Membrane anchoring of the AgrD N-terminal amphipathic region is required for its processing to produce a quorum-sensing pheromone in Staphylococcus aureus.

Quorum-sensing pheromones are signal molecules that are secreted from Gram-positive bacteria and utilized by these bacteria to communicate among individual cells to regulate their activities as a group through a cell density-sensing mechanism. Typically, these pheromones are processed from precursor polypeptides. The mechanisms of trafficking, processing, and modification of the precursor to generate a mature pheromone are unclear. In Staphylococcus aureus, AgrD is the propeptide for an autoinducing peptide (AIP) pheromone that triggers the Agr cell density-sensing system upon reaching a threshold and subsequently regulates expression of virulence factor genes. The transmembrane protein AgrB, encoded in the agr locus, is necessary for the processing of AgrD to produce mature AIP; however, it is not clear how AgrD interacts with AgrB and how this interaction results in the generation of mature AIP. In this study, we found that the AgrD propeptide was integrated into the cytoplasmic membrane by a conserved alpha-helical amphipathic motif in its N-terminal region. We demonstrated that membrane targeting of AgrD by this motif was required for the stabilization of AgrD and the production of mature AIP, although this region was not specifically involved in the interaction with AgrB. An artificial amphipathic peptide replacing the N-terminal amphipathic motif of AgrD directed the protein to the cytoplasmic membrane and enabled the production of AIP. Analysis of Bacillus ComX precursor protein sequences suggested that the amphipathic membrane-targeting motif might also exist in pheromone precursors of other Gram-positive bacteria.

Quorum sensing is a way that bacteria communicate with each other (1). Bacteria produce autoinducing molecules at basal levels. When bacterial population density increases, the autoinducing molecule concentration reaches a threshold that results in the activation of the quorum-sensing system and subsequently alters the cell activity that enables bacteria to change their behavior as a group like multicellular organism. Quorum sensing is involved in the regulation of sporulation, mating, bioluminescence, virulence factor expression, and bio-film formation. Although acylated homoserine lactones with different acyl side chains are common autoinducers for Gramnegative bacteria, Gram-positive bacteria use small peptides as pheromones (2). Well studied examples of peptide pheromones include the competence-stimulating peptide from Streptococcus pneumoniae (3,4), lantibiotics and the bacteriocin-inducing peptide of lactic acid bacteria (5), the ComX pheromone and competence and sporulation factor of Bacillus subtilis (6), sex pheromones of Enterococcus faecalis, and the autoinducing peptide (AIP) 1 in staphylococci (7). These pheromones are typically processed from precursor polypeptides with modifications in some cases. A leader peptide with a conserved double glycine sequence found in the lactobacterial lantibiotics and bacteriocin-inducing peptides (5) is considered important for propeptide trafficking. ATP-binding cassette transporters with proteolytic activity are postulated to be responsible for both processing of the propeptides and secretion of the mature pheromones. It is interesting to note that the precursor of the E. faecalis sex pheromone is first generated from a lipoprotein precursor by a signal peptidase. This precursor molecule, located in the membrane, is then processed through intramembranous proteolysis by a zinc metalloprotease (Eep) (8,9), although the catalytic mechanism has yet to be determined. The leader peptide is also predicted in the competence and sporulation factor precursor, but the trafficking and processing of this precursor are not clearly defined.
In Staphylococcus aureus, expression of the virulence factors is coordinately regulated by a quorum-sensing system encoded by the agr (accessory gene regulator) locus (7). The agr locus encodes four proteins: AgrA, AgrB, AgrC, and AgrD. AgrC and AgrA compose a two-component signal transduction pathway, in which AgrC is the sensor kinase, and AgrA resembles a response regulator (10,11). AgrD is the precursor of AIP, a peptide pheromone that is secreted from the bacteria into the culture medium (12,13) and that is the ligand of AgrC, a membrane protein with its N-terminal half, which has been proposed to contain the AIP-binding site, anchored in the cytoplasmic membrane via five transmembrane helices (14,15) and with its C-terminal half, which composes a histidine kinase domain, located in the cytoplasm (11). AgrB is required for the production of AIP (13) and functions as a protease that is involved in the proteolytic cleavage of AgrD (16). It has also been proposed that AgrB functions as an oligopeptide transporter that facilitates the secretion of mature AIP (10,16). However, the interaction between AgrD and AgrB and the mechanisms of the processing of AgrD and the secretion of mature AIP by AgrB remain unclear. Based on the DNA sequences of agr loci and the specific interactions between AIP and AgrC and between AgrD and AgrB, four S. aureus groups have been defined (13,17). AIP from an S. aureus strain activates the Agr response in itself and in the same group members, but inhibits the Agr response in heterologous group members. The four S. aureus AgrD groups have 46 or 47 residues, and the AIPs have different amino acid sequences and lengths. An alignment of these AgrD groups (group I AgrD, NCBI accession number CAA36782; group II AgrD, accession number AAB63265; group III AgrD, accession number AAB63268; and group IV AgrD, accession number AAG03056) is shown, with the AIP sequences in boldface.
The specific interaction between AgrB and AgrD is not so strict. AgrD of S. aureus group II is processed only in the presence of group II AgrB, whereas group I AgrD and group III AgrD are processed by group I AgrB and vice versa (7). In this study, we found that the AgrD propeptide was anchored in the cytoplasmic membrane by an N-terminal amphipathic region. We also demonstrate that the membrane targeting of AgrD by the amphipathic region was required for its normal processing to produce mature AIP, but was not involved in the specific interaction with AgrB. This amphipathic region is present in all staphylococcal AgrD sequences available in the Gen-Bank TM /EBI Data Bank as revealed by sequence analysis. Furthermore, analysis of B. subtilis ComX sequences revealed the existence of similar amphipathic regions in their N-terminal parts, suggesting that the membrane targeting sequence might play an important role in the processing of the propeptides to generate active mature pheromones in other Gram-positive bacteria.

EXPERIMENTAL PROCEDURES
Bacterial Strains and Culture Conditions-The S. aureus plasmids and strains used in this study are listed in Table I. S. aureus cells were grown in CY-GP broth (18), supplemented with antibiotics (5 g/ml chloramphenicol and 5 g/ml erythromycin) when necessary. Bacteria grown overnight at 37°C on GL plates (18) were routinely used to inoculate liquid cultures. Cell growth was monitored with either a Klett-Summerson colorimeter with a green (540 nm) filter (Klett, Long Island City, NY) or a VERSAmax microplate reader (Molecular Devices) at A 650 nm . Escherichia coli strain MC1061-5 (19) was grown in LB broth, supplemented with tetracycline (20 g/ml) when necessary.
Construction of S. aureus Plasmids-The group I AgrD-His 6 expression plasmid pLZ4009 was constructed as follows. A PCR product was generated using primers GJ56 (5Ј-GCTCTAGAAGCTATTACATTATT-ACC-3Ј, before the Shine-Dalgarno sequence of agrD, with the XbaI site underlined) and GJ28 (5Ј-CTAATGATGATGATGATGATGTTCGTGT-AATTGTGTAATTC-3Ј, with six histidine codons and one stop codon underlined and 3Ј of agrD italicized) and pRN6852 (12) as the template. The PCR product was then digested with XbaI and cloned into the pRN5548 XbaI and EcoRI (blunted with Klenow fragment) sites. The group II AgrD-His 6 expression plasmid pLZ4010 was made by PCR amplification of the SA502A agrD gene from pRN6958 (13) with primers GJ44 (5Ј-CTATTATTCCATGGACTTCATTTAC-3Ј, sequence around the NcoI site of the pRN5548 plasmid, with the NcoI site underlined) and LZ31 (5Ј-ATTAATGATGATGATGATGATGTTTGTCGTA-TAAATTCGTTAATT-3Ј, with six histidine codons and one stop codon underlined and 3Ј of group II agrD italicized). The PCR product was digested with NcoI and ligated to a SmaI-and NcoI-digested 2.2-kb fragment of pRN5548.
Plasmid pLZ4017 was constructed as follows. A PCR product was amplified from pLZ4011 with primers LZ48 (5Ј-AACATCGCAGCTTA-TAGT-3Ј, group I agrD sequence) and tmR (5Ј-CAGTGTGGCAATCAC- with the E. coli leader peptidase I gene sequence underlined and the pRSET-A multicloning site sequence italicized), digested with BspHI, and dephosphorylated with shrimp alkaline phosphatase (MBI Fermentas). This fragment was ligated to a BspHI-digested and 5Ј-phosphorylated PCR fragment amplified from pLZ4011 with primers GJ45 and tmF (5Ј-GTGACGGGCATTTTATGGTGCGTGCGGCGGATTTTAAAA-AACATTGGTAACATCGCAG-3Ј, with the E. coli leader peptidase I gene and two arginine codons underlined and the agrD sequence italicized). The resulting plasmid contained the coding sequence of the N-terminal transmembrane region of E. coli leader peptidase I (21 amino acids) followed by two arginine codons and the agrD sequence encoded the C-terminal 32 amino acids. This plasmid was used as a template to produce PCR products with primer pairs GJ44/LZ17 and GJ45/lepF (5Ј-GCGAATATGTTTGCCCTG-3Ј, the E. coli leader peptidase I coding sequence starting from the second codon). The PCR products were then digested with NcoI, 5Ј-phosphorylated, and ligated to produce pLZ4017. All PCR products used for plasmid construction were amplified with Pfu Turbo® high fidelity DNA polymerase (Stratagene). The correct sequence of every newly constructed plasmid was confirmed by DNA sequencing.
AIP Activity Assays-AIP activities were measured using S. aureus cells containing an agr P3-blaZ fusion on plasmid pRN6683 (20) according to the method described previously (12).

PhoA Fusion Plasmid Construction and Expression in E. coli Strains-The
AgrD-PhoA fusion protein expression plasmid pLZ5001 was constructed as follows. A PCR product amplified from pGJ4002 (16) with primers GJ93 (5Ј-GAAGATCTTCGTGTAATTGTGTTAATTC-3Ј, with the BglII site underlined) and GJ94 (5Ј-CGGGATCCATGAATA-CATTATTTAACTTATTTTTTG-3Ј, with the BamHI site underlined) was digested with BglII and BamHI. This product was then inserted into the BglII site of pAWLP-2 (16). Plasmid pLZ5002, encoding a protein with only the N-terminal region of AgrD fused to PhoA (AgrDN-PhoA), was constructed by deleting the coding sequence for the Cterminal region of AgrD in pLZ5001. A PCR product amplified from pLZ5001 with primers LZ10 (5Ј-GAAGATCTTTTTTGCAGCTCAG-3Ј, with the BglII site underlined) and LZ15 (5Ј-GAAGATCTCTGCGAT-GTTACCAATGT-3Ј, with the BglII site underlined) was digested with BglII and DpnI and then self-ligated. E. coli strain MC1061-5 transformed with pLZ5001 or pLZ5002 was cultured overnight for PhoA activity assay or cell fractionation and Western blot hybridization.
Alkaline Phosphatase (PhoA) Activity Assays-PhoA activity was measured as described (21) with modifications. E. coli MC1061-5 cells expressing AgrD-PhoA fusion proteins were grown in LB broth (19) at 37°C overnight. Cells were collected, washed once with 1 M Tris buffer (pH 8.0), and suspended in the same buffer containing 1% SDS and 10% chloroform, followed by vortexing for 15 min to extract proteins. p-Nitrophenyl phosphate (Sigma 104 solution) was then added to the mixtures. After a 30-min incubation at 37°C, K 2 HPO 4 solution (13%) was added to stop the reactions. The reaction mixtures were centrifuged, and the clear supernatant was measured at both 420 and 550 nm. Alkaline phosphatase activity units were calculated by a formula described previously (21).
Cell Fractionation-Harvested S. aureus cells were suspended in 1ϫ sucrose/sodium maleate/MgCl 2 solution containing 10 g/ml lysostaphin and incubated for 30 min at 37°C. The protoplasts were then lysed by addition of phosphate-buffered saline (PBS) or as otherwise indicated and supplemented with 1 mM phenylmethanesulfonyl fluoride and protease inhibitors. The cell lysate was briefly sonicated, and the cell debris was removed by centrifugation at 7000 ϫ g for 10 min at 4°C. Total protein levels of the lysates were measured with Bio-Rad protein assay dye reagent at A 595 nm . Cell membrane fractions were separated by ultracentrifugation at 200,000 ϫ g for 2 h at 4°C. E. coli MC1061-5 cells containing pLZ5001 or pLZ5002 suspended in PBS with 1 mg/ml lysozyme were incubated at 4°C for 1 h and then lysed by brief sonication after addition of 1 mM phenylmethanesulfonyl fluoride. Membrane fractions were prepared by ultracentrifugation. The pellets were dissolved in 1ϫ SDS sample buffer (19) or as otherwise indicated.
SDS-PAGE and Western Blotting-Protein samples dissolved in SDS sample buffer were incubated at 70°C for 10 min before being separated by either Tris/glycine/SDS-PAGE (19) or Tris/Tricine/SDS-PAGE (22). The separated proteins were then electrophoretically transferred to polyvinylidene difluoride membranes (Millipore). After blocking overnight at 4°C in Tris-buffered saline (19) containing 0.1% Tween 20 and 5% bovine albumin (Sigma), the polyvinylidene difluoride membranes were incubated in blocking buffer with a primary antibody (1:2000 dilution of mouse anti-tetrahistidine monoclonal antibody (QIAGEN Inc.) or 1:5000 dilution of anti-T7 tag monoclonal antibody (Novagen)) for 1 h at room temperature. The membranes were washed extensively with Tris-buffered saline plus Tween 20 and then probed with horseradish peroxidase-conjugated sheep anti-mouse secondary antibody (Amersham Biosciences). The immunoblots were detected with an ECL Plus Western blot detection system followed by exposure to Hyperfilm TM ECL TM (Amersham Biosciences).

RESULTS
AgrD Is an Integral Membrane Protein-To determine the subcellular location of AgrD, we made two plasmids in which six histidine codons were added to the 3Ј-end of either the group I or II agrD gene and inserted downstream of the bla promoter of the pRN5548 expression vector (23). Plasmids pLZ4009 and pLZ4010 were transformed into agr-null S. aureus strain GJ2035, creating strains LZ4009 and LZ4010, respectively. Western blot hybridization analysis with the antitetrahistidine monoclonal antibody used as a probe revealed a band with an estimated size of ϳ6 kDa from whole cell lysate of LZ4009 or LZ4010 expressing His 6 -tagged AgrD (Fig. 1A, lanes  4 and 7). No band was observed from cell lysate of LZ0001 carrying pRN5548 (Fig. 1A, lane 1). The sizes of the proteins detected were consistent with that of the predicted molecular mass of AgrD-His 6 . These ϳ6-kDa bands were detected only in the membrane fractions prepared from LZ4009 and LZ4010 (Fig. 1A, lanes 6 and 9), but not in the cytoplasmic fractions ( lanes 5 and 8), indicating that AgrD is a membrane-associated protein.
To determine whether AgrD is an integral or peripheral membrane protein, we prepared protoplasts from S. aureus LZ4009 cells and lysed the protoplasts in PBS containing 1 M sodium chloride, in 0.2 M sodium carbonate, or in PBS containing 1% sarcosyl. After incubation at 4°C for 2 h, the whole cell lysates were fractionated by ultracentrifugation. Fig. 1B showed the results of Western blot hybridization with the antitetrahistidine monoclonal antibody as a probe. AgrD-His 6 was not extracted from the membrane fraction by 1 M sodium chloride (Fig. 1B, lane 3) and was only partially extracted from the membrane fraction by 0.2 M sodium carbonate (lane 5). We note that, under both conditions, only peripheral membrane proteins are dissolvable (24,25). However, this His 6 -tagged protein was dissolved in 1% sarcosyl (Fig. 1B, lane 7), an anionic detergent proven to preferentially dissolve integral membrane proteins (26). Similar results were obtained from Western blot analysis of LZ4010 expressing group II AgrD-His 6 (data not shown). Because the His 6 tag is highly hydrophilic and unlikely to be associated with the lipid bilayer, these results strongly suggest that AgrD is an integral membrane protein.
The N-terminal Region of AgrD Is Anchored in the Cytoplasmic Membrane-The full-length RN6390B agrD gene or the 5Ј-region of the gene coding the N-terminal 23 residues was prepared by PCR and then inserted into the E. coli phoA fusion plasmid pAWLP-2, creating two translational in-frame phoA fusion plasmids, pLZ5001 and pLZ5002, respectively. Expression of the fusion proteins in E. coli MC1061-5 was evaluated by Western blot hybridization of the whole cell lysates with a rabbit anti-E. coli PhoA polyclonal antibody (kindly provided by Dr. Andrew Wright, Tufts University School of Medicine). Both AgrD-PhoA and AgrDN-PhoA fusion proteins with correct molecular masses were detected (Fig. 2, lanes 1 and 7). A nonspecific or endogenous PhoA protein band with approximate molecular mass of 55 kDa appeared in samples including the control (Fig. 2, lane 6), as we previously found in strains expressing AgrB-PhoA fusions (16). We noted that, in the pAWLP-2 plasmid, due to addition of the artificial cloning site, the phoA gene was not expressed, so the weaker anti-PhoA response band from cells with this plasmid alone might be due to the unspecific interaction between the anti-PhoA polyclonal antibody and other E. coli proteins and/or the endogenous PhoA protein since the E. coli strain we used is PhoA ϩ . The reason we saw strong bands in cells expressing AgrD-PhoA fusion proteins at the same position as in the control might be due to the degradation of AgrD-PhoA fusion proteins, and these degraded products might be recognized by the polyclonal antibody. The anti-PhoA interactive proteins appeared predominately in the membrane fraction and were completely dissolvable in 1% sarcosyl (Fig. 2, lanes 4 and 10), but were not extracted from the membrane fractions by either 1 M sodium chloride or 0.2 M sodium carbonate (data not shown). Because AgrDN-PhoA was anchored in the membrane, these results suggest that the N-terminal half of AgrD is the region that is integrated into the lipid bilayer. The alkaline phosphatase (PhoA) activity of the E. coli strain expressing either the AgrD-PhoA or AgrDN-PhoA fusion protein was measured. The PhoA activities of these two strains were no more than that of the strain carrying the vector pAWLP-2 (data not shown), indicating that the C-terminal part of AgrD is located in the cytoplasm. These results suggest that the AgrD propeptide could be either a transmembrane protein with the N terminus outside and the C terminus inside of the membrane or a membraneanchoring protein with both the N and C termini in the cytoplasm.
An N-terminal Amphipathic Motif Is Conserved in Staphylococcal AgrD Proteins-As our results indicated that AgrD was  1, 4, and 7) and cytoplasmic (lanes 2, 5, and 8) and membrane (lanes 3, 6, and 9) fractions prepared from S. aureus cell expressing AgrD-His 6 with the anti-tetrahistidine monoclonal antibody as a probe. Cells were grown and induced; protoplasts were prepared and lysed in PBS; and membrane and cytoplasmic fractions were separated by ultracentrifugation as described under "Experimental Procedures." Proteins were separated by Tris/Tricine/SDS-PAGE. Lanes 1-3, control strain LZ0001 (GJ2035 containing the cloning vector pRN5548); lanes 4 -6, strain LZ4009 (GJ2035 containing RN6390B agrD-His 6 in pRN5548); lanes 7-9, strain LZ4010 (GJ2035 containing SA502A agrD-His 6 in pRN5548). The cytoplasmic and membrane fractions were prepared from total cell lysates containing 150 g of total proteins. B, extraction of the AgrD protein from the cell membrane. Protoplasts of LZ4009 were lysed in different solutions and subject to ultracentrifugation. Each pellet was dissolved in the original solution with the same volume. Supernatant (lanes 1, 3, 5, and 7) and pellet (lanes 2, 4, 6, and 8) fractions were then mixed with the same volume of 2ϫ SDS sample buffer, and the proteins were separated by Tris/Tricine/SDS-PAGE. Western blot hybridization was performed with the anti-tetrahistidine monoclonal antibody as a probe. Lanes 1 and 2, PBS containing 150 mM NaCl; lanes 3 and 4, PBS containing 1 M NaCl; lanes 5 and 6, 0.2 M Na 2 CO 3 ; lanes 7 and 8, 1% sarcosyl in PBS.

FIG. 2. The N-terminal region of AgrD is anchored in the cytoplasmic membrane.
AgrD-PhoA fusion proteins were expressed in E. coli strain MC1061-5. Cells were harvested from overnight cultures and lysed by brief sonication in PBS plus lysozyme. The cell lysates were then fractionated by ultracentrifugation. The membrane (pellet) fraction was suspended in PBS or dissolved in 1% sarcosyl in PBS, and the latter was then separated by ultracentrifugation. All samples were separated by Tris/glycine/SDS-PAGE. The rabbit anti-PhoA antibody was used as a probe for Western blotting. Lanes 1-5, AgrDN-PhoAexpressing strain; lane 6, total cell lysate of the control strain containing the pAWLP-2 vector; lanes 7-11, AgrD-PhoA-expressing strain. Lanes 1, 6, and 7, total cell lysate in PBS; lanes 2 and 8, cytoplasmic fraction; lanes 3 and 9, membrane fraction; lanes 4 and 10, membrane fraction dissolved in 1% sarcosyl in PBS; lanes 5 and 11, pellet from 1% sarcosyl-dissolved membrane fraction after ultracentrifugation. anchored in the cytoplasmic membrane, we asked how AgrD is associated with the membrane lipid bilayer. Hydrophobicity analysis of 24 AgrD sequences from various staphylococcal species retrieved from the GenBank TM /EBI Data Bank failed to find a conserved transmembrane helix among those AgrD sequences (data not shown). A hydrophobic moment plot (27) of each of the 24 AgrD sequences predicted that the N-terminal region of AgrD could form an amphipathic ␣-helix of various lengths. Helical wheel presentations of the first 18 residues of both group I and II AgrD proteins are shown in Fig. 3. It is clear that all the hydrophilic residues are located on one side of the helix. This feature is conserved in all AgrD proteins but two. These two Staphylococcus auricularis AgrD proteins contain one threonine located in the middle of the hydrophobic side that may not interfere with the overall hydrophobic moment. The frequencies of hydrophilic residues are different among AgrD proteins; this may explain why some AgrD sequences were predicted to have transmembrane region and others were not. These results imply that a common ␣-helical amphipathic motif exists in the N-terminal region of staphylococcal AgrD proteins and that this motif is likely to be anchored in the cytoplasmic membrane.
The N-terminal ␣-Helical Amphipathic Motif of AgrD Is Required for Its Membrane Anchoring and Processing-To investigate the function of the N-terminal ␣-helical amphipathic motif of AgrD, we made several AgrD mutants with various residues deleted from the N terminus (Fig. 4). Total cell lysates were prepared from S. aureus cell expressing double-tagged wild-type group I AgrD (TLDH) or the N-terminal deletion mutants and then analyzed by Western blot hybridization using the anti-T7 monoclonal antibody as a probe. As shown in Fig. 5A, protein bands with approximately the same sizes of the predicted molecular masses of TLDH and the AgrD mutants were seen; however, the amounts of the proteins detected were significantly different. S. aureus cells produced similar amounts of TLDH-dN10 (Fig. 5A, lane 4) compared with TLDH-dN5 (lane 3), but much less than TLDH (lane 1). When 12 or more residues were deleted from the N terminus of AgrD in TLDH, the mutant protein levels were significantly lower than those of TLDH, TLDH-dN5, and TLDH-dN10. These results suggest that the N-terminal region of AgrD plays an important role in the stability of the AgrD protein.
To determine the role of the N-terminal ␣-helical amphipathic motif of AgrD in its subcellular localization, we prepared cytoplasmic and membrane fractions from S. aureus cells expressing TLDH-dN10, TLDH-dN12, or TLDH-dN14. These fractions were analyzed by Western blot hybridization using the anti-T7 monoclonal antibody as a probe. As shown in Fig.  5B (lanes 1, 4, and 7), all three mutant proteins were expressed. TLDH-dN14 was found only in the cytoplasmic fraction (Fig.  5B, lane 8), whereas TLDH-dN12 was detected predominately in the cytoplasmic fraction (lane 5). However, TLDH-dN10 was detected predominately in the membrane fraction (Fig. 5B, lane  3), indicating that the deleted N-terminal region is crucial for membrane anchoring of AgrD.
To establish the relationship between membrane anchoring of AgrD and its processing to generate mature AIP, we cotransformed plasmids encoding TLDH or AgrD N-terminal deletion mutants with plasmid carrying group I wild-type AgrB. The supernatants were then prepared from these strains, and the AIP activities were measured (Fig. 5C). S. aureus cells expressing wild-type AgrB and TLDH-dN10 produced ϳ50% AIP activities compared with cells expressing AgrB and TLDH, which was likely due to either the lower expression level of TLDH- dN10 or unstable membrane anchoring that resulted in fewer proteins anchoring in the membrane compared with TLDH, which was detected only in the membrane fraction (data not shown). Only a trace amount of TLDH-dN12 was found in the membrane (Fig. 5B, lane 6); and when this protein was coexpressed with AgrB, a very low amount of AIP activity was detected. In contrast, no TLDH-dN14 was found in the cell membrane (Fig. 5B, lane 9), and no AIP activities were detected when this protein was coexpressed with AgrB. Similar AIP activity assay results were obtained with the N-terminal deletion mutants of His 6 -tagged AgrD without the N-terminal tag (data not shown). These results indicate that N-terminally deleted AgrD that is not associated with the membrane cannot be processed to generate mature AIP.
The N-terminal ␣-Helical Amphipathic Motif of AgrD Is Not Involved in the Specific Interaction with AgrB-The interaction between AgrB and AgrD is group-specific (13), e.g. S. aureus group I AgrD can be processed only by its cognate AgrB, but not by group II AgrB and vice versa. Since AgrB is a multipass transmembrane protein (16), we attempted to address the question of whether the membrane-anchoring amphipathic motif of AgrD is involved in the group-specific interaction with AgrB. Two chimeric C-terminal His 6 -tagged AgrD proteins (DH-ISII and DH-IISI) (Fig. 4) were constructed by swapping group I and II AgrD proteins at the 19th residue (isoleucine). The two chimeric AgrD proteins were expressed at comparable levels compared with the C-terminal His 6 -tagged group I and II AgrD proteins, as confirmed by Western blot hybridization with the anti-tetrahistidine monoclonal antibody as a probe (data not shown). The AIPs produced by S. aureus cells coexpressing chimeric AgrD with either group I or II AgrB were then measured. As shown in Fig. 6, AgrD with the C-terminal region from one group was processed by its cognate AgrB no matter whether the N-terminal region was from group I or II AgrD. This result strongly suggests that the N-terminal ␣-helical amphipathic motif of AgrD is not involved in the groupspecific interaction with AgrB.

An AgrD Mutant with Its N-terminal Region
Replaced with an Artificial Amphipathic 11-Amino Acid Peptide Is Functional-An artificial 11-amino acid peptide that would form an amphipathic ␣-helix as predicted by the hydrophobic moment analysis (27) was fused in-frame with the C-terminal region of group I AgrD-His 6 (from the 15th residue (isoleucine) to the C terminus). The helical wheel analysis results of the Nterminal region of this AgrD mutant (AmphiNDH) are showed in Fig. 7A. The AmphiNDH protein was expressed in S. aureus (Fig. 7B, lane 1) and was found predominately in the membrane fraction (lane 3) and could not be extracted from the cell membrane by 1 M NaCl (lane 5), confirming that the amphipathic ␣-helix is anchored in the membrane. When the AmphiNDH protein was coexpressed with group I AgrB, cells produced comparable levels of AIP compared with cells expressing wild-type AgrD-His 6 and AgrB (Fig. 7C). These results indicate that the artificial amphipathic ␣-helix serves as a targeting motif that anchors AgrD in the membrane for its processing.
To test whether replacement of the N-terminal amphipathic region of AgrD with a hydrophobic transmembrane ␣-helix enables the AgrD C-terminal region to be processed properly, we made an AgrD mutant (tmDH) in which the N-terminal transmembrane ␣-helix (21 amino acids) of E. coli leader peptidase I (28) followed by two arginine residues was fused with the C-terminal 32 amino acids of AgrD. E. coli signal peptidase I has been widely used in transmembrane topology studies in both prokaryotic and eukaryotic cells (29,30). This membrane protein has two transmembrane ␣-helices, and the first ␣-helix has an N terminus outside and C terminus inside of the membrane topology. We chose the first ␣-helix of this protein and added two arginine residues at its C terminus according to the so-called "positive inside rule" (31) to increase the possibility that the protein fused at the C terminus of the ␣-helix would be inside the cytoplasmic membrane. The helical wheel presentation of the first transmembrane ␣-helix of E. coli signal peptidase I is shown in Fig. 7A. In S. aureus cells, the AgrD tmDH mutant was expressed at comparable levels compared with AgrD-His 6 (data not shown). As expected, the mutant protein was found only in the membrane fractions as shown by Western blot hybridization analysis using the anti-tetrahistidine antibody as a probe (Fig. 7C). However, unlike AmphiNDH, no AIPs were produced by cells expressing both tmDH and group I AgrB as determined by AIP activity assays (Fig. 7D). These  1, 4, and 7) were prepared from S. aureus cells expressing the TLDH-dN mutants, and cytoplasmic (lanes 2, 5, and 8) and membrane (lanes 3, 6, and 9) fractions were separated by ultracentrifugation. Samples were subjected to Tris/Tricine/SDS-PAGE followed by Western blotting using the anti-T7 antibody as a probe. The cytoplasmic and membrane fractions were prepared from whole cell lysates containing 600 g of total proteins. Lanes 1-3, TLDH-dN10; lanes 4 -6, TLDH-dN12; lanes 7-9, TLDH-dN14. C, AIP activities produced by S. aureus cells expressing both AgrB and TLDH or TLDH-dN mutants. results indicate that an amphipathic ␣-helix is crucial as a targeting motif that anchors AgrD in the membrane for its proper processing. However, we could not rule out the possibility that the transmembrane ␣-helix from E. coli signal peptidase I used in our studies is not translocated properly on the membrane in S. aureus.

DISCUSSION
In this study, we found that AgrD (the propeptide for the Agr quorum-sensing signal AIP) in S. aureus is an integral membrane protein, with its N-terminal region integrated into the cytoplasmic membrane. We have also demonstrated that integration of AgrD into the membrane is required for its stability and processing to generate mature AIP.
Membrane proteins integrate into lipid bilayer in several ways. The majority of them are transmembrane proteins with either ␣-helical transmembrane segment(s) or ␤-stranded transmembrane segments. Others are monotopic membrane proteins that are commonly anchored in the membrane by an amphipathic region (32). Thermodynamics studies of amphipathic peptides show that the amphipathic region is usually folded into an ␣-helix upon interaction with the membrane bilayer (33,34). Analysis of the predicted staphylococcal AgrD sequences revealed that amphipathic motifs exist in the Nterminal region of AgrD proteins, although the sequences are highly diversified, and the average hydrophobicity values of this region vary. We propose that AgrD is anchored in the cytoplasmic membrane by an N-terminal amphipathic ␣-helix either on the inner leaflet of the membrane or as a transmembrane segment (Fig. 8, A and B). We found that replacing the N-terminal 14 amino acids of AgrD with an artificial amphipathic 11-amino acid peptide enabled the protein to be targeted correctly to the membrane and processed by AgrB to produce mature AIP. However, replacement of the same AgrD N-terminal region with a well defined E. coli signal peptidase I transmembrane ␣-helix totally blocked AgrD from being processed by AgrB, even though this AgrD mutant could still be targeted to the cytoplasmic membrane. Assuming that the E. coli signal peptidase I transmembrane ␣-helix is translocated properly in S. aureus, these results favor the inner leafletanchoring model of AgrD as shown in Fig. 8A. Furthermore, the N-terminal T7 tag and the nine-amino acid linker region in TLDH-dN14 are hydrophilic, so it could be considered that the N-terminal 14 amino acids of AgrD were replaced with a 28amino acid hydrophilic peptide. The fact that this mutant could not anchor in the membrane and could not be processed to generate mature AIP suggests that the amphipathic motif does play a key role in AgrD protein trafficking.
Our previous study shows that AgrB is a multipass transmembrane protein involved in the proteolytic processing of AgrD (16); therefore, membrane localization may be a prerequisite for AgrD to work together with the AgrB protein. Considering that the two proteins may interact around or inside the cytoplasmic membrane, it is of interest to know whether the membrane-anchoring region of AgrD is involved in the interaction. However, our studies on the chimeric AgrD proteins with sequences exchanged between group I and II AgrD proteins revealed that group-specific processing of AgrD was determined only by its C-terminal region. More important, the AgrD mutant with an artificial N-terminal region containing an amphipathic motif was processed as efficiently as wild-type AgrD, suggesting that the N-terminal amphipathic region of AgrD serves only as a membrane-targeting motif, directing it to the same compartment with AgrB for further processing. Furthermore, significantly lower amounts of AgrD N-terminal deletion mutants that were defective in their abilities to anchor in the membrane were detected compared with the wild-type AgrD sequence, suggesting that anchoring of the AgrD propeptide in the membrane by its N-terminal amphipathic region prevents the nonspecific degradation of the proteins. We hypothesized that the initial interaction between AgrD and AgrB might involve the C-terminal to AIP region of AgrD. The AgrD mutant with four amino acids deleted from its C terminus had no effect on its processing by AgrB to generate mature AIP. However, deletion of six amino acids from the C terminus of FIG. 6. AIP activities produced by S. aureus strains expressing AgrB and chimeric AgrD-His 6 . Culture supernatants were prepared from cells expressing C-terminal His 6 -tagged wild-type or chimeric AgrD proteins with or without AgrB. The AIP activities were measured using RN6390B(pRN6683)(S. aureus group I reporter strain) and SA502A-(pRN6683)(S. aureus group II reporter strain) as described previously (12,13). Sa I, S. aureus group I; Sa II, S. aureus group II. DH-ISII and DH-IISI are chimeric AgrD proteins (see Fig. 4).
AgrD totally eliminated its ability to be processed. 2 The direct interaction of this region of AgrD with AgrB has yet to be confirmed experimentally.
Examples of proteins that anchor in the membrane by an amphipathic ␣-helix include plasma lipoproteins (35), ion channels (32), membrane-anchoring enzymes (36), and proteins involved in signal transduction pathways that relay signals from transmembrane receptors to downstream cytoplasmic proteins 2 J. Lin and G. Ji, unpublished data. NaCl. The membrane and cytoplasmic fractions were then separated by ultracentrifugation. Proteins were separated by Tris/Tricine/SDS-PAGE followed by Western blot analysis with the anti-tetrahistidine monoclonal antibody as a probe. In B, lane 1 contains the total cell lysate (150 g of total proteins); lanes 2 and 4 contain the cytoplasmic fractions; and lanes 3 and 5 contain the membrane fractions. The cytoplasmic and membrane fractions were prepared from total cell lysates containing 150 g of total proteins. In C, lane 1 contains the total cell lysate; lane 2 contains the cytoplasmic fractions; and lane 3 contains the membrane fractions. D, AIP activities produced by S. aureus cells expressing both group I AgrB and group I AgrD-His 6 , AmphiNDH, or tmDH. The AIP activities were measured using RN6390B (pRN6683) (S. aureus group I reporter strain) as described (12).