Rhizobium meliloti mutants that fail to succinylate their Calcofluor-binding exopolysaccharide are defective in nodule invasion

Cell, Vol. 51, 579-587,

November

20, 1987, Copyright

0 1987 by Cell Press

F. Hanks,*

Isolation of exoH Mutants We had previously used agar medium containing Calcofluor to screen for transposon TnS-induced mutants of R. meliloti that formed nonfluorescent colonies because

Results

plasmid of R. meliloti (Finan et al., 1986; Hynes et al., 1986). Alfalfa seedlings inoculated with these exo mutants formed ineffective nodules that contained few if any bacteria and none of the morphologically differentiated forms termed bacteroids. More detailed characterizations of nodules elicited by an exoB mutant revealed that no infection threads are formed following inoculation with this mutant and that plant cells in the interior of nodules elicited by this strain are empty and do not contain bacteriaor bacteroids (Finan et al., 1985). Similar findings implicating Rhizobium exopolysaccharides in normal nodule development have been reported for the nodulation of clover by R. trifolii (Chakravorty et al., 1982), the nodulation of Leucaena by a Rhizobium sp. strain NGR234 (Chen et al., 1985), and the nodulation of pea by R. leguminosarum (Borthakur et al., 1986). The R. meliloti 8x0 mutants described above uncouple the ability of the bacteria to signal the plant to initiate nodule formation from the ability of the bacteria to invade the nodule and establish an effective symbiosis. The signaling of the plant requires nod gene products (reviewed in Downie and Johnston, 1986) and involves action at a distance, perhaps by means of a diffusible substance or by some type of signal propagation mechanism. The role(s) of exopolysaccharides in nodule development has not been established, although it seems clear that they are not required for the primary recognition of the correct plant host by the bacterium (Leigh et al., 1985; McNeil et al., 1986). Djordjevic et al. (1987) have reported that it is possible to suppress the symbiotic phenotypes of exo mutants of Rhizobium sp. strain NGR234 and R. trifolii strain ANU843 by the addition of exogenous exopolysaccharide purified from the homologous wild-type strain. In contrast, we have been unable to suppress the symbiotic deficiencies of R. meliloti exo mutants by the addition of purified exopolysaccharide isolated from the exe+ parent Rm1021 suggesting that the polysaccharide may play a different or an additional role(s) in the case of R. meliloti. The structure of the acidic Calcofluor-binding exopolysaccharide of R. meliloti strain Rm1021 has been determined (Aman et al., 1981), and consists of a polymer of an octasaccharide. Each octasaccharide consists of a backbone of three glucoses and one galactose, a side chain of four glucoses, and 1-carboxyethylidene (pyruvate), acetyl, and succinyl modifications in a ratio of approximately 1:l:l. It is not clear what structural elements of this complex macromolecule are crucial for its role(s) in nodule development. In this paper, we report the isolation and characterization of R. meliloti mutants that fail to succinylate their Calcofluor-binding exopolysaccharide and form empty ineffective nodules on alfalfa.

Rhizobium meliloti Mutants That Fail to Succinylate Their Calcofluor-Binding Exopolysaccharide Are Defective in Nodule Invasion John A. Leigh,’ Jason W. Reed,7 Joanna Ann M. Hirsch,* and Graham C. Walker* * Department of Microbiology University of Washington Seattle, Washington 98195 t Biology Department Massachusetts Institute of Technology Cambridge, Massachusetts 02139 *Department of Biological Sciences Wellesley College Wellesley, Massachusetts 02181

Summary We have identified a set of TM-generated mutants of Rhizobium meliloti on the basis of their failure to form a fluorescent halo under UV light when grown on agar medium containing Calcofluor. These mutations define a new genetic locus we have termed exoH. Alfalfa seedlings inoculated with exoH mutants form ineffective nodules that do not contain intracellular bacteria or bacteroids. Root hair curling is significantly delayed and infection threads abort in the nodule cortex. Analyses of exopolysaccharide secreted by exoH mutants have shown that it is identical to the Calcofluor-binding exopolysaccharlde secreted by the exoH+ parental strain except for the fact that it completely lacks the succinyl modification. In vitro translation of total RNA isolated from nodules induced by an exoH mutant has shown that only one of the plant-encoded nodulins is induced, as compared with the 17 nodulins induced by wild-type strains. These observations suggest that succinylation of the bacterial polysaccharide is important for its role(s) in nodule invasion and possibly nodule development. Introduction The formation of legume root nodules and their invasion by Rhizobium entails a complex series of developmental changes (Newcomb, 1981). The products of the various rhizobial nod genes have been shown to be central to this process and are under active investigation by many laboratories (Downie and Johnston, 1986; Long, 1984). Recent findings have also focused attention on the critical role of rhizobial polysaccharides (Bauer, 1981; Sutherland, 1985) in nodulation. The acidic exopolysaccharide of R. meliloti (Jansson et al., 1977; Aman et al., 1981) has been strongly implicated as a requirement for nodule invasion (Leigh et al., 1985). We isolated a set of mutants of the R. meliloti strain Rm1021 on the basis of their failure to fluoresce under UV light on medium containing Calcofluor and showed that these mutants did not synthesize this acidic exopolysaccharide. The exo mutations in these strains mapped to different genetic loci (Leigh et al., 1985), at least several of which are located on the second symbiotic mega-

Cell 580

Figure

1. Wild-Type

Strain

Rm1021

(left) and exoH (haloless)

Mutant

of a deficiency in exopolysaccharide synthesis (Leigh et al., 1985). While screening, we also observed colonies (at a frequency of one per 10,000 TnScontaining transconjugants) that were fluorescent but differed from colonies of the parental strain Rm1021 by the lack of a fluorescent zone around the colony (Figure 1). Three mutant strains (Rm7l54, Rm7115, and Rm7225) exhibiting this “haloless” phenotype on Calcofluor plates were obtained; the first two strains contain an insertion of Tn5 and the last strain an insertion of the Tn5 derivative Tn5-233 (De Vos et al., 1988). For each strain, we found the haloless phenotype to be 100% cotransducible with the transposon insertion. These three insertion mutations define a new locus, which we have termed exoH. An R. meliloti cosmid clone bank was conjugally transferred into strain Rm7154, and a cosmid was isolated from a colony that had a halo. This cosmid, pEX154, complemented the haloless phenotype of the other two mutants as well. Furthermore, it also complemented the nonfluorescent phenotype of our previously described exoA mutants (Leigh et al., 1985) indicating that the exoH locus is linked to exoA. Consistent with this inference, we found that the exoH mutation of strain Rm7225 was 99% (189/191) cotransducible with the exoA mutation of strain Rm7031 (Leigh et al., 1985). In the course of a major mapping study of a region of DNA that

Rm7l54

(right)

on LB/Calcofluor

Agar Viewed

under

Ultraviolet

Light

includes the exoA, K, H, F, and B genes, the exoH mutations were mapped to a 3.5 kb EcoRl fragment of the cosmid pEX154 (J. Himawan, J. W. Reed, S. Long, and G. C. Walker, unpublished data). This fragment has been subcloned into the vector pSUP104 (Simon et al., 1983) to yield the plasmids pEX41 and pEX42 (J. Himawan et al., unpublished data). Strains mutant in exoH carrying either plasmid pEX41 or pEX42 exhibited a fluorescent halo on Calcofluor plates. By analyzing Tn5 insertion derivatives of these plasmids, we have been able to establish that the exoH locus is at least 1.O kb and less than 1.7 kb long (Figure 2). exoH Mutants Form Empty Nodules These exoH mutants were of particular interest because we found that, despite the fact that they evidently produced a Calcofluor-binding polysaccharide, they did not form normal Fix+ nodules on alfalfa (Medicago sativa). The parental strain Rm1021 formed pink cylindrical nodules within 10 days. In contrast, the nodules formed with the exoH mutants were round and white and remained so for up to a month. Although the nodules induced by exoH mutants were not cylindrical, they had peripheral vascular bundles and a nodule cortex (Figure 3). The plants inoculated with the exoH mutants remained blanched and stunted, indicating lack of nitrogen fixation, while plants

Mutants

2. Map of the exoH and exoK Loci

1.0 kb

R. meliloti 581

1

Figure Tn5 insertions we have obtained in genomic DNA are shown below the line, and those obtained in the subclones pEX41 and pEX42 are shown above the line. For genomic Tn5 insertions, open circles represent haloless, Fix- mutations in the exoH locus; half-open circles represent delayed halo, Fix+ insertions in the exoK locus (J. Himawan, J. W. Reed, S. Long, and G. C. Walker, unpublished data); and the closed circle represents an insertion having a wild-type phenotype. For plasmid Tn5 insertions, closed circles represent insertions in the subclones that complement both the haloless and Fix- phenotypes of the exoH mutants Rm7l54 and Rm7225; open circles represent insertions that fail to complement either phenotype.

inoculated with the parental strain Rm1021 grew normally. Three weeks after inoculation, plants inoculated with the exoH mutants had about 15 nodules, while plants inoculated with Rm1021 had about ten nodules. Introduction of pEX154, pEX41, or pEX42, plasmids which complemented the haloless phenotype, into the exoH mutants restored the ability of the strains to form normal Fix+ nodules. Occasionally, nodules formed with strain Rm7225 became cylindrical and pink after a long delay (20 to 35 days), and renewed plant growth occurred after 35 to 40 days. In all of the above respects, the nodules elicited by the exoH mutants resembled those elicited by exopolysaccharide-deficient (exo) mutants. Since the nodules formed by exo mutants are empty and contain only few bacteria and no bacteroids (Leigh et al., 1985; Finan et al., 1985; Leigh and Hirsch, unpublished results), we examined the nodules obtained with the exoH mutants. Crushing such nodules three weeks after inoculation released few, if any, bacteria and no bacteroids as judged by light microscopy, while turbid suspensions of bacteroids were released from nodules obtained with the parental strain Rm1021. More detailed examinations of sectioned nodules by light microscopy revealed that the majority of exoH nodules sectioned (39141) were devoid of intracellular bacteria. Thus, in these respects also, the nodules elicited by the exoH mutants resembled those elicited by exopolysaccharide-deficient mutants. Nodules Obtained with exoH Mutants Are Delayed in Root Hair Curling and Have Aborted Infection Threads Unlike strain EJ355, an exo6 mutant of R. meliloti that elicited root hair branching and hypertrophy but not shepherd’s crook formation (Finan et al., 1985), exoH mutants elicited marked root deformation and 360° curling of root hairs (Figure 4A). However, in contrast to the wild-type strain Rm1021, which curls root hairs within 24 hr of inoculation, exoH mutants did not curl root hairs until 72-96 hr

Figure duced

,loOlrm

,

an Alfalfa Root In-

cell. Other abbreviaand r (root).

3. Off Median Longitudinal Section through by the R. meliloti exo,Y Mutant Rm7l54

An infection thread (i 1) is present in a peripheral tions: nc (nodule cortex), vb (vascular bundle),

after inoculation. Seven days after inoculation, equal or greater numbers of shepherds crooks were noted on exoH-inoculated roots compared with roots inoculated with the wild-type strain (Figure 48). Moreover, root hair deformations other than 380“ curling (such as hypertrophy, corkscrew coiling, and looping) were also more frequently observed on exoH-inoculated roots (data not shown). Although shepherd’s crooks were elicited by exoH mutants, we did not observe infection threads in these curled root hairs. To examine nodule structure in more detail, we sectioned more than 40 nodules. In almost one-fourth of the nodules examined (10/41), infection threads were present within the external cells of the nodule cortex (Figure 5A). In certain cases, the infection threads were within what appeared to be remnants of root hairs (Figure 5B). Frequently, the infection threads were broader than those observed in wild-type induced nodules and contained numerous rhizobia (Figure 58). They appeared to end abruptly in the nodule cortex. In a minority of nodules (2/41 examined), bacteria were released from infection threads into the host cell cytoplasm (Figure 6). Although some mutant bacteria became singly enclosed within peribacteroid membrane, we did not observe mature bacteroids in any of the nodules examined. From observing sectioned nodules, we also found rhizobia located between host cells. Occasionally, they had penetrated several cell layers into the nodule (data not shown). These bacteria were not encapsulated by peribacteroid membranes and did not differentiate into bacteroids. The Exopolysaccharide Secreted by exoH Mutants Is Not Succinylated The fact that colonies of exoH mutants fluoresced under UV on Calcofluor plates indicated that these cells were

Cell 562

Figure 4. (A) A 360° curl of a root hair (shepherds crook) elicited by the R. meliloti exoH mutant Rm7154 3 days after inoculation of alfalfa roots. (6) Several shepherds crooks (arrows) elicited by the R. meliloti exoH mutant Rm7l54 I seven days after inoculation of alfalfa roots, vrewed with Nomarsf ii optics.

No penetration rhizobia (r) are

were altered in their ability to degrade it into smaller more diffusible fragments, or second, that an altered Calcofluorbinding polysaccharide was synthesized that was not subject to the normal cellular degradative processes. Our results are consistent with the latter explanation and

within a peripheral cell of an alfalfa root nodule elicited by R. meliloti exoH mutant Rm7f54. (6) Infection threads (i t) viewed in cross section in the bases of root hair cells. Numerous

synthesizing a polysaccharide that was capable of binding Calcofluor. The two simplest a priori explanations for the absence of a fluorescent halo around the colonies were either first, that normal Calcofluor-binding exopolysaccharide was being synthesized but the exoH mutants

Figure 5. (A) Infection thread (i t) (and branches?) of the inner nodule cells is observed. present within the thread.

Ft. meliloti Mutants 503

Figure 6. Transmission Electron Micrograph of Infection Threads (i t) within the Cytoplasm of a Cell of a Nodule Elicited on Alfalfa by the R. meliloti exoH Mutant Rm7154 Some rhizobia appear to be exiting from the infection thread via an infection drop (i d) (Newcomb. 1961). Bar = 1 Wm.

indicate that exoH mutants synthesize an altered Calcofluor-binding exopolysaccharide that differs from wildtype Calcofluor-binding exopolysaccharide by the absence of succinylation. When the exoH mutants were grown under nitrogenlimiting conditions, they secreted substantial amounts of a Calcofluor-binding exopolysaccharide into the medium as did their parent strain Rm1021. However, the exopoly saccharide produced by the exoH mutants was not as efficiently precipitated by the cationic detergent cetrimide, suggesting that exoH exopolysaccharide was not as acidic as the Rm1021 exopolysaccharide. This suspicion was confirmed by proton NMR spectroscopy. As previously reported, the exopolysaccharide synthesized by strain Rm1021 contained approximately one unit each of pyruvate, acetate, and succinate for each oligosaccharide subunit (Aman et al., 1981) (Figure 7). In contrast, the exopolysaccharide isolated from the exoH mutants lacked succinate (Figure 7). Consistent with our genetic results, we found that exoH mutants carrying the cosmid pEX154 synthesized succinylated exopolysaccharide (Figure 7). The somewhat higher degree of succinylation of the exopolysaccharide in the strain carrying the exe/-P plasmid pEX154 is consistent with the exoH gene product being the succinylating activity or a factor that positively regulates the succinylating activity. The only effect of the exoH mutation on the structure of the Calcofluor-binding exopolysaccharide seems to be the absence of the succinate. Gas chromatographiclmass spectrometric analyses of purified exoH exopolysaccharide indicated that the sugar compo-

Rm7225

Rm1021

6

5

4

/

Ppn

3

II

2

I

1

0 Figure 7. Proton NMR of Exopolysaccharide Produced by a Wild-Type R. meliloti Strain (RmlOPl), Three exoH Mutants (Rm7l54, Rm7225, and Rm7l15), and a Complemented exoH Mutant (Rm7l54[pEX154]). The assignments are as in Aman et al. (1951). The singlets at 1.46 and 2.15 ppm represent the methyl protons of the 1-carboxyethylidene (pyruvate) and acetyl groups respectively. The triplets at 2.47 and 2.63 ppm represent the methylene protons of the succlnyl group (absent in the exoH mutants). The complex region from 3.3 to 4.6 represents ring protons of the carbohydrate constituents. The solvent peak (HDO) is at 4.19 ppm.

sition and linkages were identical to those of the Calcofluor-binding exopolysaccharide synthesized by strain Rm1021. Thus the exopolysaccharide produced by strain Rm7154 contained 3-linked glucose, 4-linked glucose, 6-linked glucose, 46linked glucose, and 3-linked galactose.

Cell 584

nst-29r

We have recently observed that Tn5 insertions in the exoK locus (Figure 2) immediately downstream of exoH (Himawan et al., unpublished data), initially exhibit a phenotype on Calcofluor plates similar to that of exoH mutants. They appear haloless for the first four days of incubation, but then, unlike exoH mutants, they acquire a fluorescent halo much like that of the wild-type strain. Furthermore, these exoK mutants form Fix+ nodules on plants. We have analyzed exopolysaccharide from an exoKstrain by proton NMR and have found that it has succinate, as well as acetate and pyruvate, modifications (data not shown).

Figure

8. Identification

of Nodulin

Gene

Products Autoradiographs of two-dimensional gels of in vitro translated products from total RNA isolated from: (A) effective alfalfa root nodules induced by R. meliloti Rm1021, (6) ineffective root nodules induced by the exoH strain Rm7l54, and (C) uninoculated roots. Nodule-specific and nodule-stimulated (nst) products are indicated by arrowheads. The corresponding locations of n-38 and GS, are ‘indicated by open circles. Leghemoglobin (Lb) and nodule-specific glutamine synthetase (GS,) were previously identified by immunoprecipitation (Hanks et al., unpublished data). Three spots present on all gels are indicated with long arrows to aid in orientation.

exoH Mutants Induce the Synthesis of Only One Nodulin In vitro translation of total RNA isolated from nodules induced by wild-type Ft. meliloti strains has resulted in the identification of 17 major nodule-specific gene products (nodulins) (Lang-Unnasch and Ausubel, 1985; Vance et al., 1985; J. F. Hanks, L. A. Macol, and A. M. Hirsch, unpublished data) (Figures 8A and 8C). In contrast, when the in vitro translation products of total RNA isolated from nodules induced by exoH mutants were analyzed, only one nodule-specific translation product (nodulin n-38) was observed (Figure 86). In this respect, the plant response in-

R. meliloti 505

Mutants

duced by the exoH mutants is identical to that induced by the exopolysaccharide deficient exo mutants (J. F. Hanks, C. A. Smith, L. A. Macol, and A. Hirsch, unpublished data; Ft. Dickstein, personal communication). None of the other nodulins, including the five leghemoglobin spots or nodule-specific glutamine synthetase (GS,), was detected (Figure 86). In addition, the expression of two other nodule mRNAs was observed to be greatly stimulated in translations of RNA from exoH-induced nodules. One of these (nst-57; Figure 86) is present at about equal levels in nitrogenfixing nodules (Figure 8A) and roots (Figure 8C), and is therefore not a nodulin but stimulated in nodules induced by exoH mutants and by the exopolysaccharide-deficient 8x0 mutants. The second nodule-stimulated spot (nst-28; Figure 86) was not found in nitrogen-fixing nodules and appears to be unique to nodules induced by exoH mutants and by the exopolysaccharide-deficient exo mutants. Protein extracts of exoH-induced nodules were analyzed for nodulin proteins by enzyme-linked immunoassay. None of the leghemoglobin proteins was detected using two different preparations of leghemoglobin antisera, nor was nodule-specific glutamine synthetase present. We had previously identified these nodulins in extracts from nitrogen-fixing nodules with these same antisera. In addition, none of the other nodulin proteins previously identified using antiserum prepared against nodule extract (Hanks et al., unpublished data) was observed in exoH-induced nodules. Discussion In this paper, we report the isolation of exoH mutants of R. meliloti strain Rm1021 that fail to succinylate their acidic Calcofluor-binding exopolysaccharide. These exoH mutants are able to elicit nodule development on alfalfa but are defective in nodule invasion and thus produce “empty” nodules that do not fix nitrogen. Alfalfa seedlings inoculated with these exoH mutants exhibit delayed root hair curling compared with plants inoculated with wild-type bacteria, form infection threads that abort at a very early stage, and express only one nodulin. The most economical explanation for our results is that the exoH gene either encodes or positively regulates a succinylating activity and it is the failure of the exoH mutants to succinylate their Calcofluor-binding exopolysaccharide that is responsible for their deficiencies in nodule invasion. This straightforward model is interesting, as it suggests that a particular structural feature of this bacterial polysaccharide is required in order for it to function correctly in the developmental process of nodulation. The complex symbiotic phenotype of exoH mutants described above could result from the polysaccharide playing one or more primary roles during nodulation. We will briefly discuss some of the formal possibilities for how the polysaccharide could act in the nodulation process. For example, the delay in root hair curling and infection thread initiation seen with exoH mutants could result from deficiencies of the mutant bacteria in recognizing root hairs that are susceptible to invasion (Bauer, 1981; Halverson and Stacey,

1988) or in providing a sufficient local concentration of enzymes that might be involved in plant cell wall degradation (Ljunggren and Fahraeus, 1981; Callaham and Torrey, 1981; Ridge and Rolfe, 1985). The altered structure of the polysaccharide could affect the efficiency of recognition of infectible root hairs, while the reduced negative charge of the nonsuccinylated polysaccharide might make it a less effective binder of extracellular degradative enzymes. The early termination of infection threads could be due to a failure of the nonsuccinylated exopolysaccharide to serve as an appropriate matrix in the thread. Finally, the failure of exoH mutants to induce most of the plantencoded nodulins could be due to an inability to produce an appropriate oligosaccharide that might serve as a signal (Darvill and Albersheim, 1984; Van et al., 1985; Halverson and Stacey, 1988) in eliciting the synthesis of nodulins. With respect to this last possibility, it is interesting to note that a fix-27 mutant of R. meliloti, which appears to abort nodulation at a similar stage to exoH mutants (R. Clover and E. R. Signer, personal communication), elicits the complete spectrum of nodulin gene expression (except for glutamine synthetase) (J. F. Hanks, L. A. Macol, and A. M. ,Hirsch, unpublished data). Two other more complicated classes of models must also be considered. The second model resembles the first in assuming that the exoH gene product either has a succinylating activity or regulates a succinylating activity, but differs from it in suggesting that it is the succinylation of some other entity besides the Calcofluor-binding polysaccharide that is crucial for normal nodule development. For example, a fraction of the cyclic (l-+2)-8-D-glucan of R. meliloti is acidic (Miller at al., 1988; K. J. Miller, personal communication) and therefore might be succinylated, as are the related membrane-derived oligosaccharides of E. coli (Kennedy et al., 1978; Miller et al., 1988). Since mutants (n&A and n&S) that do not synthesize the cyclic (1-2)~f3-D-glucan (G. Ditta, personal communication) also form empty ineffective nodules (Dylan et al., 1988), it is possible that the failure of an exoH mutant to succinylate such a molecule could account for one or more of the symbiotic deficiencies of exoH mutants that we have observed. The third model assumes that the exoH gene product plays some role in normal nodulation that is completely independent of its role in succinylation, for example, functioning as a receptor. We consider this last model less likely, but it too is consistent with our present data. The three models that we have discussed are not mutually exclusive, and it is possible that the various symbiotic deficiencies of exoH mutants that we have observed could ultimately be explained by some combination of these classes of models. In contrast to the results reported by Djordjevic et al. (1987) with derivatives of Rhizobium sp. strain NGR234 and with R. trifolii, we have not been able to suppress the invasion deficiency of R. meliloti Rm1021 exo mutants by the addition of exogenous purified Calcofluor-binding exopolysaccharide. The apparent requirement for this exopolysaccharide for continued infection thread growth and induction of nodulins in R. meliloti-alfalfa nodulation might imply some internal signalling mechanism and

Cell 566

Procedures

could account for our failure to see an analogous suppression. The R. meliloti exoH mutants described in this paper were originally identified because of their haloless phenotype on plates containing Calcofluor. The data discussed in this paper indicate that exoH mutants synthesize a high molecular weight Calcofluor-binding polysaccharide that differs from that made by the parental strain by the absence of the succinyl modification. One simple explanation for this haloless phenotype is that it is due to a failure of a bacterial endoglycosylase to degrade the polysaccharide to diffusible Calcofluor-binding fragments in the normal fashion. If future work were to show this to be the case, the symbiotic deficiencies of exoH mutants would focus attention on the possible importance of exopolysaccharide degradation in normal nodulation. It is possible that there is a need to degrade the polysaccharide at some stage during nodulation in order for the rhizobia to proceed in the nodulation process. Alternatively, as speculated above, a degradation product of the polysaccharide could function as a signal to the plant. In this view, one or more of the deficiencies of exoH mutants in nodulation could be due to an inability of the nonsuccinylated polysaccharide to be degraded to the appropriate oligosaccharide by either bacterial or plant enzymes, or a requirement for the succinyl modification on such an oligosaccharide if the appropriate degradation did occur. Experimental

and Hair Curling Assays of alfalfa (Medicago saliva cv. Iroquois) were germinated on agar slants and inoculated as described previously al., 1982). Root hair curling was examined by growing alfalfa in Farhaeus slide assemblies containing liquid medium 1957).

Bacterial Strains and Genetic Techniques The strains used have been described previously (Leigh et al., 1985). Mutants Rm7154 and Rm7ll5 were obtained from matings of E. coli strain 1830 with R. meliloti Rm1021. Mutant Rm7225 was obtained from a mating of E. coli MM294A(pRK607) with Rm1021. The plasmid pRK607 is a derivative of pRK2013 (Figurski and Helinski, 1979) containing Tn5-233, a modified form of Tn5 that encodes resistance to gentamycin and spectinomycin (De Vos et al., 1986). Antibiotic levels used were neomycin sulfate (200 fig/ml), gentamycin (15 pglml), spectinomycin (100 pglml). and streptomycin (500 pg/ml). The haloless mutants were obtained by screening for colonies lacking a halo on buffered (10 mM HEPES [pH 7.41) LB agar containing 0.02% Calcofluor white MR2 (Cellufluor, Polyscience, Warrington, PA) using a handheld UV lamp (long wavelength). Transduction, isolation of complementing cosmids, plant nodulation tests, and detection of exopolysaccharide with Calcofluor and cationic detergent hexadecyl trimethylammonium bromide were performed as described (Leigh et al., 1985). Tn5 insertions in the exoH+ subclones pEX41 and pEX42 were generated by passage through the Tn5-containing strain MT614 as described elsewhere (J. Himawan, J. W. Reed, S. Long, and G. C. Walker, unpublished data). For complementation tests, the subclones and the derivatives having Tn5 insertions were first mated into the exoH mutants Rm7l54 and Rm7225. The resulting strains were streaked on LB agar plates containing Calcofluor and used to inoculate seedlings as described below. Nodulation Seedlings nitrogen-free (Meade et seedlings (Farhaeus,

Light and Electron Microscopy Nodules were excised from alfalfa roots after acetylene reduction assays were performed. Nodules were removed between 18 days and 4 weeks after inoculation and fixed and prepared for microscopy as described previously (Hirsch et al., 1983).

Detection of Nodulins Total RNA was isolated from nodules or roots by LiCl precipitation (De Vries et al., 1982) and translated in vitro using a rabbit reticulocyte lysate kit (New England Nuclear). The products were analyzed by twodimensional gel electrophoresis @‘Farrell, 1975) for nodulin gene expression by comparison of nodule and root RNA translation products as described previously (J. F. Hanks, L. A. Macol. and A. M. Hirsch, unpublished data). For enzyme-linked immunoassay of protein extracts (Western blots), 70 pg of protein extract from nodules or roots were separated on a two-dimensional gel, transferred to nitrocellulose (Towbin et al., 1979). and analyzed with specific antibody against leghemoglobin or glutamine synthetase or antibody prepared against total nodule extract using peroxide-conjugated goat anti-rabbit secondary antibody and reagents purchased from Bio-Rad (Immuno-Blot kit). Two preparations of leghemoglobin antiserum and glutamine synthetase antiserum were kindly provided by Ton Bisseling (Agricultural University, Wageningen, The Netherlands), Carroll Vance (University of Minnesota, St. Paul) and Julie Cullimore (University of Warwick, Coventry) respectively. Antiserum prepared against total nodule extract was previously used to detect nodulins in nitrogen-fixing nodules (Hanks et al., unpublished data). Proton NMR Spectroscopy Samples for proton NMR spectroscopy were prepared as follows: logphase cells were suspended in nitrogen-free MS-glucose medium (Leigh et al.. 1985) at an optical density at 660 nm of 0.5 (18 mm light path), and shaken at 30°C for two days. Supernatant was lyophilized, dialyzed against water and lyophilized. Yield was approximately 200 mg per liter of cell suspension. Each sample (approximately 50 mg) was dissolved in water, sonicated to aid in dissolution and decrease viscosity, and exchanged several times with D20. Spectra were obtained with a Bruker 500 MHz instrument with the probe heated to 75OC or with the 500 MHz instrument operated by the Francis Bitter National Magnet Laboratory at M. I. T. The chemical shift standard was sodium(trimethylsilyl-l-propanesulfonate) (trimethylsilyl protons = 0 ppm). Acknowledgments

July 21, 1987; revised

September

1, 1987.

We thank Dan Doherty for the isolation of one of the exoH mutants, Tom Pratum and David Ruben for help with the NMR spectroscopy, and Steve Levery of Senitiroh Hakomori’s laboratory for the analysis of sugar composition. Our thanks to Carol A. Smith for the microscopy and photography, Lisa A. Macol for the in vitro translations, and Stacey Steinberg for technical assistance. We are grateful to Ton Bisseling, Carroll Vance, and Julie Cullimore for their gifts of antisera. This work was supported by a U. S. Department of Energy grant DEFG06-86ER13532 to J. A. L., by a grant from the Graduate School Research Fund of the University of Washington to J. A. L., by National Science Foundation grants PCM-8316793 and DCB-8703297 to A. M. H., and by Public Health Service Grant GM31030 to Ethan R. Signer and G. C. W. J. W. R. was supported by a National Science Foundation Predoctoral Fellowship. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. Received References

of legumes

by Rhizobia.

Ann. Rev. Plant

Aman. l?, McNeil, M., Franzen, L., Darvill, A. G., and Albersheim, l? (1981). Structural elucidation, using HPLC-MS and GLC-MS, of the acidic polysaccharide secreted by Rhizobium melilofi strain 1021. Carbohydr. Res. 95, 263-282.

Barber, C. E.. Lamb, J. W.. Daniels, M. J., Downie, J. A., A. W. 6. (1986). A mutation that blocks exopolysacchaprevents nodulation of peas by Rhizobium leguminosabeans by R. phaseoli and is corrected by cloned genes or the phytopathogen Xanthomonas. Mol. Gen. Genet.

Bauer, W. D. (1981). Infection Physiol. 32, 407-409. Borthakur, D., and Johnston, ride synthesis rum but not of from Rhizobium 203, 320-323.

R. meliloti 587

Mutants

Callaham, D. A., and Torrey, J. G. (1981) The structural basis for infection of root hairs of 7iifolium repens by Rhizobium. Can. J. Bot. 59, 1647-1664.

and the Ann. Rev.

H., Batley, M., Redmond, J., and Rolfe, 8. G. (1965). Alteration effective nodulation properties of a fast growing broad host Rhizobium due to changes in exopolysaccharide synthesis. J. Physiol. 120, 331-349.

Chakravorty, A. K., Zurkowski, W., Shine, J., and Rolfe, B. G. (1982). Symbiotic nitrogen fixation: molecular cloning of Rhizobium genes involved in exopolysaccharide synthesis and effective nodulation. J. Mol. Appl. Genet. 1. 565-596. Chen, of the range Plant Darvill, A. G., and Albersheim, R (1964). Phytoalexins elicitors-a defense against microbial infection in plants, Plant Physiol. 35, 243-275.

J. G. H. (1962). Diversity of of protein synthesis in etio756. 129-135.

De Vos, G. F., Walker, G. C., and Signer, E. R. (1966). Genetic manipulation in Rhizobium me/i/o0 utilizing two new transpsosn Tn5 derivatives Mol. Gen. Genet. 204, 485-491. De Vries, S. C., Springer, J., and Wessels, abundant mRNA sequences and patterns lated and greened pea seedlings. Planta

A. W. B. (1966). Nodulation root? Cell 47; 153-154.

of legumes

Djordjevic, S. P., Chen, H., Batley, M., Redmond, J. W., and Rolfe, B. G. (1987). Nitrogen fixation ability of exopolysaccharide synthesis mutants of Rhizobium sp. strain NGR234 and Rhizobium trifolii is restored by the addition of homologous exopolysaccharides. J. Bacterial. 169, 53-60. Downie, J. A., and Johnston, by Rhizobium: the recognized Dylan, T., lelpi, L., Stanfield, S., Kahyap, L., Douglas, C., Yanovsky, M., Nester, E., Helinski, D. R., and Ditta. G. (1966). Rhizobium me/i/o0 genes required for nodule development are related to chromosomal virulence genes in Agrobacferium tumefaciens. Proc. Natl. Acad. Sci. USA 83, 4403-4407. Fahraeus, G. (1957). The infection of clover root hairs by nodule bacteria studied by a simple glass slide technique. J. Gen. Microbial. 32, 374-381. Figurski, D. H., and Helinski, D. R. (1979). Replication of an origincontaining derivative of plasmid RK2 dependent on a plasmid function provided in trans. Proc. Natl. Acad. Sci. USA 76, 1646-1652. Finan. T. M., Hirsch, A. M., Leigh, J. A., Johansen, E., Kuldau. G. A., Deegan, S.. Walker, G. C., and Signer, E. R. (1965). Symbiotic mutants of Rhizobium meliloti that uncouple plant from bacterial differentiation. Cell 40, 669-677.

J. (1986). Signal exchange Rev. 50, 193-225.

in plant-

Finan, T. M., Kunkel, B., De Vos, G. F., and Signer, E. R. (1966). A second symbiotic megaplasmid in Rhizobium meliloti encodes exopolysaccharide and thiamine genes. J. Bacterial. 767, 66-72. Halverson, L. J., and Stacey, microbe interactions. Microbial. Hirsch, A. M., Bang, M.. and Ausubel, F. M. (1963). Ultrastructure analysis of ineffective nodules formed by nif:Tn5 mutants of Rhizobium meliloti. J. Bacterial. 755, 376-380.

poly nod-

Hynes, M. F., Simon, R., Muller, t? Niehaus, K., Labes, M., and Puhler, A. (1986). The two megaplasmids of Rhizobium meliloti are involved in the effective nodulation of alfalfa Mol. Gen. Genet. 202, 356-362 Jansson, P., Kenne, L., Lindberg, B. Ljunngren, H., Lonngren, J., Ruden, U.. and Svensson, S. (1977). Demonstration of an octasaccharide repeating unit in the extracellular polysaccharide of Rhizobium meliloti by sequential degradation. J. Amer. Chem. Sot. 99, 3612381 5. Lang-Unnasch, N., and Ausubel, F. M. (1965). Nodule-specific peptides from effective alfalfa root nodules and from ineffective ules lacking nitrogenase. Plant Physiol. 77, 833-639.

H., and Fahraeus, G. (1961). The role of polygalacturonase invasion by nodule bacteria. J. Gen. Microbial. 26,521-526.

Leigh, J. A., Signer, E. R., and Walker G. C. (1965). Exopolysaccharide deficient mutants of ffhizobium meliloti that form ineffective nodules. Proc. Natl. Acad. Sci. USA 82, 6231-6235. Ljunggren, in root-hair

Long, S. (1984). Genetics of Rhizobium nodulation. In Plant-Microbe Interactions, T. Kosuge and E. W. Nester, eds. (MacMillan: New York), pp. 265-306.

McNeil, M., Darvill, J.. Darvill, A. G., Albersheim, P, van Veen, R., Hooykaas, I?, Schilperoort, Ft., and Dell, A. (1966). The discernible structural features of the acidic exopolysaccharides secreted by different Rhizobium species are the same. Carbohydr. Res. 146. 367-326.

adaptaoligosac-

Meade, H. M., Long, S. R., Ruvkun, G. B., Brown, S. E.. and Ausubel, F M. (1982). Physical and genetic characterization of symbiotic and auxotrophic mutants of Rhizobium meliloti induced by transposon Tn5 mutagenesis. J. Bacterial. 749, 114-122. Miller, K. J., Kennedy, E. P, and Reinhold, V. N. (1966)Osmotic tion by gram-negative bacteria: possible role for periplasmic charides. Science 231, 46-51.

electrophoresis

Newcomb, W. (1961). Nodule morphogenesis and differentiation. In Biology of the Rhizobiaceae, Int. Rev. Qt. Suppl. 12, K. L. Giles and A. G. Atherly, eds. (New York: Academic Press), pp. 247-296. O’Farrell, P H. (1975). High resolution two-dimensional of proteins. J. Biol. Chem. 250, 4007-4021. Ridge, R. W. and Rolfe, B. G. (1965). Rhizobium sp. degradation of legume root hair cell wall at the site of infection thread origin. Appl. Envir. Microbial. 50, 7l7-720.

of gramAnn. Rev.

Simon, R., Priefer, U., and Puhler, A. (1983). Vector plasmids for in-vivo and in-vitro manipulations of gram-negative bacteria. In Molecular Genetics of the Bacteria-Plant Interaction, A. Puhler, ed. (Berlin: Springer-Verlag), pp. 98-106. Sutherland, I. W. (1965). Biosynthesis and composition negative bacterial extracellular and wall polysaccharides. Microbial. 39, 243-270.

Towbin, H., Staehelin, T., and Gordon, J. (1979). Electrophoretic transfer of proteins from polyacrylamide gels to nitrocellulose sheets: procedure and some applications. Proc. Natl. Acad. Sci. USA 76, 43504354. Van, K. T T., Toubart, P, Cousson, A., Darvill, A. G.. Gollin, D. J., Chelf, P., and Albersheim, P Manipulation of the morphogenetic pathways of tobacco explants by oligosaccharins. Nature 314, 615-617. Vance, C. P., Boylan, K. L. M., Stade, S., and Somers, D. A. (1985). Nodule-specific proteins in alfalfa (Medicago sativa L.). Symbiosis 1, 69-64. Van Golde, L. M. G., Schulman, H., and Kennedy, E. P. (1973). Metabolism of membrane phospholipids and its relation to a novel class of oligosaccharides in Escherichia co/i. Proc. Natl. Acad. Sci. USA 70, 1368-1372.