Identification of the Genes for the Lactose-specific Components of the Phosphotransferase System in the lac Operon of Staphylococcus aureus*

The nucleotide and deduced amino acid sequences of the 1acE and lacF genes, which code for the lactose-specific Enzyme I1 and Enzyme I11 of the Staphylococcus aureus phosphotransferase system, are presented. The primary translation products consist of a hydrophobic protein of 572 amino acids (M, = 62,688) and a polypeptide of 103 amino acids (M, = 11,372), respectively. The assignment of lacF as the gene for Enzyme 111'" was based upon the known amino acid sequence of the protein. The identity of lmE as encoding Enzyme 11'" was based upon immunoreactivity of the cloned gene product with antibodies raised against purified Enzyme 11'" from S. aureus and an assay of biological function of the protein expressed in Escherichia coli. The order of the known genes of the S. aureus lac operon is lacF-lacE-lace, the latter encoding phospho-&galactosidase. Gram-positive aureus is uptake of the car-bohydrate by the PTS' Two which and are

staphylococcal chromosome (9,lO). We have cloned the phospho-P-galactosidase gene ( k c ) in Escherichia coli (5) and have determined its nucleotide sequence (11). This analysis indicated that lacG is the terminal determinant of a polycistronic operon. We present here the nucleotide sequence of the gene for EIILa' (lacE) and EIII'"' (la#). We present evidence that these cloned gene products, expressed in E. coli, are indeed EII'"" and EIII'"'. This is the first report of the primary sequence of an E11 and E111 of the PTS from a Gram-positive bacterium. The amino acid sequence of EII'"', deduced from the nucleotide sequence, displays structural similarities with EIImt' and EII"' of E. coli (12,13).

EXPERIMENTAL PROCEDURES
Bacterial Strains, Plasmids, and Bacteriophage-E. coli JMlOl was used as a host for the sequencing coliphage vectors Ml3mplO and Ml3mpll (14). Plasmids pFB34 and pFB40 (5) were maintained in E. coli LE392 (15). E. coli CMK carries the cloned EIII'" from Lactobacillus casei.2 The S. aureus strains used were wild-type strain S5601 and strain S305A, constitutive for the lactose-specific components of the PTS (16).
Determination of Nucleotide Sequence-DNA manipulations were as previously reported (5). DNA sequencing was conducted by the dideoxy chain termination method (17). Computer analysis on the sequence was carried out with "Seqaid" (18), a software package provided by D. Rhoades and D. Roufa (Kansas State University). Hydropathy analysis of the protein product of lacE was as described by Kyte and Doolittle (19). Preparation of Membranes-S. aureus strains S305A and S5601 (not induced for the lac operon) were cultivated, harvested, and disrupted as described previously (20). The crude membranes were sedimented at 13,000 X g overnight or at 170,000 X g for 4 h. E. coli LE392 harboring the plasmids pFB34 or pFB40 ( 5 ) were grown overnight in Luria broth containing 50 mg/liter ampicillin. 15 g of cells from 9 liters of culture were suspended in 40 ml of buffer A (0.05 M Tris-HC1, 0.1 mM EDTA, 0.1 mM dithiothreitol, pH 7.5) and disrupted by sonication. Membranes were sedimented as described above and washed with buffer to remove residual basal level 8galactosidase, which interfered with the S. aureus mutant complementation assay (16). The supernatant was chromatographed on a Sephadex G-75 column (70 X 5 cm) to remove E. coli @-galactosidase.
This step was necessary to assay for EM"' of S. aureus.
Purification of Denatured E I P of S. aureus-One g of crude membranes were suspended in 10 ml of buffer A. The pH of the suspension was adjusted to 12.4 by the addition of 1 ml of 1 M NaOH. EII'"' was then sedimented at 170,000 X g for 3 h at 4 "C. Up to 0.2 g of the sediment were homogenized in 2 ml of buffer B (0.08 M Tris/ glycine, 0.1 mM EDTA, 0.1 mM dithiothreitol, 0.1 mM lactose, pH 9.3) (6). One ml of S305A "dansylated membranes" and 1.5 ml of 3 X sample buffer3 were added and applied to preparative SDS-polyacrylamide gel electrophoresis. For the preparation of EII'"', 10% acryl- To produce the internal fluorescent protein standard, 15 mg (wet weight) of NaOH-extracted membranes were dissolved in 0.85 ml of NaHC03 (0.1 M, pH 9) and 0.15 ml of 20% SDS. Forty pl of a 10% solution of dansyl chloride in acetone was added with vigorous shaking. The mixture was incubated at 70 "C for 2 min; the reaction was stopped by adding 20 pl of 0-mercaptoethanol and again heating for 2 min at 70 "C. Low molecular reaction products were removed by passage through a Pasteur pipette filled with Sephadex G-25 equilibrated with gel sample buffer. The separation was followed under UV light. The mixture could be stored frozen (21).
Production of Anti-EIP Antibodies-540 pg of electroeluted concentrated EII'"' were homogenized with complete Freund's adjuvant and injected intramuscularly into a rabbit. Thirty-five days later, a subcutaneous booster injection with the same amount of EII'" in Freund's complete adjuvant was made. A second booster injection was performed at day 63 with 180 pg of protein in incomplete adjuvant. The rabbit was drugged with ether at day 72 and exsanguinated, which yielded 200 ml of blood.
Western Blot Analysis-Western blotting was carried out according to the Bio-Rad Immun-Blot protein A horseradish peroxidase conjugate manual. Antiserum was applied in a dilution of 1:60. The proteins were transferred from an analytical SDS-polyacrylamide gel to a nitrocellulose sheet (Schleicher and Schuell, 0.45 pm). The transfer was performed at 4 "C for 30 min at 10 V/cm and then for 1 h at 15 V/cm. The transfer buffer had the following composition: 20% methanol, 3 g of Tris, 14.4 g of glycine, and 1 g of SDS/lOOO ml.
Assay of E I P Actiuity-A mutant complementation assay with an extract of strain 825A, defective in EII'"', was used as previously described (22).
Phosphorylation of E I P of S. aureus with f2P]PEP-Phosphorylation of EII"' was performed in 44 p1 of 50 mM NH,HCO, with the following components: 500 nmol of MgC12, 5 pg of Enzyme I from Streptococcus fuecalis (23), 2 pg of HPr from S. fuecalis (24), 2 pg of EIII'" of S. aureus (7), 100 pg of membrane protein:EII'"", 0.1 pCi of ["PpIPEP (25). To identify labeled proteins on an SDS gel, reaction mixtures without EII, EIII, and HPr were run. To demonstrate strict PEP dependence, 5 nmol of ATP were included; specific dephosphorylation of EII'" was achieved by the addition of 5 nmol of isopropyl fhbthiogalactopyranoside to the complete phosphorylation mixture.

RESULTS AND DISCUSSION
Nucleotide Sequence of the lacF and lacE Genes-The S.
aureus lac genes were cloned on a 5.1-kilobase EcoRI-BarnHI DNA fragment (5). The sequence described here begins at an RsaI site, 370 base pairs from the EcoRI site, and extends 2151 nucleotides to the ATG initiation codon of the gene for phospho-@-galactosidase (la&). The lacG sequence has been reported (11). The restriction map of and sequencing strategy for the lac DNA are shown in Fig (19) indicates that the NH2-terminal 75% of the protein is very hydrophobic (Fig. 3). In this program, the amino acids are scored according to their hydropathy, with isoleucine scoring +4.5 and arginine scoring -4.5. The plot for the COOH-terminal 25% of the protein fluctuates about the midline. This pattern is similar to that reported for EIImt' (12) and EIIg" (13) of E. coli, although the EIImt' protein domains each comprise half of the protein. The EIIbg' of E.
colidoes not display this two-domain hydropathy pattern (27). Membrane-spanning portions of proteins have been shown to have an average hydropathy of +1.22 to +2.65 over a 19residue segment (19). At least six regions of the EIIInc have sufficient hydrophobic character to be able to traverse the membrane. Seven such possible membrane-spanning regions have been reported for EIImt' (12) and EIIg'" (13). The average     The deduced amino acid composition of the lacF and lacE gene products is presented in Table  I. Consistent with the hydropathy analysis, the la& product contains a high percentage of the nonpolar amino acids alanine, phenylalanine, glycine, isoleucine, leucine, and valine. The lacE protein thus has features suggesting it is EII'"', although it is larger than the reported size of this protein ((6) see below). The size disparity may be due to protein processing or anomalous migration of this hydrophobic protein in SDS gels. It is of interest to note that this protein was not detected by E . coli maxicell analysis (5), despite the presence of 19 methionine residues for radiolabeling. Identification of the lacE Gene Product as EII'"-SDSpolyacrylamide gel electrophoresis of NaOH-extracted membranes of the lactose-constitutive S. aureus strain S305A showed a prominent 48-kDa protein band, which was absent in the wild-type strain S5601 not induced for lactose metabolism (data not shown). This is smaller than the size previously determined for EII1aC (6). The discrepancy may be a function of the different gel systems used. Purified EII'"' was prepared and used to immunize rabbits. The resulting antiserum was used to probe lysates of E. coli carrying plasmid pFB34. As shown in Fig. 4A, the antibodies recognized a 48-kDa species in the lysate from pFR34-carrying cells, but not with a control E. coli lacking this plasmid. The protein produced in E. coli comigrates with purified EII'"' from S. aureus (Fig. 4R). The same immunoreactive protein was observed with LE392 cells harboring pFR40 ( 5 ) , a deletion subclone of pFR34 which carries the entire lacE coding sequence (data not shown).

G I K L A K T Q G A E Y I K L T R D G Q
Biological activity of the product of the cloned lacE gene was measured using a mutant complementation assay. Values obtained with E. coli LE392 (pFR34) and S. aureus S305A were 15 and 60 nmol of o-nitrophenyl-8-D-galactopyranoside 6-phosphate/min/mg of membrane protein, respectively. The biological activity and immunoreactivity thus establish the lacE gene as the determinant of EII'"' of S. aureus.
Codon Usage Analysis-Work with E. coli has established that there is a correlation between expression level of a gene and its codon usage pattern. The more highly expressed genes display a nonrandom pattern of codon usage, utilizing a restricted set of codons which are recognized by major species of isoacceptor tRNAs, while genes expressed a t low levels have a more random pattern of codon usage (28, 29). This trend may also hold true for the Gram-positive bacterium Bacillus subtilis (30). T h e codon usage patterns for the three sequenced genes of the S. aurew lac operon are presented in Table 11. T h e codon usage is biased toward A-or U-rich codons reflective of the relatively low G+C content of S . aureus DNA (32-35 mol 96). However, some codons such as AUA (isoleucine), which are rarely used in highly expressed genes in E . coli, are infrequently used in these S. aureu.9 messages despite its A+U-rich nature. This bias against AUA is not found in all S. aureus genes (1 1).
Organization of the S. aureus lac Operon and Intercistronic Region.+" stem-loop structure resembling a transcription terminator is positioned immediately 3' to lacC suggesting this is the terminal gene of this transcriptional unit (11). T h e gene order of the operon is lacF-lacE-la& Additional significant open reading frame(s) can be found upstream of lacF.  Furthermore, no sequences resembling the consensus E. coli/ R. subtilis promoter sequence are found within the sequence reported here. Therefore, the operon must contain additional determinants promoter proximal to lacF. The initiation codons for each of the known genes of the lac operon are preceded by a ribosome binding site sequence (Fig. 5 ) . Rased upon the rules of Tinoco et al. (31), the calculated free energies for the interaction of these sequences with the 3'-end of R. subtilis 16  suhtilis hosts (34). T o calculate the distances between the ribosome binding sites and the downstream initiation codons, we followed the convention of measuring from the first base to the right of the preferred E. coli ribosome binding sequence AGGA (or the equivalent position when this sequence was not present) through the base adjacent to the initiation codon (33). These distances ranged from 9 t o 11 bases, within the range of values reported for genes of R. suhtilis (33). The open reading frame of lacE shown in Fig. 2 begins with the first ATG following the ribosome binding site. The first potential initiation codon has a spacing of only a single nucleotide pair from the AGGA-equivalent sequence. This short spacing makes the ribosome binding site proximal ATG possibly not the true initiation codon (unless initiation of translation  occurs without the requirement of the ribosome binding sequence). Two additional ATG codons are found 4 and 7 nucleotides 3' to the first ATG. All three ATG codons are in the same reading frame; use of either of the latter two would omit two (Met-Thr) or three (Met-Thr-Met) amino acids from the amino terminus of the protein. The third initiation codon, 10 nucleotides from the ribosome site, has the more favorable spacing. There is little intercistronic space in this operon. The ribosome binding site sequence for lacE is within the terminal coding sequence of lacF, that of LacG overlaps by one nucleotide pair the ochre termination codon of lacE, while the ribosome binding sequence of lacF is 5 base pairs downstream of the ochre codon of the preceding open reading frame. The buried nature of the ribosome binding site for lacE raises the possibility that translating ribosomes may initiate at the first AUG codon, while a ribosome initiating translation at this ribosome binding site may require a greater spacing and thus initiate at the third AUG codon. This possibility will be resolved when the N-terminal amino acid sequence of the protein becomes available.
Phosphorylation of EII'" of S. aureus-As shown in Fig. 6  oside) which is a substrate for the lactose PTS. The phosphorylated residue is most likely histidine (35,36). The deduced amino acid sequence of the lacE gene product revealed only four histidine residues. This feature of the S. aureus EII'"' should greatly facilitate the localization of the active site histidine by means of site-directed mutagenesis.