extracted with Triton X-100 before fixation to remove depolymerized tubulin, the MTOC could be identified as a small aster-like structure (Fig. 2A,B). Only a single MTOC was seen per cell, as in sensory neurons (Jacobs and Thomas, 1982), and it was always localized at the nuclear indentation, in roughly the same position as the Golgi complex. If the nocadazole-treated cells were allowed to recover for 10-30 min, numerous straight, radially oriented microtubules extended from the MTOC in all directions, some of them entering the cells’ processes (Fig. 2C-F). Electron microscopic studies of hippocampal neurons at stages 2 and 3 of development indicate that the centriole pair is often associated with the Golgi complex, both lying near the nuclear indentation (J. Deitch and G. Banker, unpublished observations). Thus it is probable that the MTOC, identified by microtubule depolymeriza-tion-repolymerization, is associated with the centrosomal region.
We next asked whether there was a relationship between the location of the Golgi complex and the site at which the axon emerged. Only cells which had a clearly defined axon were considered. Fig. 3 summarizes our findings. In about 20 % of stage 3 cells the Golgi complex was positioned in the cytoplasmic region adjacent to the axon. In a similar percentage of cells the Golgi complex was positioned directly opposite the axon (see, for example, Fig. 1), and in the remaining cells it was localized at intermediate positions. These data offer no indication that the position of the Golgi complex is an important determinant of the site of axonal outgrowth. Because the experimental treatment necessary to identify MTOCs substantially disrupts neuronal morphology (see Fig. 2), the possible relationship between the location of the MTOC and the site of origin of the axon could not be directly assessed. However, since the position of the MTOC was always closely correlated with that of the nuclear indentation and the Golgi complex, a relationship between the location of the MTOC and the site of origin of the axon can also be excluded.
An analysis such as this, based on observations of fixed cells, could be confounded because the position of the Golgi complex and the MTOC can change over time (e.g. see Kupfer et al. 1982). To determine whether this occurred in developing hippocampal neurons, we followed individual cells by differential interference contrast microscopy during the establishment of polarity. The position of the Golgi complex and the MTOC was inferred from the location of the indentation in the nuclei, which could be readily observed in the majority of living cells. Cells were recorded at 30-60 min intervals, beginning at stage 2 and continuing for at least several hours after the axon emerged. Because the nucleus extends above the plane of the substrate on which the cells’ processes are growing, a through focal series was performed at each time point.
In all, 12 cases were obtained in which the position of the nucleus and nuclear indentation could be visualized unambiguously (Fig. 4). On average, these recordings began 8h before axonal outgrowth (range 1 to 17 h), and continued for 14 h after outgrowth (range 7 to 28 h). In about half of the cells, the position of the nuclear indentation changed during this time, ‘rotating’ by as much as 120° (Fig. 4A,B). In other cells the position of the nuclear indentation remained quite constant (Fig. 4E,F). Regardless of this, the position of the nuclear indentation at the moment when the axon emerged was no more closely correlated with the site of axonal emergence than observations of fixed cells had suggested. Nor was there any consistent tendency for the nuclear indentation to reorient toward or away from the base of the axon after the axon had emerged.
Localization of stable microtubules during the development of polarity
In a final series of experiments we analyzed the distribution of post-translationally modified forms of tubulin during the establishment of neuronal polarity. Cells were fixed after 1-2 days in culture and processed for immunofluorescence with antibodies that recognize tyrosinated tubulin (predominantly associated with dynamically unstable microtubules) and antibodies that recognize detyrosinated or acetylated tubulin (predominantly associated with stable microtubules). We first examined the distribution of acetylated microtubules. At stage 2, before the extension of the axon, acetylated microtubules were typically present in most or all of the minor processes (Fig. 5 A-B). In stage 3 cells, which have a single axon and several minor processes, acetylated microtubules were also present in all processes (Fig. 5 C-D). Occasionally microtubules appeared to extend from the region of the MTOC across the nucleus and into a process at the opposite pole of the cell (Fig. 5B). Growth cones, evident in the corresponding phase-contrast photographs, appeared devoid of acetylated microtubules (Fig. 5 C-D), as previously reported (Robson and Burgoyne, 1989; Ferreira and Caceres, 1989).
Double labeling of the cells with antibodies against tyrosinated tubulin and against detyrosinated tubulin also gave no indication of a spatial segregation of stable or unstable microtubules. In stage 2 cells all processes appeared to be stained to a similar extent with both antibodies (Fig. 6 A-C). This was also true of stage 3 cells (Fig. 6 D-F); both the axon and the minor processes contained both types of microtubules. Consistent with their lack of acetylated microtubules, growth cones also appeared devoid of detyrosinated microtubules, although immunoreactivity was observed using the antibodies against tyrosinated tubulin (compare Fig. 6 B,E with C,F). Occasional cells were observed, both at stage 2 and stage 3, that lacked acetylated or detyrosinated tubulin in one or two of their minor processes; this is not surprising since time-lapse observations show that occasionally minor processes retract entirely and new minor processes form. However, we never observed cells in which acetylated or detyrosinated microtubules were confined to a single process, as if this process were destined to become the axon.
DISCUSSION
Microtubule populations
To assess the distribution of stable microtubules during the development of neuronal polarity, we used antibodies that recognize post-translationally modified n-tubulins that are enriched in the population of dynamically stable microtubules (Schulze et al. 1987; Webster et al. 1987; Piperno et al. 1987; Black et al. 1989). Acetylated and detyrosinated tubulin was present in microtubules during stage 2 and stage 3 of development, and was found in all or nearly all of the cells’ processes. This is consistent with previous sugestions that stable microtubules are necessary for the generation of asymmetric cell shape (Kirschner and Mitchison, 1986; Piperno et al. 1987; Bulinski et al. 1988), which in neurons corresponds to the extension of neurites. We did not observe a preferential distribution of acetylated microtubules to the axon, as has been reported for cerebellar neurons in culture (Ferreira and Caceres,1989).
The distribution of stable and unstable populations of microtubules in developing hippocampal neurons is consistent with the stability of their processes, as assessed by time-lapse microscopy (Dotti et al. 1988; Goslin and Banker, 1990). Individual minor processes usually persist throughout stage 2 and stage 3, although they extend and retract at their tips (Dotti et al. 1988; Goslin and Banker,1990).Thus it is not surprising that they contain acetylated and detyrosinated microtubules. In contrast, the growth cones of these processes apparently contained dynamically unstable microtubules, as evidenced by their lack of post-translational modifications. This finding was not unexpected. Growth cones are themselves dynamically unstable processes, extending and retracting continuously; the microtubules they contain must similarly be short-lived. Moreover, growth cone microtubules are particularly sensitive to depolymerization by nocodazole (Bamburg et al. 1986). The absence of acetylated microtubules in the growth cones of dorsal root ganglion and cerebellar neurons in culture has been reported previously (Robson and Burgoyne, 1988; Ferreira and Caceres, 1989).
The Golgi complex and MTOC
By light microscopy, both the Golgi complex and the MTOC were visualized as single, small spots in the perinuclear cytoplasm, and both were frequently localized together. We presume that the structure identified by our assay as the MTOC corresponds to the centrosome, since electron microscopic studies show that the centrioles and the Golgi complex often lie in a similar position (J. Deitch and G. Banker, unpublished observations). An association between the Golgi complex and MTOC is common in many types of cells (for review, see Singer and Kupfer, 1986), including developing neurons in many regions of the nervous system (Hinds and Hinds, 1974; Shoukimas and Hinds, 1978). In most hippocampal neurons both the MTOC and the Golgi complex were localized to a particular region of the cytoplasm, adjacent to a flattening or indentation of the nuclear perimeter. The significance of this nuclear shape is unclear, but it occurs in other cell types, including lymphocytes, where the Golgi complex and centrioles are also located near the nuclear indentation (Fawcett, 1986). The polarity of microtubules at this stage of developmenmt, with plus-ends directed toward the growth cones (Baas et al. 1989), is consistent with their origin from a site in the cell body. However, it is not known what proportion of the microtubules actually arise from the centrosomal region.
In a variety of non-neuronal cell types, the positioning of the Golgi complex and the MTOC is related to the direction of cell migration and the presumed direction of transport and insertion of new membrane mass (reviewed by Singer and Kupfer, 1986). The situation for nerve cells is apparently more complex. In one of the pioneer neurons in grasshopper embryos, whose development is quite stereotyped and has been carefully studied, Lefcort and Bentley (1989) have described a polar organization of tubulin, actin and the Golgi complex which consistently lies at the base of the axon and develops before the axon emerges. The localization of this structure appears to be determined by the mitotic cleavage plane. However, this pattern of organization differs in the second pioneer cell, suggesting that such intrinsic determinants of neuronal polarity can be overridden by other factors. In mammalian neurons developing in situ no consistent relationship has been observed between the position of the Golgi complex and centrioles and the site of emergence of either axons or dendrites (Hinds and Hinds, 1974; Shoukimas and Hinds, 1978). Similarly, in PC12 cells, the position of the centrioles was unrelated to the site of neurite outgrowth (Stevens et al. 1988). However, since the position of these organelles can change, such a relationship could go undetected unless development can be followed at the single cell level. In the present study, we also failed to observe an alignment of the Golgi complex or the MTOC toward the site of the elongating axon. Even when living cells were followed by time-lapse recording, so that the location of these structures could be determined at the exact moment when the axon emerged, no correlation was apparent.
Cytoplasmic determinants in the development of hippocampal neurons
From our observations, it is possible to develop a working hypothesis concerning the role of cytoplasmic determinants in the development of hippocampal neurons in cell culture. First, it seems reasonable to conclude that stable microtubules are necessary for the generation of minor processes, which are themselves relatively stable. Consistent with this possibility, we observed that minor processes at both stage 2 and stage 3 of development contained both acetylated and detyrosinated microtubules. Presumably the growth of minor processes, like the growth of axons, also requires the delivery of materials from the Golgi complex. Thus it seems reasonable to assume that tracks of stable microtubules extend from the region of the Golgi complex into each of the minor processes, and that these are established during stage 2 of development. In fact it was occasionally possible to visualize acetylated or detyrosinated microtubules that appeared to extend from the region of the cell containing the Golgi complex and MTOC into minor processes, including minor processes that arose from the opposite side of the cell.
If these assumptions are correct, then it is not surprising that the position of the Golgi complex and MTOC has little influence in determining the identity of the axon in hippocampal neurons in culture. The localization of these organelles is thought to be significant because this can direct the transport of new membrane, derived from the Golgi complex, toward one pole of the cell (Singer and Kupfer, 1986). If, in hippocampal neurons, stable microtubules extend from the vicininty of the MTOC and Golgi complex into each of the cell’s minor processes during stage 2 of development, then transport from the Golgi region is not likely to be routed preferentially into the nearest minor process, but could enter all minor processes with equal likelihood. Any minor process could receive the materials necessary for axonal growth, and each would have an equal chance of becoming the axon.
It is important to point out that these conclusions depend on the particular features of hippocampal development, which do not apply to all types of neurons. Cultured hippocampal neurons initially extend several short, stable processes. Only after an interval of 12-24 h does one of these acquire axonal characteristics. Other neurons, including grasshopper pioneer neurons and zebrafish motoneurons, initially extend only one process, which becomes the axon (Lefcourt and Bentley, 1989; Pike and Eisen, 1990). In such cells, the localization of the Golgi complex and the MTOC might be of much greater importance in determining the site of origin of the axon.
ACKNOWLEDGEMENTS
We would like to thank the following people who generously provided the antibodies used in this study: Dr Gianni Pipemo (anti-acetylated tubulin); Dr Tomas Kreis (anti-detyrosinated and anti-tyrosinated tubulin); and Dr Keith Stanley (TGN-38). We also thank Judith Van Woert for her excellent technical assistance and Dr Stephen Smith, who kindly allowed us to use the computer-controlled microscopy system he developed. This research was supported by NIH Grant NS17112 to G.B. C.G.D. was supported by a Fogarty International Fellowship from the National Institutes of Health (USA) and by a Fellowship from the Alexander Von Humboldt Stiftung (Bonn, FRG).