Subunit Arrangement in GpsB,a Regulator of Cell Wall Biosynthesis

Bacterial strains and growth conditions

All bacterial strains used in this study are listed in Table 1. L. monocytogenes strains were grown in brain–heart infusion (BHI) broth or on BHI agar plates at 37°C. When necessary, antibiotics and supplements were added to the growth media at concentrations of 50 μg/ml for kanamycin and 1 mM for IPTG. The Escherichia coli strain TOP10 was used for all cloning procedures. ²⁴

Construction of plasmids and strains

The T88A, T88D, F91A, L94A, and F105A amino acid exchanges were introduced into L. monocytogenes gpsB, cloned previously into pSH254, by Quikchange mutagenesis ²⁵ using the primer pairs JR290/JR291, JR291/JR293, JR311/JR312, JR313/JR314, and JR315/JR316, respectively (Table 2). The presence of the desired mutations was verified by DNA sequencing. The resulting plasmids are listed in Table 1 and were introduced into L. monocytogenes EGD-e by electroporation. ²⁶ Kanamycin-resistant clones were selected, and plasmid insertion at the attB site of the tRNA^Arg locus was verified by PCR.

To generate the plasmid for expressing C-LmGpsB, the open reading frame encoding L. monocytogenes GpsB residues 87–113 was PCR amplified from genomic DNA from strain EGD-e with primers RC87f and RC113f, which contain flanking NcoI and XhoI restriction sites, and cloned between the NcoI and XhoI sites of pMAT11, a modified form of pHAT4. ²⁷ The plasmids for expressing the full-length LmGpsB_F91A, LmGpsB_F105A mutants and the double mutant LmGpsB_F91AF105A were prepared by Quikchange mutagenesis using primers RC91f, RC91r, RC105f, and RC105r; for the double mutant, two rounds of mutagenesis were used—the first round with primer pairs RC91f and RC91r and the second round with primer pairs RC105f and RC105r (Table 2).

Similarly, the full-length BsGpsB_T75E and BsGpsB_T75D mutants were prepared by Quikchange mutagenesis with primers RC75Ef and RC75Er, and RC75Df and RC75Dr, respectively; the plasmid for expressing full-length wild-type BsGpsB was used as a template for the Quikchange reaction to generate the T75D mutation. By contrast, for the T75E Quikchange reaction, the plasmid for expressing BsGpsB_T75D was used as a template. The resultant plasmids were sequenced to verify the successful introduction of the desired mutations and were subsequently used for the overproduction of the full-length GpsB variants by the same procedures as wild-type GpsB proteins. ¹¹

Isolation of cellular proteins and Western blotting

Cells were harvested by brief centrifugation in a table-top microfuge and washed with ZAP buffer (10 mM Tris-HCl pH 7.5, 200 mM NaCl). Cells were resuspended in 1 ml ZAP buffer containing 1 mM PMSF and disrupted by sonication. The cell debris was removed by centrifugation, and the supernatant was used as the soluble protein extract. Protein samples were separated by SDS polyacrylamide gel electrophoresis and then transferred onto positively charged polyvinylidene fluoride (PVDF) membranes using a semidry transfer unit. GpsB and DivIVA were detected using polyclonal rabbit antisera raised against L. monocytogenes GpsB ¹¹ and B. subtilis DivIVA, ²⁸ respectively, with anti-rabbit immunoglobulin G conjugated to horseradish peroxidase as the secondary antibody. The ECL chemiluminescence detection system (Thermo Scientific) was then used for detection of the peroxidase conjugates on the PVDF membranes in a chemiluminescence imager (Vilber Lourmat).

Protein purification

PBP1 and GpsB proteins were purified as described previously. ¹¹ C-LmGpsB proteins were expressed as fusion proteins with N-terminal His₆- and MBP-tags; the protocols for expression and purification of the fusion protein by Ni-NTA chromatography were as described previously for C-BsGpsB. ¹¹ Following Ni-NTA chromatography, the fusion protein was cleaved with TEV protease (ratio of fusion protein:TEV 50:1 by mass) overnight at 4°C while simultaneously dialyzing into a buffer of 50 mM Tris-HCl pH 8.0, 300 mM NaCl using a 2 kDa molecular weight cutoff dialysis membrane. The dialysate was passed back over a 5 ml Ni-NTA cartridge (Qiagen) to separate TEV protease, C-LmGpsB released by cleavage from the fusion protein and residual uncleaved His₆-MBP-C-LmGpsB. The flow through from the Ni-NTA column was concentrated in a centrifugal concentrator to 1 ml and loaded onto a Superdex75 XK16/60 column (GE Healthcare) equilibrated in a buffer of 10 mM Tris-HCl pH 8.0, 250 mM NaCl. Elution fractions containing C-LmGpsB were pooled, concentrated to c. 5 mg/ml in a centrifugal concentrator, and then dialyzed further into 20 mM sodium phosphate pH 7.8, 250 mM NaCl before finally flash freezing the protein in liquid nitrogen and storing at −80°C.

Size exclusion chromatography

The full-length wild-type LmGpsB and LmGpsB_F91A, LmGpsB_F105A and LmGpsB_F91AF105A proteins were analyzed on a Superdex200 Increase 10/300 GL column (GE Healthcare) equilibrated in 10 mM Tris-HCl pH 8.0, 250 mM NaCl, at a flow rate of 0.5 ml/min. Protein samples at 3 mg/ml concentration were injected onto the column through a 100 μl sample loop. Wild-type BsGpsB and the BsGpsB_T75D and BsGpsB_T75E mutants were analyzed under the same conditions, except the protein concentration was 1 mg/ml.

Size exclusion chromatography-MALS

GpsB samples (500 μl) at concentrations of 8 mg/ml or 0.5 mg/ml were loaded onto a Superdex200 Increase 10/300 GL column (GE Healthcare) equipped with a Jasco UV-2077 detector, Wyatt DAWN Heleos II EOS 18-angle laser photometer (with the 13th detector replaced with the QELS in-line dynamic light scattering detector) coupled to a Wyatt Optilab rEX refractive index detector. The flow rate was 0.75 ml/min. Molecular mass and concentrations of the peaks eluting from the column in a running buffer of 10 mM Tris-HCl pH 8.0, 250 mM NaCl were analyzed using Astra 6.2 (www.wyatt.com/products/software/astra.html).

Circular dichroism

CD thermal melts were measured in a 1-mm path length quartz cuvette in a 20 mM sodium phosphate pH 7.8, 200 mM NaCl buffer for wild-type LmGpsB, N-LmGpsB, and C-LmGpsB and in 20 mM sodium phosphate pH 7.8, 250 mM NaCl for wild-type BsGpsB, BsGpsB_T75D, and BsGpsB_T75E. The temperature was increased at a rate of 1°C/min, and ellipticity was monitored at a wavelength of 222 nm with a response time of 8 seconds and bandwidth of 2 nm. The reversibility of the melts was verified by recording spectra in the wavelength range 200–240 nm (scan speed 20 nm/min, response 2 seconds) before and after the melt, to check that the spectra were superimposable. Protein concentrations were 23 μM for wild-type LmGpsB, N-LmGpsB, and C-LmGpsB, 29 μM for LmGpsB_F91AF105, and 5 μM for wild-type BsGpsB, BsGpsB_T75D, and BsGpsB_T75E.

The CD spectra of wild-type and C-LmGpsB_F105A proteins were recorded at protein concentrations of 23 and 33 μM, respectively, in a buffer of 20 mM sodium phosphate pH 7.8 at 20°C. The scan speed was 10 nm/min, and the response time was 4 seconds. Protein concentrations were determined based on calculated extinction coefficients at 280 nm ²⁹ for LmGpsB, N-LmGpsB, and BsGpsB proteins; for the C-LmGpsB proteins, which lack tryptophan and tyrosine residues, amino acid analysis was used instead to quantify concentration. Secondary structure composition was evaluated from the spectra using the program CDSSTR ³⁰ within the DICHROWEB ³¹ server.

SAXS data collection and processing

The LmGpsB and N-LmGpsB proteins were analyzed in a buffer of 10 mM Tris-HCl pH 8.0, 150 mM NaCl at a protein concentration of 40 mg/ml. SAXS intensity data, I(q) versus q, ( \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6}\begin{document} $$q = 4 \pi sin 2 \theta / \lambda$$ \end{document} , where 2θ is the scattering angle) were collected using batch-mode SAXS on beamline P12 at PETRAIII (DESY; Hamburg), or batch mode and inline size exclusion chromatography (SEC)-SAXS on beamline B21, Diamond Light Source. Batch-mode SAXS was completed using 30 μl of LmGpsB or N-LmGpsB in a dilution series, which was loaded onto the beamline using the EMBL automated sample changer. ³² SAXS data were collected at PETRAIII with 0.5 second exposures onto a 2 M Pilatus detector (Dectris) at a distance of 3.0 m and an X-ray wavelength of 1.24 Å; these data were reduced and subtracted using in-house software. ³³

For SEC-SAXS data collection at Diamond, 50 μl of LmGpsB was loaded onto a Superdex 200 Increase 3.2/300 column and the eluent flowed through the SAXS beam at a flow rate of 0.1 ml/min; the buffer used as the blank in these measurements was the eluent after one SEC column volume. SAXS data were constantly collected at 1 second intervals using a 2 M Pilatus detector (Dectris) at a distance of 3.9 m and an X-ray wavelength of 1 Å. Subtraction of the SEC-SAXS data was completed for each frame across the elution peak, and the radius of gyration (R_g) and the integral of ratio to background were plotted. The data were scaled, merged, and averaged for each frame with a consistently similar R_g. All further processing and analysis of data were carried out using ScÅtter (www.bioisis.net/scatter).

Pair distance distribution functions were calculated with GNOM, ³⁴ with D_max values for LmGpsB and N-LmGpsB of 185 and 76 Å; the D_max values were estimated initially using DATGNOM and the atomic coordinates of N-LmGpsB, ¹¹ from which the pair distance distribution function was calculated with ScÅtter; all waters and heteroatoms were removed before the calculation.

SAXS model generation

Dummy atom models were constructed in DAMMIN in slow mode using scattering data up to a maximum q value of 8/R_g (0.15Å⁻¹ for LmGpsB, 0.36Å⁻¹ for N-LmGpsB). For N-LmGpsB, the chi-squared values calculated by DAMMIN, corresponding to the agreement between experimental scattering curves and scattering curves calculated from the dummy atom models, ranged between 1.60 and 1.64. For LmGpsB, the chi-squared values from individual DAMMIN run were in the range 1.11–1.14. For N- LmGpsB, the average normalized structural discrepancy (NSD) between models from 10 independent DAMMIN runs, after averaging them together in the DAMAVER suite, ³⁵ was 0.539 ± 0.09. For LmGpsB, the models from 15 DAMMIN runs were superimposed and averaged in the DAMAVER suite and the averaged model was then input into DAMSTART to generate a starting model for a further 15 DAMMIN runs. The final models were then again averaged together using the DAMAVER suite; the NSD between models was 0.552 ± 0.09.

Dummy atom models were visualized in CHIMERA ³⁶ by representing the atoms as beads, with the bead scale adjusted to a value equivalent to the dummy atom radius divided by 0.7. The surface mesh representation of dummy atom models was generated by using the CHIMERA “molmap” command to convert the model to the equivalent electron density map at 25 Å resolution—the contouring of this map was then adjusted to enclose the surface of the dummy atoms. The crystal structures of the N- and C-terminal domains of GpsB ¹¹ PDBids 4ug1 and 5an5, respectively, were also represented as 25 Å resolution electron density maps using the “molmap” command in CHIMERA. For the LmGpsB sample, the molecular weight of the scattering particles was calculated from the volume of correlation ³⁷ using ScÅtter.

Production of lipid II-meso-diaminopimelic acid

Lipid II-meso-diaminopimelic acid (Lipid II-m-DAP) was synthesized using UDP-MurNAc-pentapeptide isolated from Bacillus cereus, essentially as described previously^38,39 with the following modifications. Purification was performed over a DEAE-cellulose column using a linear gradient of chloroform/methanol/water (2:3:1 v/v/v) to chloroform/methanol/1 M ammonium bicarbonate (2:3:1 v/v/v). The fractions containing Lipid II-m-DAP were collected and dried under vacuum. The resulting lipid II-m-DAP was then dissolved in 1:1 chloroform:methanol and stored at −20°C until use.

Production of [ ¹⁴ C]-amidated Lipid II-m-DAP

The ability of MurG to exchange the GlcNAc group between UDP-GlcNAc and Lipid II was used to radioactively label Lipid II. Since the E. coli MurG used here displayed much lower affinity for amidated Lipid II-m-DAP than the nonamidated form, we first synthesized [ ¹⁴ C]-Lipid II-m-DAP and then amidated this using AsnB from B. subtilis. First, 1.22 μmol of purified lipid II-m-DAP was incubated with 12.5 μCi of UDP-N-acetyl-D-[1- ¹⁴ C]glucosamine (specific activity 55 mCi/mmol) (Hartmann Analytic GmbH). The labeling reaction was performed in 1 ml of 100 mM Tris-HCl, pH 8.0, 1 mM MgCl₂, and 2% (w/v) Triton X-100. The reaction was started by the addition of 1 μL purified recombinant E. coli MurG, which is able to exchange the N-acetyl-D-glucosamine of lipid II with that of UDP-N-acetyl-D-[1- ¹⁴ C]glucosamine. After incubation at room temperature for 2.5 hours, the reaction was complete as determined by liquid scintillation counting.

To amidate the carboxylic acid group of the m-DAP, lipid II-DAP was incubated with B. subtilis AsnB in the presence of ATP and glutamine (to be published elsewhere), followed by extraction using butanol/pyridine acetate, pH 4.2, and another purification step over a DEAE cellulose column using a gradient of chloroform/methanol/water (2:3:1 v/v/v) to chloroform/methanol/0.5 M ammonium bicarbonate (2:3:1 v/v/v).

In vitro peptidoglycan synthesis assays

The continuous fluorescence glycosyltransferase (GTase) assay using dansyl-labeled lipid II as the substrate was performed essentially as described previously, ⁴⁰ with only minor modifications to buffer conditions and temperature. We also used an endpoint assay to measure the GTase and transpeptidase (TPase) activity under the same experimental conditions. The activity of BsPBP1 (0.4 μM) was measured at 37°C in the absence and presence of 10 μM BsGpsB proteins in 50 mM HEPES.NaOH pH 7.5, 20 mM NaCl, 10 mM CaCl₂, and 5% (v/v) glycerol. The endpoint assay used [ ¹⁴ C]-labeled native amidated lipid II-m-DAP substrate followed by the quantification of both GTase and TPase products by high-pressure liquid chromatography (HPLC) as described previously. ⁴¹

Low resolution model of subunit arrangement in solution by SAXS

In the absence of diffracting crystals of full-length LmGpsB, SAXS was used to determine its low resolution structure since the correct assembly of subunits in LmGpsB is highly pertinent to the protein's function. To validate this approach, we first analyzed N-LmGpsB for which the crystal structure has already been solved. ¹¹ The measured scattering profile for N-LmGpsB matched favorably with the theoretical profile calculated from the corresponding atomic coordinates (Fig. 1A). From a Guinier analysis of the scattering profile, an R_g of 21.4 Å was obtained, while the pair distribution function calculated by inverse Fourier transformation with GNOM ³⁴ gave an R_g of 22.2 Å. These values compare extremely favorably with an R_g value of 21.4 Å that was calculated from the atomic coordinates with CRYSOL. ⁴² Finally, ab initio dummy atom molecular models built from the scattering data with DAMMIN ⁴³ reproduced well the shape and dimensions of the dimeric N-LmGpsB (Fig. 1B).

OPEN IN VIEWER

FIG. 1.

SAXS analysis of N-LmGpsB. (A) SAXS scattering curves (top), Guinier plots (middle), and pair distribution functions (bottom) for N-LmGpsB. The experimentally observed scattering curve is shown as a black solid line and the scattering curve, calculated from the coordinates with CRYSOL (42), as a red dashed line. The chi-squared value for the agreement between the two curves in the q range 0–0.25 Å⁻¹ is 1.8. For the pair distribution function of the N-LmGpsB, the red dashed plot represents the profile calculated from the coordinates with the ScÅtter software. (B) SAXS-derived ab-initio dummy atom models of N-LmGpsB calculated with DAMMIN from small-angle X-ray scattering profiles. Representative models are shown from three independent DAMMIN calculations. The dummy atom model is shown in surface representation rendered in mesh. The coordinates of the N-LmGpsB dimer (red) have been manually docked inside the mesh, represented either in ribbon form or as the equivalent electron density maps would appear at 25 Å. SAXS, small-angle X-ray scattering.

Analysis of LmGpsB yielded the best quality data on SAXS beamlines with an in-line SEC facility. The monodispersity of LmGpsB samples after in-line SEC was indicated by the linearity of Guinier plots at q values less than 1.3/R_g (Fig. 2A). The molecular mass of the scattering particles after in-line SEC, calculated from the volume of correlation, ³⁷ was 75,900 Da, which is within 5% of the actual mass of the expressed LmGpsB protein when assembled as a hexamer.

OPEN IN VIEWER

FIG. 2.

SAXS analysis of LmGpsB. (A) SAXS scattering curves (top), Guinier plots (middle), and pair distribution functions (bottom) for LmGpsB. (B) Analysis of the flexibility of the N-LmGpsB (top) and the LmGpsB protein (bottom) by Porod–Debye plots. The plateau in the plot for N-LmGpsB is consistent with a compact folded protein.

A Porod–Debye analysis of the comparative flexibility of LmGpsB and N-LmGpsB suggested that the full-length protein is more flexible than the isolated N-terminal domain. The Porod–Debye plot for N-LmGpsB rose to a plateau (Fig. 2B), which is characteristic of a folded compact protein. ⁴⁴ By contrast, the Porod–Debye plot for LmGpsB had a less pronounced plateau and instead conformed to a hyperbole. The relatively shallow gradient of the asymptote, however, indicated that LmGpsB is markedly less flexible than a completely unfolded protein. Given the successful crystal structure determinations of both N- and C-terminal domains of GpsB, the flexibility within LmGpsB is most likely attributable to the linker region between the domains.

The pair distribution function of LmGpsB had, like N-LmGpsB, the characteristics of an elongated molecule, ⁴⁵ with a steep initial increase in P(r) to a maxima followed by a more gradual decrease to zero at D_max. From the Guinier plot and pair distribution functions, R_g values of 51.2 and 53.1 Å were obtained for LmGpsB. Ab initio dummy atom molecular models were calculated in DAMMIN without symmetry constraints, however, no obvious symmetric relationships were apparent from the resultant models. A notable feature of the models is two distinct lobes of different sizes connected by a narrower central region (Fig. 3).

OPEN IN VIEWER

FIG. 3.

SAXS molecular envelopes for LmGpsB. (A) The average surface of 15 independently generated ab initio LmGpsB dummy atom models is represented as a mesh. Three of these models are represented in (B) to illustrate the level of variation. The coordinates of N-LmGpsB dimers (red and cyan) and the C-LmGpsB trimers (blue) were docked manually and are shown in ribbon form (right column) or as equivalent electron density maps would appear at 25 Å resolution (left, middle columns). The coordinates shown correspond to residues 5–67 (N-terminal domain) and 89–106 (C-terminal domain) of LmGpsB; LmGpsB has a total of 113 residues. The models do not include the linker between domains. In the arrangements shown in (B), with the simplest of the possible connectivities between domains, the chain termini are separated by distances of up to 55 Å. By comparison, a fully extended 21 polypeptide residue chain would be ∼80 Å in length, and therefore, the models are compatible with the domains being connected by an appropriate length linker in a semicompact conformation.

The most logical arrangement of the individual domains in the GpsB hexamer resembles a tripod with the three dimeric N-GpsB domains aligned as in the legs in a tripod with the two C-GpsB trimeric domains encompassing its base (Fig. 4A). Such an arrangement is logical on the basis that it presents a relatively straight path and near equivalent distance between connected C- and N-terminal domains of GpsB. Manually positioning the atomic coordinates of the N- and C-terminal domains of GpsB into the ab initio dummy atom model supports such a tripod arrangement (Fig. 3); the three N-GpsB dimers encompass the larger of the two lobes, and the two C-GpsB trimers correspond to the smaller one.

OPEN IN VIEWER

FIG. 4.

Interaction between GpsB domains. (A) A priori hypothetical arrangement of the N- and C-terminal domains of GpsB, as described in the text. N-LmGpsB dimers are colored cyan and red; C-LmGpsB trimers blue. The putative path taken by interdomain linkers is represented as a black dotted line. Note the N-LmGpsB dimer colored cyan is covalently linked to two separate C-LmGpsB trimers and thus forms a bridge between two trimers that would stabilize association into a hexamer. (B) The interface between adjacent (red and green) C-BsGpsB molecules as observed in its crystal lattice, with the key hydrophobic interfacial residues F78, L81, and F92 drawn as sticks and colored black. The lower panel also shows V91, a residue in the hydrophobic core of the C-BsGpsB trimer, which packs against the interfacial residue F92. In the lower panel, the numbering of the Listeria monocytogenes GpsB equivalents of BsGpsB F78, L81, V91, and F92 is shown in italics in parentheses.

The importance of hexamer formation for GpsB activity in vivo

The tripod model positions the two C-LmGpsB trimers in close proximity. The lattice interactions in the crystal structure of C-BsGpsB (encompassing residues 76–98 of BsGpsB) are thus of interest in considering how the subunits assemble to form the GpsB hexamer. Two highly conserved phenylalanines (F78, F92) stand out for being highly solvent exposed on the surface of the C-BsGpsB trimer; within the crystal lattice both F78 and F92 are buried at the interface between adjacent trimers (Fig. 4B). To probe whether burial of these residues drives the assembly of subunits within the full-length hexameric GpsB, the equivalent residues in LmGpsB, F91, and F105 were mutated to alanine, both singly and in combination. The oligomeric state of the wild-type and the mutant proteins was analyzed by SEC. Mutation of F105 to alanine both alone and in combination with the F91A mutation substantially increased the retention volume of LmGpsB mutants on SEC analysis relative to the wild-type LmGpsB (Fig. 5A). Subsequent SEC-MALS analysis of the LmGpsB_F91AF105A double mutant confirmed this is due to a change in the oligomeric state; the double mutant formed a trimer at high protein concentrations and a dimer at lower concentrations (Fig. 5B). Mutation of F91A alone had an intermediate effect on assembly, with SEC analysis revealing two species; the predominant peak has a similar retention volume as the hexameric wild-type protein, while the other peak has a retention volume more similar to the F91AF105A double mutant (Fig. 5B).

OPEN IN VIEWER

FIG. 5.

SEC and SEC-MALS analysis of GpsB proteins. (A) SEC analysis of wild-type LmGpsB (solid black line), LmGpsB_F91A (solid green line), LmGpsB_F105A (solid red line), and LmGpsB_F91AF105A (dashed black line) proteins at 3 mg/ml concentration. (B) SEC-MALS analysis of wild-type LmGpsB (solid black line) and LmGpsB_F91AF105A (dashed black line) proteins. The chromatograms represent analysis at two injected protein concentrations of 8 mg/ml (top) and 0.5 mg/ml (bottom). The deconvoluted molecular masses of the eluting species (red dashed lines) are plotted on the right hand axis. The average of 76 kDa and 38.9 kDa across the major peak for wild-type LmGpsB and LmGpsB_F91AF105A at 8 mg/ml is consistent with the theoretical mass of a hexamer (79 kDa) and trimer (38.6 kDa). At lower protein concentrations, the average mass of LmGpsB_F91AF105A is 26 kDa, consistent with the theoretical mass of a dimer (25.8 kDa), whereas the wild-type LmGpsB remains hexameric. (C) SEC of wild-type BsGpsB (solid black line), BsGpsB_T75E (dashed blue line), and BsGpsB_T75D (solid red line) mutants, all at 1 mg/ml. These phosphomimetic mutations do not introduce any significant change to the oligomeric state of BsGpsB. SEC, size exclusion chromatography.

The structural importance of F91 and F105 is further supported by thermal stability analysis using circular dichroism spectroscopy. Wild-type LmGpsB has a T_m (the temperature of the midpoint of thermal unfolding) of 50°C, whereas the LmGpsB_F91AF105A double mutant has a T_m of 45°C (Fig. 6A). At the protein concentrations used for the CD analysis, the LmGpsB_F91AF105A double mutant was observed to be a dimer on SEC-MALS analysis (Fig. 6B), which may explain why LmGpsB_F91AF105A has a similar T_m as the dimeric, isolated N-LmGpsB domain (Fig. 6A).

OPEN IN VIEWER

FIG. 6.

Circular dichroism spectra. (A) Thermal denaturation of LmGpsB (black line), N-LmGpsB (orange line), C-LmGpsB (cyan line), and LmGpsB_F91AF105A (red line), monitored by circular dichroism; the dashed line represents the summation of the individual thermal melts measured for N-LmGpsB and C-LmGpsB. Unfolding of the secondary structure is observed by monitoring ellipticity at 222 nm. The F91AF105A mutation reduces the stability of LmGpsB. (B) Circular Dichroism spectra of wild-type C-LmGpsB (top) and C-LmGpsB_F105A (bottom panel). The dashed spectrum represents the reconstructed spectra after fitting the secondary structural content with the program CDSStr (30). (C) Thermal denaturation of wt BsGpsB (black line), BsGpsB_T75D (red line), and BsGpsB_T75E (cyan line) monitored by circular dichroism ellipticity at 222 nm. The midpoint of the unfolding transition is similar for the three proteins (c. 62°C).

To further investigate the structural role of F105, the alanine substitution was made within the context of the isolated C-terminal domain. The isolated C-terminal domain forms trimers, rather than hexamers, in solution, ¹¹ and F105 is exposed on the surface of this trimer, rather than in the hydrophobic core. F105 is unlikely to have an important structural role within the isolated C-terminal domain, but will stabilize the hexameric full-length protein. However, the C-LmGpsB_F105A fold was markedly destabilized as the alpha helical content was reduced from 70% in the wild-type C-LmGpsB protein to 10% in the F105A mutant. This effect can perhaps be reconciled on closer analysis of the structure of C-BsGpsB; F92 (equivalent to F105 in L. monocytogenes GpsB) is in close proximity to the hydrophobic core residue V91, and while one face of the F92 aromatic ring is surface exposed, the other face packs against V91 (Fig. 4B). The structural importance of F105 may therefore have two components; first, F105 stabilizes indirectly the folding of the isolated C-terminal domain because of its interaction with Val104, and second, F105 is likely to stabilize further interactions between C-GpsB trimers within the GpsB hexamer, by packing against Phe91 and Leu94 in another protomer in the hexamer (Fig. 4B). The dimers and trimers observed with the LmGpsB_F91AF105A double mutant likely arose from misfolding of at least the C-terminal domain, given the loss of secondary structure associated with the F105A mutation (Fig. 6A).

To validate the effect of mutations in the interface between C-GpsB trimers, we used a complementation assay to measure the effect of gpsB mutations on GpsB activity in L. monocytogenes. Briefly, the L. monocytogenes ΔgpsB mutant grew equally well as the wild-type strain at 30°C and was still viable at 37°C; however, the L. monocytogenes ΔgpsB mutant was not viable at 42°C, neither were gpsB mutants affected in self-interactions or interactions with the L. monocytogenes orthologue of B. subtilis PBP1, PBP A1 (for simplicity referred to herein as PBP1). ¹¹ Mutations that individually exchanged the pair of phenylalanines (F91, F105, the equivalent B. subtilis residues are highlighted on the structure in Figure 4) and the adjacent L94 into alanines were introduced to L. monocytogenes gpsB, and the resulting alleles were tested in the complementation assay. As shown in Fig. 7A, the wild-type L. monocytogenes strain EGD-e proliferated readily at 42°C, whereas the ΔgpsB mutant (strain LMJR19) was unable to grow at all at this temperature. Reintroduction of wild-type gpsB into the ΔgpsB mutant (strain LMS56) repaired this defect, demonstrating successful complementation. The point mutations introduced fell into two classes: strain LMS185, expressing gpsB_F91A was able to grow at 42°C, although with increased autolysis in the stationary phase as observed by phase contrast microscopy (not shown) and by growth curve analysis (Fig. 7A), suggesting that the in vivo activity of LmGpsB_F91A was partially impaired. By contrast, mutations L94A and F105A completely prevented growth of strains LMS186 and LMS187, respectively (Fig. 7A), indicating that these amino acid exchanges generated biologically inactive GpsB proteins. Western blotting showed that all the GpsB mutant proteins were expressed both during growth at 37°C, where all strains are viable (Fig. 7C), and also 2 hours after a temperature shift to 42°C (Fig. 7D). The loss of growth at 42°C is therefore more likely to result from a loss of a functional property of GpsB rather than a loss of protein production or its degradation.

OPEN IN VIEWER

FIG. 7.

Effect of mutations in the GpsB C-terminus on growth of L. monocytogenes at 42°C. (A) L. monocytogenes strains EGD-e (wt), LMJR19 (ΔgpsB), LMS56 (ΔgpsB+gpsB), LMS185 (ΔgpsB+gpsBF91A), LMS186 (ΔgpsB+gpsBL94A), and LMS187 (ΔgpsB+gpsBF105A) were grown in BHI broth containing 1 mM IPTG at 42°C, and growth was recorded in hourly intervals. (B) Growth of L. monocytogenes strains EGD-e (wt), LMJR19 (ΔgpsB), LMS56 (ΔgpsB+gpsB), LMJR161 (ΔgpsB+gpsBT88A), and LMJR162 (ΔgpsB+gpsBT88D) was measured as described in (A). All average values and standard deviations were calculated from experiments performed in triplicate. (C) Western blots showing expression of LmGpsB proteins in the same set of strains as in (A, B) during growth at 37°C (top panel). Cells were harvested for protein isolation at an optical density of 1.0 (λ = 600 nm). A parallel Western blot shows DivIVA expression for control (bottom panel). (D) Stability of LmGpsB proteins after temperature upshift to 42°C. Western blot showing amounts of LmGpsB proteins in cells that were grown at 37°C to an optical density of 1.0 and then shifted to 42°C for 2 further hours before protein isolation (top panel). Again, a parallel Western blot shows DivIVA protein levels for control (bottom panel).

Role of the phosphorylation of GpsB

It has been reported recently that GpsB was phosphorylated in B. subtilis cells at T75 (T88 in LmGpsB) by the kinase PrkC. ¹⁹ The phosphomimetic mutations T75D and T75E appeared to have the same salt-sensitive phenotype as a gpsB deletion. ¹⁹ Likewise, we found that the introduction of the analogous T88D exchange into LmGpsB almost completely inactivated the protein. However, the phosphoablative T88A mutation was without effect on complementation activity of LmGpsB (Fig. 7B, D). T75 in BsGpsB and T88 in LmGpsB are located in the linker between the two domains. To investigate whether the T75E and T75D mutations had any impact on the hexamerization of BsGpsB, full-length, wild-type, T75E and T75D mutant BsGpsB proteins (BsGpsB_T75E and BsGpsB_T75D) were analyzed by SEC in conditions under which wild-type BsGspB was shown by SEC-MALS to form a stable hexamer. ¹¹ The T75E and the T75D mutations did not have a major effect on the size exclusion chromatograms of BsGpsB; the major species in wild-type and both mutant proteins had an identical retention volume (Fig. 5C). Similarly, thermal denaturation analysis by circular dichroism did not reveal any marked difference in the structural stability of wild-type BsGpsB and the T75D and T75E mutants (Fig. 6C).

Effect of BsGpsB on the glycosyltransferase and transpeptidase activity of BsPBP1 in vitro

To explore the functional role of GpsB and the relationship to its structure, the effect of BsGpsB on the GTase and TPase activities of BsPBP1 was investigated. BsPBP1 was used as both the TGase and TPase activities of this enzyme have successfully been recapitulated in vitro, ⁴⁶ which is not yet the case for the listeria PBPs. Both wild-type and phosphomimetic T75D/T75E mutants of BsGpsB were used in these experiments in case any modulation of PBP1 activity by GpsB requires prior phosphorylation of GpsB. The GTase activity of BsPBP1 was monitored using a fluorescence assay, in which the polymerization of fluorescently labeled lipid II followed by digestion of the resulting glycan chains by a muramidase results in a reduction of fluorescence intensity with time, correlated with the rate of the reaction. The addition of a twenty five-fold excess of either wild-type or the phosphomimetic T75D/T75E mutants of BsGpsB had no apparent effect on the rate of the GTase reaction catalyzed by BsPBP1 (Fig. 8A).

OPEN IN VIEWER

FIG. 8.

GpsB does not affect the enzyme activities of PBP1 in vitro. (A) The GTase activity of BsPBP1 in the presence (blue line) and absence of wild-type BsGpsB (red line), BsGpsB_T75D (green line), and BsGpsB_T75E (magenta line) was measured using fluorescently labeled lipid II as described previously. ⁴⁰ In the absence of BsPBP1 (black line), no lipid II consumption is observed. Each measurement is shown as the mean ± SD (n = 3). (B) HPLC chromatograms from in vitro PG synthesis assays. BsPBP1 in the presence and absence of BsGpsB was incubated with radiolabeled amidated lipid II-m-DAP. The resultant PG was digested, boiled, and reduced before the resulting muropeptides were separated by HPLC. Peak 1—disaccharide pentapeptide(NH₂) monophosphate resulting from unused substrate and glycan chain ends; peak 2—disaccharide tetrapeptide(NH₂) derived from GTase and carboxypeptidase activity; peak 3—disaccharide pentapeptide (nonamidated) from the GTase activity on contaminating nonamidated substrate; peak 4—disaccharide pentapeptide(NH₂) resulting from the GTase activity; peak 5—bis-disaccharide tetrapeptide(2NH₂) resulting from GTase, TPase, and carboxypeptidase activities; peak 6—bis-disaccharide tetrapentapeptide(2NH₂) resulting from GTase and TPase activities. HPLC, high-pressure liquid chromatography.

The effect of BsGpsB on BsPBP1 was explored further using a HPLC-based in vitro PG synthesis assay ⁴¹ to monitor the composition of peptidoglycan synthesized by BsPBP1 in the presence and absence of BsGpsB. Consistent with others' previous observations, ⁴⁶ BsPBP1 was active as a TPase when assayed with a native lipid II substrate that had been amidated at the ɛ-carboxylate of meso-diaminopimelic acid (lipid II-m-DAP). In this endpoint assay, PBP1 produced peptidoglycan with approximately 30% of the peptides present in cross-links, irrespective of the absence or the presence of a twenty five-fold molar excess of either wild-type or the phosphomimetic T75D/T75E mutants of BsGpsB (Fig. 8B). As observed with other PBPs, ³⁹ BsPBP1 also exhibited a weak DD-carboxypeptidase activity that was unaffected by the presence of any BsGpsB protein tested. Since BsGpsB interacts with BsPBP1 with a K_d of 0.7 μM, ¹¹ at the protein concentrations used in both assays BsPBP1 should be bound to BsGpsB, suggesting that the interaction has no discernible effect on either the GTase or the TPase activities of BsPBP1 in vitro.