Identification and characterization of the lipoprotein N-acyltransferase in Bacteroides

Abstract

Members of the Bacteroidota compose a large portion of the human gut microbiota, contributing to overall gut health via the degradation of various polysaccharides. This process is facilitated by lipoproteins, globular proteins anchored to the cell surface by a lipidated N-terminal cysteine. Despite their importance, lipoprotein synthesis by these bacteria is understudied. In Escherichia coli, the α-amino-linked lipid of lipoproteins is added by the lipoprotein N-acyltransferase Lnt. Herein, we have identified a protein distinct from Lnt responsible for the same process in Bacteroides, named lipoprotein N-acyltransferase in Bacteroides (Lnb). Deletion of Lnb yields cells that synthesize diacylated lipoproteins, with impacts on cell viability and morphology, growth on polysaccharides, and protein composition of membranes and outer membrane vesicles (OMVs). Our results not only challenge the accepted paradigms of lipoprotein biosynthesis in gram-negative bacteria but also suggest the existence of a new family of lipoprotein N-acyltransferases.

Keywords: Bacteroides; acyltransferase; cell surface; lipoproteins.

Fig. 1.

Identification of BF9343_0945 as the lipoprotein triacylating enzyme in Lnt-depleted E. coli. (A) DNA fragments recovered from viable Lnt-depleted cells were mapped to the B. fragilis genome. The two distinct inserts overlap on the single open reading frame BF9343_0945 (boxed and shaded). (B) Lnt-depleted KA349 was transformed with the indicated plasmids and spread on three arabinose plates and three glucose plates each. Colonies were enumerated and represented as a weighted ratio of glucose-grown to arabinose-grown colonies. (C) Whole-cell lysates of KA349 (P_BAD-lnt) grown under inducing (+) and noninducing (−) conditions, and Δlnt strains with plasmids p1, p31, and pBF9343_0945 were separated by SDS-PAGE and immunoblotted against Lpp(K58A)-Strep. Lipoprotein having three acyl chains (arrowhead marked “3”) and two acyl chains (arrowhead marked “2”) are indicated. (D) Trypsinized lipopeptides of Lpp purified from the Δlnt strain expressing BF9343_0945 were eluted from nitrocellulose and analyzed by MALDI-TOF MS. The prominent peak at m/z 1396 corresponds to the triacylated N-terminal peptide possessing acyl chains totaling 48:1 (with 48 and 1 referring to the total number of carbons and a double bond, respectively), and m/z 1424 is the same peptide with acyl chains totaling 50:1.

Fig. 2.

Evidence of lipoprotein N-acyltransferase activity by BT_4364 in B. theta Immunoblots against Lpp(K58A)-Strep (A), and BT_3736-Strep and BT_3183-Strep (B) using whole-cell lysates of B. theta WT, ΔBT_4364, and the complemented strain. Lysates from E. coli KA349 (P_BAD-lnt) grown under inducing (+) and noninducing (−) conditions were included as triacyl and diacyl controls for Lpp(K58A)-Strep. Lipoproteins having three acyl chains (arrowheads marked 3) and two acyl chains (arrowheads marked 2) are indicated.

Fig. 3.

Structural comparisons and active site residues of Lnb. (A) Side-by-side comparison of the AlphaFold-predicted structure of Lnb (Left) to the solved structure of E. coli Lnt (PDB 5xhq; Right), colored rainbow from blue (N terminus) to red (C terminus). (B) An overlay of the predicted periplasmic domain of Lnb (blue), with the AlphaFold-predicted structure of S. aureus LnsA (purple) and the solved structure of V. cholerae TseH (PDB 6v98; white). (C) Close-up of the active site of TseH (white) overlaid with Lnb (blue). (D) Whole-cell lysates of indicated B. theta strains were separated by SDS-PAGE and immunoblotted against Lpp(K58A)-Strep. Lipoprotein having three acyl chains (arrowhead marked 3) and two acyl chains (arrowhead marked 2) are indicated.

Fig. 4.

Viability, morphology, and competition of WT versus Δlnb cells. (A) The CFU/mL of WT, Δlnb, and complemented strains when grown in rich media for 7 and 24 h. Results are from two biological replicates with three technical replicates each. (B) Representative images of WT, Δlnb, and complemented cells grown in rich media for 7 and 24 h. Cells were stained with CellMask Deep Red, with images taken at 40x magnification. White arrows indicate representative cell sizes. (C) Violin plot quantifying the major axis length of WT, Δlnb, and complemented strains grown in rich media for 7 and 24 h (n = 36,692 to 81,618 objects). The average length is indicated. (D) In vitro competitions of barcoded B. theta WT and Δlnb strains in rich media. Relative abundance was calculated as the percent composition of a strain’s DNA relative to the total DNA in the sample. Data shown are representative of two biological replicates with three technical replicates each. ***P < 0.005 by the paired t test. n.s., not significant.

Fig. 5.

Growth of B. theta strains in rich or MM with various carbon sources. The OD at 600 nm (OD₆₀₀) of B. theta WT, Δlnb, and complemented strains on TY-glucose (A), MM-glucose (B), MM-amylopectin (C), and MM-arabinan (D) over time. Curves are the average of three technical replicates. An outlier well for Δlnb cells grown in MM-amylopectin was omitted from the dataset. The graphs shown are representative of three biological replicates.

Fig. 6.

Surface localization of Sus lipoproteins in B. theta. (A) Representative microscopy images of B. theta WT, Δlnb, and complemented cells when immunostained for SusG. Alexa Fluor 488 and CellMask Deep Red images were taken in the green and red channels, respectively, with equal contrast across all images within the same channel. White arrows indicate representative cells. Quantification of the integrated fluorescent intensity is graphed as a violin plot for SusG (B) and SusE (C). Averages are indicated. ***P < 0.005 by the paired t test. n.s., not significant.

Fig. 7.

Lnb mutant shows altered TM and OMV protein profiling in B. theta. (A) Coomassie blue staining after SDS-PAGE of B. theta WT and Δlnb TM and vesicle (OMV) fractions. Equivalent amounts of protein were loaded in each lane (10 μg). (B) Cartoon representation of lipoprotein comportment in B. theta WT (Left) versus Δlnb (Right) in TM and OMVs. Each shape represents approximately 3% of the total lipoproteins detected. (C) Parts-of-a-whole diagram of the same data, following the same legend as in panel B. The white wedge represents unclassified lipoproteins.