Microtubules rich in modified α-tubulin characterize the tail processes of motile fibroblasts

Tubulin undergoes three forms of post-translational modification. α-tubulin can be detyrosinated at its C terminus by the enzyme tubulin carboxypeptidase (Barra et al. 1974) and it can be acetylated at the ε-amino group of a lysine residue (LHernault & Rosenbaum, 1985). The β-monomer can be phosphorylated (Gard & Kirschner, 1985). The tt-tubulin modifications are thought to occur exclusively on assembled microtubules (detyrosination, Gundersen et al. 1987; acetylation, Piperno et al. 1987). Cells stained with antibodies against detyrosinated or acetylated α-tubulin have stained micro-tubules that are in general curly and occur in the more central regions of cells (Gundersen et al. 1984; Piperno et al. 1987). The majority (≈90%) of the microtubules in an interphase cell turn over very rapidly, growing and shrinking with a half-life of about 10 min (Schulze & Kirschner, 1986, 1988; Sammak et al. 1987; Prescott et al. 1989a; Sammak & Borisy, 1988). However, a minority of the microtubules have a much longer half-life, in the region of one hour or more (Schulze & Kirchner, 1987). For at least two cell lines it is on these stable microtubules that the modifications have been found to occur (detyro-sination, Schulze et al. 1987; acetylation, Webster & Borisy, 1989). Such microtubules have also been found to be more resistant to drug disruption (Kreis, 1987). The function(s) of this subset of stable microtubules has so far proved difficult to determine. The post-translational modifications seem to be a consequence of their stability rather than the causative mechanism (Webster et al. 1987; Khawaja et al. 1988).

The NIH–3T3–T15 cell line is a NIH–3T3-derived line that has been transfected with a construct containing multiple copies of the human N-ras proto-oncogene under the control of a mouse mammary tumour virus long terminal repeat promoter (McKay et al. 1986). This promoter can be activated by adding the steroid dexa-methasone to the medium. Induction of the promoter results in the expression of the p2l N-ras oncoprotein within 12 h. The induction of p2l results in a variety of alterations in the physiology and morphology of the NIH–3T3 cells. The cells become progressively more motile and this is reflected in a cell shape change from a flat, well-spread polygon, best expressed in confluent cultures, to a polarised shape with one or more thin distal cell processes or ‘tails’. The cells overgrow one another and ultimately form dense foci of completely spherical cells that have lost contact with the substratum.

In a recent review, Bray and White (1988) described fibroblast locomotion as one of the hardest forms of cell motility to understand, despite intensive study since the pioneering work of Abercrombie and colleagues (cf. Abercrombie, 1980). Possible roles of microtubules in this movement have been particularly difficult to determine although they are thought to be of major significance (reviews: Mareel & DeMets, 1984; Vasiliev, 1987). Microtubule-disrupting drugs are known to prevent polarized movement of fibroblasts (Gail & Boone, 1971; Vasiliev & Gelfand, 1976). This suggests some major role in cell polarization, either directly in orientating the F-actin microfilaments or indirectly, perhaps via determining the direction of intracellular organelle transport. However, microtubule inhibitor experiments on chick heart fibroblasts suggest that motility and polarisation dependence on microtubule integrity changes with time in culture (Middleton et al. 1989).

In this study monoclonal antibodies against acetylated and detyrosinated α-tubulin have been used to localise modified microtubules in transformed NIH–3T3–T15 cells and compared with their non-transformed counter-parts.

NIH–3T3–T15 cells were cultured in Dulbecco’s Modified Eagle’s Medium (DMEM), with 10 % aseptic calf serum added, in a 10% COz atmosphere at 37°C. Cells for immunofluor-escence study were seeded onto glass coverslips, placed into medium with (NIH–3T3–T15⁺) or without (NIH-3T3-TI5⁻) 2 μM-dexamethasone, and allowed to grow to confluence for at least 24 h. Major changes in the phenotype of the cells were normally not seen until 3 days had elapsed in dexametha-sone-containing medium.

The following antibodies were used: ID5, a mouse monoclonal antibody that recognises detyrosinated α-tubulin (Wehland & Weber, 1987); 6–11–Bl, a mouse monoclonal antibody that recognises acetylated <τ-tubulin (Piperno et al. 1987); YL 1/2 (Kilmartin et al. 1982), a rat monoclonal antibody that recognises tyrosinated α-tubulin, supplied by Sera Lab., Sussex.

Cells on coverslips were fixed in 90 % methanol in Mes buffer (100 mM-2-(N-morpholino)-ethane sulphonic acid (Mes), 1 mM-EGTA, 1 mM-MgSO₄, pH6·9) at —20°C for 5 mm. The cells were then extracted with acetone, also at −20°C, and washed three times with phosphate-buffered sahne/bovine serum albumin (0 ·02%, w/v) (Weber et al. 1975). Non-specific staining was prevented by blocking with a 1 in 10 dilution of rabbit serum (Sera Lab., Crawley Down) in phosphate-buffered saline. The microtubules were stained sequentially with either 6–11–Bl or ID5, rhodamine-conjugated anti-mouse IgG (Dako, High Wycombe), YL 1/2 (Sera Lab., Crawley Down) and fluorescein-conjugated anti-rat IgG (Lorne Diagnostics, Reading). Cells labelled for F-actin microfilaments were fixed in 4% paraformaldehyde in PBS and stained with either FITC- or rhodamine-conjugated phalloidin. Fixed, stained cells were mounted in Citifluor (Citifluor Ltd, London) and photographed with an Olympus OM2N camera mounted on a Zeiss standard microscope, using Kodak TMax 400 film pushed to 1600 ASA.

NIH–3T3–T15 cells on a glass coverslip were placed in DMEM with 10% aseptic calf serum buffered with 20mM-Hepes buffer. The dish containing the cells was placed on a Nikon Diaphot inverted microscope (Nikon, Japan) and kept at 37°C by using a fan heater. Temperature was monitored with a digital thermocouple (CP Instruments, Bishop’s Stortford). The phase-contrast image of the cells was recorded with an RCA SIT camera (model TC1004) using a time-lapse video recorder (RCA model TC 38OOX). Images were recorded at a rate of 0·25 picturess^-1 and replayed at 50 pictures s^-’. Real time was recorded on an integral time/date generator. Still images frozen by the video recorder were photographed directly from the monitor using an Olympus OM2N camera and a 50 mm Macro lens.

NIH-3T3-TI5⁻ cells, not induced to produce the ras protein p2l, have a generally flattened, well-spread morphology similar to their parent NIH–3T3 cell line (Linstead et al. 1988). Fig. 1A shows untransformed NIH–3T3–T15 cells stained with the YL 1/2 antibody against tyrosinated α-tubulin. The distribution is typical of static, fibroblast cells in culture and resembles the pattern seen in the parental line (Fig. 1C). In both cases the YL 1/2-stained microtubules are evenly distributed throughout the cells and extend to the cell periphery. By contrast, the microtubules rich in acetylated α-tubulin (‘acetylated microtubules’), stained with 6–11–Bl anti-body (Figs 1B, 5A), were predominantly perinuclear. Such microtubules were wavy and frequently stained in short stretches along the same microtubule, giving a fragmented appearance. This markedly resembles the pattern previously described by Piperno et al. (1987) and is similar for the majority of NIH–3T3 (Fig. 1D) and NIH–3T3–T15 (Fig. 1B) cells. However, a small minority, for example those seen in Fig. 1D, contained strongly staining bundles of acetylated microtubules.

Treatment of the NIH–3T3–T15 cell line with dexa-methasone-containing medium resulted in a marked change in cell shape over a period of about 3 days. The cells lose their polygonal, well-spread morphology, becoming polarised and frequently bearing one or more long thin cell processes or ‘tails’. The microtubule distribution seen with YL 1/2 reflected this change in cell shape but showed no gross alteration in microtubule density within the cell; all parts of the cell were evenly stained. Fields of such cells stained with YL 1/2 (Fig. 2A) showed numerous cells overgrowing one another with an even microtubule distribution. However, the distribution of acetylated microtubules seen with 6–11–Bl antibody (Figs2B, D, 5C) showed an obvious dichotomy in the staining pattern, the thin cell processes being heavily stained. Groups of acetylated microtubules usually extended from a perinuclear region and formed a dense bundle within each tail, running near the plasma-lemma in the tails. These microtubule bundles in the tails of the cells are not obvious from the YL 1/2 staining pattern of confluent cultures (Fig. 2A,C) but dominate the 6–11–Bl pattern of the same cells (Fig. 2B,D).

Because of the close proximity of the microtubules to one another within the bundles it was not possible to discern whether all the microtubules run the whole length of the tails.

The overall pattern of microtubules rich in detyrosinated α-tubulin (‘detyrosinated microtubules’) seen in NIH–3T3–T15⁺ cells was closely similar to that seen for the acetylated microtubules. Fig. 3B shows the detyro-sinated microtubules stained with ID5 antibody in cells equivalent to those seen in Fig. 2. Fig. 3A shows the same cells counterstained with the anti-tyrosinated α-tubulin antibody YL 1/2. The detyrosinated α-tubulin antibody picks out the microtubule bundles in the tail region of the cells. The, patterns seen with the 6–11–B1 and ID5 antibodies were similar but both were very different from the pattern seen with the YL 1/2 antibody. However, the ID5 staining was much less continuous and gave a less-fragmented appearance than that seen with the 6–11–Bl antibody. The number of cell body micro-tubules stained was also usually significantly greater (Fig. 3B). The ID5 also stained the centrioles and mid-bodies.

Modified tubulin microtubule bundles in cell tails were also found in other ras-transforrned cell lines. For example, MR4 cells, a Rat-l-derived line transfected with a mutant human H-ras gene under control of a metallo-thionein promoter (Reynolds et al. 1987) showed marked 6–11–Bl-stained microtubules (Fig. 3D) in the cell tails following zinc-induction of the gene. In the YL 1/2-counterstained field (Fig. 3C) these were not so distinct. Other ras-transformed cells showing similar modified microtubule bundles were EC 816 (a constitutive human H-ras overexpresser from Dr C. Marshall, Chester Beatty Institute, London) and L1043 (a constitutive mutant (EJT24) H-ras transformed cell line also from Dr C. Marshall) (not shown). Modified microtubule bundles were also seen in some non-transformed cell lines, rarely in the case of the NIH–3T3–T15 parent line NIH-3T3 (Fig. 1C,D) or Balb/c-3T3 (Aaranson & Todaro, 1968) but commonly in primary human fibroblasts (not shown).

Prolonged exposure (in excess of 5 days) of the NIH–3T3–T15 cells to dexamethasone-containing medium caused the cells to become progressively more rounded. The cells ultimately become spherical and associated in clumps or foci. These changes were progressive but did not occur in all the cells at the same time, so that any chosen field of cells contained cells at all stages in the process of morphological change until ultimately they all became rounded and lost contact with the substratum. Fig. 4 shows such a focus of cells. The image at the optical plane nearest the glass substratum showed clear staining of bundles of acetylated microtubules in the cell tails, both within the focus and in the surrounding region (Fig. 4A). The central region of the focus also showed strong staining of acetylated microtubule bundles in the cell tails (Fig. 4C). The tops of the foci contained mainly the rounded cell bodies and only a few distinct acetylated microtubules were discernible (Fig. 4E). The YL 1/2 counterstains for these images are shown in Fig. 4B,D,F. Fig. 4B and D demonstrates that diffuse YL 1/2 staining was visible in the interior of the focus while the 6–11–Bl antibody clearly distinguished cell tails (Fig. 4A,C). The rounded morphology of the cell bodies at the surface of the focus can be seen in Fig. 4D,F with YL 1/2 staining.

Fig. 5A and C shows the acetylated microtubule pattern for the cells stained for F-actin in Fig. 5B and D, respectively. Fig. 5A shows that cells with tails did occur occasionally in the NIH–3T3–T15 cells but that such cell tails were the principal feature of fields of transformed cells stained with 6–11–Bl The acetylated tubulin-rich tails sometimes appeared to stain unevenly in short lengths (Fig. 5C).

Cells not exposed to dexamethasone showed the characteristic F-actin staining pattern shown by the parent NIH-3T3 cell line. Fig. 5B shows a field of untransformed NIH–3T3–T15 cells stained with fluorescein-conjugated phalloidin. The cells were rich in F-actin stress fibres and peripheral actin associated with the ruffled membrane. Stress fibres were most marked in the giant multinucleate cells sometimes seen in this cell line (not shown). A minority of cells had no or few stress fibres. These were often found overlying other cells or in rather poor contact with the glass substratum. Stress fibre expression in the uninduced N1H–3T3–T15 cells was variable and seemed to depend on the substratum. Cells grown directly on untreated plastic dishes often had fewer stress fibres compared to cells grown on glass coverslips. However, different areas of the same cover-slips often bore groups of cells that varied in the amount of stress fibre development, presumably reflecting variability in some physical characteristic of the substratum, or possibly in the adhesion of some culture medium factor(s) to the substratum. There was generally a marked reduction or total loss of stress fibre development within NIH–3T3–T15 cells following dexamethasone addition to the medium (Fig. 5D). This reduction paralleled the increase in cellular ras oncoprotein (not shown). After 3 days in culture with dexamethasone the majority of the cells, particularly those with tails, had no stress fibres. However, as can be seen in Fig. 5D, some cells did express stress fibres. Thus the variability in expression of stress fibres seen in NIH-3T3-TI5 ⁻ cells was also apparent in the NIH–3T3–T15⁺ cells. Generally the cells that had few or no stress fibres were polarised and contained distinct bundles of acetylated microtubules in their tails.

Using a time-lapse video recorder it was possible to demonstrate that the long processes of polarised NIH–3T3–T15⁺ cells were indeed the trailing edges of motile cells. Fig. 6 show a series of stills taken from such a time-lapse sequence. Although most of the cells showed some degree of movement, the most rapidly moving cells were those with a polarized morphology (Fig. 6). These cells were characterized by long thin cell processes or ‘tail’ regions at their trailing edges. Some cells had more than one of these processes. Often a tail was seen to become a new leading edge upon reversal or change in the direction of cell movement, which occurred not infrequently. In such cases both the original and the new trailing edges showed bundles of modified microtubules. These spindle-shaped cells had two ‘tails’ and lacked an obvious polarity for a while until a new direction of movement developed. The tail regions seen in the moving cells clearly corresponded to the modified tubulin-rich tails seen in the fixed preparations, as shown by fixing and locating the videoed cells (not shown).

Transformation of the NIH–3T3–T15 cell line by dexa-methasone-induced synthesis of p2l ras oncoprotein results in a major reorganisation of a subset of micro-tubules. These microtubules are recognised by antibodies to two different types of post-translational modification of α-tubulin acetylation and detyrosination. In the untransformed state (NIH-3T3-TI5⁻) such microtubules have a predominantly perinuclear distribution and curly shape very similar to that described in well-spread fibroblasts and other cell types (acetylated microtubules, Piperno et al. 1987; detyrosinated microtubules, Gundersen et al. 1984; Kreis, 1987; Wehland & Weber, 1987). However, upon transformation the cells become polarized and motile, bearing one or more long thin trailing processes or tails, comparable to those seen in chick embryonic fibroblasts (Abercrombie, 1980; Chen, 1981; Dunn, 1980). The tails often became longer and thinner than those typically described by Chen (1981). Progressive thinning of the trailing edge of the cell may result in the formation of the arrays of microtubules present in each tail or the microtubules may be more highly organised into a distinct bundle, with cross-links occurring between individual microtubules. In either case the distribution of modified microtubules represents a dramatic shift from that seen in the untransformed NIH–3T3–T15 cells or their parental NIH-3T3 line.

A comparison between the patterns of post-trans-lationally modified microtubules seen with the two types of antibody, ID5 against detyrosinated α-tubulin and 6–11–Bl against acetylated α’-tubulin, revealed some differences in staining. Although both showed a similar bright continuous staining of the microtubule bundles, in the NIH–3T3–T15⁺ cell tails the cell body microtubules stained more continuously with ID5 than with 6–11–Bl, giving the latter a fragmented appearance. Acetylation of microtubules appears to be distinctly patchy in NIH–3T3–T15 cells compared to detyro3ination, which was more uniform. Both modifications are believed to occur on intact microtubules (Gundersen et al. 1987; Piperno et al. 1987), so the source of the difference is unclear. It may represent differences in enzyme behaviour, activity or accessibility around the individual micro-tubules, e.g. the acetylase may be excluded from some regions of the microtubule while the carboxypeptidase has almost complete access. A similar difference in the staining pattern between acetylated microtubules and detyrosinated microtubules was also noted by Bulinski et al. (1988).

Modified microtubule bundles in the tails of other fibroblast lines were also found, including the ras-transformed cell lines (MR4, EC816, L1043), untrans-formed fibroblasts (primary human fibroblasts Balb/ c–3T3) and infrequently in the untransformed parent lines (NIH–3T3). Acetylated or detyrosinated α-tubulin-rich microtubules have also been described in the processes of a variety of other cell types. These include the axons and other processes of both cultured and central nervous system neurones (Gundersen & Bulinski, 1986; Wehland & Weber, 1987; Cambray-Deakin & Burgoyne, 1987; Cummings et al. 1983; Wolf et al. 1988), sperm tails (Gundersen & Bulinski, 1986) and the process of forskolin-treated CHO cells (Wehland & Weber, 1987). Thus the presence of modified micro-tubules in cell processes is not unique to the tails of motile fibroblasts but may represent a general feature of thin cell processes containing microtubules and may also reflect an alteration in microtubule turnover.

Post-translationally modified microtubules are likely to correspond in general to the more slowly turning over microtubule population (Schulze et al. 1987). Vasiliev (1987) has described the occurrence of ‘stable domains’ without lamellae or pseudopodal activity predominantly in cell processes. Such stable microtubule domains were noted in the tails of polarised fibroblasts, the branched elongated tails of TPA-treated fibroblasts and the growing axons of neural cells. Gundersen and Bulinski (1988) have described the reorientation of stable microtubules, but within the anterior parts of cells adjacent to an artificial wound in a confluent culture. This may represent an early change in motility initiation or a stabilization of the wound edge.

The post-translational modifications on the microtubules in the tails of NIH-3T3-T1S⁺ cells might simply be a consequence of their increased length or the constricted nature of the tail. Alternatively, these microtubules may have a number of specific cellular functions: (1) structural, maintaining the integrity of the tail; (2) transport, providing a means of moving membrane and cytoplasmic material towards the front of the cell; and (3) organis-ational, acting on other cytoskeletal elements such as the cortical F-actin to prevent membrane ruffling.

Chen (1981) studied in detail the retraction of the trailing edge or ‘tail snap’ of embryonic chick heart fibroblasts. He showed that it was a dynamic process consisting of a fast and a slow component. The fast component was attributed to the passive release of tension perhaps built up in microfilament bundles, while the slow phase was an active contraction of a microfilament meshwork. During the slow phase there was a marked reduction in tail microtubules from about 80 before tail snap to less than 10. This suggests that such microtubules might act as part of the cytoskeletal organisation, which may act as a ‘stable strut’ to maintain the tail.

The terminal trailing edge of a fibroblast is characterised by a cluster of focal contacts (Chen, 1981). This cluster is left behind as a result of the cytoplasmic rupture accompanying tail snap. Since microtubules have been proposed to stabilize focal contacts in the leading lamelli-podium (Rinnerthaler et al. 1988), it seems quite possible that the modified and very possibly stabilized microtubule bundle associated with the cell tail could be stabilising and maintaining the focal contacts present in the tail until retraction. Once retraction has been initiated by loss of stress fibre bundles in the tail region the microtubules would rapidly disassemble, possibly as a result of the rupture near the end of the tail. In this way events dependent on microtubule ends could be communicated to the cell interior. Such an organization would be in agreement with the more general hypothesis of Kirschner & Schulze (1986), that close association of microtubules with the plasmalemma could alter cell morphology’ by stabilizing membrane domains to produce cell asymmetry.

Vasiliev (1985) previously proposed that microtubules directly stabilize the F-actin cortex of the posterior region of fibroblasts to prevent pseudopodal activity. The bundles of modified microtubules described in the cell tails of fibroblasts in close association with the plasma-lemma would be suitable candidates for stabilisation of cortical F-actin at the rear of the cells, quite possibly via the focal adhesion contacts. Using microtubule-disassembly drugs, Tomasek & Hay (1984) have found that microtubules are required for the elongation of corneal fibroblasts into a bipolar form and the maintenance of the long processes that result. Microtubules were seen to run along the processes and insert into a microfilament-rich cortex. These authors concluded that microtubules exert their effects via F-actin microfilaments.

Loss of F-actin stress fibres is commonly seen during fibroblast transformation (reviewed by Vasiliev, 1985). However, there is not a consistent relationship between transformation and the absence of F-actin stress fibres. It has also been found that the changes in microfilament bundle organisation can be simply a reflection of changes in cell morphology and substratum adhesion, and not of growth control properties, so caution must be exercised in interpreting the significance of any particular phenotype (Willingham et al. 1977). However, the loss of stress fibres is also associated with the acquisition of cell motility (Herman et al. 1981) and in the results reported here NIH–3T3–T15⁺ and NIH-3T3-TI5⁻ cells are being grown under identical conditions. Upon transformation, NIH–3T3–T15 cells generally show a major reduction in the size and number of stress fibres present, which we interpret as being due to the acquisition of cell motility. However, some cells do not lose their stress fibres, despite the fact that they have been shown to contain p2l oncoprotein (Prescott et al. 1989b).

The loss of F-actin stress fibres correlates well with the appearance of modified microtubules in the cell tail and also activated motility in NIH–3T3–T15⁺ cells. It is possible that the presence of modified microtubules in the cell tails correlates generally with the loss of stress fibres and with other situations where cell motility is acquired upon transformation or in embryonic cells.

The changes that occur upon the transformation of NIH–3T3–T15 cells may provide a model for the active cell movement that is believed to occur during metastasis (Liotta, 1985). Cell motility is likely to be important during metastasis in moving tumour cells into the b100d and lymph systems (Strauli & Weiss, 1977) and during metastatic invasion through the extracellular matrix (Liotta et al. 1986a). These metastatic migrations may be under the influence of factors secreted by transformed cells, such as autocrine motility factor (Liotta et al. 1986a). Interestingly, these factors are secreted at high levels by other NIH-3T3-derived cell lines transfected with ras oncogenes (Liotta et al. 1986b). The ras oncogene product p2l is known to induce a marked increase in pinocytic endocytosis (Bar-Sagi & Feramisco, 1986) and exocytosis (Bar-Sagi & Gomperts, 1988). Such changes may represent a general elevation of surface activity of which cell motility is one form. Autocrine motility factor also stimulates cell motility and induces pseudopod formation (Guirguis et al. 1987). p2l elevation in transformed cells has been linked to inositol phosphate production and kinase activity and these changes are likely to affect a whole variety of cellular processes including the cytoskeleton (Wakelam et al. 1986). In all the ras-transformed cell lines examined by us to date a major feature is the presence of bundles of modified microtubules in the cell tails. This feature, like the loss of stress fibres, may well be a direct or indirect consequence of ras gene activation. It is a matter of some interest as to whether other motile transformed or tumour cells will show bundles of modified microtubules in their tails.

The authors thank Dr J. Wehland and Dr G. Piperno for their generous gifts of the ID5 and 6–11–Bl antibodies, the M.R.C. and Cancer Research Campaign for funding, Ruth Magrath for excellent technical assistance and Jill Gorton for secretarial help.