ABSTRACT
Cytochemical methods have been used to follow the incorporation of enzymes and antigens in the cellulosic intine of the pollen grain of Gladiolus gandavensis. Acid phosphatase, esterase and ribonuclease were first detected in the developing intine during the vacuolate period, and by the end of this phase activity was present throughout the intine, especially in the area of the colpus, where this wall layer is thickened. Enzyme activity was also noted in the peripheral region of the protoplast during the period of intine growth. Succinic and NADH (reduced nicotine adeninedinucleotide) dehydrogenase activity was detected in the outer zone of the protoplast, and, at the end of pollen development, in the interbacular cavities of the exine and the pollen-coat materials, but at no time were these enzymes found in the intine.
The appearance of the antigens was traced during pollen development. Antigens were first detected in the early vacuolate period, soon after the release of the spores from the meiotic tetrads. They were found to be associated with the intine from the first deposition of poly-saccharides in this layer. As the intine thickened, so the amounts of associated antigens were found to increase, especially in the neighbourhood of the colpus.
In slurries prepared using mature pollen, enzymes and antigens were found to diffuse into the medium even with preparation times as short as 30 s. The leakage was over the whole of the exine, not only in the vicinity of the colpus; this may be related to the thinness of the exine in Gladiolus and the presence of pores in the tectum.
Electron micrographs showed the presence of vesiculate and tubular cavities in the intine, containing fibrillar or granular material. These inclusions are presumed to be the sites of the enzymes and antigens, since in a related genus, Crocus, acid phosphatase has been shown to be present in the corresponding cavities.
Gel diffusion and immunoelectrophoresis showed the presence of 4 major antigenic fractions in the leachates from mature Gladiolus pollen. Tests with the antisemm against leachates from species of the related genus, Iris, showed 1 cross-reacting fraction.
INTRODUCTION
The presence of proteins, including acid phosphatase, in various strata of the pollen grain walls of Paeonia and Amaryllis spp. was demonstrated by Tsinger & Petrovskaya Baranova (1961), who emphasized that the principal site of deposition was in the cellulosic intine. These observations have recently been confirmed and considerably extended by Knox & Heslop-Harrison (1969, 1970a) in a cytochemical study of the pollens of more than 50 flowering plant species and of a species of Pinus. In all cases the major site of the proteins, including those showing hydrolytic activity (ribonuclease, acid phosphatase, esterase, protease and amylase), is the cellulosic intine, especially at the apertures, where this inner wall layer is usually thickened. By the use of electron microscope cytochemistry, acid phosphatase activity has been shown to be associated with vesicular and tubular inclusions within the intine cellulose in Crocus vernus (Knox & Heslop-Harrison, 1970c), and it now seems likely that similar intine inclusions described in ultrastructural accounts of other flowering plant pollen walls (literature list in Knox & Heslop-Harrison, 1970a) will also prove to be the sites of wall-held enzymes.
The intine has also recently been shown by immunofluorescence methods to be the principal source of the rapidly leachable antigens of the pollen grains of Ambrosia spp. (ragweeds) and Gladiolusgandavensis (Knox, Heslop-Harrison & Reed, 1970). Antigen E, one of the most active human allergens from ragweed pollen (King, Norman & Lichtenstein, 1967), is located in the intine, from which it is speedily lost on moistening (Knox& Heslop-Harrison, 1971).
Intine synthesis begins in the early vacuolate period of pollen development (Heslop Harrison, 1968 a, b), and it has been shown for Cosmos bipinnatus that stratified, polysome-bearing endoplasmic reticulum is conspicuous in the cortical cytoplasm during the main part of intine growth (Knox & Heslop-Harrison, 1970a) at a time when ribbons or leaflets of material, probably proteinaceous in character, are being incorporated with the cellulosic strata. Gladiolus, with its large pollen grains (about 75 μ m in diameter at maturity) and thick intine, has provided ideal material for a more intensive study of the ontogeny of the intine and its included proteins, and this paper reports the results of such an investigation based upon the use of cytochemical and immunological techniques. The cytochemical methods adopted for the localization of hydrolytic enzymes in mature pollen grain walls in previous work have now been applied to earlier developmental stages in Gladiolus, and the earliest synthesis of antigens associated with the intine and other strata has also been followed using the immunofluorescence techniques previously employed (Knox et al. 1970), modified to enhance resolution.
MATERIALS AND METHODS
Plant material
Fresh pollen of Gladiolus gandavensis van Houtte cv. was taken from cut flowers obtained from a commercial source in Madison, Wisconsin, and stored Gladiolus pollen was provided by the Allergy Division of Greer Drug & Chemical Co., Lenoir, N.C. Most of the developmental study was based upon plants obtained from commercial nurseries in Australia. No differences in the pollen characteristics or ontogenetic behaviour were noted between the plants from different sources, and as mentioned below the immunological properties of pollen leachates were identical.
Specification of developmental stages
In addition to standard cytological criteria, use was made of the fluorochromatic reaction (FCR) for categorizing developmental stages in the young spores (Knox & Heslop-Harrison, 19706). Spores and pollen from anthers at the different developmental stages were dissected into a freshly made 0 · 5 – 0 · 8 M sucrose medium containing fluorescein diacetate at saturation concentration. This non-fluorescent substrate readily enters the spores and is cleaved by esterases, releasing fluorescein. Fluorescein, being polar, does not penetrate the plasmalemma and is retained within the cell, where it is readily detectable by fluorescence microscopy. The FCR allows an immediate and precise evaluation of the state of vacuolation of the spore, an important developmental criterion.
Tissue preparation
Anthers, or slurries of pollen extracted from the anther loculi, were prepared for freeze sectioning as previously described (Knox & Heslop-Harrison, 1970a; Knox, 1970). For electron microscopy, mature pollen grains were fixed in 2% OsO4 buffered at pH 7 · 2 for 12 h at 2 CC. The pollen suspension was centrifuged, washed in buffer, dehydrated and embedded in the low viscosity medium of Spurr (1969). Ultrathin sections were post-stained with uranyl acetate and lead citrate following standard methods.
Enzyme localization
Acid phosphatase was localized as described by Knox & Heslop-Harrison (1970a), using the methods of Barka & Anderson (1962) with a-naphthyl acid phosphate and naphthol AS-BI phosphate as substrates.
Ribonuclease was localized by the method of Enwright, Frye & Atwal (1965) as modified by Knox & Heslop-Harrison (1970 a).
Amylase was localized by a substrate film method. Microscope slides were coated with 6 % starch (hydrolysed starch, Connaught Laboratories, Ottawa) and air-dried. Sections 2 – 4 μ m thick, fresh, or pre-fixed for 1 h in methanol prior to freeze-sectioning, were rapidly thawed on to the starch films. The slides were incubated for 2 – 4 min at room temperature before stopping the reaction with iodine – potassium iodide solution, which also served to stain the starch. To check the extent of digestion of the films, some sections were rinsed in running tap water before or after iodine staining for comparison. Controls were run by mounting sections on films pre-stained in iodine solution and air-dried before use. Under these conditions, no digestion of the films was observed.
Dehydrogenases were detected using 0·05 mg/ml tetranitro-blue-tetrazolium as coupler,with 0·05 M sodium succinate as substrate for succinic dehydrogenase and 0·002 M NADH as substrate for NADH dehydrogenase in 0·05 M tris buffer, pH 7·2 (Bryant & Nicholas, 1966). Sections were incubated for 5 min at room temperature. Controls were run by omitting sub-strate.
Polysaccharide localization
The periodic acid-Schiff (PAS) procedure was used for the detection of wall polysaccharides (Jensen, 1962). Controls were run by omitting the periodate oxidation. Slides were incubated in Schiff reagent for 30 min followed directly by 3 changes of 0·5 % potassium metabisulphite solution, before washing in water, dehydrating and mounting.
Immunofluorescence technique
Antisera were prepared against Gladiolus pollen leachates as previously described (Knox et al. 1970). The specificity of the antisera was tested using the gel-diffusion and immune electrophoretic methods of Ouchterlony (1958) using the micro-slide modification (Gelman Instrument Co., Ann Arbor, Michigan). Gradient diffusion tests were carried out as described by Allison & Humphrey (1960) and Crowle (1961).
In cross-reaction tests made with other members of the Iridaceae, the pollen leachate was prepared as for Gladiolus (Knox et al. 1970).
Modifications of the immunofluorescence technique previously used with pollen have been introduced to improve both specificity and resolution. Rhodamine B isothiocyanate (RITC) was originally used as the fluorescent label in both direct and indirect methods (Knox et al. 1970). The choice was made because the orange fluorescence of RITC contrasted well with the green autofluorescence of the exine layers, high in aqueous mounts, while differentiation was rather more difficult with the more commonly favoured fluorescein isothiocyanate (FITC),which itself has a green fluorescence. However, there are disadvantages in the use of RITC. Its fluorescence is less intense than that of FITC (Goldman, 1968), and it has proved difficult to detect small amounts of RITC-labelled precipitins against the weak general fluorescent back ground of cell contents. Accordingly, FITC-labelling has been adopted in the present study, and the difficulty offered by the autofluorescence of the exine has been minimized by using permanent dry-mount preparations as suggested by Bastos et al. (1968). In these mounts, exine fluorescence is much reduced, and the fading of the specific fluorescence contributed by the stains is retarded. It has proved possible to localize FITC-labelled precipitins with high resolution, even when the antigens have been present in very small amounts, as in the diffusate from grains in slurries.
The complete procedure adopted was as follows. Frozen sections prepared by the methods of Knox (1970) were mounted on slides previously coated with 1 % gelatin, and air-dried before use. Sections were used without fixation, or were post-fixed in ethanol or methanol for 1 h at room temperature and re-dried. Antisera were then spread over the sections, and the preparations were incubated in Petri dishes in a moist atmosphere at room temperature for 40 min. As controls, duplicate sections were incubated with normal, pre-immunization rabbit serum.
The slides were then washed in saline for 10 min, rinsed in distilled water and air-dried to prevent the loss of sections during further processing. FITC-labelled goat anti-rabbit gamma globulin (lyophilized, Pentex, Research Products Division, Miles Laboratories, Kankakee, Illinois) was made up at a concentration of 2 mg/ml in distilled water, and spread over the sections previously treated with immune or control sera. The slides were incubated for a further period of 40 min in a moist atmosphere at room temperature, washed in saline for 10 min, rinsed in distilled water and air-dried. The sections were then mounted in Eukitt (Kindler, Freiburg), a non-fluorescing mounting medium, and stored in the dark. No appreciable fading occurred over several months storage, and fading during observation was greatly reduced, as noted by Bastos et al. (1968) for acridine orange fluorescence under similar conditions.
OBSERVATIONS
The developmental sequence
By the use of the FCR and reference to other criteria visible with the light micro scope, the developmental periods described in Table 1 can be distinguished. The young spores show no FCR while within the meiotic tetrad (Knox & Heslop-Harrison, 1970b), but on release from the callose wall the reaction is intense. During the early period of expansion without vacuolation, fluorescein accumulates through the entire volume of the protoplast. After the onset of vacuolation, the reaction is restricted to the peripheral protoplasmic layer, because, as described in the earlier paper, fluorescein does not penetrate the tonoplast and so is held out of the vacuole. At the end of the vacuolate period the reaction once more extends through the entire volume of the grain.
The synthesis of the intine does not begin in Gladiolus spores until the onset of vacuolation. Thickening continues during the period of expansion accompanying vacuolation, showing that new wall material must be added at a considerable rate. Resorption of the vacuole begins about the time of tapetum dissolution, and thence forth there is no further deposition of intine material.
Fine-structural features of the mature pollen wall
Fig. 1 is a low-magnification electron micrograph of the wall of a mature pollen grain of Gladiolus sectioned approximately radially in the neighbourhood of the colpus. The intine is seen as an electron-transparent zone between the plasmalemma and the exine, thickened markedly at the colpus. The exine is tectate and mostly smooth, although a few spines are present (Fig. 1) as well as occasional pores (Fig. 2). As in other Iridaceae, exine stratification is not well defined (Erdtman, 1966). The bacula are irregular, and the inter-bacular cavities are injected in the mature grain with a material with approximately the same electron density after osmication as sporo pollenin. This material is extractable with ether, and is probably therefore lipid in nature (see Fig. 27). It is likely to be a form of pollenkitt, derived from the tapetum towards the end of the maturation period. The nexine is well marked, and it is this layer which continues at the colpus to form a thin, irregular sheath over the thickened intine.
The microfibrillar structure of the cellulose of the intine is well seen in Figs. 2 and 3. Tubular cavities or elongated vesicles containing filamentous or amorphous electron opaque material are present in all parts of the intine. In the non-apertural regions these lie towards the outside of the intine, and frequently appear to be directly apposed to the nexine (Fig. 2). It is to be noted, however, that in the pollen of the Iridaceae, as in many other types, there is some interbedding of the nexine material and the outer layers of the intine proper. The long axes of the cavities in the non-apertural regions appear mostly to be oriented tangentially. In the region of the colpus where the intine is thicker, the outer tangentially oriented cavities are stacked 4 or 5 in depth, and in addition there are irregular tubular cavities within, lying with the long axes at right angles to the plasmalemma (Fig. 3).
The relationship of these intine inclusions with those described in other species and the likelihood that they represent the physical sites of intine-associated proteins are matters discussed further below.
Wall-associated enzyme activity during spore development
Acid phosphatase, ribonuclease and amylase
There was no indication of the association of these enzymes with spore walls in the pre-vacuolate and early vacuolate period. Examples of spores in the early vacuolate period (anther length 8-9 mm) are shown in Fig. 7 following the reaction for acid phosphatase. Some cytoplasmic activity is present, but there is none associated as yet with the polysaccharide of the intine (Fig. 6).
Spores of anthers 10 – 11 mm in length are in the mid-vacuolate period, and these show intense cytoplasmic acid-phosphatase activity, with what is presumably intine incorporation particularly at the colpial side of the spore (Fig. 12). A thin line of amylase activity was observed in substrate film tests made with spores at this develop mental stage.
Spores in the late-vacuolate period (anther length 13 – 15 mm) show conspicuous wefts of acid phosphatase activity in the intine, and it is possible even by optical microscopy to discern that these are drawn out in a plane at right angles to the plasma lemma in the vicinity of the colpus (Fig. 15). The cortical region of the spore proto plast, the probable site of enzyme synthesis, also shows activity at this time. Mature pollen of Gladiolus from anthers 18 – 21 mm in length shows intense intine-associated activity (Knox & Heslop-Harrison, 1970a). By the time of pollen maturation (Fig. 18) activity in the outer zone of the protoplast of the vegetative cell has subsided, pre sumably with the termination of enzyme synthesis in this site.
The results with acid phosphatase left no doubt about the sites and timing of the incorporation of this wall-associated enzyme, and those obtained from the ribonuclease and amylase tests showed that the synthesis and incorporation of these 2 enzymes must follow a broadly similar pattern.
Evidence of diffusion of the enzymes from the wall sites in mature pollen grains was observed in both cytochemical and substrate film tests. Amylase activity is detectable by the substrate film method in the intine, especially at the colpus, and also in the cavities of the exine (Fig. 28). In rapidly prepared slurries where the period of moistening before freezing was 15 – 30 s, activity was restricted to the immediate vicinity of the exine surface (Fig. 28D, F), but when the slurry was left for 5 min at 5 °C before freezing, activity was clearly evident in a wide zone around the grains (Fig. 28 c, E).
Only when the grains were briefly pre-fixed in methanol was activity largely confined to the intine and the exine cavities (Fig. 28A). Although the effect is best seen with the substrate film method used for amylase, there were indications that the other hydro lases behaved in a similar fashion, since the gelatine around the grains often showed some reaction product.
As demonstrated by Knox & Heslop-Harrison (1970 a), proteolytic enzymes also diffuse from the grains and digest the gelatine, permitting a more rapid release of the other enzymes and proteins into the surrounding medium. It is of interest that the amylase activity detected in these substrate films is released into the gelatine in funnel like wisps of activity, as clearly seen in Fig. 28 F, from a slurry prepared within 15 – 30 s. Fig. 28 c and E shows slurries prepared and stored for 5 min at 5 °C prior to freezing.
In the lower end of Fig. 28 c, wisps of activity in gelatine, conspicuously funnel-shaped, can be seen in transverse section, emanating from a grain below the plane of sectioning. Amylase activity is demonstrable only at the periphery of the ‘funnels’.
Dehydrogenases
Succinic dehydrogenase and NADH dehydrogenase were found to be confined to the peripheral region of the protoplast during the early vacuolate period. Intense activity was invariably present in the surrounding tapetal tissue at this time (Fig. 25). At pollen maturation, these enzymes were detected in 3 sites: (a) in the protoplast of the vegetative cell, and especially at the periphery; (b) in the inter-bacular cavities of the exine, clearly to be seen in comparison with control sections; and (c) most strikingly in the surface materials of the exine (tryphine). The localization may be seen in Fig. 26. At no time was activity detected in the intine, so it must be supposed that these enzymes have a different ontogenetic history from that of the hydrolases, as is indeed to be expected from their different metabolic roles. The presence of the activity in the tryphine and cavities of the exine is perhaps not unexpected because the surface materials of the mature pollen are derived from the breakdown of the tapetum, which as already noted shows intense dehydrogenase activity as an intact tissue in earlier developmental stages. The distribution of lipids closely parallels the dehydro genase pattern (Fig. 27).
Immunological characterization of the diffusible antigens of Gladiolus and Iris pollen
Immunodiffusion tests of leachates of Gladiolus pollen from both American and Australian sources against the anti-Gladiolus serum show 3 major precipitin lines. That nearest the antiserum well is characteristically thick and ‘fuzzy’ (Figs. 29-31). Tests with leachates of pollen from 10-mm anthers (early vacuolate period) showed 3 precipitin bands as in mature pollen. In order to test the relative mobilities of the various fractions, a gradient diffusion test was carried out (Fig. 32). From this test, it is clear that the thick, fuzzy fraction shows the greatest mobility through the gel, the precipitin line forming at an angle of about 62 ° (when antiserum trough is 90°), compared with the other 2 fractions whose precipitin lines have smaller angles, indicating differences in diffusion kinetics (Allison & Humphrey, 1960; Crowle, 1961). When subjected to immunoelectrophoresis in tris-barbital buffer, pH 8 · 8, ionic strength 0 · 03, 4 precipitin arcs are clearly distinguishable (Fig. 33). A dense precipitin arc is present from the origin and trails towards the cathode, while 3 arcs are present towards the anode. The major protein fraction thus appears to be positively charged, and basic, with a high isoelectric point above pH 8 · 8, since it migrates towards the cathode during the electrophoresis. The other 3 anodal fractions are negatively charged and presumably acidic.
When leachates prepared from pollen of Iris unguicularis were tested by immunodiffusion against anti-Gladiolus serum, only one precipitin line developed. This appeared as a simple’ spur’ connected with the thick ‘fuzzy’ precipitin line of Gladiolus antigen (Figs. 29, 30), which gradient diffusion tests have shown is the most rapidly diffusing fraction. When electrophoresed alongside Gladiolus antigen, the common Iris antigen arc corresponds to the acidic protein fraction of Gladiolus leachate which shows the greatest mobility, being closest to the anode. Leachates from cultivars of Iris xiphium and I. germanica showed similar immunodiffusion patterns to that of I. unguicularis when tested against the anti-Gladiolus serum.
Wall-associated antigens during spore development
The matrix material of the primexine, as seen in the spores of the late tetrad, at the time of release and during the pre-vacuolate period, is PAS-positive, indicating its polysaccharide nature. However, there were no associated antigens, and the spore wall showed only faint, non-specific staining with FITC-labelled antibody (Fig. 4).
Specific labelling of wall antigens was first observed in spores at the early vacuolate period, with a diameter of about 36 /<m, derived from anthers 8-9 mm in length. In these spores, a rather diffuse zone of fluorescence could be detected over the entire inner surface of the exine (Fig. 5), and tests on neighbouring sections showed that polysaccharide was already present in the same site (Fig. 6), indicating that intine synthesis had begun. The poor definition of the zone of precipitin in these early stages probably reflected the fact that the antigens were being synthesized in the cortical regions of the cytoplasm and being incorporated simultaneously with the wall poly saccharides at the plasmalemma.
Spores in the mid-vacuolate period from anthers 10-12 mm in length showed a thicker precipitin layer over the inner surface of the exine (Figs. 9, 10), again corres ponding closely to the distribution of the PAS-positive material (Figs. 8, 11). A mark edly thicker deposition was clearly apparent at the colpial end in favourably oriented sections. In the late-vacuolate interval (spores from anthers about 13 mm in length) the deposition of the antigens appeared to be nearing completion, judging from the thickness of the zone of labelled precipitin, especially at the colpus (Fig. 13), comparable to that of the intine polysaccharides (Fig. 14).
With pollen mitosis and the resorption of the vacuole, the pollen grain enters the final phase of maturation (Table 1, p. 213). Micrographs showing the localization of the intine-associated antigens in mature pollen of Gladiolus obtained with RITC labelled antiserum by both direct and indirect methods have been published elsewhere (Knox et al. 1970). While there is no question that the antigens are mostly associated with the intine, their ready mobility in later developmental stages must be emphasized. Even in rapidly prepared slurries (Figs. 16, 17) the precipitins often formed diffusion halos around the grains, and there were losses even with pre-fixation for 1 h in methanol (Fig. 19). When slurries were incubated for 5 min at 5 °C before freezing, the precipitins were mostly present in the surrounding medium (Fig. 20), showing that the antigens diffuse out very rapidly under these conditions. It is noteworthy that with Gladiolus pollen, the exine of which is comparatively thin, the antigens are lost from all over the surface, not only from the region of the colpus. This is clearly seen in Figs. 16, 17, 19 and 20, where the precipitins appear as radially oriented wisps. Antigens could not be detected at all in preparations of Gladiolus pollen leached in saline for 2 h prior to freeze-sectioning, showing that there is no tenaciously held component. In earlier work it was shown that the wall-held enzyme activity of Gladiolus pollen may be removed completely by leaching or electrophoresis (Knox & Heslop-Harrison, 1970a).
Wall-associated antigens in Iris unguicularis
As mentioned above, the diffusible antigens from the pollen of I. unguicularis cross react with antibodies present in the anti-Gladiolus serum, and the immunofluorescence method was used to localize the cross-reacting fraction. The pollen grain of I. un guicularis is about twice the diameter (145 μ m) of that of Gladiolus, and extremely good resolution was often obtained. In rapidly prepared slurries, most of the cross-reacting material remained associated with the thick intine, with comparatively little diffusion into the medium (Fig. 21). When diffusion did occur the antigens moved out in radially oriented streams (Figs. 22, 23), and the localization of these was such as to suggest that the passage was through particular areas of the sexine, perhaps through pores of the kind illustrated for Gladiolus in the electron micrograph of Fig. 2.
DISCUSSION
The main finding from the developmental study is that the incorporation of the acid hydrolases and the diffusible antigens is closely correlated with the growth of the intine in the young spore. From the early vacuolate period when the polysaccharide material of the intine could first be detected cytochemically, the immunofluorescence method showed the presence of associated antigens. Acid phosphatase could not be detected in the very earliest stage of intine deposition, perhaps because of the relative insensitivity of the method, but thereafter the enzyme could be localized quite pre cisely in the intine throughout its development. The other acid hydrolases follow the same pattern as acid phosphatase, and it is entirely probable that they also are incorporated in the growing intine almost from the beginning.
The light-microscopic evidence shows that the incorporation of enzymes and anti gens into the intine of Gladiolus spores is not restricted to the area of the colpus, but extends over the entire surface, including the non-apertural region. Only towards the end of wall development is there a marked difference in the thickness of the intine at the colpus, or in the concentration of enzymes and antigens there. The fine-structural features of the mature intine can readily be interpreted in the light of the develop mental results. A high-resolution cytochemical study of the intine of Crocus vernus, also of the Iridaceae, has shown that acid phosphatase is localized in tubular and vesiculate inclusions in the intine (Knox & Heslop-Harrison, 1970c), so it may be confidently accepted that the corresponding cavities of the Gladiolus intine contain proteinaceous material inserted during intine growth (see also the developmental evidence from Cosmos bipinnatus, Knox & Heslop-Harrison, 1970a). In the non apertural intine the vesicles are concentrated in the outermost, and so youngest, zone; and this may be correlated with the fact that the incorporation of enzymes and antigens begins with the earliest deposition of the intine material. At the colpus the tangentially oriented peripheral tubules and vesicles are present, and also in the inner part, radially oriented inclusions which must be incorporated during later growth. Again, the observation is in accord with the fact that a concentration of enzyme activity and antigens in the colpial region becomes apparent only in the late-vacuolate developmental period. A striking feature is that in certain of the acid phosphatase preparations the reaction product in the inner part of the colpus was distributed in radially oriented ribbons, resolvable even with the light microscope. This strongly supports the view that the enzyme is located in the tubular inclusions visible in electron micrographs, as has been shown for Crocus vernus (Knox & Heslop-Harrison, 1970c).
The pathway of movement of the enzymes and antigens out from the wall sites in Gladiolus and Iris merits comment, since observation suggests that the loss is from all over the surface of the exine, and not wholly through the colpus. The contrast with species possessing a thicker exine may be noted; with such pollens the bulk of the intine-held proteins passes out through the apertures on moistening, as was found for the tricolpate grains of the Compositae and for the polyporate grains of the Malvaceae (Knox & Heslop-Harrison, 1970 a and unpublished observations). Because the intine held proteins do pass through the exine layers in Gladiolus, it is not possible to say whether or not the enzyme activity detectable in the inter-bacular spaces results from incorporation during development or from movement during processing. In Crocus vernus, a species with a very thick intine, enzyme activity in the mid-zone of the intine is readily distinguishable from that in the exine cavities, both by optical and electron microscopical methods (Knox & Heslop-Harrison, 1969, 1970a, c); similarly, in the Compositae there seems little doubt that the enzymes held in the cavea of the exine are injected from outside of the grain during maturation and do not move there only at the time of imbibition. A convincing demonstration for Gladiolus must await the development of more effective methods for stabilizing the wall-associated proteins in mature grains.
Turning to the question of the identity of the antigens present in the Gladiolus pollen leachate, their heterogeneity is evident from the immunological tests. Using gel diffusion methods, 3 precipitin bands were detected, one of which, the most rapidly diffusing, cross-reacts with Iris pollen leachate (Figs. 29-30). The spur formed indicated only partial identity between the antigens of Gladiolus and Iris, the thick fuzzy precipitin band suggesting that several antigens may be present, in addition to that with the common antigenic determinant. By electrophoresis in agarose at pH 8 · 8, 4 precipitin arcs were detected, each immunologically distinct. All except the arc nearest the anode were specific to Gladiolus in tests against I. unguicularis pollen leachate. The anti-Gladiolus serum showed a cross-reaction with an Iris antigen which had migrated towards the anode in an exactly similar way to the most rapidly migrating Gladiolus antigen. This specificity is not unusual with pollens of flowering plants, as shown by the extensive gel-diffusion studies of Wodehouse (1954a, b, 1955, 1957), who observed cross-reactions between various genera and species of Compositae, Chenopodiaceae and Gramineae. Cross-reaction does not necessarily indicate that the antigens are identical (Feinberg & Grayson, 1959). Indeed Antigens E and K of Ambrosia (ragweed) pollen are chromatographically separable but immunologically similar (King et al. 967). A striking feature of the Gladiolus results is that the protein fraction with apparently a common antigenic determinant with Iris showed the greatest mobility with all 3 methods used (gradient diffusion, immunoelectrophoresis and immuno fluorescence), presumably indicating a lower molecular weight.
The possibility that the different antigens of Gladiolus occupy different sites in the wall, or are synthesized at different periods, must of course remain open. The methods adopted so far would prevent detection of any such differences, although immunodiffusion tests of pollen from early vacuolate period showed similar precipitin bands to those of mature pollen. It might well be possible to gain evidence for some of the fractions, for example, by titrating out the cross-reacting component using Iris pollen leachates before attempting localization.
Although the wall-held enzymes and antigens occupy the same general sites, positive identification of one with the other has been avoided in the foregoing. The hydro-lytic enzymes may be among the antigens, but it is probable that they do not contribute greatly to the total antigenic activity of leachates. Assays for individual enzymes in such leachates show that activity is relatively low for the total amount of protein present (R. B. Knox & J. V. Jacobsen, unpublished data), and further reason for sup posing that the enzyme proteins represent only a small fraction of the total comes from inter-familial comparison, as between Iridaceae and Compositae. Several enzymes have been shown to be common to the intines of Gladiolus and Ambrosia (Knox & Heslop
Harrison, 1970a), but in the reciprocal tests using anti-Gladiolus and anti-Ambrosia sera only a trivial level of cross-reaction has been found (Knox et al. 1970). This result could arise from immunological differences in proteins possessing the same enzymic properties in the 2 species, but it is more readily explained on the assumption that the bulk of the antigenic material has other functions, perhaps being concerned with inter and intra-specific incompatibility reactions – that they are, in fact ‘recognition’ sub stances (Heslop-Harrison, 1967; Knox et al. 1970). The ingenious work of Lewis, Burrage & Walls (1967) on the proteins implicated in the intraspecific incompatibility reactions of Oenothera show that these rapidly diffuse from single moistened grains in a manner strongly suggesting that they are derived from wall sites.
Certain of the antigenic fractions present in Gladiolus pollen leachates are probably quite similar to those responsible for human allergies in various wind-borne pollens. These have been characterized as acidic globular proteins with molecular weights between 30000 and 40000 (King et al. 1967; Johnson & Marsh, 1966). Evidence is given elsewhere (Knox& Heslop-Harrison, 1971) that Antigen-E, one of the principal antigens present in Ambrosia pollen, is located in the intine.
ACKNOWLEDGEMENTS
I am particularly grateful to Professor J. Heslop-Harrison, F.R. S., for his enthusiastic support of this work and for valued advice and criticism of the manuscript. Part of this work was supported by the U.S. National Science Foundation (grant no. GB 7775 to Professor Heslop Harrison) and by the Graduate School of the University of Wisconsin; I thank both these agencies. I wish to thank Dr Charles E. Reed of the Department of Medicine, University of Wisconsin, for preparing the antisera and for helpful comments on the results, and Dr W. F. Dudman of C.S.I.R.O. Division of Plant Industry, Canberra, and Dr M. D. Poulik of the Child Research Center, Detroit, Michigan, for assistance with immunological procedures; Mr A. Pyliotis of C.S.I.R.O., Canberra, for preparing the electron micrographs; and Mrs K. Marshall for helpful technical assistance. The material for the developmental study was kindly supplied by Bear’s Nurseries, Brisbane, Queensland.
Figs. 1 – 3. Electron micrographs of osmium-fixed Gladiolus pollen.
REFERENCES
Figs. 5 – 7. Freeze-sectioned pollen of Gladiolus. Spores in early vacuolate period from 8-mm anthers.
Figs. 10 – 12. Freeze-sectioned pollen of Gladiolus. Pollen in mid-vacuolate period, from 11– 12 mm anthers.
Figs. 13– 15. Freeze-sectioned pollen of Gladiolus. Pollen in late vacuolate period from 13– 15 mm anthers, × 1200 approx.
Figs. 16– 18. Freeze-sectioned mature pollen of Gladiolus prepared from 18– 21 mm anthers. Pollen from slurry prepared as rapidly as possible - only 15– 30 s before freezing.
Figs. 21-24. Freeze-sectioned pollen of Iris unguiciilaris. (ci, colpial intine; i, intine.) × 500 approx.
Figs. 25-27. Freeze-sectioned pollen of Gladiolus.
Figs. 29 – 31. Immunological reactions of the diffusible antigens from Gladiolus and Iris pollen. Serological reactions by Ouchterlony gel diffusion.