A biological system producing a self-assembling cholesteric protein liquid crystal

Previous studies on mantid oothecal protein have shown that it is an a-protein, which can be converted into ribbons by treatment in vitro with calcium ions (Rudall, 1956); also, within the colleterial glands it exists in spherulites, suitable sections of which show a double birefringent spiral (Kenchington & Flower, 1969), like those seen in cholesteric liquid crystal spherulites (Robinson, 1966; Wilkins, 1963). Electron micrographs of thin sections of such spherulites showed a parabolic pattern (Kenchington & Flower, 1969), which can be derived from oblique sections through a helicoidal structure (Bouligand, 1965). (The basis of helicoidal structure is shown in Fig. 1, p. 95). Our own work, which has proceeded independently from that of Kenchington & Flower (1969) but which was inspired by a polarized-light micro graph of a double spiral in oothecal protein from Sphodromantis centralis (Kenching ton, 1965), supports the helicoidal interpretation of the protein spherulites; but our interpretation of the units which form such helicoids differs radically from that of the above workers. Furthermore, we wish to demonstrate the self-assembly of such helicoids, and to emphasize their liquid crystalline nature.

Protein samples were obtained from the left colleterial gland of adult female Sphodromantis tenuidentata, reared from eggs in the laboratory by Mr C. W. Berg. Results were confirmed on a South African mantis, Miomantis monacha. Material for electron microscopy was prepared by fixing whole gland tubules for 2 h in 2·5 % glutaraldehyde in 0·05 M cacodylate buffer at pH 7·2 at 4°C. It was washed in several changes of buffer at 4 °C for 0·5 h, then immersed in 1 % aqueous osmium tetroxide for 1 h. The tubules were transferred directly to 70 % ethanol, dehydrated, transferred to propylene oxide, and embedded in Araldite. Thin sections were stained either in a saturated solution of uranyl acetate in 50 % ethanol, or in 2 % aqueous potassium permanganate, followed in both cases by lead citrate. Sections were examined in an AEI EM 6 B electron microscope. Microdensitometric scans of electron micrographs were made on a Joyce double-beam automatic recording microdensitometer. Polarization and phase-contrast microscopy were carried out with Zeiss polarizing and phase-contrast microscopes, and observations at controlled temperature were made using a Kofler hot microscope stage.

Fixed material was sectioned in situ in the oothecal gland. The system appears lamellar with a periodicity of 1–3 μm, and double spirals of lamellae were frequently seen both in phase contrast (Fig. 3) and in the electron microscope (Fig. 4).

Observation of unfixed material in freshly killed mantids shows that the oothecal gland is packed with lamellar spherulites best observed between crossed polaroids (Fig. 5). A typical isolated spherulite is photographed between crossed polaroids in Fig. 6. The distortion from truly spherical shape was typical in our material. Use of a 550-nm retardation plate shows the orientation of construction units to be circum ferential, birefringence being positive parallel to the lamellae. The single nature of the extinction rings (lamellae) shows the system to be uniaxial, as distinct from biaxial (double rings). The spherulites were stable for many weeks when mounted in glycerol. Electron micrographs (Fig. 14) show that the edges of spherulites appear ‘sticky’, with strands of material perhaps forming by contact with neighbouring spherulites. Rudall (1956) described the globules, which he showed to be protein, as corpuscles moving in the gland serum. We confirm this observation, but wish to extend it by distinguishing between the corpuscular mobility and the actual flow of the protein itself; the difference, which is an important one, is easily seen with a polarizing microscope. A fresh colleterial gland tubule was punctured, and the spherulites seen to coalesce as they flowed out through the puncture, forming a system with preferred orientation in the direction of flow. This system coiled backwards and forwards upon itself some distance outside the tubule. These observations show that the protein in the gland is in a liquid crystalline state.

The construction units of spherulites have been described by Rudall (1956) as fibrils 0·2–0·5 μm in diameter, and by Kenchington & Flower (1969) as fibrils 0·05–0·1 μm in diameter. Such fibrils were never seen in any of our electron micro graphs. On the contrary, the oothecal protein forms a continuous phase which can be traced at high magnification (×250000) with no discontinuities throughout a section, and which appears electron-dense after staining with potassium permanganate. The less electron-dense discontinuous phase seen in the micrographs (Figs. 4, 8, 9) has not yet been identified. It is probably pushed into a helicoidal array by the surrounding continuous phase protein. A comparable system is the moulding of pore canals into twisted ribbons by the crystalline helicoidal array of microfibrils in insect cuticles (Neville & Luke, 1969a).

The electron-light component forms a useful natural marker system, displaying the parabolic patterns typical of a helicoidal system (Fig. 8). Such patterns have been analysed for crustacean cuticle microfibrils (Bouligand, 1965), and for insect cuticle pore canals (Neville, Thomas & Zelazny, 1969).

The electron-density variation across such micrographs as Fig. 8 was measured with a microdensitometer. The rhythmical variation (Fig. 2 A) is consistent with the steady rotation (theoretically sinusoidal) of the units in a helicoidal structure as in Fig. 1. Superimposed upon this are seen the troughs caused by the electron-light discon tinuous phase (arrows in Fig. 2 A). The presence of a double spiral at the centre of a spherulite (Fig. 4) also supports a helicoidal explanation for this structure by com parison with light-microscopical observations on cholesteric liquid crystals, and is discussed below. The evidence supports the hypothesis for the helicoidal nature of cholesteric liquid crystals.

Our electron micrographs show that helicoidal structure arises extracellularly (Fig. 11), no parabolic patterns appearing in the cells. Extensive areas of parabolic patterning are restricted to the lumen of the gland, occurring at a distance of 3 μm from the apical border of the gland cells and 1 μm from the luminal cells in which the gland cells are embedded. The intervening space is filled with the so-called serum (Fig. 13). The amount of secretion present within the end-apparatus of the gland cells is insufficient to enable us to determine whether it is already helicoidal. Since there are no discontinuities over extensive volumes of the final spherulites, the products extruding from the gland cells must be capable of assembling on to the material which has already been secreted. Thus the formation of helicoidal architecture occurs by an extracellular self-assembly process in the gland lumen.

Lamellar surface structures formed in vitro by self-assembly on a glass coverslip or on the surface of a glass tube, when oothecal protein was left in locust saline for several days. These surface structures (Fig. 7) closely resemble those formed on glass surfaces by solutions of poly-y-benzyl D-glutamate and poly-y-benzyl L-glutamate in dioxan (Robinson, 1958). They provide further evidence for the self-assembling properties of mantis oothecal protein.

When extracted oothecal protein is heated on a microscope slide with a Kofler hot stage, during continuous observation between crossed polaroids, a dramatic change is seen at a critical temperature of 55 °C. The previously flowing and birefringent liquid crystalline phase abruptly changes into a static and isotropic gel. Lowering the temperature showed the change to be irreversible. The system has changed in physical state from a liquid crystal to a hydrated rubber-like gel of low tensile strength, which develops cracks on deformation, and shows reversible strain birefringence when stresses lower than that causing tensile failure are applied. Identical results were obtained with protein from both species of mantid tested.

The above procedure was repeated and the resulting gel fixed for electron micro scopy. Thin sections showed that the helicoidal pattern had disappeared, leaving a random matrix with the electron-light discontinuous phase still present (Fig. 9). (Gelling prior to fixation resulted in harder material causing the scratches in Fig. 9.) Microdensitometiic scanning confirmed the abolition of the rhythmical variation in electron density, but the electron-light component was still represented by troughs (arrows, Fig. 2B).

The general histological appearance of the secretory cells at the electron-microscope level has been described for the closely similar cockroach left colleterial gland in a pioneer paper on insect gland cell ultrastructure (Mercer & Brunet, 1959). The equivalent details of the mantid left colleterial gland have been given by Kenchington & Flower (1969). Whilst confirming the fundamentals of these descriptions, we wish to add the following details.

Prior to secretion, the oothecal protein occurs as vesicles in the cells (Fig. 12), which, by contrast with normal epidermal cells secreting cuticle, are very rich in rough endoplasmic reticulum, suggesting that the protein is synthesized for export in the gland cells themselves. (We note this feature because epidermal cells in general may obtain some of the proteins which they subsequently secrete from else where in the body via the haemocoel. This may be deduced from the electrophoresis results of Fox & Mills, 1969.) The microvillate end-apparatus through which secre tion of the mantis oothecal protein occurs is typical of insect gland cells in general (Mercer & Brunet, 1959; Gupta & Smith, 1969). As in the colleterial glands of Saturniid moths (Berry, 1968), several of the mantid colleterial gland cells contain cytolysomes with myelin-like figures.

Colleterial gland cells are set in an epithelium otherwise composed of so-called ‘chitinogenous’ cells. Whilst agreeing with this general layout, we disagree with previous workers (Mercer & Brunet, 1959; Kenchington & Flower, 1969) in the naming of these cells. The structure bordering the lumen of the organ resembles an epicuticle in ultrastructure and thickness (Figs. 10, 11, 13). Since epicuticle does not contain chitin, we therefore propose to call the cells responsible for this structure ‘lumen cells’. In the mantids they contain numerous microtubules oriented parallel to the surface of the lumen. This has previously been noted by Berry (1968) in Saturniid moth colleterial glands.

The above ultrastructural evidence supports the theory, based upon optical pro perties, of the helicoidal structure of cholesteric liquid crystals (Friedel, 1922). The mantis oothecal protein appears potentially useful for building chemical and archi tectural models of cuticle, and for experiments on helicoid self-assembly. It emerges that protein can form a helicoid in the absence of chitin (Hackman & Goldberg, 1960, have shown that chitin is absent from the oothecal protein of a mantid, Orthodera ministralis), but this does not necessarily imply that protein is the prime factor governing assembly of helicoids in arthropod cuticle.

We have suggested that the helicoidal structure of arthropod cuticle in general might arise by subsequent stabilization of a self-assembling cholesteric liquid crystal line deposition zone present as a thin region next to the cuticle-secreting epidermal cells (Neville & Luke, 19690,6; Neville & Caveney, 1969). Electron-microscope images show evidence of helicoidal structure in this deposition zone. The fact that we have shown above that it is possible to fix and visualize a cholesteric liquid crystal line system in the electron microscope, does not detract from the hypothesis.

Kenchington & Flower (1969) briefly mention the similarity between the double spiral seen in light-microscope preparations of mantis spherulites, with that of polypeptide spherulites (Robinson, 1966). We wish to extend the comparison to include the double spiral patterns in transfer RNA spherulites (Wilkins, 1963), and to stress that such spiral patterns arise because of geometrical reasons. The mathe matical derivation of double spirals in helicoidal spherulites is discussed by Pryce & Frank in Robinson, Ward & Beevers (1958). They show that sections through a helicoidal spherulite will always contain a double spiral pattern except in the plane of the single radial line of disinclination, which is also a geometrical consequence of their construction. Double spirals have also been seen in sections of tubercles in crab cuticle and their origin is beautifully demonstrated in diagrams by Bouligand (1965). They also occur in sections of corneal lenses of some arthropod eyes (Horridge, 1969; S. Caveney, unpublished), where they arise from a hemisphere of cuticle with helicoidal construction.

With regard to the units from which the helicoids are built, we disagree with the interpretation of Kenchington & Flower (1969). They described the units as electron densely staining ‘fibrils’ arranged at angles of 18 ° to each other, for which they have specifically constructed a perspex model. By contrast, we regard the construction units as unresolvable by electron microscopy of thin sections. A more likely candidate for a unit could be the twin-coiled a-helices postulated by Rudall (1956) on the basis of X-ray diffraction of artificial fibres pulled out from the viscous protein in the gland. It is significant that the building units of other helicoidal systems are also asymmetrical (cholesteryl derivatives), often with helical components (transfer RNA) or even totally helical (synthetic polypeptides and DNA: Robinson, 1961).

We find no discontinuities in the electron densely-staining matrix, and this is supported by the gelling experiments in which a continuous isotropic gel was formed. One common feature of all the helicoidal systems we have worked with is that they are capable of showing a very wide range of pitch (see Fig. 1, p. 95), reflecting a wide variation of angle between successive planes of units.

The formation of a gel from mantis oothecal protein clearly differs from the shrinkage of collagen, since the latter phenomenon is readily reversible. A closer analogy is the formation of gelatin from collagen (Harkness, 1961), which involves the irreversible denaturation of the secondary structure, and the formation of a new random configuration with higher entropy. We appear to have converted an ordered liquid crystal into a random gel, perhaps by the breakdown of the H-bonds in the secondary structure of the original units, followed by the formation of a new random secondary structure. The breaking of the original H-bonding (for instance in the twin a-helices postulated by Rudall, 1956), could explain the dramatically sudden disappearance of the helicoidal architecture.

We wish to thank Mr C. W. Berg for rearing the mantids and Mr S. Caveney for obtaining Miomantis monacha. We thank Prof. J. W. S. Pringle, F.R.S., for his comments on the manu-script. Finally, our grateful thanks to the Agricultural Research Council for full financial support.