Lipid Requirements for the Enzymatic Activity of MraY Translocases and in Vitro Reconstitution of the Lipid II Synthesis Pathway*

Screening of new compounds directed against key protein targets must continually keep pace with emerging antibiotic resistances. Although periplasmic enzymes of bacterial cell wall biosynthesis have been among the first drug targets, compounds directed against the membrane-integrated catalysts are hardly available. A promising future target is the integral membrane protein MraY catalyzing the first membrane associated step within the cytoplasmic pathway of bacterial peptidoglycan biosynthesis. However, the expression of most MraY homologues in cellular expression systems is challenging and limits biochemical analysis. We report the efficient production of MraY homologues from various human pathogens by synthetic cell-free expression approaches and their subsequent characterization. MraY homologues originating from Bordetella pertussis, Helicobacter pylori, Chlamydia pneumoniae, Borrelia burgdorferi, and Escherichia coli as well as Bacillus subtilis were co-translationally solubilized using either detergent micelles or preformed nanodiscs assembled with defined membranes. All MraY enzymes originating from Gram-negative bacteria were sensitive to detergents and required nanodiscs containing negatively charged lipids for obtaining a stable and functionally folded conformation. In contrast, the Gram-positive B. subtilis MraY not only tolerates detergent but is also less specific for its lipid environment. The MraY·nanodisc complexes were able to reconstitute a complete in vitro lipid I and lipid II forming pipeline in combination with the cell-free expressed soluble enzymes MurA-F and with the membrane-associated protein MurG. As a proof of principle for future screening platforms, we demonstrate the inhibition of the in vitro lipid II biosynthesis with the specific inhibitors fosfomycin, feglymycin, and tunicamycin.

Screening of new compounds directed against key protein targets must continually keep pace with emerging antibiotic resistances. Although periplasmic enzymes of bacterial cell wall biosynthesis have been among the first drug targets, compounds directed against the membrane-integrated catalysts are hardly available. A promising future target is the integral membrane protein MraY catalyzing the first membrane associated step within the cytoplasmic pathway of bacterial peptidoglycan biosynthesis. However, the expression of most MraY homologues in cellular expression systems is challenging and limits biochemical analysis. We report the efficient production of MraY homologues from various human pathogens by synthetic cellfree expression approaches and their subsequent characterization. MraY homologues originating from Bordetella pertussis, Helicobacter pylori, Chlamydia pneumoniae, Borrelia burgdorferi, and Escherichia coli as well as Bacillus subtilis were cotranslationally solubilized using either detergent micelles or preformed nanodiscs assembled with defined membranes. All MraY enzymes originating from Gram-negative bacteria were sensitive to detergents and required nanodiscs containing negatively charged lipids for obtaining a stable and functionally folded conformation. In contrast, the Gram-positive B. subtilis MraY not only tolerates detergent but is also less specific for its lipid environment. The MraY⅐nanodisc complexes were able to reconstitute a complete in vitro lipid I and lipid II forming pipeline in combination with the cell-free expressed soluble enzymes MurA-F and with the membrane-associated protein MurG. As a proof of principle for future screening platforms, we demonstrate the inhibition of the in vitro lipid II biosynthesis with the specific inhibitors fosfomycin, feglymycin, and tunicamycin.
Proteins involved in bacterial cell wall biosynthesis are highly conserved and often essential. In stage I, the cytoplasmic proteins MurA-F are responsible for the formation of the soluble peptidoglycan precursor (Park's nucleotide). In stage II, the integral membrane protein MraY catalyzes the ligation of Park's nucleotide with the membrane-bound undecaprenylphosphate (C 55 -P) 4 resulting into lipid I. The membrane-associated globular protein MurG catalyzes subsequent glycosyl transfer forming lipid II (Fig. 1). As the final monomeric building block in bacterial cell wall biosynthesis, lipid II will be transported through the membrane into the periplasmic space for further polymerization in stage III of cell wall biosynthesis (1).
The bacterial cell wall biosynthetic pathway is an important and traditional target for antibiotics and numerous drugs directed against key catalysts for clinical applications have been developed. However, most known drugs such as the penicillin derivatives affect the periplasmic or extracellular steps of peptidoglycan synthesis catalyzed by soluble enzymes. In contrast, almost no clinical drugs have been developed that are targeted against the integral membrane proteins involved in formation and translocation of precursors. Several MraY inhibitors such as tunicamycin (2), capuramycin (3,4), mureidomycin (5), liposidomycin (6), or michellamine B (7) have been identified but could not enter clinical applications. Intensive systematic screening programs have been hindered by the difficult nature of these enzymes restricting their detailed structural and functional characterization. Recent screens resulted into a first crystal structure of MraY from the thermophilic Aquifex aeolicus (8), but no structural information of MraY from bacterial pathogens is so far available.
Synthetic biology approaches based on cell-free expression technologies have been developed in the past years for the efficient production of preparative scale amounts of membrane proteins (9). Individual environments can be designed for cotranslational solubilization of membrane proteins. Generally, membrane proteins could either be solubilized in the presence of micelles formed by amphiphilic compounds (D-CF, detergent-based cell-free expression) or in the presence of membranes of defined or complex lipid compositions (L-CF, lipidbased cell-free expression). In particular, the combination of the nanodisc technology with cell-free expression opens promising perspectives for the systematic screening of lipid effects on the functional folding of membrane proteins (10,11). If synthesized without any supplied hydrophobic environment, membrane proteins will precipitate (P-CF, precipitate forming cellfree expression) and often form type I aggregates that may become post-translationally solubilized into functional proteins (12)(13)(14). In previous studies we have analyzed the activity of cell-free-expressed MraY homologues from Escherichia coli and Bacillus subtilis (11,15). The two proteins have striking differences in their requirements for functional folding. B. subtilis MraY (Bs-MraY) was active in a wide variety of conditions including numerous detergent micelles as well as complex liposomes, whereas the E. coli MraY (Ec-MraY) was completely inactive in the presence of all tested detergent micelles. Lipid I forming Ec-MraY could only be obtained after L-CF expression in the presence of preformed nanodiscs composed of anionic lipids with phosphoglycerol headgroups.
In the present work, (i) we analyzed the lipid specificity of the Gram-positive Bs-MraY, (ii) we quantified the requirement for anionic lipids of Ec-MraY, and (iii) we in particular analyzed whether the observed characteristics of Ec-MraY could be transferred to other MraY enzymes from human pathogens. Cell-free expression protocols for the MraY homologues from Helicobacter pylori causing ulcer, Bordetella pertussis causing whooping cough, Chlamydia pneumoniae as a common cause of pneumonia implicated in cardiovascular disease, and Borrelia burgdorferi causing persistent infections after tick bites have been established. We could demonstrate for the first time that the co-translational formation of nanodisc complexes by cellfree expression is an efficient tool to generate functionally folded MraY enzymes from different human pathogens for their subsequent in vitro characterization. We further isolated the MurA-F enzymes necessary for biosynthesis of the lipid I pentapeptide precursor as well as MurG necessary for subse-quent lipid II formation after cell-free expression. We demonstrate that MraY⅐nanodisc complexes interact with the purified cell-free synthesized Mur enzymes for the reconstitution of a functional lipid I and lipid II forming pathways. As proof of principle for subsequent screening approaches, we demonstrate the inhibition of lipid II formation by the reconstituted in vitro pathway with a variety of known inhibitors.
Cell-free Expression of Proteins-The continuous exchange cell-free configuration (CECF) was used based on lysates from E. coli A19 cells and T7-RNA polymerase transcription. Lysate preparation, expression, and purification of T7-polymerase and CECF expression were performed following standard protocols as described elsewhere (9,17). Concentrations for amino acids and the potassium acetate stock solution where adjusted to 25 mM and 10 M, respectively (11). Reactions were performed in 55 l with a ratio of reaction mixture (RM) to feeding mixture (FM) of 1:15 using mini-CECF containers incubated at 30°C under 200 rpm shaking for around 16 h in 24-well cell-culture plates (Greiner, Frickenhausen, Germany) containing the FM (17). FM and RM were separated by a cellulose membrane with a cut-off of 12-14 kDa (Spectrum Laboratories, Rancho Dominguez, CA). Initial expression screens for efficiency optimization were performed in the P-CF mode. For D-CF expression detergents were supplemented to the FM and RM to give a final concentration of 0.4% (w/v) for digitonin, Brij35, and Brij78 and 0.1% (w/v) for DDM. Nanodiscs were added to the RM in concentrations of 80 -100 M for sufficient MraY solubilization (11). After expression RMs were centrifuged (22,000 ϫ g, 10 min) to separate the soluble fraction from the precipitate.
In Vivo Expression and Purification of Mur Enzymes-The expression of MurA to MurF was carried out as described previously (18). MurA to MurF constructs from Staphylococcus aureus with C-terminal His 6 tags were transformed into E. coli BL21(DE3) and cells were grown on LB medium at 30°C and induced at an optical density of 0.6 with 0.5 mM isopropyl 1-thio-␤-D-galactopyranoside. Cells were harvested after 3 h and lysed by sonication in buffer M (50 mM NaH 2 PO 4 , pH 7.8, 300 mM NaCl, 10 mM imidazole) after 30 min incubation with 200 mg/ml of lysozyme, 100 mg/ml of DNase, and 10 mg/ml of RNase. After separation of cell debris the supernatant was applied to Ni-NTA Superflow (Qiagen, Hilden, Germany) and washed with buffer M and buffer M supplemented with 20 mM imidazole. Proteins were eluted with buffer M containing 200 mM imidazole. Proteins were stored in 50% glycerol at Ϫ20°C.
Preparation of Nanodiscs-The membrane scaffold protein derivative MSP1E3D1 containing an N-terminal His 6 tag (19) was expressed in E. coli BL21(DE3) Star cells. 2.4 Liters of LB medium supplemented with 0.5% (w/v) glucose and 30 g/ml of kanamycin sulfate were inoculated with a pre-culture and incubated at 37°C under vigorous shaking. Cells were induced at an optical density of Ϸ1.0 with 1 mM isopropyl 1-thio-␤-D-galactopyranoside and further grown for 1 h at 37°C followed by 4 h at 28°C. After incubation cells were harvested (7,000 ϫ g, 10 min) and ϳ15 g of cell pellet were resuspended in 50 ml of buffer A (40 mM Tris-HCl, pH 8.0, 300 mM NaCl) containing one tablet of Complete protease inhibitor (Roche, Penzberg, Germany) and 1% (v/v) Triton X-100. Cell breakage was performed by sonication with two times three cycles of 1 min and 45 s with 1 min resting time. To remove cell debris, lysate was centrifuged (30,000 ϫ g, 20 min) and the resulting supernatant was filtered (0.45 m). Filtered lysate was loaded onto a Histrap column equilibrated with 5 column volumes of buffer A with 1% (v/v) Triton X-100. After loading the column was washed with 5 column volumes of each buffer A with 1% (v/v) Triton X-100, buffer A at pH 8.9 with the addition of 50 mM cholic acid, buffer A, and buffer A containing 50 mM imidazole. Protein was eluted with 5 column volumes of buffer A supplemented with 300 mM imidazole. MSP1E3D1 containing fractions were pooled and glycerol was added to a final concentration of 10% (v/v). Pooled fractions were dialyzed against two times 5 liters of buffer A with 10% (v/v) glycerol. MSP1E3D1 concentration was determined by A 280 ⑀ ϭ 27,310 M Ϫ1 with the molecular mass of 31.96 kDa using a NanoDrop device (Peqlab, Erlangen, Germany).
For nanodisc reconstitution MSP1E3D1 was mixed with lipid stock solutions at distinct protein to lipid ratios as previously described (11). Protein to lipid ratios for mixtures of DMPC and DMPG were 1:115 and 1:110 and for DPPC, DPPG, and DPPE 1:100, 1:110, and 1:80, respectively. The different ratios have been determined by pilot screening experiments and analyzed by size exclusion chromatography with a Superdex 200 3.2/30 column (GE Healthcare, Muenchen, Germany) with DF buffer (40 mM Tris-HCl, pH 8.0, 100 mM NaCl) as running buffer and a flow rate of 0.05 ml/min. Reconstitution mixtures of protein, lipid, and 0.1% (w/v) DPC were incubated for 1 h at room temperature. Nanodiscs were formed by dialysis of the reconstitution solution against 5 liters of DF buffer for 3 days with two times buffer exchange. To remove aggregates, nanodiscs were centrifuged (22,000 ϫ g, 20 min) and the supernatant was concentrated with Centriprep tubes with 10-kDa MWCO (Merck Millipore, Darmstadt, Germany) by centrifugation (2,000 ϫ g) up to 0.5-1.0 mM. For long time storage nanodiscs were shock frozen with liquid nitrogen and stored at Ϫ80°C (11).
MraY Purification-P-CF generated precipitates were washed twice with RM volume of buffer Y1 (10 mM Tris-HCl, supplemented with LMPG (1 or 0.75%), DDM (1%), DPC (2%), LS (0.8%), or a mixture of LMPG (0.5%) and DDM (1%). Suspensions were incubated for 1 h at room temperature for solubilization. After incubation the suspension was centrifuged again (22,000 ϫ g for 20 min) and the supernatant was used for affinity purification. Soluble fractions of D-CF and L-CF expressions could directly be used for further purification. MraY enzymes were purified utilizing the C-terminal His 10 tag or StrepII tag. For both purification strategies the protein solution was diluted 5 times in buffer P (20 mM sodium phosphate, pH 7.2, 150 mM NaCl, 2 mM ␤-mercaptoethanol) and incubated with equilibrated Ni-NTA Superflow (Qiagen, Hilden, Germany) or StrepII-Tactin resin (IBA, Goettingen, Germany) with a bead volume corresponding to around 1/3 of the RM for 3 h at room temperature on an overhead shaker. Resin with bound protein was washed three times with five times the bed volume of buffer P for StrepII-tagged proteins and His 10 -tagged proteins with buffer P supplemented with 40, 60, and 80 mM imidazole. His 10 -tagged proteins were eluted with a volume of buffer P containing 300 mM imidazole corresponding to the RM and elution from StrepII-Tactin resin was performed with buffer P with the addition of 5 mM desthiobiotin. All buffers for purification of solubilized proteins after P-CF expression or D-CF expressed samples contained DDM (0.2%). Purified samples were shock frozen with liquid nitrogen and stored at Ϫ80°C prior to activity test.
SDS-PAGE and Immunoblotting-Protein samples supplemented with SDS loading buffer (100 mM Tris-HCl, pH 6.8, 8 M urea, 0.12% (w/v) bromphenol blue, 20% (w/v) SDS, 15% (v/v) glycerol, 20% (v/v) ␤-mercaptoethanol) were loaded on 11% Tris Tricine SDS gels. Gels were run for 20 min at 80 V and 45-55 min at 150 V. Afterward they were incubated with a solution of 50% (v/v) ethanol and 10% (v/v) acetic acid and stained with a colloidal staining solution containing 0.02% (w/v) Commassie Brilliant Blue G250, 5% (w/v) aluminum sulfate-(14 -18)-hydrate, 10% (v/v) ethanol, and 2% (v/v) orthophosphoric acid. For Western blot analysis proteins were transferred from gels onto a 0.45 m polyvinylidene difluoride membrane (Merck Millipore, Darmstadt, Germany) by wet blotting for 35 min at 340 mA. Membranes were blocked for 1 h at room temperature on a shaker using 25 ml of PBST (1.8 mM KH 2 PO 4 , 10 mM Na 2 HPO 4 , 137 mM NaCl, 2.6 mM KCl, 0.05% (v/v) Tween 20) with the addition of 4% (w/v) skim milk powder. Afterward for detection of His 10 -tagged proteins the anti-His antibody (Qiagen, Hilden, Germany) was applied as first antibody in a dilution of 1:2,000 in 10 ml of PBST including 0.002% (w/v) skim milk powder for overnight incubation at 4°C on a shaker. The membrane was washed two times with 25 ml of PBST for 15 min at room temperature on a shaker before anti-mouse IgG horseradish peroxidase (HRP) conjugate (Sigma) was applied as second antibody in a dilution of 1:5,000 in 10 ml of PBST including 0.002% (w/v) skim milk powder for 1 h at room temperature on a shaker. For detection of StrepII-tagged proteins an anti-StrepII HRP-conjugate from Bio-Rad (Muenchen, Germany) in a dilution of 1:7,500 in 10 ml of PBST including 0.002% (w/v) skim milk powder was used. Subsequently, the membrane was washed two times with 25 ml of PBST for 15 min and with 25 ml of PBS. Blots were analyzed by chemiluminescence using selfmade ECL detection reagents and a Lumi-imager F1 TM (Roche Diagnostics).
MraY Functional Assay-The MraY concentration needed for calculation of specific activities of MraY in nanodiscs was quantified by immunodetection after Western blotting. As protein standards for concentration determination Precision Plus Protein Unstained standards (Bio-Rad, Muenchen, Germany) for StrepII-tagged MraYs and Roti-Mark 10 -150 (Carl-Roth, Karlsruhe, Germany) for His 10 -tagged MraYs were used. Band intensities of Western blot bands were analyzed by densitometric measurements utilizing ImageJ software (National Institute of Health, Bethesda, MD). The functional analysis of MraY in nanodiscs was performed as described before (15). MraY reactions contained 5 nmol of C 55 -P, 10 mM LS, 100 mM Tris-HCl, pH 8.4, 40 mM MgCl 2 , 50 nmol of UDP-MurNAc-pentapeptide isolated from Staphylococcus simulans (20) and varying amounts of MraY in a total volume of 50 l. Activities of Cp-MraY⅐nanodisc complexes were analyzed in 0.4% Triton X-100 (w/v), 75 mM Tris-HCl, pH 7.5, 6 mM MgCl 2 , 10% dimethyl sulfoxide and with precursor concentrations as above.
In Vitro Synthesis of Lipid II-Lipid II was synthesized as described with minor modifications (21). In brief 100 nmol of UDP-GlcNAc in 50 mM Bistris propane, pH 8.  The MraY homologues were first expressed in the P-CF mode and the resulting precipitates were solubilized in LMPG, DDM, or in a LMPG/DDM mixture. DDM was selected as it was previously found to be efficient in the post-translational solubilization of P-CF-expressed Bs-MraY resulting into functionally folded enzyme (15). Although post-translational solubilization efficiency of the three MraY proteins in LMPG was 100%, it was significantly lower with DDM and estimated at ϳ10%. In an alternative approach the P-CF precipitates of Hp-MraY and Bp-MraY were first quantitatively solubilized in relatively harsh detergents such as LMPG, DPC, or LS and then exchanged against the milder DDM as a second detergent upon immobilization by affinity chromatography purification. Furthermore Hp-MraY, Bp-MraY, Cp-MraY, and Bb-MraY were D-CF expressed and co-translationally solubilized in the presence of digitonin, Brij35, or Brij78. The solubilization efficiency of Hp-MraY in these detergents was quantitative. Solubiliza-tion of the other MraY enzymes was lower and in particular with digitonin less than 50% ( Table 2). All solubilized samples were exchanged into DDM micelles upon affinity chromatography purification.

Cell-free Production and Functional
The enzymatic activity of MraY samples was analyzed for lipid I formation with in vitro assays using the precursors UDP-MurNAc-pentapeptide purified from S. simulans cells and LSsolubilized C 55 -P. The formation of lipid I was monitored by TLC and quantified using radiolabeled precursors if appropriate. For detergent-solubilized P-CF-synthesized samples of Hp-MraY some minor lipid I formation could be detected (Table 2). Although a faint spot of lipid I was verified after TLC separation, the incorporated radiolabel in parallel experiments was below the detection limit allowing for quantification. Stability of the detergent-solubilized enzyme was furthermore, too low for detection of activity in purified samples or in samples transferred to DDM upon purification. For Bp-MraY and Bb-MraY, no lipid I formation could be detected in P-CF-synthesized samples. The D-CF samples of all tested MraY enzymes did not show any lipid I formation.

Co-expression of the Hp-MraY, Bp-MraY, Cp-MraY, and Bb-MraY Homologues in the Presence of Nanodiscs-Ec-MraY
could efficiently be solubilized if co-expressed in the presence of nanodiscs assembled with the MSP1E3D1 derivative and resulting in an estimated nanodisc diameter of ϳ12 nm (11). The functional folding of Ec-MraY in nanodiscs was verified by lipid I formation activity. The four enzymes Hp-MraY, Bp-MraY, Cp-MraY, and Bb-MraY were therefore expressed in the presence of 80 -100 M MSP1E3D1-nanodiscs. The nanodiscs had been pre-assembled with membranes composed out  of either the eukaryotic neutral lipid DMPC or the anionic lipid DMPG present in bacterial membranes. Co-translational solubilization efficiencies of the MraY enzymes with the various nanodisc preparations was between 60 and 90% and was therefore in agreement with previous results obtained with Ec-MraY ( Table 2). The MraY⅐nanodisc complexes were purified from the RMs by affinity chromatography by taking advantage of terminal His 10 tags or StrepII tags. The purified complexes were further analyzed by SDS-PAGE ( Fig. 2A) and the solubilized MraY proteins were identified by immunoblotting with antibodies directed against the purification tag (Fig. 2B). The activity of the purified MraY⅐nanodisc complexes was analyzed for lipid I formation by the TLC assay (Fig. 2C). Despite similar efficiencies in the association of Hp-MraY with nanodiscs assembled with either DMPC or DMPG, the characteristic lipid I formation activity was only detected with the Hp-MraY⅐nanodisc (DMPG) complexes. Lipid I formation was further obtained with Bp-MraY⅐nanodisc (DMPG) as well as with Cp-MraY⅐nanodisc (DMPG) complexes. Accordingly to Hp-MraY, Bp-MraY⅐nanodisc (DMPC) and Cp-MraY⅐ nanodisc (DMPC) complexes remained inactive. Samples of Bb-MraY associated with nanodiscs containing either DMPG or DMPC failed to synthesize detectable lipid I. Lipid I formation was furthermore, quantified by radioassays and the obtained values were comparable with data of Bs-MraY after cellular expression (22). Lipid I synthesis of the Hp-MraY⅐nanodisc (DMPG) and Cp-MraY⅐nanodisc (DMPG) samples was relatively high with 9.7 and 10.8 nmol/g of MraY.
The activity of Bp-MraY⅐nanodisc (DMPG) complexes was lower with ϳ2.3 nmol of lipid I/g of MraY. Hp-MraY and Bp-MraY samples were stable for at least 1 week at 4°C, whereas Cp-MraY complexes were only stable for a few days ( Table 2).

Lipid Selectivity of Bs-MraY-Within the characterized
MraY enzymes, the Bs-MraY homologue appears to be exceptional as it tolerates various detergents for its functional folding. In contrast, all analyzed enzymes derived from Gram-negative bacteria were only active if solubilized in the anionic lipid DMPG or closely related derivatives. The enzymes remained inactive if solubilized in detergents or if reconstituted into membranes composed out of phosphocholines. Due to this observation we analyzed whether the Gram-positive enzyme Bs-MraY, despite its detergent tolerance, shows a lipid preference as well. Bs-MraY was co-expressed in the presence of nanodiscs assembled with lipids of phosphocholine, phosphoglycerol, or phosphoethanolamine families. Chain length and flexibility of the anionic phosphoglycerol derivatives was varied by preparing nanodiscs with membranes composed out of DMPG, DPPG, DOPG, and POPG. In addition, nanodiscs were prepared with membranes composed out of the non-charged lipids DMPC, DPPC, and DOPC as well as DPPE and DOPE. The phosphoethanolamine derivatives were included in the assay as they represent the most prevalent lipid type in bacteria.
In contrast to the homologues from Gram-negative bacteria, Bs-MraY exhibited a strong lipid I forming activity in the range of 6 -10 nmol/g of enzyme in membranes composed out of all analyzed lipid types (Fig. 3A). No clear influence of the lipid headgroup as observed for Ec-MraY was detectable. Exceptions were nanodisc complexes with membranes composed out of saturated long chain lipids such as DPPG containing two palmitoyl chains, resulting in weak and unstable Bs-MraY activity (data not shown). However, dipalmitoyl lipids are generally problematic for the assembly of nanodiscs due to their high transition temperatures of above 40°C. Heterogeneities as well as rigidity of the bilayer in corresponding nanodisc preparations could prevent proper protein insertion or dynamics.  JANUARY 29, 2016 • VOLUME 291 • NUMBER 5

JOURNAL OF BIOLOGICAL CHEMISTRY 2541
Ec-MraY Activity Requires Membranes with Quantitative DMPG Composition-Lipids with phosphoglycerol headgroups such as DMPG act either as a kind of a chemical stabilizer or they could provide an overall suitable membrane environment for MraY folding and activity. Nanodiscs with varying DMPC:DMPG ratios were used as supplements for L-CF expression reactions with Ec-MraY in final concentrations of 100 M. The resulting Ec-MraY⅐nanodisc complexes were purified and analyzed for lipid I formation. Translocase activity of samples in membranes composed out of equimolar ratios of DMPG and DMPC were reduced to ϳ75% if compared with Ec-MraY samples in DMPG membranes (Fig. 3B). However, membranes with DMPG concentrations of 40% and below were obviously not sufficient to maintain Ec-MraY in a folded conformation.
Development of Cell-free Production Protocols for the E. coli MurA-F and MurG Enzymes-The PCR amplified Mur genes were cloned into vectors piVEX 2.4c for MurA to MurF and pET21a for MurG. MurA, -B, -D, -E, and -F were expressed with a C-terminal poly(His) 6 tag for purification and detection. The MurC and MurG proteins were produced with a N-terminal poly(His) 6 tag and a C-terminal StrepII tag or a C-terminal poly(His) 10 tag, respectively (Table 1). All enzymes were synthesized in standard CECF reactions without additives. For highest production efficiency, Mg 2ϩ ion optima were determined for all targets within the 14 to 22 mM range. The solubility of the enzymes in the cell-free reaction mixture was almost quantitative for MurA, MurC, MurD, and MurE. Approximately 50% of the synthesized protein was soluble for MurB and MurG and the lowest solubility of 30% was obtained for MurF. Subsequent experiments were exclusively performed with the soluble fractions of the enzymes. As MurG has been described as a potentially membrane-associated protein containing a significant patch of hydrophobic residues (23) we attempted to improve the solubility of MurG by performing D-CF reactions in the presence of DDM, digitonin, Brij35, or Brij78. Only minor improvements of solubility were obtained with digitonin and with the two Brij derivatives, whereas DDM had no effect. Subsequent expression experiments were thus performed in the P-CF mode.
All expressed proteins contained terminal purification tags and the reaction mixtures were applied to Ni-NTA Superflow resin or StrepII-Tactin resin for affinity purification. Apparent pure protein samples were obtained after one step affinity purification (Fig. 4A). The final yield of the purified proteins was within the range of 1 to 3 mg/1 ml of cell-free reaction for MurA, MurC, MurD, MurE, and MurG. Lower yields between the range of 0.1 and 1 mg/ml of cell-free reaction were obtained with MurB and MurF. Purified proteins could be stored without notable loss of activity at Ϫ80°C for months.
Coupled in Vitro Synthesis of Lipid I and Lipid II by Cell-free Expressed Proteins-The preparation of MraY⅐nanodisc complexes by L-CF expression allows in vitro biosynthesis of the peptidoglycan precursors with samples totally devoid of detergent. Cell-free synthesized and purified MurA-F from E. coli, Bs-MraY⅐nanodisc (DMPG), or Ec-MraY⅐nanodisc (DMPG) complexes and E. coli MurG were mixed in vitro for lipid I and lipid II formation assays. As a control, the MurA-F enzymes of S. aureus expressed in E. coli cells in combination with cell-free expressed MraY⅐nanodisc complexes and MurG were used. Both, lipid I as well as lipid II formation by the pipeline of cellfree synthesized enzymes and starting with UDP-GlcNAc as substrate could be revealed by TLC monitoring (Fig. 4B). The results demonstrate the correct interaction and assembly of the soluble enzymes with the MraY⅐nanodisc complexes from B. subtilis as well as from E. coli and with the membrane-associated MurG protein into a functional biosynthetic pathway.
Inhibitor Monitoring of in Vitro Lipid II Synthesis-To demonstrate the potential of the in vitro reconstituted lipid II synthesis pathway as a drug screening platform, we analyzed the effects of three known inhibitors targeting different steps of the pathway. The soluble Mur enzymes were addressed by the MurA inhibitor fosfomycin (24) and the MurA/MurC inhibitor feglymycin (25). Both inhibitors lead to a drastic decrease in lipid II synthesis below 200 and 10 M, respectively (Fig. 5A). MraY activity was inhibited by addition of tunicamycin as previously reported (15) (Fig. 5B).

Discussion
For several decades, cell wall biosynthetic enzymes have provoked the interest of pharmaceutical research. The production of soluble enzymes such as the cytoplasmic MurA-F and MurG proteins in E. coli cells is very efficient and crystal structures are available (23, 26 -31). Recently, the first three-dimensional MraY structure could be solved from the thermophilic species A. aeolicus after an extensive expression screen comprising many different MraY homologues (8). We demonstrate that cell-free expression provides access to a larger variety of MraY enzymes in reasonable time frames and in efficiencies approaching milligrams of protein/1 ml of reaction. The availability of MraY enzymes for the characterization of inhibitors as well as for their directed engineering could open new pipelines for drug development alternatively to strategies involving the soluble cytoplasmic enzymes of the biosynthetic chain (32) or the cell wall precursors themselves (33)(34)(35)(36).
Conformational folding of membrane proteins as well as the formation and stability of biosynthetic complexes can strictly depend on particular lipid components (37). The demonstrated approach by combining cell-free expression with the nanodisc technology is an ideal platform for the rapid screening of such lipid preferences. The strategy is based on the co-translational association of synthesized membrane proteins with preformed nanodiscs to prevent any contact of the nascent proteins with detergents. In this study we could show that all analyzed MraY enzymes originating from Gram-negative bacteria are highly sensitive against detergent contact. We assume that this characteristic might have played a major role in the reported problems to extract significant amounts of MraY proteins as functionally folded enzymes out of cellular membranes after in vivo expression approaches. The reported process is therefore currently the only method to obtain a variety of these enzymes originated from different bacteria in a functionally active form for in vitro studies in defined environments.
Anionic charge of lipids within the membrane bilayer was clearly necessary for the enzymatic activity of the Gram-negative MraY homologues. Facilitating proper membrane integration and/or promoting subsequent functional folding of the enzymes could be speculated as responsible underlying mechanisms. The requirement of Ec-MraY for membranes predominantly composed out of DMPG gives rather evidence of effects supporting membrane integration. However, the co-translational solubilization was also possible with membranes composed out of non-charged phosphocholines, although the solubilization event might be different from a complete and proper membrane integration. The lack of any Ec-MraY activity with DMPC membranes indicates that the final conformation of MraY in DMPC membranes is different from that in DMPG membranes. Specific interactions of the enzymes with the anionic lipid and/or lipid headgroups as observed for other membrane proteins can furthermore, not be ruled out (38,39). The different behavior of Bs-MraY and the Gram-negative homologues is unexpected. Alignments show an overall homology  between the primary structures of the analyzed MraY proteins of 40%. The homology of Ec-MraY to Bs-MraY as well as to Hp-MraY is similar with ϳ45%. Different membrane integration mechanisms of MraY in between Gram-negative and Gram-positive bacteria could be speculated. Alternatively, differential requirements for cysteine oxidation could be a specific folding determinant of MraY homologues. For MurA, active site cysteines are essential (26,40), whereas the role of cysteines in the enzymatic activity of MraY is still unclear. Membrane localization of MraY could allow the disulfide bridge formation in periplasmic loops but Bs-MraY does not contain any cysteine residues and only one cysteine is present in Bb-MraY. Different numbers of cysteines are present in the other MraY homologues, but their non-conserved localization especially in putative transmembrane regions makes an involvement in structural disulfide bridges or in the catalytic mechanisms rather unlikely.
A further hint toward the observed variation in folding or stability of MraY homologues comes from the recently published structure of A. aeolicus MraY (8). The enzyme forms a dimer having a hydrophobic tunnel between the dimer interface. Electron density mapping indicated the positioning of putative lipid molecules within that tunnel. Lipids might thus play a role in the dimerization of MraY or in maintaining dimeric structures. MraY dimerization might be necessary for proper enzymatic activity or for the stability of the protein, although this speculation still awaits further experimental approval.
Despite high expression levels, Bb-MraY remained inactive in numerous attempts of optimizing the expression environment. The Lyme disease pathogen B. burgdorferi is an intracellular parasite of various eukaryotes. It could be speculated that particular eukaryotic lipids might be necessary for Bb-MraY folding and screening with nanodiscs containing corresponding lipid compositions could be a future perspective. Furthermore, a precursor selectivity of MraY proteins might be considered. The structure of the UDP-MurNAc-pentapeptide precursor for lipid I formation by MraY has some variation within the peptide moiety. The amino acid lysine is included in many Gram-positive bacteria, whereas others use diaminopimelic acid or ornithine (41). There are further variations in the content of D-amino acids. However, it is not clear which determinants of the precursor structure are mandatory for MraY recognition, as at least some enzymes appear to be rather nonselective in terms of the sugar-modified pentapeptide (16) as well as the lipid carrier molecule (42).
In contrast to Bb-MraY, the MraY enzyme of the intracellular human parasite C. pneumoniae was functional when co-translationally integrated into nanodiscs assembled with the lipid DMPG. The presence of functional MraY enzymes in the reduced proteome of the cell wall-less Chlamydia cells was for a long time a matter of debate and only recently the presence and function of the enzyme was demonstrated (16). It is speculated that the function of MraY in C. pneumoniae might be necessary for cell division, rather than for biosynthesis of peptidoglycan in its classical function as a osmotic stabilizing cell wall envelope. Recent studies provided evidence for a minimal ring-shaped peptidoglycan structure in Chlamydia that might localize at the cell division site (43,44). Here we present the preparative scale in vitro production of this enzyme, which opens new perspectives for its characterization in defined environments.
The rapid access to MraY homologues by cell-free expression as well as the streamlined in vitro production of precursors (42) could become interesting for the design of new throughput drug screening platforms. We achieved the reconstitution of the basic lipid II biosynthesis pathway after production of all essential proteins by cell-free expression. The modulation and detailed analysis of this pathway with enzymes originating from different pathogenic bacteria appears to become feasible (42). We could further extend previous studies with the in vitro expressed Mur enzymes that were only able to reconstitute the reaction catalyzing the formation of the UDP-MurNAc-pentapeptide precursor but lacking the synthesis of the final lipid I and lipid II products by the integral membrane protein MraY and the membrane-associated protein MurG (45). Moreover in comparison to previous approaches (42) we present a fast twostep method for lipid II synthesis, which in addition does not need extensive enzyme preparations and facilitates the engineering of the biosynthetic proteins. A complete cell wall precursor biosynthesis platform could help to develop throughput assays for drug identification and the analysis of individual MraY homologues from pathogenic organisms could detect species-specific characteristics. In addition, by using MraY⅐ nanodisc complexes, such assays could be performed in a detergent-free environment and in membranes of defined composition.