Lon protease of the {alpha}-proteobacterium Agrobacterium tumefaciens is required for normal growth, cellular morphology and full virulence

The ATP-dependent Lon (La) protease is a conserved protein present in diverse organisms, including archaea, bacteria, yeast, plants and animals. Lon from Escherichia coli (Lon_Ec) is an 87 kDa homo-oligomeric protein and contains highly conserved N-terminal ATPase and C-terminal protease domains (Goldberg et al., 1994). Lon_Ec plays a primary role in the degradation of many abnormal proteins as well as unstable regulatory proteins. These latter targets include the λ N protein, the SulA cell division regulator, the positive regulator of capsule synthesis, RcsA, and the F plasmid antidote protein CcdA (Gottesman, 1996). lon mutants of E. coli exhibit a pleiotropic phenotype that includes increased sensitivity to UV irradiation (Howard-Flanders et al., 1964), cell division defects resulting in filamentation (Howard-Flanders et al., 1964), mucoidy (Markovitz, 1964) and reduced degradation of various abnormal proteins (Gottesman & Zipser, 1978). The phenotypic properties of lon mutants are the result of a decreased ability to degrade specific regulatory proteins; SulA in the case of filamentation and RcsA in the case of overproduction of capsular polysaccharide.

Lon plays important roles in the regulation of developmental functions in a number of other bacteria, including normal cell morphogenesis and developmental progression in Caulobacter crescentus (Wright et al., 1996), fruiting body formation in Myxococcus xanthus (Gill et al., 1993) and lateral flagellar biosynthesis in Vibrio parahaemolyticus (Stewart et al., 1997). Recent studies have shown that Lon from Brucella abortus functions as a general stress-response protease and is required for wild-type virulence during the initial stages of infection in BALB/c mice (Roberston et al., 2000). In Sinorhizobium meliloti, Lon is involved in the regulation of exopolysaccharide synthesis and is required for effective nodulation of alfalfa (Summers et al., 2000).

The phytopathogenic α-proteobacterium Agrobacterium tumefaciens, a very close relative of S. meliloti, causes crown gall tumours on susceptible plants by transferring T-strand DNA from its tumour-inducing (Ti) plasmid into a susceptible host plant cell. This bacterium has served as a model for studies concerning a number of important biological phenomena, including plasmid conjugation, pathogenhost signalling, Type IV secretion, trans-kingdom gene transfer and quorum sensing. Given the requirement for Lon protease in microbehost interactions of other α-Proteobacteria, we constructed a lon null mutant of A. tumefaciens and characterized the effect of this mutation on a number of traits, including growth, cellular morphology and pathogenicity.

Bacterial strains, media and growth conditions.
Strains of A. tumefaciens and E. coli and the plasmids used in this study are listed in Table 1. Strains of E. coli were grown at 37 °C in Luria broth (LB; Invitrogen) or in minimal A glucose medium (Miller, 1972), while strains of A. tumefaciens were grown at 28 °C in low salt LB, in MG/L medium (Cangelosi et al., 1991), in AB minimal medium (Chilton et al., 1974) supplemented with 0·2 % mannitol as sole carbon source (ABM) and on nutrient agar (NA; Difco). All liquid cultures were grown with agitation to ensure adequate aeration. Growth in liquid cultures was monitored turbidimetrically using a KlettSummerson colorimeter fitted with a red filter or with a Spectronic 20 spectrophotometer at 600 nm. Antibiotics were used at the following concentrations in µg ml¹: for E. coli, kanamycin, 50; tetracycline, 10; ampicillin, 100; and gentamicin, 10; for A. tumefaciens, kanamycin, 50; tetracycline, 2; carbenicillin, 100; and gentamicin, 50. Calcofluor (Sigma) was added to LB plates at a final concentration of 200 µg ml¹. L-Arabinose was added at a final concentration of 0·1 % to induce expression of the lon gene of A. tumefaciens from P_BAD.

Table 1. Bacterial strains and plasmids used in this study

Plasmids.
Plasmid pBADlon, a derivative of pBAD22 (Guzman et al., 1995) expressing Lon from A. tumefaciens C58 (Lon_At) under the P_BAD arabinose-controlled promoter, was constructed as follows. The lon gene (lon_At) was amplified from A. tumefaciens NTL4 chromosomal DNA with Pfu DNA Polymerase (Stratagene) and cloned into the XbaI site of pBAD22. A clone containing lon_At correctly oriented downstream to the BAD promoter was identified and named pBADlon (Table 1). A second PCR-amplified lon gene (lon_At) along with its own promoter region was cloned into the XbaI site of pAW50 (Wise et al., 2005), creating plasmid pAWlon. This vector replicates at very low copy number in A. tumefaciens (Wise et al., 2005). Site-directed mutations were introduced onto lon_At using the QuikChange Kit (Stratagene), according to the instructions of the manufacturer. Briefly, for construction of pBADlonK364Q, containing a mutation at lysine-364 in the ATP-binding motif, primers 5'-CTCCCGGCGTCGGCCAGACCTCGCTCGCCAAG-3' and 5'-CTTGGCGAGCGAGGTCTGGCCGACGCCGGGAGG-3' were used. pBADlonS680A, carrying a mutation at the active site serine-680, was constructed by using primers 5'-CCGAAGGACGGACCGGCCGCCGGTGTTGCCATG-3' and 5'-CATGGCAACACCGGCGGCCGGTCCGTCCTTCGG-3'. Similar strategies were used to introduce the two mutations into lon_At cloned in pAW50. Cloned inserts and the mutant alleles were confirmed by automated DNA sequencing performed at the Biotechnology Center of the University of Illinois at Urbana-Champaign.

Genetic manipulations.
Plasmids were introduced into E. coli by CaCl₂-mediated transformation and into A. tumefaciens strains by electroporation or by biparental mating using E. coli S17-1 (Simon et al., 1983).

Disruption of the A. tumefaciens chromosomal lon gene.
Genomic DNA was prepared from an overnight culture of A. tumefaciens NTL4. A 4988 bp chromosomal fragment consisting of the entire coding sequences of lon and clpX along with the intergenic region between the two genes and an 879 bp sequence downstream from lon was amplified using PfuTurbo DNA polymerase (Stratagene) (Fig. 1a). The PCR product was cloned directly between the BamHI and XbaI sites of pUC19 to create pSlon (Fig. 1a). The gentamicin resistance cassette from pMGm (Murillo et al., 1994) was excised by PstI, made blunt by T4 DNA polymerase and inserted into the unique MscI site within the lon gene in pSlon at a position corresponding to amino acid residue alanine-532, creating pSlonG (Fig. 1a). The BamHIXbaI fragment containing clpX and the disrupted lon gene was excised from pSlonG and cloned between the BamHI and SpeI sites of pSR47s (Merriam et al., 1997). The resulting plasmid, pSRlonG, was transformed into S17-1 λpir and the resultant strain was mated with NTL4 as described previously (Cook & Farrand, 1992). NTL4 carrying the chromosomal disruption in lon was selected by plating on medium containing the appropriate antibiotics and 5 % sucrose. Allelic exchange of the mutant lon gene for the wild-type allele was confirmed by Southern hybridization (data not shown). A similar strategy was used to construct NTL4Δfla, a fla, non-motile strain of A. tumefaciens in which the three tandem flagellum genes flaA, flaB and flaC were replaced with a tetracycline resistance cassette.

(19K): Fig. 1. The lon gene of A. tumefaciens C58 and its protein product. (a) Genetic organization of the lon region from the chromosome of A. tumefaciens strain C58. The arrows indicate the direction of transcription. The fragment consisting of the entire coding sequences of lon and clpX along with the intergenic region between the two genes and an 879 bp sequence downstream from lon was cloned between the BamHI and XbaI site of pUC19 to create pSlon. A gentamicin resistance cassette was inserted into the unique MscI site to disrupt the coding sequence of the lonAt gene in pSlon, creating pSlonG. (b) Location of the two major consensus domains in LonAt. The ATPase Walker A motif and the proteolytic active site serine are shown as well as the sites of the two amino acid substitution mutations, LonK364Q and LonS680A. (c) Proteins in total cell lysates of A. tumefaciens NTL4 (lon+) and NTS2 (lon) were separated by 7·5 % SDS-PAGE and Lon was detected by immunoblotting with polyclonal rabbit antiserum directed against B. abortus Lon as described in Methods.

UV sensitivity.
Sensitivity to UV irradiation was assayed as described by Miranda & Kuzminov (2003). Briefly, a fresh overnight culture was diluted 100-fold into 2 ml ABM and grown with shaking at 28 °C to 2x10⁸ cells ml¹. Tenfold serial dilutions were made in 0·9 % NaCl and 10 µl volumes of a set of the serial dilutions were spotted in rows of six onto a square Petri dish containing LB agar. The liquid in each spot was allowed to dry down, the plates were partially covered with a screen and exposed to a gradient of doses of UV light in the direction perpendicular to the dilution gradient, so that every dilution column (from 10¹ to 10⁶) received its own dose. A UV cross-linker (Amersham-Pharmacia), in which all the lamps except the central one were removed and 90 % of the remaining lamp was shielded, was used to deliver precise doses of UV (measured by the internal UV sensor). Immediately after exposure, the plate was covered with aluminium foil and incubated at 28 °C for 24 h. The titre of the culture at the zero dose was used to determine the survival at various UV doses.

Motility and chemotaxis assays.
Motility assays were conducted in ABM medium solidified with 0·3 % agar as described by Ding & Christie (2003). Cell cultures were normalized to an OD₆₀₀ of 0·5 and a 2 µl volume of each strain was inoculated onto the surface of the motility plates. The plates were examined after 12, 24 and 48 h incubation at 28 °C. Chemotaxis assays were performed on 0·3 % soft agar plates without carbon sources. A sterile Whatman paper disk saturated with 15 % (w/v) sucrose, glucose or mannitol was placed at the centre of the agar plate. Two microlitres of each cell culture normalized to an OD₆₀₀ of 0·5 was inoculated onto the swarm plate 4 cm from the paper disk. The chemotaxis plates were maintained at 28 °C and examined throughout a 96 h incubation period.

Electron microscopy.
Agrobacterium strains were grown overnight with shaking at 28 °C in LB containing appropriate antibiotics to an OD₆₀₀ of approximately 0·8. One millilitre of each culture was washed three times and resuspended in sterile PBS buffer. Cells from the washed cultures were absorbed to Formvar-coated nickel grids (EM sciences) for 1 min. Excess culture was blotted with Whatman filter paper and the grid was placed face down on a drop of 1 % uranyl acetate (EM Sciences) for 1 min. Excess stain was blotted with Whatman filter paper, the grid was air-dried and viewed with an LEO 906e transmission electron microscope operating at an accelerating voltage of 80 kV.

β-Galactosidase assay.
Production of β-galactosidase by E. coli strains grown in A medium was quantified using a modification of the Miller method (Miller, 1972). Activity was expressed as units of enzyme per 10⁹ c.f.u. Samples were assayed in triplicate and the experiments were repeated three times. Results from a single representative experiment are shown.

RNA preparation.
Cultures of A. tumefaciens strains were grown to an OD₆₀₀ of 0·5 in ABM medium at 20 °C or shifted to 37 °C and sampled after 5, 10, 15 and 30 min incubation at the elevated temperature. Total RNA was isolated using Trizol reagent (Invitrogen) as follows. Each 50 ml culture sample was rapidly chilled on ice and the cells were collected by centrifugation. The cell pellet was homogenized in 5 ml hot Trizol reagent (65 °C) and the mixture was shaken at 65 °C for 10 min. Chloroform (1 ml) was added, the tube was shaken vigorously by hand for 15 s and incubated at room temperature for 23 min. The phases were separated by centrifugation at 12 000 g for 15 min at 4 °C and the upper phase was removed to a new tube. One volume of isopropyl alcohol (2·5 ml) was added, mixed well and the mixture was incubated at room temperature for 10 min. The precipitated RNA was collected by centrifugation at 12 000 g for 10 min at 4 °C. The supernatant was removed and the pellet was washed once with 5 ml 75 % ethanol and air-dried to near completion. The RNA was dissolved in 300 µl RNase-free water and the preparation was treated with RNase-free DNase [RQ1 DNase (Promega)] to remove contaminating DNA. The preparation was extracted with phenol/chloroform to remove the DNase. The RNA was precipitated with isopropyl alcohol, collected by centrifugation, dried to near completion and dissolved in 100 µl RNase-free water. The RNA concentration was quantified using a Lambda 3B spectrophotometer (Perkin Elmer).

RNA blotting and hybridization.
RNA samples (20 µg) were separated on 1·5 % agarose gels containing 2·2 M formaldehyde and transferred to a positively charged nylon membrane (Roche) by capillary action in 20x SSC (3 M NaCl, 0·3 M tri-sodium citrate). An 887 bp groEL DNA probe and a 1·2 kb lon_At DNA probe were purified from agarose gels and labelled with digoxigenin [(DIG)-11-dUTP] using the DIG-High Prime labelling kit (Roche). DIG labelling, prehybridization, hybridization, posthybridization washes and detection with chemiluminescent substrate were performed according to the instructions of the manufacturer (Roche). DIG-labelled RNA molecular mass marker set II (Roche) was used as the RNA ladder.

Immunoblotting.
Cultures of A. tumefaciens were grown in ABM at 20 or 37 °C as described above. Proteins in total cell lysates were resolved by 7·5 % SDS-PAGE polyacrylamide gels loaded on a per cell equivalent basis and the separated proteins were transferred to nitrocellulose membrane after electrophoresis. Immunoblots were developed with polyclonal rabbit antiserum directed against Brucella abortus Lon obtained from Dr R. Martin Roop II (Department of Microbiology and Immunology, Louisiana State University Health Sciences Center, USA). Primary antibody was detected with horseradish peroxidase (HRP)-conjugated monkey anti-rabbit antiserum. Antibodyantigen interactions were visualized by chemiluminescence with an ECL kit (GE Healthcare). The intensity of protein bands was quantified by digital scanning and analysis of developed X-ray films with NIH Image 1.63 software.

Virulence assays.
Tumorigenesis was assessed on leaves of Kalanchoë diagremontiana using the culture dilution method as described by Nair et al. (2003). Strains were grown in MG/L overnight at 28 °C with the appropriate antibiotics. Cells were collected by centrifugation, resuspended in 0·9 % (w/v) NaCl to an OD₆₀₀ of 1·0 and further diluted as necessary in 0·9 % NaCl. Two-centimetre-long wounds made with an 18-gauge needle on the youngest expanded leaves of a 5-week-old K. diagremontiana plant with three pairs of leaves were inoculated with 3 µl of these dilutions. Tumour formation was monitored over a period of 25 days. Assays were done in triplicate and the experiments were repeated three times.

A. tumefaciens lon locus
Based on the whole-genome sequence of A. tumefaciens C58 (Wood et al., 2001), the lon_At gene (Atu1261), located on the circular chromosome, is 2418 bp in length and encodes a protein of 805 aa with a calculated molecular weight of 88 750. Lon_At shares 86 % amino acid identity and 92 % similarity with the Lon protease from S. meliloti (Lon_Sm) (Summers et al., 2000), 81 % identity and 90 % similarity with Lon from B. abortus (Lon_Ba) (Roberston et al., 2000), and 61 % identity and 75 % similarity with Lon_Ec from E. coli (Chin et al., 1988). The deduced amino acid sequence of Lon_At contains a typical ATP-binding motif, GPPGVGK₃₆₄T and KAKKANPLFLLD (Chin et al., 1988) at amino acid coordinates 358365 and 414425, respectively, and a consensus serine protease active site motif IHVHVPEGATPKDGPS₆₈₀AG (Amerik et al., 1991) at amino acid coordinates 665682 (Fig. 1b). These features are conserved in all known Lon proteases. The A. tumefaciens lon gene is preceded by an ORF encoding a homologue of ClpX, another ATP-dependent protease (Fig. 1a), which is separated from lon_At by a 414 bp intergenic region predicted to encode the promoter of the A. tumefaciens lon gene. Located 275 bp downstream from the lon gene, a hupB homologue could encode a histone-like protein whose counterpart in E. coli is involved in regulating transcription by constraining DNA supercoils and DNA accessibility to regulatory proteins (Drlica & Rouvière-Yaniv, 1987).

The lon mutation impairs cell growth
To explore the role of Lon protease in A. tumefaciens, we constructed a null mutant by inserting a gentamicin resistance cassette into the lon gene by double-crossover homologous recombination as described in Methods (Fig. 1a). On the selection plates, we noted two obviously different-sized colonies, both of which grew slower than the wild-type parent. One such clone, named NTS1, formed very small colonies and grew very slowly. The second, named NTS2, formed relatively larger colonies and grew faster than NTS1. When streaked out for single colony isolation, NTS1 gave two colony sizes, the small one like its parent and larger sized colonies resembling NTS2. When similarly streaked out, NTS2 gave colonies of homogeneous size which grew faster than those of NTS1, but slower than the wild-type parent. Southern blot analysis indicated that the lon gene of both NTS1 and NTS2 was disrupted as predicted (data not shown). Strain NTL4 grew with a doubling time of about 2·5 h in ABM minimal medium (Fig. 2). However, strains NTS1 and NTS2 grew with a doubling time of around 8 and 4 h, respectively, in this medium (Fig. 2). To verify that the slow growth phenotype of the lon mutants is due to the lon mutation, the two strains were complemented with lon_At cloned with its own promoter in the low-copy-number plasmid pAW50 (Table 1). Derivatives of NTS1 and NTS2 harbouring pAWlon exhibited growth rates similar to that of the wild-type parent (Fig. 2). Viability assays showed that the mutant gave the same number of c.f.u. as its parent strain at the equivalent levels of turbidity (data not shown). Because of its consistent growth properties, we chose NTS2 for further studies. As assessed by Western analysis, NTS2 failed to produce any protein that interacts with the Lon antiserum (Fig. 1c). The wild-type strain NTL4 yielded a single reactive protein with a mobility corresponding to 89 kDa.

(21K): Fig. 2. Growth of A. tumefaciens wild-type and lon mutant strains in ABM minimal medium. Turbidity values of cultures, expressed as Klett units, were determined at the given times. Data points are from a single culture for each strain and are representative of one of three experiments. , Wild-type strain NTL4; , lon mutant NTS1; , NTS1(pAWlon); •, lon mutant NTS2; , NTS2(pAWlon).

Exopolysacchride production
Succinoglycan is the major acidic exopolysacchride produced by A. tumefaciens and can be easily detected by the binding of the optical brightening agent Calcofluor. When streaked out on LB medium containing Calcofluor, colonies of NTS2 fluoresced under UV light with an intensity indistinguishable from those of NTL4 (data not shown).

UV sensitivity and cellular morphology
lon mutants of E. coli exhibit enhanced UV sensitivity and filamentation. Strain NTS2 was no more sensitive to UV irradiation at doses ranging from 20 to 60 J m² than its parent, NTL4 (data not shown). Unlike lon mutants of E. coli, NTS2 did not produce filaments even following UV irradiation (Fig. 3a, panels 1 and 2, and data not shown). However, as seen by phase-contrast microscopy, the lon mutant, grown under normal conditions, displayed obvious differences in cell morphology (Fig. 3a, panel 2). Electron microscopic examination showed that cells of the wild-type strain NTL4 exhibited a typical short rod shape (Fig. 3b, row 1). However, approximately 8085 % of the mutant cells were branched, appearing as distinct Y shapes (Fig. 3b, row 2). NTS2 harbouring pAWlon took on a near wild-type shape, although the complemented cells exhibited a somewhat swollen appearance (Fig. 3a, panel 3, and Fig. 3b, row 3). Plasmids expressing the two Lon_At mutants, LonK364Q and LonS680A, did not restore normal cell morphology (Fig. 3a, panel 4, and data not shown).

(100K): Fig. 3. A lon mutation affects cell morphology in A. tumefaciens. (a) Phase-contrast microscopy; (b) electron microscopy. Cells of wild-type strain NTL4 (a, 1; b, row 1) show a typical short rod shape while most cells of the lon mutant NTS2 (a, 2; b, row 2) are branched, appearing as Y shapes. Complementation of the lon mutant cells by expression of a wild-type lon gene from pAWlon restored the wild-type cell morphology (a, 3; b, row 3). No complementation was observed in the lon mutant complemented with the clone expressing LonAtS680A (a, 4). The arrows in (a) 2 and 4 indicate aberrant cell shapes. The bars in (b) indicate 1 µm.

Motility and chemotaxis
Given its aberrant cell morphology, we assessed the influence of the lon mutation on motility and chemotaxis using soft agar medium. As shown in Fig. 4(a), the fla strain NTL4Δfla, which is deleted for the three flagellum genes flaA, flaB and flaC, is completely non-motile. In contrast, wild-type strain NTL4 is strongly motile. The lon mutant NTS2 is also motile, although the zone of spreading is smaller than that of its parent (Fig. 4a). The complemented mutant NTS2(pAWlon) is as motile as the wild-type NTL4. When viewed by phase-contrast microscopy, the percentage of NTS2 cells that were motile appeared to be similar to that of its wild-type parent (data not shown). The wild-type strain NTL4, the mutant NTS2 and the complemented mutant NTS2(pAWlon) all exhibited chemotaxis toward glucose (Fig. 4b), as well as toward sucrose and mannitol (data not shown).

(46K): Fig. 4. The lon mutant of A. tumefaciens is motile and exhibits chemotaxis. (a) Motility was assayed on 0·3 % agar plates as described in Methods. (b) Taxis by the strains toward glucose applied to a Whatman filter disk at the centre of plate was assessed as described in Methods.

Genetic complementation of an E. coli lon mutant
We determined whether Lon_At is equivalent to its E. coli homologue using two tests. In the first, we examined whether Lon_At could correct the SulA-associated filamentous phenotype exhibited by E. coli lon mutants. E. coli HDB98 (lon510) formed long filaments (Fig. 5a, panel 2), while the strain expressing Lon_At exhibited normal cell morphology (Fig. 5a, panel 3). However, neither of the two Lon_At mutants complemented the cell division defect (Fig. 5a, panel 4, and data not shown). We also assessed the ability of Lon_At to complement the defect in regulation of cpsB by RcsA. The Lon⁺ E. coli strain HDB97, which contains a cpsB : : lacZ fusion, expressed very little β-galactosidase activity, whereas its near isogenic lon mutant, HDB98 in which RcsA is not turned over, expressed the reporter at a high level (Fig. 5b). Expression of Lon_At from pBADlon in HDB98 resulted in a strong depression of expression of the cpsB : : lacZ reporter while the same strain expressing either of the two Lon_At mutants exhibited high levels of β-galactosidase activity characteristic of the host lon mutant (Fig. 5b).

(48K): Fig. 5. LonAt complements the filamentous and mucoid phenotypes of a lon mutant of E. coli. (a) Phase-contrast microscopy of strains: 1, HDB97 (lon+); 2, HDB98 (lon); 3, HDB98(pBADlon); 4, HDB98(pBADlonS680A). (b) Cultures of E. coli HDB97 (cpsB : : lacZ) and its lon mutant HDB98 with or without a lonAt-expressing plasmid as noted, were grown in minimal A medium with (black bars) or without (white bars) 0·1 % arabinose. Cultures were sampled and assayed for β-galactosidase activity.

Transcription of the A. tumefaciens lon gene is induced by growth at elevated temperature
In E. coli, lon is part of the heat-shock regulon. We assessed the effect of a temperature shift from 20 to 37 °C on the expression of lon_At in A. tumefaciens using two assays. In the first, we examined the level of mRNA transcripts of lon accumulated in the cells upon shifting to elevated temperature. In Northern blot analysis, levels of groEL mRNA, which appeared as two bands at 2·1 and 1·7 kb, increased to a maximum after 15 min incubation at the elevated temperature (Fig. 6a). In a similar analysis using the lon probe, the lon-specific mRNA was not detectable in cells grown at 20 °C, but produced a strong signal at 2·5 kb in cells grown for 15 min at the elevated temperature (Fig. 6b). The lon mRNA was not detectable in cells grown for 5 and 10 min at 37 °C and disappeared from cells grown for 30 min at the elevated temperature (Fig. 6b). In the second, we evaluated the level of Lon protein after temperature shift using an immunoblot assay. The amount of Lon protein increased slightly but reproducibly after heat shock, reaching its maximum at 15 min followed by a small decrease in level after 30 min (Fig. 6c and d).

(47K): Fig. 6. Expression of lonAt responds to elevated temperature. (a, b) Northern analysis of the groEL and lon genes. Total RNA of A. tumefaciens was isolated before and at different times (5, 10, 15 and 30 min) after raising the incubation temperature from 20 to 37 °C. An 887 bp groEL DNA fragment and a 1·2 kb lon DNA fragment were labelled with digoxigenin (DIG)-11-dUTP and used as probes for hybridization. (a) Hybridization with the groEL probe; (b) hybridization with the lon probe. (c) Immunoblot analysis of the levels of Lon protein in cells before and at different times (5, 10, 15 and 30 min) after raising the incubation temperature from 20 to 37 °C. Proteins in total cell lysates were separated by 7·5 % SDS-PAGE and Lon was detected by immunoblotting with polyclonal rabbit antiserum directed against B. abortus Lon as described in Methods. (d) The intensity of protein bands in (c) was quantified by digital scanning and analysis of developed X-ray films with NIH image 1.63 software. Experiments were repeated two times and data are representative of one of two experiments.

The lon mutant is highly attenuated for virulence
Given that lon mutants of B. abortus and S. meliloti exhibit defects in interaction with their eukaryotic hosts (Roberston et al., 2000; Summers et al., 2000), we determined if the lon mutation influences the virulence of A. tumefaciens. Strains NTL4, NTS2 and their derivatives harbouring pTiC58 or pTiR10 were inoculated onto K. diagremontiana leaves at decreasing cell concentrations as described in Methods. Strains NTL4 and NTS2 which lack Ti plasmids were avirulent (Fig. 7). Strains NTL4(pTiC58) and NTL4(pTiR10) induced tumours at all infection doses, although NTL4(pTiC58) yielded less tumour mass than did NTL4(pTiR10) at the lowest inoculation dose (OD₆₀₀=0·01). However, NTS2(pTiC58) and NTS2(pTiR10) failed to induce tumours at the lowest inoculation dose (OD₆₀₀=0·01) and induced much less tumour mass at the higher inoculation doses (OD₆₀₀=0·1 and 1·0, respectively) as compared to the wild-type parent (Fig. 7). Introducing pAWlon into strains NTS2(pTiC58) and NTS2(pTiR10) restored wild-type levels of virulence to the lon mutant harbouring either Ti plasmid (Fig. 7). Mutations in the ATPase or the proteolytic domains of Lon expressed from pAWlonK364Q or pAWlonS680A abolished complementation of the attenuated phenotype (Fig. 8b).

(77K): Fig. 7. The lon mutant of A. tumefaciens is attenuated for virulence. Strains were grown and inoculated onto K. diagremontiana leaves as described in Methods. Wounds in leaves were inoculated with strains diluted to population sizes indicated by the OD600 values at the bottom: A, NTL4; B, NTS2; C, NTL4(pTiC58); D, NTL4(pTiR10); E, NTS2(pTiC58); F, NTS2(pTiR10); G, NTS2(pTiC58, pAWlon); H, NTS2(pTiR10, pAWlon). Labels to the left indicate strains inoculated on the left side of each leaf while labels to the right indicate strains inoculated on the right side of each leaf.

(23K): Fig. 8. The attenuated virulence of the lon mutant NTS2 is not due to its slow growth. (a) Growth of Lon+ A. tumefaciens strains NTL4 and A136 and the lon mutant NTS2 in ABM minimal medium. The optical density at 600 nm of the cultures was determined at the indicated times. Data are representative of one of three experiments. , NTL4; , A136; , lon mutant NTS2. (b) Virulence assays on K. diagremontiana leaves. Wounds in leaves were inoculated with strains: A, NTL4; B, NTS2; C, A136; D, NTS2(pTiR10); E, A136(pTiR10); F, NTS2(pTiR10, pAWlon); G, NTS2(pTiR10, pAWlonK364Q); H, NTS2(pTiR10, pAWlonS680A). All strains were inoculated at a population size corresponding to an OD600 of 1·0.

We considered the possibility that the decrease in virulence of NTS2 is due to its slow growth. Our isolate of A. tumefaciens A136, a Rif^R Nal^R derivative of strain C58 (Watson et al., 1975), grows slower than its parent strain NTL4 and, with a generation time of 3·5 h in minimal medium, grows at a rate similar to that of NTS2 (Fig. 8a). The presence of pTiR10 had no effect on their respective growth rates (data not shown). While NTS2 harbouring pTiR10 was highly attenuated, A136 harbouring the same Ti plasmid was strongly tumorigenic (Fig. 8b). Given its strong conservation across phylogenetic lines, it is not surprising that Lon protease from A. tumefaciens is closely related to that of E. coli and in particular to that of S. meliloti, a very close relative in the family Rhizobiaceae. It also is not surprising that Lon_At requires both the Walker box motif and the active site serine; both domains are required for the activity of the E. coli enzyme. Consistent with these genetic, phylogenetic and enzymic relationships, the organization of the entire lon locus is conserved; the region flanking lon encodes clpX and hupB (Fig. 1) on the circular chromosome of A. tumefaciens C58 as well as on the chromosomes of many prokaryotes (Chin et al., 1988; Roberston et al., 2000; Stewart et al., 1997; Summers et al., 2000). However, despite these similarities, Lon from A. tumefaciens targets at least a subset of proteins quite different from those turned over by this protease in E. coli.

The influence of Lon on cell shape provides an example of such target diversity. Lon mutants of A. tumefaciens and E. coli both exhibit altered cell morphologies. However, while lon mutants of E. coli produce long filaments, cells of the A. tumefaciens lon mutant exhibit swellings and branching morphologies (Figs 3 and 5). Clearly, at the cellular level Lon is important to the regulation of cell division in both E. coli and A. tumefaciens. However, the protease contributes to this process by recognizing quite different targets in these two bacteria. In E. coli, lon mutants fail to degrade SulA, a cell division inhibitor, resulting in filamentation (Mizusawa & Gottesman, 1983; Schoemaker et al., 1984). On the other hand, among the α-Proteobacteria, in the most well-studied relative of A. tumefaciens, C. crescentus, Lon contributes to the regulation of cell division by controlling the intracellular levels of CcrM, a DNA methylase. The levels and thereby the activity of this enzyme signal the initiation of DNA replication (Wright et al., 1996). CcrM is a target of Lon and the unregulated accumulation of this essential methylase in lon mutants results in abnormal timing of initiation of DNA replication (Wright et al., 1996). The defect in the timing of initiation most probably results in the altered cellular morphology through a pathway that couples chromosome replication to cell division. While A. tumefaciens lacks SulA, it contains an orthologue of the C. crescentus ccrM gene (Kahng & Shapiro, 2001). This gene also is essential in A. tumefaciens and overexpression of ccrM results in altered cellular morphologies very similar to that of the lon mutant (Kahng & Shapiro, 2001). Moreover, overexpression of CcrM in S. meliloti results in cells with morphologies indistinguishable from that of the Agrobacterium lon mutant (Wright et al., 1995). We hypothesize that in A. tumefaciens CcrM is degraded by Lon and that aberrant accumulation of CcrM in the lon mutant causes inappropriate DNA methylation which interferes with the timing of DNA replication and subsequent cell division. CcrM, but not SulA, is present and strongly conserved in all of the α-Proteobacteria for which sequence data are available (Wright et al., 1995). We conclude that while Lon plays a central role in controlling cell division in both groups of prokaryotes, the targets of this protease, and therefore the mechanism of regulation, differs between the α- and γ-Proteobacteria. Moreover, while the role of Lon in modulating cell morphology is most pronounced during the SOS response within the γProteobacteria, the protease is more completely integrated into normal cell division processes in the α-subgroup.

Lon_At complements E. coli lon mutants for filamentation and regulation of expression of cps, the operon encoding production of colonic acid capsule (Fig. 5). Thus, while the protease controls cell division in the two groups of bacteria by targeting different regulatory elements, Lon from Agrobacterium retains its ability to recognize SulA and also RcsA, the positive activator of the cps regulon from E. coli. Lon also is required for proper regulation of extracellular polysaccharide synthesis in the close relative, S. meliloti (Summers et al., 2000). However, Lon apparently does not control extracellular polysaccharide in A. tumefaciens; as judged by Calcofluor binding, the level of extracellular polysaccharide production by the lon mutant is not detectably altered from that of its wild-type parent.

On initial isolation, the lon mutants formed two colony morphotypes, one slow-growing and unstable, the other growing somewhat faster but stable (Fig. 2 and data not shown). Given the instability of the NTS1-type and the stability of the NTS2-type colonies, we consider it likely that the stable morphotype contains a second, spontaneous mutation that suppresses some strongly deleterious effect of the lon mutation on growth. The nature of this suppressor mutation is not known, but it is unlikely to contribute to the other lon phenotypes exhibited by NTS2. All such phenotypes are complementable by wild-type lon expressed from its own promoter from a low-copy-number vector.

Two lines of evidence indicate that Lon is a member of the heat-shock regulon in A. tumefaciens. First, the transcription of lon_At was transiently stimulated by elevated temperature (Fig. 6b). Second, the increase in mRNA level was accompanied by a small but reproducibly detectable increase in the intracellular levels of the Lon protein (Fig. 6c and d). Analysis of the nucleotide sequence upstream of lon_At revealed an element similar to the consensus σ³² heat-shock promoter of the γ-proteobacterium, E. coli (Fig. 9a). Significantly, a σ³² homologue, RpoH, plays a major and global role in regulating the heat-shock response in A. tumefaciens (Mantis & Winans, 1992; Segal & Ron, 1995; Rosen et al., 2002; Nakahigashi et al., 1999). No sequences upstream of the 35 region of the lon promoter were found conserved in the corresponding upstream region of other heat-shock promoters in Agrobacterium or E. coli. However, a 16 bp inverted repeat (IR) is located between the putative 10 and 35 regions, and a second 14 bp IR is located downstream of the putative transcriptional start site (Fig. 9b). While these IR elements are not conserved in other α-Proteobacteria, they may play a role in the transcriptional or posttranscriptional regulation of lon in A. tumefaciens. Expression of lon is also induced by elevated temperature in B. abortus (Roberston et al., 2000) and S. meliloti (Mitsui et al., 2004), suggesting that, as in the γProteobacteria, inclusion of lon in the heat-shock regulon may be a common feature in the α-Proteobacteria. Interestingly, given that NTS2 does not display increased sensitivity to UV irradiation it is unlikely that Lon is part of the SOS response in A. tumefaciens.

(22K): Fig. 9. Promoter region of A. tumefaciens lon contains a σ32-like sigma factor recognition site and two inverted repeats. (a) Alignment ofpromoter sequences of lon in α-Proteobacteria. Bacteria are abbreviated as follows: Ba, B. abortus; Cc, C. crescentus; Sm, S. meliloti; At, A. tumefaciens. The identified and putative start sites of heat-inducible transcripts are underlined and boxed, respectively. Nxatg indicates the number of nucleotide bases between the transcriptional start site and the translation start codon of lon. Bases found in at least three of four promoters are indicated in upper case and bold type. Consensus promoter sequences deduced from upstream untranslated regions of lon in α-Proteobacteria and those of the RpoH-dependent promoters in A. tumefaciens (Nakahigashi et al., 1999) and E. coli σ32 (Yura et al., 1993) are shown for comparison. (b) A 16 bp inverted repeat (IR) located between the putative 10 and 35 regions and a second, 14 bp IR located downstream of the putative transcriptional start site are shown.

Lon in its wild-type form is required for full pathogenicity by A. tumefaciens (Fig. 7). Hence, ours joins a growing list of reports regarding the influence of this protease on the interaction of bacteria with their eukaryotic hosts. For example, among the α-Proteobacteria, Lon of B. abortus is required during the initial stages of infection in BALB/c mice (Roberston et al., 2000) while in S. meliloti Lon is required for formation of effective, nitrogen-fixing nodules in alfalfa (Summers et al., 2000). In the γ-Proteobacteria Lon is important for systemic infection of mice by Salmonella enterica serovar Typhimurium (Takaya et al., 2003). Pathogenesis of A. tumefaciens is a complex process involving signalling, gene regulation, assembly and function of a Type IV secretion system (T4SS), and physical interactions between the bacterium and its host plant. At present we do not know how the lon mutation affects these processes. Lon might target some protein involved in controlling expression of the vir regulon, the functions of which are involved in T-strand processing, transfer and integration into the plant genome. Alternatively, the mutation might interfere mechanistically with T-strand processing or with the assembly or function of the VirB T4SS. Interestingly, a mutant of strain C58 defective in the TAT transport system shows defects in cell morphology and virulence that phenocopy those of the lon mutant (Ding & Christie, 2003). It will be interesting to determine whether the loss of virulence associated with defects in Lon protease and the TAT secretory system reflect common or different targets. We thank Dr R. Martin Roop II, Department of Microbiology and Immunology, Louisiana State University Health Sciences Center for providing us with the polyclonal rabbit antiserum directed against B. abortus Lon, and Dr. Andrei Kuzminov, Department of Microbiology, University of Illinois at Urbana-Champaign, for kind assistance with the UV sensitivity assay. We also thank Dr Zhao-Qing Luo, Department of Biological Sciences, Purdue University, for the generous gift of plasmid pSR47s, and the other members of the laboratory for helpful discussions. This work was supported by grant No. R01 GM52465 from the NIH to S. K. F.