G. Stacey, J. Loh, B. Zhang, Y.-H. Lee, C. Bickley, D. Lohar, G. Liao, G. Copley,
Center for Legume Research, Department of Microbiology, University of Tennessee, Knoxville, TN, 37996-0845, USA
1. Regulation of the Bradyrhizobium japonicum Nodulation Genes
The regulation of nod gene expression in several (Sino)Rhizobium! Bradyrhizobium! Azorhizobium species has been extensively studied. Current evidence shows that most nod genes are involved in the production of substituted lipo-chitin Nod signals that induce root hair curling and cortical cell division. The generic model for the regulation of these genes predicts that the nodD gene product is constitutively expressed and interacts with plant-produced flavonoid chemicals (e.g. isoflavones) leading to induction of other nodulation genes. This system of nod gene transcriptional control is essential for initiation of the infection process and for determining host specificity.
Continuing research in our laboratory has shown that the generic model of nod gene regulation defined in (Sino)Rhizobium species does not adequately account for the control of nod gene transcription in B. japonicum. Indeed, nod gene regulation in B. japonicum shows surprising complexity (Figure 1). Our data show that B. japonicum uses members of three different global regulatory protein families to control nod gene expression. Specifically, a LysR-type regulator,
NodDl, and a two-component regulatory system, NodVW, positively regulate nodYABC gene expression in response to plant produced isoflavone signals (Loh et al. 1997, 1999, 2001; Loh, Stacey 2001). In addition, NolA, a member of the MerR-type regulatory family (see below), acts indirectly to repress nod gene expression (Dockendorff et al. 1994; Garcia et al. 1996). The no I A gene is a rare case in bacteria where a single gene encodes three, distinct polypeptides (Loh et al. 1999). The largest protein, NolAl, is a transcriptional regulatory protein that is required to induce nolA2,3 as well as nodD2. The smallest of the NolA proteins, NolA3, appears to have an important, although undefined function, in nodulation. NodD2 is a repressor of nod gene expression. Thus, expression of NolA 1 ultimately leads to a down-regulation of nod gene expression.
Recent data from our laboratory, described below, show that NolA is a central regulatory component that allows the cell to respond to a variety of signal molecules and finely regulate nod gene expression. Figure 1 presents our current working model for the regulation of the nod genes of B. japonicum. The net effect of this complex regulatory circuitry is to provide the organism with responsive mechanisms for fine regulation of nod gene expression and to integrate such regulation into the overall metabolism of the cell.
B. japonicum possesses two nodD genes arranged tandemly as nodDxnodDj. NodD2 is a repressor of nod gene expression in B. japonicum (Garcia et al. 1996), Bradyrhizobium sp. (Arachis) NC92 (Gillette and Elkan 1996) and Sinorhizobium sp. strain NGR234 (Fellay et al. 1998). Since NolAl positively activates NodD2 expression, regulation of NolA expression effectively controls repression of the nod genes. Hence, NolA is a central, negative regulator of nodulation.
During the course of experiments designed to measure nod-lacZ expression, we made the observation that a nodY-lacZ fusion consistently showed higher activity when assayed in a NodC mutant. NodC is a chitin synthase involved in synthesizing the chitin backbone of the lipo-chitin Nod signal. These data, along with other observations, led us to postulate that Nod signals could act to feedback regulate Nod signal synthesis through the NolA-NodD2 repression pathway. As a first test of this hypothesis, we tested various chitin oligomers (d.p.-l-8) for their ability to induce nolA-lacZ and nodD2-lacZ expression. The data obtained indicated that chitin oligomers do induce nolA expression, with the chitin tetramer being the most active (Loh and Stacey 2001). The chitin tetramer was active as a nolA inducer at concentrations as low as 1 nM, with an optimal activity at 10 nM. Further experiments showed that the effect of chitin was specific for NolAl expression (data not shown).
Different rhizobia produce NodC proteins that determine the chain length of the Nod signal. For example, the NodC from S. meliloti synthesizes primarily a chitin tetramer. In contrast, the NodC from S. sp. strain NGR234 produces primarily a chitin pentamer, while B. japonicum produces both chitin tetramer and pentamer. As a means to increase the level of chitin produced intracellularly, we expressed the S. meliloti, S. sp. NGR234, and the B. japonicum nodC genes in B. japonicum under the control of the constitutive trp promoter. Transconjugants expressing either the S. meliloti or B. japonicum NodC showed a 4-5 fold lower level of nodC-lacZ expression than a strain transformed with vector alone. In contrast, expression of the S. sp. NGR234 NodC protein did not affect nodC-lacZ induction by genistein (data not shown). These data are consistent with the previous results showing that a chitin tetramer is the most active inducer of NolA. Indeed, constitutive expression of the S. meliloti NodC had no effect on nodC expression when assayed in a NolA mutant background (data not shown). Addition of the Nod signal (an acylated, fucosylated pentamer) at concentrations as high as 1 |iM to wild-type cultures did not induce nolA expression. However, addition of synthetic tetrameric LCOs (kindly provided by T. Ogawa, Riken Institute, Japan) activated NolA expression and repressed nodC-lacZ expression. This result was found regardless of chemical substitutions of the chitin backbone.
The above experiments suggest that NolA mediates feedback inhibition of nod gene expression in response to the level of intracellular chitin tetramer. Although the major Nod signal produced by B. japonicum is a substituted, chitin pentamer, our previous work documented the production of tetrameric Nod signals by this bacterium (Cohn et al. 1999).
A puzzling aspect of our studies on nod gene regulation was the requirement that cells be induced (with genistein or S SE) at low cell density in order to see maximal nod-lacZ expression. This phenomenon suggested that nod gene expression could be regulated in a population density-dependent manner, being repressed at high cell density. The discovery of the negative regulation system of NolA and NodD2 prompted us to look more closely at this question. As shown diagrammatically in Figure 1, the inducibility (fold-induction) of a nodY-lacZ fusion drops in response to the production of a quorum signal released from B. japonicum cells (Loh et al. 2001). Population density does not affect nodY-lacZ expression in a NolA mutant background. Thus, NolA appears to be directly involved in the repression of nod gene expression at high population density. Indeed, other experiments using strains with either a nolA-lacZ or nodD2-lacZ fusion showed that the level of these proteins increases with increasing cell density (data not shown)
Filtered, culture supernatants contain an inducer of nolA-lacZ which we have termed CDF, cell density factor (Loh et al. 2001). In most gram-negative bacteria such quorum sensing compounds are modified homoserine lactones. However, repeated attempts using a variety of methods failed to isolate a homoserine lactone from B. japonicum cultures. We have now succeeded in purifying small amounts of CDF, which exhibited the ability to induce a nolA-lacZ fusion. The purified CDF, as expected, also has the ability to repress nod gene expression. The chemical identity of this compound is currently under investigation.
A significant commercial industry exists for the production of B. japonicum inoculants for agricultural use. The companies that produce these inoculants routinely grow their cultures to high cell densities before eventually packaging the cells, together with a carrier (e.g. peat). A common problem with such commercial inoculants is that they compete very poorly for nodule occupancy against indigenous, soil B. japonicum strains. This is referred to as the "competition problem". Indeed, in some cases, the B. japonicum strain in the commercial inoculant may occupy only 0 to 1-2% of the nodules formed.
Upon the discovery of the CDF, its ability to induce NolA synthesis and, concomitantly, repress nod gene expression, we obtained several batches of commercial inoculant. The inoculants and an uninoculated peat (control) were extracted and the extract was tested for its ability to induce nolA-lacZ expression. Significant inducer activity was found in all of the inoculants tested (Loh et al. 2001). Controls of uninoculated peat showed little or no inducer activity. The conclusions from this study are that commercial soybean inoculant contains significant levels of a compound that represses nod gene expression and, therefore, likely reduces the efficacy of the inoculant strain. In order for the inoculant bacterium to nodulate the plant, it will have to initiate growth and overcome the action of the CDF. This delay could be a significant factor in the inability of the inoculant strain to compete against indigenous, soil rhizobia that are commonly present as low population densities (e.g. 102"5 cfug"1).
It has been known for some time that, although nod gene expression is required for the earliest stages of nodulation, the expression of these genes is repressed within the nodule. Various authors have speculated about what may be shutting off nod gene expression in planta. Recently, we obtained convincing evidence that NolAl, likely through its role in quorum sensing, is required to repress in planta expression of nodY. We expressed a nodY-GUS fusion in wild-type B. japonicum and the NolA mutant, BjB3. Soybean plants Were inoculated with both strains, nodules harvested 21 days later and stained for GUS activity. GUS activity was only detected in nodules formed by the NolA mutant strain. No nodY expression was seen in nodules formed by the wild type. Compared to broth cultures, B. japonicum does not attain a comparable cell density within the nodule. However, in planta each bacterium is enclosed in a symbiosome with little space between the bacterium and symbiosome membrane. Therefore, even low synthesis of the quorum signal in planta could translate into a very high, localized concentration in the symbiosome.
The known lipo-chitin Nod signals produced by rhizobia are substituted chitin oligomers, usually of four to five iV-acetylglucosamine (GlcNAc) residues, mono-TV-acylated at the non-reducing end and carrying a variety of substitutions at both the reducing and non-reducing terminal GlcNAc residues. Each rhizobial species produces a variety of Nod signals with specific substitutions.
Our laboratory extensively studied the chemical specificity required for Nod signal action on soybean. In summary, these data showed that the presence of a 2-O-methylfucose residue on the terminal, reducing GlcNAc was critical for biological activity on soybean only when the LCO Qipo-chitooligosaccharide) was a pentamer (reviewed in Cohn et al. 1997). If the LCO was a tetramer, then fucosylation rendered the molecule inactive at eliciting root hair curling or cortical cell division in soybean. Hence, both the chemical substitutions and the LCO chain length were critical. Moreover, our results support the hypothesis that Nod signal perception in soybean involves at least two discrete recognition events with differing chemical specificity. For example, we found that a mixture of non-fucosylated Nod signals could surmount the requirement for a fucosylated Nod signal with regard to rice bean (Vigna umbellata) nodulation (Cohn et al. 1999). Other work examined nodulin gene expression in response to Nod signal addition. These studies showed that two, chemically distinct recognition events are involved in Nod signal action. One signaling step is somewhat non-specific in that even chitin oligomers (e.g. pentamer) would induce transient expression of the early nodulins. However, sustained expression of ENOD40 required a soybean-specific Nod signal. The existence of two recognition events was supported by finding that expression of the early nodulin ENOD2 could not be induced by the addition of any single LCO or Nod signal. ENOD2 expression was clearly induced only when a mixture of LCO was added. One member of this mixture could be a simple, chitin oligomer, if the other member was a soybean-specific Nod signal.
Currently, a bona fide Nod signal receptor has not been identified. Etzler et al. (1999) reported a unique lectin (DB46) isolated from the roots of the legume Dolichos biflorus. This lectin was found to bind to Nod signals from a variety of rhizobia. Etzler et al. (1999) demonstrated that the lectin possessed ATPase activity (i.e. apyrase activity), which was significantly increased upon addition of the Nod signal. For this reason, the lectin was termed a lectin-nucleotide phosphohydrolase (LNP). The D. biflorus LNP (i.e. DB46) was found on the surface of root hairs using fluorescent antibody labeling. Antibody directly against LNP blocked nodulation. Recently, we extended these studies by demonstrating the presence of orthologs of the D. biflorus apyrase (LNP) in other legumes [e.g. GS50, GS52 in soybean (Day et al. 2000) and Mtapyl-5 in M. truncatula (Cohn et al. 2001)]. We showed that GS52 in soybean and Mtapyl and Mtapy4 from M. truncatula are early nodulins, induced within 3 h by rhizobial inoculation. Moreover, antibody against GS52 blocked soybean nodulation. M. truncatula mutants defective in very early nodulation events also were defective in Mtapyl and Mtapy4 expression. Therefore, legume apyrases must be considered as candidates for a Nod signal receptor. The Cohn et al. (2001) and Day et al. (2000) studies showed the benefit of analyzing multiple legumes. Isolation of both the soybean and M. truncatula apyrases allowed the identification of a region of microsynteny between the two genomes containing a cluster of apyrase genes.
Chitin oligomers, which can be generated from fungal cell walls by endochitinase, can induce defense responses or related cellular responses in many monocots and some dicots. Our previous work on the specificity of Nod signal action (Cohn et al. 1997) suggested that a chitin binding protein could be involved in Nod signal recognition (i.e. as measured by ENOD40 induction). Therefore, Day et al. (2001) used 125I-labeled APEA (aminophenyl ethylamine)
conjugates of ¿V-acetylchitooctaose and iV-acetylchitopentaose as ligands to identify a chitin-binding site in microsomal membrane preparations from both soybean suspension cultured cells, as well as root preparations. Binding to this site was saturable with an apparently Kd of approximately 40 nM. Competition experiments using chitin oligomers (d.p.=2-8) demonstrated that this binding site preferred the higher molecular weight oligomers (d.p.=7-8). Affinity labeling using a 1251-labeled yV-acetylchitooctoase ligand identified an 85 kDa chitin-binding protein in the plasma membrane. The binding specificity of this 85 kDa protein for various chitin oligomers correlated with the ability of the same oligomers to induce an oxidative burst response and medium alkalinization in soybean suspension cultured cells. Treatment of soybean suspension cells with the pentameric, B. japonicum Nod signal also resulted in medium alkalinization. The response was similar to that shown by treatment with chitin pentamer or tetramer. Indeed, Nod signal binding to soybean plasma membrane preparations (as measured by competition against labeled ligand) also showed an affinity approximately that of the chitin tetramer. From these data, we concluded that the 85 kDa, chitin-binding protein was likely involved in the soybean defense response mediated by chitin fragments released from fungal pathogens. Although this protein appears to bind to the Nod signal, it does so at low affinity and is likely not involved in Nod signal action. However, Nod signal does act as a significant elicitor of defense-related responses in soybean. Therefore, the possibility arises that Nod signal elicitation of a defense response could affect nodulation.
To examine this possibility, we recently constructed transgenic Lotus japonicus plants expressing the bacterial NahG protein. This enzyme catalyzes the breakdown of salicylic acid (SA) to catechol. SA is a known mediator of induced defense responses in plants. Previous research showed that transgenic nahG plants were unable to accumulate SA and showed a reduced resistance to pathogen attack. Our preliminary data indicate that the L. japonicus nahG transgenic plants accumulated significantly less SA in their roots and showed an approximately 50% increase in nodulation. These results are consistent with a role for SA-mediated defense mechanisms in controlling legume infection by rhizobia.
Was this article helpful?