Introduction

The pollen wall is the most complex wall found in plants. To understand regulation of pollen wall development, a detailed ontogenetic study, capturing each stage of development at a high level of resolution, is needed. The tetrad period, understood as crucial for pattern determination in the pollen wall by some authors (Dickinson 1970; Heslop-Harrison 1972; Blackmore and Barnes 1987; Gabarayeva 2000, 2014; Vinckier and Smets 2005; Taylor and Osborn 2006; Taylor et al. 2013, 2015), is most often underestimated (or represented by one stage) by many investigators, whereas almost all the main characters of the future exine are defined in this period. Consequently, observations presented here focus on the tetrad period of development.

Our next point of special interest concerns mechanisms, underlying sporoderm construction processes, which are not under direct genetic control. The pioneering idea of the universal importance of self-organization for pattern formation in nature by Thompson (1961) was then picked up by several biologists (Heslop-Harrison 1972; Gerasimova-Navashina 1973; Sheldon and Dickinson 1983; Dickinson and Sheldon 1986; van Uffelen 1991; Hemsley et al. 1992b, 2003; Collinson et al. 1993; Gabarayeva 1993, 2000, 2014; Southworth and Jernstedt 1995; Hemsley and Griffiths 2000) and discussed in reviews (Wellman 2004; Blackmore et al. 2007; Ariizumi and Toriyama 2011). The essence of our hypothesis (Gabarayeva and Hemsley 2006; Hemsley and Gabarayeva 2007) on the role of self-assembly processes in spore and pollen wall development is that the sequence of sporoderm developmental events represents the sequence of self-assembling micellar mesophases, initiated by genomically given parameters of surfactant glycoproteins and sporopollenin (SP) precursors and monomers (phenyl-propanoids and fatty acids) at increasing concentrations. Correspondingly, our goal in this study is to determine whether our hypothesis, which proved to be applicable to a wide diversity of other plants (about 30 species from remote taxa, viz. from Cyatheaceae (leptosporangiate fern), Cycadaceae, gymnosperms, basal and advanced angiosperms—see our papers in References), also applies in Polemonium.

Another point in question for discussion, based on observations of developmental events in Polemonium and other species in the course of exine development, is the localization of SP precursor synthesis. Is this synthesis localized exclusively to the tapetum, or does it also proceed in microspores? Pro and contra reasons for this are discussed as one of the fundamental questions.

Polemonium caeruleum L. is a hardy perennial herbaceous flowering plant and produces cup-shaped, lavender-colored, or white flowers. The plant usually reaches a height from 45 to 60 cm. The plant is native to damp grasslands, woodlands, meadows, and rocky areas in temperate areas of Europe and Asia. The plant flowers in June–July during 15–20 days.

We have chosen this species for our study because its pollen wall has not been ontogenetically studied and because it was easy to collect many flower-buds at different stages of development. We believe that self-assembly is a kind of universal mechanism, taking part in spore and pollen development. In essence, any species is suitable for our studies, if only a plant is rich in flower-buds/sporangium.

Pollen grains of Polemonium caeruleum are spherical, pantoaperturate, with ornamented aperture membrane; the structure of the pollen wall is semitectate-columellate (Halbritter and Svojtka 2005). No ontogenetic studies on pollen wall of Polemonium have been previously undertaken.

Material and methods

Flower-buds of Polemonium caeruleum (L.) were obtained from the Garden of the Komarov Botanical Institute, St. Petersburg for 3 years, 80–100 buds every year. Stamens were fixed for 24 h at 20 °C in 3% glutaraldehyde in 0.1 M cacodylate buffer pH 7.4 with 2.5% sucrose and lanthanum nitrate (1%) for better preservation of the plasma membrane glycocalyx. After post-fixation in 2% osmium tetroxide for 4 h at 20 °C followed by acetone dehydration, the samples were embedded in a mixture of Epon and Araldite. Ultrathin sections were contrasted with 1% aquatic solution of uranyl acetate and 0.2% lead citrate. Sections were examined with a Hitachi H-600 transmission electron microscope (TEM). For scanning (SEM) observations, mature pollen grains were mounted on stubs, coated with gold/palladium and examined with Jeol JSM-6390.

Results

Early tetrad stage

After the completion of meiosis, four early microspores of each tetrad, surrounded by the callose special cell wall, are united in planar (Fig. 1a) or tetrahedral tetrads. Remnants of the cytomictic channels which united the late spore mother cells are still evident (Fig. 1a, arrows). A large nucleus with a nucleolus occupies a considerable part of the microspore volume (Fig. 1b).

Fig. 1
figure 1

Early post-meiotic tetrads of microspores in Polemonium caeruleum. a A planar tetrad with four microspores, surrounded by callose (Ca). Adjacent tetrads are still connected with each other (arrows) via cytomictic channels. b A microspore of a tetrad. Large nucleus (N) with nucleolus, plastids (P), vacuoles (V), cisternae of ER, and dictyosomes (D) are seen in the cytoplasm. Bars at 1 μm

Full size image

When first signs of the microspore surface coat—the glycocalyx—appear in the periplasmic space between the callose layer and the plasma membrane, the latter has an even profile (Fig. 2a). Its contour becomes uneven sometime later (Fig. 2b) and then takes a sinusoidal profile (Fig. 2c). In parallel, the glycocalyx layer becomes gradually thicker (Fig. 2b, c). Some plastids are in tight contact with the plasma membrane, others have a cup-like form (Fig. 2c). Then, the structure of the glycocalyx changes: Spherical units with diameter appr. 16 nm appear near the plasma membrane (Fig. 2d and magnified fragment as inset, arrows) which in the outer portion of the glycocalyx are arranged into rod-like units—columns of spherical units (Fig. 2d, arrowheads).

Fig. 2
figure 2

Young tetrad stage in Polemonium caeruleum. a A thin layer of the glycocalyx (G) appears on the plasma membrane (PM). Plenty of ribosomes (R) are seen in the cytoplasm; ER cisterna is parallel to the plasma membrane. b Some later step. The plasma membrane is slightly wavy, discontinuous glycocalyx becomes thicker (arrowheads). c The plasma membrane acquires sinusoidal outline. The glycocalyx appears more distinct. Cisternae of RER persist their orientation. d and inset Tangential section reveals tightly packed spherical units (arrows) of the glycocalyx, which are rearranged into rod-like structures on the surface (arrowheads). For details at higher magnification, see inset. CP cup-like plastid, Mi mitochondrion, P plastid. Bars in a–d at 0.25 μm. Inset in d at 0.1 μm

Full size image

Middle tetrad stage

Initial columellae appear on the plasma membrane (Fig. 3a, arrowheads). The substructure of this procolumella, pointed by the left arrowhead, is shown in the inset: a tuft of rod-like units, each unit consists of columns of spherical subunits of appr. 16 nm. Many rough endoplasmic reticulum (RER) cisternae are seen in the microspore cytoplasm (Fig. 3a, b, arrows). Inside the plasma membrane invaginations, the glycocalyx appears as a network with roundish meshes (Fig. 3b). Slightly later, the procolumellae acquire more complex substructure (Fig. 3c): They appear as a tuft of rod-like units with diameter appr. 60 nm, tied by another subunit like a sheaf (left arrowhead and left inset, tiny white arrowheads), or have a wigwam-like form (the right arrowhead and the right inset). It is well discernible in Fig. 3c and insets that rod-like units consist of columns of spherical units. RER cisternae are often outstretched to the plasma membrane under the developing columellae. A very thin protectum (Fig. 3c, arrows) is a newcomer at this stage.

Fig. 3
figure 3

Middle tetrad stage in Polemonium caeruleum. a and inset The initial procolumellae appear in the glycocalyx (G) on pinnacles of the plasma membrane (arrowheads). The procolumella in inset reveals the substructure as several columns of spherical units (white arrowhead). RER cisternae are parallel to the plasma membrane (arrows). b Deep invaginations of the plasma membrane appear, filled with the glycocalyx (G). Note RER cisternae, outstretched to the plasma membrane (arrows). c Well-discernible procolumellae, consisting of rod-like units (arrowheads; see them also in insets), surrounded by another subunit (tiny white arrowheads), support a thin initial protectum (arrows). Ca callose, MC microspore cytoplasm. Bars in a–c at 0.25 μm. Insets in a at 0.1 μm. Both insets in c at 0.1 μm

Full size image

Late tetrad stage

The tetrahedral tetrad is shown in Fig. 4a. The cytoplasm of two adjacent tetrads is full of organelles (Fig. 4b), and the developing sporoderm is noticeable (Fig. 4b, arrowhead). Higher magnification reveals the primexine with a well discernible tectum, its thickness is appr. 0.07 μ (Fig. 4c, d, arrowheads) and short columellae. Many dictyosomes and plastids (some are cup-like) are seen in the microspore cytoplasm (Fig. 4c). RER cisternae are in direct contact with the nuclear envelope (Fig. 4e, arrows).

Fig. 4
figure 4

Late tetrad stage in Polemonium caeruleum. a A survey of a tetrahedral tetrad. b Two adjacent tetrads, their cytoplasm contains many small vacuoles (V), lipid globules (LG), and cup-like plastids (CP). c The primexine is well discernible with procolumellae and protectum (arrowheads). d Proximal sides of the adjacent tetrad microspores. The protectum (arrowheads) is well developed. e A border of the microspore nucleus. Note contacts (arrows) of ER cisternae with the nuclear envelope (NE). AL anther loculus, Ca callose, N nucleus. Bars at 1 μm

Full size image

The contacts of RER cisternae and plastids with the plasma membrane attract special attention (Fig. 5a–d). Many images of RER cisternae, extended to and contacting with the plasma membrane, are observed (Fig. 5a, c, d, arrows); some cisternae open directly to the periplasmic space (Fig. 5b, arrowhead). Plastids are often seen in the close vicinity to the plasma membrane (Fig. 5e–g). Plastids produce clusters of tiny osmiophilic vesicles (Fig. 5e–g, arrowheads) which evidently are directed to the plasma membrane.

Fig. 5
figure 5

Late tetrad stage in Polemonium caeruleum. a Contacts of ER cisternae (arrows) with the plasma membrane. b A fragment of ER cisternae, directly contacting the plasma membrane and opening to the periplasmic space (arrowhead). c, d Cisternae of ER in contact with the plasma membrane, some in sites of developing procolumellae. Autolytic vacuoles (AV) (c) and endocytotic channel (EC in d). e–g Plastids (P) in close vicinity (arrowheads) to the plasma membrane. Ca callose, G glycocalyx, GV Golgi vesicles, MC microspore cytoplasm, PEx primexine, PG plastoglobule, PT protectum. Bars at 0.25 μm

Full size image

The aperture sites lack glycocalyx, and correspondingly, primexine is not developed (Fig. 6a, b). The cytoplasm of the tapetal cells is overloaded with RER cisternae (Fig. 6c).

Fig. 6
figure 6

Late tetrad stage in Polemonium caeruleum. Aperture sites (a, b) and tapetum (c). a, b At the sites of the future apertures the layer of the glycocalyx is scanty (a) or absent (b) and primexine is absent. c The cytoplasm of the tapetal cells (Ta) is full of RER cisternae. AL anther loculus, Ca callose, CP cup-like plastid, G glycocalyx, MC microspore cytoplasm, Mi mitochondrion, Pex primexine. Bars at 0.25 μm

Full size image

Stage of disintegrating tetrad

The callose envelope appears discontinuous: It is partly dissolved (Fig. 7a) and partly remains in remnants (Fig. 7b). Because the tectum in Polemonium caeruleum is discontinuous and represented by short striae, oriented in different directions, it appears on sections differently, depending on the plane of section: as widenings of tops of columellae (Fig. 7a, b) or as rather long fragments of typical tectum (Fig. 7c). A new layer, which is a forerunner of the first endexine lamella (Fig. 7a, arrows), occurs inward to the glycocalyx (arrowheads). Double-membrane organelles, containing needle-like precipitations, attract attention (Fig. 7a, c, stars). Similar crystals were observed in analogous organelles of Passiflora racemosa at young tetrad stages (Fig. 7a, corner insets). Large lipoid globules are seen beyond the plasma membrane, inside its invaginations (Fig. 7a). Large plastids with dark stroma and osmiophilic inclusions are plentiful in the microspore cytoplasm (Fig. 7a, b).

Fig. 7
figure 7

Stage of disintegrating tetrad (a, b) and young free microspore stage (c) in Polemonium caeruleum. a Initial discontinuous endexine layer (arrows) appears between the dark contrasted plasma membrane (PM) and the glycocalyx (arrowheads). Large lipid globule (L) in the periplasmic space. Some double-membrane organelles (stars) contain osmiophilic needle-like crystals, the left one is in contact with the plasma membrane. Similar crystals in plastids of Passiflora racemosa at young tetrad stages (corner insets in a). b and inset The initial fragments of the endexine as tripartite layer with central white line (arrows). c The ectexine consists of a thick tectum and thin columellae with spiral-like shield (tiny arrowheads). The endexine is represented by primordial white-lined lamella (arrowheads). Note dark-contrasted lipid droplets on the plasma membrane, in the peripheral cytoplasm (white asterisks) and on the plastid surface (P). Double membrane organelles with needle-like crystals (stars). CaR callose remnants, D dictyosome, G glycocalyx, MC microspore cytoplasm, N nucleus. Bars in a–c at 0.25 μm. Insets in a, b at 0.1 μm

Full size image

Slightly later well-discernible fragments of the tripartite endexine lamella with typical central white line occur (Fig. 7b and inset, arrows).

Young free microspore stage

The columellae become higher and the tectum more prominent: Its thickness has increased up to 0.12 μ (Fig. 7c). Note that young columellae still preserve their substructure: Careful observation reveals a kind of spiral shield around columellae, initiated at the middle tetrad stage (Fig. 3c). The white-lined primordial endexine lamella appears continuous (arrowheads). The plasma membrane is overloaded with lipid globules (Fig. 7c, asterisks); such globules are seen also in the cortical cytoplasm and as inclusions in plastids. The participation of plastids in synthesis and release of lipid substance beyond the plasma membrane seems evident (Fig. 8a–d). Plastids, situated in close vicinity to the plasma membrane, pinch off lipid globules (Fig. 8a, asterisks) which are also observed on the plasma membrane. Other plastids are seen in direct contact with the plasma membrane (Fig. 8b–d). Note that double-membrane organelles with needle-like crystals—most probably another generation of plastids or mitochondria—persist in the cortical cytoplasm (Fig. 8a–d, stars). Similar organelles were found in the microspore cortical cytoplasm of Chamaedorea microspadix (Fig. 8e, star), where the plasma membrane was also loaded with lipids.

Fig. 8
figure 8

Young free microspore stage in Polemonium caeruleum (a–d) and late tetrad microspore in Chamaedorea microspadix (e). a Plastids (P) evidently synthesize some lipid substance, pinching them off as globules (asterisks) and governing them to the plasma membrane, the latter is heavily loaded with lipid substance. Note double-membrane organelles with needle-like crystals (stars). b–d Close contacts of plastids with the plasma membrane which looks unusually thick because of heavy accumulation of lipids (asterisks). Double-membrane organelles (stars) with needle-like crystals accompany plastids, some of them are in contact with the plasma membrane. The endexine in a–d shown by arrowheads. e Similar double-membrane organelles (star) with needle-like crystals in Chamaedorea microspadix late tetrad microspores (Fig. 6c from Gabarayeva and Grigorjeva 2010, Grana 49: 91–114). C columella, GR glycocalyx remnants, MC microspore cytoplasm, T tectum. Bars at 0.5 μm

Full size image

Middle free microspore stage

The endexine is considerably thicker: It consists of several lamellae, visible by their central white lines (Fig. 9a, arrowhead). Dark-contrasted particles appear on the tectum surface (Fig. 9a, white arrows). A slightly oblique section shows that such particles cover the whole tectum surface (Fig. 9b, black arrows). Many plastids, some of them are in contact with RER cisternae (Fig. 9, black arrowhead), persist in the cortical cytoplasm, sliding probably along the bundles of microtubules towards the plasma membrane (Fig. 9b, white arrowheads). Golgi stacks are very active, and many Golgi vesicles are accumulated under the plasma membrane.

Fig. 9
figure 9

Middle free microspore stages in Polemonium caeruleum. a The endexine is more developed and contains several lamellae with white lines (arrowhead). Small roundish osmiophilic particles—so-called sporopollenin-acceptor particles (SAPs)—appear on the surface of tectal elements (white arrows). A bundle of microtubules (black arrow) contacts the plasma membrane. b An oblique section through the microspore surface reveals the tectum surface with SAPs (arrows) and the surface of the endexine lamella (End). RER cisterna is in contact with plastid (black arrowhead). Note microtubules (white arrowheads), some are connected with plastids. C columella, D dictyosome, GR glycocalyx remnants, GV Golgi vesicles, MC microspore cytoplasm, PM plasma membrane, T tectum. Bars at 0.25 μm

Full size image

In the aperture sites, the endexine lamellae are separated from each other and fragmented (Fig. 10a, arrowheads). In the thick zone of the aperture, glycocalyx clusters of dark-contrasted particles (Fig. 10a, arrows), similar to those on the tectum (Fig. 10b, arrows), appear. The tectum is as thick as 0.17 μ. Similar tectal particles in developing microspores of Borago officinalis are shown in Fig. 10c (arrows).

Fig. 10
figure 10

Middle free microspore stages in Polemonium caeruleum. a An aperture site (A) of a microspore is filled with the glycocalyx (G). The endexine lamellae (End) divergent and are partly fragmented (arrowheads). Clusters of SAPs appear in the outer part of the glycocalyx (arrows). b The ectexine is represented by tectum and columellae. Note SAPs on the tectum surface (black arrows). The endexine is shown by white arrows. c similar clusters of SAPs on tops of columellae in Borago officinalis free microspore (arrows). d Tapetum is fragmented, its remnants (TaR) include plastids (P) and many lipid globules: large (LG) and small dark-contrasted (arrows), the latter on their way to developing exine (Ex). AL anther loculus, C columella, G glycocalyx, MC microspore cytoplasm, PM plasma membrane, T tectum. Bars in a–c at 0.25 μm and in d at 0.5 μm

Full size image

The tapetum has lost its integrity and appears as fragments, mainly as plastids and lipid globules of different size and osmiophility. Small high-osmiophilic droplets are evidently on their way to the developing exine (Fig. 10d, arrows).

Late free microspore stage

Two new layers appear at the beginning of this stage: a dark-contrasted, sponge-like layer on the surface of the endexine (Fig. 11a–d), and the channeled intine at aperture sites (Fig. 11a, b). Later on, this sponge-like layer is transformed into a foot layer with thickness of 0.13 μ (Fig. 11e, f). The tectum thickness has reached 0.35 μ, and the maximal thickness of the wavy endexine is 0.15 μ. Semi-mature exine of total thickness about 1 μ consists of ectexine and a comparatively thin endexine (Fig. 12a–c). The ectexine (Fig. 12a, double-headed arrow) includes a discontinuous tectum of appr. 0.35 μ in thickness, supported by columellae, and a foot layer of the same thickness. The whole ectexine is penetrated by microchannels (Fig. 12a–c). The lamellated endexine is discernible by its white lines (Fig. 12a–c) and compressed in result of the increasing of the cytoplasmic volume: Its thickness has decreased from 0.15 μ (Fig. 11e) to 0.08 μ. In the aperture sites (Fig. 12b), the endexine is seen in fragments. Spinules appear on the tectum (Fig. 12b, arrowheads).

Fig. 11
figure 11

Late free microspore stage in Polemonium caeruleum. a, b Two new layers—osmiophilic sponge-like layer (SL) and channeled intine (Int) at aperture sites appear. c, d Interaperture regions with sponge layer. e, f Later developmental step. Rather thick foot layer (FL), penetrated by microchannels, appears on the base of sponge layer. The lamellated endexine is discernible by their white lines (restricted by white arrows). SAPs on the tectum surface (black arrows). A aperture site, C columella, End endexine, MC microspore cytoplasm, P plastid, PM plasma membrane, T tectum. Bars at 0.25 μm

Full size image
Fig. 12
figure 12

Late free microspore stage in Polemonium caeruleum. a An interaperture zone of the exine. The ectexine (double-headed arrows) consists of a very thick tectum and foot layer, penetrated by microchannels, and of thin columellae. The endexine is comparatively thin and consists of several lamellae, discernible by their white lines. b A fragment of the exine with an aperture site (A) in the middle. Note that the foot layer and the endexine are completely absent in the aperture. Spinules occur on the tectum on the base of SAPs. c Note that the foot layer, developed on the base of sponge layer, is ultimately very thick. White-lined endexine shown by arrowheads. Columellae are thin (arrows). End endexine, FL foot layer, MC microspore cytoplasm, T tectum. Bars at 0.25 μm

Full size image

Mature pollen grains contain two sperm nuclei, and the cytoplasm of vegetative cell is overloaded with starch grains and lipid globules (Fig. 13a). The endexine is not evident. A SEM image of a pollen grain shows multiple apertures (Fig. 13b, arrowheads). Higher magnification shows tiny spinules on the tectal striae (Fig. 13c, arrows).

Fig. 13
figure 13

Mature pollen grains in Polemonium caeruleum. a Two sperm nuclei (SN) in a pollen grain after the completion of mitosis. The cytoplasm of the vegetative cell (VC) with starch grains (SG) and lipid globules (LG). Note that endexine is indiscernible in mature pollen grain. b SEM image of a pollen grain with irregular striato-reticulate sculpture (rugulate) and multiple apertures (arrowheads). c Pollen grain surface. Note tiny spinules on the tectum (arrows) which were developed on the base of SAPs. Apertures (asterisks). A aperture, Ex exine, Int intine. Bars in a at 2 μm, in b at 10 μm, and in c at 1 μm

Full size image

Discussion

Sequence of exine construction

The main steps of sporoderm ontogeny in Polemonium caeruleum are as follows: (1) spherical units of the glycocalyx appear on the plasma membrane (Fig. 14a); (2) spherical units gradually transform into columns on the surface of the glycocalyx (Fig. 14a), then to tufts or wigwam-like groups of rod-like units at sites of the future columellae (Fig. 14b). Simultaneously, the initial tectum appears; (3) groups of these rod-like (cylindrical) units start to extend, with a corresponding growth of columellae in height. The tectum increases in thickness (Fig. 14c); (4) the first endexine lamella appears simultaneously with massive overloading of the plasma membrane with lipids (Fig. 14d) and with multiple contacts of lipid-producing microspores’ plastids with the plasma membrane (Fig. 8); (5) accumulation of SP monomers in the periplasmic space, the multiplication of the endexine lamellae, and the appearance of dark-contrasted particles (SAPs) on the tectum (Fig. 14e); (6) further SP monomers accumulation in the periplasmic space, the appearance of a sponge-like layer and of the channeled intine in aperture sites (Fig. 14f); and (7) the appearance of the foot layer on the base of sponge-like layer and of spinules on the tectum on the base of SAPs (Fig. 14g); massive SP accumulation completes the maturation of the exine.

Fig. 14
figure 14

Semi-scheme showing the main developmental stages in Polemonium caeruleum (left column) and micellar interpretation of the processes (right column). a-c Tetrad stages. a Fragment of Fig. 2d. Spherical units (arrows) of the glycocalyx are accumulated on the plasma membrane at early tetrad stage: These units are the first micellar mesophase (a′) of the self-assembling sequence—spherical micelles. A transition to the next mesophase—columns of spherical micelles (a″)—is seen on the glycocalyx surface (arrowheads). b Fragment of Fig. 3c. Initial columellae are arranged from several columns of spherical micelles (right arrowhead), which self-arrange into cylindrical micelles, collected into a tuft (left arrowhead in b, b′). Initial tectum layer (arrow). c Fragment of Fig. 4d. Tightly packed cylindrical micelles—the next, hexagonal mesophase—form the layer of the glycocalyx, where procolumellae stand out (arrows) as tufts of cylindrical micelles (c′). Sporopollenin precursors are accumulated on the primexine. d Fragment of Fig. 7c. At the foot of columellae, the first endexine lamella with central white line appears (arrow), based on the next mesophase—laminate micelles (d′) with typical gap between bilayers. e Fragment of Fig. 9b. The endexine consists of several white-lined lamellae (arrow), based on laminate micelles with their typical gaps (e′). SAPs appear on the tectum surface: the sites of future spinules, accumulating sporopollenin (SP). f Fragment of Fig. 11a. New sponge-like layer (asterisk) is the base of the foot layer. It represents sponge mesophase (f′)—a bicontinual structure with unequal subvolumes of immiscible liquids. g Semi-mature ectexine includes a thick discontinuous tectum, columellae, and very thick foot layer (asterisk), interrupted at aperture sites. The foot layer appears on the base of sponge mesophase (f, asterisk). Lamellate endexine is seen by their central white lines. Spinules (arrowheads) on the tectum appear on the base of SAPs (g′). Microchannels pierce the whole ectexine. They appear in semi-liquid ectexine on the base of bicontinuous double-labyrinth structure, where oil and water are mixed, stabilized by surfactants at the interfaces (g″ a fragment from P. Ball, A soft and sticky world. 1994)

Full size image

Exine construction, regarded as a sequence of micellar self-assembly

The supposition that the microspore glycocalyx, this framework for the future exine appearing in the tetrad period, is a self-assembling colloidal system (Gabarayeva 1990, 1993) was based on histochemical data showing that the microspore glycocalyx consisted mainly of glycoproteins, with the addition of lipopolysaccharides (Rowley 1971, 1973, 1975; Pettitt and Jermy 1974; Rowley and Dahl 1977; Pettitt 1979). Many glycoproteins and lipopolysaccharides are known to be surface-active substances (surfactants): Their molecules are diphilic. This means that at some definite concentration (so-called critical micelle concentration—Fridrichsberg 1995), these molecules are capable of self-association, leading to the appearance of micelles.

Micelles, these supra-molecular aggregates, appear in colloidal solutions of surfactants in definite sequence, as their concentration increases. The sequence of the main mesophases (pseudophases) is as follows: spherical micelles, transitive “columns of coins,” cylindrical micelles, and layer of close-packed cylindrical micelles (hexagonal mesophase). Further increase in the concentration and/or addition of lipid substances leads to the appearance of laminate micelles, formed by parallel packing of bimolecular layers, separated by a stratum of water. The latter are seen with TEM as typical “central white lines” of endexine lamellae.

The sequence of sporoderm developmental events represents, in essence, the unfolding sequence of self-assembling micellar mesophases, initiated by genomically given physico-chemical parameters and induced by surfactant glycoproteins at increasing concentrations.

At the right vertical column of Fig. 14, a sequence of micellar mesophases is shown which corresponds to the main developmental events in the course of exine ontogeny in Polemonium caeruleum. If we trace this sequence from a′ to e′, it is clear that the sequence includes the main set of micellar mesophases. If one now regards the sequence of the main developmental steps at the left vertical column and compares it with the corresponding mesophases on the right, the conformity between structural units in the micrographs and schemes appears evident. The first glycocalyx units, spherical ones (Fig. 14a, arrows), observed at the early tetrad stage, are most probably spherical micelles (Fig. 14a′). The latter, as the concentration of glycoproteins, delivered by Golgi vesicles into the periplasmic space increases, re-arrange first into transitional mesophase—columns of spherical micelles (Fig. 14a, arrowheads and a″), then to cylindrical micelles (Fig. 14b, arrowheads and b′). Simultaneously, a thin protectum appears (Fig. 14b, arrow). Singular cylindrical micelles grow in number and self-assemble into tufts of parallel, close-packed ones, forming the next mesophase—hexagonal (Fig. 14c′): These are the base for future columellae (Fig. 14c, arrows). Meanwhile, SP precursors and monomers accumulate on the protectum.

When the tetrads disintegrate, tufts of cylindrical micelles grow up in height: Columellae appear taller (Fig. 14d). The concentration of lipophilic SP precursors in the periplasmic space increases, hence the next micellar mesophase—laminate one, with typical gaps filled with supporting liquid (Fig. 14d′) and seen with TEM as white lines. The first lamella of the endexine is formed on the base of such a laminate micelle after SP accumulation (Fig. 14d, arrow). Then, the number of laminate micelles (Fig. 14e′) and—correspondingly—of the endexine lamellae increases (Fig. 14e, arrow).

Then, a thick sponge-like layer appears (Fig. 14f, asterisk)—the so-called sponge mesophase (Fig. 14f′). This is most probably a highly concentrated colloid emulsion which is formed of two immiscible liquids (of hydrophobic SP precursors and water-based medium). Sponge mesophases are also regarded as bicontinuous structures with uneven subvolumes. This sponge-like layer is the base of the belated foot layer (Fig. 14g, asterisk)—an unusual case when the endexine appears earlier than the foot layer.

The last ontogenetic step of the exine in Polemonium is connected with the appearance of microchannels, piercing the foot layer, columellae, and tectum (Figs. 14g and 12 for higher magnification). How could it be that microchannels appear inside a solid ectexine? The fact is that semi-mature microspore exines before anther dihescence remain semi-liquid (Rowley and Rowley 1998) and ultimate SP polymerization proceeds after the contact of pollen grains with air. When an additional amount of SP hydrophobic precursors are added from the tapetum, the balance in semi-liquid exine mixture changes, and a bicontinuous double-labyrinth structure may occur (Fig. 14g″). The oil and water phases may form a continuous interpenetrating network (Ball 1994).

Spinules (Fig. 14g, arrowheads) on the tectum appear on the base of SAPs after SP accumulation (g′).

Comparison with ontogenetic sequences in other species

Ours and other authors’ ontogenetic studies of a number of species have shown that in spite of great differences in the ultimate exine structure and sculpture of mature pollen grains, there is much in common between their developmental sequences. A surprising uniformity is observed in the chain of events in the course of exine development, with some variations. The reason for this, in our opinion, is that self-assembly interferes in genomically controlled process and is an intrinsic phenomenon which assists the exine ontogeny. We have studied the exine development in about 30 species from remote taxa (ferns, cycads, Gnetopsida, gymnosperms, basal and advanced angiosperms—e.g. Gabarayeva 1991, 1995, 1996, 2014; Zavada and Gabarayeva 1991; Gabarayeva and Rowley 1994; Gabarayeva and Grigorjeva 2002, 2004, 2010, 2011, 2012, 2014, 2016; Gabarayeva et al. 1998, 2003b, 2009a, b, 2010a, b, 2011a, b, 2013a, b, 2014, 2016; Grigorjeva and Gabarayeva 2015), and in all of them, the constructive units (or “building blocks”) are the same and appear in the same sequence. All these building blocks fit well to the sequence of self-assembling micellar mesophases. In some species, reiteration of the sequence, partial or total, is observed—another feature of the self-assembling systems. Some other, more complex self-assembling structures as nematic liquid-crystals and bicontinuous phases also occur.

Rowley, who examined substructure of developing exines in tens of species during tens of years, has shown (1990) that the universal unit of exines is rod-like and called it a tuft, because this unit consisted of several rod-like subunits, twined round by another subunits like a sheaf; Rowley called them correspondingly core subunits and binder subunit. We suggested (Gabarayeva and Hemsley 2006; Hemsley and Gabarayeva 2007; Gabarayeva et al. 2016) that the tuft of Rowley corresponds most probably to a group (a tuft) of cylindrical micelles, braided with another one or by a filament (long cylindrical micelle). Images in Figs. 3c and inset and 7c (follow tiny arrowheads) confirm Rowley’s ideas.

Laminate micelles with their gaps between bilayers are evidently not only the base for all species with lamellate endexines, but they occur also in orbicule-like formations in Magnolia sieboldii (Gabarayeva and Grigorjeva 2012), as tripartite lamellae lying free in sporangium of Alsophila cetosa (Gabarayeva et al. 2011b), in anther loculus in Plantago major (Gabarayeva et al. 2017), in primexine (Dickinson 1976), and as radial lines in Hedycarya arborea exine (Sampson 1977, 2000). Earlier Scott (1994) suggested self-assembly for tripartite lamellae.

Sponge-like layers were also observed in Passiflora racemosa during the formation of the second endexine layer (Gabarayeva et al. 2013a, b, Fig. 4). In both Polemonium and Passiflora, SP gradually impregnates sponge-like layers, turning their friable structure to an unbroken one.

The concept of SP-acceptor particles (SAPs) was introduced by Rowley and coauthors (Skvarla and Rowley 1987; Rowley and Claugher 1991). They showed that in many species, early SP accumulation in the middle tetrad stage was connected with some osmiophilic particles (SAPs). Particles of similar size and electron density were also observed during exine formation in Asimina (Waha 1987), Nuphar (Rowley et al. 1995), Borago (Gabarayeva et al. 1998; Rowley et al. 1999), Stangeria (Gabarayeva and Grigorjeva 2002), Encephalartos (Gabarayeva and Grigorjeva 2004), and Symphytum (Gabarayeva et al. 2011b). SAPs probably include enzyme molecules for SP polymerization (Gabarayeva and Hemsley 2006). The proteinaceous nature of these particles was suggested based on results of oxidative experiments on a variety of species (Rowley et al. 2001; Gabarayeva et al. 2003a, b). SAPs, observed in Polemonium exine development, determine the sites of the future spinules. In Borago and Symphytum, they are sites of gemmae appearance. Receptive sites of the primexine (a boundary layer) were also described in Compositae and other species using oxidation by potassium permanganate (Blackmore and Barnes 1987; Blackmore and Claugher 1987). SAPs determine SP accumulation at specific sites. It is probable that in the Arabidopsis male-sterile mutant efd (Hu et al. 2014), where such “specific sites” of SP accumulation are disturbed, SAPs are lost. Therefore, the role of the genome is not restricted to the initial stage but applies also to later stages, determining specific positions of SP accumulation.

The involvement of self-assembly processes in exine development is supported by (1) morphological-ontogenetic data and (2) in vitro experimental simulations of spore-like and exine-like patterns (Hemsley et al. 1996a, 1998, 2000; Griffiths and Hemsley 2001; Hemsley et al. 2003; Moore et al. 2009; Gabarayeva and Grigorjeva 2013, 2016). Simulations, obtained in vitro conditions of colloidal mixtures, when the influence of genome was absent, represented a number of patterns, mimicking structure of spores and separate layers of microspore exines—ectexine, endexine, sponge-like layers, spine-like, and gemma-like structures—at different development stages.

The evidence suggests that genome and self-assembly share control of exine formation.

Self-assembly as an instrument of evolution

Self-assembly is a physico-chemical phenomenon, operating in colloidal systems. Because the cytoplasm and spore/microspore periplasmic space are colloidal systems, self-assembly a priori should function there (self-assembly of microtubules, microfilaments, plasma membrane, etc.). Mandelbrot (1982) suggested that nature need not overload genetic code with detail, much is predetermined by self-assembly. The genetic code need not describe the complexity of the ultimate structure, but only the initial conditions (Regier and Hatzopoulos 1988). The role of self-organization in living nature was discussed by Kurakin (2005), in an interdisciplinary review of Benítez (2013). Lintilhac (2014) confirms that physical forces play a prominent role in development.

The intrinsic feature of self-assembly is its nonlinear regularities. Self-assembly mechanisms are governed by attractors—a limited number of possible end states (Kauffman 1993). Hemsley has shown (1998) that a sudden, spasmodic character of self-assembly may result in well-known phenomena: appearance of similar patterns in genetically remote species and sharply different patterns in genetically close species. Slight deviation in genetic control (of substance concentrations, etc.) may cause an abrupt or no change in exine development. Indeed, a switch from mainly radially oriented ectexine constructive elements (columellae, alveolae, etc.) to tangentially oriented endexine lamellae is reminiscent of a phase transition in physics when gradual changes in concentration or in other parameters produce abrupt macroscopic changes between discrete stable states (Ingber 2003).

Why has evolution saved such a “wilful” instrument in its toolbox? Hemsley and Griffiths suggested (2000) that the incorporation of self-assembly mechanisms in development is advantageous to organisms in some respects. First, such systems are energy-saving in comparison with those where for ultimate structure the encoding in genome and then decoding via enzymes are needed. The energy needed for aggregation of two colloidal particles in a colloidal system is less than the energy required for their dissociation, so such a system will tend to form aggregates. Therefore, if an organism synthesizes a colloidal dispersion (e.g. in the microspore periplasmic space), the formation of colloidal crystals would be a regular thing. Second, self-assembly systems can switch from the production of one pattern to another with only minor modification of the initial conditions. The latter is supported by simulation experiments (see above). This confirms the opinion (Hemsley 1998) that evolutionary change (in the form of speciation) could be relatively rapid and that the deviation from the primary type of development may be multiplied up many times as the consequence of nonlinear character of self-assembly.

Role of plastids and RER in exine development

It was shown (Piffanelli et al. 1998; Murphy and Vance 1999) and currently generally accepted by numerous studies on Arabidopsis and rice mutants that SP monomers are synthesized in the tapetum and are delivered into the developing microspore walls after tetrad disintegration (e.g. Ariizumi et al. 2004; Ariizumi and Toriyama 2011; Dobritsa et al. 2011; Liu and Fan 2013; Lallemand et al. 2013; Lou et al. 2014; Quilichini et al. 2014; Shi et al. 2015). It was shown that in Arabidopsis tapetal plastids, de novo synthesized fatty acids undergo consecutive modifications in ER or in plastids themselves (Dobritsa et al. 2009; Chen et al. 2011). A number of genes proteins form a sporopollenin metabolon in the ER to effectively produce lipidic and phenolic precursors for sporopollenin biosynthesis (Lallemand et al. 2013).

Though it has been shown that the synthesis of the sporopollenin and tryphine precursors (fatty acids and phenylpropanois) was controlled by the sporophytic genes expressed in tapetal cells (Li and Zhang 2010), some authors hold the opinion that the source of the SP monomer during the tetrad period can be the microspore itself (Heslop-Harrison and Dickinson 1969; Echlin 1971; Heslop-Harrison 1976; Dickinson 1976; Dickinson and Potter 1976; Audran 1981; Gabarayeva 1991, 1993, 1996; Gabarayeva and Grigorjeva 2002; Gabarayeva and Hemsley 2006; Hemsley and Gabarayeva 2007). An interesting remark appeared in the review of Wallace and coauthors (2011, p. 10): “The observation that many of the genes … are predominately expressed sporophytically is somewhat at odds with the fact that exine development mostly occurs at the surface of individual microspores after meiosis”. These authors have concluded that wall development is controlled by both the diploid tapetum and haploid microspores, and identified microspores as an important additional site of SP biosynthesis. Tapetum is the major tissue for synthesizing sporopollenin precursors, as evidenced by expression analysis (Shi et al. 2015), but additional issue could be microspores themselves, even if low in quantity. Indeed, it is strange enough that the tetrad period is persistently ignored or neglected in the bulk of studies on mutants (a rare exclusion see in Fig. 1 of Shi et al. 2015) apparently because SP accumulation in the tetrad period is scanty in comparison with that in the post-tetrad one. But, an advantage in quantity far from always is the main point in comparison with the advantage in quality. The oxidation degradation experiments of a number of authors (Southworth 1974, 1986; Rowley and Prijianto 1977; Rowley and Dahl 1977; Blackmore and Claugher 1987; Rowley and Claugher 1991; Gabarayeva et al. 2003a, b) have unambiguously shown that the so-called receptor-dependent SP, which accumulates in small amount exclusively in the tetrad period, is very resistant to severe degradation—unlike massive amounts of receptor-independent, tapetum-derived SP, accumulating in the post-tetrad period. If the process of complex exine development is imagined in architect-engineering terms, all the main work—preparation of blueprints (genome-determined concentrations of substances in the periplasmic space), house basement and installation of the scaffold (the appearance of the glycocalyx by self-assembly), marking (the appearance of SAPs, under genome control), erection of walls, and installation of overlapping (the appearance of the primexine, with all but fragile characters of the mature ectexine)—all these key operations proceed in the tetrad period, whereas dress work or decorating (massive accumulation of tapetum-originated SP on the primexine) is an auxillary, cumulative process, a matter of post-tetrad time.

Another doubt is cast upon exclusively tapetal origin of SP precursors because of the problem with penetration of fatty acids and phenylpropanoids through the thick callose envelope to the periplasmic space in the tetrad period. Recent investigations indicate that lipid transfer protein (LTP) proteins, may be responsible for the transport of sporopollenin precursors through the plasma membrane (Zhang and Li 2014), not through callose.

Our data on Polemonium show that plastids (Fig. 5e–g) and RER (Fig. 5a–d) of the microspores apparently participate in synthesis of some lipoid substances which are delivered to the microspore plasma membrane and transferred (probably by active monomolecular transport—Markham et al. 2000a, b) into the periplasmic space: Both are observed in direct contact with the plasma membrane at the late tetrad stage and slightly later (Figs. 7 and 8), and plastids are often in close contacts with RER cisternae (Fig. 9b). These tight contacts bring about heavy lipid accumulations on the plasma membrane. Lipid transfer proteins (LTPs) may be also involved in the process (Li-Beisson et al. 2010). A generation of small plastids (or mitochondria—these double-membrane organelles is difficult to specify) bears needle-like crystals, like in developing microspores in Passiflora racemosa (Fig. 7a, corner insets) and in Chamaedorea microspadix (Fig. 8a, corner inset). The explanation for this could be that these organelles synthesize fatty acids, which are known to be turned on certain conditions to lead soaps, and the latter have the form of needle-like crystals (Geyer 1973). The evidence that long-chain fatty acid synthesis is delegated to the plastids has been suggested by Roughan and Slack (1982) and confirmed later: In Arabidopsis, de novo fatty acid synthesis is localized in plastids (Li-Beisson et al. 2010; Ariizumi and Toriyama 2011).

The main groups of lipoid SP precursors and monomers are aliphatic units—fatty acids (de Leeuw et al. 2006) and phenyl-propanoids—and aromatic units, especially cinnamic acid derivations (Gubatz et al. 1986; Herminghaus et al. 1988; Gubatz and Wiermann 1992, 1993; Wiermann and Gubatz 1992; Gubatz et al. 1993; Hemsley et al. 1992a, 1993, 1996b; Wilmesmeier and Wiermann 1995, 1997; Niester-Nyveld et al. 1997; Van Bergen et al. 2004; Grienenberger et al. 2010; Legrand 2010; Quilichini et al. 2015).

It is known that enzymes of phenyl-propanoid metabolism are located on ER membranes in Arabidopsis thaliana (Burbulis and Winkel-Shirley 1999) and Nicotiana tabacum (Achnine et al. 2004). It has been also shown that plastids take part in the synthesis of different phenolics, including p-coumaric acid (Goodwin and Mercer 1983; Mousdale and Coggins 1985; Zaprometov 1993; Muravnik and Shavarda 2011). These data, together with our morphological evidence, add some weight to the supposition that plastids of microspores, in tandem with RER cisternae, synthesize some small but very important part of SP precursors/ monomers, accumulated on the primexine in the tetrad period.

Tapetum

The tapetum during the pollen wall development in Polemonium caeruleum is surprisingly not impressive: Its ultrastructural features do not change significantly from the early tetrad period up to middle free microspore stage and are shown in Fig. 6c. In free microspores, the tapetum is already disintegrated, supplying actually completed exine with additional SP precursors. No orbicules or triphyne are evident.

Conclusions

  1. 1.

    The sequence of the exine developmental events in Polemonium caeruleum, traced in detail, has shown that it fits well to the sequence of self-assembling micellar mesophases. The same phenomenon was observed in all other species under our (and other authors’) studies. These observations, together with experimental mimicking of exine in vitro in colloidal solutions (Gabarayeva and Grigorjeva 2013, 2016), provide strong evidence for the important role of self-assembly in exine development.

  2. 2.

    The evidence suggests that genome and self-assembly probably share control of exine formation.

  3. 3.

    It is highly probable that self-assembly is an intrinsic instrument of evolution. Being an energy-favorable physico-chemical phenomenon operating in colloidal systems, self-assembly plays a prominent role in living nature, particularly in sporoderm development. One of advantages, conferred by self-assembly, is a high potential of variability, inherent in non-linear systems.

  4. 4.

    The data obtained allow to conclude that microspore plastids, as well as tapetal ones, participate in the synthesis of lipid substances—probably fatty acids and phenyl-propanoids (sporopollenin monomers) for developing exine. Close contacts of two types of plastids and RER cisternae with the microspore plasma membrane suggest transfer of these substances to the periplasmic space.