Applied Microbiology and Biotechnology

, Volume 65, Issue 5, pp 497–503

Gibberellin production by bacteria and its involvement in plant growth promotion and yield increase


    • Cátedra de Química Orgánica y Biológica, Facultad de Ciencias AgrariasUniversidad Nacional de Cuyo
  • Fabricio Cassán
    • Laboratorio de Fisiología Vegetal, Facultad de Ciencias Exactas, Físico-Químicas y NaturalesUniversidad Nacional de Río Cuarto
  • Patricia Piccoli
    • Cátedra de Química Orgánica y Biológica, Facultad de Ciencias AgrariasUniversidad Nacional de Cuyo
    • Area de InvestigacionesUniversidad Juan A. Maza

DOI: 10.1007/s00253-004-1696-1

Cite this article as:
Bottini, R., Cassán, F. & Piccoli, P. Appl Microbiol Biotechnol (2004) 65: 497. doi:10.1007/s00253-004-1696-1


This review focuses on studies with bacteria for which biosynthesis/production of the plant hormones gibberellins have been demonstrated. Actual data on gibberellin metabolism by bacteria are analyzed in comparison with the biosynthetic pathways known for vascular plants and fungi. The potential involvement of gibberellins produced by symbiotic and soil-endophytic microorganisms in plant growth promotion and yield increase is also discussed.

Gibberellins as plant hormones

Gibberellins are tetracyclic diterpenoid acids that are involved in a number of developmental and physiological processes in plants (Crozier et al. 2000; Davies 1995). These processes include seed germination, seedling emergence, stem and leaf growth, floral induction and flower and fruit growth (King and Evans 2003; Pharis and King 1985; Sponsel 2003). Gibberellins are also implicated in promotion of root growth, root hair abundance, inhibition of floral bud differentiation in woody angiosperms, regulation of vegetative and reproductive bud dormancy and delay of senescence in many organs of a range of plant species (Bottini and Luna 1993; Fulchieri et al. 1993; Reinoso et al. 2002; Tanimoto 1987). In most (if not all) of these processes gibberellins act in combination with other phytohormones and additional regulatory factors, so that the signaling pathways are highly integrated (Trewavas 2000). A major problem in understanding the role of gibberellins is that the scientific information available utilizes many different species and diverse experimental models, making it difficult to extrapolate results across species, genera and family boundaries. To date, 136 different chemical structures have been characterized as naturally occurring gibberellins ( To this figure should be added an unknown number of glucose-conjugate forms (Schneider 1983) and likely intermediates yet to be identified in biosynthetic and catabolic pathways (Sponsel and Hedden 2004). Of these 136 gibberellins, the 3β-hydroxylated, C19 gibberellins GA1, GA3 and GA4 (Fig. 1), have been reported by studies with single gene dwarf mutants as being directly effective in promotion of shoot elongation in plants (Crozier et al. 2000). However, it is very likely that other 3β-hydroxylated C19 gibberellins also function per se as effectors of shoot elongation in a wide range of plant species.
Fig. 1

Chemical structure of GA1, GA3 and GA4, the three gibberellins reported by studies with single gene dwarf mutants as being directly effective in promotion of shoot elongation in plants (Crozier et al. 2000)

Historically, gibberellins were discovered in culture filtrates of the fungus Fusarium moniliforme (Gibberella fujikuroi in the sexual form) in 1926 in Japan by Kurosawa (cited in Tamura 1990) and their chemical structure was partially elucidated 10 years later (Yabuta and Sumiki 1938, cited in Tamura 1990). The Western world, however, only became aware of their existence some two decades later when the structure of gibberellic acid was confirmed (GA3, Fig. 1, Curtis and Cross 1954). Soon afterwards, the first plant gibberellin (GA1, Fig. 1) was identified by Macmillan and Suter (1958) from Phaseolus coccineus seeds.

However, gibberellins are produced not only by higher plants and fungi (MacMillan 2002) but also by bacteria (Atzorn et al. 1988; Bastián et al. 1998; Bottini et al. 1989; Gutiérrez-Mañero et al. 2001; Janzen et al. 1992; MacMillan 2002). In fungi and bacteria there is no known role for gibberellins, rather they seem to be secondary metabolites that may play a role as signaling factors towards the host plant.

Bacteria known to improve plant growth and crop yield

In the twentieth century there was a global tendency toward increased use of fertilizers, especially nitrogen, as a way to improve crop productivity. This tendency first developed in industrialized countries, but in the 1960s the “green revolution” extended it to the “third world”. However, the progressive increase in use of mineral fertilizers has posed a severe treat to a wide range of ecological systems. Thus, more recently there has been growing attention focused on more “environmentally friendly” N2-fixing bacteria as a mode of increasing crop yield (Okon and Labandera-González 1994).

For many years most efforts emphasized Rhizobiaceae symbiotic associations. In fact, their use now extends worldwide and the benefits from increased nitrogen supply are unanimously recognized. Since the 1970s, however, an increasing interest has developed in “free-living” rhizosphere bacteria (i.e., Pseudomonas, Bacillus, Azotobacter, Azospirillum; De-Polli et al. 1977; Döbereiner et al. 1976; Ludden et al. 1978; Okon et al. 1976a,b), including the so-called plant growth promoting rhizobacteria (PGPR, reviewed by Glick et al. 1999). While use of Rhizobiaceae symbiotic bacteria is restricted almost exclusively to the Leguminosae, PGPR are being utilized with cereals like wheat or rice, the primary global source of food. Nevertheless, it has been postulated that the beneficial effects of PGPR microorganisms are not due solely to N2 fixation (Bashan and Levanony 1990; Okon and Labandera-González 1994, and literature cited therein). The benefits of PGPR use seem to be a consequence of a complex mix of different mechanisms, which may include N2 fixation by nitrogenase, nitrate reductase activity, siderophore production, and phytohormone synthesis/metabolism and release to the plant (Cassán et al. 2001a,b; Fulchieri et al. 1993). Other mechanisms that have been proposed, such as increasing water and mineral uptake, are more likely to be the result of root growth promotion by any (or all) of the above mentioned mechanisms (Cassán et al. 2003). Nonetheless, although the mechanisms by which PGPR promote increases in crop yield are not fully elucidated, the synthesis of phytohormones, including gibberellins (Bottini et al. 1989), and the absorption of these hormones by the crop plant are considered to be key causal factors (Cassán et al. 2001a,b; Fulchieri et al. 1993; Okon and Labandera-González 1994).

Gibberellin characterization and metabolism in bacteria

The first report of gibberellin characterization in bacteria using physico-chemical methods was by Atzorn et al. (1988), who demonstrated the presence of GA1, GA4, GA9 and GA20 in gnotobiotic cultures of Rhizobium meliloti.

In Azospirillum sp. several studies have characterized gibberellins by capillary gas chromatography-mass spectrometry (GC-MS), i.e., GA1, GA3, GA9, GA19 and GA20 using gnotobiotic cultures of A. lipoferum (Bottini et al. 1989; Piccoli et al. 1996, 1997) and of GA1 and GA3 from gnotobiotic cultures of A. brasilense (Janzen et al. 1992). The iso-lactone of GA3 has also been identified in these studies, but its presence may be an artifact produced by the analytical purification or GC conditions. Using biological assays the total amount of gibberellins produced in pure cultures of ca. 108 cfu ml−1 ranged from 20 pg ml−1 to 400 pg ml−1 (Bottini et al. 1989; Piccoli et al. 1996). However, in co-culture with other bacteria (Cacciari et al. 1989; Flouri et al. 1995) or with fungi (Janzen et al. 1992) the total amount of gibberellins produced increased ca. 10-fold.

Apart from Azospirillum sp. and Rhizobium sp., production of gibberellin-like substances has also been claimed in numerous bacterial genera, although the techniques used (TLC, bioassays, HPLC-UV) are of poor resolution and/or reliability. Using unequivocal physico-chemical methods, such as GC-MS, production of gibberellins has been confirmed in Acetobacter diazotrophicus, Herbaspirillum seropedicae (Bastián et al. 1998) and Bacillus sp. (Gutiérrez-Mañero et al. 2001) in addition to Azospirillum sp. (see above).

In higher plants gibberellin biosynthesis (Fig. 2; Crozier et al. 2000; Hedden and Phillips 2000) begins with the cyclization of a C20 precursor, geranyl geranyl diphosphate (GGPP). This intermediate is synthesized in plastids starting from isopentenyl diphosphate (IPP) coming either from cytosol-formed mevalonic acid or via the plastid deoxylulose 5-phosphate pathway (Litchtenthaler 1999; Sponsel 2002). In actively growing tissues, cyclization of GGPP yields ent-kaurene (ent-K) in a two-step synthesis that requires two enzymes: copalyl diphosphate synthase (CPS), which yields copalyl diphosphate (CPP), and ent-kaurene synthase (KS), which gives the final product. Subsequently, ent-K is converted into “true” gibberellins by a series of oxidative reactions catalyzed by two types of enzymes. The first type are membrane-related cytochrome P450 monooxygenases (ent-K oxidase, Kox, and ent-kaurenoic acid oxidase, KAox) and lead to formation of the first gibberellin, GA12-aldehyde, which is then converted by a 13-hydroxylase (GA13ox) to GA53-aldehyde or GA53 via GA12. Subsequent metabolism at the C20 stage is accomplished by 2-oxoglutarate-dependent soluble dioxygenases (GA20ox) and 3β-hydroxylases (GA3ox, Fig. 2). A third group of 2-oxoglutarate-dependent dioxygenases, 2β-hydroxylases (GA2ox) hydroxylates the GA molecule at carbon 2, removing its growth-promoting activity (MacMillan 1997; Hedden and Phillips 2000; Sponsel and Hedden 2004).
Fig. 2

Putative biosynthetic pathway for gibberellins in Azospirillum sp. based on well-established steps in vascular plants and fungi, and data available on gibberellin metabolism studies with this bacterium. Gibberellins already characterized by GC-MS as produced by Azospirillum sp are boxed. The conversion of GA9 to GA3 has been demonstrated in vitro with gnotobiotic cultures of the bacterium, while the metabolism of GA20 to GA1 has been established both in vitro and in vivo (i.e., in association with dy mutants of rice)

In fungi, the general pathway is similar to that of higher plants, although the genes and enzymes involved differ. Recently, Tudzynski et al. (2003) completed the cloning of six genes of the gibberellin biosynthesis gene cluster in Gibberella fujikuroi and determined the functions of these genes, thus defining the complete gibberellin biosynthetic pathway in this fungus. Notably, all the enzymes involved are membrane-related cytochrome P450 monooxygenases and none are soluble dioxygenases. These enzymes comprise the gibberellin-specific GGPP synthase (GGS2), ent-K synthase (CPS/KS), and four cytochrome P450 monooxygenase genes (P450-1P450-4) closely linked in a gene cluster (Mende et al. 1997; Linnemannstöns et al. 1999). P450-4 encodes Kox, catalyzing the three oxidation steps between ent-K and ent-KA (Tudzynski et al. 2001), while P450-1 encodes a highly multifunctional monooxygenase, which catalyzes four steps involving oxidation at two carbon atoms, in the main pathway from ent-KA to GA14 via GA12-aldehyde (Rojas et al. 2001). P450-2 was shown to encode a GA20ox, which converts GA14 to GA4 by removal of C-20 (Tudzynski et al. 2002). P450-3, encodes the 13-hydroxylase that converts GA7 to the end product, GA3, while another gene, orf3, encodes the desaturase that converts GA4 to GA7 (Tudzynski et al. 2003).

Thus, although information on the enzymes and their genetic control involved in the metabolism of gibberellins in higher plants and fungi is abundant, evidence for bacterial biosynthesis is scarce. Tully et al. (1998) sequenced a cluster of three complete P450 genes (CYP112, CYP114, and CYP117) in Bradyrhizobium japonicum, plus a partial P450 gene fragment (CYP115P) previously shown to encode a cytochrome P450. Although the biochemical functions of the products of these genes are uncertain, similarities in structure with other genes suggest an operon involved in terpenoid synthesis bearing some resemblance to plant and Gibberella genes for ent-KS.

For Azospirillum sp. it has been demonstrated that the bacterium can metabolize gibberellins in vitro (Piccoli and Bottini 1994a; Piccoli et al. 1996, 1997) as well as in vivo, i.e., in association with a higher plant (Cassán et al 2001a,b). The finding by Piccoli and Bottini (1994b) that the bacteria produce GA9 and GA19 in chemically defined culture may indicate the existence of two branches for the biosynthetic pathway in Azospirillum sp. (Fig. 2), i.e., early 13-hydroxylation involving the metabolism of GA19 (and its metabolite, GA20) to GA1, and an early non-hydroxylation branch where GA9 is (presumably) the precursor of GA3. In fact, when the chemically defined medium was supplemented with deuterium-labeled GA20, putative deuterium-labeled GA1 was identified based on the mass spectrum profile (Piccoli and Bottini 1994a). Consistent with this scheme, the feeding of A. lipoferum cultures with tritium- and deuterium-labeled gibberellins also indicated that GA9 produced GA3, but not GA1, while GA1 (but not GA3) was obtained with GA20 as substrate (Piccoli et al. 1996). The concept of two branches in the gibberellin biosynthetic pathway is also reinforced by the effect of blue light on A. lipoferum cultures, i.e., a 2- to 3-fold increase in the amount of GA3, relative to GA1 (Piccoli and Bottini 1996).

Other environmental factors, such as N supply (Piccoli and Bottini 1994b), O2 availability and osmotic potential (Piccoli et al. 1999), can influence both the quantity and type of gibberellin produced by Azospirillum cultures. High concentrations of NH4Cl reduced the amount of GA3 produced (Piccoli and Bottini 1994b), i.e., like in vascular plants and fungi (Candau et al. 1992), gibberellin synthesis initiates when N availability decreases. The quantity of GA3 produced was also severely reduced by restricted gas exchange. In the presence of PEG as an osmotic agent (Ψw=−1.21 MPa), the total amount of GA3 was reduced only 50% despite a 90% reduction in the number of cells per milliliter of culture medium (Piccoli et al. 1999). This implies a compensatory mechanism in the bacterium’s ability to produce GA3 under drought conditions, which in turn may explain the positive effects of gibberellin produced by endophytic Azospirillum sp. noted in water-stressed maize seedlings (Cohen et al. 2001).

Additionally A. lipoferum grown in vitro in chemically defined medium hydrolyzed both ether and ester glycosides of GA20 that were isotopically labeled with deuterium (Piccoli et al. 1997). This finding was confirmed in vivo using two mutants of rice deficient in gibberellin synthesis. A reversion of the dwarf phenotype to wild-type (tall) was obtained when these dwarf mutants were fed with deutero GA20-glycosides, and liberation of the aglycone deutero GA20 occurred, as did its subsequent 3β-hydroxylation to deutero GA1 (Cassán et al. 2001a,b, 2003). When prohexadione-Ca, a specific inhibitor of the 2β-hydroxylation and 3β-hydroxylation steps, was added to the plant system, no reversion of dwarf phenotype was observed in the rice mutant, nor was deutero GA1 identified by GC-MS. This suggests that the enzymes that mediate the 3β hydroxylation step in both the rice plant and the bacterium correspond to the family of 2-oxoglutarate-dependent dioxygenases (Cassán et al. 2001b). Cassán et al. (2003) also found that inoculation with A. lipoferum and A. brasilense of the dy mutant of rice fed with GA12-aldehyde reversed the dwarf phenotype, while addition of Uniconazole-P (an inhibitor of membrane-related cytochrome P450 monooxygenases) repressed the reversion of the dwarf phenotype. Taken together, these results suggest that, as in higher plants, early steps of the gibberellin biosynthetic pathway in the bacterium may be regulated by membrane-related cytochrome P450 monooxygenases, and the late hydroxylative steps by soluble 2-oxoglutarate-dependent dioxygenases.

The involvement of gibberellin produced by bacteria in plant growth and yield promotion

Several papers (reviewed by Cassán et al. 2003) have claimed that nodules of different Leguminosae species contain more gibberellin-like substances than do adjacent roots, suggesting that the microorganism modifies hormonal levels in the nodules, either by affecting plant cell metabolism or by affecting gibberellin production by the bacterium. For example, Dobert et al. (1992) demonstrated that Phaseolus lunatus plants inoculated with a specific strain of Bradyrhizobium sp. showed a marked internode elongation that was not observed in plants inoculated with other compatible bradyrhizobia. Measurement of gibberellin content using deuterated internal standards, and GC-MS analysis, showed that increased levels of GA1, GA19, GA20, and GA44 in nodules formed by the two bacterial strains that enhanced elongation growth (Dobert et al. 1992). Yanni et al. (2001) noted that indigenous Rhizobium leguminosarum bv. trifoli can colonize rice roots in the Egyptian Nile delta where rice has been rotated with Trifolium alexandrinum L. since antiquity. The Rhizobium-rice combination promotes root and shoot growth, thereby improving seedling vigor and increasing grain yield. Yanni et al. (2001) also found that pure cultures of these Rhizobium strains produced auxin (IAA) and gibberellin (tentatively identified as GA7).

Gibberellin production by Azospirillum sp. and Bacillus sp. has been implicated in the increased 15N uptake seen after inoculation of wheat roots (Kucey 1988). Application of GA3 to the roots, in concentrations similar to those produced by the microorganisms, promoted root growth in maize seedlings, and inoculation with different Azospirillum strains increased levels of GA3 in maize roots (Fulchieri et al. 1993); in contrast, non-inoculated seedlings contained predominantly conjugated GA3. Furthermore, reversal of dwarfism, both genetic and induced by inhibitors of gibberellin biosynthesis, was demonstrated in both rice and maize seedlings that had been inoculated with Azospirillum sp., and showed the endophytic presence of the bacteria (Lucangeli and Bottini 1996, 1997) A reversal of the dwarf phenotype was also obtained in these inoculated dwarf rice mutants fed with deutero GA20-glycosides, and associated with the increased growth was a liberation of the aglycone, deutero GA20, and its 3β-hydroxylated metabolite, deutero GA1 (Cassán et al. 2001a,b, 2003). It was not possible, however, to determine whether the plant growth response to bacterial inoculation was due to bacterial gibberellin production and deconjugation of gibberellin glycosides by enzymes of the microorganism. It should be noted that the inoculated rice seedlings showed a much lesser, and not significant, response to the applied deutero GA20-gycloside.

As noted above, gibberellins are known to interact with other hormones. Cohen et al. (2001) tested the ability of A. lipoferum to alleviate temporary drought in maize seedlings. In this case, Prohexadione-Ca and fluridone were used to block gibberellin and ABA synthesis, respectively. Inhibition of ABA synthesis was detrimental to the plant mainly because stomatal closure was reduced. However, the most harmful situation was obtained under drought stress, when the synthesis of both hormones had been reduced. Prior inoculation with A. lipoferum promoted growth of both roots and shoots under drought, partially reversing the effects of the two biosynthesis inhibitors. Alleviation of water stress symptoms in wheat plants has been reported previously by Creus et al (1997), with the alleviation effects being attributed, at least in part, to gibberellin production by the bacteria.

The effect of inoculation with A. diazotrophicus and of applications of GA3 at several doses on total carbohydrates, sucrose, glucose and fructose was assessed in shoots of Sorghum bicolor (Bastián et al. 1999). Both A. diazotrophicus and application of GA3 were effective in promoting total carbohydrate accumulation, but neither technique yielded an increase in sucrose levels. In contrast, fructose and glucose levels were significantly enhanced by both A. diazotrophicus and GA3 treatments, relative to controls.

Bacillus pumilus and Bacillus licheniformis, isolated from the rhizosphere of Alnus glutinosa L. Gaertn., both have strong growth-promoting activity. Gutiérrez-Mañero et al. (2001) showed that the dwarf phenotype induced in A. glutinosa seedlings by Paclobutrazol (an inhibitor of gibberellin biosynthesis) was effectively reversed by applications of extracts from medium incubated with both bacteria and also by exogenous GA3. GC-MS analysis of extracts of these media showed the presence of GA1, GA3, GA4 and GA20. Probanza et al. (2002) also reported that inoculation with Bacillus licheniformis and B. pumilus enhanced growth of Pinus pinea plants, presumably by bacterial gibberellin production.

In conclusion, the beneficial effect of PGPR on growth and yield (Okon and Labandera-Gonzáles 1994; Cassán et al. 2003) of many crop plants can likely be explained, at least in part, by: (1) gibberellin production by endophytic bacteria (Bottini et al. 1989; Fulchieri et al. 1993; Janzen et al. 1992; Kucey 1988; Lucangeli and Bottini 1997, Piccoli et al. 1999), (2) deconjugation of gibberellin-glucosyl conjugates exuded by the roots, or in the plant (Piccoli et al. 1997), and (3) 3β-hydroxylation by bacterial enzymes of inactive 3-deoxy gibberellins present in roots, to active forms such as GA1, GA3 and GA4 (Piccoli and Bottini 1994a; Piccoli et al. 1996; Cassán et al. 2001a,b).

Copyright information

© Springer-Verlag 2004