Eur. J. Biochem. 205, 375-381 (1992)

0 FEBS 1992

Purification and properties of the mercuric-ion-binding protein MerP Lena SAHLMAN and Bcngt-Harald JONSSON Department of Biochemistry, University of UmeB, Sweden (Received October 31/Dcccmber 18,1991) - EJB 91 1470

The gene merP, coding for a mercuric-ion-binding periplasmic protein (P protein), was cloned into the expression vector pCA. In an Escherichia coli strain bearing the resulting plasmid, the P protein constitutes about 20% of total soluble protein. P protein was purified using ammonium sulfate precipitation and two chromatography steps. Typical yields were 20 - 30 mg from 7.5 1 bacterial culture. The protein is a monomer with a molecular mass of 7500 Da. The periplasmic signal peptide was processed identically in both the recombinant and the wild-type proteins. CD spectra of both proteins were identical and indicated that the structure is highly ordered, containing approximately 80% a-helix. Purification in the presence of excess cysteine resulted in a form of the protein containing two reduced thiols, in agreement with the published sequence which has two cysteine residues. When cysteine was omitted from the purification buffers, no reduced thiol groups could be detected suggesting that the cysteine residues are oxidized. Both of these forms of the protein were found to bind approximately five Hg2 ions/protein molecule in an apparently non-specific manner. However, in the presence of external thiol compounds, the protein with reduced thiols bound only one Hg2+ ion/protein molecule with an apparent Kd of 3.7 1.3 pM. Under these conditions, the protein with oxidized thiols did not bind H g 2 + .The possible physiological role of this protein in Hg2+ detoxification is discussed. +

It has been known for some time that the bacterial enzyme mercuric reductase confers resistance to mercuric ions. The enzyme reduces Hg2+ to metallic mercury, which in itself is not toxic [l].Mercuric reductase is localized in the cytoplasm, and uses NADPH as reducing agent. Bacteria that are resistant to mercury also contain proteins apparently necessary for the transport of mercuric ions into the cytoplasm. These transport proteins seem to be required for full resistance. Cells that contain mercuric reductase but lack the transport proteins are only slightly more resistant to Hg2+than cells without the iner genes [2]. The existence of these other proteins was first discovered in bacteria that were hypersensitive to mercuric ions. These hypersensitive bacteria lack a functional reductase, but have an intact transport system, i.e. they can transport the toxic compound into the cytoplasm but lack the ability to detoxify it [3]. The genes encoding the proteins necessary for mercuric ion resistance are often located on transposons. The mer operon of the transposon Tn21 consists of five structural genes, Correspondence to L. Sahlman, Department of Biochemistry, University of Umci, S-90187 Umci, Swcden Ahhrevzutions. N bs2, 5,5'-dithiobis-(2-nitrobcnzoic acid); protein P and MerP, mercuric-ion-binding protcin. Enzjwzes. Mercuric reductase (EC 3.1 6.1 . l ) ; dcoxyribonucleasc I, DNase (EC 3.1.21.1); ribonuckdsc A, RNase (EC 3.1.27.5); carbonic anhydrdse (EC 4.2.1.I); trypsin (EC 3.4.21.4); T7 RNA polymerase (EC 2.7.7.7); restriction endonuclcases (EC 3.1.21.4).

merTPCAD [4-61. The merD-encoded protein has no known function, although a role in gene regulation has been suggested [7, 81. Transcription is regulated by the merR gene product, which acts as a repressor in the absence of mercuric ion, but as an inducer when Hg2+is present [9, lo]. Mercuric reductase, encoded by merA, is the only protein besides the merR polypeptide to have been studied at the protein level [Ill. The genes merT and merC code for two hydrophobic, presumably membrane-bound proteins involved in the transport of Hg2 across the inner membrane [ I l , 321. The gene merP, 273 bp, encodes a small polypeptide with a molecular mass of 7474 after processing of a periplasmic signal sequence 14, 131. It has been proposed that the merP-encoded P protein scavenges mercuric ions as soon as they enter the periplasmic space. The P-protein - mercuric-ion complex is postulated to interact with the two membrane-bound proteins, encoded by merT and merC, which then transfer the ions across the inner membrane [ll, 121. Interestingly, the mer operon from transposon Tn501 lacks the merC gene [14], while the rner operon from Thiobacillus lacks the merT gene 1151. Since both of these systems have functional Hg2+ transport, the exact polypeptide requirements for transport over the membrane is an enigma. The function proposed for the P protein requires it to bind mercuric ions and then to interact with the membrane protein(s). There have been no published reports of attempts to verify this theory using purified protein. This paper describes the cloning of the merP gene into an overexpressing +

316 system, purification of the corresponding protein, and the Hgz+ binding properties of the P protein in vitro. MATERIALS AND METHODS Materials Tryptone was obtained from Difco and yeast extract from Oxoid. Restriction enzymes were purchased from BoehringerMannheim. Phenyl-Sepharose, Sephadex G-50, DEAESepharose Fast Flow, SDS gel molecular mass standards, and isopropyl B-D-thiogalactoside were from Pharmacia-LKB. 203Hg(N03)was from New England Nuclear. Pyronin Y, 5 3 ’ dithiobis-(2-nitrobenzoic acid) (Nbs,), trypsin, ribonuclease (RNase) and deoxyribonuclease (DNase) were obtained from Sigma. Myoglobin and cytochrome c were from Serva Feinbiochemica. Dialysis tubing Spectro/Por 3 was from Spectrum Medical Industries, Inc. Microsep concentrators were purchased from Filtron Technology Corporation. Picofluor was obtained from Canberra Packard. Oligonucleotides were synthesized by Symbicon, Umei. Ultrafilter YM5 was obtained from Amicon. Human carbonic anhydrase I1 was a gift from Dr B.-H. Jonsson. Bacterial strains and plasmids The Escherichia coli strain bearing plasmid pDUl003 [5] was a gift from Dr Simon Silver (University of Illinois at Chicago). This plasmid includes the complete mer operon from the resistance plasmid R100, cloned into pBR322. Esclierichicl coli strain CJ236 was used in the mutagenesis. This strain lacks the dUTP ‘repair’ system, so it incorporates dUTP instead of dTTP into DNA. E. coli strain BL21/DE3 was used for the overproduction of protein. This strain contains the gene for T7 RNA polymerase, under the control of the kic promoter, The vector pCA [16] was used for overexpression. This plasmid/phage hybrid vector contains the T7-RNA polymerase promoter. The plasmid pNH6, which was used for the cloning, was kindly provided by Dr Nancy Hamlett (Harvey Mudd College, California). This plasmid contains the mer structural genes from transposon Tn21, cloned into pTZ19K. Genetic techniques The plasmid pNH6 was transformed into competent cells of E. coli CJ236. Mutagenesis was carried out using the method of Kunkel [17]. DNA sequencing was performed directly on the plasmid after strand separation, using the dideoxy-chain-terminating method of Sanger et al. [I 81. Growth of bacteria Cells were grown in Luria broth [tryptone (10 g/l), yeast extract (5 g/l), NaCl(10 g/l)] containing ampicillin (50 pg/inl). Culture turbidity was monitored at 660 nm, using a Hitachi 100-60 spectrophotometer. Protein electrophoresis Samples were run from the anode towards the cathode on non-denaturing gels using the following conditions : separating gel, 15% acrylamide, 0.385% bisacrylamide, 0.22 M Tricine/KOH pH 7.0; stacking gel, 3.4% acrylamide, 0.09% bisacrylamide, 43.5 mM Tris/28.5 mM Tricine pH 7.4;

anode buffer, 0.04 M Bistris, 0.025 mM Tricine, pH 7.5; cathode buffer, 0.0625 M Tricine, 0.05 M KOH, pH 8.9 [19]. Pyronin Y was used as tracking dye. At this pH value (7.5) the P protein is positively charged, and most other proteins are negatively charged: only positively charged proteins will move into the gel towards the cathode. Only one band with a low molecular mass can be detected on the gel, and this band constitutes the P protein. This method can thus be used to detect the presence of P protein in crude extracts or fractions from chromatography (Dr Nancy Hamlett, personal communication). SDS/PAGE was performed as described by Schagger and von Jagow [20], using 16% acrylamide, 0.5% bisacrylamide. Gels were scanned using a LKB 2202 Ultroscan laser densitometer. Purification of P protein Cells were grown in 7.5 1 Luria broth to an absorbance of 0.8 at 660 nm, whereupon isopropyl p-D-thiogalactoside was added to a final concentration of 0.2 mM. After 6 h of induction, the cells were harvested. Two different procedures were used to isolate the protein: in one all buffers contained 1 mM cysteine, and in the other no cysteine was added to the buffers. Otherwise the procedures were identical. The cells were suspended in 70 ml 10 mM potassium phosphate pH 7.3. DNase and RNase were added to a final concentration of 0.01 mg/ ml, respectively, and Mg2+was added to a final concentration of 10mM. The nuclease treatment is important, as intact nucleic acids do not separate from the protein in the following steps. The cells were broken by sonication and cell debris was removed by centrifugation at 15000 rpm (28000 x g) for 15 min. Ammonium sulfate was added in three steps. The crude extract was first brought to 20% saturation with ammonium sulfate and the precipitated material was removed by centrifugation and discarded. Ammonium sulfate was then added to the supernatant to bring it to 50% saturation. After centrifugation, the supernatant was set aside and the precipitated material was resuspended in 50 mM potassium phosphate pH 7.3. The second step was repeated, and the supernatant was pooled with that obtained in the previous step. This three-step procedure gave a better yield than a single step. The pooled supernatant fractions were diluted with 50 mM potassium phosphate pH 7.3 so that saturation with ammonium sulfate was reduced to 45%. This preparation was then loaded onto a Phenyl-Sepharose column (20 cm x 5 cm), equilibrated with 50 mM potassium phosphate pH 7.3, containing ammonium sulfate to 45% saturation. The column was washed with this buffer until all of the protein, as monitored by A l g O ,had eluted. The P protein does not bind to PhenylSepharose at this ammonium sulfate concentration in a largescale preparation. Fractions containing P protein, as determined by native gel electrophoresis, were pooled and concentrated by ultrafiltration using a YM5 filter. The concentrated protein was dialysed against a total of 4 1 50 mM potassium phosphate pH 7.3 and then applied to a DEAE-Sepharose column (15 cm x 2.5 cm), equilibrated in the same buffer. The P protein did not bind and came through the column in the wash, Fractions containing P protein were pooled and concentrated by ultrafiltration. The protein was stored at - 80‘ C. N-terminal sequencing N-terminal sequencing was performed using an Applied Biosystems model 477A sequencing system.

311

Thiol determinations Protein purified using cysteine-containing buffers was first passed over a Sephadex G-25 column to remove cysteine from the buffer. Thiol determinations were subsequently performed in 70 mM Tris/Cl pH 8.0 containing 3 mM EDTA, using protein that was either in the native state or denatured by 0.1% SDS. Nbsz was added to a concentration of 0.19 mM, which represents a %fold excess over the protein. The change in absorbance was monitored at 412 nm, using a blank without the protein as a reference. The molar concentration of thiols was then calculated using a molar absorption coefficient for thionitrobenzoate of 13600 M-’ cm-’ [21].

Native molecular mass estimations A Sephadex G-50 column (90 cm x 2.5 cm) equilibrated in 50 mM potassium phosphate pH 7.3 was used to estimate the molecular mass of the native protein. The column was calibrated with the following proteins: trypsin (24000 Da), myoglobin (17 800 Da), ribonuclease A (13700 Da), and cytochromc c ( 3 2 300 Da).

Spectroscopy CD spectra were recorded at 25°C using a Jasco 5720 spectropolarimeter. The spectrometer is equipped with software for the analysis of spectra with regard to their content of secondary structure. Ultraviolet/visible spectra were obtained with a Perkin-Elmer 320 spectrometer.

Binding of Hg2+ A solution of 1.5 in1 containing 20 pM protein in 50 mM potassium phosphate pH 7.3 was mixed with 1.5 ml buffer containing 5 -200 pM HgC12. In control experimcnts, buffer was substituted for the protein solution. This mixture contained ’03Hg2+ at a concentration of less than 1% of the total Hg2+ concentration. In some experiments the solution also contained a fourfold molar excess of cysteine over mercuric ions. The protein was mixed with the HgC12 solution and immediately placed in Microsep microconcentrator cells. These consist of an upper plastic container with a maximum volume of 4.5 ml. There is an ultrafilter at the bottom of the upper container: in this case, a filter with a molecular mass cutoff of 3000 Da was used. This filter-containing unit was placed into a reservoir for filtrate, and ultrafiltration was achieved by centrifugation. In these assays the Microsep devices wcre centrifuged at 4500 g for 3 min. After centrifugation approximately 100 pl, or 3% of the starting volume, had filtered through to the lower reservoir. Samples were removed from both the upper and lower reservoirs and added to 2.5 ml Picofluor, and then counted in a LKB-Wallac liquid scintillation counter 1214 Rackbeta. Duplicate samples were prepared for each Hgz+ concentration. The data was analyzed using the following equation, describing tight binding: J’ = c(&

$.

+ c) - b’(&

+ f + c)2- 4 x f x c ) / 2

where y is the concentration of bound Hg”, Kd the dissociation constant, t the total concentration of added Hg2+, and C the capacity for binding Hg2+ [22]. This equation is derived from the mass law, but uses total concentrations rather than free concentrations of protein and ligand. It was used because it is not possible to measure the amount of free Hg2+

in the presence of cysteine, since there is no reliable value for the binding constant for the cysteine-Hg” complex. In these experiments, the protein and the protein-mercury complex remained in the top reservoir, since the 7500-Da protein does not pass through the 3000-Da filter, while solution containing unbound Hg2+ filtered through. By measuring the mercuric ion concentrations in both the upper and lower reservoirs, the concentration of bound mercuric ions could be calculated by subtracting the concentration of non-bound mercuric ions. The background, that is the concentration of mercuric ions apparently bound in the samples containing only buffer, was subtracted from the binding data before the analysis was performed. For the samples with cysteine but no protein present, the concentration of mercury was only approximately 5% lower in the filtrate than in the starting solution. The corresponding value for samples without cysteine or protein was 20%.

RESULTS Cloning and overexpression of mcvP The gene merP starts at position 968 and ends at position 1240 of plasmid RlOO [13]. Two new sites for the restriction enzyme NcoI were introduced into pNH6, one at the position corresponding to 970 (without changing the ATG start codon) and the other at the position corresponding to 1280. A 31 I-bp NcoI fragment was then cleaved out of pNH6 and introduced into the plasmid/phage hybrid vector, pCA, behind the T7RNA polymerase promoter. The new plasmid was transformed into Escherichia coli strain BL21/DE3, which contains the gene for T7 RNA polymerase under the control of the lac promoter. Extracts of cells from 19 colonies were analyzed for the presence of P protein using native polyacrylamide electrophoresis. Five clones expressed the P protein, seven contained the Ncol fragment in the wrong orientation, and the remaining seven contained the religated pCA vector. Plasmid DNA was prepared from two of the five positive clones, and the DNA sequences were determined. There are two sequences published for the merP gene [4, 131. They are both identical except in the positions 1113 and 1114 (numbered according to Misra et al. [13]). Misra et al. have published the sequence G followed by C, while Barrineau et al. [4] have the reversed order, C and G. In this work half of the merP gene was sequenced using a primer for the sense strand, and the other half using a primer for the antisense strand. The primers allowed for overlap in the middle of the gene, including positions 1113-1114. At position 1113-1114 in the sequence of the sense strand, G and C ran in the same position on the sequencing gel. The sequence from the other primer distinguished the order as G followed by C for the sense strand. This agrees with the sequence published by Misra et al. [13]. The order G and C would result in serine incorporation, but the other sequence [4] would code for a threonine. The amino acid analysis performed on the purified protein indicated the presence of serine rather than an additional threonine (Table 1). Cells were grown to an absorbance of 0.8 at 660 nm. Synthesis of the merP gene product was then induced by adding isopropyl P-D-thiogalactoside using concentrations of 0.1, 0.2 and 0.5 mM. Cells were harvested after 2, 4, 6, or 8 h. There was no discernible increase in the proportion of P protein expressed with time, although 0.1 mM isopropyl P-Dthiogalactoside seemed to induce somewhat lower amounts.

378 Table 1. Amino acid composition. A quantitative amino acid analysis was performed on 2.9 pI MerP solution with A 2 , , = 0.701. The number of residues was calculated using c2,, = 2.8 m M ' cm :nd means not determined ~

Amino acid

Number of

Amount

residues from

in 2.9 p1

contained 16% acrylamide, 0.5% hisacrylamide. Lane A, molecular mass standards (value in Da on left). Lane B, supernatant after sonication and centrifugation. Lane C. supernatant after 20% ammonium sulfate precipitation. Lane D, supernatant after 50% ammonium sulfate precipitation. Lane E. supernatant from second precipitation with 50% ammonium sulfate. Lane F, pooled, concentrated, and dialyzed fractions from Phenyl-Sepharose column. Lane G, pooled and concentrated fractions from DEAE-Sepharose. Lane H, molecular mass standards (values in Da on right). Fig. 1. SDS gel of fractions from the purification. The gel

As estimated by scanning SDS gels, P protein made up about 20% of the total soluble protein. Purification of the merP gene product

Thr

+ Asn

Ser Glu + Gln

Pro GIY Ala

CYS Val Met Ile Leu

TYr Phe

nmol 9" 5" 3+3 3 4 10

Spectroscopic studies The ultraviolet/visible spectrum of the protein between 600 nm and 230 nm was obtained. The sample exhibited no absorbance above 310 nm, with a small peak in the aromatic region at 277 nm. There was a trough at 260 nm in the purest preparations, indicating the absence of nucleic acids. The ratio of absorbance at 211 nm divided by that at 260 nm was 1.4 in preparations of pure protein. Two samples of 2.9 pl and 5.9 p1

3.48 4.53

2.01 2.86 7.15

2 21 1 1

nd 7.68 0.74 0.70 2.13 0.71 1.48 0.06 6.87 0.73

3 1

2 0

LYS

9 1 0

nd ~

The P protein was purified as described in Materials and Methods. Two different procedures were used, one including cysteine in all the buffers, and one without cysteine. The presence or absence of cysteine did not seem to affect the yield. Typical yields from 7.5 1 were 20-30 mg protein. The protein prepar2tion contained a small amount of contaminants after the Phenyl-Sepharose column. These were removed in the following step using DEAE-Sepharose. The pure protein can be seen together with fractions from various steps in the purification procedure on the SDS/polyacrylamide gel in Fig. 1. A peak at 260 nm was detected in the spectrum of some preparations of P protein. This indicates the presence of contaminating nucleic acids. These can be removed by binding the protein to a hydroxyapatite column equilibrated in 10 mM potassium phosphate pH 7.3. The protein elutes off with 100mM buffer and the nucleic acids remain bound to the column. This step is usually not necessary, provided that the nuclease treatment is successful. A sample of the purified recombinant protein was subjected to N-terminal sequence analysis. The N-terminal sequence obtained, Ala-Thr-Gln-Thr-Val, corresponded to the N-terminus expected when the periplasmic signal peptide has been cleaved off [ll]. Protein was also prepared from a wild-type strain containing the plasmid pDU1003. In this preparation cysteine was omitted from the buffers throughout the preparation. The N-terminal sequence was also determined for a sample of this purified protein and found to be the same as for the overproduced protein.

2.98 5.85

4+0

His Arg TrP

Number of residuesi

molecule

DNA sequence u31 Asp

~

4.1 8.1 4.8

6.2 2.8 3.9 9.8

nd 10.6 1.o

1.o 2.9 1.o 2.0 0 Y.4 1.o nd

~~~

a According to the DNA sequence by Barrineau et al. [4], there would be 10 Thr and 4 Ser.

with an absorbance at 277 nm of 0.701 were submitted for quantitative amino acid analysis (Table 1). From these results, the absorption coefficient at 277 nm was calculated to be 2.8 mM-'cm-'. The CD spectrum of P protein prepared in the presence of cysteine from the overproducing strain is shown in Fig. 2. The spectrum was analyzed with the software provided with the spectrometer. There is a high degree of structure, approximately 80% a-helix, but also some p structure. CD spectra were also recorded of P protein prepared without cysteine present in the buffer, both from the overproducing strain and from a wild-type strain. The two spectra were very similar to each other, as well as to the spectrum of P protein purified in the presence of cysteine.

Thiol determinations The protein purified with cysteine present in the buffers was found to contain 1.9 thiol group/molecule that react with Nbsz in the native state. Protein denatured with SDS also contained 1.9 thiol group/molecule. The protein contained the same number of reactive thiol groups after storage for 20 h at 4°C in a cysteine-free buffer. The protein has two cysteines, according to the amino acid composition deduced from the DNA sequence [13]. When the cysteine was omitted from the protein purification buffers, there were no reactive thiols detectable on either the native or denatured protein. Protein with reactive thiol groups will from now on be referred to as 'reduced protein', while 'oxidized protein' refers to protein with no reactive thiol groups. It is not possible to reduce the oxidized protein with an excess of cysteine. The oxidized protein was incubated for 30 min at room temperature with a fivefold molar excess of cysteine, prior to gel filtration and thiol determination. The

379 15

50 40

10

30 n

c n 5

20

0

-0

E

v

0

0

10 0 0

-5

-10

.w--./-L,' \.I

I

I

190

200

210

220

I

1

230

240

250

Wavelength (nm) Fig. 2. CD spectra of P protein. The protein was in 50 mM potassium phosphate pH 7.3. Spectra were recorded at 25°C.Two of the spectra were multiplied by the factors indicated, to compensate for the difference in concentration. (----) Overproduced P protein, reduced, 11.4 pM; (. . . . . .) overproduced P, oxidized x 0.95; (-) wild-

Fig.3. Binding studies in the absence of external thiol compounds. 1.5 ml20 pM protein was mixed with 1.5 ml Hgz+ solution in 50 mM potassium phosphate pH 7.3. The mixture was centrifuged in Microsep microconcentrators. Approximately 100 pl filtered through. The concentration of bound Hg2+could be calculated measuring the concentration of non-bound Hg2+ in the filtrate.A blank with buffer instead of protein was made for each measurement and the value from the blank subtracted before analysis. (V)Oxidized P; ( A ) reduced P; (--)calculated curve based on the non-linear curve-fitting analysis giving Kd = 10.1 pM and a binding capacity of 48.2 pM; (----) calculated curve using Kd = 27.5 pM and binding capacity of 52.6 pM.

type P, oxidized x 4.

10 times the protein concentration. It was not possible to use higher concentrations of HgZ+,since the protein starts precipitating, as indicated by changes in the CD spectrum at these higher concentrations. The experiment was also performed with human carbonic anhydrase 11, a protein which has one cysteine residue/molecule, is inhibited by mercuric ions, and would not be expected to bind specifically an excess Native molecular mass of mercuric ions [23]. Carbonic anhydrase bound Hg2+ in a The molecular mass of the monomer can be calculated to fashion essentially identical to that observed for the P protein, be 7474 Da from the amino acid sequence. The molecular with similar K d and binding capacity (data not shown). The mass of the native protein was determined by gel filtration on curve-fitting analysis for the data from the P proteins gave a a Sephadex G-50 column. Both the reduced and the oxidized good fit with one Kd value. However, since there are only two protein were chromatographed under the same conditions as cysteines and few other amino acids with known high affinity the standard proteins. Both forms of the P protein eluted at for Hg2+ in the protein, it seems highly unlikely that there a volume corresponding to a molecular mass of 8700 Da. This would be five binding sites with equal Kd values. The determiresult shows that the P protein exists as a monomer in solution. nation of Kd values in this case is therefore probably not very meaningful, although the data points fit the curve. Together. these results suggest that although mercuric ions bind to the Binding studies P protein in the absence of excess thiol, there does not seem Binding studies were first attempted using the commonly to be a specific binding site with strong affinity for mercuric used technique of equilibrium dialysis. This technique had the ions. problem that the background values in the absence of protein The results shown in Fig. 4 are from an experiment where were very high: there was more Hg2+ bound to the dialysis cysteine was added to the mercuric ion solutions at a constant membrane than to the protein. Therefore, a different method cysteine/Hg2+ molar ratio of 4: 1. Here there is a clear differusing microconcentrators was utilized. ence between the reduced and the oxidized proteins. The Hg2+ binding studies were performed with both oxidized oxidized protein clearly has little affinity for mercuric ion. and reduced protein. Fig. 3 shows the concentration of bound Curve-fitting analysis of the data from the reduced protein versus total concentration of mercuric ions, when no external (10 pM) gave a binding capacity of 8.8 0.6 pM and an apcysteine is present. Non-linear regression analysis gave a bind- parent Kd of 3.7 1.3 pM. These results show that the reduced ing capacity of 48.2 2.3 pM for the oxidized protein. This protein has the capacity to bind approximately 1 Hg2+ ion/ is almost five times higher than the protein concentration protein molecule in the presence of excess thiol, but that (10 FM). The Kd value was 10.1 & 2.1 pM. The corresponding the oxidized protein only binds minute quantities of Hgz+. values for the reduced protein, which contains two thiol Considering that mercuric ions have a very high affinity for groups, are 52.6 f 6.2 pM and 27.5 f 8.3 pM, respectively. thiol compounds [24], this indicates that the binding to the The highest concentrations of Hg2 in these experiments were reduced P protein must be very strong. In view of this result,

number of reactive thiol groups/molecule had increased from 0 to 0.3. This indicates that the cysteine residues are readily accessible for oxidation (or reduction) only during the purification, perhaps during the sonification procedure.

*

+

380

Fig. 4. Binding studies in the presence of cysteine. The experiments were carried out as in Fig. 3, except that the Hgz+ solution also contained a fourfold excess of cysteine over H g 2 + .The data points are from two sets of experiments. ( 0 )Oxidized P; (0)reduced P; (-) curve calculatcd from the curve-fitting analysis giving Kd = 3.7 pM and a total binding capacity of 8.8 pM.

it is interesting to note that this specific binding cannot be detected as a distinct binding site in the experiment without external cysteine. Perhaps this indicates that a thiol compound is a necessary part of the protein -mercuric-ion complex.

DISCUSSION The cloning of the gene rnL>rPinto an overexpressing system and the purification of the P protein is described in this paper. The amino acid content of the protein obtained corresponded to that expected from the sequence of merP [13]. There was no indication that the protein had been proteolyzed. There have been reports that some periplasmic proteins do not become properly processed in overproducing strains, or that overproduced proteins are sometimes packaged in inclusion bodies [25]. There was no evidence that the P protein produced and purified as described was improperly processed. The native as well as the SDS gels showed a single band. Nterminal sequencing of protein prepared either from a wildtype strain or the overproducing strain showed that the Nterminal signal sequence had been cleaved off as expected between Ah19 and Ah20 [ll], corresponding to bases 10181939 [13]. The fact that the CD spectrum of the protein from the overproducing strain is identical with the spectrum of the protein from a wild-type strain, and that they both show a highly ordered structure, indicates that the protein is not only processed correctly but also folded correctly. The P protein was shown to be a monomer with a molecular mass of 7500 Da after processing of the periplasmic signal sequence. This is quite small compared to other purified periplasmic binding proteins which have molecular masses varying between 25000 - 56000 Da 1261. These other binding proteins often bind fairly large molecules, such as maltose, whereas Hg2+ is a comparatively small ion. Perhaps this is a Factor in the size difference. As shown in this work, there are two forms of the MerP protein. One contains no reactive thiol groups, while the other form has two. Both forms of the protein show binding of mercuric ions. However, of these two forms, only the reduced protein exhibits 1: 1 binding of mercuric ions specifically in the

prcsence of external thiol compounds. These thiol compounds may be a necessary component of the protein-ligand complex, since this strong binding site does not appear to exist in the absence of thiols. The question is, of course, which form dominates in the cell? Studies on the components of the periplasm have shown that it contains several proteins involved in electron transport and in the creation of a proton electrochemical gradient [27]. This indicates that the environment could be reducing, so that the thiols of the P protein are kept reduced. The oxidized form may then represent an artefact from the breakage of the cells, or the purification procedure. Bacterial transport systems can be divided into two groups: one that is insensitive to, and one that is sensitive to, osmotic shock. This latter group contains a periplasmic component, vital for transport, that is released when the outer cell membrane is destroyed. The periplasmic component is a protein that binds the substance that will be transported, and the transport is thus said to be 'binding-protein-dependent'. The most well-characterized systems in this group are the histidine and maltose transport systems [26].These consist of two membrane-bound proteins, a periplasmic binding protein, and a third membrane-associated protein believed to provide the energy necessary for transport. According to the proposed model [26], the molecule to be transported binds to the periplasmic binding protein. This induces a conformational change in the protein, which is then able to recognize one or both of the membrane-bound proteins. The protein/proteinligand interaction causes the ligand to be released from the binding protein, and transported over the inner membrane with the help of the two membrane proteins. A membraneassociated protein provides the necessary energy from ATP hydrolysis. What features does the mercuric ion transport system have in common with the histidine or maltose transport? This paper shows that the MerP protein has the ability to bind mercuric ions in vitro. Of the two described forms of the protein, only the reduced form binds H g 2 +in the presence of external thiol compounds. Thus the MerP protein is in possession of one of the functions proposed for periplasmic binding proteins. There are also two membrane proteins, MerT and MerC, in the mer operon of transposon Tn21, but only MerT in the operon from TnSUI and only MerC in the operon from Thiohacillus. All of these operons confer resistance to H g Z f , so the transport over the inner membrane does not seem absolutely to require two different proteins. One important feature of the histidine and maltose systems is the membraneassociated protein that can hydrolyze ATP. There has been no such component found in any of the mer operons sequenced so far. It is not likely that mercuric ion transport is passive, without the requirement for energy, but it is not clear where the energy comes from. The transport could be dependent upon the proton gradient across the inner membrane. So far the only similarity between Hg" and histidine or maltose transport systems is the presence of a periplasmic binding protein and the differences are quite important. The other transport systems studied involve transport of nutricnts into the bacteria, e. g. histidine, but Hg2'-resistant bacteria have a transport system for the uptake of a toxic compound. Hgz+ must be transported into the cell in order to be detoxified, since mercuric reductase, which reduces Hg2+ to Hg', is located in the cytoplasm and requires NADPH which is produced in the cytoplasm. Mercuric ions have a very high affinity for thiol compounds, and also the ability to rapidly interchange between thiols [24], so it is important that the

382 mercuric-ion-resistant bacteria have a system for preventing undesirable binding. Since the periplasm contains many different components, it seems likely that several of these could bind the mercuric ions and become inactivated. For the survival of the cell, it is therefore important that the MerP protein has a higher affinity for mercuric ions than other thiol groups have. Bacteria that are resistant to mercuric ions usually survive at levels up to 50 pM, whereas sensitive bacteria show growth inhibition at concentrations above 2 pM [3]. In this work it was shown that the reduced MerP protein has an apparent Kd of 3 pM in the presence of a fourfold excess of cysteine over Hg2 . This is low enough to be biologically relevant. The toxicity of the ion involved in transport could be the reason that there is a special protein with the sole function to ‘disarm’ the mercuric ion. Whether the MerP protein is also able to undergo conformational change upon Hg2+ binding and interact with the membrane proteins, remains to be investigated. Amino acid analysis was performed by Mr Ove Schedin (Dept. of

’

Medical Chemistry, Ume& Universitet). Wc are grateful to Dr Nancy Hamlett for help with the native gels, to Ms Eleonore Granstrom Skarfstad for exccllenl technical assistance, and to Drs S. Lindskog and J. Powlowski for comments on the manuscript. This project was supported by the Swedish Natural Science Research Council, (K-KU 8966-302) to Lena Sahlman.

REFERENCES 1. Summers, A. 0. & Silver, S. (1978) Annu. Rev. Micrmbiol. 32, 637 - 672. 2. Lund, P. A. & Brown, N. L. (1987) Gene 52,207-214. 3. Nakahara, H., Silvcr, S., Miki, T. & Rownd, R. H. (1979) J . Bucteriol. 140, 161 - 166. 4. Barrincau, P., Gilbert, P., Jackson, W. J., Jones, C. S., Summers, A . 0 . &Wisdom, S. (1984) J . Mol. Appl. Genet. 2,601 -619. 5. Ni’Bhriain, N. N., Silver. S. & Foster, T. J. (1983) J . Bacteriol. 155, 690-703. 6. Brown, N. L., Misra, T. K., Winnie, J. N., Schmidt, A,, Seiff, M . & Silver, S. (1986) Mol. Gen. Genet. 202, 143-151. 7. Lee, I. W., Gambill, B. D. & Summcrs, A. 0. (1989) J. Bacteriol. 17I, 2222-2225.

8. Nucifora, G., Silver, S. & Misra, T. K. (1989) Mol. G07.Genet. 220, 69-12. 9. Foster, T. J. & Ginnity, F. (1985) J . Bacteriol. 162, 773-776. 10. Lund, P. A,, Ford, S. J. & Brown, N. L. (1986) J . Gen. Microhiol. 132,465-480. 11. Summcrs, A. 0. (1 986) Annu. Rev. Microbiol. 40,607 - 634. 12. Brown, N. L. (1985) Trends Biochem. Sci. 10,400-403. 13. Misra, T. K., Brown, N . L., Fritzinger, D. C . , Pridmorc, R. D., Barnes, W. M., Haberstroh, L. & Silver, S. (1984) Proc. Null Acud. Sci. USA 81, 5975-5979. 14. Misrd, T. K., Brown, N. L., Haberstroh, L.. Schmidt, A,, Goddctte, D. & Silvcr. S. (1985) Gene 34, 253-262. 15. Shiratori, T., Inoue, C., Sugawara, K., Kusano. T. & Kitagawa, Y . (1989) J . Bacteriol. I71, 3458 - 3464. 16. Fierke, C. A., Krebs, J . F. & Venters, R. A. (1991) in Carbonic anhydrase: ,from biochemistry and genetics ro physiology und clinical medicine (Botrt, F., Gros, G. & Storey, B. T., eds) pp. 22 - 36, VCH Verlagsgesellschaft, Weinheim. 17. Kunkel, T. A. (1985) Proc. Nut1 Acad. Sci. U S A 82,488-492. 18. Sanger, F., Nicklen, S. & Coulson, A. R. (1977) Proc. Nut1 Acad. Sci. USA 74,5463 - 5467. 19. Jovin, T. M., Dante, M. L. & Chrambach, A. (1970) Nationul Technicullnforrnution Service, Public Board Numbers 196085 106092,203016,258309-259312, Springfield, VA. 20. Schiggcr, H. & Von Jagow, G. (1987) Anal. Biochem. 166,368319. 21. Beutler, E., Duron, 0. & Kelly, B. M. (1963) J . Lab. Clin. Med. 61,882-888. 22. Leatherbarrow, R. J. (1 990) CraAr Version 2.0. Erilhacus Softwarc Ltd., Staines UK. 23. Eriksson, A. E., Kylsten, P. M., Jones, A. T. & Liljas, A. (1988) Proteins Struct. Funct. Genet. 4 , 283 - 293. 24. Carty, A. J. & Malonc, S. F. (1979) in The hiogeachemistvy uf mercury in the environment (Nriagu, J . 0.. ed.) pp. 433-479, Elsevier/North-Holland Biomedical Press, Amsterdam. 25. Gribskov, M. &L Burgess, R. R. (1983) Gene 26, 109-118. 26. Ames, G . F.-L., Mimura, C. S. & Shyamala, V . (1990) F E M S Microhiol. Rev. 75,429 -446. 27. Ferguson, S. J. (1 988) in Bucterial energy trunsd~cction(Anthony, C., ed.) pp. 151 - 182, Academic Press, London.