Convergent Evolution of Amadori Opine Catabolic Systems in Plasmids of Agrobacterium tumefaciens

Abstract

Deoxyfructosyl glutamine (DFG, referred to elsewhere as dfg) is a naturally occurring Amadori compound found in rotting fruits and vegetables. DFG also is an opine and is found in tumors induced by chrysopine-type strains of Agrobacterium tumefaciens. Such strains catabolize this opine via a pathway coded for by their plasmids. NT1, a derivative of the nopaline-type A. tumefaciens strain C58 lacking pTiC58, can utilize DFG as the sole carbon source. Genes for utilization of DFG were mapped to the 543-kb accessory plasmid pAtC58. Two cosmid clones of pAtC58 allowed UIA5, a plasmid-free derivative of C58, harboring pSa-C that expresses MocC (mannopine [MOP] oxidoreductase that oxidizes MOP to DFG), to grow by using MOP as the sole carbon source. Genetic analysis of subclones indicated that the genes for utilization of DFG are located in a 6.2-kb BglII (Bg2) region adjacent to repABC-type genes probably responsible for the replication of pAtC58. This region contains five open reading frames organized into at least two transcriptional soc (santhopine catabolism) groups: socR and socABCD. Nucleotide sequence analysis and analyses of transposon-insertion mutations in the region showed that SocR negatively regulates the expression of socR itself and socABCD. SocA and SocB are responsible for transport of DFG and MOP. SocA is a homolog of known periplasmic amino acid binding proteins. The N-terminal half of SocB is a homolog of the transmembrane transporter proteins for several amino acids, and the C-terminal half is a homolog of the transporter-associated ATP-binding proteins. SocC and SocD could be responsible for the enzymatic degradation of DFG, being homologs of sugar oxidoreductases and an amadoriase from Corynebacterium sp., respectively. The protein products of socABCD are not related at the amino acid sequence level to those of the moc and mot genes of Ti plasmids responsible for utilization of DFG and MOP, indicating that these two sets of genes and their catabolic pathways have evolved convergently from independent origins.

Opines are produced by crown gall tumor cells, which are induced by pathogenic Agrobacterium spp. (8, 56). Genes for the biosynthesis of these compounds are encoded by the T region of Ti (tumor-inducing) or Ri (root-inducing) plasmids, which are present in the plant pathogens. During infection, the T region is transferred to plant cells and stably integrated into the nuclear genome. The genes for production of opines and plant hormones encoded in the integrated DNA are expressed, and opines are released from the transformed plant cells into soil, where they can be utilized by the pathogenic bacterium as carbon and energy sources (8, 42, 51). The genes for the utilization of the opines by the bacterium also are located on the virulence plasmids, but in segments outside of the T-DNA region. Opines play key roles in the interkingdom interactions between Agrobacterium spp. and infected plants, promoting selective growth of the pathogenic bacterium (15, 39, 45, 54), inducing Ti plasmid conjugal transfer (11, 42), acting as attractants for agrobacterial strains (27), and inhibiting growth of certain agrobacterial strains (16, 29).

Among the more than 20 known opines, the chrysopine family (5, 7) is produced by tumors induced by A. tumefaciens Chry strains, Ficus strains, and IIBV7 (5, 52). This family of Amadori-type compounds includes N-1-deoxy-d-fructosyl-l-glutamine (DFG, referred to elsewhere as dfg), commonly called santhopine; N-1-deoxy-d-fructosyl-l-glutamate (DFGA, referred to elsewhere as dfga); N-1-deoxy-d-fructosyl-5-oxo-l-proline (DFOP, referred to elsewhere as dfop); and chrysopine, the spiropyranosyl lactone of DFG (5). Chemically, these compounds are closely related to the mannityl opines (Fig. 1). DFG and DFGA are the deoxyfructosyl analogs of mannopine (MOP) and mannopinic acid (MOA), respectively, and chrysopine and DFOP are deoxyfructosyl analogs of agropine (AGR) and agropinic acid, respectively. These similarities suggest a close relatedness in catabolic pathways and also in the genes coding for the enzymes for their catabolism between the two families of opines. In this regard, DFG is an intermediate in the pathway for the catabolism of AGR and MOP coded for by the octopine/mannityl opine-type Ti plasmids such as pTi15955. MOP either converted from AGR (18, 19) or taken up by the MOP transport system encoded by mot genes (40) is oxidized to DFG by MOP oxidoreductase encoded by mocC (Fig. 1) (30). The product, DFG, is further degraded by functions encoded by MocD and MocE, the putative DFG deconjugase and a kinase, respectively (30). The mot and moc genes cluster to a 25-kb region of the octopine/mannityl opine-type Ti plasmids (9, 30).

FIG. 1.

Structures of opines belonging to the chrysopine family (A) and the mannityl opine family (B). MOP produced from AGR by MOP cyclase is oxidized to DFG by MOP oxidoreductase in MOP-utilizing A. tumefaciens strains.

Although the catabolism of the mannityl opines is coded for by mannityl opine-type Ti and Ri plasmids (30, 35), a second set of genes for utilization of DFG apparently is located elsewhere in the genome of many agrobacterial strains (53). Strain NT1, a derivative of C58 that lacks the Ti plasmid but harbors a large accessory plasmid called pAtC58, can utilize DFG as the sole carbon source. In contrast, UIA5, which lacks both plasmids, cannot utilize this compound (28, 53). This observation suggested that pAtC58 codes for transport and degradation of DFG.

In the present study, we identified and characterized the genes on pAtC58 that code for the utilization of the Amadori opine. These genes differ phylogenetically from the mot and moc genes required for catabolism of MOP and DFG, coded for by the octopine/mannityl opine-type Ti plasmids.

MATERIALS AND METHODS

Bacterial strains and plasmids.

The bacterial strains and plasmids used in the present study are listed in Table 1. Plasmid pSa-C was constructed by cointegrating the broad-host-range IncW vector pSa4ΔH (16) with pKS-C that contains mocC under control of the lac promoter of pUC19 (31). Plasmid pDSK-C was constructed by cointegrating pKS-C with pDSK519, a derivative of the IncQ plasmid RSF1010 (26), by using unique KpnI sites in these plasmids. Plasmid pYDH208 is a cosmid clone of pTi15955 that encodes all of the enzymes essential for the catabolism of MOP and AGR, as well as transporters for these opines (7). Plasmid pYDPH208 is a derivative of pYDH208 which has a nonpolar deletion mutation in agcA (19).

TABLE 1.

Bacterial strains and plasmids used in this study

Strains and plasmids	Relevant genotype and characteristicsa	Source or reference
Strains
Escherichia coli
LE392	el4− (McrA−) hsdR514 supE44 supF58 lacY1 or Δ(lacIZY)6 galK2 galT22 metB1 trpR55	44
S17-1	pro Res− Mod+ Mob+ Tpr Smr	47
Agrobacterium tumefaciens
NT1	pTiC58−, pAtC58+ derivative of the nopaline-type strain C58, DFG+	Our collection
UIA5	pTiC58−, pAtC58− derivative of the nopaline-type strain C58, DFG−	Our collection
15955	Wild-type isolate; catabolizes all mannityl opines	Our collection
Plasmids
pBS-Bg2	6.2-kb BglII fragment 2 from pCHAt56 cloned into pBluescript SK(−)	This study
pBSG2	Derivative of ColE1 containing a promoterless lacZ downstream of a unique BamHI site, Apr Gmr	This study
pBSG2-PA	ColE1 derivative in which the promoter region of socA is transcriptionally fused to a promoterless lacZ, Apr Gmr	This study
pBSG2-PR	ColE1 derivative in which the promoter region of socR is transcriptionally fused to a promoterless lacZ, Apr Gmr	This study
pCH-1	13.6-kb BglII-HindIII fragment from pCHAt56 cloned into pRK415	This study
pCHAt31	Cosmid clone of pAtC58, Tcr	This study
pCHAt56	Cosmid clone of pAtC58, Tcr	This study
pCH-Bg2	6.2-kb BglII fragment 2 from pCHAt56 cloned into pRK415	This study
pCH-Bg2ΔR	Derivative of pCH-Bg2 with a deletion in socR	This study
pCH-H1	11.2-kb HindIII fragment 1 from pCHAt56 cloned into pRK415	This study
pCH-P1	4-kb PstI fragment 1 from pCHAt56 cloned into pRK415	This study
pCH-P2	4-kb PstI fragment 2 from pCHAt56 cloned into pRK415	This study
pCP13/B	Broad-host-range IncP cosmid vector, Tcr	9
pDSK519	Broad-host-range IncQ cloning vector, Kmr	26
pDSK-Bg2	6.2-kb BglII fragment 2 from pCHAt56 cloned into pDSK519	This study
pDSK-C	Concatamer of pKS-C and pDSK519, Kmr	This study
pKNG101	Suicide vector with R6K ori, sacB+ Smr	25
pKS-C	Derivative of pUC19 containing mocC cloned under the lac promoter, Apr	31
pRK415	Broad-host-range IncP cloning vector, Tcr	26
pSa-C	Concatamer of pKS-C and pSa4ΔH, Cbr Cmr	31
pYDPH208	Derivative of pYDH208 with a non-polar deletion in agcA	19
pYDPH208::Tn3#10	Derivative of pYDPH208 with a Tn3HoHo1 insertion in mocC	17
pYDPH208::Tn3#1	Derivative of pYDPH208 with a Tn3HoHo1 insertion in mocD	17
pYDPH208::Tn3#94	Derivative of pYDPH208 with a Tn3HoHo1 insertion in mocE	17
pYDPH208::Tn3#5	Derivative of pYDPH208 with a Tn3HoHo1 insertion in the vector	17
pAtC58-socA::lacZ	Derivative of pAtC58, in which the promoter region of socA is transcriptionally fused with lacZ	This study
pAtC58-socR::lacZ	Derivative of pAtC58, in which the promoter region of socR is transcriptionally fused with lacZ	This study
pAtC58ΔsocR	socR deletion derivative of pAtC58	This study
pAtC58ΔsocR-socA::lacZ	Derivative of pAtC58ΔsocR, in which the promoter region of socA is transcriptionally fused with lacZ	This study
pAtC58ΔsocR-socR::lacZ	Derivative of pAtC58ΔsocR, in which the promoter region of socR is transcriptionally fused with lacZ	This study

^aAbbreviations: Tp^r, trimethoprim resistance; Ap^r, ampicillin resistance; Cb^r, carbenicillin resistance; Cm^r, chloramphenicol resistance; Km^r, kanamycin resistance; Gm^r, gentamicin resistance; Sm^r, streptomycin resistance; Tc^r, tetracycline resistance; DFG⁺, utilizes DFG as a sole carbon source.

Culture media, growth conditions, and chemicals.

Nutrient broth (Difco Laboratories, Detroit, Mich.) and L broth (LB) (Difco) were used as rich media for Agrobacterium spp. and Escherichia coli, respectively. AT minimal medium supplemented with 0.15% (NH₄)₂SO₄ (ATN) was used to grow strains of Agrobacterium for opine utilization studies (9). Noble agar (Difco; final concentration, 1.8%) was used to solidify the minimal medium. MOP was added to minimal medium at a final concentration of 5 mM for growth studies and at a final concentration of 1 mM for induction studies. Strains of E. coli were grown at 37°C, whereas strains of Agrobacterium were grown at 28 ^oC. For E. coli, ampicillin at 100 μg/ml, kanamycin at 25 μg/ml, and tetracycline at 10 μg/ml were used. For Agrobacterium, carbenicillin at 100 μg/ml, kanamycin at 100 μg/ml, gentamicin at 30 μg/ml, and tetracycline at 2 μg/ml were used. All antibiotics, mannopine, X-Gal (5-bromo-4-chloro-3-indolyl-β-d-galactopyranoside), and ONPG (o-nitrophenyl-β-d-galactopyranoside) were purchased from Sigma Chemical Company (St. Louis, Mo.). DFG was chemically synthesized as described previously (5).

DNA manipulation and transformation.

Large- and small-scale isolations of plasmid DNA from Agrobacterium cells were performed by a rapid alkaline lysis procedure as described previously (9). Restriction enzyme digestions and ligations were carried out as recommended by the manufacturer (Promega, Madison, Wis.). Transformation of Agrobacterium strains and E. coli was performed as previously described (4, 44, 47).

Isolation of pAtC58 from strain NT1.

Cells of NT1 from 500 ml of a late-exponential-phase culture in nutrient broth were harvested by centrifugation for 10 min at 4°C and 10,000 × g, and the pellet was resuspended in 20 ml of Agrobacterium wash buffer (0.5 M NaCl, 0.05 M Tris-HCl, 0.02 M Na₂EDTA [pH 8.0], 0.05% Na-Sarkosyl) (48). The suspension was recentrifuged, and the pellet was resuspended in 24 ml of solution I (50 mM glucose, 10 mM EDTA, 25 mM Tris-HCl; pH 8.0) containing lysozyme (5 mg/ml). The suspension was incubated at 37°C for 10 min, and 48 ml of fresh Solution II (1% sodium dodecyl sulfate [wt/vol] in 0.2 N NaOH) was added to lyse the cells. To the lysate, 2 M Tris-HCl (pH 7.0) was added with gentle mixing until the pH reached 7.0. Then, 10 ml of 5 M NaCl was added, and the mixture was extracted by gentle agitation once with 40 ml of phenol saturated with 3% NaCl. The emulsion was centrifuged in a Beckman GSA rotor for 30 min at 4°C at 8,000 rpm, the aqueous phase was removed, and DNA was precipitated and collected by centrifugation after the addition of 2 volumes of 95% ethanol, followed by incubation at −20°C for at least 1 h. Plasmid pAtC58 in the precipitated DNA was purified by two sequential centrifugations to equilibrium in CsCl-ethidium bromide (44).

Construction of a cosmid library of pAtC58.

A 60-μg portion of pAtC58 DNA was partially digested with Sau3AI, and the digested DNA fragments were size fractionated in a 5 to 40% (wt/vol) sucrose gradient by centrifugation at 26,000 rpm at 20°C for 24 h in a Beckman SW41 rotor. Fragments larger than 15 kb were ligated into the unique BamHI site of the cosmid vector pCP13/B (9), and the products were packaged into phage λ by using an in vitro packaging system (Promega) as recommended by the manufacturer. Clones were recovered by transfection into E. coli LE392, and transfectants were pooled and stored. Total plasmid DNA isolated from the pooled cells was introduced into E. coli strain S17-1 (47) by electroporation (4), and the pooled transformants were used to introduce the cosmid clones into A. tumefaciens UIA5 by conjugation.

Subcloning of regions associated with DFG utilization from cosmid clones of pAtC58.

Plasmids pCH-1, pCH-Bg2, pCH-H1, pCH-P1, and pCH-P2 were constructed by inserting a 13.6-kb BglII/HindIII partial-digest fragment, a 6.2-kb BglII (Bg2) fragment, an 11.2-kb HindIII (H1) fragment, a 4-kb PstI (P1) fragment, and a 4-kb PstI (P2) fragment (Fig. 2), respectively, from pCHAt56 into the broad-host-range IncP cloning vector pRK415 (26).

FIG. 2.

Physical map of the region of pAtC58 shared by two cosmid clones pCHAt31 and pCHAt56, which confer catabolism of DFG. Plasmids pCHAt31 and pCHAt56 are cosmid clones of pAtC58 that allow UIA5 harboring a mocC clone to utilize MOP. Each of the subclones, constructed as described in Materials and Methods, was transformed into UIA5(pDSK-C), and transformants were tested for growth on solid medium containing MOP as the sole carbon source. Growth was examined visually for 1 week and was scored as follows: −, no growth; +, slow growth; and ++, wild-type growth (i.e., it grew as well as strain 15955, the positive control).

Mannopine utilization.

Growth was assessed by visual inspection at daily intervals of colonies growing on solid AT medium containing MOP (5 mM) as the sole carbon source. Growth was recorded on the basis of comparisons with appropriate positive (strain 15955) and negative (strain UIA5) controls and scored as follows: −, no growth; +, poor growth; ++, good (wild-type) growth; and +++, growth better than wild-type (30).

Mutagenesis with Tn3HoHo1.

Plasmid pCH-1 was mutagenized with Tn3HoHo1 as described previously (49). This transposon carries a promoterless lacZ gene, and generates transcriptional and translational fusions when inserted into a gene in the appropriate orientation (49).

Opine uptake analysis.

Uptake of opines was estimated indirectly by measuring the disappearance of the substrates from liquid medium as previously described (31). Tested strains were grown in 100 μl of AT minimal medium containing the opine as sole carbon source at a final concentration of 1 mM, and the cultures were incubated at 28°C with shaking. After 4 h, cells were removed by centrifugation, and the supernatants were spotted onto Whatman 3MM filter paper. The filter paper strips were air dried, and the opine remaining in the samples was visualized by the alkaline silver nitrate staining method as described previously (10, 29). Opine-utilizing and -nonutilizing strains were included as positive and negative controls.

Nucleotide sequence analysis.

For DNA sequence analysis, a set of nested deletions of pBS-Bg2 was generated as described by Sambrook et al. (44). Exonuclease III, S1 nuclease, and Klenow fragment were purchased from MBI Fermentas (Amherst, N.Y.). Plasmid DNA was isolated by using the Wizard plus DNA purification system (Promega). Nucleotide sequences were determined by using an ABI-373A automated DNA sequencer (Perkin-Elmer Corp., Foster City, Calif.). The nucleotide sequences were assembled and open reading frames were identified by using the DNASIS program (Hitachi Software, Inc., Tokyo, Japan). Nucleotide sequences were examined for putative promoters and ribosome-binding site (RBS) sequences by visual inspection. Nucleotide and amino acid sequences were compared to those in the GenBank database by using the BLASTA algorithm (1). The nucleotide sequences of the soc genes and rep genes are deposited into the GenBank database under accession numbers AF151698 and AF283811, respectively. (While this article was in preparation, the nucleotide sequence of the entire genome of strain C58 was published by two groups [14, 55]. Our sequences of the soc and rep regions matched perfectly with those of pAtC58 deposited in GenBank under accession numbers AE007872 and AE008689.)

Construction of a socR deletion mutant of pAtC58.

A 4.1-kb segment spanning socB, socC, socD, and the 3′ end of socA was removed by using ExoIII from pBS-Bg2, which is a derivative of pBluescript SK(−) containing the 6.2-kb BglII fragment from pCHAt31 (Fig. 2). The resulting plasmid, pBS-BG2ND83, contains the intact socR gene, the intergenic region between socR and socA, and the 5′ end of socA. This plasmid was digested with EcoRV, which cuts at two sites in the middle of socR, and religated. The resulting plasmid, pBSΔsocR, contains an allele of socR with a 427-bp deletion, which was confirmed by nucleotide sequence analysis (data not shown). Plasmid pBSΔsocR was concatamerized with the sacB suicide vector pKNG101 (25) by using unique BamHI sites, to create pKNGΔsocR. This construction was introduced into NT1, and the resulting transformants were spread on AB minimal medium containing 0.5% glucose and 5% sucrose. Among the sucrose-resistant colonies, we obtained one clone that is sensitive to carbenicillin and streptomycin. This isolate, named NT1(pAtC58ΔsocR), contains a derivative of pAtC58 with the deletion allele of socR generated by a double-crossover event between pKNGΔsocR and pAtC58. The deletion was confirmed by determination of the nucleotide sequence of a PCR product of the socR region on pAtC58 in the strain (data not shown).

Construction of socR::lacZ and socA::lacZ fusions in pAtC58 and pAtC58ΔsocR.

A 775-bp DNA fragment between the 5′ ends of socR and socA, which contains divergently oriented promoters upstream to these two genes, was amplified by PCR with the following primers: 5′-CCGGATCCTCGCCCGAATTGATGAG-3′ and 5′-CTGGATCCCGAAACCGGTAAAGTTGC-3′. The resulting DNA fragment was digested with BamHI (recognition sites indicated by underlining) and ligated in both orientations into the unique BamHI site of the suicide vector pBSG2, which is a derivative of ColE1 and contains the gene for resistance to gentamicin and a unique BamHI site followed by a promoterless lacZ gene. The resulting constructs, pBSG2-P_R and pBSG2-P_A, contain the lacZ genes transcriptionally fused to the regions containing the promoter sequences upstream of socR and socA, respectively. These two plasmids were individually introduced into strains NT1 and NT1(pAtC58ΔsocR), and transformants resistant to gentamicin were selected. Because pBSG2 cannot replicate in Agrobacterium sp., these cells were assumed to have the plasmids incorporated into pAtC58 by single-crossover events, thereby generating transcriptional lacZ fusions to socR and socA in pAtC58, respectively. The constructions were confirmed by Southern hybridization analyses (data not shown).

Measuring β-galactosidase activity.

The β-galactosidase activities, scored as “−” (white), “+” (pale blue), or “++” (dark blue), were qualitatively measured by using X-Gal in solid ATN medium with or without MOP (1 mM). For quantitative measurements of β-galactosidase activity, cells to be tested were grown to early exponential phase in ATNG broth. The culture was split into two subcultures, with MOP at a final concentration of 1 mM added to one, and the two subcultures were grown in parallel for an additional 6 h. The levels of β-galactosidase activities, expressed as Miller units, were determined as described previously (30). All of these assays were repeated four times.

RESULTS

Rationale for the use of a mocC clone for intracellular conversion of MOP to DFG.

The chemical instability of DFG is a major obstacle to studies concerning the pathway and its genes responsible for transport and catabolism of this Amadori opine. Since MOP is catabolized via DFG, to overcome this technical barrier, we employed strains harboring a clone expressing mocC. The product of this gene catalyzes the conversion of MOP to DFG, allowing us to use the mannityl opine as the substrate to search for genes on pAtC58 responsible for DFG catabolism.

Isolation of cosmid clones of pAtC58 that allow UIA5 harboring mocC to utilize MOP as the sole carbon source.

After construction of a cosmid library of pAtC58, we isolated two cosmid clones, pCHAt31 and pCHAt56, that allowed A. tumefaciens UIA5(pSa-C), which lacks pAtC58, to grow by using MOP as the sole carbon source. Southern analyses showed that these two clones hybridized with the genome of A. tumefaciens NT1 but not with that of UIA5 (data not shown), indicating that they contain fragments from pAtC58. Restriction fragment analysis revealed that the inserts in these two clones share a ∼12.3-kb region in common (Fig. 2). A series of subclones was constructed, and each of the clones was tested for the ability to confer MOP utilization on UIA5(pSa-C). pCH-Bg2, that contains a 6.2-kb BglII (Bg2) fragment, confers slow growth with MOP as sole carbon source, whereas pCH-1 that contains a 13.6-kb BglII-HindIII fragment confers wild-type growth on the strain (Fig. 2). These results suggest that the 13.6-kb region contains all of the genes responsible for uptake of MOP and DFG and degradation of DFG.

Molecular genetic characterization of Tn3HoHo1-insertion derivatives of pCH-1.

We generated Tn3HoHo1 insertion mutations in pCH-1, and the constructs were introduced into UIA5 harboring a compatible mocC clone, pDSK-C. Each of the insertion mutants was tested for expression of β-galactosidase activity generated by fusion to the lacZ gene of Tn3HoHo1 in the presence or absence of MOP and for the ability to take up and utilize MOP as sole carbon source (Fig. 3A). Based on the phenotypes of these mutants, the 13.6-kb region could be divided into four regions (Fig. 3B). β-Galactosidase activities from the lacZ fusions and phenotypes of insertion mutants indicated that regions I and IV contain genes transcribed from right to left, whereas regions II and III contain genes transcribed from left to right.

FIG. 3.

Molecular genetic analyses of the 13.6-kb BglII-HindIII region of pAtC58 that confers utilization of DFG. (A) Phenotypes of Tn3HoHo1 insertion derivatives of pCH-1. The positions of the transposon insertions are indicated by the numbered vertical lines, and the direction of the lacZ-fusion generated by each insertion is indicated by arrowheads. Each derivative of pCH-1 was introduced into UIA5(pDSK-C), and transformants were tested for growth on MOP as described in the legend to Fig. 2. The transformants also were tested for the ability to take up MOP from liquid media as described in Materials and Methods and scored as follows: −, no MOP uptake; +, slow uptake; and ++, uptake as rapid as that by strain 15955. The β-galactosidase activity was examined on solid ATNG medium supplemented with X-Gal in the absence or presence of MOP at a final concentration of 1 mM as described in Materials and Methods. (B) Genetic organization of pCH-1 based on phenotypic characteristics of insertion mutants and nucleotide sequence analysis (GenBank accession numbers AF151698 and AF283811). Open reading frames identified in regions I, II, and III associated with utilization of DFG are indicated. The rep genes of pAtC58 in region IV also are indicated. The unfilled arrowheads represent the positions and directions of putative promoter regions. Numbers below the genes represent the coordinates of the region with respect to the complete nucleotide sequence of pAtC58 in the GenBank database under accession number AE007872 (14).

Among more than 500 derivatives examined, we identified only one insertion (insertion 196) in region I. UIA5(pDSK-C) harboring insertion 196 grew with MOP as the sole carbon source faster than wild-type strain 15955 or the strain harboring wild-type pCH-1. The lacZ fusion was expressed constitutively at a relatively high level (Table 2). Mutants with insertions in regions II and III failed to grow on medium containing MOP as the sole carbon source. lacZ fusions in region II at best responded only poorly to MOP, even in strains harboring the mocC clone. In mutants harboring the mocC clone β-galactosidase activities from lacZ-fusions in region III increased more than twofold upon the addition of MOP, whereas in the absence of the mocC clone, no significant increase in β-galactosidase activities were observed (Table 2). These results suggest that genes in this region are inducible and that DFG or one of its metabolites, but not MOP, is the inducer. Insertions in region IV had no effect on uptake of MOP or growth on media containing MOP as the sole carbon source.

TABLE 2.

Inducibility of soc genes by MOP in the presence or absence of a mocC clone

Host strain	pCH-1 mutanta	Insertion site	lacZ fusion geneb	Pres- ence of mocCc	β-Galactosidase activity (Miller units)d		Fold induc- tion
					−MOP	+MOP
UIA5	196	Region I	socR	−	1,055	895
				+	969	783
UIA5	482	Region III	socC	−	24	24
				+	23	60	2.6
NT1				+	6	117	19.5
UIA5	492	Region III	socD	−	28	24
				+	21	56	2.7
NT1				+	6	136	22.6
UIA5	49	Vector	Vector	−	20	24
				+	37	35

^aPositions of these insertions in pCH-1 are shown in Fig. 3.

^bThat is, the gene to which lacZ was fused.

^cmocC was provided by pDSK-C.

^dTested cells were grown to exponential phase in ATNG minimal medium at 28°C with shaking. The cultures were split into two portions, and MOP was either not added (−MOP) or added at a final concentration of 1 mM to one portion (+MOP). The cultures were grown for an additional 6 h, and β-galactosidase activities were measured and are expressed as Miller units (36). Values from A. tumefaciens NT1 are averages of results from two independent experiments. In the case of A. tumefaciens UIA5, values from a single representative experiment from four independent experiments are shown. However, the patterns of expression seen in this experiment are representative of the patterns seen in each repeat.

Region II confers transport of DFG and also MOP in the presence of a mocC clone.

Strains C58 and NT1 took up DFG but not MOP from media (Fig. 4A), whereas these same two strains harboring the mocC clone took up both opines (Fig. 4B and C). However, UIA5 failed to take up detectable amounts of MOP or DFG (Fig. 4A and C), even when provided with a mocC clone (Fig. 4B). UIA5(pCH-1) took up DFG, but not MOP, whereas UIA5(pCH-1) harboring the MocC clone took up both opines (Fig. 4A and C).

FIG. 4.

SocA and SocB confer transport of MOP and DFG. Uptake of MOP and DFG was measured semiquantitatively by detecting remaining opine in the spent culture fluid with silver nitrate as described in Materials and Methods. (A) Uptake of MOP by cultures of C58, NT1, UIA5, and UIA5 harboring various plasmids. The “196” denotes the derivative of pCH-1 with an insertion in socR as shown in Fig. 3. (B) Uptake of MOP by cultures of C58, NT1, and UIA5, all harboring the mocC clone, pDSK-C, and representative Tn3HoHo1-insertion derivatives of pCH-1. (C) Uptake of DFG by the same set of strains as in panel B, but lacking the mocC clone.

We also examined UIA5 harboring representative derivatives of pCH-1 with Tn3HoHo1 insertions in each of the four regions. Insertion 196 in region I allowed cells to take up both DFG and MOP even without mocC (Fig. 4A and C). A derivative of pCH-1 with an insertion in region II (insertion 420) did not allow detectable uptake of DFG even in strains that also harbored the mocC clone (Fig. 4B). UIA5 harboring pCH-1 with insertions in region III failed to take up MOP in the absence of the mocC clone (data not shown). However, addition of a mocC clone allowed these mutants to take up the opine weakly. When uptake of DFG was examined, a similar pattern was observed. UIA5 harboring pCH-1 with an insertion in region I rapidly took up the Amadori opine, whereas a strain harboring pCH-1 with an insertion in region III took up the opine slowly. Insertions in region II completely abolished the ability of UIA5 to take up DFG (Fig. 4C). These results indicate that region II is responsible for the transport of DFG and also MOP.

The 6.2-kb BglII fragment restores MOP catabolism to mocD and mocE mutants.

In strains harboring octopine/mannityl opine-type Ti plasmids, DFG produced from MOP by MocC is degraded by the products of mocD and mocE (30). Given that MOP is catabolized via DFG, we expected that defects in mocD and mocE should be complemented by the 6.2-kb BglII (Bg2) fragment if it codes for genes essential for the degradation of the Amadori opine. Cosmid clone pYDPH208 from pTi15955 encodes all of the functions necessary for uptake and catabolism of MOP (9, 19, 20, 30). In cells harboring this plasmid, MOP taken up by the Mot transport system is degraded by MocC, MocD, and MocE. Derivatives of pYDPH208 with insertions in mocC, mocD, mocE, or the vector (17) were introduced into UIA5 harboring pDSK-Bg2, a plasmid containing the 6.2-kb Bg2 fragment from pAtC58, and each of the resulting strains was examined for growth with MOP as the sole carbon source. pDSK-Bg2 complemented mutations in mocD and mocE of pYDPH208 but did not complement the defect in mocC (data not shown). These results suggest that the 6.2-kb BglII region in pDSK-Bg2 contains the genes necessary and sufficient for the degradation of DFG but lacks genes required for conversion of MOP to DFG.

Nucleotide sequence analysis of the 13.6-kb BglII-HindIII region: soc genes are distinct from moc genes.

Analysis of the nucleotide sequence of the 13.6-kb BglII-HindIII region of identified five open reading frames in regions I, II, and III, and these putative genes were named soc (for “santhopine catabolism”): socR, socA, socB, socC, and socD. The orientations of the genes are in agreement with predictions based on the lacZ-fusions (Fig. 3A); socR in region I is transcribed from right to left, whereas socAB in region II and socCD in region III are oriented from left to right (Fig. 3B). All of these genes initiate with ATGs and are preceded by sequences with strong matches with the RBS consensus sequence. The 139-bp intergenic region between the divergently oriented socA and socR genes does not contain sequences with matches to canonical −10 and −35 promoter elements. However, this region does contain a thirteen-base pair perfectly inverted repeat (IR) (Fig. 5A), with the initiation codon for socR located between the two arms. The IR is not related to other nucleotide sequences in the databases.

FIG. 5.

Construction of lacZ fusions to promoters in the region upstream of socR and socA. (A) Nucleotide sequence of the intergenic region between socR and socA. The putative promoter regions (P_A and P_R), RBSs, and the start codons for socR and socA are indicated. A 13-bp IR sequence is indicated by the dotted arrows. (B) Cloning of the region containing promoters P_R and P_A. A 755-bp PCR product of the region between the 5′ ends of socR and socA (▥), which contains the divergently oriented promoters, was cloned into pBSG2 in both orientations to construct pBSG2-PA and pBSG2-PR. In these plasmids, the promoters of socA and socR, respectively, are positioned upstream to a promoterless lacZ gene. Coordinate numbers are indicated in parentheses. (C) These constructions were introduced into NT1, and single crossovers were selected. As a consequence, the reporter gene is transcriptionally fused to P_A and to P_R, respectively, on pAtC58. Plasmids pAtC58ΔsocR-socA::lacZ and pAtC58ΔsocR-socR::lacZ, in which lacZ is fused to P_A and P_R, respectively, were constructed in the same manner.

SocR is a homolog of MocS and MocR (Table 3), which are repressors responsible for negative regulation of the moc genes of pTi15955 (22, 30). SocA is related to a number of periplasmic amino acid-binding proteins, especially those for histidine and lysine-arginine-ornithine (LAO), which are the substrate recognition components of ABC-type transporters (Table 3). The N-terminal half of SocB is similar to the permease components of several ABC-type amino acid transporters, whereas the C-terminal half is homologous to the ATP-binding protein components, most notably, GlnQ (Table 3). SocC is related to a large number of oxidoreductases of the GFO-IDH-MocA family, whereas SocD is related to a family of FAD-requiring oxidoreductases and shows reasonably strong homology to Faox-C, a fructosyl-amino acid amadoriase from Corynebacterium sp. (Table 3). With the exception of SocR, none of the Soc proteins show significant homology at the amino acid sequence level to the products of the moc genes responsible for the catabolism of MOP and its product, DFG, present on octopine/mannityl opine-type Ti plasmids.

TABLE 3.

Characteristics of proteins translatable from soc and rep genes

Protein (no. of amino acids)	Homologous protein	Source	Function	% Identity (% similarity)	Reference
SocR (383)	MocS	A. tumefaciens	Repressor in moc operon	39 (53)	30
	MocR	A. tumefaciens	Repressor in moc operon	36 (46)	30
SocA (277)	HisJ	E. coli K-12	Histidine-binding protein	32 (48)	24
	ArgT	Salmonella enterica	LAO-binding protein	30 (48)	41
SocB (502)a
N-terminal half	Lin2352	Listeria innocua	Amino acid permease	32 (54)	13
	SAV1858	Staphylococcus aureus	Glutamine permease	31 (55)	33
C-terminal half	GlnQ	Archaeglobus fulgidus	Glutamine transporter (ATPase)	58 (71)	32
	GlnQ	Thermoanaerobacter tengcongensis	Glutamine transporter (ATPase)	58 (72)	3
SocC (335)	Orf334	Sinorhizobium meliloti	Oxidoreductase	46 (62)	43
	BMEI2003	Brucella melitensis	d-Xylose dehydrogenase (NADP+ dependent)	40 (54)	6
SocD (375)	Faox-C	Corynebacterium sp.	Fructosyl-amino acid oxidase	54 (69)	21
RepC (404)	RepC	Agrobacterium rhizogenes	Replication protein	68 (81)	38
	RepC	Rhizobium etli	Replication protein	65 (80)	12
RepB (343)	RepB	Agrobacterium rhizogenes	Replication protein	37 (54)	37
	RepB	Agrobacterium tumefaciens	Replication protein	40 (55)	34
RepA (435)	RepA	Mesorhizobium loti	Replication protein	55 (76)	23
	RepA	Agrobacterium tumefaciens	Replication protein	57 (71)	34

^aThe deduced amino acid sequences of this protein consist of three distinct regions: an N-terminal region (amino acids 1 to 223), a junction region (amino acids 224 to 265), and a C-terminal region (amino acids 266 to 502). The junction region does not show any significant homology with genes in databases.

Region IV contains three genes (Fig. 3B), the protein products of which are closely related to those of the repABC-type replication systems from plasmids of Agrobacterium spp. and other α-proteobacteria (Table 3). The three genes are oriented opposite to those of the adjacent soc genes, a finding consistent with β-galactosidase activities from the Tn3HoHo1 insertion mutants. The nucleotide sequence of pAtC58 immediately to the left of region I did not identify genes with significant homologies with any known genes in the databases (14, 55).

SocR is responsible for the negative regulation of the soc operon.

The phenotypes of socR mutants (region I) suggested that this gene codes for a regulator that controls expression of the soc genes. To test this hypothesis, promoterless lacZ genes were transcriptionally fused to the 5′ ends of socR and socA in pAtC58 to construct pAtC58-socR::lacZ and pAtC58-socA::lacZ, respectively (Fig. 5). Cells of NT1 harboring each of these derivatives of pAtC58 exhibited low basal levels of β-galactosidase activities when grown with or without MOP (Table 4). When the mocC clone pRK-C was introduced into these strains, the resulting cells continued to express the reporter genes at a low basal level when grown in the absence of MOP. However, when grown with MOP, the levels of β-galactosidase activity in these two strains were significantly increased (Table 4). pAtC58ΔsocR-socR::lacZ and pAtC58 ΔsocR-socA::lacZ are socR deletion derivatives of the respective lacZ fusion derivatives. NT1 harboring either of these lacZ fusion derivatives of pAtC58ΔsocR expressed the reporters at high constitutive levels when grown in media with or without MOP regardless of the presence of a mocC clone (Table 4).

TABLE 4.

SocR represses transcription from promoters upstream of socR and socA and a catabolic intermediate of MOP is an inducer

pAtC58 derivatives	Relevant genotypea	Presence of mocC cloneb	Mean β-galactosidase activity ± SD (Miller units)c		Fold induction	Growth on MOPd
			−MOP	+MOP
pAtC58	socR+, socA::lacZ	−	12 ± 1.72	6 ± 0.37		−
pAtC58ΔsocR	ΔsocR, socA::lacZ	−	1,266 ± 55.26	1,189 ± 37.19		−
pAtC58	socR+, socA::lacZ	+	6 ± 1.53	135 ± 1.31	23	+
pAtC58ΔsocR	ΔsocR, socA::lacZ	+	906 ± 8.35	927 ± 12.66		++
pAtC58	socR+, socR::lacZ	−	60 ± 0.07	61 ± 0.72		−
pAtC58ΔsocR	ΔsocR, socR::lacZ	−	478 ± 17.90	460 ± 11.72		−
pAtC58	socR+, socR::lacZ	+	47 ± 2.88	91 ± 6.25	2	+
pAtC58ΔsocR	ΔsocR, socR::lacZ	+	404 ± 18.84	379 ± 0.41		++

^asocR deletion derivatives and lacZ fusions to the promoters P_A and P_R, located in the regions upstream to socA and socR, respectively, were constructed as described in Materials and Methods and also in Fig. 5.

^bmocC was provided in pRK-C.

^cCells were grown to exponential phase in ATN broth containing 0.5% glucose. The culture was split into two portions, and MOP was either not added (−MOP) or added to one portion at a final concentration of 1 mM (+MOP). The cultures were grown for an additional 6 hs. The β-galactosidase activity of each culture was measured as described previously (36). Values were derived from two independent experiments.

^dCultures were grown on ATN solid medium containing 5 mM MOP as the sole carbon source. Growth was visually observed over a 2-week period and recorded on the following scale: −, no growth; +, slow growth; and ++, wild-type growth (i.e., it grew as fast as strain 15955 on MOP).

Full induction of Soc expression requires another function of pAtC58.

The levels of induction by MOP seen in UIA5 derivatives harboring pCH-1 reporter fusions (Table 2) were considerably lower than those observed when the reporter fusions were contained within pAtC58 (Table 4). These observations suggested that pAtC58 provides some active factor required for full induction of the soc genes. To test this possibility, we introduced the two pCH-1-derived reporter fusions into NT1(pDSK-C) and determined the effect of MOP on the induction of socC and socD. As shown in Table 2, the presence of pAtC58 resulted in induced levels of expression two to three times higher than those observed in the UIA5 background. Moreover, basal levels of expression were lower in the NT1 background, resulting in relative levels of induction in response to MOP that were considerably higher than those observed in the UIA5 background (Table 2). However, the induction levels observed in the NT1 background were similar to those seen in strains in which the reporters were contained within pAtC58 (compare Tables 2 and 4).

DISCUSSION

DFG, an Amadori compound and an opine associated with the chrysopine-type Ti plasmids, is structurally related to MOP, an opine associated with the mannityl-opine-type Ti and Ri plasmids. Moreover, the catabolism of MOP coded for by the mannityl opine-type Ti plasmids proceeds through DFG as an intermediate (Fig. 1). Certain isolates of Agrobacterium lacking Ti, Ri, or known opine-catabolic plasmids can utilize this substrate (28, 30, 53), raising the question as to the nature of this DFG catabolic pathway and the relationship of the genes and their enzymes with those of the MOP catabolic system.

The DFG catabolism system of pAtC58 is composed of four genes. The first two, socA and socB, code for components of a periplasmic protein-dependent ABC-type transporter. Consistent with this conclusion, mutations in either gene block the uptake of DFG as measured by a substrate depletion assay (Fig. 4). In addition, the system apparently also transports MOP; strains harboring pCH-1 and a clone coding for MocC, the enzyme that oxidizes MOP to DFG, grow well with MOP as the sole carbon source (Fig. 2 and 3). Consistent with this conclusion, mutations in socA or in socB abolish growth on MOP (Fig. 3).

The DFG transporter is atypical among the ABC family in being composed of only two gene products. Based on sequence relatedness, we predict SocA to be the periplasmic substrate binding protein. As with most such elements, SocA shows low levels of sequence relatedness to other periplasmic binding proteins, but it is most closely related to HisJ of E. coli and to the multisubstrate LAO-binding protein, ArgT of Salmonella enterica serovar Typhimurium (Table 3). SocA is less closely related to MotA, the periplasmic binding protein of the Ti plasmid-coded MOP transporter, suggesting that the two opine recognition elements have arisen from different lineages. SocB, the apparent transporter protein, contains regions related to the trans-membrane transporter and the ABC-energizer components of classical ABC transport systems (Table 3). Presumably, this protein arose from a fusion between genes encoding the separate components. SocB is not closely related to other members of either family of transporter elements. However, among the best-match homologs are transporters for amino acids, and glutamine in particular (Table 3). This latter observation may be significant since both MOP and DFG contain a glutamine residue.

socC and socD, probably code for the enzymes responsible for degrading DFG. Consistent with this conclusion, strains harboring derivatives of pCH-1 with insertions in either gene fail to grow on MOP (Fig. 3). However, such mutants take up small amounts of both opines as one would expect for strains able to transport but not catabolize these substrates (Fig. 4). SocC is most closely related to a large family of NAD/NADP-dependent oxidoreductases. More significantly, SocD is most closely related to Faox-C, a fructosyl-amino acid oxidoreductase from Corynebacterium sp. strain 2-1-4 (Table 3). This enzyme degrades Amadori compounds by cleaving the imine linkage between the sugar and the amino acid (21).

Given their arrangements and spacings, we predict that the set of four soc genes comprises an operon expressed from a regulated promoter located directly upstream of socA. This 139-bp region contains a putative ribosomal binding sequence just upstream of socA but no recognizable promoter elements (Fig. 5). However, fusions of this region to a lacZ reporter clearly show that this sequence confers reasonably strong promoter activity (Fig. 5 and Table 4). That this promoter is regulated is suggested by our observation that lacZ fusions to socC and socD are expressed at higher levels in cells grown with glucose and MOP compared to cells grown on glucose only (Fig. 3 and Table 2). Strains with correctly oriented reporter insertions in socA and socB, the two transporter genes, do not exhibit such inducibility, suggesting that MOP or its degradation product, DFG, is the inducer. Directly upstream of socA and oriented in the opposite direction is a gene encoding a putative transcriptional regulatory element, SocR. Consistent with this assignment, insertions in or deletions of socR result in increased levels of growth on medium containing MOP as the sole carbon source (Fig. 3), as well as constitutive expression of reporter fusions to socA (Table 4). SocR, a member of the LacI family, is most closely related to MocS, the repressor that controls expression of the MOP catabolic regulon of the mannityl opine-type Ti plasmids (22, 30). However, sequence relatedness is not strong (Table 3), suggesting that SocR and MocS diverged from a common ancestor relatively early on. It is likely that SocR recognizes DFG and not MOP as its effector ligand. Although MOP induces expression of the soc genes, it does so only in strains that also express MocC, the enzyme that converts MOP to DFG (Tables 2 and 4).

While reporter fusions to soc genes within pAtC58 are inducible to high levels (Table 4), they are expressed at very low levels, even under inducing conditions, when located in the subclone, pCH-1 (Table 2). This difference in expression levels apparently is due to the requirement for some positive regulator coded for by pAtC58. Compared to UIA5, which lacks pAtC58, the soc reporters were induced to significantly higher levels when tested in NT1, a strain that also harbors this element (Table 2). The lower levels of basal expression of the two reporters observed in NT1 derivatives grown in the absence of MOP most likely is due to higher levels of SocR, which is expressed from both pAtC58 and the pCH-1 reporter construct present in these strains.

We had proposed, based on their similar chemistries and the fact that MOP is degraded through DFG, that the pathways for the catabolism of the Amadori compound and the mannityl opine would share enzymes in common (30). However, our results indicate that the pathway coded for by pAtC58 is quite different from that conferred by the mannityl opine-type Ti plasmids (Fig. 6). The two key enzymes, MocD and SocD, are unrelated at the amino acid sequence level. Moreover, in the Ti plasmid-encoded pathway, DFG produced from MOP by MocC is degraded to unknown products via the concerted action of MocD and MocE (28, 29). MocD resembles sugar aminotransferases, whereas MocE is related to sugar kinases, suggesting that a phosphorylated intermediate is involved in the catabolism of DFG. This phosphorylation is essential; mocE mutants fail to utilize MOP (30). SocD, on the other hand, is most closely related to an FAD-dependent fructosyl-amino acid amadoriase that catalyzes the oxidative cleavage of imine bonds to form a free primary amine and the glucosone of the sugar (21). No phosphorylated intermediate is involved. We suggest that SocD catalyzes a similar reaction, producing glutamine and the glucosone of fructose from DFG. SocC, which is related to a family of oxidoreductases, may convert the resulting glucosone to fructose by reducing the C1 aldehyde to the alcohol.

FIG. 6.

Comparisons in the metabolism of AGR, MOP, and DFG conferred by mannityl opine-type Ti plasmids and pAtC58 and in the organization of related genes. The genes responsible for the biosynthesis of mannityl opines in plant tumor cells are coded for in the T_R region of the Ti plasmids (10). In Agrobacterium spp., independent ABC-type transporters are responsible for the uptake of each of the three mannityl opines, and separate genes sets coding for enzymes are involved for each of the opines. Two unrelated sets of genes, one encoded by Ti plasmids and the other by pAtC58, confer catabolism of DFG. Genes on the T region that code for biosynthesis of the opines in plant cells are indicated by dotted arrows. Genes responsible for the transport of the mannityl and Amadori opines are indicated by open arrows, those responsible for catabolism of the compounds are indicated by solid arrows, and those responsible for the regulation of these genes are indicated by shaded arrows. Open reading frames with unidentified functions are indicated by open boxes. Genes: soc, DFG utilization gene; mot, MOP transport genes; agt, AGR transport genes; agcA, catabolic MOP cyclase; moc, MOP degradation genes.

Clearly, the two gene sets conferring the transport and catabolism of DFG are of different phylogenetic origins. However, the pathways are physiologically redundant. The soc gene set will compensate for mutations in mocD and mocE of the MOP catabolic pathway (data not shown). Why Agrobacterium spp. retain two independent pathways for the uptake and utilization of Amadori compounds remains a mystery. However, it is likely that strong selective pressures exist for the evolution and maintenance of the DFG catabolism systems. Amadori compounds, including DFG, form spontaneously in rhizospheres and in rotting plant materials (2). These compounds constitute an excellent source of carbon, nitrogen, and energy, and their utilization could play a role in competition for niche occupancy. In this regard, a number of soil fungi and bacteria have been isolated that either utilize Amadori compounds or contain amadoriases (21, 46, 50). Among the agrobacteria, many isolates utilize DFG regardless of biovar, their opine types, or even the presence of a Ti plasmid (52; C.-H. Baek, unpublished results). Regardless of the mechanisms involved, our results emphasize the metabolic diversity and complexity of A. tumefaciens. Given that the complete genome sequence of strain C58 is available (14, 55), this bacterium offers a rich resource for examining the genetic and the concomitant physiological components that go to make up a well-adapted and successful soil-saprophytic and plant-colonizing bacterium.