F-actin distribution in polymorphonuclear leucocytes

The acquisition of polarity, and subsequent locomotion of neutrophil leucocytes is closely connected with chemotaxis, since chemotactic factors stimulate both morphological polarization and locomotion, even in the absence of a gradient of attractant (Zigmond et al. 1981; Shields & Haston, 1985). Polarization and locomotion are accompanied by the passage of waves of contraction that pass antero-posteriorly down the length of the cell (Haston & Shields, 1984). While the occurrence of contraction waves is easily demonstrated by analysis of film data, cytoskeletal organization during wave propagation has proved more difficult to investigate. A major reason for this, at the level of light microscopy, is that good fixation with cross-linking agents (for example, glutaraldehyde) reduces or destroys the capacity of cytoskeletal proteins to interact with antibodies or fluorochrome-labelled phallotoxins that bind to F-actin. At the electron-microscope (EM) level, post-fixation staining with osmium tetroxide has been shown to destroy microfilaments, although this can be overcome to some extent by stabilizing F-actin with tannic acid and fungal toxins (Maupin-Szamier & Pollard, 1978).

We have previously failed to demonstrate any differences in F-actin organization at the constriction ring, which marks a fixed contraction wave, in motile lymphocytes at either the EM or the light-microscope level (Haston & Shields, 1984). However, formaldehyde fixation was employed before staining cells with phallotoxin and this has proved inadequate for fixing motile cells in suspension or in three-dimensional collagen gels. This paper reports on the organization of F-actin visualized with fluorochrome-labelled phalloi-din (Barak et al. 1981) in neutrophil leucocytes moving on different substrata, using improved fixation methods.

Human neutrophil leucocytes were obtained from peripheral venous blood; 2 vol. of heparinized blood were mixed with 1 vol. of 6% dextran and after sedimenting at 1 g for 45 min, the supernatant was layered onto lymphocyte separation mixture (Flow Labs, Irvine, Scotland) and centrifuged at 400 g for 30 min. The pellet was washed once in HBSS/Mops (see below) and the remaining red blood cells were lysed in 1ml distilled water for 30 s. The remaining cells (>95% neutrophil leucocytes) were washed twice in HBSS/Mops.

The medium used throughout was Hanks’ Balanced Salt Solution (HBSS, Flow Labs, Irvine, Scotland) buffered with 10 mM-morpholinopropane sulphonic acid (Mops, Sigma Chemicals Ltd, Poole, Dorset, England). The chemotactic peptide used to stimulate locomotion was N-formylmethio-nyl-leucyl-phenylalanine (fMLP, Sigma Chemicals Ltd) which was made up fresh daily from 10⁻³ M stock dissolved in dimethyl sulphoxide kept at —20°C. Purified human serum albumin (HSA), used to coat glass substrata was obtained from Behringwerke, Marburg, West Germany. Tetramethvl rhodamine isothiocyanate-labelled phalloidin (TRITC-phal-loidin) was obtained from Sigma Chemical Co. Ltd. Phalloidin stock was dissolved in methanol (3 ml methanol/300 units phalloidin) and stored at —20°C. Staining solution was prepared from stock by drying 50 μl of stock phalloidin in a glass test tube, adding 1ml HBSS/Mops and 100 μg lyso-phosphatidyl choline (Sigma Chemical Co. Ltd). For fixation of cells electron microscope grade 25% glutaraldehyde was obtained from Sigma Chemical Co. Ltd.

Round glass coverslips (16 mm) were washed in chromic acid, rinsed in water then ethanol, dried and incubated with 5 mg ml⁻¹ HSA in HBSS/Mops for 30 min at 37°C, rinsed and kept in HBSS/Mops before use. Cells suspensions (2× 10⁵ ml⁻¹) were added to plastic multiwell dishes containing the coverslips, with 1× 10⁻⁸ M-fMLP.

Collagen gels were prepared as described (Elsdale & Bard, 1972). For these experiments collagen was used at 2×5 mg ml⁻¹ and layered into 30mm plastic tissue culture dishes to give a depth of 3 mm. Cells were added to the top of the gel at 1×10⁶ ml⁻¹ with 1 × 10⁻⁸ M-fMLP.

Cells in suspension were incubated at 1×10⁶ cells ml⁻¹ with 1×10⁻⁸M-fMLP in HBSS/Mops; 2ml samples of cell suspension were incubated in 10 ml plastic conical bottom test tubes.

After incubating for 30 min at 37°C the cell preparations were fixed by adding equal volumes of 0·2% glutaraldehyde in HBSS/Mops (pH 7·2) for 10 min at room temperature. Cells, whether in suspension, adherent to coverslips or inside a collagen gel, were washed after fixation with 0–05 M-glycine in HBSS/Mops and then incubated at 4°C for 15 min in 0·05 M-glycine. This quenches free aldehyde groups, which contribute to non-specific staining after fixation with glutaraldehyde (Sullivan et al. 1984).

Cell suspensions were stained by adding one drop of staining solution to cells pelleted at 400 g for 15 min after which the suspension was washed three times in HBSS/Mops. Coverslips were stained by adding one drop of staining solution to each coverslip for 15 min after which they were rinsed in HBSS/Mops and mounted. Pieces of fixed collagen gel were removed from the dish of collagen (approx. 5 mm × 5 mm), placed in the well of multiwell plastic trays and one drop of staining solution was added for 15 min. After staining, the pieces of gel were rinsed repeatedly in HBSS/Mops over 1 h and mounted on a slide by gently squeezing the fluid from the gel with a coverslip.

For immunofluorescence staining with TG1, a monoclonal antibody against a common granulocyte surface antigen, collagen gels with invaded neutrophils were fixed as described above. A 1:10 (v/v) dilution of antibody in HBSS/Mops was added to pieces of collagen after 30 min at 4°C. After washing in HBSS/Mops over 1 h a second layer, FITC rabbit anti-mouse IgG (Miles Scientific, Slough, England) diluted 1:200 (v/v) in HBSS/Mops, was added. The monoclonal antibody TG 1 was a kind gift from Dr P. Beverley, University College, London.

Cell preparations were examined with an Olympus BH2 fluorescence microscope with epi-illumination, using a × 100 objective (NA 1–3). Photographs were taken using Ilford XPI 400 film with 90s exposure times.

Our previous attempts to examine F-actin distribution in motile leucocytes have been unsatisfactory because of poor fixation (Haston & Shields, 1984). Although round, unstimulated neutrophils could be fixed with 4% formaldehyde, most motile cells consistently lysed. Cells not actually lysed had aberrant morphology lacking the characteristic polarized shape of motile leucocytes. Lysis always occurred at the leading lamel-lipodium, the cell posterior being easily distinguished by the uropod. Neutrophils that were adherent to HSA-coated glass were not lysed to the same extent as those in suspension or in collagen gels, but preservation, as determined by phase-contrast microscopy, was not good. Although formaldehyde-fixed neutrophils could be stained with TRITC-phalloidin, staining was most successful on very adherent, well-spread neutrophils. Staining of the most motile cells was not satisfactory because of the post-fixation morphology described above.

Although glutaraldehyde, a strong cross-linking agent, gives excellent morphological preservation, certain antibody or probe binding sites are destroyed. In addition, remaining free aldehyde groups significantly reduce probe specificity and increase the background. In these experiments, good morphological preservation was achieved by fixing with 0·1–0·2% glutaraldehyde. Below this, the preservation was not sufficient to prevent morphological deterioration within 1–2 h. Above 0·2% glutaraldehyde, background fluorescence was significantly increased in spite of post-fixation quenching of free aldehyde groups with 0–05 M-glycine.

Neutrophils fixed after stimulation by 1× 10⁻⁸ M-fMLP in suspension, i.e. non-adherent, had a consistent polarized morphology with a ruffled lamellipodium and knob-like uropod marking the cell posterior. TRITC-phalloidin-stained neutrophils had an intensely stained ruffled lamellipodium; the ruffles often extended down the length of the cell (Fig. 1A). In addition, the cortex and uropod were brightly stained (Fig. IB).

Neutrophils that invaded a collagen gel on stimulation with 1× 10⁻⁸ M-fMLP were less consistent in their morphology than cells in suspension and many different cell shapes were seen. The lamellipodium, however, was a consistent feature and, although less prominent than in cells in suspension, it stained brightly for F-actin. The cortex was stained, sometimes in localized broad bands (Fig. 2A), and in many cells the cortex anterior to a constriction was more intensely stained than the posterior of the cell (Fig. 2A,C). Often, a brightly stained band of F-actin was apparent at constriction rings (Fig. 2B-D) but in many cells no complete band was seen (Fig. 2A).

Fig. 3A-C shows neutrophils fixed while settling onto albumin-coated glass within 1, 3 and 5 min, respectively, after stimulation with fMLP. Within 1 min, the cells were almost spread with intensely stained peripheral ruffles, already showing signs of polarity (Fig. 3A). After 2min the ruffled periphery became more obviously polarized (Fig. 3B) and within 5 min the majority of neutrophils were fully polarized with an anterior, very ruffled lamellipodium.

Fig. 4A is of neutrophils fixed 15 min after stimulation with fMLP when polarity was fully established. A minority of cells became too spread for locomotion (Fig. 4B). In these cells F-actin was prominent around the periphery. One problem in staining motile adherent cells is that the most motile are the least adherent. Many of these will then be lost during fixation and staining. Fig. 4A,B indicates how staining for F-actin varies between adherent (Fig. 4B) and less adherent (4A) cells.

A major difficulty encountered when viewing fluorescently labelled cells in three-dimensional matrices is the appearance of optical artifacts as the observer focuses down through the specimen. For example, in a cell that has a more or less cylindrical shape, some parts may be perpendicular to the observer and other parts may be horizontal. End-on views of membrane sections, together with flare from out-of-focus regions, can have the appearance of bright bands, which may be mistaken for increased labelling.

Although this type of artifact can be distinguished by careful focusing it was considered necessary to investigate whether brightly stained constrictions were, in fact, end-on views of the cortex.

Flurochrome-labelled antibodies against cell surface antigens produce superficially similar staining to the cortical staining of TRITC-phalloidin and comparisons can therefore be made in the observed staining patterns. The antibody used here was the monoclonal TGI, against a common granulocyte surface antigen that is resistant to glutaraldehyde fixation. This antibody was a kind gift from Dr P. Beverley, University College, London. Staining of neutrophils with TGI is shown in Fig. 5. The surface is brightly stained with an indication of more intense staining at the front, a phenomenon that we have previously observed in motile lymphocytes stained with anti-Thy 1,2 (Haston & Shields, 1984). Although there is a distinct morphological constriction, the staining in this region does not resemble that shown in Fig. 2.

The* mechanism of the essentially amoeboid locomotion of leucocytes, slime-moulds and, of course, free-living amoebae has proved extremely difficult to investigate. Clearly, sol—gel transformations of actin, regulated by a number of actin-fragmenting and binding proteins are involved in cytoplasmic streaming, locomotion and phagocytosis and it now seems certain that force generation also involves an actomyosin-based contractility. A major difficulty in understanding amoeboid locomotion is the absence of any stable cytoskeletal structures and so the coordination of contractile or other force-generating events cannot be dissected.

In neutrophil leucocytes, it has been shown that the acquisition of polarity and locomotion is accompanied by the passage of contractile waves from the anterior to the posterior of the cell (Senda et al. 1975; Shields & Haston, 1985). These waves are stimulated by chemotactic peptides. In a cell moving on a substratum the wave appears to be fixed but in motile cells in suspension the wave moves while the cell is fixed. It should be emphasized that stimulated neutrophils floating in suspension not only take up a polarized morphology but also go through the same cycles of movement as seen in neutrophils translocating over a substratum (Keller & Cottier, 1981; Shields & Haston, 1985). In motile leucocytes, polarity can be rapidly reversed and this, together with the absence of permanently polarized cytoskeletal features, led us to suggest that there was no structural basis to polarity in that its maintenance depends on the continued passage of waves of peristaltic contraction propagated through a contractile network. The lack of any discernible organization of microfilaments at glutaraldehyde-fixed contraction waves makes it difficult to understand how the equatorial contraction occurs. We, and others (Senda et al. 1975; Haston & Shields, 1984), have suggested that the cyclical contractile events superficially resemble those of earthworm locomotion in that a peristaltic wave of contraction acts on the hydrostatic skeleton resulting in forward extension of an elastic anterior. In earthworms, the contraction passes through bands of permanently organized circular muscles. In neutrophils and other amoeboid cells, however, the only permanently organized cytoskeletal structure is the cortical sheath of randomly oriented, organelle-excluding microfilaments, which runs under the membrane and is particularly thick under the lamellipodium (reviewed by Bray et al. 1986). Although the cortex of microfilaments is a candidate for the contractile network it is also possible that cyclical sol—gel transformations of actin pass as a wave down the length of the cell. Previous studies have shown that the waves of contraction that occur during leucocyte locomotion are clearly visible by phasecontrast microscopy in cells moving on glass, in collagen or in suspension. It is surprising, therefore, that the constriction was only brightly stained on cells in collagen gels. Many constriction rings were very intensely stained with phalloidin. Why collagen substrata should produce such a distinct staining pattern is not clear. Neutrophils moving through a three-dimensional collagen matrix must squeeze through small gaps and the cortical actin may be compressed, giving rise to more intense staining. Alternatively, F-actin may be recruited to constriction rings that coincide with gaps in the matrix to resist compression by the stressed collagen or to permit outward force to be exerted against the collagen, as suggested by Brown (1982). Another possibility is that the cytoskeletal organization of a contraction wave is extremely labile and not preserved during fixation. The extra mechanical support provided by a collagen matrix may stabilize the cytoskeleton at the constrictions marking fixed contraction waves. Centripetally moving ‘arcs’ of microfilaments have been observed in motile fibroblasts. It has been suggested that arcs result from the gross movements of bands of polymerized actin (Heath, 1983). The wave form of the arcs resembles the wave of contraction in motile leucocytes, but it is difficult to see how gross backward transport of actin could lead to forward extension.

An investigation of F-actin distribution in motile viable cells would provide more accurate information. Preliminary results suggest that neutrophil leucocytes are useful cells for this approach: (1) because they move fast enough to see in real time; and (2) because chemotactic factors that stimulate locomotion also increase the pinocytotic uptake of TRITC-phalloidin, a method that has been used to label F-actin in living cells (Barak et al. 1981).

I am grateful to Dr P. Beverley for the kind gift of monoclonal antibody TGI. I thank Professor P. C. Wilkinson, in whose laboratory this work was performed, for critical reading of this manuscript. This work is supported by the British Medical Research Council.