Correlating the Structure and Activity of Y. pestis Ail in a Bacterial Cell Envelope

Abstract

Understanding microbe-host interactions at the molecular level is a major goal of fundamental biology and therapeutic drug development. Structural biology strives to capture biomolecular structures in action, but the samples are often highly simplified versions of the complex native environment. Here, we present an Escherichia coli model system that allows us to probe the structure and function of Ail, the major surface protein of the deadly pathogen Yersinia pestis. We show that cell surface expression of Ail produces Y. pestis virulence phenotypes in E. coli, including resistance to human serum, cosedimentation of human vitronectin, and pellicle formation. Moreover, isolated bacterial cell envelopes, encompassing inner and outer membranes, yield high-resolution solid-state NMR spectra that reflect the structure of Ail and reveal Ail sites that are sensitive to the bacterial membrane environment and involved in the interactions with human serum components. The data capture the structure and function of Ail in a bacterial outer membrane and set the stage for probing its interactions with the complex milieu of immune response proteins present in human serum.

Significance

Ail is a critical virulence factor of Yersinia pestis, and its interactions with human serum are central for promoting the immune resistance of bacteria to the human host defenses. Here, we capture the action of Ail in a functional bacterial environment and set the stage for probing its interactions with the complex milieu of immune response proteins present in human serum. The development of an Escherichia coli model system of Y. pestis for biophysical studies is new, to our knowledge, and biologically important. The work extends the range in situ NMR spectroscopy to include models of microbial infection.

Introduction

The protein and lipid components of bacterial outer membranes work together to support microbial cell survival in a wide range of host environments and represent key virulence factors. This phenomenon is especially striking in the case of Yersinia pestis—the agent responsible for multiple devastating human plague pandemics throughout history. The Y. pestis outer membrane protein adhesion invasion locus (Ail) and lipopolysaccharide (LPS) have coevolved to enhance microbial resistance to human innate immunity, enabling the bacterium to produce high-level septicemia in its mammalian hosts (1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11). Ail and LPS jointly promote evasion of the host immune defenses and adhesion and invasion of host cells (9, 10, 11). Ail and LPS mutants exhibit altered sensitivity to human serum, antibiotics, and cell wall stress (12,13), and we have identified LPS-recognition motifs on the surface of Ail important for establishing mutually reinforcing Ail-LPS interactions that promote microbial survival in human serum, antibiotic resistance and cell envelope integrity (14). These features highlight the role of the outer membrane environment as a critical regulator of Ail activity and underscore the importance of characterizing the native Ail-membrane assembly as a whole, rather than its individual components.

Previous structural studies have relied on purified Ail refolded in detergent micelles, detergent-free lipid nanodiscs, or liposomes for x-ray diffraction and NMR (15, 16, 17, 18, 19). The structure of Ail—an eight-stranded β-barrel with four extracellular loops and three intracellular turns—is conserved in all cases, but these chemically defined samples preclude NMR studies aimed at probing the molecular interactions of Ail with either bacterial membrane components, such as LPS or human ligands, in the context of the native environment. Importantly, this situation negates a major advantage of NMR compared to other structure determination technologies: the high sensitivity of NMR signals to their local environment, which makes them extremely useful for characterizing even weak intermolecular interactions by direct spectroscopic detection.

We have shown that nanodiscs and liposomes can incorporate Ail with various types of LPS, including native Y. pestis LPS (14,19), but even though these detergent-free lipid platforms represent important advances in membrane complexity, they remain distant substitutes for the native outer membrane, which is highly anisotropic, heterogeneous, and connected to the cytoskeletal peptidoglycan layer. The emerging field of in situ NMR (20, 21, 22, 23, 24, 25, 26) presents new opportunities for examining the structural properties and interactions of Ail in a native bacterial outer membrane.

Here, we describe an Escherichia coli model system in which Ail is natively folded in the bacterial outer membrane for parallel solid-state NMR structural studies and microbial functional assays in situ. Using this model, we demonstrate that Ail expression produces Y. pestis virulence phenotypes in E. coli and that isolated cell envelopes expressing ¹⁵N- and ¹³C-labeled Ail yield atomic-resolution NMR spectra that allow us to probe the structure of Ail in a native bacterial membrane and its interactions with human serum components. This model sheds light on the interactions of Ail with components of human serum and provides a platform for advancing structure-activity NMR studies of Ail in the native environment, including its various physiological modulations that can be elicited by factors such as temperature, antibiotics, or exposure to serum.

Materials and Methods

Expression of Ail in the E. coli outer membrane

The gene encoding mature Y. pestis Ail was cloned in the NcoI and XhoI restriction sites of the E. coli plasmid pET-22b(+), downstream of the signal sequence of pectate lyase B (27). This plasmid was transformed into E. coli cells by heat shock, and positive clones were selected by plating on LB agar with ampicillin (100 μg/mL) and chloramphenicol (35 μg/mL) in the case of Lemo21(DE3) cells. Three E. coli cell strains derived from BL21(DE3) (28) were tested: C41(DE3) (29), BL21(DE3)ΔACF (30), and Lemo21(DE3) (31). Transformed cells were cultured overnight at 30°C, with vigorous shaking, in 50 mL of M9 minimal media supplemented with Basal Medium Eagle vitamin solution (1% by volume), ampicillin (100 μg/mL), and chloramphenicol (50 μg/mL). This overnight culture was used to inoculate 500 mL of fresh M9 media (supplemented as above), and grown from OD₆₀₀ ∼0.05 to OD₆₀₀ ∼0.4 at 30°C, at which point the cells were harvested by low-speed centrifugation (5000 × g, 4°C, 20 min) and then resuspended in 500 mL of fresh supplemented M9 media. Protein expression was induced by adding 0.4 mM isopropyl β-D-1-thiogalactopyranoside (IPTG), after which the shaking speed was reduced and the temperature lowered to 25°C. The cells were cultured for 20 min before adding rifampicin (100 μg/mL) and then for an additional 20 h in the dark. The cells were harvested by low-speed centrifugation and resuspended in HEPES buffer (10 mM, pH 7.4).

To obtain uniformly ¹⁵N,¹³C-labeled Ail, the cells were grown in unlabeled M9 media and only transferred to ¹⁵N,¹³C-labeled M9 prepared with (¹⁵NH₄)₂SO₄ (1 g/L) and ¹³C₆-glucose (5 g/L) before induction with IPTG. To obtain ²H labeling of endogenous E. coli biomolecules, cells were grown in 100 mL of ²H₂O and then switched to H₂O before induction with IPTG.

Isolation of bacterial membrane fractions

Ail-expressing cells were lysed by three passes through a French press. After removing cellular debris by centrifugation (20,000 × g, 4°C, 1 h), the total cell envelope fraction, including inner and outer membranes, was harvested by ultracentrifugation (100,000 × g, 4°C, 1 h), washed three times with sodium phosphate buffer (20 mM, pH 6.5), and then collected by ultracentrifugation. To further isolate the outer membrane fraction, the total cell envelope was suspended in 7 mL of buffer I (10 mM HEPES, pH 7.4, 3.4 mM N-lauroylsarcosine) and then separated by ultracentrifugation. The supernatant containing detergent-solubilized inner membrane was removed, and the pellet was washed three times with sodium phosphate buffer (20 mM, pH 6.5) and then collected by ultracentrifugation. All three fractions—total cell envelope, inner membrane, and outer membrane—were analyzed by polyacrylamide gel electrophoresis (PAGE) in sodium dodecyl sulfate (SDS) and immunoblotting with the α-Ail-EL2 antibody (17).

NMR sample preparation

Samples for solid-state NMR studies were packed into 4, 3.2, or 1.3 mm magic angle spinning (MAS) rotors. For samples incubated with human serum—either normal human serum (NHS) or heat-inactivated human serum (HIS)—isolated cell envelopes expressing ¹³C,¹⁵N Ail were suspended in 7 mL of serum and gently mixed for 4 h at room temperature before harvesting and washing with sodium phosphate buffer by ultracentrifugation and packing into the MAS rotor. The preparation of Ail liposome samples has been described (18).

Solid-state NMR spectroscopy

Solid-state NMR experiments with ¹³C detection were performed on a 750 MHz Bruker Avance III HD spectrometer with a Bruker 3.2 mm ¹H,¹³C,¹⁵N E-free MAS probe or a 500 MHz Bruker Avance spectrometer with a Bruker 4 mm ¹H,¹³C,¹⁵N E-free MAS probe, at an effective sample temperature of 7 ± 5°C, with a spin rate of 11,111 Hz. Typical π/2 pulse lengths for ¹H, ¹³C, and ¹⁵N were 2.5, 2.5, and 5 μs, respectively. ¹H decoupling was implemented with the SPINAL64 sequence, with a radio frequency field strength of 90 kHz during acquisition. Two-dimensional ¹⁵N/¹³C NCA spectra were acquired using the SPECIFIC-CP pulse program, with contact times of 2 ms and a 70–100% ramp, for ¹H/¹⁵N cross polarization (CP), and 4.9 ms for ¹⁵N/¹³C CP. Two-dimensional ¹³C-¹³C PDSD (proton-driven spin diffusion) spectra were acquired with a ¹H-¹³C contact time of 1 ms, a ¹³C-¹³C mixing time of 50 ms, and 20 ms of SWf-TPPM ¹H decoupling.

Experiments with ¹H detection were performed on a 900 MHz Bruker Avance III HD spectrometer equipped with a Bruker 1.3 mm MAS probe, operating at an effective sample temperature of 30 ± 5°C, with a spin rate of 57,000 ± 15 Hz. Two-dimensional ¹H/¹⁵N CP-HSQC spectra were acquired as described previously (18), using the MISSISSIPPI sequence (32) for water suppression.

Bacterial cell assays

E. coli Lemo21(DE3) cells transformed with Ail-expressing plasmid pET-22b(+) or with empty plasmid were cultured as described above, then harvested by low-speed centrifugation, washed twice in ice-cold phosphate-buffered saline (PBS), and resuspended in ice-cold PBS to OD₆₀₀ = 1.0.

To assay binding to human vitronectin (Vn), washed bacteria (250 μL) were added to an equal volume of NHS (H4522; Sigma-Aldrich, St. Louis, MO), HIS (H3667; Sigma-Aldrich), purified full-length Vn (1 μM in PBS; SRP3186; Sigma-Aldrich), or purified Vn-HX (2 μM in PBS) prepared as described (33). The binding reactions were incubated for 30 min at 37°C, then placed on ice. Bacterial cells and bound proteins were then collected by centrifugation (6000 × g, 5 min, 4°C), washed three times with 1 mL of ice-cold PBS containing 0.1% Tween-20, and lysed by boiling in 100 μL of SDS-PAGE sample buffer (50 mM Tris-HCl, pH 6.8, 2% SDS, 5% glycerol, 1% β-mercaptoethanol). Bacterial cell lysates and cosedimented proteins were subjected to SDS-PAGE and immunoblot analysis with rabbit polyclonal anti-Vn (R12-2413; Assay Biotech, Fremont, CA) and anti-Ail-EL2 (34) antibodies.

To assay serum resistance, E. coli cells collected after incubation with NHS or HIS were diluted 100-fold with PBS, and 25 μL of this dilution mixture was plated on LB agar supplemented with ampicillin (100 μg/mL) and chloramphenicol (35 μg/mL). After incubating at 37°C overnight, the percentage of cell survival was estimated by counting the number of bacterial colonies present on each plate using ImageJ (35). The percent survival represents the number of colonies that survive in NHS divided by the number that survive in HIS.

To assay pellicle formation, E. coli cells were harvested, then resuspended in 2 mL of M9 minimal media to OD₆₀₀ = 0.5 in 15 × 100 mm glass tubes and incubated for 16 h at 37°C. After removing the cells by centrifugation, the interior glass walls were treated with methanol and air dried overnight, then washed three times with PBS and again air dried overnight. The tubes were treated with 2.5 mL of crystal violet solution (0.1%) for 10 min, then washed with PBS and air dried. Pellicle formation was detected qualitatively as a violet-stained rim at the air-liquid interface.

Results and Discussion

Production of folded Ail in the outer membrane of E. coli

High-resolution, multidimensional NMR studies require isotopically ¹⁵N- and ¹³C-labeled biomolecules. For studies of proteins in native cell membranes, targeted isotopic labeling is critical for suppressing background NMR signals from other cellular components, and solid-state NMR methods are needed to overcome the correlation time limitations posed by samples that are effectively immobilized on a timescale greater than microseconds. Moreover, the inherently low sensitivity of NMR necessitates samples that are enriched in target protein. This requirement is compatible with the properties of Ail, which naturally comprises more than 30% of the Y. pestis outer membrane proteome at the mammalian infection temperature of 37°C (6,9,36, 37, 38).

E. coli is widely used as a model for assaying the functions of Y. pestis proteins, including Ail. E. coli strains derived from BL21(DE3) have a rough-type LPS that lacks an extended O-antigen, similar to that of Y. pestis, and we have shown that E. coli rough-type LPS and Y. pestis LPS both perturb the NMR spectra of Ail in a similar manner (14). The use of E. coli also facilitates isotope labeling for NMR.

To drive the production of folded ¹⁵N,¹³C-Ail in the E. coli outer membrane, we cloned the sequence of the pectate lyase B (PelB) leader peptide (27) before the N-terminus of mature Ail and relied on the E. coli native β-barrel assembly machinery (BAM) to insert Ail across the bacterial outer membrane. The Ail-enriched outer membrane or the entire cell envelope—including inner and outer membranes—was isolated and taken directly for NMR (Fig. 1 A). To reduce isotopic labeling of endogenous E. coli components, the cells were grown in unlabeled minimal media and transferred to isotopically labeled media only before induction with IPTG. We tested Ail expression in three derivative strains of E. coli BL21(DE3) (28), in which recombinant protein synthesis is controlled by a chromosomally encoded, IPTG-inducible polymerase from bacteriophage T7 and its plasmid-encoded T7 promoter.

Figure 1.

Expression of folded Ail in the E. coli outer membrane. (A) Depiction of the bacterial cell envelope (CE) isolated from whole cell (WC) for NMR studies, including outer membrane (OM) with embedded Ail (cyan), inner membrane (IM), peptidoglycan (PG), and periplasm. (B–G) SDS-PAGE of CE, WC, or OM fractions isolated from E. coli C41(DE3), BL21(DE3)ΔACF, or Lemo21(DE3) cells. Ail was visualized with Coomassie stain (A–D and F) or by immunoblotting with Ail-specific antibody (E and G). Cells incubated with NHS for 30 min were analyzed by SDS-PAGE and immunoblotting after heat denaturation (G). To see this figure in color, go online.

Our initial attempts utilized C41(DE3) cells (29), which promote folded protein insertion in the E. coli membrane because of a mutation that weakens the production of T7 polymerase (31). These cells yield high expression of properly folded Ail (Fig. 1 B) but also produce appreciable amounts of endogenous E. coli proteins that are isotopically labeled. Background NMR signal can be reduced by expressing Ail in BL21(DE3)ΔACF, a mutant cell strain that lacks the major E. coli outer membrane proteins OmpA, OmpC, and OmpF (30). Compared with C41 strains, this mutant is compatible with the use of rifampicin, a potent inhibitor of E. coli RNA polymerase (39) that suppresses endogenous protein production and was first used to enable targeted isotopic labeling of heterologous proteins for solution (40) NMR studies in situ. These cells produce very high levels of Ail (Fig. 1 C), but their usefulness is limited by the production of large amounts of misfolded protein, detectable as a band migrating just above 17 kDa in SDS-PAGE that is distinguishable from folded Ail at 14 kDa (17). The addition of rifampicin to BL21(DE3)ΔACF cells reduces protein expression levels overall but does not enhance the proportion of folded to unfolded Ail. Although endogenous background signal can be reduced by isolating the outer membrane from the cell envelope, this process exposes the sample to the compromising effects of detergent on biomolecular structure and activity (41,42). The two-dimensional (2D) ¹⁵N/¹³C NCA solid-state NMR spectra from outer membrane preparations of either C41(DE3) or BL21(DE3)ΔACF cells reflect the overall signature of Ail (17, 18, 19) but suffer from suboptimal resolution and sensitivity (Fig. S1).

To avoid these complications, we tested Lemo21(DE3) cells (31), in which T7 RNA polymerase can be controlled by its natural inhibitor T7 lysozyme and rifampicin can be used to effectively halt transcription of the E. coli genome. By initially growing the cells in unlabeled minimal media, then transferring them to isotopically labeled media supplemented with both rifampicin and IPTG, expression of ¹⁵N/¹³C-Ail was induced and allowed to continue under control of the bacteriophage T7 promoter, but endogenous protein production was blocked. As an additional important advantage, we found that rifampicin moderates the level of Ail overexpression and reduces protein misfolding to negligible levels, resulting in cell envelope and outer membrane fractions that are highly enriched in folded Ail (Fig. 1, D and E).

Solid-state NMR of Ail in bacterial cell envelopes

Cell envelope preparations from Lemo21(DE3) cells yield high-quality 2D ¹⁵N/¹³CA solid-state NMR spectra with respect to both signal intensity and resolution (Fig. 2 A, black). The line widths of well-resolved signals, such as those from Ala and Gly (Fig. S2), are in the range of 0.6–0.8 ppm for ¹³C and 1.3–1.8 ppm for ¹⁵N. They compare favorably with those measured from purified Ail reconstituted in liposomes (18), demonstrating that the combined use of this cell strain with rifampicin is highly beneficial for producing folded, isotopically labeled Ail. Based on comparison of the one-dimensional cross sections of the ¹⁵N/¹³C peaks (Figs. S2 and S6) with those from Ail liposomes (18), we estimate that the cell envelope sample in the 3.2 mm MAS rotor contains ∼0.7 mg of Ail or 25% of the Ail liposomes.

Figure 2.

Solid-state ¹⁵N/¹³C/¹H NMR of Ail in E. coli cell envelopes. (A) 2D ¹⁵N/¹³C NCA spectra of ¹⁵N,¹³C-Ail in E. coli CE (black) or reconstituted liposomes (red). Spectra were recorded at 750 MHz, 7°C, with an MAS rate of 11 kHz and 640 transients for CE or 128 transients for liposomes. Resolvable assigned peaks are marked. Asterisks denote new unassigned peaks. (B) ¹H-detected ¹H/¹⁵N CP-HSQC spectra of ¹⁵N-Ail in E. coli CE (black) or reconstituted liposomes (red). Spectra were recorded at 900 MHz, 30°C, with an MAS rate of 57 kHz and 1600 transients for CE or 160 transients for liposomes. (C) Structural model of Ail embedded in the Y. pestis OM taken from previous MD simulations (14). Spheres denote resolved and assigned amide N atoms that undergo ¹H/¹⁵N chemical shift perturbations from 0 ppm (cyan) to 0.15 ppm (red) between the cellular and liposome environments. Side chains forming two clusters of LPS-recognition motifs are shown as yellow sticks. The boundaries of the OM phospholipid and LPS layers are marked (gray lines). Residue numbers, from E1 to F156, correspond to the mature sequence of Ail. To see this figure in color, go online.

Notably, the one-dimensional ¹³C spectra (Fig. S3) acquired for E. coli cell envelopes prepared from Ail(−) bacteria, transformed with empty plasmid but otherwise grown in an identical manner as Ail(+) bacteria, show no evidence of protein signals when compared with the spectra of Ail(+) preparations. These cells do not express Ail (Fig. 1 F), and together with the NMR data, they demonstrate that the expression system is tightly regulated and leads to highly selective isotope labeling of the target Ail, with minimal background.

Many resonances in the 2D ¹⁵N/¹³CA spectrum can be assigned by direct comparison with the solid-state and solution NMR spectra of purified Ail reconstituted in liposomes (Fig. 2 A, red) or nanodiscs (18,19), indicating that the same overall structure—an eight-stranded β-barrel with four extracellular loops (EL1–EL4) and three intracellular turns (T1–T3) (15,16,19)—is preserved in the bacterial outer membrane. Within the 2D ¹⁵N/¹³CA spectrum, there are some new signals, and 47 crosspeaks are sufficiently resolved to facilitate the detection of site-specific chemical shift perturbations (relative to liposomes) for residues dispersed across the Ail sequence and its β-barrel topology (Fig. S4 A).

The ¹H, ¹⁵N, and ¹³C chemical shifts from backbone sites are sensitive to chemical environment, ligand binding, and protein conformational changes (43). To examine the effects of the native membrane environment on Ail, we acquired a ¹H-detected ¹H/¹⁵N CP solid-state NMR spectrum of Ail in bacterial cell envelopes (Fig. 2 B, black). Typically, ²H labeling of side-chain protons is needed to achieve line narrowing and high sensitivity in ¹H-detected solid-state NMR experiments with MAS rates ∼60 kHz (44,45). For purified and reconstituted membrane proteins, this can be achieved by labeling with ²H and back exchanging the amide hydrogens to ¹H during refolding from denaturing conditions. This approach, however, is not readily applicable to cellular membrane proteins, for which membrane-embedded segments remain water inaccessible throughout the sample preparation protocol. In this study, we prepared the sample by initially growing cells in ²H₂O and then transferring the culture to ¹H₂O, with ¹⁵N salts and rifampicin, for Ail induction. The goal was to silence the background by ²H labeling (∼70%) of endogenous E. coli components while achieving targeted ¹⁵N labeling of Ail. The resulting 2D CP-HSQC spectrum is strikingly good, despite the lack of Ail deuteration.

The cell envelope spectrum, obtained at 900 MHz ¹H frequency and 30°C, compares favorably with the CP-HSQC solid-state NMR spectrum of fractionally deuterated (∼70%) Ail in liposomes (Fig. 2 B, red), which was obtained at the same field and temperature (18), and also with the TROSY-HSQC solution NMR spectrum of uniformly deuterated (∼99%) Ail in nanodiscs obtained at 800 MHz and 45°C (Fig. S5; (18,19)). Individual resonances have line widths in the range of 0.15 ppm for ¹H and 1.5 ppm for ¹⁵N, in line with recent reports of ¹H-detected spectra of membrane proteins in native cell membranes (46). A total of 22 ¹H/¹⁵N peaks can be resolved and assigned by direct comparison with the previously assigned solution NMR spectra of nanodiscs and solid-state NMR spectra of liposomes (14,18,19). Improvements in ¹H/¹⁵N resolution should be available through the implementation of solid-state NMR experiments with MAS rates greater than 60 kHz as described (44,45) and ²H labeling schemes such as those that enable selective visualization of the water-accessible loops, obtained by growing cells in ²H-labeled media, followed by incubation in H₂O to back exchange water-accessible sites.

Previously, we showed that Ail acquires conformational order in the presence of LPS, leading to substantial enhancement of solid-state NMR CP signal intensity (19). Moreover, all-atom molecular dynamics (MD) simulations (14) indicate that the extracellular loops of Ail exhibit dramatically reduced root mean-square fluctuations in a native bacterial membrane. In line with these earlier observations, the NCA and CP-HSQC spectra of Ail in cell envelopes contain some new signals, which were not previously observed in the spectra from reconstituted liposomes. These include HN peaks assigned in the solution NMR spectra of nanodiscs (G2, G60, G66, G105, S114, and the side chain of W148), as well as new unassigned peaks. These results suggest that Ail has more restricted conformational dynamics in E. coli outer membranes, leading to enhanced CP transfer efficiency and signal intensity.

Spectral comparisons reveal chemical shift differences between the chemically defined liposome membrane environment and the bacterial cell envelope (Fig. 2 C; Table S1). Independent spectroscopic assignments will be needed to fully and precisely map the effects of the bacterial membrane environment on the protein, but some prominent differences are detectable at this stage. These map to sites in the extracellular loops (D23, G60, G66, and G105) and near the extracellular membrane-water interface (A11, G28, A48, G92, and G94), just below the two clusters of positively charged residues that we have identified as LPS-recognition motifs on opposite poles of the Ail β-barrel (14): cluster I (R14, K16, K144, and K139), which is tightly localized to the base of the two short extracellular loops EL1 and EL4, and cluster II, which occupies a broader region on the barrel surface extending from the base (R27, R51, and H95) to the outer extremities (K69, K97, and K99) of the two long loops EL2 and EL3.

Notable spectral differences are also observed for membrane-embedded sites of the Ail β-barrel (G9, G89, D132, G151, and G153) and sites in the intracellular turns (R80, I81, F123, and N124), suggesting that the entire protein senses the physical-chemical properties of its membrane environment. The results are in line with our recent report (14) that specific interactions of outer membrane LPS with cluster I and cluster II Ail sites enhance the conformational order of Ail, dampen LPS dynamics, and cause an overall thickening of the outer membrane around its β-barrel, whereas mutations in the Ail binding sites for LPS compromise cell envelope integrity, reduce Y. pestis survival in serum, and enhance antibiotic susceptibility.

To further examine the side-chain conformations of Ail in the bacterial cell envelope, we acquired a 2D ¹³C/¹³C correlation PDSD spectrum (Fig. 3, black). The spectrum, obtained in 25 h, has excellent resolution. The overall pattern of crosspeaks is conserved between bacterial cell envelope and liposome preparations (Fig. 3, red), indicating that the side-chain conformations are broadly similar. Resolved signals from Ala, Ile, Pro, Ser, Thr, and Val spin systems can be easily identified based on their chemical shifts, which reflect the β-barrel structure of Ail, and 54 intraresidue crosspeaks can be assigned (Fig. S6) by direct transfer from the liposome NMR data that were assigned previously (18). By contrast, the spectrum of Ail(−) cell envelopes contains no protein signals (Fig. S6), confirming the isotopic labeling selectivity of the expression system.

Figure 3.

Solid-state ¹³C/¹³C NMR of Ail in E. coli cell envelopes. 2D ¹³C/¹³C PDSD spectra were recorded at 750 MHz, 7°C, with an MAS rate of 11 kHz, and with 204 t1 increments and 304 transients for CEs (black) or 512 t1 increments and 64 transients for reconstituted liposomes (red). Resolvable signals from Ala, Ile, Ser, Thr, and Val residues of Ail (blue) and non-Ail bacterial CE components (gold) are marked. (A) Aliphatic region. (B) Expanded spectral regions. To see this figure in color, go online.

The ¹³C/¹³C correlation PDSD spectrum also contains signals from cell envelope components other than Ail, including lipids, LPS, and peptidoglycan, which make up a large proportion of the bacterial cell mass and incorporate ¹³C even with the use of rifampicin because of the relatively fast doubling time of E. coli cells and the lag time between IPTG induction and rifampicin addition. Although the flexible regions of these molecules are detectable only by through-bond polarization transfer, their more rigid groups are visible in the NMR spectra obtained with CP (23,47) and are present in the PDSD spectra of both Ail(+) and Ail(−) cell envelopes (Fig. S6).

Treatment with cerulenin, an inhibitor of fatty acid biosynthesis has been used to inhibit ¹³C labeling of E. coli lipids and simplify ¹³C/¹³C correlation spectra of proteins (22). Moreover, spectroscopic approaches have been developed to silence lipid signals based on ¹⁵N-filtering (48), and NMR data acquisition above the gel-to-liquid phase transition of the lipids has also been shown to suppress lipid signal intensity (49). These cell envelope signals, however, also present an opportunity to probe specific interactions of Ail with the cell envelope.

Notably, a set of ¹³C/¹³C cross peaks observed in the region between 30–45 and 60–75 ppm of the Ail(+) cell envelope spectrum are absent from the Ail(−) spectrum. Resonance assignments will be needed to determine the network of interatomic interactions in the Ail outer membrane, but there are two plausible explanations for this apparently Ail-dependent observation. Firstly, these resonances could reflect contacts between LPS sugar moieties (60–75 ppm) and His, Lys, Arg, and Tyr side chains of Ail (30–45 ppm), in line with previous observations based on chemical shift perturbation analysis (14). Secondly, they could reflect inter- or intramolecular contacts of LPS and lipids, resulting from the general ordering effect of Ail on the outer membrane of E. coli. These effects need not be exclusive but could synergize such that Ail-LPS interactions rigidify bound LPS molecules, thereby enhancing LPS intramolecular CP transfer.

Emergence of Y. pestis phenotypes in E. coli through Ail expression

One important advantage of in situ NMR spectroscopy is the ability to probe the structural underpinnings of biology by assaying protein activity in the same samples that are analyzed by high-resolution spectroscopy. Having confirmed the structural integrity of Ail expressed in E. coli cells, we asked whether Ail also presents its characteristic virulence phenotypes observed in Y. pestis.

The role of Ail in providing serum resistance to Y. pestis is well known, and expression of Ail in E. coli DH5α cells is sufficient to render them immune to serum-mediated killing (34). Similarly, we found that Ail(+) Lemo-21(DE3) cells, expressing plasmid-encoded Ail, exhibit an 87 ± 7% survival after incubation with NHS relative to HIS. By contrast, Ail(−) cells, transformed with empty plasmid, formed no colonies after incubation with NHS (Fig. 4 A).

Figure 4.

Y. pestis phenotypes induced by Ail in E. coli. Assays were performed with E. coli Lemo21(DE3) cells transformed with Ail-bearing plasmid, Ail(+), or empty plasmid, Ail(−). Each data set is representative of at least triplicate experiments. (A) Survival of E. coli cells in NHS relative to HIS was assayed by incubating cells with serum, then plating on agar and counting the surviving bacterial colonies. Survival is reported as percent serum resistance relative to the number of surviving colonies in HIS. (B) Pellicle formation was observed in Ail(+), but not Ail(−), cells. Cells were suspended in 2 mL of M9 minimal media (OD₆₀₀ = 0.5, in 15 × 100 mm glass tubes) and incubated for 16 h at 37°C. After removing the cells by centrifugation, the interior glass walls were treated with methanol and air dried overnight, then washed three times with buffer and air dried overnight. Finally, the tubes were treated with 2.5 mL of crystal violet solution (0.1% for 10 min), then washed with buffer and air dried. Pellicle formation was detected as a violet-stained rim at the air-liquid interface (arrow). (C) Vn binding activity was assayed by co-sedimentation of Ail(+) or Ail(−) bacterial cells with NHS, HIS, purified full-length Vn, or purified Vn-HX. Immunoblots were probed with anti-Vn antibodies. (D) ¹H-detected ¹H/¹⁵N CP-HSQC solid-state NMR spectra of Ail in E. coli cell envelopes acquired before (black) or after (blue) incubation with NHS. Spectra were recorded at 900 MHz, 30°C, with an MAS rate of 57 kHz and 1600 (black) or 2048 (blue) transients. (E) Structural model of Ail embedded in the Y. pestis OM taken from a recent MD simulation (14). Spheres denote resolved and assigned amide N atoms that undergo ¹H/¹⁵N chemical shift perturbations from 0 ppm (cyan) to 0.2 ppm (blue) in the presence of NHS. The boundaries of the OM phospholipid and LPS layers are marked. To see this figure in color, go online.

Ail is also known to promote bacterial cell autoaggregation, pellicle formation, and flocculent bacterial growth (50), three phenotypes that are associated with virulence in Y. pestis strains and many other microbial pathogens. In the case of Ail, this property is thought to be related to the distribution of charged amino acids in its extracellular loops. Notably, Ail(+) Lemo21(DE3) cells displayed marked autoaggregation and formed a pellicle at the air-liquid interface that can be readily visualized by staining with crystal violet (Fig. 4 B). By contrast, neither Ail(−) or native nontransformed cells formed pellicles.

Finally we tested the ability of Ail(+) Lemo-21(DE3) cells to bind known ligands of Ail. Previous studies have shown that Y. pestis relies on Ail to associate with many human host proteins, including fibronectin (51), C4b-binding protein (C4BP; 52), and Vn (33,34). Incubation of Ail(+) cells with NHS, purified full-length Vn, or purified Vn-HX—corresponding to the HX domain of Vn (33)—resulted in Vn cosedimentation in all three cases (Fig. 4 C). By contrast, no cosedimentation was observed for either Ail(−) cells or Ail(+) cells incubated with HIS, in which Vn is presumably denatured by heat treatment (34). We conclude that Ail confers specific virulence-related Y. pestis phenotypes to Lemo21(DE3) cells. Notably, this implies that the samples for structural analysis are taken from a biologically active cellular system.

Interaction of Ail with human serum components

Identifying specific Ail sites involved in serum resistance is important. To reconstitute the host-pathogen interface that exists when Y. pestis enters the blood stream and study this microenvironment at atomic resolution, we incubated Ail(+) cell envelopes with NHS, washed extensively with buffer to eliminate nonspecific binding, and then transferred the sample to the NMR rotor for spectroscopic analysis. Previous studies (34) have shown no evidence of Ail proteolysis in the presence of serum, and we confirmed that this was indeed the case by immunoblot analysis of Ail after incubation with NHS (Fig. 1 G; note that the sample was heat denatured before SDS-PAGE and immunoblotting, and thus, Ail migrates at the apparent molecular weight of unfolded protein). Incubation with NHS produced several chemical shift perturbations in the ¹H/¹⁵N spectrum of Ail (Fig. 4, D and E; Fig. S4 B).

Ail is known to bind at least two serum components, C4BP (34,52) and Vn (33,34). Although additional studies with purified Ail ligands will be needed to assign the observed perturbations to specific serum molecules, it is notable that a few ¹H/¹⁵N signals are appreciably perturbed. These include signals from G66, G92, and G94, residues that localize near F54, F68, S102, and F104, which have been shown to be critical for binding serum Vn and for the serum resistance phenotype (53). Interestingly, intramembrane perturbations near the center of the Ail transmembrane β-barrel are also observed, indicating that the interactions of serum molecules with the extracellular loops of Ail may relay allosterically to membrane-embedded sites. Moreover, it is possible that the Ail β-barrel senses outer membrane perturbations caused by the association of serum components with the bacterial cell surface.

Conclusions

The ability to probe the molecular structure and interactions of Ail in its native cellular environment is critical for gaining mechanistic insights into Ail-mediated pathogenesis and for developing medical countermeasures. Previously, using NMR and MD simulations with artificial membranes of defined molecular composition, we identified specific molecular interactions between Ail and LPS that are important for Y. pestis resistance to serum and antibiotics (14). Here, we have shown that Ail and its virulence phenotypes can be expressed in a bacterial outer membrane for parallel in situ microbiology assays and solid-state NMR structural studies, with atomic resolution. These tools allow us to expand the size and complexity of the Ail microenvironment further than previously possible, and identify key residues that are sensitive to both the bacterial membrane environment and the interactions of human serum.

In NMR studies with bacterial cell envelopes, we detect specific Ail signals that firmly establish the sensitivity of the protein's conformation to its microenvironment. Perturbations across distinct topological regions of Ail can be detected as a response to its expression in the asymmetric bacterial outer membrane and, although the overall structure of Ail is unchanged, the native membrane appears to have subtle ordering effects on the extracellular loops. The NMR data further reveal sensitivity of specific Ail sites to serum exposure and, although the identity of the serum components responsible for these effects remains to be determined, human Vn is a good candidate because it is avidly recruited by Ail to the bacterial cell surface. Finally, the NMR data indicate that Ail may sense serum-induced perturbations of the surrounding membrane environment. While the mechanism underlying these observations is unknown, the recruitment of complement components at the cell surface could be responsible for such an effect.

These results provide a molecular window to the Ail-membrane assembly. They reaffirm the importance of the membrane environment as a key regulator of biological function and underscore the importance of characterizing the native assembly as a whole rather than its individual components.

Editor: Timothy Cross.

Author Contributions

J.E.K., L.M.F., K.S., C.S., Y.Y., and S.P., experiment execution. J.E.K., S.J.O., G.V.P., and F.M.M., research design, data analysis, writing, and editing.