Generation of New Hydrogen-Recycling Rhizobiaceae Strains by Introduction of a Novel hup Minitransposon

Abstract

Hydrogen evolution by nitrogenase is a source of inefficiency for the nitrogen fixation process by the Rhizobium-legume symbiosis. To develop a strategy to generate rhizobial strains with H₂-recycling ability, we have constructed a Tn5 derivative minitransposon (TnHB100) that contains the ca. 18-kb H₂ uptake (hup) gene cluster from Rhizobium leguminosarum bv. viciae UPM791. Bacteroids from TnHB100-containing strains of R. leguminosarum bv. viciae PRE, Bradyrhizobium japonicum, R. etli, and Mesorhizobium loti expressed high levels of hydrogenase activity that resulted in full recycling of the hydrogen evolved by nitrogenase in nodules. Efficient processing of the hydrogenase large subunit (HupL) in these strains was shown by immunoblot analysis of bacteroid extracts. In contrast, Sinorhizobium meliloti, M. ciceri, and R. leguminosarum bv. viciae UML2 strains showed poor expression of the hup system that resulted in H₂-evolving nodules. For the latter group of strains, no immunoreactive material was detected in bacteroid extracts using anti-HupL antiserum, suggesting a low level of transcription of hup genes or HupL instability. A general procedure for the characterization of the minitransposon insertion site and removal of antibiotic resistance gene included in TnHB100 has been developed and used to generate engineered strains suitable for field release.

Legume plants utilize atmospheric nitrogen through their association with diazotrophic bacteria. However, this is an energy expensive process that requires a significant fraction (up to 22%) of the plant net photosynthate (37). A minimum of 25% of the electron flux through nitrogenase is directed to the reduction of protons into hydrogen gas (52). This nitrogenase-dependent hydrogen production is one of the major factors that influence the efficiency of symbiotic nitrogen fixation (50). A few Rhizobium and many Bradyrhizobium strains induce in nodule bacteroids a hydrogen uptake (Hup) system that is able to recycle the hydrogen evolved by nitrogenase (reviewed in references 16 and 46). This recycling results in a more efficient use of energy in the process. It has been shown that in the Bradyrhizobium japonicum-soybean system, symbioses with hydrogenase-positive strains result in significant increases of yield and nitrogen fixation compared to those with hydrogenase-negative isogenic mutants (16, 32), although experiments of this type have led to conflicting conclusions in other laboratories (12). Most Rhizobium strains potentially useful as legume inoculants are Hup⁻, and they produce root nodules that evolve large amounts of hydrogen (4).

The physiology and genetics of the hydrogen oxidation system in endosymbiotic bacteria have been studied in B. japonicum and Rhizobium leguminosarum bv. viciae (15, 31, 46). The first component of this system is a [NiFe] hydrogenase that oxidizes hydrogen evolved from nitrogenase and transfers the electrons to oxygen through an electron transport chain. The mature enzyme is an αβ heterodimer with subunits of 30 and 60 kDa. The hydrogenase active site contains a heterometallic NiFe cluster along with CN and CO groups (38). The overall hydrogenase structure is highly conserved in different bacterial systems.

The genetic determinants responsible for hydrogen oxidation in R. leguminosarum bv. viciae are located in a region of the symbiotic plasmid spanning ca. 18 kb of DNA. Within this region, a cluster of 18 genes (hupSLCDEFGHIJK hypABFCDEX) required for hydrogenase synthesis has been identified and sequenced (21, 22, 39, 40, 42). The hupS and hupL genes encode the hydrogenase small and large structural subunits, respectively (21). Other functions have been assigned to different Hup and Hyp proteins, based on either direct analysis or information obtained from conserved hup systems from other bacteria: electron transport (HupC [13]), HupL-specific proteolytic processing (HupD [43]), subunit scaffolding (HupK [23]), and Ni binding (HypB [41]). As is the case for other metalloenzymes, the synthesis of hydrogenase is a complex process, and a general model explaining it is still incomplete.

Analysis of the regulation of R. leguminosarum hydrogen oxidation genes has revealed the presence of two main promoters (designated P1 and P5) that control the expression of hup and hyp genes. P1 controls the symbiotic expression of hydrogenase structural genes and downstream hup accessory genes. This promoter is activated by NifA, the general activator of nitrogen fixation genes (5). P5 directs the expression of hypBFCDEX genes and is dependent on FnrN, a transcriptional activator of the Fnr-FixK family involved in gene activation in response to microaerobic or anaerobic conditions (9, 18, 19). The ubiquity of NifA and FixK in rhizobia (17) suggests that the incorporation of the H₂-recycling ability into other Rhizobium strains that are potentially useful as legume inoculants can be approached by using the whole R. leguminosarum hup cluster without further modification to enhance expression. In fact, different strategies for heterologous hup expression have been developed in the past either by using plasmid-borne constructs harboring the hup gene cluster from R. leguminosarum (6, 28) and B. japonicum (24, 26) or by integrating the hup cluster into the chromosome (1, 25, 56). The introduction of new symbiotic traits into Rhizobium strains by using plasmid constructs has a major drawback in the rapid loss of plasmids in bacteroids due to the absence of antibiotic selection pressure (26). This limitation can be overcome by the use of a stabilizing system (par locus) to enhance plasmid maintenance under these conditions (24). However, since a large number of proteins are involved in hydrogenase synthesis, the presence of the system on multicopy plasmids could impose a large metabolic load on recipient strains. Additionally, antibiotic markers required for plasmid selection in the laboratory might make these inoculants undesirable for field release. Chromosomal integration of the hup DNA region is likely to be a better alternative, provided that expression from a single copy results in hydrogenase levels high enough to recycle all the hydrogen evolved by nitrogenase. The hup genes from B. japonicum have been integrated into chickpea-nodulating Rhizobium strains by a recombination-dependent method using specific sequences from the recipient strain (1). However, this method requires the construction of different hup-containing vehicles adapted to each particular recipient strain. Integration of new genes into the bacterial chromosome has been simplified by the development of mini-Tn5 transposons based on the pUT vector (11, 20). This vector allows the cloning of different genes between the 19-bp inverted repeats required for Tn5 transposition. There are several advantages associated to the use of these vectors. First, the gene for the transposase is located outside of the inverted repeats, and so it is not transposed to the recipient genome. This ensures that no further transpositions will occur. Second, the pUT vector also contains the oriT region, so that this type of minitransposon can be introduced into gram-negative bacteria by conjugation. Third, pUT-derivative constructs require R6K-specific π protein for replication that must be provided by the host strain. Thus, they behave as suicide plasmids in Rhizobium. The system was further modified by Wilson et al. (57), who constructed a series of derivative vectors designed for the study of Rhizobium-plant interactions.

In this paper we report the generation of rhizobial strains with the ability to recycle hydrogen by using a novel minitransposon able to integrate the entire R. leguminosarum hup cluster.

MATERIALS AND METHODS

Bacterial strains, plasmids, and growth conditions.

The bacterial strains and plasmids used in this study are listed in Table 1. Escherichia coli strains were routinely grown at 37°C in Luria-Bertani medium. Rhizobium strains were grown in tryptone-yeast extract (TY), yeast mannitol broth (YMB), or Rhizobium minimal (Rmin) medium under conditions described previously (28). For bacterial conjugations, fresh cell cultures of donor and recipient strains were mixed on TY plates and incubated overnight at 28°C. Transconjugants were selected on Rmin plates supplemented with the corresponding antibiotics. To determine insertion stability in nodules, bacteroid suspensions were serially diluted and plated onto YMB agar. Two hundred colonies were streaked onto YMB agar plates with or without spectinomycin, and the frequency of spectinomycin-resistant colonies was calculated. The antibiotic concentrations used were as follows (in micrograms per milliliter): ampicillin, 150; cloramphenicol, 30; kanamycin, 50; spectinomycin, 100; and tetracycline, 12.

TABLE 1.

Bacterial strains and plasmids used in this study

Strain or plasmid	Relevant characteristics	Source or reference
E. coli
S17-1	thi pro hsdR hsdM+ recA RP4 2−Tc::Mu-Km::Tn7	51
S17-1 λ-pir	S17-1 λ pir; host for pir-dependent plasmids	51
B. japonicum
UPM744	CB1802, Hup− on soybean	44
UPM744-H series	UPM744::TnHB100	This work
UPM752	3.15 B3, Hup− on soybean	44
UPM752-H series	UPM752::TnHB100	This work
M. ciceri
UPMCa7	Wild type; Hup− on chickpea	33
UPMCa7-H series	UPMCa7::TnHB100	This work
M. loti
U226	Wild type; Hup− on birdsfoot trefoil	34
U226-H series	U226::TnHB100	This work
R. etli
CE3	Wild type; Hup− on bean	35
CE3-H series	CE3::TnHB100	This work
R. leguminosarum bv. viciae
UPM791	128C53 Strr	27
UML2	Wild type; Hup− on pea	45
UML2-H series	UML2::TnHB100	This work
PRE	Wild type; Hup− on pea	29
PRE-H series	PRE::TnHB100	This work
S. meliloti
102F34	Wild type; Hup− on alfalfa	10
102F34-H series	102F34::TnHB100	This work
Plasmids
pAL618	pLAFR1 cosmid, hup hyp region from UPM791	27
pCAM 140	mTn5SSgusA40 in pUT/mini-Tn5 Spcr	57
pCE7	pUC18 carrying DNA flanking Spcr in CE3-H7	This work
pK18mobsacB	pK18 derivative, sacB Kmr	49
pKmsCE7	pK18mobsac containing DNA flanking TnHB100 insertion in CE3-H7 and hupSL′ region	This work
pRL161	pUC18 derivative containing CK1 cassette, Apr Kmr	14
pRL487	pUC18 derivative containing CC1 cassette, Cmr Kmr	14
pRLH43	pBS derivative carrying a 6.4-kb EcoRI DNA fragment from pAL618	21
pUC18	Apr, α-lac/MCS	58
pUC18Not	Apr, pUC18 with NotI sites flanking the polylinker	20
pTnB1	pCAM140 derivative, Spcr/Apr	This work
pTnHB100	pTnB1 with R. leguminosarum hup cluster, Spcr Apr	This work

DNA manipulation techniques.

Plasmid DNA preparations, restriction enzyme digestions, DNA cloning, transformation of DNA into E. coli, agarose gel electrophoresis, and Southern hybridization of genomic DNA were performed by standard methods (47). Labeling of DNA probes with digoxigenin, hybridization of DNA on nylon filters, and chemiluminescent detection of hybridizing bands were performed by using a DIG detection kit as specified by the manufacturer (Roche Molecular Biochemicals, Mannheim, Germany). Total DNA from Rhizobium strains was isolated as previously described (27).

The DNA sequence was determined with a DNA-sequencing kit (dRhodamine Terminator Cycle-Sequencing Ready Reaction) and an ABI377 automatic sequencer (PE Biosystems, Foster City, Calif.). For the determination of nucleotide sequence flanking the TnHB100 insertion in R. etli CE3-H7, a primer complementary to 3′ end of the spectinomycin resistance cassette (5′-GCTGGCTTTTTCTTGTTATCG-3′) was used.

Construction of the TnHB100 minitransposon.

An outline of the construction process for TnHB100 is shown in Fig. 1. The whole R. leguminosarum hup cluster (hupSLCDEFGHIJKhypABFCDEX) was isolated from cosmid pAL618 as a PvuII restriction fragment. This fragment was modified into an XbaI fragment by intermediate cloning into the symmetrical polylinker of EcoRV-digested pRL487. This 19,148-bp XbaI fragment contains the 18 genes of the hup cluster along with flanking DNA: downstream from hypX, 87 nucleotides corresponding to the 5′ end of the ψ-hoxA pseudogene (39) and 1,502 nucleotides corresponding to the pLAFR1 vector sequence; upstream of hupS gene, a 1,085-bp fragment that includes the 3′ end of orf3 and orf4 (7). Plasmid pCAM140 was used as the pUT-like vector. In this plasmid, an XbaI site located outside the Tn5 inverted repeats was inactivated by XbaI digestion followed by Klenow enzyme filling and religation, generating plasmid pCAM140X. Then, a single XbaI site was introduced upstream of the Spc^r gene by using a NotI fragment containing pRL161 symmetrical polylinker previously cloned in pUC18Not. This fragment was cloned in pCAM140X, replacing the gusA gene, and further XbaI digestion and religation led to pTnB1. Plasmid pTnHB100 was obtained by cloning the 19,148-bp XbaI fragment into pTnB1. The construction and use of the TnHB100 minitransposon have been included in patent application P9902819 (Spain, December 1999).

FIG. 1.

Construction of pTnHB100. The minitransposon TnHB100 was obtained by cloning the hup region from cosmid pAL618 into a pUT derivative (pCAM140) as described in Materials and Methods. It comprises a 21,376-bp DNA region located between the 19-bp Tn5 inverted repeats, represented by solid boxes. The thick line downstream from the hup hyp region indicates DNA from the vector pLAFR1. The shaded bar over the gene cluster represents the restriction fragment used for deletion of the spectinomycin resistance gene (see Fig. 4). Abbreviations for restriction sites: C, CelII; E, EcoRI; N, NotI; P, PstI; X, XbaI.

Plant tests and hydrogen metabolism determinations.

Tests were carried out using plants of pea (Pisum sativum cv. Frisson), bean (Phaseolus vulgaris cv. Contender), chickpea (Cicer arietinum cv. Atalaya), birdsfoot trefoil (Lotus corniculatus cv. La Estanzuela San Gabriel), alfalfa (Medicago sativa ecotype Aragón), and soybean (Glycine max cv. Williams) inoculated with R. leguminosarum bv. viciae, R. etli, Mesorhizobium ciceri, M. loti, Sinorhizobium meliloti, and B. japonicum, respectively. Conditions for inoculation with rhizobial strains, cultivation of plants, and bacteroid preparation were as previously described (6). Plants were maintained on a standard nutrient solution (8) that was supplemented with 85 μM NiCl₂ from day 10 after inoculation. Hydrogenase activity in bacteroid suspensions was determined by measuring O₂-dependent H₂ uptake by the amperometric method (45), and hydrogen evolution in intact nodules was determined by gas chromatography (6). The protein content of bacteroid suspensions was determined by the bicinchoninic acid method (53) after digestion in 1 N NaOH at 90°C for 10 min, with bovine serum albumin as the standard.

Immunological detection of hydrogenase.

HupL protein was detected in bacteroid crude cell extracts using antiserum raised against B. japonicum HupL. Immunoblot assays were carried out as described by Brito et al. (6).

Nucleotide sequence accession numbers.

The nucleotide sequence corresponding to the 21,376 bp of TnHB100 has been compiled from data from our laboratory and other laboratories. This sequence has been submitted to GenBank database (accession no. AF246703). The nucleotide sequence of the DNA region flanking the insertion site of TnHB100 in CE3-H7 has been deposited in the same database (accession no. AF247185).

RESULTS

Construction and use of a hup minitransposon.

To develop a vehicle for the generation of stable insertions of hydrogenase genes into different rhizobia, we constructed a minitransposon (TnHB100) containing the hydrogenase gene cluster from R. leguminosarum bv. viciae UPM791. The design of the strategy for cloning the entire hup cluster into a minitransposon was greatly facilitated by the availability of the complete sequence of the R. leguminosarum hup gene cluster. The whole set of 18 genes (hupSLCDEFGHIJK hypABFCDEX) required for hydrogenase synthesis in R. leguminosarum was contained in a single 19,148-bp XbaI fragment, which was constructed as described in Materials and Methods and Fig. 1. This DNA fragment was cloned in plasmid pTnB1, a modified version of pCAM140 (57), resulting in plasmid pTnHB100. This plasmid harbors the minitransposon TnHB100 on a pUT-like vector. TnHB100 includes a spectinomycin resistance gene (Spc^r) and the 18 genes required for hydrogenase synthesis preceded by the 3′ region of orf3 and orf4 identified in cosmid pAL618 insert DNA (7). Due to the cloning procedure, a ca. 1.5-kb region from vector pLAFR1 is present downstream from the hup hyp region.

Minitransposon TnHB100 was introduced into different Hup⁻ rhizobial strains by conjugation using E. coli S17-1λ⁻pir(pTnHB100) as the donor strain. The recipient strains used were R. leguminosarum bv. viciae PRE and UML2, R. etli CE3, M. loti U226, S. meliloti 102F34, B. japonicum UPM752 and UPM744, and M. ciceri UPMCa7. Spectinomycin-resistant transconjugants arose at a frequency of ca. 10⁻⁸ per recipient cell with slight variations depending on the recipient species (data not shown). These transconjugants were checked for the presence of the hydrogenase gene cluster by Southern analysis using plasmid pAL618 DNA as probe. Representative examples of the results obtained are shown in Fig. 2A. Over 50% of the spectinomycin-resistant transconjugants of all recipient strains tested displayed the expected pattern of hup-specific hybridizing bands. Confirmation of transposition rather than recombination of the hup cluster was obtained by checking for the absence of pUT vector sequences by Southern hybridization analysis. Over 90% of the spectinomycin-resistant transconjugants that acquired the hup cluster showed no detectable pUT sequences (data not shown).

FIG. 2.

Analysis of TnHB100 insertion into Rhizobium strains. (A) Southern hybridization of EcoRI-digested total DNA from different wild-type and engineered rhizobial strains with pAL618 cosmid DNA as the probe. Strains: R. leguminosarum bv. viciae UPM791 (lane 1), PRE (lane 2), and PRE-H2 (lane 3); R. etli CE3 (lane 4) and CE3-H2 (lane 5); B. japonicum UPM744 (lane 6), and UPM744-H1 (lane 7); S. meliloti 102F34 (lane 8) and 102F34-H1 (lane 9). (B) Analysis for randomness of the TnHB100 insertion. Southern hybridization of PstI-digested total DNA from different TnHB100-containing clones using the spectinomycin resistance cassette as probe is shown. Strains: R. etli CE3 (lane 1) and CE3::TnHB100 derivatives (lanes 2 through 5); B. japonicum UPM752 (lanes 6) and UPM752::TnHB100 derivatives (lanes 7 through 10).

The strategy described above allowed us to obtain hup-bearing transconjugants from all the Hup⁻ strains tested. To check whether the insertions occurred at random sites, the variability of the site of insertion was analyzed by hybridization of PstI-digested total DNA from the transconjugants using a 2-kb EcoRI DNA fragment containing the spectinomycin gene as probe. As shown in Fig. 2B for R. etli CE3 and B. japonicum UPM752 derivatives, all the transconjugants analyzed gave different hybridization patterns, indicating that the transposon had inserted at different sites in the genome. The presence of single hybridizing bands on each transconjugant also indicates that no multiple insertion events took place during the process.

Expression of hydrogenase activity in engineered rhizobial strains.

To assess the expression of the introduced hup genes, two TnHB100-bearing transconjugants per recipient strain were used as inocula of the respective legume hosts and the rates of hydrogen evolution from nodules and hydrogen oxidation activity in bacteroid suspensions were determined (Table 2). High levels of hydrogenase activity were detected in TnHB100-containing derivative strains of B. japonicum, M. loti, R. etli, and R. leguminosarum bv. viciae PRE. The hydrogenase activity expressed by these engineered strains was capable of recycling all hydrogen produced by nitrogenase, as deduced from the undetectable levels of hydrogen evolution by nodules. In contrast, only moderate expression, leading to partial recycling of hydrogen, was observed in R. leguminosarum bv. viciae UML2, while very low hydrogenase activity was observed in M. ciceri and S. meliloti strains.

TABLE 2.

Hydrogen metabolism in bacteroids and nodules induced by TnHB100-containing rhizobial strains

Strain	Bacteroid hydrogenase activitya	Nodule H2 evolutionb
B. japonicum
752	<10	4.97 ± 3.19
752-H1	4,400 ± 1,250	<0.1
752-H2	3,200 ± 500	<0.1
744	<10	13.09 ± 1.42
744-H1	4,300 ± 450	<0.1
744-H11	2,850 ± 250	<0.1
M. ciceri
UPM Ca7	<10	4.43 ± 0.79
UPM Ca7-H1	20 ± 10	3.24 ± 0.43
UPM Ca7-H2	40 ± 10	4.19 ± 0.23
M. loti
U226	<10	5.88 ± 0.53
U226-H3	21,300 ± 2,450	<0.1
U226-H7	18,075 ± 1,450	<0.1
R. etli
CE3	<10	4.48 ± 2.07
CE3-H2	10,400 ± 3,850	<0.1
CE3-H7	10,350 ± 3,000	<0.1
CE3-H71	11,075 ± 850	<0.1
R. leguminosarum bv. viciae
UPM791	3,000 ± 200	<0.1
PRE	<10	10.50 ± 3.3
PRE-H4′	2,350 ± 500	<0.1
PRE-H2	2,550 ± 1,220	<0.1
UML2	<10	5.49 ± 0.19
UML2-H3	400 ± 100	1.85 ± 0.17
UML2-H8	350 ± 100	2.31 ± 0.16
S. meliloti
102F34	<10	13.24 ± 1.64
102F34-H1	<10	14.24 ± 4.17
102F34-H3	50 ± 20	13.65 ± 1.74

^aExpressed as nanomoles of H₂ consumed per hour per milligram of protein. Values are averages ± standard deviations of four determinations. 

^bExpressed as micromoles of H₂ evolved per hour per gram (fresh weight) of nodule. 

The amount of hydrogenase protein in bacteroids induced by the engineered strains was estimated by immunoblotting using antiserum raised against the hydrogenase large subunit. Mature hydrogenase polypeptide in bacteroids, as determined by the presence of an immunoreactive band of 65 kDa, was apparent in bacteroid extracts from TnHB100-containing derivatives of R. leguminosarum bv. viciae PRE, R. etli CE3, and M. loti U226 (Fig. 3, lanes 3, 7, and 12, respectively). In contrast, no detectable signal was observed in derivative strains of R. leguminosarum UML2 (lane 5), S. meliloti 102F34 (lane 10), or M. ciceri UPMCa7 (data not shown). The amount of immunoreactive material in crude cell extracts showed a good correlation with hydrogen oxidation activities previously determined (Table 2).

FIG. 3.

Immunodetection of the hydrogenase large subunit (HupL) in TnHB100-containing strains. The immunoblot shows bands immunoreactive with antiserum raised against B. japonicum HupL in bacteroid crude extracts resolved by sodium dodecyl sulfate-polyacrylamide gel electrophoresis on a 10% acrylamide gel. Numbers on the left of the panel indicate the molecular masses of unprocessed (66-kDa) and processed (65-kDa) forms of hydrogenase large subunit as deduced from the comparison to standard molecular mass markers. Strains: R. leguminosarum bv. viciae UPM791 (lane 1), PRE (lane 2), PRE-H4 (lane 3), UML2 (lane 4), and UML2-H3 (lane 5); R. etli CE3 (lane 6), CE3-H7 (lane 7), and CE3-H71 (lane 8); S. meliloti 102F34 (lane 9) and 102F34-H3 (lane 10); M. loti U226 (lane 11) and U226-H3 (lane 12).

It was expected that insertion of TnHB100 into the genome of the recipient strains led to stable integration of the hup cluster in the absence of antibiotic selection pressure. To confirm this, we analyzed one of the engineered strains (R. etli CE3-H7) for the presence of TnHB100 in nodules by assessing the percentage of spectinomycin-resistant isolates in bacteria recovered from nodules. All isolates analyzed were resistant to spectinomycin, indicating that TnHB100 was fully stable under symbiotic conditions.

Modification of engineered rhizobial strains for field release.

To generate rhizobial strains engineered for hydrogen recycling that are useful as field inoculants, additional modification and characterization of the TnHB100-containing strains was required. Integration of the minitransposon led to stable incorporation of the hydrogen oxidation gene cluster along with the spectinomycin and streptomycin resistance gene (Spc^r) used as marker. This gene should be removed from the engineered strains, since antibiotic resistance markers are undesirable for inoculants to be released into the environment. Additionally, an adequate description of engineered inoculant strains requires characterization of the site of insertion of the minitransposon in each case.

To eliminate the spectinomycin resistance marker, a strategy involving in vitro deletion of this gene and subsequent marker exchange of the deletion was designed and applied to R. etli transconjugant strain CE3-H7 (Fig. 4). First, the DNA region flanking the TnHB100 insertion in R. etli CE3-H7 was cloned in plasmid pCE7 by en masse ligation of PstI-digested CE3-H7 total DNA to pUC18 DNA digested with the same enzyme and selection of spectinomycin-resistant transformants. An EcoRV-CelII fragment containing R. leguminosarum bv. viciae hupS (including the P1 promoter region) and part of hupL was obtained from pRLH43. This fragment was cloned adjacent to R. etli DNA downstream from the spectinomycin resistance gene in pCE7 in the vector pK18mobsac, obtaining pKmsCE7. Plasmid pKmsCE7 was used to induce the deletion of the spectinomycin resistance gene from CE3-H7 by selection of double recombinants by using sucrose sensitivity mediated by the sacB gene. The absence of the resistance gene in the resulting spectinomycin-sensitive derivative strains was confirmed by Southern blotting using as a probe a ca. 2-kb EcoRI DNA fragment containing this gene (data not shown). One of the antibiotic resistance-cured derivatives, CE3-H71, was used as the inoculum for bean plants, and bacteroids from nodules were analyzed for hydrogenase activity (Table 2). Bean bacteroids of CE3-H71 showed levels of hydrogenase activity comparable to those from the parental strain, CE3-H7. Also, similar amounts of hydrogenase enzyme were detected in immunoblot experiments (Fig. 3, lane 8). These results indicate that the process of elimination of the spectinomycin resistance gene did not affect the expression of the adjacent hydrogenase gene cluster.

FIG. 4.

Elimination of the spectinomycin resistance gene from TnHB100-containing strains. The scheme shows the outline of the procedure for the elimination of the spectinomycin resistance gene (Spc^r) in R. etli CE3-H7. The narrow shaded box in the R. etli CE3-H7 genome indicates the DNA region flanking the TnHB100 transposon cloned in pCE7 along with the Spc^r gene. The striped box corresponds to the DNA region from pAL618 upstream from the hup hyp region, which is also removed by this procedure. Plasmid pKmsCE7 is a pK18mobsac derivative used for the transfer of the deletion to the genome (see the text for details). Only relevant regions of the plasmids are shown.

A major drawback of systems for strain engineering based on random insertion of new genes into the genome is the potential negative effects associated with the insertion into a particular site. In our work, inspection of the symbiotic performance of the different hup-containing transconjugants did not reveal significant differences in plant growth or nodule development. However, genetically engineered rhizobial inoculants generated by this system require characterization of the insertion site of TnHB100 in each case. The above-described strategy for the elimination of antibiotic resistance genes generated an intermediate plasmid, pCE7, that allows a facile characterization of the integration site by DNA sequencing. For the R. etli CE3-H7 insertion, plasmid pCE7 contained a 656-bp DNA region flanking TnHB100. This DNA region was sequenced, and the resulting nucleotide sequence was compared to DNA data banks. The analysis showed that TnHB100 insertion in CE3-H7 is located within an open reading frame (orf1) homologous to genes encoding hypothetical proteins identified in genome-sequencing projects from Haemophilus influenzae (P44861), Yersinia enterocolitica (P31485), and E. coli (P77570). The potential phenotypic effect associated with this mutation is currently under investigation.

DISCUSSION

In this paper we present a novel strategy for the incorporation of the R. leguminosarum bv. viciae hydrogen oxidation gene cluster into different members of the family Rhizobiaceae based on the use of a new hup minitransposon (TnHB100). The available data indicate that this 21-kb DNA minitransposon is efficiently transposed into different rhizobial strains. This is about twice the maximum size of DNA fragments previously transposed using pUT-based systems (11), stressing the idea that Tn5-derived minitransposons are powerful tools for the incorporation of new phenotypic traits into gram-negative bacteria, including traits, like hydrogenase, encoded by complex multigenic clusters.

The use of TnHB100 has allowed the generation of engineered R. leguminosarum, R. etli, M. loti, and B. japonicum strains that are able to recycle the hydrogen evolved from nitrogenase, thus showing maximum values of relative efficiency (defined as 1 − [H₂ evolved/C₂H₂ reduced]). Regarding the first three species, these results are consistent with previous data from Brito et al. (6) on the analysis of heterologous expression of a cosmid-based R. leguminosarum hup cluster in different rhizobia. This hup cosmid (pAL618) is present at ca. 15 copies per cell in Rhizobium (36), with a stability in Rhizobium bacteroids ranging from 50 to 80% (6). When the hup cluster is transposed into the genome using TnHB100, only one copy per cell is present. The data presented in this work demonstrate that the expression of this single copy is enough to recycle all the hydrogen evolved by nitrogenase in the species mentioned above. This result indicates that the hup and hyp promoters are efficiently activated by endogenous NifA and FnrN-FixK regulators, respectively. This is particularly relevant for B. japonicum. Previous attempts to efficiently express hup genes in Hup⁻ strains of this bacterium were hampered by the low stability of pLAFR1-derivative plasmids in bacteroids (26), which led to low levels of hydrogenase activity. This limitation has been overcome by the approach developed in this work.

Introduction of TnHB100 resulted in very low to undetectable levels of hydrogenase activity in R. leguminosarum bv. viciae UML2, S. meliloti, and M. ciceri strains. Low levels of expression of cosmid-based hup genes, leading to only partial recycling of the hydrogen evolved by nitrogenase, had been previously observed in R. leguminosarum UML2 and S. meliloti 102F34 (6). Now that these strains have been engineered for hup activity by using a single-gene-copy strategy, the observed level of hydrogenase activity is even lower. These data are consistent with the absence of hydrogenase in immunoblot experiments. A possible explanation for these results is a poor transcription of hup genes in these strains, which might be due to low levels of available NifA protein. Limitation of the expression of foreign genes in rhizobia due to low availability of NifA protein has been previously reported for other NifA-dependent genes from S. meliloti, such as dicarboxylic acid transport or nodulation efficiency genes (2, 48). However, since transcription of transposed hup genes in heterologous backgrounds has not been quantitated directly, the possibility of the existence of additional limitations at the posttranscriptional or posttranslational level cannot be discarded. There may be potential limitations on the efficiency of HupL processing by HupD (6) and also on the translocation of hydrogenase through the membrane. A similar situation might exist in M. ciceri strains, where neither hydrogenase activity nor immunoblot signals were detected. In contrast, other studies on heterologous expression of hup genes in chickpea rhizobia showed a significant reduction in the amount of hydrogen evolved by nodules. Introduction of plasmid pIJ1008 into Rhizobium sp. (Cicer) strains resulted in the induction of significant hydrogenase activity (55). Although this plasmid has not been fully characterized at the molecular level, its ability to complement the Nod⁻ Fix⁻ phenotype of R. leguminosarum deletion strains (3) indicates that it contains a large number of genetic determinants for nodulation and nitrogen fixation, probably including mifA. Also, the B. japonicum hup cluster induces significant hydrogenase activity in Rhizobium sp. (Cicer) (25). Again, the hup cluster from B. japonicum contained in cosmid pHU52 includes the hoxA gene, which encodes the transcriptional regulator responsible for expression of hydrogenase structural genes (54). The lack of hup-specific regulatory genes in the minitransposon used here could contribute to the lower expression of hup genes observed in our work.

The use of antibiotic resistance genes as markers greatly facilitates the selection of recombinants in the laboratory. However, once the strains have been developed, removal of antibiotic resistance genes from the engineered strains is desirable for two reasons. First, public perception issues and legal regulations in some countries are against the presence of genes for antibiotic resistance in bacteria to be released into soils as legume inoculants. Second, the presence of antibiotic resistance marker determinants may impose an extra metabolic load on the cells, which could suppress the potential benefits associated with the introduced trait. In fact, competition analyses performed with E. coli have demonstrated that the insertion of antibiotic resistance markers into a particular gene does interfere with bacterial competitiveness for growth compared to that in strains bearing deletions on the same gene (30). The standard procedure reported here for elimination of the spectinomycin resistance gene allows the generation of strains acceptable for field dissemination. Furthermore, intermediate constructs generated during this process are useful for the characterization of the transposon insertion site.

An additional question regarding the use of the system developed in this work is the effect of the minitransposon insertion. In all cases tested in this study, inspection of the plants inoculated with the engineered strains revealed no differences in either shoot growth or nodulation by modified strains compared to those of plants inoculated with their corresponding wild-type counterparts. Ideally, such insertion should occur in a symbiotically silent site (2). However, the identification of this type of locus is difficult and must be made for each recipient strain, since the same site could be present in some rhizobia but not in others (24). In our approach, random insertion of TnHB100 should be followed by analysis of the insertion site. Such analysis can be performed by standard molecular biology techniques as presented here. In the example reported in this work, there is no information about the potential phenotype associated with the mutation in the identified open reading frame. The molecular characterization of the loci where the insertions have taken place, along with information obtained from analysis of the effects of the insertion on plant growth, should lead to the identification of engineered strains where the effect of the insertion is negligible. Confirmation of the lack of a negative effect associated with specific insertions might come from the analysis of control strains generated by removal of hup genes and part of the spectinomycin resistance gene. These strains would carry the same mutated loci without acquisition of new functional genes. A single construct, carrying part of the spectinomycin resistance gene and the DNA downstream from hypX, can be designed to mediate this modification in any TnHB100-containing strain through double recombination. This construct, currently being tested in our laboratory, will allow the generation of adequate control strains for productivity tests aimed at the field evaluation of the actual contribution of the hydrogenase system to legume productivity in different Rhizobium-legume systems.