Lipoarabinomannan modification as a source of phenotypic heterogeneity in host-adapted Mycobacterium abscessus isolates

Significance

Difficult-to-treat pulmonary infections caused by Mycobacterium abscessus have been steadily increasing in the U.S. and globally. Genomic analysis of serially isolated strains from the lung of infected patients has begun to shed light on the molecular mechanisms underlying the pathoevolution of this emerging pathogen. We show that, upon lung infection, M. abscessus acquires nonsynonymous mutations in a gene required for the biosynthesis of a major cell envelope lipoglycan. These mutations lead to pleiotropic changes in the ability of M. abscessus to form serpentine cords and biofilms, replicate in innate immune cells, and induce an inflammatory response. The phenotypic diversity that ensues is expected to enhance survival in the hostile environment of the host.

Abstract

Mycobacterium abscessus is increasingly recognized as the causative agent of chronic pulmonary infections in humans. One of the genes found to be under strong evolutionary pressure during adaptation of M. abscessus to the human lung is embC which encodes an arabinosyltransferase required for the biosynthesis of the cell envelope lipoglycan, lipoarabinomannan (LAM). To assess the impact of patient-derived embC mutations on the physiology and virulence of M. abscessus, mutations were introduced in the isogenic background of M. abscessus ATCC 19977 and the resulting strains probed for phenotypic changes in a variety of in vitro and host cell-based assays relevant to infection. We show that patient-derived mutational variations in EmbC result in an unexpectedly large number of changes in the physiology of M. abscessus, and its interactions with innate immune cells. Not only did the mutants produce previously unknown forms of LAM with a truncated arabinan domain and 3-linked oligomannoside chains, they also displayed significantly altered cording, sliding motility, and biofilm-forming capacities. The mutants further differed from wild-type M. abscessus in their ability to replicate and induce inflammatory responses in human monocyte–derived macrophages and epithelial cells. The fact that different embC mutations were associated with distinct physiologic and pathogenic outcomes indicates that structural alterations in LAM caused by nonsynonymous nucleotide polymorphisms in embC may be a rapid, one-step, way for M. abscessus to generate broad-spectrum diversity beneficial to survival within the heterogeneous and constantly evolving environment of the infected human airway.

Mycobacterium abscessus (Mabs), comprising M. abscessus subsp. abscessus, M. abscessus subsp. massiliense, and M. abscessus subsp. bolletii, are considered the most virulent and most difficult to treat rapidly growing mycobacteria and, together with Mycobacterium avium complex species, are responsible for 70 to 95% of pulmonary infections caused by nontuberculous mycobacteria (NTM) (1, 2). People with structural lung diseases such as those with chronic obstructive pulmonary disease, cystic fibrosis (CF), and non-CF bronchiectasis are particularly susceptible (1–3). A large-scale population genomics study has revealed that genetically clustered Mabs isolates of increased pathogenicity have emerged in the last 50 y and are currently responsible for more than 70% of Mabs pulmonary infections in the CF population (4–6).

Mabs is an intracellular pathogen known to infect macrophages, epithelial cells, neutrophils, and endothelial cells (7–10). Mabs can also survive extracellularly in host tissues as evidenced by biofilm-like structures in the thickened alveolar walls, airways, and lung cavity of patients (11–13). While the capacity of Mabs to switch between different intracellular and extracellular lifestyles likely promotes in vivo survival and tolerance to antibiotics, relatively little is known of the specific strategies evolved by this bacterium to colonize the host and establish a long-term infection in the human lung.

Acquisition of adaptive mutations is a common theme in microbial persistence. Clinical studies with Gram-negative CF pathogens such as Pseudomonas aeruginosa and the Burkholderia cepacia complex have shown that, following initial infection with environmental isolates, specific phenotypes are selected in the CF airway. Changes undergone by these isolates in the host are many and include loss of flagellar-dependent motility, mucoidy, increased auxotrophy, increased biofilm formation and resistance to antibiotics and antimicrobial peptides, decreased secretion of virulence factors, and alterations in the structure of lipopolysaccharide (LPS) leading to immunomodulation (14, 15). By comparison, disease progression in Mabs-infected patients has thus far essentially been linked to a decrease in outer membrane glycopeptidolipid (GPL) content resulting in changes in the aggregative and biofilm-forming capacity of the bacilli and their subsequent interactions with innate immune cells (10, 16–21). Other changes potentially relevant to the pathoevolution of Mabs were only recently revealed by genomic analysis of serially isolated strains from the lung of infected CF patients (4, 22). Two of the genes found to accumulate an excess of nonsynonymous single nucleotide polymorphisms (SNPs) in the course of infection, ubiA and embC, participate in critical steps of the biosynthesis of essential polysaccharides found in the cell envelope of all mycobacteria (23). EmbC, the focus of the present study, is an arabinosyltransferase responsible for much of the linear elongation of the arabinan domain of lipoarabinomannan (LAM) (Fig. 1A). In Mycobacterium tuberculosis, where the role of LAM in infection has been better defined, this lipoglycan modulates key aspects of the host innate and adaptive immune responses. In addition to mediating the binding and entry of M. tuberculosis inside phagocytic cells through C-type lectins and modulating phagosome maturation inside macrophages, LAM and its biosynthetic precursor, lipomannan (LM), contribute to driving proinflammatory or anti-inflammatory responses depending on the fine details of their structures (24–28). The fact that mutations in embC were reported in M. abscessus subsp. abscessus, M. abscessus subsp. massiliense and M. abscessus subsp. bolletii isolates and identified in 17 different Mabs-infected CF patients across three independent studies (4, 29, 30) is further indication that mutational variations in embC are likely to confer a broad adaptive advantage to Mabs during infection. To test this hypothesis and gain insight into the changes undergone by the bacterium as a result of mutations in embC, we introduced clinically relevant embC nonsynonymous SNPs in the isogenic background of Mabs American Type Culture Collection (ATCC) 19977. We then analyzed the effects of these mutations on the biosynthesis of LAM and other phenotypes relevant to persistent infection, including drug resistance, aggregative and biofilm-forming capacity, and the ability of Mabs to establish an infection and stimulate an immune response in human monocyte–derived macrophages and epithelial cells.

Fig. 1.

Schematic representation of the structure of Mabs LAM and analysis of EmbC mutations identified in host-adapted Mabs isolates. (A) Proposed structure of Mabs LAM showing acetyl and succinyl substituents on the Ara₄ and Ara₆ arabinan termini and the tentative location and structure of 3-linked mannoside chains on the mannan domain. Ara₅ motifs, not shown in this figure, are thought to be extended linear Ara₄ motifs harboring one additional Araf residue (31). (B) Mapping of the structural locations of the wild-type residues corresponding to the 15 patient-derived Mabs EmbC mutations. The predicted AlphaFold model is shown as cartoon. Individual domains corresponding to the N-terminal periplasmic domain (PN), transmembrane domain (TM) and C-terminal periplasmic domain (PC) are colored red, yellow, and cyan, respectively. The location of ethambutol inferred from the structural alignment of the crystal structure from M. smegmatis (PDB 7VBE) is shown as ball and stick representation along with the transparent surface representation. The regions predicted to be disordered are shown in magenta. (C) SDM and FoldX mutant stability prediction. SDM scores are unitless whereas the FoldX scores are given in kcal/mol. Positive and negative values correspond to stabilizing effects in SDM and FoldX, respectively, and vice versa. (D) Same predicted AlphaFold model as in panel (B) but the cartoon representation is differentially colored based on the allosteric coupling intensity values calculated using Ohm. Its corresponding color key is shown at the Bottom. Higher positive values of ACI (shown in red) correspond to potential allosteric hotspots linked to the ligand binding site.

Results

Mutational Variations in the embC Gene of Mabs Isolates Recovered from the Lung of Long-Term Infected Patients.

Whole genome sequencing of isolates from 201 CF patients (5) with longitudinal samples identified embC (MAB_0189c) as one of the genes accumulating more nonsynonymous SNPs than would be expected by chance in as many as 15 patients (4). A screening for other nonsynonymous SNPs potentially impacting LAM biosynthesis identified ubiA as the only other gene from this pathway in which mutations occurred at a higher rate than would be expected by chance (4). embC mutations were present in all three Mabs subspecies [M. abscessus subsp. abscessus (n = 7), M. abscessus subsp. massiliense (n = 2), and M. abscessus subsp. bolletii (n = 6)]. The fact that nonsynonymous SNPs occurred at a significantly lower rate in longitudinal Mabs isolates from laparoscopy-associated wound infections suggests that embC mutations may specifically be beneficial to survival in the lung environment (4).

Mapping of the 15 nonsynonymous SNPs (D192N, Y222C, M310T, I402V, L404R, A444V, D514N, T526A, A605T, F646L, G647S, R921P, D1046G, R1082H, and S1084R) on the predicted AlphaFold model of EmbC from Mabs indicated that they impacted all three structural regions of the enzyme, namely the transmembrane domain (TM) (I402, L404, A444, D514, T526, A605, F646, G647), the N-terminal periplasmic domain (PN) (D192, Y222, M310) and the C-terminal periplasmic domain (PC) (R921, D1046, R1082, S1084) (32) (Fig. 1B). Several of the mutated residues mapped at the interface of two or more domains, including Y222 at the interface of all three domains, F646 and G647 at the interface of PC and TM, and R1082, S1084, and D192 at the interface of PN and PC (Fig. 1B and SI Appendix, Fig. S1). Additionally, TM residues D514 and T526 are relatively solvent exposed and, thus, close to the periplasmic surface of the protein. While the PC domain contains carbohydrate-binding modules important for activity (33–35), the PN domain linking TM1 and TM2 was shown to adopt the characteristic fold of polysaccharide-binding units. The TM domain participates in the formation of an active site pocket at the interface with PC and periplasmic loops 2 to 6 which contains the binding sites for the donor and acceptor substrates and the catalytically relevant DDX motif (32, 34). SDM and FoldX were used to predict the impact of mutations on protein stability. For 13 out of 15 mutations, both programs predicted the mutations to destabilize the protein structure (Fig. 1C). Disagreement between SDM and FoldX was only seen in five cases (A444V, D1046G, D192N, D514N, and R1082H). A summary of other structural properties of the mutated residues, including relative solvent accessibility, residue depth, residue packing density, and distance to the substrate binding site is provided in SI Appendix, Table S1. The predicted impacts of mutations on surrounding hydrogen and cation pi interactions are shown in SI Appendix, Fig. S1. Of all mutated residues, M310 is the closest to the substrate binding site (symbolized in Fig. 1B as the binding site of ethambutol which overlaps with that of the arabinose donor and acceptor substrates) (32), at a distance of 13.80 Å. Importantly, all four TM residues, I402, L404, A444, and A605, were predicted to be destabilizing or neutral, and all were predicted to be allosterically coupled with the active site, implying a deleterious impact of the mutations on arabinoside substrate binding (Fig. 1D). None of the mutations were predicted to impact the dimerization of EmbC or to affect the binding affinity with AcpM (32).

Generation of Isogenic Mutants of Mabs ATCC 19977 Expressing Patient-Derived embC Mutated Variants.

To study the impact of patient-derived embC mutations on the physiology and virulence of Mabs, isogenic strains expressing different mutated forms of embC were generated by recombineering in the background of Mabs ATCC 19977 rough. While mutational embC variants have been reported in both smooth (S) and rough (R) variants of Mabs (29, 30), a rough morphotype parent was selected for its relevance to persistent infection (18, 36) and likely enhanced surface exposure of lipoglycans in the absence of GPL in the outer membrane (20). Six different mutants (EmbC^Y222C, EmbC^M310T, EmbC^I402V, EmbC^D514N, EmbC^F646L, and EmbC^D1046G) spanning the different regions of EmbC affected by the nonsynonymous SNPs were generated. In these strains, the endogenous wild-type (WT) copy of embC is deleted and replaced by mutated versions of the gene expressed from an integrative plasmid (pMV306-xylE) under control of the constitutive Phsp60 promoter. An isogenic strain expressing a WT copy of embC instead of a mutated variant (MabsΔembC::pMV306H-embC^WT) was also generated as WT control. Finally, a Mabs embC null mutant containing an empty plasmid (MabsΔembC::pMV306-xylE) was generated as a strain totally deficient in EmbC activity (SI Appendix, Fig. S2). All recombinant strains were successfully achieved indicating that, similar to the situation in other rapidly growing mycobacteria (Mycobacterium smegmatis; Mycobacterium neoaurum, and Mabs subsp. abscessus clinical isolate GZ002) (37–39), but in contrast to that in M. tuberculosis (40), embC is not required for Mabs ATCC 19977 growth under axenic conditions.

Patient-Derived embC Mutations Do Not Impact the Antibiotic Susceptibility of Mabs.

Because chronically infected patients from which Mabs embC mutants were isolated typically underwent long courses of antibiotic treatment, we first determined whether nonsynonymous SNPs in embC affected antibiotic susceptibility. In line with a recent report (37), the embC null mutant proved to be 4- to 16-fold more susceptible to a number of antibiotics (clarithromycin, levofloxacin, vancomycin, linezolid, cefoxitin, rifampin and rifabutin) in 7H9-ADC-Tween 80 compared to the control strains, Mabs ATCC 19977 and MabsΔembC::pMV306H-embC^WT (Table 1). For most drugs, this increase in susceptibility was even more pronounced in cation-adjusted Mueller Hinton II broth recommended by the Clinical and Laboratory Standards Institute for MIC determinations in NTM (SI Appendix, Table S2). Whether in 7H9-ADC-Tween 80 or in Mueller Hinton II, the MIC values of standard-of-care antibiotics (clarithromycin, azithromycin, linezolid, cefoxitin, tigecycline, rifampin and rifabutin) against the six patient-derived mutants, EmbC^Y222C, EmbC^M310T, EmbC^I402V, EmbC^D514N, EmbC^F646L and EmbC^D1046G, did not differ more than two-fold from that of the control strains (Table 1 and SI Appendix, Table S2). We thus conclude that antibiotic resistance was not the primary driver for the selection of embC mutants during infection.

Table 1.

Susceptibility to antibiotics of Mabs ATCC 19977 expressing different variants of embC

Drugs	ATCC 19977	WT	Y222C	M310T	I402V	D514N	F646L	D1046G	ΔembC
Clarithromycin	32	16	16	16	16	16	16	16	4
Levofloxacin	4	4	2	2	2	4	2	4	1
Vancomycin	16	16	8	4	4	16	16	16	1
Linezolid	8	4	8	2	2	2	2	2	1
Cefoxitin	16	8	8	8	8	8	16	8	4
Rifampin	16	8	16	4	4	8	8	4	1
Tigecycline	2	2	2	1	2	2	2	1	1
Rifabutin	2	2	2	2	2	2	2	2	0.5
Azithromycin	64	128	128	128	128	128	128	128	64
Bedaquiline	1	1	2	2	2	1	2	1	1

MIC values (in μg/mL) were determined in 7H9-ADC-Tween 80. Mabs ATCC 19977 (rough morphotype) is the parent Mabs strain. ΔembC is the embC knockout mutant. Other Mabs strains have had their endogenous embC gene deleted and replaced by either a wild-type (“WT”) or mutated version of the gene expressed from an integrative plasmid under control of the Phsp60 promoter.

Effect of Clinically Relevant embC Mutations on LAM Biosynthesis.

The potential impact of patient-derived nonsynonymous SNPs on the arabinosyltransferase activity of EmbC was assessed by purifying the LAM from the embC mutants and comparing their structure to those purified from the control strains expressing WT embC genes (Mabs ATCC 19977 and MabsΔembC::pMV306H-embC^WT). SDS-PAGE analysis of the purified LAM from the different strains pointed to the faster migration of the material purified from EmbC^M310T, EmbC^I402V, EmbC^D514N, and EmbC^D1046G, indicative of changes in size and/or charge (Fig. 2A). The embC knockout mutant, which is expected to produce a “LAM” devoid of arabinan domain (39), produced the smallest lipoglycan of all strains. Normal size LAM was restored upon complementation with pMV306H-embC^WT (WT; second lane) (Fig. 2A).

Fig. 2.

Electrophoretic mobility of LAM from Mabs ATCC 19977, the embC null mutant, and the recombinant Mabs ATCC 19977 strains expressing different mutated variants of embC. (A) Purified LAMs from the different strains were run on a 10 to 20% Tricine gel followed by periodic acid-silver staining. The results presented are representative of three separate lipoglycan extractions and SDS-PAGE analyses. MWM, molecular weight marker. (B) Araf/Manp ratio of the purified LAMs from the different strains derived from the glycosyl linkage analysis presented in Table 2. Shown on the graph are the means ± SD of three analytical replicates. Asterisks denote statistically significant differences between the control strain, MabsΔembC::pMV306H-embC^WT, and the embC mutants (****P < 0.0001; ***P < 0.0005; **P < 0.005; *P < 0.05; ordinary one-way ANOVA, Dunnett’s multiple comparison). (C) Gas chromatograms showing the presence of 3-linked mannose residues in the Mabs ATCC 19977 parent strain, the control WT strain MabsΔembC::pMV306H-embC^WT (WT), the EmbC^M310T mutant and the embC knockout mutant, comigrating with the authentic standard, 2,4,6-tri-O-methylD-mannitol acetate.

Quantitative analyses of per-O-methylated alditol acetate derivatives of the control and mutant LAMs were undertaken to gain insight into their Manp and Araf content and the degree of branching of their mannan and arabinan domains. Compared to the control strains, significant decreases in the Araf to Manp ratios of LAM from four embC mutants were notable, with EmbC^M310T displaying the lowest ratio, followed by EmbC^D1046G, EmbC^D514N, and EmbC^F646L (Fig. 2B). Although the Araf to Manp ratio of LAM prepared from EmbC^I402V was also slightly lower than that measured in the control strain, MabsΔembC::pMV306H-embC^WT, the difference was not statistically significant. As expected, the embC null mutant was essentially devoid of Araf (Fig. 2B). Glycosyl linkage analyses confirmed the absence of an arabinan domain in the null mutant and, consistent with Araf to Manp ratios, pointed to important alterations in the structure of the arabinan domain of the EmbC^M310T, EmbC^I402V, EmbC^D514N, EmbC^F646L, and EmbC^D1046G mutant LAMs. Compared to the controls, all five mutant LAMs presented a significant reduction in the relative percentage of 5-linked Araf indicative of reduced EmbC activity (Table 2). EmbC^M310T further showed a significant reduction in the relative percentage of 3,5-linked Araf indicative of a less branched arabinan domain. A higher relative percentage of 3,6-linked Manp was also observed in the LAM prepared from EmbC^M310T.

Table 2.

Glycosyl linkage analysis of per-O-methylated LAM

Strain	t-Ara	2-Ara	5-Ara	t-Man	3,5-Ara	3-Man	6-Man	3,6-Man
ATCC19977	13.34 ± 0.9	6.20 ± 0.4	37.50 ± 0.96	10.01 ± 0.5	13.60 ± 0.4	2.78 ± 0.5	8.7 ± 0.7	7.74 ± 0.7
WT	10.17 ± 2.62	5.62 ± 0.66	42.26 ± 2.02	9.21 ± 2.13	13.25 ± 0.88	2.39 ± 0.31	8.78 ± 0.80	8.26 ± 0.14
ΔembC	1.71 ± 0.11*	1.30 ± 0.27*	2.56 ± 0.06*	32.41 ± 0.7*	0.66 ± 0.12*	14.39 ± 0.7*	19.11 ± 1.2*	27.85 ± 1.4*
Y222C	12.33 ± 0.7	6.29 ± 1.5	38.43 ± 1.5	9.65 ± 0.8	12.39 ± 0.5	3.14 ± 0.5	7.75 ± 0.6	10 ± 1.2
M310T	10.16 ± 0.82	6.38 ± 3.4	26.46 ± 0.4*	10.08 ± 0.1	10.44 ± 0.6*	16.05 ± 1.9*	6.097 ± 0.6	14.38 ± 0.2*
I402V	11.27 ± 0.4	6.13 ± 1.5	35.70 ± 2.8*	12.18 ± 0.1*	13.99 ± 0.8	4.39 ± 0.3	8.75 ± 1.7	8.16 ± 0.9
D514N	10.76 ± 0.21	6.11 ± 0.23	31.95 ± 0.9*	11.38 ± 0.3	13.47 ± 0.5	4.5 ± 0.2	11.94 ± 1.4	8.82 ± 0.5
F646L	12.39 ± 2.1	6.61 ± 0.37	33.93 ± 2.7*	9.76 ± 0.3	12.34 ± 0.7	10.57 ± 1.7*	7.48 ± 1.2	6.91 ± 0.7
D1046G	11.08 ± 0.29	5.52 ± 0.65	29.9 ± 2.9*	11.75 ± 0.4	12.93 ± 0.6	10.76 ± 1.8*	9.6 ± 0.5	9.58 ± 0.3

Reported values are averages ± SD of three technical repeats and represent relative distribution in %. Asterisks denote statistically significant differences between the WT (MabsΔembC::pMV306H-embC^WT) and mutant LAMs (*P < 0.05; ordinary one-way ANOVA, Dunnett’s multiple comparison).

Intriguingly, a predominant peak appeared on the chromatograms of a subset of mutants (EmbC^M310T, EmbC^F646L, and EmbC^D1046G, and the embC null mutant) whose gas chromatography/mass spectrometry (GC/MS) fragmentation pattern was suggestive of 3-linked Manp (Table 2 and Fig. 2C). Although detectable in other strains, this peak was ~2.4 to 6.7-fold less abundant (Table 2). Comigration with an authentic, chemically synthesized, 2,4,6-trimethyl-mannitol acetate confirmed its identity as 3-linked Manp. This result points to the existence in Mabs LAM of previously undetected α-(1→3)-linked oligomannoside chains. Given that this substituent was found in abundance on the arabinan-deficient lipoglycan produced by the embC knockout mutant, it is reasonable to assume that these α-(1→3)-linked oligomannoside(s) is(are) attached to the mannan core of LAM.

Analysis of the nonreducing arabinan termini of LAM upon Cellulomonas gelida endoarabinanase digestion also revealed important differences between strains in that four mutants (EmbC^M310T, EmbC^I402V, EmbC^D514N, and EmbC^D1046G) presented considerably less Ara₄ termini and more Ara₆ termini compared to the control strain, MabsΔembC::pMV306H-embC^WT (Table 3). This decrease in Ara₄ relative to Ara₆ termini is reminiscent of changes observed in M. smegmatis mutants expressing truncated forms of EmbC (35). Detailed analysis of the Ara₄, Ara₅ and Ara₆ termini of Mabs LAM for the presence of acetyl and succinyl substituents further showed a complete absence of succinates on the Ara₄ and Ara₆ termini of five mutants (EmbC^M310T, EmbC^I402V, EmbC^D514N, EmbC^F646L and EmbC^D1046G) with residual succinylation only being found on the Ara₅ termini of mutants EmbC^D514N, EmbC^F646L, and EmbC^D1046G (Table 3). Furthermore, no unmodified nonreducing Ara₄ termini and fewer unmodified Ara₆ termini were present in mutants EmbC^M310T and EmbC^I402V, which instead showed more acetylation of both oligoarabinosides. In contrast, more unmodified nonreducing Ara₄ termini were found in mutants EmbC^D514N, EmbC^F646L, and EmbC^D1046G.

Table 3.

LC/MS analysis of the unmodified and covalently modified oligoarabinosides released from the LAM of Mabs ATCC 19977 and the Mabs ATCC 19977 strains expressing different WT or mutated variants of embC upon C. gelida endoarabinanase digestion

	ATCC19977	WT	Y222C	M310T	I402V	D514N	F646L	D1046G
Total Ara4	35.2	31.7	31	19.6	14.4	16.9	32.3	23.2
Unmodified Ara4	35.6	29.7	37	0	0	41.5	55.6	47.4
Ara4+succinate	0	4.9	0	0	0	0	0	0
Ara4+acetate	61.5	61.4	60.3	100	100	58.5	44.4	52.6
Ara4+succinate+acetate	2.9	4.0	2.8	0	0	0	0	0
Total Ara6	38.1	40.1	41.7	56.3	59.6	54.7	40.9	45.9
Unmodified Ara6	88.0	97.6	98.6	75.4	78.5	87.0	96.3	100
Ara6+succinate	0.6	2.1	1.2	0	0	0	0	0
Ara6+acetate	11.4	0.7	0.5	24.6	21.5	13	3.7	0
Total Ara5	26.7	28.2	27.4	24.0	26	28.4	26.8	30.9
Unmodified Ara5	65.3	64.5	69.5	86.3	72.4	87.9	73.8	80.3
Ara5+succinate	6.4	6.0	5.3	0	0	3.0	4.3	2.0
Ara5+acetate	23.6	23.9	21.3	13.7	27.6	9.2	21.9	17.7
Ara5+succinate+acetate	4.7	5.6	3.8	0	0	0	0	0
Ara4 + Ara5 + Ara6	100	100	100	100	100	100	100	100

Shown in bold letters are the relative percentages of total (including, unmodified and covalently modified) Ara₄, Ara₅ and Ara₆ oligoarabinosides released upon C. gelida endoarabinanase digestion of LAM from the different strains and the individual representation (expressed as percentages) of unmodified and covalently modified oligoarabinosides within each group.

In conclusion, all patient-derived embC mutants, with the exception of EmbC^Y222C, produced modified forms of LAM whose arabinan domain was truncated to various degrees. Structural alterations in the arabinan domain which extended to its nonreducing termini were accompanied in the case of three patient-derived mutants and the embC knockout by an increased abundance of α-(1→3)-linked oligomannoside chain(s) on the mannan backbone. While some structural changes in the mutant LAMs (e.g., decreased Araf/Manp ratio, reduced relative percentage of 5-linked Araf; decrease in Ara₄ to Ara₆ termini) are likely attributable to the decreased EmbC activity of the bacilli, other changes (e.g., formation of α-(1→3)-linked oligomannoside chains; increased or decreased acetylation and succinylation of LAM) are suggestive of alterations in the activity of other LAM biosynthetic enzymes. Since EmbC belongs to a multiprotein complex which includes other arabinosyltransferases and AcpM among other protein partners (32, 41), it is indeed possible that changes in the structure and stability of the mutated forms of EmbC alter their ability to productively interact with other enzymes, resulting in a diversity of (SNP-specific) changes in LAM structure. The fact that disrupting EmbC-AcpM interactions negatively impacts the elongation of the arabinan domain of LAM clearly illustrates the importance of protein interactions involving EmbC in the biosynthesis of LAM (32). Alternatively or in addition, changes in the degree of acetylation/succinylation of LAM and abundance of α-(1→3)-linked oligomannosides could result from the activation of compensatory metabolic pathways when the activity of EmbC and/or protein partners is decreased.

Mutational Variations in embC Are Associated with Alterations in Cording, Sliding Motility, and Biofilm Formation.

Having established that patient-derived mutations have an impact on the arabinosyltransferase activity of EmbC and LAM structure, we next proceeded to compare the various mutants to the control strains expressing a WT version of the embC gene in a variety of in vitro assays aimed at determining how clinically relevant changes in the structure of LAM might affect the growth, morphotype, biofilm-forming capacity, and aggregative properties of Mabs.

The growth rates of the different strains at 37 °C were compared both in standard laboratory medium (7H9-ADC-Tween 80) and in Synthetic CF Medium (SCFM), which we found to be a better mimic of the composition of the CF airway (42). With the exception of the embC knockout mutant whose growth was slightly delayed in SCFM, no notable differences in growth were seen between the control and the patient-derived mutant strains (SI Appendix, Fig. S3). Likewise, the surface hydrophobicity of the mutants harboring nonsynonymous SNPs in embC, as assessed by Congo Red binding, was comparable to that of the control strains (SI Appendix, Fig. S4). When plated on 7H11-OADC agar, the colonial morphology of all strains was also comparable except for the EmbC^M310T and EmbC^I402V mutants whose colonies appeared less wrinkled (but were clearly not smooth when compared to the parent Mabs ATCC 19977 smooth morphotype strain) and the embC knockout mutant whose colonies were even smoother, but not dry, with translucent edges (Fig. 3A).

Fig. 3.

Impact of patient-derived embC mutations on the morphology, sliding motility, and ability of Mabs ATCC 19977 to form biofilms and serpentine cords. (A) Colony morphology of the control and embC mutant strains on 7H11-OADC agar plates after 5 d of incubation at 37 °C. Mabs ATCC 19977 R is the rough morphotype parent strain used throughout this study (and in all other figures). Mabs ATCC 19977 S is the reference smooth morphotype strain shown here for comparison. (B) Sliding motility of the control and embC mutant strains on agar medium. Mabs strains were drop-inoculated from liquid cultures diluted to 10⁶ CFU/mL onto 7H9-ADC medium containing 0.34% agar and incubated at 37 °C for 5 d at which point the diameters of spread of the different strains were compared. (C) Biofilm formation in SCFM by the different strains was analyzed after 5 d of incubation at 37 °C by crystal violet staining as described previously (43). Shown are the means ± SD of absorbances measured at 562 nm for six biological replicates. Asterisks denote statistically significant differences between the control strain, MabsΔembC::pMV306H-embC^WT, and the embC mutants (****P < 0.0001; ***P < 0.001; ordinary one-way ANOVA; Dunnett’s multiple comparison). (D) The control and mutant strains were inoculated in Tryptic Soy broth at a concentration of 10⁴ CFU/mL and their ability to form serpentine cords was compared after 3 d of incubation at 37 °C. All assays were performed as described under Materials and Methods and the results presented are representative of two to three independent experiments.

When compared for sliding motility, four mutants stood out as consistently spreading less than the control strains on the surface of 7H9-agar plates: EmbC^M310T, EmbC^I402V, EmbC^D514N, and EmbC^D1046G (Fig. 3B). Two of these mutants, EmbC^M310T and EmbC^I402V, were further found to be more proficient at forming biofilms in SCFM, as was the embC knockout mutant (Fig. 3C). Strikingly, the same two patient-derived mutants displayed a notable cording defect in tryptic soy broth (Fig. 3D), with EmbC^I402V displaying the greatest defect of the two.

Incidentally, the EmbC^M310T and EmbC^I402V mutants are, with the embC knockout, the only three mutants to be devoid of succinylated arabinan termini (Table 3), suggestive of an impact of LAM succinylation on the biofilm-forming capacity, colony morphology and, perhaps, cording and sliding motility of rough Mabs strains. Of note, M. avium, Mycobacterium marinum, M. smegmatis, and M. abscessus ATCC 19977 (smooth) strains deficient in LAM and arabinogalactan (AG) succinylation, were reported to display various, species-specific, alterations in aggregative properties, colony morphology, surface rigidity, surface hydrophobicity, and biofilm formation (44–47). It was proposed that the succinylation of the two major cell envelope polysaccharides of mycobacteria serves to modulate, most likely through indirect charge-mediated effects, the cell surface properties of the bacilli (46). To determine whether a defect in LAM succinylation might account for the altered phenotypic traits observed in some patient-derived EmbC mutants, a sucT (MAB_2689) mutant was generated by allelic replacement in the Mabs ATCC 19977 rough background, yielding Mabs^RΔsucT (Fig. 4A). We previously established that MAB_2689 (SucT) was the succinyltransferase responsible for the succinylation of LAM and AG in Mabs ATCC 19977 smooth (46). Compared to the WT parent strain, Mabs^RΔsucT formed significantly more biofilms in SCFM (Fig. 4B) and displayed colonies on 7H11-OADC that were less wrinkled with translucid edges, more comparable to those formed by the EmbC^M310T mutant (Figs. 3A and 4C). The WT strain and the sucT KO, in contrast, did not differ in their cording properties or sliding motility (Fig. 4 D and E). These observations implicate the loss of succinyl substituents as a contributing driver of at least some of the altered phenotypes of the EmbC^M310T, EmbC^I402V, and embC knockout mutants. The differences in in vitro phenotypes displayed by Mabs^RΔsucT and the patient-derived embC mutants may tentatively be explained by the fact that Mabs^RΔsucT is totally deficient in the succinylation of both LAM and AG, while embC mutants are expected to only lack succinyl substituents on LAM as a result of altered interactions between EmbC and SucT. Compared to Mabs^RΔsucT, EmbC mutants are also expected to display different changes in their surface lipid, polysaccharide, or protein composition as a result of additional EmbC-related alterations in LAM structure impacting cell envelope organization.

Fig. 4.

Phenotypic properties of an Mabs ATCC 19977 rough sucT knockout mutant. (A) Allelic replacement at the sucT locus of Mabs ATCC 19977 (rough). Genomic DNA was extracted from the WT Mabs ATCC 19977 rough strain and knock-out (ΔsucT) mutant and allelic replacement in the mutant was confirmed by PCR as detailed in Materials and Methods. The expected sizes of the PCR products are 2,474 bp and 3,415 bp in the WT strain and knockout mutant, respectively. The WT and mutant strains were compared for their ability to form biofilms in SCFM (B), colony morphology (C), ability to form serpentine cords in Tryptic Soy broth (D), and sliding motility on 7H9-ADC agar plates (E) as described in Fig. 3. The M310T and I402V EmbC mutants were included in the biofilm assay for comparison. Shown on the graph are the averages ± SD of crystal violet absorbances measured at 562 nm for four biological replicates. Asterisks denote statistically significant differences between WT strain and mutant strains (****P < 0.0001; **P < 0.001; *P < 0.01; ordinary one-way ANOVA; Dunnett’s multiple comparison).

In search of potential differences between the cell surface composition of the controls and EmbC mutants, the surface lipids from the different strains were extracted and compared. α and α′ mycolates are with surface trehalose polyphleates (TPP) and GPLs the only three cell envelope constituents proposed thus far to impact the ability of Mabs to form serpentine cords (18, 48, 49). While control and mutant strains presented very similar mycolate, TPP and GPL contents, some variations were noted at the level of two unknown surface glycolipids (annotated as X and Y on SI Appendix, Fig. S5). Compared to the isogenic control strain, MabsΔembC::pMV306H-embC^WT, EmbC^M310T and EmbC^I402V presented less glycolipids X and Y at their surface, whereas EmbC^F646L had less X but more Y, and EmbC^D1046G had less glycolipid Y. To what extent these cell surface glycolipids contribute to the cording, sliding motility and biofilm-forming capacity of Mabs remains to be determined. Nevertheless, an important conclusion that may be drawn from our analyses of embC and sucT mutants is that the controlled elongation and succinylation of the arabinan domain of LAM is a strategy used by Mabs to modulate its cell surface composition and properties in response to host stresses.

Impact of embC Mutations on the Interactions of Mabs with Toll-Like Receptor 2 (TLR2).

Mononuclear phagocytes and respiratory epithelial cells lining the lung airways play a critical role in the surveillance and innate immune responses to pulmonary pathogens. Shin et al. (50) reported that, in murine macrophages infected with Mabs, there is a transient interaction between the pattern recognition receptors TLR2 and dectin-1, resulting in the induction of host protective cytokines, such as TNFα, IL-6, and p40 subunit of IL-12, and enhanced phagocytosis. In both murine macrophages and primary human monocytes, Mabs induction of TNFα is dependent on TLR2 (20, 21, 51). Respiratory epithelial cells, on the other hand, respond to Mabs infection with expression of IL-8 and human β-defensin (HβD2) in a TLR2-dependent manner (7). Mycobacterial LM is a known TLR2 agonist (24). Comparatively, LAM is a weaker agonist owing to its arabinan domain masking the bioactive mannan core (52–54). Since truncations in the LAM arabinan domain of some of the patient-derived Mabs embC mutants could increase the accessibility of their mannan domain to pattern recognition receptors and/or indirectly alter the outer membrane organization and surface topology of Mabs and, thus, accessibility of other TLR2 ligands, we compared the relative ability of the control strains and a subset of embC mutants to stimulate HEK293 cells stably transfected with human TLR2 and CD14 genes (HEK-TLR2 cells) and a NF-κB-inducible reporter system (secreted alkaline phosphatase). The mutants selected for this study included EmbC^M310T, EmbC^I402V, EmbC^D1046G, and the embC knockout mutant, whose LAM structures and in vitro phenotypic traits differed most markedly from the controls. The results indicated that the EmbC^M310T mutant was consistently the least potent activator of TLR2, while the embC knockout mutant activated TLR2 the most, and the EmbC^I402V mutant activated TLR2 slightly but significantly more than the MabsΔembC::pMV306H-embC^WT control. The activity of the EmbC^D1046G mutant did not significantly differ from that of the control strain (Fig. 5A).

Fig. 5.

NF-κB activation in HEK-TLR2 cells by the Mabs ATCC 19977 control and mutant embC strains and surface extracts derived thereof. (A) The NF-κB activity of HEK-TLR2 cells stimulated with the control and a subset of embC mutant strains (MOI of 1) at 37 °C for 16 h was determined by reading the absorbance of the medium at 650 nm. Shown are the means ± SD of absorbances measured at 650 nm for three biological replicates. (B) NF-κB activity of HEK-TLR2 cells stimulated with TSE, either untreated or treated with papain, prepared from the MabsΔembC::pMV306H-embC^WT control strain and three patient-derived EmbC mutants. Shown are the means ± SD of absorbances measured at 650 nm for three biological replicates. Asterisks denote statistically significant differences between the control strain, MabsΔembC::pMV306H-embC^WT (or TSE derived thereof), and the embC mutants (or TSE derived thereof) (****P < 0.0001; **P < 0.01; *P < 0.05; ordinary one-way ANOVA; Dunnett’s multiple comparison). The results presented in (A) and (B) are representative of two to three independent experiments.

Cell surface proteins, and lipoproteins in particular, were reported to play a predominant role in the TLR2 agonist activity of rough and smooth Mabs ATCC 19977 (21). To determine whether quantitative and/or qualitative changes in surface protein content might account for the differential ability of the EmbC^M310T and EmbC^I402V mutants to activate TLR2, surface extracts were prepared from these mutants, the EmbC^D1046G mutant, and the MabsΔembC::pMV306H-embC^WT control strain, and compared for their ability to stimulate TLR2 before and after treatment with protease (papain). The same volume of surface extracts prepared from the same amount of bacilli for each strain (as measured by dry cell weight) were added to the cells to account both for potential qualitative and quantitative differences in protein contents across strains. Fig. 5B shows that the ability of the different surface extracts to stimulate HEK-TLR2 cells followed the trends reported in Fig. 5A for the intact bacilli. Compared to surface extracts prepared from the control strain, TLR2 stimulation was significantly more pronounced with extracts prepared from the EmbC^I402V mutant and significantly less pronounced with extracts prepared from EmbC^M310T. Statistically significant differences between surface extracts were abolished upon treatment with papain (Fig. 5B). We conclude from this experiment that alterations in the surface protein content of the EmbC^M310T and EmbC^I402V mutants relative to the control account for their different ability to activate TLR2.

Impact of embC Mutations on the Interactions of Mabs with Human Monocyte–Derived Macrophages and Epithelial Cells.

To further investigate the consequences of mutations in embC on the inflammatory activity and virulence of Mabs, we next compared the control and mutant strains for intracellular replication and release of cytokines and chemokines by human monocyte–derived THP-1 cells and A549 epithelial cells, and for activation of THP-1 cells. Experiments were conducted with the same subset of mutants as in Fig. 5.

THP-1 cells were infected at an multiplicity of infection (MOI) of 1. Viability, as assessed by propidium iodide staining, was comparable for all infecting bacterial strains and was >99% throughout the course of infection. Determining intracellular CFUs 2 and 48 h postinfection showed that the intracellular replication rates of the EmbC^I402V and embC knockout mutants were significantly less than that of the control strains expressing WT embC genes (Fig. 6A). Compared to THP-1 cells infected with other strains, EmbC^D1046G- and embC knockout-infected cells displayed significant increases in human antigen presentation/processing activation markers CD80 and CD40. embC knockout-infected cells also displayed a significant increase in CD86 expression, while THP-1 cells infected with EmbC^M310T displayed, on the contrary, reduced CD86 and CD40 expression compared to the MabsΔembC::pMV306H-embC^WT control (Fig. 6B). Consistent with the reduced activation of EmbC^M310T -infected THP-1s, these cells released significantly less of the chemokines RANTES and MCP-1 and also secreted significantly less proinflammatory cytokines (IFNγ, TNFα, IL-8, IL-1β and IL-12p40) compared to cells infected with the control. EmbC^M310T-infected cells also released less anti-inflammatory cytokine IL-10 (Fig. 6C). Reflective of their increased activation state, EmbC^D1046G- and embC knockout-infected cells released, on the contrary, significantly more RANTES, IL-1β and TNFα than the cells infected with MabsΔembC::pMV306H-embC^WT. The embC knockout mutant further induced increased secretion of IL-12p40, IL-10, and MCP-1. EmbC^I402V -infected cells also released significantly more TNFα than the control but less IL-1β (Fig. 6C).

Fig. 6.

Intracellular replication, immune activation, and chemokine/cytokine secretion induced by the control and embC mutant strains in monocyte-derived THP-1 macrophages. THP-1 cells were infected at an MOI of 1. (A) Two and 48 h post infection, the macrophages were lysed, and intracellular bacteria enumerated by CFU plating. Shown is the fold-change in CFUs for each strain between 48 h and 2 h. (B) Levels of expression of activation markers for HLADR, CD80, CD86, and CD40 48 h postinfection was determined by flow cytometry. (C) Culture supernatants were analyzed 48 h postinfection for chemokine/cytokine secretion by multiplex immunoassay using Luminex. The results presented are the means ± SD of triplicate wells from one experiment and are representative of three independent experiments. Asterisks denote statistically significant differences between the control strain, MabsΔembC::pMV306H-embC^WT, and the embC mutants (****P < 0.0001; ***P < 0.0005; **P < 0.005; ordinary one-way ANOVA; Dunnett’s multiple comparison).

Infection of A549 epithelial cells with the same strains reproducibly revealed a slightly but significantly higher rate of replication of mutants EmbC^M310T and EmbC^I402V relative to the control, MabsΔembC::pMV306H-embC^WT, after 48 h of infection. In contrast, reduced replication was observed for the embC knockout mutant (Fig. 7). All infected cells, however, behaved similarly with regard to IL-8 secretion (SI Appendix, Fig. S6).

Fig. 7.

Intracellular replication of embC mutants in human A549 epithelial cells. A549 lung alveolar type II epithelial cells were infected at an MOI of 1. Two and 48 h post infection, the cells were lysed and intracellular bacteria enumerated by CFU plating. The results presented are the means ± SD of triplicate wells from one experiment and are representative of two independent experiments. Asterisks denote statistically significant differences between the control strain, MabsΔembC::pMV306H-embC^WT, and the embC mutants (****P < 0.0001; ***P < 0.0006; **P < 0.005; ordinary one-way ANOVA; Dunnett’s multiple comparison).

Collectively, our results point to the variable outcome of embC mutations on the virulence and immunomodulatory properties of Mabs and to the important, direct or indirect, role of LAM in modulating these activities. While the embC knockout mutant consistently displayed a virulence attenuation phenotype in both cellular models used in this study, was a more potent TLR2 agonist and stimulated the production of key proinflammatory cytokines in THP-1 cells, EmbC^I402V, which also replicated less extensively in THP-1 cells but slightly more in A549 epithelial cells, similarly displayed increased TLR2 agonist activity and induced more TNFα secretion in macrophages, but was a less potent inducer of IL-1 than the control strain. EmbC^M310T, which presented no signs of virulence attenuation in THP-1 cells and slightly increased replication in A549 epithelial cells, displayed, on the contrary, a less pronounced inflammatory phenotype in THP-1 macrophages in line with its reduced ability to activate THP-1 cells and less potent TLR2 agonist activity. Finally, EmbC^D1046G, which replicated similarly to the control strains over a period of 48 h in both cellular models, tended to present a proinflammatory phenotype in THP-1 cells despite no change in TLR2 agonist activity.

Discussion

A higher rate of nonsynonymous SNPs than would be expected by chance occurs in the embC gene of Mabs in the course of human lung infection. Analysis of a subset of these SNPs in the isogenic background of Mabs ATCC 19977 rough indicates that these mutations are not functionally silent as they affect, to various degrees, the structure of LAM, the physiological properties of Mabs and its interactions with innate immune cells.

The nature and diversity of LAM structural changes that follow the introduction of nonsynonymous SNPs in embC point to the alteration of the activity of more enzymes than the only EmbC arabinosyltransferase. Such far-reaching consequences of EmbC SNPs may be explained by the fact that EmbC functions as part of a multiprotein complex and/or by changes in the metabolism of the bacilli in response to stresses imposed by these mutations. Two patient-derived mutants in particular, EmbC^M310T and EmbC^I402V, produced LAM variants devoid of succinyl substituents suggestive of loss of succinyltransferase (SucT) activity on LAM. Through the generation of a sucT knockout mutant, we were able to establish LAM succinylation as a—most likely indirect—modulator of the colony morphology and biofilm-forming capacity of Mabs. Whether triggered by changes in LAM succinylation or other structural alterations of the lipoglycan, the mechanisms underlying the large number of physiological and immunomodulatory changes undergone by the various patient-derived EmbC mutants are likely to be pleiotropic. As is the case with other major cell envelope constituents, alterations in the structure of LAM are expected to significantly impact the organization of both the inner and outer membranes of mycobacteria, with consequences on the permeability of the cell envelope, the export of envelope constituents and import of nutrients, and the surface composition and topology of the bacilli (55). Such changes may in turn impact their biofilm-forming capacity, cording, sliding motility, interactions with host immune cells, susceptibility to host defense mechanisms and virulence. In support of this hypothesis, differences were noted between the surface lipid content of the control strain and those of at least four EmbC mutants. Furthermore, the differential ability of the EmbC^M310Tand EmbC^I402V mutants to activate TLR2 was found to be attributable to their surface protein contents. Finally, the dramatic increases in the susceptibility of the embC knock-out mutant to a variety of antibiotics are evidence of the impact of LAM on the permeability of the cell envelope.

One of the most striking physiological changes associated with the EmbC^M310T and EmbC^I402V mutants was their decreased or loss of ability to form serpentine cords which, predictably, correlated with an increase in biofilm-forming capacity (56) and changes in colony morphology. The finding that LAM structure may modulate (directly or indirectly) these aspects of the physiology of Mabs is significant for at least two reasons. First is the impact that cording is known to have on the interactions of Mabs with macrophages. Cord formation confers upon Mabs the ability to withstand phagocytosis thereby promoting extracellular growth and immune evasion (10, 48, 57). More generally, the propensity of M. tuberculosis and NTM, including Mabs, to grow as serpentine cords correlates with persistent, progressive, infection (10, 18, 58–63). Second, the increase in biofilm-forming capacity that accompanies a loss or reduction in cording has precedents in other pathogens of the CF lung such as P. aeruginosa and B. cepacia where regain of biofilm formation during chronic infection is frequently observed and thought to enhance persistence in the lung tissue (14, 15).

The modulation of cell envelope (lipo)polysaccharide synthesis and structure is a conserved theme in chronic bacterial infections. In general, these changes contribute to host colonization, immune evasion, persisting inflammation, and antimicrobial resistance. P. aeruginosa and other pathogens of the CF lung, for instance, alter the structure of their LPS to modulate biofilm formation, proinflammatory signaling, macrophage invasiveness, and other innate immune defenses during chronic colonization of the airway. Multiple clinically relevant mutations affecting the structure of LPS have been reported, each typically associated with pleiotropic effects on the physiology of Gram-negative pathogens and their interactions with the host (14, 15). Our work indicates that Mabs uses a similar, albeit LAM-mediated, strategy to adapt to the lung environment. The number of phenotypic changes associated with single patient-derived SNPs in EmbC qualifies embC as a pleiotropic adaptive gene whose mutation can lead to a coordinated response impacting not only the host colonizing capacity of Mabs, but also its drug tolerance (through biofilm formation and cording) and immunomodulatory properties. Such coordinated responses aimed at enhancing in vivo persistence are otherwise typically achieved by the activation of global regulators or the development of hypermutability (64). It is also noteworthy that different SNPs in EmbC lead to different, sometimes divergent, phenotypic traits. This observation is suggestive of specializing adaptation to the different compartments of the human airway as has been reported with P. aeruginosa strains from patients with CF (64). Human airways are indeed highly compartmentalized, each compartment being associated with different environmental characteristics and stresses (64) to which Mabs must adapt. In the course of infection, the CF airway is further known to evolve, as reflected by changes in airway inflammation and the worsening of structural damage to the lung. In this context, the diversity of phenotypes conferred by different SNPs in EmbC may confer upon Mabs an adaptive advantage.

An unexpected finding of our studies is the identification of α-(1→3)-linked mannoside side chains substituting the mannan backbone of LAM in a number of mutants including the embC knockout strain. While Mabs, like Mycobacterium chelonae, was known to display α-(1→3)-linked monomannoside substituents in place of the more widespread α-(1→2)-linked monomannoside substituents of the mannan domain of LAM (46, 65), α-(1→3)-linked oligomannoside side chains have not previously been reported in any mycobacterial LAMs. This feature of Mabs LAM is distinct from the extended α-(1→6)-linked mannoside side chains reported in M. tuberculosis and M. smegmatis LAM (66) and the α-(1→2)-linked dimannoside substituents of M. kansasii LM (67). The impact that an increase in the relative proportion of α-(1→3)-linked oligomannoside side chains may have on the biological activities of Mabs LAM remains to be investigated. Nevertheless, this finding highlights the very dynamic nature of Mabs LAM, the structure of which appears to evolve in response to adaptive changes in the arabinan core domain.

In conclusion, our results establish LAM as an important modulator of Mabs virulence and host inflammatory responses to infection. They further indicate that nonsynonymous SNPs in a single gene (embC) may be a rapid way for Mabs to generate broad spectrum physiologic and pathogenic diversity required for persistence in the heterogeneous environment of the infected airway. This work also adds to our understanding of the structure of Mabs LAM and its dynamic remodeling in the course of infection. Critical aspects of the role of Mabs LAM in infection that future studies should address include a comparative assessment of the pathogenicity and persistence of the different embC mutants in immunocompetent, CF and non-CF, chronic animal models of Mabs infection. Given that embC mutations occur in both rough and smooth morphotype isolates (4, 29, 30), it would also be of interest to determine how morphotype affects the phenotypic and immunomodulatory outcomes of the same LAM mutants. Finally, the availability of LAMs presenting various degrees of modification by α-(1→3)-linked oligomannoside side chains opens the way to structure–function relationship studies aimed at understanding how these naturally occurring structural features impact the biological activities of the entire molecule.

Materials and Methods

Construction of the Mabs embC and sucT Mutants.

The embC (MAB_0189c) and sucT genes (MAB_2689) from Mabs ATCC 19977 were disrupted using the ORBIT system (68). Mabs ATCC 19977 strains expressing mutated forms of embC were generated by expressing, under control of the Phsp60 promoter from the integrative pMV306-xylE plasmid, the embC^Y222C, embC^M310T, embC^I402V, embC^D514N, embC^F646L, and embC^D1046G variants in the Mabs embC null mutant. See additional details in SI Appendix, SI Materials and Methods.

EmbC Mutants Modeling and Analysis.

EmbC mutants modeling and analysis was performed as described in SI Appendix, SI Materials and Methods.

Purification and Analysis of LAM.

Lipoglycans were extracted and purified from Mabs cells following procedures described earlier (45). Other procedures related to the structural analysis of LAM are detailed in SI Appendix, SI Materials and Methods.

Preparation of Surface Extracts.

Bacteria grown in 7H9-ADC-Tween 80 medium to an OD600 nm of 0.8 were harvested, resuspended in sterile water and gently vortexed for 1 min with 5 g glass beads (4 mm diameter) per 1 g wet cells as described (21). The resulting suspension was centrifuged, and the total surface extract was collected and filtered through a 0.2 μm sterile filter. Protein digestion with papain was performed by incubating total surface extracts (TSE) with papain (Spectrum Chemical Mfg. Corp.; 1 mg per mg of total surface extract proteins) at 37 °C for 16 h, after which the protease was inactivated by heating at 100 °C for 15 min.

Synthesis of 2,4,6-Tri-O-Methyl-D-Mannose.

The synthesis of 2,4,6-tri-O-methyl-D-mannose used as a standard for 3-linked mannose in the glycosyl linkage analyses is described in SI Appendix, SI Materials and Methods.

Drug Susceptibility Testing.

MIC values were determined in 7H9-ADC-Tween 80 broth and in cation-adjusted Mueller−Hinton II broth in a total volume of 100 μL in 96-well microtiter plates as described (46). MICs were determined using the resazurin blue test (69) and confirmed by visually scanning for bacterial growth. MIC is defined as the lowest concentration inhibiting growth.

Congo Red Binding and Sliding Motility.

Mabs strains were tested for Congo red binding in tryptic soy agar and for sliding motility on 7H9-ADC medium containing 0.34% agar as described (46).

Cording Assay.

The ability of the different Mabs strains to form serpentine cords was assessed by inoculating 10⁴ CFU/mL of each of the strains in tryptic soy broth and analyzing their growth characteristics after 3 d of incubation at 37 °C (70). The formation of serpentine cords was monitored using an Olympus CKX41 microscope. Pictures were taken using an Olympus U-CMAD3 camera.

Biofilm Formation.

Biofilm formation in SCFM was monitored by crystal violet staining as described by Belardinelli et al. (43).

Determination of hTLR2-Mediated NF-κB Activity Using HEK-Blue hTLR2 Reporter Cells.

The HEK-Blue™ human TLR2 (hTLR2) cell line (InvivoGen, San Diego, CA), a derivative of HEK293 cells that stably expresses the human TLR2 and CD14 genes along with a NF- κB-inducible reporter system (secreted alkaline phosphatase) was used as recommended by the manufacturer. Flat-bottom 96-well plates were used and 50,000 cells were seeded per well. A known TLR2 agonist, purified LM from M. tuberculosis H37Rv (0.2 µg per well), was used as a positive control for NF-κB induction. Mabs strains were added to the wells at a MOI of 1. Alternatively, TSE prepared from the control and mutant strains as described above, before and after treatment with papain, were added to the wells. To account for potential qualitative and quantitative differences in surface extract materials between the strains, the total volume of surface extract prepared for each strain was normalized to the dry weight of the bacteria pellet, and the same volume of surface extract for each strain was added to the cells. Plates were incubated at 37 °C in a humidified incubator with 5% CO₂ and hTLR2-mediated NF-κB activity was monitored spectrophotometrically at 650 nm after 16 h.

Macrophage Infections, Flow Cytometric Analysis of Activation Markers and Multiplex Immunoassay for Cytokines and Chemokines.

Human monocyte THP-1 cells (ATCC TIB-202) were grown in 24-well plates to ~90% confluence in RPMI-1640 (Corning) supplemented with 10% fetal bovine serum (FBS) and 1% penicillin-streptomycin and subsequently differentiated for 24 to 36 h with 50 ng/mL phorbol-12-myristate-13-acetate (Sigma-Aldrich) to induce differentiation into adherent macrophages. After 24 to 36 h, the culture medium was replaced and cells were allowed to rest for 24 h prior to infection. THP-1 cells were washed with warm PBS and infected with well-dispersed suspensions of the control and mutant strains (from frozen tittered stocks) in RPMI at an MOI of 1 for 2 h at 37 °C. Cells were then washed three times with warm PBS and the wells replenished with RPMI containing 250 µg/mL amikacin to kill extracellular bacteria. After an hour of incubation, cells were washed three more times with PBS and incubated in RPMI containing 50 µg/mL amikacin for the remainder of the experiment. Two and 48 h postinfection, viable intracellular Mabs was assessed by washing monocyte-derived macrophages twice with sterile PBS, lysing the cells in sterile water, and plating serial dilutions on 7H11 agar plates to enumerate CFU. All experiments were performed in biological triplicate and repeated three times.

Analysis of activation markers (MHCII, CD80, CD86, or CD40) and multiplex immunoassay for cytokines and chemokines essentially followed earlier procedures (71). See SI Appendix, SI Materials and Methods for details.

Infection of A549 Lung Alveolar Type II Epithelial Cells.

A549 human lung alveolar type II epithelial cells (ATCC CCL-185) were grown in 24-well plates to ~90% confluence in Dulbecco’s Modified Eagle Medium (DMEM) (Gibco) supplemented with 10% FBS and 1% penicillin–streptomycin. Cells were washed with warm PBS two to three times prior to infection with well-dispersed suspensions of the control and mutant strains (from frozen tittered stocks) at an MOI of 1 in DMEM at 37 °C. After 2 h of incubation, cells were washed three times with warm PBS and the wells replenished with DMEM containing 250 μg/mL amikacin to kill extracellular bacteria. After another hour of incubation, cells were finally washed three more times with PBS and incubated in DMEM containing 50 µg/mL amikacin for the remainder of the experiment. Two or 48-h postinfection, viable intracellular bacteria were assessed by washing the cells thrice with sterile PBS prior to lysis in sterile water and plating of serial dilutions on 7H11-OADC agar to enumerate CFU. Culture supernatants were collected 48 h postinfection and assayed for IL-8 secretion using a commercial human IL-8 DuoSet ELISA kit (R&D systems). All experiments were performed in biological triplicate and repeated two times.

Statistical Analysis.

Statistical tests were performed as indicated in the figure legends. Calculations were performed using GraphPad Prism version 9.5.1 for Windows (San Diego, CA).

Supplementary Material

Appendix 01 (PDF)

Data, Materials, and Software Availability

All study data are included in the article and/or SI Appendix.