Year: 2007
Type: review
Abstract: Recent findings on legume biogeography and the timing of evolution of key legume tribes have supported a new view of the evolution of nodule processes. It is suggested that an initial infection process not involving root hairs led to two branches of legume nodule development, one that subsequently developed transcellular infection threads (ITs) to carry bacteria to young nodule cells and one in which such ITs were not formed. Two types of nodules, with indeterminate or determinate growth, evolved from each of these. Knowledge of the diversity of bacteria known to nodulate legumes and their relations with other bacteria is expanding rapidly, posing new questions about nodulation in the field. Ectomycorrhizas (ECMs) are found in both nodulating and non-nodulating legumes and may be important in some environments. This Update will address the following topics: (1) when and where nodulation evolved in legumes; (2) the key processes that led to nodule structures found in extant legumes; (3) the growing number of nitrogen-fixing bacteria known to nodulate legumes; and (4) the role of ECMs and endomycorrhizas in certain legume groups. Structure of the main types of legume nodules. A, Sesbania macrantha root nodule. Although morphologically similar to the aeschynomenoid type of nodule seen in B, the infected tissue contains uninfected cells and bacteria are transmitted to infected cells by ITs. B, Aeschynomene rostrata stem nodule. This is typical of a clade of dalbergioid legumes. ITs are never formed and infected tissue contains no uninfected cells. Infection occurs through breaks where lateral or adventitious root initials protrude and a few infected cells divide repeatedly. C, Mimosa himalayana. This structure is typical of all mimosoid and many papilionoid nodules and in most cases follows from root hair infection. There is a clear apical meristem (arrow), and the infected tissue contains a mixture of infected and uninfected cells. ITs convey bacteria to cells newly formed by the meristem. D, Cytisus garden hybrid, typical of many genistoid legumes. ITs are never seen and infected tissue contains no uninfected cells. There is a distinct apical meristem (arrow), which may divide, forming branched nodules or in some cases encircle the root (Lupinus, Lotononis). E, L. uliginosus, a typical determinate nodule as found in many members of tribe Loteae and in phaseoloid legumes such as soybean (Glycine max). Meristematic activity is short lived, infection is via root hairs, and infected tissue contains uninfected cells. F, Erythrophleum ivorense, a typical caesalpinioid nodule with a blunt apex, a clear apical meristem (arrow), and uninfected cells in the infected tissue. Infected cells retain bacteria in modified ITs, known as fixation threads. They may branch repeatedly and be lignified in the outer layers. Doyle and Luckow (2003) suggested that there may have been four separate events leading to nodulation in legumes, but also pointed out that resolution of the basal, non-nodulating clades of legumes is still incomplete and may change in the coming years. So far there is no additional information to resolve this question, so in this Update we concentrate on the processes leading to formation of nodules, rather than the number of nodulation events. A tentative scheme for the evolution of different types of nodule structure. Dashed line, Pathway not fully demonstrated. In addition to asking when and where legumes evolved, it is also relevant to ask why nodulation evolved in some groups. As the process of nitrogen fixation uses a significant amount of the total carbon fixed by the host plant, one driving force could have been an excess of carbon dioxide coupled with a deficit of combined nitrogen. Several lines of evidence suggest that, at about 55 million years ago, when nodulate legumes may have evolved, there was a major peak in atmospheric carbon dioxide, temperature, and humidity (Bowen et al., 2004; Sprent, 2007). First, compatible rhizobia needed to gain entry into the legume root. The most widely studied mode of entry is via root hairs and involves transcellular ITs. However, even the species that normally use this pathway may, under certain circumstances (usually a form of stress), become infected through breaks in the epidermis or wounds where lateral roots emerge (crack entry). Examples include white clover (Trifolium repens; Mathesius et al., 2000), Lotus uliginosus (James and Sprent, 1999), and the mimosoid aquatic legume Neptunia natans (= Neptunia oleracea; Subba-Rao et al., 1995). For the latter species, the passage of bacteria between host cells and then the formation of transcellular ITs is clearly illustrated. We postulate that the default position for infection is directly between epidermal or cortical cells, and that this may lead to one of two patterns of nodule development. The first never involves transcellular ITs, although as bacteria pass between cells they may be surrounded by some of the extracellular components normally found in transcellular ITs (Brewin, 2004). This mode of infection is found in the dalbergioid and genistoid legumes, and may account for approximately 25% of all legume genera (Sprent, 2007). Considering that these groups include important grain (some species of Lupinus, Arachis) and forage (Stylosanthes) legumes, they have been surprisingly little studied. Arachis and Stylosanthes have aeschynomenoid nodules, formed following crack infection where lateral (occasionally adventitious) roots emerge. A few cells are infected by bacteria as they pass between cells. These host cells divide repeatedly to give the characteristic uniformly infected central tissue, with loss of meristematic activity (Fig. 1B; see Lavin et al. [2001] and Sprent [2001] for further details and references). Although the general structure of the indeterminate nodules of several genistoid legumes has been known for many years (Sprent, 2007, and refs. therein), their detailed development has only recently been described. Studies on Lupinus albus and Chamaecytisus proliferus (now included in Cytisus) describe infection directly via the epidermis or at the bases of root hairs (Vega-Hernández et al., 2001; González-Sama et al., 2004), with a few host cells being infected and these dividing repeatedly to give uniform infected tissue, but with some cells retaining meristematic activity. Genista tinctoria nodules are very similar in structure to those of Cytisus (Fig. 1D), and Kalita et al. (2006) show clearly how infected cells in the apical meristem divide, forming new nitrogen-fixing tissue as the nodule grows. The second type of nodule development involves development of transcellular ITs. Although generally associated with root hair infection, they may not always be. Lonchocarpus muehlbergianus is a member of the important tropical tribe Millettieae. It does not produce root hairs and infection probably occurs between epidermal cells, with later formation of transcellular ITs (Cordeiro et al., 1996). Subsequently, as in indeterminate nodules with root hair infection, individual cells are infected by branches of the transcellular ITs and active nitrogen-fixing tissue contains a mixture of infected and uninfected cells. This pattern of development has been studied in detail for many papilionoid legumes and also appears common in at least some Mimosoideae and all Caesalpinioideae (Fig. 1, C and F). The position is similar in the determinate nodules of phaseoloid legumes (including Glycine, Phaseolus, and Vigna) and Lotus in tribe Loteae, except that meristematic activity is short lived (Fig. 1E). Entry of transcellular ITs into newly formed meristematic cells is accompanied by cessation of later phases of mitotic division, so that cells become polyploid and greatly enlarged, enabling them to house vast numbers of nitrogen-fixing bacteria. In indeterminate nodules, bacteria also show high levels of DNA replication and this is accompanied by loss of viability. In determinate nodules this does not occur (Mergaert et al., 2006). However, endoreduplication also occurs in Lupinus nodules, which do not have transcellular ITs (Gonzáles-Sama et al., 2006). The universal presence of uninfected cells in the infected tissue of nodules with transcellular ITs suggests that these may have a role in nodule functioning. This is certainly true of determinate ureide-exporting nodules (those in the phaseoloid group) where these interstitial cells are the main site of synthesis of the ureides allantoin and allantoic acid, the chief export products from such nodules (Sprent, 2001). The function of interstitial cells in determinate Lotus nodules (these export amides, not ureides) and indeterminate nodules is not clear, but they seem to be a required structural feature. Further, genetic information for differentiation of nodules in the absence of rhizobia, including the formation of large (normally infected) and small (normally interstitial) cells in the central tissue, is located in the host legume for at least some of the more recently evolved vicioid (galegoid) legumes (Pa et al. [1991] for alfalfa [Medicago sativa]; Blauenfeldt et al. [1994] for white clover; Gleason et al. [2006] for Medicago truncatula; and Tiricine et al. [2006] for Lotus japonicus). There have been occasional reports (Allen and Allen, 1981; Bryan et al., 1995) that roots of the caesalpinioid legumes Gleditsia and Peltophorum can be invaded by rhizobia, followed by formation of ITs but without the formation of nodules. Bryan et al. (1995) thought that such processes could be an early stage in nodule evolution and ITs are certainly a feature of all nodulated caesalpinioid legumes (Sprent, 2001). In nearly all of the latter, bacteria are not released from ITs and nitrogen fixation takes place within modified ITs, called fixation threads (Sprent, 2001). This led to the suggestion (Sprent, 2007) that ITs were initially a defense response to an invading organism. In the caesalpinioid genus Chamaecrista, there is a spectrum of structures from fixation threads in arboreal species to full release of bacteria into symbiosomes in herbaceous species (Naisbitt et al., 1992). A more detailed analysis of these species might provide information on the evolution of symbiosomes. In evolutionary terms, the formation of transcellular ITs is a necessary prerequisite for root hair infection. The role of cell wall materials in this process has been reviewed by Brewin (2004). Discussion on evolution of nodulation has hitherto taken into account presence or absence of nodules and nodule morphology (for example, Doyle and Luckow, 2003). Determinate and indeterminate nodule growth has proven to be a useful criterion. However, within these two groups, it is now clear that there are distinct differences in how individual host cells are infected and whether the infected cells are interspersed with uninfected cells. Figure 2 summarizes a hypothesis that legume nodules were first initiated from direct epidermal or crack infection and that this led to two distinct branches of nodule development, one involving transcellular ITs and one not. Further details can be found in Sprent (2007), available (free) online. Although most legume databases are confined to species with a root hair infection, there are some that are more widely based. Of those tabulated by Stacey et al. (2006), one (PlantGDB; www.plantgdb.org) includes the dalbergioid legume Arachis hypogea and the Australian legumeDB (Moolhuijzen et al., 2006) includes the genistoid legume Lupinus angustifolius. As more data are added to these databases, it may be possible to dissect out the genes and processes responsible for the major differences in nodule characteristics between these and root hair-infected species and, hence, test our hypothesis. Doyle and Luckow (2003) titled their paper “The Rest of the Iceberg,” indicating that the vast majority of legumes were under (or often not) studied. The same could be said of the bacteria that nodulate legumes, although this iceberg is melting rapidly. In the beginning (over a century ago), only one nodulating bacterium had been described, Bacillus radicicola. Shortly after, fast- and slow-growing rhizobia were distinguished and were subsequently given different generic names (Rhizobium and Bradyrhizobium). There are now several more genera of rhizobia, with numerous species, together with other bacteria from the α-proteobacteria, plus an increasing number from the β-proteobacteria (Table I Nodulation of legumes by defined species of β-rhizobia Nodulation of legumes by defined species of β-rhizobia The close similarities between plant and animal infection strategies in α-proteobacteria, including probably the best known example, Rhizobium and Brucella, are discussed by Batut et al. (2004), with a more detailed examination of rhizobia and plant pathogens by Soto et al. (2006). Horizontal (lateral) gene transfer in the form of genomic islands between various bacteria is considered by Dobrindt et al. (2004). It is likely that such a transfer of symbiotic islands accounts for many of the nodulation reports in β-rhizobia. Bernier et al. (2003) suggested that the common opportunist human pathogen Burkholderia cepacia, which can also enter wounds in plants, causing disease, could well be studied using alfalfa as a model system. Since then a strain of this organism has been isolated from nodules of species of Dalbergia in Madagascar (Rasolomampianina et al., 2005). Thus, many of the differences among bacteria invading eukaryotes are rapidly turning into similarities. It is not appropriate to go into details of these systems here, except to emphasize that not only are there numerous legume plants whose nodulation details are largely unknown (Doyle and Luckow, 2003; Sprent, 2007), but also there are far more nodulating bacteria than have yet been studied for their interaction with host genotypes. If the benefits of nitrogen fixation in legumes are to be more widely understood and exploited, especially in some tropical soils, these interactions need to be explored. It is too early to speculate how these β-rhizobia relate to either legume phylogeny or evolution, but it may be relevant that, so far, they have only been found in tropical areas. There are two main types of mycorrhiza in legumes, arbuscular mycorrhizas (AMs) and ECMs. As AMs evolved long before legumes, we may assume that all legumes have the potential to produce them (Lupinus is the only known legume genus in which this ability has been lost). Similarities between initial processes involving infection by AM fungi and rhizobia are being extensively investigated and reviewed (Kinkema et al., 2006; Stacey et al., 2006), and will not be considered here. Further, some of these processes, including endoreduplication, may have been hijacked by root-knot nematodes (Weerasinghe et al., 2005). The occurrence of ECMs in legumes is sporadic. The typical ECM, with a sheathing mantle and Hartig net, is characteristic of some of the Caesalpinioideae. In the analysis of Lavin et al. (2005), a branch of this subfamily that is non-nodulating has one section that includes only ECM genera, the others being AM (Sprent 2007). This suggests a common origin of ECMs in this branch, the plants of which are mainly trees of African rainforests. Here, their ECMs are found principally in the litter layers, as are nodules on some of the few, but profusely, nodulated legume species (Sprent, 2005). ECM legumes are a vital part of the phosphorus dynamics of such forests (Newbery et al., 1997). Although it used to be thought that ECMs and nodulation in legumes were mutually exclusive (Malloch et al., 1980), this is now known not to be true. For example, Högberg and Pierce (1986) reported that Pericopsis angolensis (a woody species in papilionoid tribe Sophoreae), normally ECM and non-nodulating, can also form AMs on plants that can then also nodulate (as can other species of this genus; Sprent, 2001). Even ECM legumes that cannot nodulate may be able to form AMs in certain locations (Moyersoen and Fitter, 1999), a feature that may enable them to exchange nutrients with nodulated AM legumes in some tropical forests (Sprent, 2005, 2007). A switch between forms of mycorrhiza according to environment is known from other, nonlegume species, such as Populus angustifolia (Gehring et al., 2006). The genome of another species of this genus, Populus trichocarpa, has recently been published (Tuskan et al., 2006), raising the possibility that genes controlling different types of mycorrhizal formation may soon be identified. Before these can be aligned with legumes, however, we need information on the genomes of legume species that can form both types. There are reports of ECMs in the other legume subfamilies, in plants from soils rather low in nutrients and water and without a pronounced litter layer. All are Australian endemics, although some acacias can form ECMs with local fungi when grown in countries as far apart as Brazil and East Africa (Sprent, 2001, and refs. therein). Papilionoid tribes Mirbelieae and Bossiaeeae have a number of genera capable of forming typical ECMs and rather looser associations, as found in species of Australian Acacia, subgenus Phyllodineae (Alexander, 1989; Sprent, 2001). All can also form AMs and, in some cases, cluster roots. Thus, many legumes appear to have the potential to form both AMs and ECMs, as do many other nodulated members of the Rosid 1 clade (Wang and Qui [2006] list these, although not using the cladistic analysis of Soltis et al. [2000]). Thus, evidence now suggests that legumes are very versatile in their symbioses. Unfortunately, the molecular aspects of ECM development have been far less studied than those for AMs, with no studies at all on ECMs in legumes. Nodulation has a significant requirement for phosphorus (P), so it would seem sensible to have P-acquiring symbioses (AM and/or ECM) near to nodules. This is true of AMs, and there have been occasional reports that AM hyphae colonize nodules. However, this appears only to be true for nonfunctional nodules (Scheublin and van der Heijden, 2006; our group has sectioned many thousands of nodules from hundreds of legume genera, and we have never seen AM hyphae in them). On the other hand, ECMs may be spatially separated from nodules (for example, in the case of P. angolensis [above]). Cluster roots can assist in the uptake of P, and, in soils with low available P, nodules are formed among them (Schulze et al. [2006] for L. albus; F. Dakora [personal communication] for Aspalathus linearis).
Source: PLANT PHYSIOLOGY
Authors: Janet I. Sprent, Euan K. James
Institution: University of Dundee
Language: English
Subfield: Plant Science
Domain: Life Sciences
SDG: Zero hunger
Open Access status: bronze
Award: G8233038880
OpenAlex ID: W2111306132
