Introduction

Bacteria of the genus Azospirillum possess numerous properties allowing them to survive and thrive in the nutrient rich and protective environment that exists in the rhizosphere of plants (Steenhoudt and Vanderleyden 2000). With the advantage of multiple plant growth promoting mechanisms (Bashan and de-Bashan 2010), Azospirillum strains have been tested and used as inoculants in crop production, initially with cereals, but later with other plants.

When the genus Azospirillum (Beijerinck 1925) was re-discovered by the late Johana Döbereiner and her colleagues in Brazil in the 1970s it was heralded as an associative plant growth-promoter for cereals (Döbereiner and Day 1976). Consequently, initial studies regarding the agronomic potential of these plant growth-promoting bacteria (PGPB) were conducted exclusively on cereals, as is often still the case (for reviews: Bashan and Levanony 1990; Okon and Labandera-Gonzalez 1994; Bashan and Dubrovsky 1996; Dobbelaere et al. 2001; Bashan et al. 2004; Massena-Reis et al. 2011, Table 1). The wealth of data concerning its interaction with cereals led to the conclusion that Azospirillum sp. has some as yet undefined specificity for gramineous plants (Baldani and Döbereiner 1980). By comparison to the specificity of Rhizobium-legume symbiosis, which is evident at early stages of the infection and involves specific molecular signaling among the bacteria and their host (Lerouge et al. 1990), caution is required when considering potential specificity of Azospirillum-cereal interactions, as the evidence better supports a possible affinity of some strains for cereals, rather than any specificity (Drogue et al. 2012).

Table 1 A sampler of the effect of inoculation of species of Azospirillum spp. on different plant species
Full size table

In subsequent studies Azospirillum species and strains known to affect the growth of cereals have been also tested on other species of plants worldwide. It has since became clear that many isolates of this genus can improve the growth and influence the metabolism of many plant species across many families, including annuals, perennials, trees, ornamental, spices, wild plants and even single cell microalgae (for documented examples see Table 1 and Table S1). Consequently, it is clear that Azospirillum can interact with a wide variety of plants and its species serve as a general plant growth-promoting bacteria (PGPB). However, the question still remains whether genus/species affinity for specific plants is evident and whether host specificity exists.

Given the pervasive contention that cereal-specificity of Azospirillum exists, critical re-evaluation of published research over the past four decades of implied specificity or preferred affinity of Azospirillum is essential. This review was driven by comparing four hypotheses derived and proposed from the prevailing assumptions within the literature: (1) Azospirillum as a genus has inherently a higher affinity for cereals and the effects recorded on other plant species are the exception, (2) Azospirillum has higher affinity to certain plant species but the increased affinity is at the species or strain level, (3) Azospirillum strains colonize and use a narrow range of plants as hosts, thus demonstrate host specificity and (4) Azospirillum species/strains are non-specific plant growth promoting bacteria affecting the metabolism of plants in general and some species/strain have a wide host range.

To examine these hypotheses we evaluated the literature using the following criteria: (1) the variety of plant species that show response to Azospirillum inoculation in general, (2) evidence for specific/unique reaction of plant only to certain Azospirillum species or strain (3) plants that have been shown to be colonized by Azospirillum and whether the attachment was exclusively by particular species/strains and (4) any knowledge of species/strain specific properties, including molecular traits, that are related to Azospirillum attachment to plants. Consequently, this essay was organized, as follows: historic background of the topic, phenomena of interaction of common strains of A. brasilense with multiple hosts, attachment and initial colonization as parameters of potential affinity/specificity, and molecular studies providing potential indicators of affinity of strains to plants.

The historic theme of “Specificity” of Azospirillum

The genus Azospirillum has currently 12 species (Lavrinenko et al. 2010), with the most studied species including: A. brasilense, A. lipoferum (Tarrand et al. 1978), A. halopraeferens (Reinhold et al. 1987), and A. oryzae (Xie and Yokota 2005). Recently, two species A. amazonense (Magalhães et al. 1983) and A. irakense (Khammas et al. 1989) were re-classified as Niveispirillum irakense and Nitrospirillum amazonense (Lin et al. 2014). Early claims of Azospirillum specific affinity for certain cereal species (Bashan and Levanony 1990) relied on inoculation studies that focused on strains of A. brasilense and A. lipoferum and were based on the following cases: (1) When responses of C3 and C4 plants were tested, A. lipoferum predominantly colonized C4 plants while A. brasilense predominantly associated with C3 plants both in tropical (Baldani and Döbereiner 1980; Baldani et al. 1986) and in temperate zones (Haahtela et al. 1981; Lamm and Neyra 1981). (2) Enhanced performances of cereal plants were more frequent when specific plant-bacterial species combination was used (Baldani et al. 1983, 1987; Reynders and Vlassak 1982; Pereira et al. 1988).

Moreover, strains of A. brasilense (SpT60, JM6A2 and Cd), isolated from different plants: wheat (Triticum spp.), maize (Zea mays), and Bermuda grass (Cynodon dactylon (L.) Pers.), respectively, showed distinct chemotactic response to organic acids, which correlated with the exudates of the respective plants of origin (Reinhold et al. 1985). Specificity can occur at the plant cultivar level; only a few of many tested cultivars of the same plants responded to inoculation with a given strain of Azospirillum (Bouton et al. 1979; Wani et al. 1985; Millet et al. 1986; Walker et al. 2011; Chamam et al. 2013), and plant genotype affected the response to Azospirillum inoculation (Garcia de Salamone and Dobereiner 1996; Garcia de Salamone et al. 1996).

The proposal of host specificity of Azospirillum was reinforced by discoveries in molecular biology. Finding sequences on Azospirillum plasmids with homology to rhizobial nodulation genes nodPQ and nodG during the late 1980s to early 1990s (Vieille and Elmerich 1990) supported the contention that specificity existed, arguing for an ability of Azospirillum to nodulate certain hosts. However, further analysis demonstrated no involvement of nod genes in Azospirillum-root interactions (Vieille and Elmerich 1992). Use of fluorescent probes demonstrated that the very common A. brasilense strain Sp245 is an internal root colonizer (Schloter et al. 1994), whereas the common strain Sp7 only colonized the root surface. Yet, this might differ from variety to variety or species to species. Contemporaneous research attempted to facilitate the interaction using synthetic auxins to create para-nodules on cereal roots that were colonized by Azospirillum (Tchan et al. 1991; Christiansen-Weniger 1992; Kennedy and Tchan 1992; Zeman et al. 1992; Sriskandarajah et al. 1993; Yu et al. 1993; Christiansen-Weniger and Vanderleyden 1994; Katupitiya et al. 1995a, b; Kennedy et al. 1997). Although this procedure did not result in endophytic colonization by strain Sp7 or establish a long-term colonization with other species, this approach created a new dimension in N2-fixation, which unfortunately was not developed further.

While the vast majority of publications have been on Azospirillum-cereal interaction, the proposal that the genus Azospirillum exclusively or mainly enhances growth of cereals has been confronted with newer evidence demonstrating effect on numerous other plant species from a variety of families (Table 1, Table S1). Currently, Azospirillum species are known to positively affect 113 plant species of which 14 are cereals and the rest non-cereals (Fig. 1a) for 34 additional botanical families (Fig. 1b).

Fig. 1
figure 1

Plant species, cereals vs. non-cereals inoculated with Azospirillum sp. showing plant beneficial effects (a). Botanical families on which Azospirillum sp. exerted beneficial effects (b). Families: 1. Asteraceae; 2. Fabaceae; 3. Poaceae (gramíneas); 4. Brassicaceae; 5. Apiaceae; 6. Solanaceae; 7. Amaranthaceae; 8. Malvaceae; 9. Piperaceae; 10. Cucurbitaceae; 11. Rosaceae; 12. Cactaceae; 13. Euphorbiaceae; 14. Convolvulaceae; 15. Caryophyllaceae. 16. Urticaceae; 17. Chlorophyceae; 18. Musaceae; 19. Casuarinaceae; 20. Cistaceae; 21. Linaceae; 22. Myrtaceae; 23. Phyllanthaceae; 24. Moraceae; 25. Fagaceae; 26. Arecaceae; 27. Pedaliaceae; 28. Acanthaceae; 29. Agavaceae; 30. Zingiberaceae; 31. Ranunculaceae; 32. Papaveraceae; 33. Iridaceae; 34. Geraniaceae; 35. Polygonaceae

Full size image

The accumulative data of the last three decades indicates that Azospirillum as a genus has the ability to interact with a wide variety of plants, including crop plants, weeds, annuals and perennials, and can be successfully applied to plants that have no previous history of Azospirillum in their roots. It appears that Azospirillum is a general rhizosphere colonizer and a general plant growth-promoter and its interaction with plants does not resemble legume-rhizobia specific interactions. This data does not preclude the possibility that Azospirillum species and strains may demonstrate plant preference, a possibility that must be investigated.

Interaction of Azospirillum strains with multiple host plants

Only a small variety of strains, including strains of A. brasilense and A. lipoferum have been commonly used in inoculation trials, some of which are commercially available for a variety of crops. Several examples of strains can demonstrate the multifaceted of activity on plants.

One of the best examples of a relatively promiscuous isolate is A. brasilense Cd/Sp7. A. brasilense Cd was isolated from plants inoculated with strain Sp7, thus they were sometimes considered to be one strain; however they have been known to display different phenotypes. They constitute one of the most studied strains for A. brasilense, having been isolated originally from a gramineous weed (Cynodon dactylon, Eskew et al. 1977) and commonly used as a reference strain. Initially, this strain was shown to colonize and enhance the growth and the yield of many winter and spring cereals (Kapulnik et al. 1981, 1983; Lin et al. 1983; Smith et al. 1984; Yahalom et al. 1984; Bashan 1986a; Assmus et al. 1995) and to move in soil towards wheat plants (Bashan 1986b; Bashan and Levanony 1987; Bashan and Holguin 1994). It had a marked capacity to enhance growth and yield of vegetables, industrial crop plants (Bashan et al. 1989b, c; 1991), burr medic (Medicago polymorpha L.) seedlings (Yahalom et al. 1990), common bean (Phaseolus vulgaris L.) (Burdman et al. 1996), environmental plants (Bashan et al. 2009b, 2012) and sunflower (Helianthus annuus L.) (Itzigsohn et al. 1995). Additionally, A. brasilense Sp7 could attach to arbuscular mycorrhizal structures (Bianciotto et al. 2001). The most unexpected enhancement was that of the rootless, single cell microalgae Chlorella vulgaris Beijerinck and C. sorokiniana Shihira et Krauss; the resemblance to its effects on growth, photosynthesis and metabolite content of plants (Bashan and Dubrovsky 1996; Gonzalez and Bashan 2000; de-Bashan et al. 2002; Bashan et al. 2006; Choix et al. 2012a, b) and its phenotypic cell-cell attachment employing fibrills (de-Bashan et al. 2011) made the combination of this strain with microalgae a proposed general model for plant-bacterial interaction (de-Bashan and Bashan 2008).

Another strain interacting with multiple hosts and multiple activities on plants is A. brasilense Sp245, which is able to colonize the surface of roots (Pereg Gerk et al. 2000) despite the fact that it was originally isolated from surface sterilized wheat roots and was shown to colonize wheat roots as an endophyte (Baldani et al. 1983, 1986; Schloter et al. 1993; Assmus et al. 1995). The strain is able to increase nitrate assimilation (Ferreira et al. 1987), alter membrane potential and proton efflux (Bashan 1990; Bashan and Levanony 1991; Bashan et al. 1989a) and enhance nitrogen accumulation (Boddey et al. 1986) and yield of wheat (Baldani et al. 1987; Okon and Labandera-Gonzalez 1994). Subsequently, this strain was used for inoculation of other plants and was capable of enhancing their growth. Strain Sp245 could increase the hormone abscisic acid content in Arabidopsis thaliana (L.) Heynh plants (Dubrovsky et al. 1994; Cohen et al. 2008) and have even more attributes. Those include: increasing growth and mineral content in soybeans (Glycine max (L.) Merr.) (Bashan et al. 1990), improving vigor of aged seed of lettuce (Lactuca sativa L.) (Carrozzi et al. 2012) enhancing germination and growth of the giant cardon cactus (Pachycereus pringlei (S.Watson) Britton & Rose) (Puente and Bashan 1993), improving the establishment of three cactus species in the field (Bashan et al. 1999) and promoting the growth of the halophyte Salicornia bigelovii Torr (Bashan et al. 2000). It survived well in the rhizosphere of tomato (Solanum lycopersicum L.) (Bashan et al. 1995), colonizing numerous weed species of different families (Bashan and Holguin 1995), inducing ammonium transporter in tomato root (Becker et al. 2002), mitigating salt effects on lettuce (Barassi et al. 2006) and promoted the growth of tomato, pepper (Capsicum annuum L.), and cotton (Gossypium hirsutum L.) (Bashan 1998; Bashan et al. 1989b, c; Bashan and de-Bashan 2005). Finally, it enhanced accumulation of intracellular nitrogen, phosphorus and enzymatic activities in a unicellular microalgae Chlorella vulgaris (de-Bashan et al. 2008b; Meza et al. 2015a, b).

A third example is A. brasilense Sp6. Originally isolated from maize, it promoted significant growth in maize (Barbieri and Galli 1993) and growth of roots of wheat (Barbieri and Galli 1993; Barbieri et al. 1986) and sorghum (Sorghum bicolor (L.) Moench) (Basaglia et al. 2003). The same isolate improved the growth of the shrub quailbush, Atriplex lentiformis (Torr.) S.Wats (de-Bashan et al. 2010a) and significantly changed the metabolism of the microalgae C. vulgaris and C. sorokiniana (de-Bashan and Bashan 2008; Meza et al. 2015a, b).

A fourth example is A. brasilense Az39, the most common commercial strain in Argentina, was tested on several cereals with significantly improved yield results (Fulchieri and Frioni 1994; Cassan et al. 2009a; Díaz-Zorita and Fernández-Canigia 2009; Zawoznik et al. 2011; Garcia de Salamone et al. 2012; Masciarelli et al. 2013) or in combination with legume nodule microorganisms (Cervantes and Rodriguez-Barrueco 1991; Perrig et al. 2007). This strain was proven growth promoter for soybean (Cassan et al. 2009b), and Casuarina sp. (Rodríguez-Barrueco et al. 1991). Numerous local publications from Argentina and India indicate the successful use of this strain for sunflower, tomato, cucumber (Cucumis sativus L.), pepper, squash (Cucurbita sp. L.), cabbage (Brassica oleracea L.), radish (Raphanus sativus L.), cotton, peanuts (Arachis hypogaea L.), alfalfa (Medicago sativa L.), Achicoria (Cichorium intybus L.), and the flowers of the day lily (Hemerocallis lilioasphodelus L.) and Nierembergia linariaefolia (Graham). (Table S1, supplementary material).

While there is evidence to show that strains of certain A. brasilense affect more than one group of plants, the main difficulties in assessing Azospirillum specificity or even affinity are the lack of studies methodically testing different strains of the same species (obtained from different sources) on specific host plant and specific strains on different host species.

Physiological and biochemical studies of attachment and initial colonization as parameters of potential affinity

Survival in the plant’s rhizosphere and promotion of plant growth do not necessarily qualify the plant as a host. To be a true host, the plant has to harbor the bacteria attached to or inside its tissue, ensuring long-term association between the two. The ability of Azospirillum species to attach to plant roots in various ways is well documented (multiple reference in Table 2) making some plants genuine Azospirillum hosts.

Table 2 Studies that showed actual attachment of Azospirillum strains to host roots, host cells or inert surfaces
Full size table

Azospirillum-mediated plant growth promotion utilizes various mechanisms that clearly demonstrate benefits to the host plant from this association, of which four are particularly important. Firstly, if Azospirillum is not attached to root epidermal cells, growth promoting substances excreted by the bacteria diffuse into the rhizosphere, consumed by nutritionally versatile microorganisms before reaching the plant and there is no mutual beneficial interaction (Bashan 1986a). With physical attachment part of these substances is diffused into the intercellular spaces of the root cortex. Secondly, without a secure attachment, rain or irrigation water may dislocate the bacteria from the rhizoplane to perish in the surrounding, nutrient-deficient soil. Azospirillum poorly survives in many soils without host plants (Bashan et al. 1995; Bashan and Vazquez 2000). Thirdly, association sites on roots having no attached Azospirillum represent a target for other aggressive rhizosphere colonizers, which are not necessarily beneficial for the plant. Finally, the root provides Azospirillum with microaerophilic niches that suit the physiological properties of this genus. Azospirillum proliferates under both aerobic and anaerobic conditions, but is preferentially microaerophilic in the presence or absence of combined nitrogen in the medium (Okon and Itzigsohn 1992) and it shows a strong aerotactic response towards the zones with reduced oxygen tension on roots (Okon et al. 1980; Patriquin et al. 1983; Reiner and Okon 1986; Zhulin et al. 1996; Alexandre et al. 2000; Stephens et al. 2006). There does not appear to be any evidence to suggest that Azospirillum is harmful to plants (Bashan 1998). The association between Azospirillum and the plant can thus be defined, in general, as mutualism suggesting a possible host specificity involved in this mutualistic relationship between the attached Azospirillum cells and the host at any level (genus, species or strain). The mode of attachment of Azospirillum to its host plant, as well as to other substrates, is an essential element to be ascertained.

Azospirillum strains can colonize roots externally and/or internally or can colonize the stem as an endophyte, as seen in rice (Oryza sativa), while some strains doing both (Table 2; Ramos et al. 2002; Zhu et al. 2002; Xie and Yokota 2005). Fluorescently labeled probes and monoclonal antibodies have confirmed the presence of Azospirillum strains in both the plant interior and the rhizosphere (Schloter et al. 1993; Assmus et al. 1995). Specifically, A. brasilense strain Sp245 was found in the root xylem, while Sp7 could only be detected on the root surface (Schloter et al. 1994). X-gal staining of labeled bacteria revealed that strain Sp7 initially colonized the sites of lateral root emergence and the root hair zone (Katupitiya et al. 1995a; Pereg Gerk et al. 2000) as does strain Sp245 (Vande Broek et al. 1993; Pereg Gerk et al. 2000). Washing wheat roots colonized by A. brasilense Cd removed most of the root-external bacteria and revealed a smaller internal root population (Bashan et al. 1986). A. brasilense Cd was also detected internally, within the cortex, using immuno-gold labeling (Levanony et al. 1989).

Root surface colonization is more common, in which the bacteria form small aggregates, although many single cells are scattered on the root surface. These surface colonizers are embedded in the external mucigel layer of the root (Umali-Garcia et al. 1981; Berg et al. 1979; Schank et al. 1979; Bashan et al. 1986; Murty and Ladha 1987; Pereg Gerk et al. 2000). Interestingly, both live and dead roots can be colonized (Bashan et al. 1986) suggesting that, while Azospirillum is attracted to root exudates, bacterial-host signaling is not essential for the actual attachment to roots, making a saprophytic growth phase of this bacterium probable.

When examining colonization by various strains, factors other than specificity can influence colonization and therefore can be responsible for differences among strains in separate studies. Factors important for such consideration include the culture age, experimental procedures and/or environmental conditions (multiple references in Table 2), and even the presence of other endophytic bacteria on the roots (Bacilio-Jimenez et al. 2001). In addition, some strains isolated from plant or rhizosphere (thus considered as environmental strains), such as A. brasilense strains Sp6 and Sp35 and A. lipoferum strains RG20, S28, and Br17 were not able to aggregate/flocculate (Pereg Gerk et al. 1998). Flocculation is both related to the production of exopolysaccharide (EPS) in Azospirillum and considered to be essential for firm attachment to root surface, suggesting that effects on plants observed following inoculation with these strains were not necessarily due to attachment to the roots but rather to the presence of these strains in the rhizosphere.

Azospirillum preferentially colonizes root elongation zones, root-hair zones and emergence of lateral roots with colonization patterns that depend on the host plant and bacteria strain (Bashan et al. 1986; Okon and Kapulnik 1986; Assmus et al. 1995; Pereg Gerk et al. 2000; Trejo et al. 2012). In wheat, colonization is mainly on the root surface and very few bacteria are attached to the root hairs (Okon and Kapulnik 1986; Bashan and Levanony 1989b), whereas in rice, massive root-hair colonization was frequently observed (Murty and Ladha 1987). In pearl millet (Pennisetum glaucum (L.) R.Br.) (Matthews et al. 1983), Kallar grass (Panicum antidotale Retz.) (Reinhold et al. 1986) and sugarcane (Saccharum sp.) callus (Berg et al. 1979; Vasil et al. 1979) most of the Azospirillum population was concentrated on the root surface. The colonization sites in some grasses corresponded to the areas where root mucigel was present, while the area around the point of emergence of lateral roots usually shows high colonization (Bilal et al. 1993).

It is clear that various plants can host bacteria from the genus Azospirillum and, evidence shows that species and even strains of Azospirillum can colonize more than one plant, suggesting a wide host range for each species/strain. Despite this, the colonization of plants has been visually demonstrated with only few strains of Azospirillum (Table 2) and a more thorough investigation of a large number of strains is required in order to conclude regarding host specificity/affinity of Azospirillum strains.

Mechanisms of attachment in relation to potential affinity

Examination of the mechanisms of attachment of Azospirillum to various substrates compared with plant roots better addresses the specificity of the interaction with hosts. Electron microscopic studies on several plant species have demonstrated that Azospirillum cells are connected to the root surface and to each other within the bacterial aggregate by a massive network of fibrillar material (Bashan et al. 1986; Levanony et al. 1989). Although azospirilla do not always show a uniform pattern of attachment in different experiments, even when the same strain is used on the same host (Michiels et al. 1989), it seems that aggregation by fibrillar material is the characteristic root surface colonization of this genus regardless the species or the strain (Umali-Garcia et al. 1980; Patriquin et al. 1983; Bashan et al. 1986; Gafni et al. 1986; Okon and Kapulnik 1986; Hadas and Okon 1987; Levanony et al. 1989). The chemical nature of these fibrils is not fully defined, but there are indications that they contain proteinaceous compounds (Bashan and Levanony 1989b) and polysaccharides that are responsible for the attachment phenomenon (Katupitiya et al. 1995a; Pereg Gerk et al. 1998, 2000). Fibrillar attachment by the bacteria is primarily dependent on active bacterial metabolism; killed bacteria did not attach to roots, but live bacteria attached to dead plant material (Bashan et al. 1986; Gafni et al. 1986). Initial root surface attachment is relatively weak and a slight rinsing of the roots releases most of the bacteria (Bashan et al. 1986) probably because of cell surface hydrophobicity, cell surface charges and cell surface lectins (Castellanos et al. 1997, 1998). Less thermodynamically stable polar attachment of Azospirillum cells to roots (Patriquin et al. 1983; Whallon et al. 1985; Levanony et al. 1989) comprised only a small fraction of the cells. Most of the root surface was colonized by bacteria in a horizontal, thermodynamically more stable, position. Examination of several strains of A. brasilense and A. lipoferum showed that although surface colonizers and endophyte strains had similar ability to anchor to wheat roots, strains with a proven ability to invade the root interior were more competitive in attaching to adsorption sites (de Oliveira Pinheiro et al. 2002).

Azospirillum has two different phases of attachment to wheat roots. The primary adsorption phase is fast but weak, reaching a maximum within 2 h of incubation, and likely governed by bacterial proteins. The second or anchoring phase takes longer, beginning after 8 h of incubation and reaching a maximum after 16 h, is stronger and irreversible, and appears to involve bacterial extracellular surface polysaccharides yielding long fibrils and a large amount of mucigel-like substances (Umali-Garcia et al. 1980; Zaady and Okon 1990; Gafni et al. 1986; Bashan and Levanony 1988b; Eyers et al. 1988a, b; Del Gallo and Haegi 1990; Michiels et al. 1990; 1991; Bashan et al. 1991; Levanony and Bashan 1991; Skvortsov et al. 1995; Puente et al. 1999). This type of attachment is not only to roots but also to plants cells as in the case of production of anchoring material when A. brasilense interacts with the single cell aquatic microalgae Chlorella vulgaris (de-Bashan et al. 2011). During this phase movement of Azospirillum along the root surface is minimal owing to formation of multistranded fibrils, although several single cells are capable of migrating among the different sections of the root system (Bashan 1986b; Bashan and Levanony 1991; Bashan and Holguin 1994) and among individual plants (Bashan and Levanony 1987; Bashan and Holguin 1995). These holdfast fibrils ensure vertical bacterial transfer by the growing root tip to deeper soil layers (Bashan and Levanony 1989a, 1991).

Adsorption and anchoring are probably different phenomena (Michiels et al. 1990, 1991) and have been observed in the roots of tomato, pepper, cotton, and soybean (Bashan et al. 1989b, c, 1991). The polar flagellum of A. brasilense, which is primarily used for swimming, was also involved in the initial attachment process of the bacteria to roots (Croes et al. 1993).

Several physiological, environmental, nutritional, and chemical factors modify A. brasilense attachment to the roots. Lectin and hydrophobic binding have been suggested as possible mechanistic mediators (Umali-Garcia et al. 1980; Tabary et al. 1984; Antonyuk et al. 1993; Karpati et al. 1995; Castellanos et al. 1997, 1998, 2000).

There are at least two different quantitative types of anchoring by this bacterium: a weak attachment to a non-biological surface and a stronger attachment to roots even though microscopically they resemble each other. The anchoring of A. brasilense Cd to hydrophobic polystyrene was significantly less than to roots and this is likely due to the hydrophobicity of the polystyrene (Bashan and Holguin 1993). Although most inoculated Azospirillum spp. survived only for a limited time in the soil (Bashan et al. 1995), some strains are soil dwellers especially in the tropics (Döbereiner et al. 1976; Döbereiner 1988). Upon inoculation to the soil Azospirillum cells are usually irreversibly adsorbed by the upper fraction of the soil profile in a charge-charge interaction mainly with clays and organic matter. Later they form attachments to soil particles such as sand, organic matter and clays using fibrillar material in a manner similar to attachment to roots. Physical and chemical soil conditions such as pH, flooding, dry regime, and availability of bacterial chemo-attractants greatly affect adsorption of Azospirillum to different degrees (Bashan and Levanony 1988a; Horemans et al. 1988). Attachment of Azospirillum to pure sand, which lacked clays and organic matter, was weaker and accomplished by a network of protein bridges produced between the bacteria cell and the quartz particles and mainly controlled by nutrient availability (Bashan and Levanony 1988b). Out of several strains examined, attachment to glass mediated by pili seems to be exclusive to A. brasilense Sp245 (Wisniewski-Dyé et al. 2011).

In summary, the phenotypic attachment of Azospirillum is mainly characterized by two-phase attachment, perhaps with the aid of the polar flagellum and a higher affinity of the attachment process to roots rather than to inert or soil surfaces. However, evidence suggests that there is a close resemblance between Azospirillum attachments to the root surfaces of different plant species, to the surfaces of single cells and to soil particles. Consequently, attachment to roots is potentially unsuitable for assessing the specificity at the species or strain levels. Affinity to roots in general may be concluded but there is insufficient biochemical and physiological data to suggest a particular affinity to cereal roots.

As a final note, endophytic Azospirillum strains may be the key to understanding the specificity of these strains to particular hosts, since here the bacterium has progressed beyond the attachment stage to enter the root system (for example, A. brasilense Sp245 in wheat roots, (Schloter et al. 1994) or the stem (for example A. oryzae COC8 and its relative Azospirillum sp. B510, Kaneko et al. 2010). Such advanced interaction suggests that the endophytic strains may communicate with the plant during the process of infection. Unfortunately, despite observations of internal colonization and advances in genomics, there is no information available on the actual mechanism of internal colonization by Azospirillum and its regulation. Further examination of Azospirillum at the molecular level may provide insights despite this lack of mechanistic information.

Molecular Azospirillum-plant interaction as potential indicator of affinity of Azospirillum strains

To date, only a small cohort of genes and molecular factors has been investigated for their involvement in the interactions between Azospirillum and plants, particularly those pertaining to potential insights regarding specificity/affinity. The lack of an easily detected plant phenotype that could be used to select bacterial mutants after inoculation with Azospirillum has complicated investigation of the genetic basis for the interaction. Consequently, mutations in traits considered to play a role in plant association have been mostly investigated and include genes involved in the production of auxins and surface compounds, genes sharing DNA homology with other plant-associative bacteria, such as nod, involved in nodulation by Rhizobium (Onyeocha et al. 1990), Rhizobium exo genes, involved in EPS production (Michiels et al. 1988; Petersen et al. 1992) and genes responsible for nitrogen fixation (nif and fix genes) (Vande Broek and Vanderleyden 1995). The p90 megaplasmid of Azospirillum was shown to carry genes such as exoBC, nodPQ, mot1,2,3 genes (production of polar and lateral flagella) and genes involved in IAA synthesis and in chemotaxis (Michiels et al. 1989; Katsy et al. 1990; Onyeocha et al. 1990; Van Rhijn et al. 1990; Vieille and Elmerich 1990; Elmerich et al. 1991; De Troch et al. 1994). Although no role was found for the nodPQ and exoBC homologous genes in Azospirillum colonization of roots, the plasmid p90 was named the rhizocoenotic plasmid, pRhico by Croes et al. (1991).

Attachment of Azospirillum to wheat roots is mainly dependent on two factors: the existence of a polar flagellum that allows the bacteria to attach to the roots and produce EPS, allowing bacteria to firmly attach to the root surface (Michiels et al. 1990, 1991; Croes et al. 1993). EPS production is regulated by the flcA gene, although the mechanism by which this regulation occurs is not fully understood (Pereg Gerk et al. 1998). This is the only regulatory gene that is known to be related to the attachment process.

The response regulator protein, FlcA, controls the shift of Azospirillum from vegetative state to cyst-like forms, both in cultures and in association with plants. Tn5 transposon-induced flcA − mutants do not flocculate, do not transform from motile vibriod cells into non-motile cyst-like forms and lack the EPS material on the cell surface under all conditions (Pereg Gerk et al. 1998). This leads to significantly reduced colonization efficiency of plant roots by Azospirillum, as they depend on the production of EPS to firmly attach to the root surface (Katupitiya et al. 1995b; Pereg Gerk et al. 1998, 2000). Development of reliable RT-PCR reference genes for Azospirillum (McMillan and Pereg 2014), facilitated the demonstration that flcA is involved in both the stress response and carbohydrate and nitrogen metabolism in Azospirillum during flocculation (Hou et al. 2014). Proteomics, RT-PCR (Hou et al. 2014) and cDNA-AFLP (Valverde et al. 2006) analyses to compare wild type A. brasilense Sp7 and non-flocculating flcA − mutants, have identified genes and proteins involved in the flocculation and aggregation of strain Sp7 (Hou et al. 2014), amongst which was a chemotaxis-like che1 homologue. Interestingly the Che1 pathway has been suggested to play a role in the adhesive cell properties of A. brasilense (Siuti et al. 2011) and controls swimming velocity, which affects transient cell-to-cell clumping (Bible et al. 2012). Similarly the nitrite/nitrate transporter NarK appears involved in aggregation (Valverde et al. 2006; Hou et al. 2014) and interestingly, the narL homologue, which possibly regulates respiratory membrane-bound nitrate-reductase, is highly expressed in wheat-bound A. brasilense FP2 cells (Camilios-Neto et al. 2014). While it has been established that flcA responds to environmental cues (Pereg Gerk 2004), its expression in response to various hosts has yet to be established. Further analysis of the conditions that affect flcA expression may shed light on Azospirillum affinity for various hosts.

Surface properties of Azospirillum are important in the attachment process. Two genes, noeJ (mannose-6-phosphate isomerase) and noeL (GDP-mannose 4,6-dehydratase), were related to EPS biosynthesis and biofilm formation (Lerner et al. 2009), and are present in the genome of several strains of Azospirillum (Sant’Anna et al. 2011; Wisniewski-Dyé et al. 2012). Disruption of dTDP-rhamnose biosynthesis by Tn5 mutagenesis modified the lipopolysaccharide core, EPS production and root colonisation in A. brasilense Cd (Jofre et al. 2004). Tlp1, an energy transfer transducer is required for taxis and for colonisation of wheat roots by A. brasilense Sp7 (Greer-Phillips et al. 2004). The pili mutant (cpaB −) of A. brasilense Sp245 has reduced biofilm formation and attachment to glass (Wisniewski-Dye et al. 2011). All of the above may be important in attachment to host roots. It is interesting to note that a spontaneous mutant of A. brasilence Sp7 lacking its lateral flagella, namely PM23, lost its ability to swarm through semi-solid medium but was able to attach to root surface (Pereg Gerk et al. 2000). Comparative analysis of the genomes of various species and strains of Azospirillum may link the molecular motifs related to attachment with the ability of different species to colonise different hosts as well as explain the different modes of colonisation discussed above (root surface versus root interior colonization).

Despite the great interest in this bacterium and it use as commercial inoculant (Bashan et al. 2014; Calvo et al. 2014), investigations into the molecular traits involved in its attachment to plants have been lacking. The value of such research was identified almost three decades ago (Elmerich et al. 1987) but the paucity of such studies persists to the present.

New insight into Azospirillum–plant interactions using high throughput technology: genomics and transcriptomics in relation to affinity for plants

The Azospirillum strain B510, an isolate of rice and a close relative of A. oryzae, was the first complete genomic analysis of an Azospirillum strain published (Kaneko et al. 2010). Subsequently, the growing number of genomic sequences available for Azospirillum species has facilitated valuable comparative studies, potentially illuminating the specificity of the Azospirillum-host interaction.

The detection of a tripartite ATP-independent periplasmic transport system and a diverse range of malic enzymes in the genome of Azospirillum strain B510 were implicated in the utilisation of C4-dicarboxylate during its interaction with rice (Kaneko et al. 2010). Genomic analysis of A. brasilense Sp245 suggested transition of the genus from aquatic to terrestrial environments at approximately the same time as the emergence of vascular plants on land (Wisniewski-Dye et al. 2011). A proposed high frequency of horizontal gene transfer (HGT) from soil and plant-associated bacteria is the suggested mechanism of adaptation to the rhizosphere and to the host plant. Such a mechanism would allow the bacteria to gain, rearrange and lose genetic traits as required for success in their ecological niche. Indeed, a high frequency of plasmid DNA rearrangements was reported for A. brasilense Sp7 that affected biofilm formation on glass and roots (Petrova et al. 2010). Moreover, higher genomic plasticity was shown in Azospirillum genomes compared to rhizobial genomes known for their genome plasticity (Wisniewski-Dye et al. 2011), strengthening the suggested link between the appearance of phenotypic variants and plasmid loss or reorganization (Vial et al. 2006b). High genomic plasticity supports the suggestion that Azospirillum possesses mechanisms of adaptation to it various hosts.

The demonstration that the Azospirillum genome acquired a substantial number of glycosyl hydrolases by HGT that are essential for decomposition of plant cell walls and that the A. brasilense Sp245 genome contains three enzymes that are orthologous to cellulases (Wisniewski-Dyé et al. 2011), supports previous suggestions that cellulolytic activity may be crucial to the ability of some Azospirillum strains to penetrate plant roots (Skvortsov and Ignatov 1998). A comparison between whole genomes of four Azospirillum strains, A. brasilense strains CBG497 (from maize grown on alkaline soil) and Sp245, A. lipoferum 4B and Azospirillum sp. B510 (Wisniewski-Dyé et al. 2012), reveal that the Azospirillum core genome (AZO-core) is dominated by proteins of ancestral origin (74 %) and 22 % of proteins acquired by HGT. While the 62–65 % of the AZO-core (strain dependent) is mainly chromosomally-encoded, the non-chromosomal proportion of AZO-core is unevenly distributed among strains. Several strain-specific genes were found to be involved in the colonization of plant roots; in a comparison among the four strains, additional flagellation and chemotaxis operons were found in A. lipoferum 4B and in Azospirillum sp. B510, while additional genes involved in EPS biosynthesis and/or transport and LPS biosynthesis were present in the A. brasilense strains, possibly acquired by HGT (Wisniewski-Dyé et al. 2012). Such strain-specific genes may suggest variability in the mode of interaction of strains with their hosts. It remains unclear, however, what effect, if any, this has on host specificity.

Although only four A. lipoferum strains out of 40 Azospirillum strains demonstrated acyl-homoserine lactone (AHL) biosynthesis ability (Vial et al. 2006a), it was suggested that strain-specific quorum sensing regulates functions linked to rhizosphere competence and adaptation to plant roots in Azospirillum (Boyer et al. 2008). Indeed, the genome analysis of A. amazonense Y2 revealed the presence of genes encoding for LuxI and LuxR homologs proteins suggesting it could synthesize AHLs and respond to their presence in the environment. The genome of A. amazonense also presents a Klebsiella pneumoniae ahlK homologue, possibly encoding a putative homoserine lactonase implicated in AHL degradation (Sant’Anna et al. 2011).

Although no difference was found in the anchoring ability of surface colonizers and endophytes of A. brasilense and A. lipoferum (de Oliveira Pinheiro et al. 2002), genomic comparisons suggest that Azospirillum strains have gained different root-adhesion mechanisms (Wisniewski-Dyé et al. 2012). Tight adherence (TAD) pili, essential for colonization and biofilm formation, are exclusive to the A. brasilense species while genes involved in cellulose synthesis probably acquired by HGT are found exclusively in A. lipoferum 4B and Azospirillum sp. B510. Specificity was also found in the range of cellulases and hemicellulases produced by the different strains, with both A. brasilense strains encoding glycosyl hydrolase-encoding genes with no orthologues in any other Azospirillum genomes. A. lipoferum 4B and Azospirillum sp. B510 seem to be more versatile for aromatic compound degradation than A. brasilense strains, with a wider range of aromatic ring-hydroxylating dioxygenases, proposed to be related to the composition of the host plant exudates, as a result of niche-specific adaptation and environmental conditions (Wisniewski-Dyé et al. 2012).

Profiling of plant secondary metabolites of maize–Azospirillum (Walker et al. 2011) and rice–Azospirillum (Chamam et al. 2013) associations revealed strain-specific responses and suggest specific interaction between Azosprillum strains and their original host cultivar. The response of two Oryza sativa japonica (rice) cultivars, Cigalon and Nipponbare, to a root surface colonizer A. lipoferum 4B (isolated from Cigalon) and an edophytic Azospirillum sp. B510 (isolated from Nipponbare), investigated using root transcriptome profiling, revealed not only strain-specific responses of rice, but also combination specific responses (Drogue et al. 2014). Most of the differentially expressed genes were related to primary metabolism, transport and gene regulation; however, strain specific response was also observed for genes related to auxin and ethylene signaling suggesting complex response to hormone signaling (Drogue et al. 2014). When considering differentially expressed genes to that of un-inoculated plants, inoculations by Azospirillum lead to the expression of genes related to stress response and plant defense in both rice inoculated with A. lipoferum 4B (Drogue et al. 2014) and wheat inoculated with A. brasilense FP2 (Camilios-Neto et al. 2014). However, inoculation of rice by the endophytic strain B510 seems to lead to the repression of a wider set of genes than A. lipoferum 4B (Drogue et al. 2014). A possible explanation is the ability of this strain to colonize the plant internally. Specificity may occur in the molecular responses of Azospirillum strains to their hosts, even though Azospirillum, as a genus, can interact with a wide range of hosts. Further investigation is required to explain the cellular changes of specific strains during association of this PGPB with plants.

Conclusions

Analysis of the extensive published data over four decades has facilitated certain conclusions about the specificity and/or affinity of organisms in the genus Azospirillum. This genus represents general plant growth-promoting bacteria improving the growth of 113 plant species of 35 botanical families, without any solid evidence of species specificity to selected plant species. Previous claims for specificity/affinity of Azospirillum for cereals, proposed in the early days of the field, are unsupported and may represent historical assumptions, likely related to the original isolation of Azospirillum organisms from cereals and experimentation performed almost exclusively on cereals. Numerous studies on Azospirillum-cereals association merely reflect the economic importance of cereals as crops, resulting in inaccurate claims that the main effect of this genus is on cereals. The demonstration of Azospirillum-plant interactions that stretch beyond cereals is expected to drive future research and greatly expand our knowledge of this important crop-enhancing genus.

It remains to be seen; however, the degree to which affinity may exist between different Azospirillum strains and various plant species, as does exist for many other PGPB for which specific strains performed better with specific plant genotypes or cultivars. Regardless of the wealth of research on Azospirillum–plant interactions, there is insufficient data regarding comparative bacterial strains-plant species or bacterial strain-plant genotypes. Additionally, comparative molecular analysis of different strains is in its infancy because only a handful of strains of Azospirillum are fully sequenced.

Future potential useful lines of research in this topic

Although no new specificity of Azospirillum to specific plant species is expected to emerge, the affinity of strains to plant genotypes or to plant species is worth investigating. This will involve examination of:

  • A large number of strains tested on one plant species or a specific strain tested on multiple hosts and isolation of the strain after plant growth are required to propose affinity of Azospirillum strains.

  • Molecular comparison at the entire genome level should be done between strains claimed to have affinity to specific plant species or to plant genotypes.

  • A comparison regarding the differences between flocculating wild-type strains and non-flocculating mutants (often impaired in colonization) of the same strain regarding their effects on plant.

  • A comparison between motile strains and strains naturally impaired in motility (affected by production of polar flagella that are required for initial Azospirillum-root interaction) on plants exuding large amount of exudates and those limited in the quantity and variability of these exudates.

  • A comparison of cross inoculation between strains of different geographical origin (tropical, desert, temperate, aquatic) and plants from the same zone or with plants with Azospirillum strains isolated from another origin.

So far, the lack of knowledge regarding the specificity/affinity of Azospirillum does not detract from the numerous companies that offer commercial products for inoculation with Azospirillum. This knowledge will make possible tailoring of a better, future market product for sustainable agriculture in common agriculture practices and organic farming.