[Index Spaink] [IMP]

THE MOLECULAR BASIS OF THE HOST SPECIFICITY OF RHIZOBIUM BACTERIA

HERMAN P. SPAINK

Institute of Molecular Plant Sciences, Clusius Laboratory, Leiden University, Wassenaarseweg 64, 2333 AL Leiden, The Netherlands

CONTENTS

ABSTRACT

The interaction between soil bacteria belonging to the genera Rhizobium, Bradyrhizobium and Azorhizobium and leguminous plants results in the induction of a new plant organ, the root nodule. After invading these root nodules via infection threads the bacteria start to fix atmospheric nitrogen into ammonia which is beneficial for the host plant. This symbiotic interaction is highly host-specific in that each rhizobial strain is able to associate with only a limited number of host plant species. The subject of this presentation is the molecular mechanism by which the bacterium determines its host-specific characteristics. This mechanism appears to be based on at least two stages of molecular signaling between the bacterium and the plant host. In the first stage, flavonoids secreted by the plant root induce, in a host specific way, the transcription of bacterial genes which are involved in nodulation, the so-called nod genes. This leads to the second step of the signaling system: the production and secretion of lipo-oligosaccharide molecules by the Rhizobium bacteria. These signal molecules, which are acylated forms of small fragments of chitin, have various discernable effects on the roots of the host plants. One of these effects is the dedifferentiation of groups of cells located in the cortex which leads to the formation of nodule meristems. In their mitogenic activity the bacterial signals resemble several well-known plant hormones like auxins and cytokinins. However, there are two major differences:
The nod genes determine this stage of host specificity by their essential role in the biosynthesis of the signal molecules. They appear to encode enzymes which are involved in the processes of fatty acid biosynthesis, fatty acid transfer, chitin synthesis and chitin modification. I will illustrate the statement that the nod gene products are ideal model enzymes for the study of these important processes because they are not needed in the free-living state of the bacteria.

INTRODUCTION

Bacteria belonging to the genera Rhizobium, Bradyrhizobium and Azorhizobium, collectively called rhizobia, are able to invade the roots (or adventitious roots) of their leguminous host plants where they trigger the formation of a new organ called the root nodule (reviewed in Spaink, 1989). In these root nodules a differentiated form of the rhizobia, the bacteroid, is able to fix nitrogen into ammonia, which can then be utilized by the plant. The host-specific aspect of this symbiosis is very pronounced and has led to the definition of cross-inoculation groups in which the bacterial species are classified according to their group of host plants. Examples of such cross-inoculation groups are R.leguminosarum biovar viciae with peas, vetches, lentils and sweet peas as hosts, R.leguminosarum biovar trifolii with clovers as hosts and R.meliloti with alfalfa and sweet clovers as hosts. At least two steps of signal exchange between plant and bacteria appear to be the basis of the determination of host-specific nodulation (Fig.1). In the first step, flavonoids excreted by the plant induce the transcription of the bacterial nodulation genes (called nod or nol genes) (Firmin et al. 1986; Peters et al. 1986; Redmond et al. 1986; Zaat et al. 1987) (Fig. 2). The host specificity of this induction process involves the bacterial NodD protein, a transcriptional regulatory protein that presumably directly interacts with the flavonoids (Spaink et al. 1987, Horvath et al. 1987). In the second step, the bacterium, by means of the nod or nol genes, produces lipo-oligosaccharide signals (Fig. 3, Table 1) that can induce the various root responses which are listed in Table 2. The response which is most obviously relevant for the establishment of the rhizobium-plant symbiosis is the induction of the nodule primordium. It has been shown that purified lipo-oligosaccharides, externally applied to the plant in nanomolar concentrations, are able to induce nodule primordia which are indistinguishable from those observed in the first stage of normal nodule organogenesis (Spaink et al., 1991; Truchet et al. 1991; Van Brussel et al. 1992). The common nodulation genes nodABC are essential for the biosynthesis of the lipo-oligosaccharides (also called Nod metabolites), consistent with their pivotal role in the infection and nodulation process. Several nod genes which were shown to be involved in the determination of the host specificity of nodulation also appear to be involved in the production of lipo-oligosaccharide signals. Most noteworthy is the gene nodE, which was shown to be the major determinant of the difference between the host specificities of the R.leguminosarum biovars viciae and trifolii (Spaink et al. 1989) and the gene nodH, which is a major determinant of host specificity in R.meliloti (Horvath et al. 1986; Roche et al. 1991; Dénarié and Cullimore 1993). In this presentation I will highlight some results of my study of the molecular basis of host specificity of Rhizobium bacteria which are the basis for the model presented in Fig.1.

I. TRANSCRIPTIONAL REGULATION OF THE NOD GENES AS THE FIRST DETERMINANT OF HOST SPECIFICITY

I.1. PROMOTERS IN THE NODULATION REGION OF THE RHIZOBIUM LEGUMINOSARUM SYM PLASMID PRL1JI

A region of 16.8 kb of the Sym(biosis) plasmid PRL1JI of R. leguminosarum biovar viciae, consisting of 9.7 kb nodulation region which confers nodulation ability on Vicia hirsuta and a region of 7.1 kb which appeared to be necessary for nodulation on V.sativa and Trifolium subterraneum, was subcloned as fragments of maximally 2.5 kb in a newly developed IncQ transcriptional fusion vector (Spaink et al. 1987a). The expression of these fragments was studied in Rhizobium. One constitutive promoter, pr.nodD, and three plant-exudate inducible promoters were found, namely pr.nodA, pr.nodF and pr. nodM. The latter promoters regulate the transcription of the operons nodABCIJ , nodFE and of an operon of at least 2.5 kb which contains the nodMNT genes (Canter Cremers et al. 1989) (Fig.2). Induction of the three inducible operons required plant exudate and a functional nodD product. The flavanone naringenin could replace plant exudate. Each of the three inducible promoters contained a nod-box. A consensus for the nod-box sequence, based on known sequences, was proposed (Fig.4). This consensus is also based on the promoter region of the nodO gene which is localized outside the originally cloned 16.8 kb fragment (De Maagd et al. 1989a; De Maagd et al. 1989b) (Fig.2). A 114 bp fragment which contains pr.nodA activity was used to localize pr.nodA by means of deletion mapping. The fragment which appeared necessary for complete pr.nodA activity is 72 bp in size, contains the complete nod-box and in addition a region of 21 bp downstream of the nod-box, in which the loosely conserved sequence AT(T)AG appears to be important for promoter activity (Fig.5) (Spaink et al. 1987a).

I.2. RHIZOBIUM NODULATION GENE nodD AS A DETERMINANT OF HOST SPECIFICITY

The function of the nodD product in the recognition of the plant flavonoid inducer of the nod genes is unknown. It is striking that, although nodD is considered to be a common nod gene, the structural requirements for inducers of nod genes of R.leguminosarum biovar viciae (Zaat et al. et al. 1987) R.leguminosarum biovar trifolii (Redmond et al. 1986) and R.meliloti (Peters et al. 1986) seem to be different. Therefore, the nodD genes of the fast-growing Rhizobium species R.leguminosarum biovar viciae, R.leguminosarum biovar trifolii and R.meliloti (the nodD1 gene) were cloned in a broad host range IncP class vector and the resulting plasmids pMP280, pMP283 and pMP284 were subsequently transferred to Rhizobium strain LPR5045 which lack a Sym plasmid (Spaink et al. 1987b). The inserts contain a complete nodD gene, including their own constitutively expressed promoter and some surrounding sequences, but no other complete nod gene. To measure the expression of inducible nod promoters, plasmid pMP154 was used, an IncQ transcriptional fusion vector in which a 114-base pair (bp) fragment containing the promoter region of the nodABCIJ operon of the R.leguminosarum biovar viciae Sym plasmid PRL1JI was cloned upstream of lacZ. In PMP154, nod promoter activity could be detected as ß-galactosidase activity. Plasmid pMP154 was mobilized to the Rhizobium LPR5045 derivatives harbouring the nodD-containing plasmids, yielding an isogenic set of strains which differ only in the source of the nodD genes. The results of the induction experiments (Table 3) show that luteolin can induce the nod promoter to a substantial level regardless of the nodD source. For the other flavonoids the source of nodD determines which substances efficiently activate the inducible nod promoter. The strains containing the nodD gene of R.leguminosarum biovar viciae or biovar trifolii can be most clearly distinguished by their reactions with eriodictyol and 7-hydroxy-flavone in that eriodictyol is the better inducer in the presence of R.leguminosarum biovar viciae nodD (ratio eriodictyol/7-hydroxyflavone is 2.8) whereas 7-hydroxyflavone is a much better inducer in the presence of R.leguminosarum biovar trifolii nodD (ratio eriodictyol/7-hydroxyflavone is 0.16). In contrast, neither these two substances nor naringenin were efficient inducers for the strain containing the R.meliloti nodD gene (Table 3). To see whether this specificity could also be reflected in nature, exudates of Melilotus alba, Pisum sativum, Vicia hirsuta, Trifolium repens and T.pratense were tested for their ability to induce the nod promoter of pMP154 in the presence of the various nodD clones. The former four exudates induced in each nodD strain the nod promoter to at least 50% of the induction level of luteolin. In contrast, exudate of T.pratense (red clover) was only a good inducer in the presence of the nodD gene of R.leguminosarum biovar trifolii (Table 3). It was also shown that the absence of induction with red clover exudate was not due to inhibitors. Because the control with T.repens (white clover) exudate shows that induction with clover exudate is possible in all cases, these results confirm that the nodD genes of the tested plasmids have cross-inoculation group specific characteristics. The results with the T.pratense exudate cast doubt on the original assumption that nodD is a common nod gene. We therefore investigated whether the nodD genes of either R.leguminosarum biovar viciae or R.meliloti can function in the nodulation of red clover. Therefore the plasmids with the different nodD genes were mobilized into R.leguminosarum biovar trifolii strain ANU851 in which the original nodD gene is inactivated by a transposon insertion. When the resulting strains were tested for nodulation ability on red and white clover, it appeared that the nodD genes of R.leguminosarum biovar viciae and R.meliloti are only able to complement the mutation of strain ANU851 for nodulation on white clover (and the R.meliloti nodD not completely) whereas the (control) nodD gene of R.leguminosarum biovar trifolii can complement for nodulation on both Trifolium species (Table 4). From these results it can be concluded that the host range of R.leguminosarum biovar trifolii is narrowed when its own nodD gene is replaced by that of R.leguminosarum biovar viciae or R.meliloti and that therefore nodD cannot be considered as a common nod gene despite the fact that the nodD products of R.leguminosarum biovars viciae and trifolii and R.meliloti have a homology with each other of at least 75% (Spaink et al. 1987b).

I.3. LOCALIZATION OF FUNCTIONAL REGIONS OF THE RHIZOBIUM nodD PRODUCT USING HYBRID nodD GENES

The nodD genes are functionally different in (i) their response to various exogenously added flavonoid inducers, (ii) the extent to which they mediate the activation of the flavonoid-inducible promoters, and (iii) the extent to which they repress their own constitutive transcription. In order to localize the regions of the nodD product which determine these differences, two series of nodD hybrid genes have been constructed using the method which is described in Fig.6 (Spaink et al. 1989c) In one series the 5' moiety is derived from the R.meliloti nodD1 gene and the 3' moiety from the R.leguminosarum biovar trifolii nodD gene. In the other series, the origins of the nodD moieties are reversed. Two regions of the nodD product appeared to be involved in autoregula- tion (Fig.7). Many regions, dispersed over the entire nodD product, are involved in the specificity of activation by flavonoids. Several hybrid nodD genes were characterized which activate trans- cription with novel inducers. Furthermore, two classes of hybrid nodD genes were found from which the activation characteristics differ completely from those of the parental nodD genes. The first class activates the nodABCIJ promoter to the maximum level in the absence of flavonoid in- ducer. This level can no longer be influenced, positively or negatively, by the presence of (iso-)flavonoids. With the second class of hybrids, activation of the nodABCIJ promoter, even in the presence of flavonoid inducers, is no longer possible. The results from these experiments are summarized in Fig.7. The presumption of Horvath et al. (1987) that the specificity of exudate recognition is only determined by the carboxy-terminal region of nodD is consistent with the high conservation of the amino-terminal region of the nodD product. In strong contrast, the results summarized in Fig. 7 show that both the carboxy-terminal region and regions located in the amino-terminal part, upstream of amino acid 90, are involved in this specificity. The results therefore show that there are many regions in the nodD product which are involved in the specificity of activation by flavonoids and that the regions determined to be involved in autoregul- ation are not clearly separated from regions involved in the flavonoid specificity (Spaink et al.1989c).

I.4. SYMBIOTIC PROPERTIES OF RHIZOBIA CONTAINING A FLAVONOID-INDEPENDENT HYBRID NODD GENE.

A hybrid nodD gene consisting of 75 percent of the nodD1 gene of R.meliloti at the 5' end and 27 percent of the nodD gene of R.leguminosarum biovar trifolii at the 3' end (Fig.8) activates the six tested inducible nod promoters of R.leguminosarum biovars viciae and trifolii and R.meliloti to maximal levels, even in the absence of flavonoids (Spaink et al. 1989a). In strains containing such a FITA (for Flavonoid Independent Transcription Activation) nodD gene, transcription of nod genes started at the same site as in induced strains containing a wild type nodD gene. In contrast to heterologous wild type nodD products, the FITA nodD gene does not cause a limitation of the host range (Table 4). Furthermore, R.leguminosarum biovars viciae and trifolii and R.meliloti strains containing the FITA nodD604 gene induce (pseudo)nodules on tropical leguminous plants. Comparison of the symbiotic properties of rhizobia containing a FITA nodD gene with those of rhizobia containing various wild type nodD genes indicates that the activation of the nodD product is of crucial importance during the process of infection thread formation and, surprisingly, also during nitrogen fixation. The observations that (i) both the level of nitrogen fixation and the number and morphology of the bacteroids depend on the source of the nodD gene, that (ii) a FITA nodD gene confers superior nitrogen fixation abilities and that (iii) this difference in nitrogen fixation was not caused by a difference in nodule number or nodulation kinetics, suggests that the presence of an activated nodD product is also of crucial importance during later stages of symbiosis. In conclusion, the nodD gene, in addition to its essential role in the early stages of the symbiosis, is also an important factor which determines the level of nitrogen fixation (Spaink et al. 1989a). Since Rhizobium strains containing a FITA nodD gene are apparently no longer limited in host range by the flavonoids produced by the host plants they are very suitable for studying other factors which limit the host range of Rhizobium or which limit the bacteroid development in non-host plants. Furthermore, the demonstrated advantages of the nodD604 gene for nitrogen fixation, if they hold under field conditions, could be of great practical importance (Spaink et al. 1988).

I.5. DISCUSSION

The initial approach of this study was to analyze the regulation of the promoters in the nodulation region of the R.leguminosarum Sym plasmid PRL1JI. From this study it was concluded that the transcription of all operons located in this region, except nodD, are induced by compounds which are exuded by the roots of the host plant. These compounds could be replaced by various commercially available flavones or flavanones. The nodD product appeared to be of crucial importance in this process of nod gene induction. This has led to a general model which essentially describes the NodD protein as a classical positive transcriptional regulator (Spaink 1989). Indeed the NodD protein is homologous to a large group of prokaryotic regulatory proteins like LysR, IlvY and CysB of E.coli, AmpR of Enterobacter cloacae, MetR of Salmonella typhimurium and NahR of Pseudomonas putida especially in their presumed DNA binding domain which is located near the N-terminus (Henikoff et al., 1988). As demonstrated in this paper, the interaction between flavonoid inducer and the NodD protein involves many regions of the NodD protein which are dispersed over the entire protein. These regions surround the predicted DNA binding region of NodD located at position 23 to 42 in the protein (Henikoff et al., 1988). Note that this DNA binding region is not the same as the region which we showed to be involved in the autoregulation by NodD. In conclusion, the relationship between the structure of the NodD protein and its known functions, i.e. activation of transcription, autoregulation and flavonoid recognition, is very complex. Furthermore, it is known that the nodD gene is not able to activate the inducible nod promoters in E.coli , P.putida , P.savastanoi or Xanthomonas campestris (Long 1989; Spaink 1989) indicating that another specific factor is also involved in the nod gene induction process. Such a factor, possibly a specific transcription factor, is common for Rhizobium, Azorhizobium, Bradyrhizobium, Agrobacterium and Phyllobacterium (Spaink 1989). Recently, evidence has been presented by Recourt et al. (1989) that the flavonoid inducer molecules are accumulated in the inner membrane of the Rhizobium cell. Furthermore, the NodD protein binds constitutively to the nod box sequences (see Long, 1989) which are the highly conserved sequences upstream of the inducible nod genes described in Fig.4. Therefore, an important aspect of the signal transduction process mediated by the NodD protein remains to be discussed: how can the NodD protein which is constitutively bound to the nod-box DNA sequence in the cytoplasm be activated by the inducer molecules present in the inner membrane? A solution for this controversy was given by Schlaman et al. (1989) who showed that NodD protein is localized exclusively in the inner membrane of the bacterium. On basis of these results a model is proposed which is given in Fig. 9. In this model NodD is an amphitropic protein with domains capable of interacting with the flavonoid compounds present in the cytoplasmic membrane. A postulated cytoplasmic domain is constitutively bound to the nod-box DNA. We suggest that, by binding to specific flavonoids present in the inner membrane, the NodD protein undergoes a conformational change resulting in an activated form of the NodD protein which is capable of activating transcription by an unknown mechanism. A further confirmation of this model comes from the observation that a NodD FITA protein, which activates transcription of the nod genes independently of added flavonoids is also localized exclusively in the inner membrane of the cell (Helmi Schlaman pers. comm.). For two reasons flavonoids do not seem likely candidates for functioning as a host-specific signal towards Rhizobium. Firstly, flavonoids are not unique for the roots of leguminous plants. On the contrary, they are present in several organs of a wide range of plant species (see Harborne, 1967). Secondly, it has been shown that several leguminous plants, like T.repens and V.sativa, exude flavonoid compounds which can also activate NodD proteins of rhizobia which are not able to nodulate these host plants (Zaat et al., 1988). Nevertheless, as described in this paper, recognition of specific flavonoids by NodD protein is an important basis of host-specificity. As shown by the results given in table 4, the nodD products of the fast-growing R.leguminosarum biovars viciae and trifolii appear to be adapted to the various plants of their cross-inoculation group, in that only the endogenous nodD genes are able to function in the nodulation of all plant species of the respective cross-inoculation groups. This adaptation suggests that the roots of plants belonging to the same cross-inoculation group have in common that they contain one or more flavonoids with a specific structural feature which is only recognized by the homologous NodD proteins. This hypothesis is in agreement with the results of Zaat et al. (1988) who show that the exudates of several host plants of R.leguminosarum biovar viciae contain inducers which are exclusively recognized by the NodD protein of R.leguminosarum. biovar viciae. Confirming the importance of flavonoids as host-specific signals is the demonstration that the fast-growing Rhizobium strains described under section I.4, which contain a FITA nodD gene, are able to nodulate several tropical leguminous plants. However, it is not clear whether the inability of the original Rhizobium strains to nodulate the tropical legumes is due to the absence of an inducing signal or is due to the presence of anti-inducing compounds. The latter explanation is favoured by the identification of isoflavones, which are known anti-inducers, as the natural inducers of the nod genes of the Bradyrhizobium species which nodulate the tested tropical legumes (see Long 1989). It was also shown that activation of the NodD product is of crucial importance during the process of infection thread formation and, surprisingly, is also important during bacteroid development and nitrogen fixation (Spaink 1989a). The latter observation indicates that the control of gene regulation by NodD can be considered as a means of host recognition during the entire symbiosis.
Herman Spaink, Last Update: Wednesday, June 26, 1996 10:24:59 AM