What does Physcomitrella look like ?

REM image

BOTANICA ACTA 111, 1-15.


Development, Genetics and Molecular Biology of Mosses

Ralf Reski

Institut für Biologie II der Albert-Ludwigs-Universität Freiburg, Germany

"Ein verdienter Bryologe des 18. Jahrhunderts, HEDWIG, klagt, die Moose seien "omnium diutissime neglecta vegetabilia" gewesen. Es hat auch wirklich lange gedauert, bis sich Ihnen die Aufmerksamkeit der Botaniker zuwandte, und die, welche dies taten, pflegten sich in den Vorreden ihrer Werke noch besonders damit zu entschuldigen, daß man auch bei diesen unscheinbaren Lebewesen doch die Weisheit des Schöpfers deutlich nachweisen und bewundern könne." (Goebel, 1915)

"Nachdem dem Zusammenarbeiten von Entwicklungsphysiologie und Vererbung wohl mehr denn je die Zukunft unseres genetischen Arbeitens gehört, werden die Moose als Objekte die besonderen Lieblingskinder mancher Genetiker bleiben." (von Wettstein, 1932)


The first descriptions of sex chromosomes in plants, of the continuity of chromosomes during the mitotic cycle and of non-Mendelian inheritance as well as the introduction of UV-mutagenesis to genetic research are landmarks in biological sciences first achieved by scientists working on bryophytes. Haploidy of the tissue facilitates mutant isolation and many developmental moss mutants have been isolated. Early moss development is triggered by auxin, by cytokinin and by light, mainly acting via phytochrome and a blue-light receptor. Due to the simplicity of the plants, development can be pinpointed to the differentiation of a single cell and be analysed in living tissue, making mosses ideal candidates for the analysis of development in an integrated approach of cell and molecular biology. Molecular genetic techniques have been applied mainly to Physcomitrella patens (Hedw.)B.S.G., where efficient protocols for transformation of nuclear DNA have been established and several nuclear, chloroplast and mitochondrial genes have been analysed. These studies reveal that Physcomitrella may be an appropriate model to study plant development in molecular terms. Recently, it has been shown that, in this species, nuclear genes can be targeted very efficiently by homologous recombination, now opening the door to reverse genetics for plant biologists.

Key words

auxin, Ceratodon, cytokinin, developmental mutants, Funaria, gene targeting, haploid genetics, land plant evolution, photoreceptors, Physcomitrella.


From time to time, bryophytes, mosses and liverworts, gained the broad interest of the scientific community. Between these oases of interest, however, bryologists had to walk through stony deserts of disregard. Only a few made these walks, while the majority changed to different model plants, promising quicker progress to be made. Thus, statements like "Well, this is nice for mosses. But does it hold true for real plants as well?" are no rarity. Apparently, we are approaching a new oasis of interest. It needs no prophecy to predict that several groups will start to work with mosses which have never been in touch with these "lower organisms" before. It is not the aim of this review to convince the already converted with data from the last two years. Instead, I will attempt to highlight some of the major achievements that were made in the last century. We do not have to reinvent moss development again and again. Most of the data are already there. They just have to be dug out of the sand.

A Short History of Moss Research

The first report on germinating moss spores can be traced back 274 years when Stähelin (1724) described the moss capsule as an urn, isolated globules from them and obtained moss plants from these germinating "Urnkügelchen" (citation according to Agardh, 1831). Hedwig (1779; reprinted in Hedwig, 1793) documented in detail the germination process and subsequent development, but he described the spore as a seed and the outgrowth as a cotyledon. It was Agardh (1831) who consequently argued that the filamentous green plants known as "protonema" (Confervenfäden in German) are no algae but the preliminary, juvenile form of leafy moss plants. Correns (1899) found differences in plastid shape and colour as morphological characteristics of different protonemal cell types and Goebel (1915) and Haberlandt (1918) analysed and discussed in detail further morphological differences, namely the orientation (oblique or transverse) of the cross walls, and their possible physiological significance but it was Sironval (1947) who suggested the terminology used until today: chloronema, caulonema, and rhizoids.

According to Goebel (1915), three scientists were most active in their attempts to find the "hidden flowers" of mosses (=cryptogams): Dillen (1687-1747) was successful in bryophyte taxonomy but described the spore capsule as an anther (and was supported in his view by Linné), while Schmiedel (1718-1792) described it as the fruit of the moss and the spores as their seeds. He discovered antheridia in bryophytes but did not succeed in finding unfertilized archegonia. Hedwig (1736-1799) was the first to find reproductive organs in various mosses and was proud to have shown that even these lower plants possess flowers ("in universum omnia vegetabilia flore et fructa instrui, ut nulla species, etiam ultima suae classis his destituta sit."). According to this view, the "moss fruit" derived from the fertilized archegonium. It was Hofmeister´s outstanding contribution (Hofmeister, 1851), describing egg cells, embryo development and the Generationswechsel, the alternation of generations, which imprints our current view of bryophyte development.

In the first decades of this century, bryophyte genetics was at the forefront of genetical research: Allen (1917) described the first sex chromosomes in plants, Heitz (1928) demonstrated the continuity of chromosomes during the mitotic cycle, and Knapp (1936) employed X-ray mutagenesis for genetic research. Non-Mendelian inheritance was first postulated by von Wettstein (1928) who analysed Funaria hygrometrica and related mosses. Likewise, von Wettstein (1924; 1932) recognised the great potential of the haploid moss protonema to genetically dissect differentiation processes. Such analyses started with mutant induction in Physcomitrium piriforme, Funaria hygrometrica, and Physcomitrella patens (Barthelmess, 1940; Oehlkers and Bopp, 1957; Oehlkers, 1965; Engel 1968). Since then moss protonemata have been widely used in plant physiology. In the last decades efforts have concentrated upon three species from two different orders: Ceratodon purpureus (Hedw.) Brid. (Dicranales), Funaria hygrometrica Hedw. (FunArial,Helveticaes) and Physcomitrella patens (Hedw.) B.S.G. (FunArial,Helveticaes). Wherever possible, I will focus on these three in the following and will neglect others, e.g. Sphagnum. Two of them (Funaria and Physcomitrella) were most intensively studied: Bopp and coworkers concentrated on the physiology of Funaria development (Bopp, 1990; Bopp and Atzorn, 1992), Schnepf and coworkers analysed Funaria development at the ultrastructural level (Schnepf, 1982), and Cove and coworkers focussed on mutant induction and analysis in Physcomitrella (Cove, 1992). These studies led to the suggestion that the genetic dissection of Physcomitrella could make a significant contribution to understanding the evolution of plant developmental processes (Goldberg, 1988; Cove et al., 1997).

Two Generations-One Plant: Moss Development

Characteristic for mosses (and puzzling for people working with higher plants only) is their heteromorphic Generationswechsel, the alternation of two generations which are distinct from each other in terms of nuclear DNA amounts and morphology. The diploid sporophyte is photosynthetically active only in its youth and requires supply from the dominating, green, haploid gametophyte. Moss gametophytes exist in two morphologically distinct forms: The juvenile form is filamentous and grows by apical cell division whereas the adult form, which is the leafy moss plant, grows by division of three faced apical cells. As the adult form bears the sex organs (=gametangia), the female "archegonia" and the male "antheridia", it is called a "gametophore". The transition from the juvenile "primitive" gametophyte (protonema) to the adult "advanced" gametophyte (=gametophore) is a single cell event, the first occurrence of a three-face apical cell, the so-called "bud", which develops into a leafy moss plant. Unlike morphogenesis in higher plants, morphogenesis in the moss protonema can be pinpointed to one single cell event in each case (Bopp, 1968).

Protonema: The Juvenile Gametophyte


Moss spores are single cells with a thick wall consisting mainly of an inner layer (intine) and an outer one (exine). They contain several small plastids with reduced lamellar systems and lipid bodies, starch grains and sometimes protein lumps as reserve material (Knoop, 1984). In the presence of water and light, the spore swells until the exosporium bursts. Subsequently, the intine is disrupted, a filamentous germ tube is put forth (Fig. 1a) and a new wall is formed along the surface of the protoplast (Schnepf et al., 1982). Obviously, spore germination is a phytochrome-dependent process (Bauer and Mohr, 1959). In the light, photosynthesis increases rapidly although chloroplasts develop even in the dark (Valanne, 1984). Sugars and phytohormones are thought to overcome dark inhibition of germination (Knoop, 1984).


The germ tube grows out to form a germ rhizoid or a chloronema cell, depending on culture conditions. Chloronema cells are relatively short, have cross walls perpendicular to the growth axis, contain numerous round chloroplasts (Fig. 1b) and multiply by apical cell division thus establishing a direct cell lineage. Chloronemal filaments have multiple, irregular branches (Fig. 1c). They grow in darkness when a carbon source is supplied (Berthier et al., 1976) but become elongated and lack fully developed chloroplasts. This etiolation can be overcome by red light, thus phytochrome may be necessary to maintain chloronema identity (Nebel, 1968; Jenkins and Cove, 1983b). Likewise, branching of the primary chloronema is under phytochrome control (Berthier et al., 1976; Jenkins and Cove 1983b).

Several lines of evidence have proposed that the growing protonema produces morphogenetic substances which are secreted into the substrate, and act from there to regulate the development of all filaments as well as to coordinate growth and differentiation of closely neighbouring plants: 1) Cell density affects cellular differentiation in moss suspension cultures (Johri and Desai, 1973; Sharma et al., 1979; Bopp, 1980), 2) Spores germinating close together generate one large protonema, although the different plants have no direct cellular connections (Klein, 1967), 3) In continuous medium replacement Physcomitrella grows to large plants without differentiating further than to chloronema. Differentiation can be initiated by the addition of auxin and cytokinin (Cove and Ashton, 1984).

By 1963 Bopp had already proposed two leachable morphogenetic factors from Funaria. In line with this, Neuenschwander et al. (1994) suggested that protonema development may be induced via specific extracellular proteins in Physcomitrella. Recently, we quantified auxins and cytokinins from tissues as well as from culture media of Physcomitrella and confirmed previous reports (e.g. Wang et al., 1981) that isopentenyladenine-type cytokinins are the major cytokinins of mosses, whereas the major cytokinins of higher plants, the zeatin-type hormones, occur only in minor quantities. Surprisingly, the majority of the hormone pool was extracellular: More than 70% of the cytokinin and more than 90% of the auxin had accumulated in the medium (Reutter et al., 1997). From that we speculated that in the non-vascular, filamentous moss protonema, completely dependent on a humid environment, communication between different filaments of one plant and between different plants is mediated by extracellular auxins and cytokinins.


Auxins induce the next differentiation step, the transition from chloronema to caulonema (Johri and Desai, 1973; Ashton et al., 1979a). Cells produced by the apical cell become longer and sometimes thinner and contain fewer and smaller plastids. Simultaneously, the orientation of the cross walls is changed by microtubules from perpendicular to oblique to the growth axis (Fig. 1c). The region of transition is about four to five cells corresponding to a growth period of about 24 h. This new cell type is morphologically and physiologically different, e.g. in contrast to chloronema, caulonemal filaments show regular branching (Bopp, 1980; Knoop, 1984). Addition of the antiauxin PCIB (p-chlorophenoxyisobutyric acid) retransforms caulonema into chloronema: cells stop growing temporarily, have reduced cell lengths and accumulate chlorophyll. Subsequently, the first new-formed cross wall remains perpendicular (Sood and Hackenberg, 1979; Bopp, 1980). The polar auxin transport inhibitors 3,4,5-TIBA (triiodobenzoic acid) and NPA (1-N-naphthylphtalamic acid) have similar effects, whereas surprisingly 2,3,5-TIBA seems to have no effect (Cove and Ashton, 1984). Several reports have suggested that endogenous levels of auxin and cAMP (cyclic adenosine-monophosphate) determine relative proportions of chloronema and caulonema during protonemal growth (e.g. Handa and Johri, 1976).

Not only hormones, but also environmental factors regulate protonema differentiation: low light intensities as well as ammonium tartrate as a nitrogen-source favour chloronema formation and retard differentiation into caulonema (Jenkins and Cove, 1983a). Lack of phosphate or nitrate enhances formation of cells with oblique cross walls, caulonema and rhizoids (Schoene, 1906), whereas lack of calcium leads to inhibition of caulonema formation. Thus, caulonema formation appears to be a response to a deficiency that needs calcium gradients to be established (Bopp, 1981). In Ceratodon, both protonemal cell types, chloronema and caulonema, show a positive phototropic response mediated by phytochrome and intracellular Ca2+- gradients (Hartmann et al., 1983). The response is cell-autonomous and associated with reorganisation of the cytoskeleton (Hartmann and Weber, 1988; Meske et al., 1996).

Light is needed to obtain branching of caulonemal filaments. In Physcomitrella red light triggers, within 2-15 s, a transient membrane depolarisation, and 3 days later the development of side branch initial cells. Both events are Ca2+-dependent, phytochrome-mediated and effectively inhibited by cation- as well as anion-channel blockers, suggesting that membrane depolarisation is involved in phytochrome signal transduction leading to caulonema branching in Physcomitrella (Ermolayeva et al., 1996; Johannes et al., 1997). Primary side branches usually become caulonema again but when the cells of the main filament get older and more basal, they can form a second side branch arising near the base of the first one. This second one usually becomes secondary chloronema thus maintaining assimilatory capacity during protonemal growth.

After having developed a second side branch, caulonema cells of the main axis of Funaria become endopolyploid. The chloronema, however, remains haploid and varies only between 1 and 2 C, depending on the mitotic phase of the nucleus (Knoop, 1984). Recent flow cytometric analyses (Gorr and Reski, 1997) revealed that Physcomitrella protonemata grown in liquid culture under a 16h light/8h dark regime are highly synchronized and exhibit extraordinary features making them ideal candidates for cell cycle analyses: chloronema cells are arrested in G2 for most of the day with mitosis occurring mostly between 2 and 4 a.m. In contrast, young caulonema cells remain in G1, whereas older caulonema cells become polyploid. It is unknown whether these profound differences in the mitotic cycle determine or reflect the physiological differences between chloronema and caulonema. However, in mosses auxins very specifically induce a set of cellular events that lead to well defined cell differentiation.


The next step in differentiation, bud production, is the transition from growth via apical cells to growth via a three-faced apical cell. This bud is the meristem initial of the erect moss plant and thus marks the transition from the filamentous juvenile to the leafy adult gametophyte (Fig. 1d). Gorton and Eakin (1957) were the first to report that a cytokinin greatly stimulated bud formation in a moss (Tortella caespitosa). Subsequently, it has been shown that bud induction is a specific cytokinin-effect in several mosses, mainly using "potent" but artificial cytokinins (Bopp, 1963; Brandes, 1967; Hahn and Bopp, 1968). From these studies it has been suggested that only caulonema cells react by budding (Bopp, 1981) and that they may be specific target cells for cytokinin action because they have cytokinin-binding proteins which are not present in chloronema cells (Erichsen et al., 1977). However, 6-(D2 isopentenyl)adenine (IP), the main native cytokinin of mosses, induced budding on chloronemata and on caulonemata of Physcomitrella and Funaria in a specific manner. At low concentrations, chloronema cells were the targets of cytokinin action whereas at high concentrations a threefold increase in the number of buds was induced on caulonema only (Reski and Abel, 1985), revealing that the competence of one tissue type depends on hormone concentration. Artificial cytokinins may be "more potent" as they are not easily degraded in planta and, therefore, even at low concentrations induce buds at caulonema cells only. The basis of this tissue-specific cytokinin response is still enigmatic.

Application of cytokinin induces morphological and biochemical alterations within 3h (Brandes, 1967; Bopp, 1981), but if plants are released from hormone treatment within 72 h, most of the newly-induced buds revert to protonemal filaments (Brandes and Kende, 1968; Sussman and Kende, 1977), indicating that the developmental fate of cytokinin-induced buds is reversable during a lengthy period. Commonly, buds induced by application of cytokinin do not develop into leafy gametophores but become callus-like and necrotic. However, if endogenous cytokinin-levels are elevated by expression of the bacterial ipt-gene, buds do develop into leafy structures (Reutter et al., 1997). High concentrations of auxin also inhibit normal development of buds, either all cells grow out as protonema, forming a "spiky" appearance or in somewhat older buds, short, thick, leafless stems appear (Ashton et al., 1979; Bopp, 1980). Occasionally, there have been reports on the effect of activated charcoal in the medium on distribution of cytokinin target cells (Klein and Bopp, 1971). In line with this, Hadeler et al. (1995) reported a significant effect of the gellant in differentiation of Physcomitrella. Grown on Gelrite-media, protonemata showed enhanced and premature budding, thus mimicking cytokinin application. It is conceivable that the substrate interferes with the exogenous hormone pools detected by Reutter et al. (1997) and thus influences protonema differentiation markedly.

Although cAMP had been discussed as a possible messenger in moss development, it alone cannot promote bud formation (Schneider et al., 1975). Calcium, on the other hand is an intracellular messenger in hormone-induced bud formation (Saunders and Hepler, 1982) and cell division is correlated with an inward current at the nuclear zone as shown in Funaria (Saunders, 1986). Calcium transport and ligand binding studies in Physcomitrella have implicated plasma membrane-localized calcium channels in the regulation of cellular calcium (Schumaker and Gizinski, 1993). Based on inhibitor studies, Schumaker and Gizinski (1996) suggested that G proteins may act via a membrane-delimited pathway to regulate calcium channels in the moss plasma membranes.

It has been known for 100 years that light is required for gametophore production (Klebs, 1893). Although some reports claimed, that exogenous cytokinin can bypass the light requirement for bud induction, it is now clear that cytokinin and light must interact in bud production (Ashton et al., 1979b; Simon and Naef, 1981; Hartmann and Jenkins, 1984). Most probably, light perception in this case is by phytochrome in a variety of mosses. An alternative suggestion was that an inhibitor of bud induction is produced in blue light (Jahn, 1964). Some authors found that buds are induced by cytokinins in blue light (Simon and Naef, 1981).


Rhizoids are similar to caulonema cells at the morphological level. They contain only few chloroplasts and possess oblique cross walls. In contrast to caulonema cells, rhizoids usually are unbranched. A further distinctive feature is their place of origin: rhizoids either occur as a single germ rhizoid cell directly out of the spore (Fig. 1b) or are more regularly formed at the basis of buds and subsequently leafy gametophores (Fig. 1d; Schoene, 1906). They show positive geotropism.

Tmema Cells

Ageing Funaria protonemata can fragment by the formation of short-lived tmema cells which are built in the proximal part of an intercalar chloronema cell by a highly unequal cell division (Correns, 1899; Bopp et al., 1991). While preprophase bands (PPBs) do not occur during tip cell division, branching or intercalary regeneration division, they occur during tmema formation as well as in the adult gametophyte (Sawidis et al., 1991).

Gametophore: The adult gametophyte

Stems and leaves

The protonemal bud is the meristem initial of the leafy shoot. As the daughter cells of its apical cell continue to divide more frequently than the central cell itself, they form a mound around it, from which the leaf primordia are derived (Figs. 1d,e). Each of the primary daughter cells gives rise to one primordium, so that the temporal and spatial leaf pattern reflects the regular division of the apical cell (Janzen, 1929). The mature stem consists of different tissues: hydroids in the center, a cortex, and the epidermal cell layer. The leaves are usually one cell thick and mostly, but not always, with a costa more than one cell thick. Except in the Polytrichales, the costa usually has no connection to stem hydroids but ends blindly in the cortex (Frey, 1981; Nyman and Cutter, 1981). The detailed organisation of the gametophore is more complicated than described here (Janzen, 1929), but little work has been carried out on its developmental biology.

Cells which have not developed into leaves form resting meristems which may consist of an apical cell and some smaller derivatives surrounding it, or may have fallen dormant at a later stage of development, e.g., as already complete primordia. These meristems can give rise to side branches when dormancy is abolished (Nyman and Cutter, 1981). Rhizoids and protonemal filaments also escape from normal tissues, especially from the "epidermis". When the apex of the leafy gametophore is removed, lateral buds grow out, forming a new plant tip and side branches. In decapitated gametophores of Splachnum the inhibitory influence of the apex can be replaced by an agar block containing auxin. This suppression of lateral bud development is antagonised by the simultaneous application of cytokinin (von Maltzahn, 1959). Bud development can also be released from apical dominance in intact plants when internal auxin transport is blocked by a ring of TIBA (Nyman and Cutter, 1981). Thus, maintenance and release of apical dominance in mosses is similar to that in higher plants.

Additionally, the stem of a gametophore suppresses regeneration in the leaves. This "central dominance" is not mediated by the apex but needs transport between the stem and the leaf (von Maltzahn, 1961). Leaves isolated from different positions of a stem have different regeneration capacities. Normally, young leaves from near the apex regenerate more easily than older ones from the stem base but in some species a second maximum of regeneration is found in basal-derived leaves. However, basal leaves regenerate from the lamina and apical ones from the midrib only. Moreover, single, isolated leaves also exhibit a regenerative gradient in that either the basal or the apical part is more efficient in producing new filaments (Bopp, 1955).

Phytochrome mediates positive phototropism as well as deetiolation in gametophores of Physcomitrella (Cove et al., 1978) and affects leaf length and width in a number of moss species.

Sex organs

The adult gametophyte, the leafy gametophore, bears the sex organs. Male gametes (= spermatozoids) are produced within antheridia (Figs. 2a-c) and female gametes (= oogonia or egg cells) are produced within archegonia (Figs. 2d,e). As already described by Hofmeister (1851) spermatozoids affect fertilization by swimming through a surface water film and down the neck of the archegonia which normally contains one egg cell (Fig. 2d far right). The zygote develops into an embryo (Figs. 2f-i) which grows out to the diploid sporophyte (Figs. 2j,k). Sex organ differentiation and embryo development in Funaria have been analysed in detail by Janzen (1929).

The induction of antheridia and archegonia depends on inherited characteristics of the species, and on external conditions; although light is needed, most mosses are day length neutral and so far only one inductive stimulus has been reproducibly proven effective in sex organ induction: low temperature. The internal conditions necessary for the induction of sex organs are completely unknown (Bopp and Bhatla, 1990).

There are dioecius mosses which have male and female individuals as well as monoecius species in which one plant bears both types of sex organs. Correns (1899) analysed 915 species of European mosses and found 54.6% to be dioecius and the remainder to be monoecius in its broadest sense. As isolated antheridia as well as isolated archegonia can be regenerated to normal monoecius Funaria (Correns, 1920), the potency to determine both kinds of sex organs must be present in every cell. From further analyses, von Wettstein (1932) claimed that sex determination in monoecious mosses is environmentally controlled whereas in dioecius mosses it is genetically determined. Wild-type strains of the monoecious Funaria and Physcomitrella are normally self-fertile in culture, whereas the dioecius Ceratodon so far did not produce sporophytes in any in-vitro culture. In Physcomitrella, auxotrophic mutants requiring p-amino benzoic acid, nicotinic acid, or thiamine are self-sterile. When crossed to other auxotrophic strains, the offspring is self-fertile, but only if the mutant alleles of the genes in the two strains complement one another (Engel, 1968; Ashton and Cove, 1977). Cove and Knight (1993) suggested that this phenomenon results from partial metabolic isolation of the sporophyte from the gametophyte.

Sporophyte: The Next Generation

Sporophytes, consisting of a small stalk (=seta) that bears the spore capsule (=sporogonium) (Figs. 2j,k), typically develop from a fertilised egg inside the archegonium (Fig. 2d far right) by differentiating an apical cell with two cutting faces (Fig. 2f). This is the first obvious difference in pattern formation between gametophytic and sporophytic development. Subsequently, near a subapical meristem, the seta of the sporogonium starts to grow. The stalk zone in between later develops the spore capsule. A detailed analysis of sporophyte development can be found in Janzen (1929). The foot of the seta, the haustorium, is embedded into gametophytic tissue, thus connecting the two generations physically.

As there are no plasmodemata between gametophytic and sporophytic tissue, specialised structures of the haustorium, such as transfer cells, may have different duties in the supply of the sporophyte (Wiencke and Schulz, 1975; Browning and Gunning, 1979). The Physcomitrella sporophyte has a comparatively short seta of few millimeters, a sporogonium of about 2 mm in diameter with no specialized structures for dehiscence and produces about 5000 spores.

When the sporophyte grows through the archegonium, a cap of gametophytic tissue, the calyptra, covers the spore capsule (Figs. 2i,j). Early removal of the calyptra results in misformed setae and spore capsules in different moss species. Interestingly, the acting principle seems to be the physical restraint exerted on the tip tissue of the young sporophyte rather than an organic compound, as dead and thoroughly extracted calyptrae have the same effect as fresh ones when placed back on the tip of a young naked sporophyte (Bopp, 1957).

Mosses do not produce wound tissue and are not able to regenerate wounded organs. However, almost all cells, whether gametophytic or sporophytic, are capable of regeneration. In contrast to higher plants, small moss tissue explants, whatever their former developmental fate, regenerate by a developmental pathway similar to spore germination, with chloronema being generated first, followed by caulonema, buds and gametophores (Correns, 1899; von Wettstein, 1924). Large fragments, however, only shift back to intermediate stages such as caulonema or even buds. Therefore, Knoop (1984) suggested using the term "redifferentiation" for these processes to indicate the shift to an earlier phase in ontogeny. The term "dedifferentiation" can not be applied for mosses, as unlike in higher plants, regeneration normally does not pass through "undifferentiated" callus cells.

Thus, diploid protonemata and gametophores can be obtained by regenerating sporophyte tissue (= aposporous regeneration) revealing that neither ploidy and the process of fertilisation nor special activities of the surrounding tissue is responsible for the differences between gametophyte and sporophyte development, the so-called Heteromorphose (Correns, 1899). Diploid gametophytes, either produced aposporously or by fusion of two haploid protoplasts, are mostly fertile but development is slower and sporophytes are produced after at least 12 months. Segregation ratios gave evidence that such sporophytes are either produced from the fusion of diploid gametes, resulting in tetraploid sporophytes, or by parthenogenetic development of a diploid archegonium into a diploid sporophyte (von Wettstein, 1932).

Polarity and Protoplast Regeneration

One of the beauties of the moss system is that cell polarity and subsequent differentiation can be studied in a complete cell lineage at the single cell level (Schnepf, 1986). Such experiments have been mostly carried out with caulonema cells. The polarity of a caulonema cell is stable as long as the cell remains in contact with neighbouring ones. The cytoskeleton is organized in a polar way and controls the position of the nucleus which in turn controls the site of new side branch formations (Schmiedel and Schnepf, 1980; Doonan et al., 1988; Quader and Schnepf, 1989). When young caulonema cells are isolated even as 10-cell filaments, intercalary cell divisions occur within only 8 h (Bopp and Böhrs, 1965). These newly formed cross walls are no longer oblique but perpendicular to the growth axis, and subsequently, plastids multiply and accumulate chlorophyll. Thus, two or even more chloronema-like daughter cells are formed within the old, regenerated caulonema cell, raising the photosynthetic capacity of the tissue as a prerequisite for further development. Until the first intercalary division, cells do not grow and protein synthesis is not needed (Knoop, 1984). When a caulonema cell exceeds a certain age, it becomes endopolyploid, produces brown wall pigments and loses its redifferentiation capacity (Knoop, 1978). In intact plants, redifferentiation can be induced by treatments which reduce the growth velocity of the tip cell below a certain limit (Knoop, 1984). Decapitation experiments in branched caulonema indicate that every tip cell produces a signal, probably auxin, which is transported basipetally and there maintains the caulonema state. As long as the cells are integrated in the filament, their polar organisation is also expressed in regeneration: In filaments, the intercalary cross wall frequently is formed in the apical part of the cell, whereas a single isolated caulonema cell regenerates by symmetric division (Knoop, 1984).

This cell polarity also affects plastid division, as the dividing plastids of an apical caulonema cell are found anterior to the nucleus at least in Physcomitrium turbinatum (Jensen, 1981). So far it is unknown whether plastids are physiologically different fore and aft of the nucleus, and how the plastids at the base of the apical cell behave with respect to further replication when they are part of the second cell of the filament (Paolillo, 1984). Although moss plastids do not exhibit an elaborate differentiation programme as in higher plants, histologically different plastid types do exist during protonema development (Tewinkel and Volkmann, 1987) and may be involved in gravitropic resonses (Walker and Sack, 1990).

Spores are asymmetric according to distribution of lipid vacuoles, organelles and spore wall construction (Brown and Lemmon, 1980). In an uniform environment they germinate according to this preformed polarity pattern and the germ tube appears at that part of the exine which was thinnest and therefore has burst during swelling (Fig. 1a). Outer gradients can easily change this polarity, most effective being light via phytochrome (Bauer and Mohr, 1959). Under low light conditions most spores develop a chloronema filament from the illuminated side of the spore (Fig. 1a). Under high-light conditions some species (e.g. Funaria) germinate bipolarly by developing, before or simultaneously with the chloronema, a pale, negatively phototropic germ rhizoid from the shaded side of the spore (Fig. 1b; Heitz, 1942). Electrophysiological studies reveal that rhizoids are formed at the side of the cell where calcium entry is enhanced (positive electrode). However, subsequent growth of both rhizoids and chloronemata is directed towards the negative electrode (Chen and Jaffe, 1979; Burgess and Linstead, 1982).

Regenerating moss protoplasts behave very much like germinating spores with respect to polarity, light and calcium gradients. They can be isolated from protonemata by enzymatic degradation of the cell walls. A quite rapid regeneration has been reported for Physcomitrella (Stumm et al., 1975; Grimsley et al., 1977; Burgess and Linstead, 1981) and Funaria (Batra and Abel, 1981). Interestingly, moss protoplasts do not need any phytohormones for regeneration. Physcomitrella protoplasts become asymmetric before dividing and extending to produce a new chloronemal filament (Abel et al., 1989). During regeneration a new polar axis of development is established in the cells, triggered by relatively high intensities of red or blue light. Regeneration stops in darkness, because the division process itself is mediated by phytochrome (Jenkins and Cove, 1983a). More detailed analyses in Ceratodon revealed that two separate gradients are established to determine the axis of alignment and axis polarity independently (Cove et al., 1996).

Mutants and More: Classical Genetics of Haploids

If two haploid strains are crossed, a diploid sporophyte arises which in turn produces haploid spores. The haploid progeny of this cross segregate differently from known Mendelian genetics: when only two alleles of one gene are analysed, the F1 segregates 1:1; when mutant alleles of two unlinked genes are analysed, a 3:1 ratio of mutant to wild-type phenotypes will occur. Von Wettstein (1932) already distinguished 6 pairs of characters in Funaria and by segregation ratios provided evidence for the functional haploidy of mosses. He found that genes governing spore-size and rate of division of protonemal cells were completely linked or, alternatively, the same gene influenced both characters. A similar relation held between leaf-form (a gametophytic character) and the form of seta and capsule (sporophytic). Other characters studied were form of paraphyses and capsule-color. Self-fertilisation of a diploid gametophyte, either produced aposporously or by fusion of two haploid protoplasts, gives rise to a tetraploid sporophyte. If the gametophyte was heterozygous, the progeny of spores from such a tetraploid sporophyte show segregation ratios consistent with their being the diploid products of meioses which have occurred in tetraploid cells as have been shown in Physcomitrella. Thus, the nature of several somatic hybrids has been identified (Cove, 1983), but also chromosome stains have proven the successful karyogamy after protoplast fusion in this species (Reski et al., 1994).

Compiling 3863 reports on chromosome counts in mosses, Newton (1984) argued that species with n = 6 or 7 chromosomes are basically haploid and those with about 10 to 14 chromosomes in the gametophytic development may be basically diploid. She suggested that autopolyploidy, allopolyploidy and amphidiploidy may have contributed to the evolution of bryophytes. This view was supported by Wyatt et al. (1988) presenting biochemical data on allopolyploidy and multiple origins of Plagiomnium medium. Newton (1984) further suggested that gametophytic tissue of Physcomitrella may have 26 chromosomes and thus be tetraploid. However, all crosses so far gave evidence that segregation ratios are that of a true haploid species (Grimsley et al., 1977; Cove 1983). Chromosomes of mosses, especially of the Funariaceae are very small and hard to analyse (Newton, 1984). Consequently, there have been varying chromosome counts in several species like Physcomitrella. However, based on mitotic as well as meiotic chromosome counts it has been established that this species has n=27 chromosomes which all are 1-2 mm in size (Reski et al., 1994). A compilation of chromosome counts from about 1300 moss species, covering 18.5% of all known moss species, is found in Fritsch (1982).

Several mosses are dioecious, in that there are male and female individuals. Consequently, sex chromosomes have been reported from about 16 species (Newton, 1984). However, proof of their role in sex determination is still limited to the liverwort Sphaerocarpos donellii (Allen, 1917, Allen, 1935). In diploid dioecy Y chromosomes are unique in occurring only in the hemizygous condition whereas in haploid dioecy neither the X nor the Y chromosome is ever homozygous. This suggests that sex chromosome evolution may have been different between diploid and haploid plants. Sporophytes of dioecius species contain both parental sex chromosomes and by aposporous regeneration can give rise to diploid gametophytic tissue. Such diploid gametophytes are regularly monoecious, leading to the hypothesis that the monoecious moss species may have arisen in nature from dioecius haploid species. However, this implies that monoecius forms would have generally diploid gametophytes and tetraploid sporophytes (Allen, 1935), which at least for the monoecious Physcomitrella is not in accordance with the segregation ratios obtained so far (Cove, 1983).

The first mutants used to study the role of hormones in bryophyte development were obtained in Funaria by Hatanaka-Ernst (1966). She used X-rays to induce developmental mutants, i.e. mutants defective in caulonema formation or in bud production. As in the haploid mosses such mutants are not fertile, the term "mutant" has to be used with care. The method of choice for the analysis of such "mutants" is somatic genetics, i.e. the fusion of two protoplasts of different type and the regeneration of diploid plants. This technique has been used in Physcomitrella (Grimsley et al., 1977; Rother et al., 1994) and in Funaria (Mejia et al., 1988). These hybrids are self-fertile but like aposporously produced diploids are slow in completing their life cycles. This is partly due to the reduced number of sex organs, particularly archegonia, and partly to the more sluggish movement of diploid spermatozoids (von Wettstein, 1924; Allen, 1935). However, tetraploid embryo development has never been studied in great detail and may be abnormal too.

In most studies, the haploid spores have been the preferred material for mutagenic treatment (Engel, 1968; Ashton and Cove, 1977). However, mutagenic treatment of protoplasts was also successful in Physcomitrella (Boyd et al., 1988), in Funaria (Atzorn et al., 1989), and in Ceratodon (Lamparter et al., 1996). Morphologically abnormal mutants are self-identifying and thus easy to isolate. Auxotrophic mutants have been isolated following laborious mass-screening programmes. A number of classes of developmentally-abnormal mutants of Physcomitrella have been selected directly by growth in the presence of inhibitory concentrations of auxins or cytokinins (Ashton et al., 1979a) as well as auxin-mutants in Funaria (Atzorn et al., 1989). Recently Lamparter et al. (1996; 1997) isolated aphototropic mutants of Ceratodon which fall into two classes: one is thought to be affected in the synthesis of the chromophore itself, the other is thought to be affected in phytochrome signalling. Some developmental Physcomitrella mutants need extra cytokinin to trigger bud production (Ashton et al., 1979a) and/or chloroplast division properly (Abel et al., 1989; Kasten et al., 1997), possibly due to a mutation in cytokinin signal-transduction (Reutter et al., 1997).

Over the past thirty years a number of mutants have been isolated, auxotrophic ones as well as developmental ones. Sometimes, nomenclature of developmental mutants isolated by different laboratories has been inconsistent. It is suggested to use a coherent nomenclature for developmental moss mutants based upon mutant description. A three-letter gene abbreviation in italics should discriminate the gene from the protein. Wild-type genes should be written in capitals and mutant genes in lower case. Suggestions for the most common developmental moss mutants are given in Table 1.

Finally: Molecular Biology of Mosses

Although a wealth of physiological data have been accumulated for different moss species, there has been a serious delay in the use of new techniques of plant molecular biology and the molecular analyses have concentrated on Physcomitrella, a species with a genome size of about three times that of Arabidopsis (480 Mbp; Gorr and Reski, 1997), distributed on n=27 chromosomes (Reski et al., 1994).

The first report on the organellar DNA of a moss was a physical map of the plastid DNA from Physcomitrella (Calie and Hughes, 1987). Subsequently, efficient protocols for the isolation of nuclear, chloroplast and mitochondrial DNA from this species were established (Marienfeld et al., 1989) and the genomic sequence of a Physcomitrella cab-gene was published (Long et al., 1989). Reski et al. (1991) applied RFLP-techniques to developmental Physcomitrella mutants and presented the first transcriptional analyses of a moss. Physcomitrella can be stably transformed by PEG-mediated DNA transfer to protoplasts (Schaefer et al., 1991) and by the particle gun (Sawahel et al., 1992), but so far not by Agrobacterium-mediated DNA-transfer (Reutter, 1994). However, it is accessible to microinjection (Abel et al., 1989). Subsequently, this moss was transformed in order to study the role of ABA and osmotic stress by using promoter elements from the wheat Em gene (Knight et al., 1995), of endogenous calcium (Russell et al., 1996), to cure developmental mutants by the bacterial ipt gene (Reutter et al., 1997) and for biotechnological purposes (Reutter and Reski, 1996). Zeidler et al. (1996) demonstrated that the widely used inducible TET-promoter is functional in Physcomitrella. Thümmler et al. (1992a) reported on the transformation of Ceratodon with an oat phyA cDNA using the PEG-mediated DNA-uptake into protoplasts. These studies reveal that there are no restraints from promoter elements or codon usage when using constructs developed for the transformation of higher plants.

The first mitochondrial DNA sequence from an archegoniate (cox3) was obtained from Physcomitrella (Marienfeld et al., 1991), as well as the first chloroplast nucleotide sequences from mosses (Kasten et al., 1992). Subsequently, nucleotide data from this species accumulated (GAPDH: Martin et al., 1993; myb-related genes: Leech et al., 1993; phyB: Kolukisaoglu et al., 1993; rbcS, cab and 25 S rDNA: Reski et al., 1994) as well as amino acid data from chloroplast proteins (Kasten et al., 1997). By functional complementation of an E. coli mutant, von Schwartzenberg et al. (1997) isolated a gene so far unknown in plants (adenosine kinase) and demonstrated that the protein is involved in cytokinin metabolism. Different cDNA libraries have been established from Physcomitrella (Martin et al., 1993; Reski et al., 1994) and molecular subtraction of two libraries from differently cytokinin-treated tissues resulted in about 80 ESTs, which comprise 39 genes so far unknown for plants. Comparison of the moss nucleotide sequences for known genes like cab and ribosomal proteins to their higher plant homologues revealed homologies between 48% and 75%, regardless of whether the higher plant was a gymnosperm, a dicot or a monocot. Likewise, codon usage in Physcomitrella was similar to what is known for seed plants (Reski et al., 1997).

Thümmler et al. (1992b) isolated a cDNA from Ceratodon which encodes a novel type of phytochrome, and Algarra et al. (1993) and Thümmler et al. (1995) reported biochemical evidence that this cDNA, phyCer or CpPHY1, encodes a light-regulated protein kinase. However, Physcomitrella expresses a conventional, distantly B-type related phytochrome lacking any homology to protein kinases (Kolukisaoglu et al., 1993) and the predominant phytochrome in Ceratodon, CpPHY2, appears to be similar to that in Physcomitrella (Lamparter et al., 1995).

Partial rDNA-sequences have been identified from several mosses for cladistic purposes (Hori et al., 1985) and analysis of complete 18 S rDNA sequences shed new light on bryophyte systematics (Bopp and Capesius, 1995). Subsequently, molecular analyses in Funaria concentrated on the analysis of ribosomal RNA (Capesius, 1997).

Are They Green Yeasts?

Working on the PEG-mediated transformation of Physcomitrella, Schaefer (1994) noticed that re-transformation of transgenic lines was significantly enhanced when transforming plasmids were used which share, besides carrying different selection markers, extensive sequence homology. Segregation analysis of the double-resistant transgenic plants gave genetic evidence that the two selection markers were tightly linked in about 86% of the clones tested. This genetic evidence for highly efficient homologous recombination was confirmed by Kammerer and Cove (1996) who extended the genetic analysis and demonstrated linkage between two transgenes but not to a mutation leading to auxotrophy. Subsequently, Schaefer and Zryd (1997) presented molecular evidence that highly efficient homologous recombination occurs in Physcomitrella. As these data were restricted to targeting of unidentified genomic DNA it may be argued that evidence for gene knock-out of a functional gene was still missing. However, such reports are now appearing: Hofmann et al. (1997) reported homologous recombination with a member of the cab gene family and Strepp et al. (1997) disrupted a novel gene for plastid division, leading to altered transcription of this gene and subsequently to plants with undivided chloroplasts. More reports will surely be published in a short time.

Meanwhile there is no doubt that homologous recombination in Physcomitrella is as efficient as in Saccharomycescerevisiae where effective gene targeting was first demonstrated by Hinnen et al. (1978). Such ratios have not been reported for algae, flowering plants or mammalian cells, including mouse embryonic stem cells. The use of efficient homologous recombination permits application of reverse genetics to molecular biology. This powerful tool allows the study of gene function directly and efficiently by silencing genes via targeted knock-out or by mutating them in vitro and analysing mutated gene function by allele replacement. This approach avoids problems with "position effects", cosuppression or unstable phenotypes in anti-sense plants commonly encountered in plant molecular biology.

From the physiological data it is obvious that many basic aspects can be studied in the moss like in seed plants. Several genes have been cloned now from Physcomitrella and they are remarkably homologous to their cognate higher plant genes. All transformation experiments so far demonstrated that significant differences in promoter usage or in codon usage between Physcomitrella and dicotyledonous angiosperms such as Arabidopsis or Nicotiana do not exist. Furthermore, simplicity of the system allows developmental analysis at the cellular level to be carried out combining the methods of plant physiology and molecular genetics with those of modern cell biology in one organism.

The success story of yeast as being a widely used model system is based on the possibility to use reverse genetics in an eukaryote. Physcomitrella is not a microbiological organism and therefore can not compete with yeast in terms of growth rate and facility of handling. However, as a multicellular land plant, Physcomitrella obviously has added value.

A "Missing Link" between Algae and Arabidopsis?

Green algae, especially Chlamydomonas, are used as model systems due to their perfect simplicity as photosynthetic unicellular eukaryotes. Arabidopsis is used as a model, mainly because of its genome being the smallest known genome of land plants. Beside the arguments given above, are there more features that might qualify Physcomitrella as a new model for plant molecular biologists filling the gap between algae and Arabidopsis?

Bryophytes are very old land plants and some of them appeared 350 million years ago. It is known from the fossil record that there have been only small morphological changes since then, making them the most conservative group of land plants (Miller, 1984; Frahm, 1994), and thus ideal candidates for evolutionary studies. In no other group can three major evolutionary steps during early land plant development (filamentous growth of the juvenile gametophyte, "kormophytic structures" of the adult gametophyte, evolution of the diploid sporophyte) be studied independently. The presence of a multicellular gametophyte invites research on putative differences between gametophytic and sporophytic gene regulation and at the subcellular level evolution can be traced, as plastid ontogeny in mosses is distinctly different from that in vascular plants. In mosses, gametophytic meristems, including apical cells, contain chloroplasts closely similar to those in fully differentiated photosynthetic tissue (Duckett and Renzaglia, 1988). Thus, mosses facilitate studies on chloroplast evolution in land plants. Already Haberlandt (1918) pointed to the connection between plastid evolution and the evolution of land plants. Often, however, the plastid differentiation of angiosperms was taken pars pro toto and the study of plastid evolution is still in its infancy (Reski, 1994; Kasten and Reski, 1997). The same is even more true for mitochondrial evolution (Marienfeld et al., 1991; Malek et al., 1996).

A lot of data have been accumulated on the development, genetics and physiology of mosses over the last decades. It is clear since then, that mosses have several unique features:

· haploidy of the tissue facilitates mutant screens,

· simplicity of the system allows the observation of individual cell lines, and

· simplicity of the system pinpoints plant developmental processes to the differentiation of one single cell.

In combination with the highly efficient homologous recombination system in Physcomitrella the doors are now wide open to address a broad range of fundamental questions in a straightforward way that was hitherto unavailable.


I am indebted to the Deutsche Forschungsgemeinschaft for a Heisenberg-Fellowship (Re 837/3-1) and for funding several projects (Re 837/2, Re 837/4, Re 837/5, Graduiertenkolleg "Biotechnologie"). Some of the work in my lab was supported by the Commission of the European Union (AMICA PTP-project EUROMOSS), by the Freie und Hansestadt Hamburg and by the BASF AG. Special thanks go to all former and present members of the group for their enthusiasm, for their results and for stimulating discussions. I am delighted to thank Arthur Bräutigam (Hamburg) who prepared the artwork for this article and Susan Reynolds (Lausanne) for helpful comments and for correcting the English. Helpful suggestions on the manuscript came also from Martin Bopp (Heidelberg), David Cove (Leeds), and Eberhard Schnepf (Heidelberg). The consistent nomenclature of developmental genes has been suggested to and approved by my colleagues in EUROMOSS (David Cove, Leeds; Raffaele Gambardella, Naples; Michel Laloue, Versailles; Enzo Russo, Berlin; Trevor Wang, Norwich and Jean-Pierre Zryd, Lausanne) and several other colleagues attending an EUROMOSS meeting in Les Diablerets (Switzerland).

References and Figures

Please refer to the printed version of this article which appeared in Botanica Acta 111, 1-15.