Mutational Analysis of TraM Correlates Oligomerization and DNA Binding with Autoregulation and Conjugative DNA Transfer*

F plasmid TraM, an autoregulatory homotetramer, is essential for F plasmid bacterial conjugative transfer, one of the major mechanisms for horizontal gene dissemination. TraM cooperatively binds to three sites (sbmA, -B, and -C) near the origin of transfer in the F plasmid. To examine whether or not tetramerization of TraM is required for autoregulation and F conjugation, we used a two-plasmid system to screen for autoregulation-defective traM mutants generated by random PCR mutagenesis. A total of 72 missense mutations in TraM affecting autoregulation were selected, all of which also resulted in a loss of TraM function during F conjugation. Mutational analysis of TraM defined three regions important for F conjugation, including residues 3–10 (region I), 31–53 (region II), and 80–121 (region III); in addition, residues 3–47 were also important for the immunoreactivity of TraM. Biochemical analysis of mutant proteins indicated that region I defined a DNA binding domain that was not involved in tetramerization, whereas regions II and III were important for both tetramerization and efficient DNA binding. Mutations in region III affected the cooperativity of binding of TraM to sbmA, -B, and -C. Our results suggest that tetramerization is important for specific DNA binding, which, in turn, is essential for traM autoregulation and F conjugation. These findings support the hypothesis that TraM functions as a “signaling” factor that triggers DNA transport during F conjugation.

Bacterial conjugation is a major mechanism for horizontal gene transfer and is defined as the unidirectional transfer of single-stranded DNA from a donor to a recipient cell. This process occurs in response to an uncharacterized mating signal generated by mating pair formation initiated by intimate cellto-cell contact (1,2). Conjugation requires an origin of transfer (oriT) in cis to the transferred DNA that contains a nic site as well as multiple binding sites for DNA-binding proteins that together form the relaxosome. Cleavage and re-ligation at the nic site by TraI, the relaxase, is in equilibrium until a stable mating pair has formed, which triggers DNA unwinding and transport between cells (2,3).
The F plasmid is the paradigm for a large group of conjugative plasmids in the IncF incompatibility complex that carries genes important for human and veterinary medicine, such as antibiotic resistance and toxin production (4). The genes responsible for F conjugation are located in the 33-kb F major transfer (tra) region with traM and traJ upstream of the major tra operon (traY to traI) (5). F plasmid TraM, which is essential for F conjugation, is not required for pilus synthesis, mating pair formation, or nicking at oriT in vivo (1, 6 -8). Consequently, TraM is thought to signal that donor and recipient cell contact has been made, thereby activating DNA transfer (1).
TraM forms tetramers in solution; the oligomerization equilibrium between tetramers and monomers can be described as two-state, suggesting that other oligomerization intermediates are not detectable (13). This is consistent with evidence that R1 TraM binds to DNA as a tetramer (14) but argues against a model in which TraM initially binds to DNA as a dimer (10). Deletion analysis has indicated that a TraM fragment of aa residues 2-55 forms dimers, whereas a TraM fragment consisting of the C-terminal 70 aa (58 -127) residues forms tetramers (13), suggesting that both regions are involved in TraM oligomerization.
The N-terminal region of TraM is important for specific DNA binding and is required for traM autoregulation and F conjugation (13,15); however, the role of the C-terminal region remains ambiguous. The importance of TraM tetramerization for autoregulation and F conjugation is unclear; similarly, the structural domains of TraM responsible for the specificity and cooperativity of DNA binding are unknown. A preliminary mutational analysis of TraM has shown that a missense mutation in either the C-or N-terminal regions can affect TraM in both autoregulation and F conjugation (16). In this work, 76 point mutations in traM were characterized for their effect on autoregulation and F conjugation; selected mutants were also characterized for their immunoreactivity, oligomerization, and DNA binding specificity. The results suggest that oligomerization is crucial for TraM function in DNA binding that, in turn, is essential for autoregulation and F conjugation.
DNA Manipulation, PCR, DNA Sequencing, and Sequence Analysis-DNA purification, manipulation, and PCR reactions followed standard procedures (17) or protocols from the manufacturers. Vent DNA polymerase (New England Biolabs) was used for all PCR reactions except for random PCR mutagenesis that used Taq DNA polymerase. Miniprep and Gel Extraction kits (Qiagen) were used for plasmid purification and extraction of DNA fragments from agarose gels, respectively. DNA sequencing was performed using the Amersham Biosciences DY-Enamic TM ET terminator cycle sequencing kit and an Applied Biosystems 373-S DNA Sequencer with XL upgrade. DNA and protein sequences were compiled and analyzed using Genetool® and Peptool® software.
Plasmids, Oligonucleotides, and Plasmid Construction-Oligonucleotides were synthesized on an Applied Biosystems 392 DNA synthesizer in the Department of Biological Sciences. Plasmids and oligonucleotides used in this work are listed in Table I. The 0.7-kb EcoRI-HindIII fragments of the PCR products generated by random PCR mutagenesis of traM (see below) were ligated to a 2.5-kb EcoRI-HindIII fragment of pRFM200, resulting in pRFM200 derivatives named after the corresponding traM mutations. The 0.7-kb EcoRI-BamHI fragments containing wild type or mutated traM from pRFM200 or its derivatives were cloned into the EcoRI-BamHI sites of pBluescript KSϩ, resulting in pJLM400 or pJLM400 derivatives, respectively. To reduce TraM expression from pJLM400 or its derivatives, glucose was added to the growth medium to a final concentration of 0.4% (w/v). A 0.7-kb BstBI-KpnI fragment from pRFM200, K31E, or I109T was used to replace the 0.5-kb BstBI-KpnI fragment in pLDLF007, resulting in pJLM102, pJLM103, or pJLM104, respectively.
Random PCR Mutagenesis of traM-Oligonucleotides JLU3 and JLU4 were used as primers for amplification of traM under error-prone PCR conditions (16). Reaction mixtures contained: 5 ng of pRFM200, 7 l of 10 mM dNTPs, 2 l of 100 mM MgSO 4 , 50 pmol of JLU3 and JLU4 each, 1 l of Taq DNA polymerase (5 units/l; Roche Diagnostics), 10 l of 10ϫ Taq incubation buffer, and double distilled H 2 O to take the total volume to 100 l. The PCR reactions were performed in the following steps: 1) 95°C for 30 s; 2) 55°C for 30 s; 3) 72°C for 1 min; and 4) repeat of steps 1, 2, and 3 for 40 cycles.
Selection of Autoregulation-defective traM Mutants-The detailed procedure for selecting autoregulation-defective traM mutants was de-scribed previously (16). Briefly, potential traM mutants cloned in pRFM200 derivatives were transformed into DH5␣ cells containing pACPM24fs::lacZ. The transformed cells were plated on solid LB medium containing X-gal, chloramphenicol, and ampicillin. Transformants were incubated for 24 h at 37°C; dark blue colonies were collected for further characterization. Plasmid DNA from each dark blue colony was extracted and sequenced using primers JLU3 and JLU4 to locate mutations in traM.
␤-Galactosidase Assays-A fresh, single colony was inoculated into LB broth containing appropriate antibiotics and grown at 37°C with shaking for 16 h. A 200-l sample was used for determining ␤-galactosidase activity as described by Miller (21) and reported as Miller units. The values were calculated using the equation: 1000 (A 420 /tvA 600 ), where t ϭ time of reaction (minutes) and v ϭ volume of culture added (ml).
SDS-PAGE and Immunoblot Analysis-0.1 A 600 (except where specified) of exponentially growing cells or a specified amount of pure protein were separated by a 15% SDS-polyacrylamide gel with a 7% stacking gel. TraM was assayed by Coomassie Blue staining (17) or by immunoblot as described by Penfold et al. (11). Immunoblot analysis was performed using rabbit anti-TraM antiserum (9) at a 1:10,000 dilution and donkey anti-rabbit secondary antibodies linked to horseradish peroxidase (Amersham Biosciences) at a 1:10,000 dilution.
Donor Ability Assays-E. coli XK1200 and ED24 were used as donor and recipient strains, respectively. The mating experiments were performed as previously described (22). Donor ability was calculated as the number of transconjugants divided by the number of donors.
Blue Native PAGE-Native gel samples were prepared as previously described (16). A 6 to 15% gradient polyacrylamide gel (pH 7.0) with a 4% stacking gel (pH 7.0) was used for electrophoresis following the procedures described by Schagger and von Jagow (23) and Schagger et al. (24). Proteins separated by blue native gels were transferred to polyvinylidene difluoride membranes for immunoblot analysis with anti-TraM antiserum.
Overexpression and Purification of TraM and Its Mutants-BL21-DE3 cells containing pRFM200 or its derivatives were grown in 250 ml of LB broth containing ampicillin at 37°C with vigorous shaking. After 3 h, isopropylthio-␤-D-galactoside was added to the culture to a final concentration of 1 mM, and the culture was grown for another 2 h before harvesting. Approximately 150 A 600 of cells were used for purification of each protein. The cell pellet was suspended in 6 ml of B-Per® bacterial protein extraction reagent (Pierce) with one tablet of Complete, Miniprotease inhibitor mixture (Roche Applied Science); the soluble fraction of the cells was extracted according to the manufacturer's instructions. All the following steps were performed at 4°C or on ice. Ammonium sulfate (720 mg) was dissolved in the extracted soluble fraction. After centrifugation at 27,000 ϫ g for 10 min, the supernatant was transferred into a new centrifuge tube, in which 480 mg of ammonium sulfate was dissolved. After centrifugation at 27,000 ϫ g for 10 min, the supernatant was carefully aspirated and discarded. The precipitate was dissolved in 2 ml of malonic acid (50 mM, pH 5.5), and the solution was pJLM102 derivative with a traM mutation, I109T (this work) pJLM400 pBluescript KSϩ with traM expressed from the lac promoter (this work) pLDLF007 pT7-4 with a DraI-DraI fragment containing F plasmid P traM and traM (9) pOX38-Km The F transfer region and RepFIA fused to a kanamycin resistance gene (35) a pOX38-MK3 pOX38 with traM disrupted by a kanamycin resistance gene (11) pRF911 pBR322-derived pBend2 with sbmA (10) pRF940 pBR322-derived pBend2 with sbmA, -B, and -C (10) pRFM200 pT7-5 with traM and the traJ promoter (16) pRFM200-Mdel pRFM200 derivative with a deletion containing most of traM (16) pT7-4 and pT7-5 Medium-copy cloning vectors (36) JLU3 5Ј-CTATAGGGAGACCGGAATTCG-3Ј; including the EcoRI site (underlined) in pT7-5 JLU4 5Ј-CGATAAGCTTGGGCTGCAGG-3Ј, including the HindIII site (underlined) in pT7-5 JLU80 5Ј-TAGGCGTATCACGAGGCCC-3Ј, 5Ј beginning at nucleotide 4328 in pBR322 a JLU81 5Ј-GGTGCCTGACTGCGTTAGC-3Ј, 5Ј complementing nucleotide 64 in pBR322 a a The nucleotide sequences of the F plasmid and pBR322 are available under GenBank TM accession numbers NC_002483 and J01749, respectively. centrifuged at 27,000 ϫ g for 10 min. The supernatant was brought to 2.5 ml with malonic acid (50 mM, pH 5.5) and was desalted on a PD10 column (Amersham Biosciences). After passing through a 0.45-M Mil-lex® syringe-driven filter (Millipore), the desalted protein extract was loaded onto a cation-exchange column (Mono S HR 5/5, Amersham Biosciences) using an Amersham Biosciences fast protein liquid chromatograph, model LCC-500. The column was eluted with malonic acid (50 mM, pH 5.5) and a 0 to 1 M NaCl gradient. Because TraM has very low UV absorbance, eluted fractions were examined by 15% SDS-polyacrylamide gels with Coomassie Blue staining, and protein peaks were further confirmed by immunoblotting with anti-TraM antiserum. Wild type TraM and the mutants V4A, F35S, A37V, Q53L, and F120L were eluted at 0.5-0.7 M NaCl. Mutants N5D, N10D, R48C, and S79* were eluted at 0.3-0.5 M NaCl. I109T, F120S, F121V, and F121S were eluted at 0.45-0.65 M NaCl. The pooled Mono S fractions (3-5 ml) of TraM or its mutant proteins were loaded onto a size exclusion column (Hiload® 16/60 Superdex 75 prep grade, Amersham Biosciences). The column was eluted with SEC buffer (50 mM sodium phosphate, 150 mM NaCl, pH 7.2), and the eluate was collected in 2-ml fractions. Fractions from B9 (void volume) to E12 (one column volume) were examined. The major peak fractions of each protein were concentrated and desalted, and the buffer was exchanged for Tris-HCl (50 mM, pH 7.6) using an Amicon® ultracentrifuge filter (Millipore) to a final volume of 50 l. Protein concentration was determined using BCA protein assays (Pierce) following the manufacturer's instructions.
Analytical Size Exclusion Chromatography-Purified TraM or its mutants (5 g) were brought to 1 ml with SEC buffer and loaded onto a Hiload® 16/60 Superdex 75 prep grade column at 4°C in fast protein liquid chromatography. The column was eluted at 0.5 ml/min with 120 ml (one column volume) of SEC buffer, and the eluate was collected in 2-ml fractions. Samples (10 or 30 l) from different fractions were separated on a 15% SDS-polyacrylamide gel and analyzed by immunoblot with anti-TraM antiserum. The column was calibrated with molecular weight markers (Sigma) under the same chromatographic conditions.
Electrophoretic Mobility Shift Assays-DNA fragments containing sbmA and sbmABC were amplified by PCR from pRF911 and pRF940, respectively, using primers JLU80 and JLU81. The resulting mixtures were concentrated in a Savant SpeedVac and separated by a 2% agarose gel. The sbmA and sbmABC fragments were isolated from the agarose gel and were quantified using an Ultrospec 3000 (Amersham Biosciences). Each binding reaction contained 40 nM of sbmA or sbmABC, 50 mM Tris-HCl (pH 7.6), 10% glycerol, 1 mM dithiothreitol, 30 g/ml bovine serum albumin, and 1.5 g poly(dI:dC) with a final volume of 15 l. After addition of a specified amount of purified TraM or its mutant proteins, each reaction mixture was incubated at 30°C for 20 min. The resulting mixture was added with 3 l of 6ϫ load dye (0.25% bromphenol blue, 30% glycerol) and loaded onto a 2% agarose gel that had been pre-run at 4°C and 30 mA in TBE (90 mM Tris borate, 1 mM EDTA) for 30 min. The gel was run at 4°C and 30 mA until the bromphenol blue dye reached the bottom of the gel. DNA was visualized by ethidium bromide staining.

RESULTS
Isolating Autoregulation-defective traM Mutants-A twoplasmid system was developed to select autoregulation-defective traM mutants generated by random PCR mutagenesis (16). In this system, pRFM200 expresses traM at a level comparable to that of the F plasmid. pACPM24fs::lacZ is a transcriptional fusion vector that carries the P traM promoter fused to lacZ with a Ϫ1 frameshift mutation at the beginning of lacZ to drastically reduce (ϳ500-fold) its activity and allow differentiation of regulated and unregulated lacZ expression. This results in a simple detection method for autoregulation-defective TraM mutants that give dark blue colonies rather than the pale blue colonies characteristic of wild-type TraM (see "Experimental Procedures").
A total of 234 mutant colonies were selected from ϳ25,000 transformants. DNA sequencing analysis revealed 135 pRFM200 derivatives with point mutations in traM, whereas the remaining 99 colonies contained frameshift mutations, multiple point mutations, or mutations within the ribosome binding site or start codon of traM, which were discarded. The 135 point mutants included 72 missense, 4 nonsense, and 59 redundant mutations that were located in 56 different amino acid residues of TraM (Fig. 1).
To assay the effect of the mutations on autoregulation, ␤-galactosidase (LacZ) activity assays were performed on cells containing pACPM24fs::lacZ and the mutant derivatives of pRFM200. Cells containing pRFM200 derivatives had LacZ activities ranging from 18 to 45 Miller units, whereas cells containing pRFM200 (wild type traM) or pRFM200-Mdel (a traM-deleted pRFM200 derivative) had activities of 8 Miller units  Some N-terminal Mutations Reduced TraM Immunoreactivity-As previously shown (16), pRFM200 and its mutant derivatives expressed similar levels of TraM as determined by immunoblot analysis (Table II). Thus, loss of autoregulation of P traM was not due to decreased levels of TraM. Some mutants with missense mutations within the first 47 codons of traM were difficult to detect by immunoblot using polyclonal anti-TraM antiserum, whereas S79*, one of the four nonsense mutants, was fully detectable (Table II ( 16)). To distinguish be- Ordinal numbers are shown below the corresponding residues in the TraM amino acid sequence. The residue substitutions resulted from different point mutations are indicated above the corresponding residues. An asterisk represents a stop codon generated by a nonsense mutation at the corresponding residue. If a TraM mutant complemented pOX38-MK3 transfer to a frequency at or higher than 10 Ϫ5 transconjugants per donor (Table II) (16), the corresponding residue substitution is shown in a smaller font size. Mutants with a nonsense mutation did not complement pOX38-MK3 transfer to detectable levels. Black boxes above the TraM sequence indicate the three regions (I, II, and III), which contain most of the missense mutations that decreased the ability of TraM to complement pOX38-MK3 transfer to a frequency lower than 10 Ϫ5 transconjugants per donor. tween protein instability and reduced immunoreactivity with TraM antiserum, wild-type traM and selected mutants were cloned into pBluescript KSϩ to give pJLM400 and its mutant derivatives, respectively, which expressed TraM at levels sufficient for detection by Coomassie Blue staining of SDS-polyacrylamide gels ( Fig. 2A). In the absence of glucose, the pJLM400 mutant derivatives expressed TraM at levels equivalent to pJLM400 (wild type TraM); however, immunoblot analysis of an equivalent gel revealed differing levels of immunoreactivity to the TraM antiserum (Fig. 2B). Thus, mutations in the first 47 aa of TraM affected the immunoreactivity of TraM rather than reducing protein stability.
Three Regions of TraM Are Important for F Conjugation-To assess the effects of different mutations on the function of TraM in F conjugation, pRFM200 or its mutant derivatives was transformed into E. coli XK1200 containing the traM-deficient F plasmid pOX38-MK3. Mating efficiency assays showed that pRFM200 (traM) restored pOX38-MK3 transfer to a frequency of 4 ϫ 10 Ϫ1 transconjugants per donor (100%), whereas none of the pRFM200 derivatives complemented pOX38-MK3 transfer completely (Table II). The most severe mutations were concentrated in three regions of TraM, I, II, and III, which correspond to residues 3-10, 31-53, and 80 -121, respectively (Fig. 1).
Two mutants were selected to determine the relationship between TraM levels, P traM autoregulation and function during conjugation. A region II mutant, K31E, and a region III mutant, I109T, which were highly defective for F conjugation ( Fig.  1 and Table II), were cloned downstream of P traM in a mediumcopy vector, pT7-4 (Table III). Both pJLM103 (P traM K31E) and pJLM104 (P traM I109T) overexpressed TraM compared with wild-type pJLM102 without affecting the growth rate of the cells (data not shown). In cells containing pACPM24fs::lacZ, pJLM104 (I109T) repressed P traM almost as well as wild type TraM (pJLM102), whereas pJLM103 (K31E) did not. In contrast, overexpression of either I109T or K31E resulted in low levels of complementation of pOX38-MK3 for transfer, suggesting that autoregulation and F conjugation are genetically distinct functions of TraM. When co-resident with pOX38-K m , a wild-type F plasmid derivative, both K31E and I109T decreased the transfer frequency of pOX38-K m by more than 400-fold (Table III), suggesting negative dominance of the mutants over wild-type TraM.
Analysis of TraM Mutants Using Blue Native PAGE-Autoregulation could be affected by mutations that alter either DNA binding affinity or the oligomerization ability of TraM. The ability of TraM and its mutants to form tetramers was inves- a Mutants are named after their codon changes. b Determined by assaying the ␤-galactosidase activity of cells containing pACPM24fs::lacZ and pRFM200 or one of its derivatives. c Determined by immunoblot analysis of 0.1 A 600 cells containing pRFM200 or one of its mutant derivatives using anti-TraM antiserum. Immunoblot bands comparable in intensity to that of 0.1 A 600 cells containing pRFM200 are considered "ϩ". Those mutants that had no or barely detectable bands and later were determined to have lower immunoreactivity than TraM as shown in Fig. 2 are referred as "Ϫ" (16). "ND", not determined. d Determined by assaying donor ability of cells containing pOX38-MK3 (TraM Ϫ ) and pRFM200 or one of its mutant derivatives. "Ͻ1 ϫ 10 Ϫ7 " refers to no detectable donor ability. e ND, not determined. f A stop codon. tigated using blue native gel electrophoresis (Fig. 3). During blue native gel electrophoresis, the proteins are bound with negatively charged Coomassie Blue G-250. This masks the effect of mutations on the intrinsic charge of a protein that would alter its electrophoretic mobility during conventional native gel electrophoresis (23). As expected, wild-type TraM migrated predominantly as a single band that was similar in size to BSA monomers (66 kDa), indicating that TraM formed tetramers (ϳ58 kDa; Fig. 3A). Higher order species of TraM were also detected that are characteristic of the propensity of TraM to aggregate. Most of the mutants tested, especially C-terminal mutants, also exhibited species smaller than tetramers. These smaller species were not degradation products, because SDS-PAGE of equivalent samples revealed no degradation of TraM or its mutants (Fig. 3B). Due to the resolution limit of blue native gel electrophoresis (23, 24), these smaller species could be either dimers (ϳ29 kDa) or monomers (ϳ14.5 kDa).

FIG. 2.
Overexpression of wild type and mutant TraM. Cells containing pJLM400 (traM) or its derivatives (traM mutants) were grown in LB broth for 3 h before harvest. All mutants are named after their mutations (Table II)  a Determined by immunoblot analysis of 0.1 A 600 of cells containing the corresponding construct using anti-TraM antiserum. b Determined by assaying the ␤-galactosidase activity of cells containing pACPM24fs::lacZ and pJLM102 or one of its derivatives. Wild-type repression is represented by pJLM102. Complete loss of repression is represented by vector control, pT7-4.
c Determined by assaying the donor ability of cells containing pOX38-MK3 and pJLM102 or one of its derivatives. d Determined by assaying the donor ability of cells containing pOX38-Km (expressing wild-type TraM) and pJLM102 or one of its derivatives.

Oligomerization of Purified TraM and Its
Mutants-Twentyone TraM mutants with point mutations in regions I, II, or III were selected for purification and further characterization. Wild type TraM and 13 mutants were successfully purified (see "Experimental Procedures"), whereas R24Q, K31E, L42P, L47P, S89F, S95P, F100S, and S114P were deemed to have significant changes in charge or conformation because of their inability to bind to the cation exchange column. During preparative size exclusion chromatography (SEC), TraM and the 13 mutants were present as a single major peak with one or two minor peaks. Two characteristic patterns for wild-type and mutant TraM, with F121S serving as an example, are shown in Fig. 4A. Small amounts of TraM and the mutant proteins were consistently found in the void volume fractions (B11), suggesting that TraM either aggregated or bound nonspecifically to protein contaminants that could be detected by Coomassie Blue staining (Fig. 4A).
The oligomerization status of TraM was further examined by analytical size exclusion chromatography using protein in the major peak from preparative SEC (see "Experimental Procedures"). TraM was detected by immunoblot, because the molar extinction coefficient of TraM, which lacks tryptophan, is very low. TraM was resolved as a single peak at fraction C4 (Fig.  4B), which corresponded in size to a tetramer (ϳ58 kDa). Almost no material corresponding to the peak containing higher aggregates (B11) during preparative SEC of either TraM or F121S was detectable (Fig. 4B).
The elution profiles of the TraM mutants during analytical SEC could be divided into four groups. F121S, in the first group, had a major peak at C6 corresponding to 45 kDa and a minor one corresponding in size to a dimer (29 kDa; Fig. 4B). Because the shape of a protein can affect its mobility during SEC (29), the major peak for F121S was thought to contain tetramers with altered conformation. This first group also contained I109T, F120S, and F121V that shared identical elution profiles with F121S (Table IV). In the second group, N5D (Fig.  4C) as well as V4A, N10D, F35S, R48C, and F120L (Table IV) had elution profiles identical to wild-type TraM, indicating the presence of tetramers. In the third group, A37V (Fig. 4C) and Q53L (Table IV) had a major peak identical to that of TraM tetramers and a minor peak at C7, which was thought to be dimers that differed from those of F121S in conformation. In the fourth group, S79*, a truncated TraM fragment of ϳ9 kDa, had a major peak at D2 corresponding to 18-kDa dimers and a minor peak at C6 that was close to the expected position of 36-kDa tetramers (Fig. 4D). These results indicated that the smaller-than-tetramer species observed during blue native gel electrophoresis were probably dimers. Thus mutations that affected tetramerization correlated with their defects in autoregulation and F conjugation, suggesting that tetramerization is important for the TraM function.
Specific DNA Binding of TraM and Its Mutants-Electrophoretic mobility shift assays were performed to determine the ability of TraM and its mutants to bind DNA. They could be categorized into four groups according to their patterns of binding to DNA fragments containing sbmABC, the three TraM binding sites. In group A (Fig. 5A), wild-type TraM bound to sbmABC with high affinity. At 600 nM TraM (equal to 150 nM TraM tetramers), 40 nM sbmABC was almost completely shifted to a position corresponding to binding to sbmA (10). A 3-fold increase of protein concentration resulted in a second shift to a position corresponding to binding to all three sites in sbmABC (10). F120L bound to sbmABC with an affinity slightly lower than wild-type but had the two shifted bands characteristic of the cooperative binding of TraM (Fig. 5A).
Group B proteins (N5D and S79*) did not shift sbmABC significantly at the highest concentration of protein tested (6000 nM), indicating a loss of DNA binding ability for these two mutants (Fig. 5B). In group C, V4A, N10D, F35S, A37V, R48C, and Q53L required 3-to 10-fold more protein than wild-type TraM to completely shift sbmABC, indicating decreased DNA binding affinity for these mutants (Fig. 5C and Table IV). In group D (Fig. 5D), I109T, F120S, F121V, and F121S, had lowered binding affinities for sbmABC similar to group C mutants. They shifted sbmABC into smeared bands, which could represent a mixture of binding complexes of different electrophoretic mobilities suggesting a loss of cooperativity during binding (Fig. 5D). The cooperativity of binding to DNA by the mutants I109T and F121V was further investigated by electrophoretic mobility shift assay using a DNA fragment containing only sbmA, the site with the highest affinity for TraM. Both mutants shifted sbmA to a single band rather than a smear, similar to wild-type TraM (Fig. 5E), although the affinity for the DNA was reduced in both cases. Thus, the smeared shifted bands, characteristic of group D mutants, appeared to be the result of a loss of cooperativity during DNA binding, with the mutant proteins binding all three sites in no particular order (10,25). All three regions (I, II, and III; Fig. 1) appeared to be important for TraM binding, with region III defining the ability of TraM to bind to sbmA, -B, and -C in order, which is a feature of cooperativity.

DISCUSSION
The analysis of point mutations in TraM revealed that distinct regions of the protein were involved in defining reactivity with anti-TraM antiserum as well as the properties of autoregulation and competence for DNA transfer. The first 47 aa of TraM appeared to be important for the immunoreactivity of TraM with polyclonal anti-TraM antiserum and could contain the principle antigenic determinants in this protein. 72 distinct missense mutations (including a 78% redundancy), covering 44% of TraM residues, clustered into three regions (I, II, and  Fig. 1. b Oligomerization status of TraM and its mutant proteins according to their native sizes is determined by analytical SEC using Hiload® 16/60 Superdex® 75 prep grade as shown in Fig. 4B, C, and D. c Determined by EMSA as shown in Fig. 5A, B, C, and D. Group A, "wild-type binding"; Group B, "No detectable binding"; Group C, "lowered affinity"; Group D, "lowered and non-differentiated affinity to sbmA, B, and C." d Determined by assaying donor ability of cells containing pOX38-MK3 and pRFM200 or one of its mutant derivatives as shown in Table II. e NA, not applicable. f A stop codon. III; Fig. 1), which defined four different phenotypes for oligomerization, DNA binding, cooperativity of DNA binding, and transfer. Mutations in all three regions were defective in binding to DNA fragments containing sbmA, -B, and -C, suggesting that binding to these sites is required for TraM to function in autoregulation and F conjugation. However, the three regions of TraM contributed to DNA binding via different mechanisms. Mutations within region I did not affect TraM tetramerization but decreased the binding affinity of TraM for sbmABC, suggesting that this region participates in direct contact with the DNA (Table IV). Asparagine is (Asn) the third most common residue to form hydrogen bonds with bases and DNA backbones in protein-DNA complexes (26). The Asn residue at position 5 might interact with the negatively charged DNA; thus, replacement of this residue with an aspartic acid (Asp) could abolish TraM binding to sbmABC (Fig. 5B). When Asn-10 of TraM was replaced by Asp, the amino acid at the equivalent position in R1 TraM (27), the resulting mutant protein bound to sbmABC less efficiently (Fig. 5C), suggesting that Asn-10 contributes to the allelic specificity of the F-like TraMs for their cognate DNA (28). This agreed with the previous finding that the first 24 aa residues of R1 TraM define its DNA binding specificity (15).
Mutations F35S and R48C in region II did not affect TraM tetramerization but decreased the binding affinity of TraM for sbmABC, suggesting that these residues also directly contribute to TraM binding to DNA (Table IV). Some other mutations within region II affected TraM tetramerization with A37V and Q53L forming small amounts of dimers as well as tetramers (Table IV). Thus region II appears to define a dimerization domain near the N terminus. Mutations in this region would give aberrant dimers resulting from interactions involving the second dimerization domain nearer the C terminus (Fig. 4C) (29). The presence of a dimerization domain near the N terminus is supported by the ability of a TraM fragment of aa 2-55 to form dimers (13). Thus the region II mutations probably affected protein-protein interactions during tetramerization. The decreased affinity of A37V or Q53L for sbmABC suggests that tetramerization is also important for binding to DNA.
Missense mutations within region III appeared to dramatically affect the quaternary structure of TraM. S89F, S95P, F100S, and S114P did not bind efficiently to the Mono S column under the conditions used in this work, suggesting that major structural perturbations occurred in these mutant proteins. I109T, F120S, F121V, and F121S appeared to form conformationally altered tetramers as well as smaller amount of dimers as determined by analytical SEC (Fig. 4B and Table IV). This is consistent with the finding that a C-terminal fragment of F TraM formed tetramers (13). Because tetramers require at least two distinct dimerization domains in each monomer, the loss of the ability to form tetramers suggests that these mutations affected one of these domains. Defects in tetramerization or conformational changes in the tetramers appear to affect DNA binding by TraM as well as transfer ( Fig. 5D and Table IV).
It could be argued that the major peak for F121S (or similar mutants) during analytical SEC contained trimers or aberrantly large dimers; however, other evidence suggests that this peak contained conformationally altered tetramers. First, because the N-terminal region of TraM exists as dimers rather than monomers (13), oligomerization of the TraM mutant proteins containing an intact N-terminal region should result in dimers and tetramers not trimers. Second, because S79*, which lacks the C-terminal 49 residues of TraM, formed a minor peak corresponding to tetramers (Fig. 4D), F121S, with only one amino acid substitution, would be expected to also form tetramers. Third, I109T, F120S, F121V, and F121S were eluted from the cation-exchange column at slightly lower salt concentrations than TraM (see "Experimental Procedures"), suggesting these mutations resulted in conformational changes.
Blue native polyacrylamide gel electrophoresis suggested that I109T and F121S formed more dimers than tetramers (Fig. 3), which was opposite to the results obtained by analytical SEC (Fig. 4B). Native gel samples were prepared from cells expressing levels of wild type or mutant TraM comparable to the physiological levels of TraM expressed by the F plasmid (16). The concentration of TraM on the gel was comparatively low, and TraM could only be detected by immunoblot. Although analytical SEC is usually accurate in estimating sizes of native protein complexes, because the potential effects of differences in charge are negligible, it could be affected by the high concentrations of protein in the sample applied to the column. Because TraM has a tendency to aggregate, this might favor the formation of tetramers via mass action. Thus, I109T and F121S probably form more dimers than tetramers with the equilibrium shifted toward tetramers if aggregation is a factor.
Mutations that decreased TraM immunoreactivity appear to be located within the first 47 aa of TraM (Fig. 2). Because prolines disrupt higher order structure, mutations such as L42P and L47P could alter the structure of TraM during immunoblot analysis. Alternatively, mutations in the first 47 aa might alter the immunogenicity of TraM with a consequent reduction in immunoreactivity to specific preparations of antiserum. Certainly R100 TraM, which differs from F TraM within the first 40 aa, reacts poorly with anti-F TraM antiserum, which agrees with this observation (data not shown). The N-terminal region also defines the DNA binding domain of the different alleles of TraM expressed by the various F-like plasmids (F, R1, and P307 (14,15,30)), suggesting that these residues are exposed on the surface of the protein and form contacts with the DNA.
Nearly normal repression for P traM was observed when I109T was overexpressed from P traM , indicating that the increased dosage of I109T compensated for its inability to autoregulate the traM promoter. This agreed with previous observations that expressing mutants of an autoregulatory gene product from its native promoter compromises its characterization because of the compensatory effect of protein overexpression (16). This result also suggested that a mass action effect of I109T at P traM resulted in repression even though I109T was defective for tetramerization. However, overexpression of I109T from P traM could not complement pOX38-MK3 for conjugative DNA transfer, indicating that F conjugation requires the quaternary structural integrity of TraM.
This work has defined three domains of TraM involved in tetramerization or DNA binding, two properties important for TraM function in autoregulation and F conjugation. The Nterminal region of TraM (region I) is required for DNA binding and defines its allelic specificity, agreeing with previous results showing that the N terminus of TraM from F-like R1 is responsible for specific DNA binding (15,30). Tetramerization of F plasmid TraM increases the binding affinity of TraM for its cognate sites, consistent with the finding that R1 TraM binds to DNA as a tetramer (14). Involvement of extensive regions (II and III) in oligomerization presumably ensures efficient tetramerization of TraM, which could explain why monomers and tetramers but not dimers were detectable in the equilibrium studies of Miller and Schildbach (13). The C-terminal domain (region III) of TraM is not only important for tetramerization but also for the cooperativity of binding to DNA, because it appears to have a role in defining the order that TraM binds to the three different sites, sbmA, -B, and -C (10).
We propose that TraM forms a DNA binding domain at the N terminus (region I) through tetramerization at regions II and III. Through cooperative binding, TraM occupies all three sites at oriT during relaxosome formation. The F relaxosome normally resides at the cell center or quarter position during exponential growth (31) when there is no occasion for conjugative DNA synthesis (7). After donor and recipient cells contact each other, TraM might connect the relaxosome to the transferosome through interactions with the coupling protein, TraD (32). TraM-TraD interactions might further change the binding properties of TraM at oriT, causing localized denaturation between sbmABC and the nic site. This localized melting at oriT could provide a region of single-stranded DNA that is required for inducing DNA helicase activity of TraI to further unwind the DNA for conjugative transfer (33). In this manner, TraM might indeed act as a "signaling factor" to trigger conjugative DNA transfer as proposed by Willetts and Wilkins (1).