Characterization of Posttranslationally Modified Multidrug Efflux Pumps Reveals an Unexpected Link between Glycosylation and Antimicrobial Resistance

Abstract

The substantial rise in multidrug-resistant bacterial infections is a current global imperative. Cumulative efforts to characterize antimicrobial resistance in bacteria has demonstrated the spread of six families of multidrug efflux pumps, of which resistance-nodulation-cell division (RND) is the major mechanism of multidrug resistance in Gram-negative bacteria. RND is composed of a tripartite protein assembly and confers resistance to a range of unrelated compounds. In the major enteric pathogen Campylobacter jejuni, the three protein components of RND are posttranslationally modified with N-linked glycans. The direct role of N-linked glycans in C. jejuni and other bacteria has long been elusive. Here, we present the first detailed account of the role of N-linked glycans and the link between N-glycosylation and antimicrobial resistance in C. jejuni We demonstrate the multifunctional role of N-linked glycans in enhancing protein thermostability, stabilizing protein complexes and the promotion of protein-protein interaction, thus mediating antimicrobial resistance via enhancing multidrug efflux pump activity. This affirms that glycosylation is critical for multidrug efflux pump assembly. We present a generalized strategy that could be used to investigate general glycosylation system in Campylobacter genus and a potential target to develop antimicrobials against multidrug-resistant pathogens.IMPORTANCE Nearly all bacterial species have at least a single glycosylation system, but the direct effects of these posttranslational protein modifications are unresolved. Glycoproteome-wide analysis of several bacterial pathogens has revealed general glycan modifications of virulence factors and protein assemblies. Using Campylobacter jejuni as a model organism, we have studied the role of general N-linked glycans in the multidrug efflux pump commonly found in Gram-negative bacteria. We show, for the first time, the direct link between N-linked glycans and multidrug efflux pump activity. At the protein level, we demonstrate that N-linked glycans play a role in enhancing protein thermostability and mediating the assembly of the multidrug efflux pump to promote antimicrobial resistance, highlighting the importance of this posttranslational modification in bacterial physiology. Similar roles for glycans are expected to be found in other Gram-negative pathogens that possess general protein glycosylation systems.

Keywords: N-linked glycans; glycosylation; multidrug efflux pump.

FIG 1

Schematic diagram of N-linked glycosylation pathway in C. jejuni and model for the role of glycosylation in the functioning of multidrug efflux pumps. Each protein component of CmeABC carries two glycosylation sites whereby PglB catalyzes the transfer of the N-linked glycans to CmeABC. CmeA is glycosylated at positions ¹²¹DFNRS¹²⁵ and ²⁷¹DNNNS²⁷⁵, CmeB is glycosylated at position ⁶³⁴DRNVS⁶⁴⁸ and theoretically at ⁶⁶³DRNAS⁶⁶⁷, and CmeC is glycosylated at position ⁴⁷ETNSS⁵¹ and theoretically at ³⁰EANYS³⁴.

FIG 2

Functional studies and effect of glycosylation on WTCmeABC and g0CmeABC. (A) DSSP analysis of CmeC (PDB 4TM4). Glycosylation sites ⁴⁷ETNSS⁵¹ (reported) and ⁰EANYS³⁴ are located in a flexible loop. Glycosylation sites in CmeC are denoted by asterisk. (B) Construction of C. jejuni variants. Inactivation of cmeD was achieved by introducing chloramphenicol cassette in the middle of the gene. This strain was later used as a parent strain to construct WTCmeABC and g0CmeABC, whereby glycosylation of CmeABC was disabled by introducing N→Q amino acid alteration to asparagine in D/E-X-N-X-S/T (X = any amino acid other than proline). (C) Western blot detection of CmeC. WTCmeABC and g0CmeABC strains were grown overnight in brucella broth media, and cells were pelleted and incubated with 2% sodium dodecyl sulfate and sodium Sarkosyl for 2 h at room temperature. The cell debris was then pelleted by centrifugation, and supernatants were mixed 1:1 with Laemmli loading buffer supplemented with dithiothreitol. Proteins were then separated by SDS-PAGE, followed by electroblotting to a polyvinylidene difluoride membrane. His₆-tagged CmeC was probed by primary anti-His₆ mouse antibody and visualized by using a LI-COR Odyssey apparatus. (D) Ethidium bromide accumulation test in C. jejuni strains. Brucella broth (30 ml) was separately inoculated with an overnight culture of C. jejuni WTCmeABC (black) and C. jejuni g0CmeABC (gray) to an OD₆₀₀ of 0.1. Cells were grown until reaching an OD₆₀₀ of 0.4 to 0.5 and then spun down, washed, and resuspended to an OD₆₀₀ of 0.2 in 10 mM sodium phosphate buffer (pH 7). The cells were incubated in a VAIN apparatus for 15 min at 37°C. Ethidium bromide was then added to final concentration of 0.2 mg/ml. (D) The fluorescence was read at excitation and emission for 20 min at 37°C. (E) Eithidum bromide accumulation in C. jejuni strains at 15 min. (F to I) MICs of C. jejuni WTCmeABC and C. jejuni g0CmeABC. The MIC was read directly from the strip at the point where the zone of inhibition of bacterial growth intersected with the antibiotic concentration on the strip. The data represent the means of three biological replicates, with two technical replicates for each. Significance was calculated using Mann-Whitney test (**, P < 0.01).

FIG 3

Generation and mass spectrometry analysis of fully glycosylated CmeA. CmeA was purified using IMAC, followed by concentration and buffer exchange using Amicon Ultra 0.5-ml centrifugal filter units. Proteins were then separated by SDS-PAGE and visualized by Coomassie blue staining (A) or electroblotted to a polyvinylidene difluoride membrane (B). CmeA-His₆ was probed using anti-His₆ mouse antibody (B) or SBA lectin (C) and then visualized by a Li-COR Odyssey, which allowed two-colors immunoblot (D). Lane 1, g0CmeA from E. coli DH10B; lane 2, gCmeA from SDB1 carrying pACYC(pgl) and pWA2; lane 3, fully glycosylated CmeA was produced in E. coli SDB1 carrying pEXT21CjpglB, pACYC(pgl), and pWA2. (E and F) Mass spectrometry analysis of glycopeptides from CmeA. Spectra were produced by fragmentation of the glycan structure attached to two glycosylation sites in g2CmeA digested by trypsin (DFNRS) (E) or chymotrypsin (DNNNS) (F). Peaks indicative of fragmentation of the N-glycans are highlighted by red arrows, whereas peptide m/z values and peptides with diNAcBac are indicated in blue.

FIG 4

CD spectra of CmeA variants in 10 mM sodium phosphate, 75 mM sodium chloride, and 10% glycerol (pH 8.0). (A) Far-UV CD spectra were collected for g0CmeA (0.124 mg/ml) and g2CmeA (0.174 mg/ml) variants using a 0.5-mm rectangular cell pathlength. The molar ellipticity was calculated and corrected for protein concentrations. Each point represents an average of collected reads made every 25 μs for 1.5 ms; the data represent the averages of three biological repeats. Thermal melts of CmeA variants for far-UV CD spectra were collected for g0CmeA (0.124 mg/ml) and g2CmeA (0.174 mg/ml) variants using a 0.5-mm rectangular cell pathlength. CD mdeg values were recorded as a function of temperature from blue (6°C) to red (94°C) for g0CmeA (B) and g2CmeA (C). Each color in between was obtained at rate 1°C/min with a 2°C stepwise increase. The reversibility of thermal unfolding was achieved recorded at 20°C, raised to T_m, and recooled to 20°C sequentially. The CD spectra were collected for 5 min at each temperature interval for g0CmeA (D) and g2CmeA (E). The CD spectra of g2CmeA stabilized after 30 min at T_m3, indicating a more resilient behavior thermal unfolding process.

FIG 5

Glycosylation enhances interactions between CmeA variants and CmeC. (A and B) SPR analysis of CM5 chip with 900 RU of g2cmeA immobilized (A) and 1,040 RU of g0cmeA immobilized (B). Association of CmeC at pH 7.4 was assessed for 2 min, and dissociation was monitored for 5 min. The concentrations of CmeC were 2-fold dilutions from 2 × 10⁻⁷ M (red) to 1.25 × 10⁻⁸ M (blue) or 2.5 × 10⁻⁸ M (green). (C and D) SPR analysis of CM5 chip at pH 6.0 with 900 RU of g2CmeA immobilized (C) and 1,040 RU of g0CmeA immobilized (D). The association of CmeC was assessed for 2 min, and dissociation was monitored for 5 min. The concentrations of CmeC were 2-fold dilutions from 2 × 10⁻⁷ M (red) to 0.6 × 10⁻⁸ M (purple).

FIG 6

Analysis of binding sites in CmeA and CmeC. (A) Amino acid alignment of signal peptide processed CmeA orthologues. Conserved amino acids are denoted by an asterisk, similar amino acids are denoted by a colon, and weak amino acid similarity is denoted by a period. The amino acid sequences were retrieved from UniProt and aligned using Clustal Omega (52). The RLS attachment site is shown to be conserved among periplasmic accessory proteins from different strains. The localization of XRLS is highlighted in a blue box, showing the presence of ¹²³N at X_–1 in the conserved RLS motif in C. jejuni and C. coli (Cj_CmeA and Cc_CmeA, respectively) but not C. fetus or C. lari (Cf_CmeA and Cl_CmeA, respectively). (B) Structural representation focusing on chain A of the CmeC trimer (PDB 4MT4). Chain A is highlighted in cyan; ³²N and ⁴⁹N are highlighted in red and blue, respectively. The proposed attachment site VGA motif is highlighted in magenta, showing its distance from both glycosylation sites. (C) Structural prediction of CmeA. The signal-processed amino acid sequence was deposited in I-TASSER, and the best structural fit was based on the MexA model. The RLS motif is highlighted in dark red; ¹²³N and ²⁷³N are highlighted in blue and light green, respectively, showing the close proximity of ¹²³N to RLS motif in CmeA. (D) Analysis of outer membrane channels. CmeC, TolC, and OrpM show the conservation of Gly structurally located at the tip region of the coiled-coil α-hairpin domain among Campylobacter species, E. coli, and P. aeruginosa.