Coumarin–phalloidin: A new actin probe permitting triple immunofluorescence microscopy of the cytoskeleton

The mode of involvement of the cytoskeleton in diverse cellular functions is an area of central interest in contemporary cell biology. Mediation of function is now generally presumed to involve an ever increasing number of proteins that are associated with the three fibrillar systems of the cell: the actin filaments, micro-tubules and intermediate filaments (Pollard & Cooper, 1986; Traub, 1985; Soifer, 1986). An integral part of studies on these associated proteins is their localization in the cytoskeleton, in one or more of the fibrillar systems, and for this purpose standard immunological and other probes for the primary fibrillar proteins are now commercially available and in general use. How-ever, localization studies have been restricted to double fluorescence microscopy, using the most commonly available fluorescein and rhodamine-labelled conjugates. Among these, different phalloidin analogues, emitting in the red (Wieland et al. 1983; Faulstich et al. 1983) or green (Wulf et al. 1979; Barak & Yocum, 1981) have proved invaluable as markers of the actin system.

In the course of our recent studies on fibroblast locomotion it became necessary to visualize actin in addition to two other cytoskeletal components (vincu-lin and tubulin) and for this purpose a compromise procedure was followed that involved overstaining one of the first two labels (after photography) with a third label carrying the same fluorophore (Small & Rinner-thaler, 1985; G. Rinnerthaler, B. Geiger & J. V. Small, unpublished data). This experience prompted us to consider the employment, instead, of a third, blue fluorescent probe. In an earlier study, Namihisa et al. (1980) had indeed described the use of such a probe, a thiol-reactive coumarin derivative (Machida et al. 1975), on heavy meromyosin or myosin subfragment-1 for the demonstration of actin in frozen sections. Since, however, myosin subfragments are not very stable in the long term the availability of this probe relies on a ready supply of the appropriate subfragments. In an alternative approach we have tested the possibility of coupling coumarin to phalloidin and here describe results obtained with the most successful of the derivatives synthesized: 7-diethylamino-3-(4-isothiocyanato-phenyl)-4-methylcoumarin (CPITC)-phalloidin. By the use of commercially available filters, triple immunofluorescence microscopy is readily achieved with this reagent in combination with fluorescein-and rhodamine-labelled probes.

Coumarin-labelled phalloidin (M_r 1240) was prepared by a method similar to that used for tetramethylrhodaminyl phalloidin (Faulstich et al. 1983).

A 10 mg sample of aminomethyldithiolanophalloidin hydrochloride (Wieland el al. 1980) was dissolved in 1·5 ml 70% dimethylsulphoxide. The pH was adjusted to 8·5 with lM-NaOH, and 10mg 7-diethylamino-3-(4-isothiocyanato-phenyl)-4-methylcoumarin (D-347, Molecular Probes) dis-solved in 1 ml dimethylsulphoxide was added. After reacting the mixture for 12 h at 4°C, the solvent was removed in vacuo and the residue dissolved in methanol. The solution was applied on two precoated thin-layer chromatography (tic) plates (20cm×20cm, SILICIA GEL 60 F-254, Merck, Darmstadt) and developed with chloroform/NH₃-saturated methanol/water (65:25:4 by vol.). The product appeared as three bands with greenish fluorescence (RF+ 0·6–0·7), which were scraped off and eluted with methanol. After final purification on a small column (15 cm × 1 cm) with Sephadex LH ₂₀ (Pharmacia, Uppsala) eluted with methanol, coumarin-labelled phalloidin (7·6 mg, 55%) was obtained as a mixture of 3–4 isomers. Coumarin–phalloidin was stored after drying in vacuo or as a stock of 0·1 mg ml^-1 in methanol, at −20°C (coumarin–phalloidin will be distributed by SIGMA Chem. Co.)

Two other coumarin reagents were successfully coupled to aminomethyldithiolanophalloidin, 7-((4-chloro-6-diethyl-amino)-s-triazin-2-yl)amino)-3-phenylcoumarin (C-192, Molecular Probes), and 7-methoxycoumarin-4-acetic acid (M-188. Molecular Probes). However, the affinity of these derivatives for actin filaments was poor, or non-existent.

Beside the coumarin fluorophore reagents, other blue-fluorescent chromophores were tested; among them stilbene, anthracene, acridine and pyrene derivatives. Specifically, these were: 4-acetamido-4-isothiocyanatostilbene-2,2-disul-phonic acid (A-339, Molecular Probes, SITS); 2-anthracene-sulphonyl chloride (A-448, Molecular Probes); succinimidyl acridine-9-carboxylate (S-1127, Molecular Probes); succinic-midyl 1-pyrenebutyrate (S-130, Molecular Probes). All these reagents yielded the corresponding phalloidin derivatives but were not investigated further because they were found unsuitable as actin stains.

Absorption and fluorescence spectra of coumarin–phalloidin were recorded in water using a Beckman 25 spectropho-tometer and a Perkin Elmer MPF-3 fluorescence spectropho-tometer, respectively.

The affinity of binding of coumarin–phalloidin to actin was carried out according to a procedure to be fully described (Faulstich et al. unpublished). In brief, the assay relied on the measurement of displacement of unconjugated, tritiated [³H]phalloidin from F-actin filaments by progressively higher concentrations of coumarin–phalloidin (see Fig. 1).

Human skin fibroblasts (provided by Dr M. Malecki, War-saw), plated on 10 mm diameter glass coverslips, were fixed for 2 min at room temperature in a glutaraldehydeμriton X-100 mixture (0·25% glutaraldehyde/0·3% Triton), followed by 10 min in 1% glutaraldehyde, both made up in a cytoskel-eton buffer (see Small el al. 1986). They were then treated with NaBH₄. (Weber et al. 1978), 0·5mgml⁻¹ in ice-cold cytoskeleton buffer (three times, 5 min) prior to immune-labelling. The coverslips were incubated in the first antibody mixture containing mouse anti-vimentin (a gift from Dr S. Blose) and rabbit anti-tubulin (a gift from Dr J. De Mey) diluted in a Tris-buffered saline (GB, see Small et al. 1986) with added 1% bovine serum albumin (BSA) and 2% normal goat serum, for 40 min at room temperature, on parafilm. After washing (twice, 10 min) in GB they were transferred to sheep anti-mouse biotin (Amersham International, UK) in GB containing 1% BSA for 30 min at room temperature and then, after a further wash, to a mixture of goat anti-rabbit rhodamine (a gift from Dr B. Geiger), streptavidin-FITC (Amersham International) and CPITC-phalloidin, all diluted in GB containing 1% BSA. The CPITC-phalloidin was used at a concentration of 1–2 μgml⁻¹ (1–2 μM). The last incubation of reagents was carried out on ground-glass slides instead of parafilm, since the latter produced detrimental staining with phalloidin derivatives. After the final wash, the coverslips were mounted in Gelvatol (20–30, Monsanto Corp., NY) made up, as described for Elvanol by Rodriguez & Deinhardt (1968), and with added n-propyl gallate (Giloh & Sedat, 1983), l,4-diazobicyclo-(2,2,2)-octane (DABCO) or phenylenediamine (Johnson et al. 1982) as anti-bleach agents. Effective concentrations of additives were 2-·5mgml⁻¹ for n-propyl gallate, 100 mgml⁻¹ for DABCO (Langanger et al. 1983) and 1 mg ml⁻¹ for phenylenediamine.

Preparations were observed and photographed on a Zeiss photomicroscope III fitted with an epi-condensor and side mounted 50 W Hg Lamp. The following Zeiss filter combinations were used: for rhodamine, set no. 14 (intensive); for fluorescein, set no. 17 (selective); and for coumarin, exciter filter G 365 (ultraviolet) and as barrier filter the ‘intensive’ FITC exciter filter BP 450-490. For observation and photography a Zeiss Neofluar (63 ×/1·25) antiflex lens, that transmits in the ultraviolet region of the spectrum, was employed. Micrographs were recorded on black and white Agfapan professional 400 negative film or Kodak Ektakrome 400 slide film with the following DIN settings on the exposure meter: rhodamine, 36; fluorescein, 27–30; cou-marin, 21–24.

As indicated in Materials and methods, a number of dyes exhibiting emission in the blue region of the spectrum were investigated for their suitability as phalloidin fluorophores. Although all of them were complexed successfully with phalloidin, very few then retained their capacity to bind to actin, as judged by positive labelling of aldehyde-fixed cells or unfixed striated muscle myofibrils in the fluorescence micro-scope. Of the three amino-selective coumarin derivatives tried, only one (CPITC) labelled cells strongly enough to make prolonged observation and photogra-phy feasible. Measurements of binding affinity showed this phalloidin analogue to bind 125 times less strongly than unconjugated phalloidin (Fig. 1).

The absorption and emission characteristics of cou-marin–phalloidin are shown in Fig. 2. For practical purposes the two spectra fit closely enough to the absorption spectra of the Zeiss u.v. excitation (365 nm) and fluorescein wide band (450–490 nm) excitation filters for these two filters to be used, respectively, as the excitation and barrier filters in the coumarin channel. Using the standard u.v. filter set (no. 02) from Zeiss, single immunofluorescence images are readily obtainable. However, overlap then occurs between the coumarin and fluorescein channels. This can only be avoided with the above filter combination for coumarin and a narrow band excitation filter for fluorescein (see also Materials and methods). Under these conditions, successful photographic recording was achieved with exposure times that were comparable with those used with rhodamine-labelled phalloidin, in the range of 5–15 s (oil immersion, 63× objective). An example of a human skin fibroblast triple-labelled for actin (cou-marin–phalloidin) vimentin (fluorescein) and tubulin (rhodamine) is shown in Fig. 3. The lack of crossover into the coumarin channel is evident. In those cases where a very dense vimentin coil dominated one part of the cell we did, however, note a depletion of actin stain in the same region. We attribute this to absorption of the coumarin fluorescence by the neighbouring, excess fluorescein fluophore; in other regions this effect was not seen.

Three anti-bleach agents were tested for their suitability with the triple fluorophore combination: n-propyl gallate (Giloh & Sedat, 1982), 1,4-diazobicyclo-(2,2,2)-octane (DABCO) and phenylenediamine (Johnson et al. 1982), using Gelvatol as mounting medium. Although DABCO (at 100 mgml⁻¹) had a marked inhibitory effect with fluorescein and Rhoda-mine, it did not noticeably reduce the bleaching of coumarin. Reduction of fading was obtained with n-propyl gallate (2·5 mg ml⁻¹) and phenylenediamine (Imgml^-1), permitting typical exposure times for recording an image on 400 ASAL27 DIN film of 7–15 s. For coumarin it was, in addition, found important to keep the pH around neutrality (pH 6·5 to 7·5) since at more alkaline values (pH 8·0) an increase in background fluorescence in blue as well as a reduced staining intensity was observed.

In conclusion, the phalloidin derivative described here should prove generally useful for studies in which the distribution of actin relative to two other components in the cell may provide extra details of cyto-skeletal interactions. This has already proved to be the case in our own recent studies on cell locomotion (Small & Rinnerthaler, 1985; Rinnerthaler, Small & Geiger, unpublished). Owing to its small size, phalloidin readily penetrates into densely packed actin networks and because of its additional stabilizing action on actin filaments it constitutes an ideal probe for the thin filament system. Hitherto, only thiol-reactive coumarin derivatives have been applied in histo-chemical studies (Namihisa et al. 1980; Sippel, 1981). Since the present dye (CPITC) used for phalloidin is an isothiocyanate it should be equally suitable as an antibody probe and thus of more general use in immunocytochemistry.

We thank Ms M. Hattenberger and Mrs H. Kunka for valuable technical assistance, and Ms K. Koppelstatter for photography.