Neisseria gonorrhoeae MlaA influences gonococcal virulence and membrane vesicle production

The six-component maintenance of lipid asymmetry (Mla) system is responsible for retrograde transport of phospholipids, ensuring the barrier function of the Gram-negative cell envelope. Located within the outer membrane, MlaA (VacJ) acts as a channel to shuttle phospholipids from the outer leaflet. We identified Neisseria gonorrhoeae MlaA (ngo2121) during high-throughput proteomic mining for potential therapeutic targets against this medically important human pathogen. Our follow-up phenotypic microarrays revealed that lack of MlaA results in a complex sensitivity phenome. Herein we focused on MlaA function in cell envelope biogenesis and pathogenesis. We demonstrate the existence of two MlaA classes among 21 bacterial species, characterized by the presence or lack of a lipoprotein signal peptide. Purified truncated N. gonorrhoeae MlaA elicited antibodies that cross-reacted with a panel of different Neisseria. Little is known about MlaA expression; we provide the first evidence that MlaA levels increase in stationary phase and under anaerobiosis but decrease during iron starvation. Lack of MlaA resulted in higher cell counts during conditions mimicking different host niches; however, it also significantly decreased colony size. Antimicrobial peptides such as polymyxin B exacerbated the size difference while human defensin was detrimental to mutant viability. Consistent with the proposed role of MlaA in vesicle biogenesis, the ΔmlaA mutant released 1.7-fold more membrane vesicles. Comparative proteomics of cell envelopes and native membrane vesicles derived from ΔmlaA and wild type bacteria revealed enrichment of TadA–which recodes proteins through mRNA editing–as well as increased levels of adhesins and virulence factors. MlaA-deficient gonococci significantly outcompeted (up to 16-fold) wild-type bacteria in the murine lower genital tract, suggesting the growth advantage or increased expression of virulence factors afforded by inactivation of mlaA is advantageous in vivo. Based on these results, we propose N. gonorrhoeae restricts MlaA levels to modulate cell envelope homeostasis and fine-tune virulence.


Introduction
The Gram-negative cell envelope (CE) plays an important role in bacterial physiology. Not only does it prevent cell lysis through the structure of the peptidoglycan cell wall [1], but it also acts to prevent entry of toxic lipophilic, hydrophilic, and amphipathic molecules. This barrier function is accomplished by the asymmetric outer membrane, which, in contrast to the phospholipid bilayer common to eukaryotic cells, is composed of an outer leaflet of lipopolysaccharide (LPS) or lipooligosaccharide (LOS) and an inner leaflet of phospholipids [2]. Due to the saturated fatty acids found in the hexa-acylated lipid A portion of LPS/LOS, the lipid interior of the LPS/LOS layer is less fluid than that of a phospholipid layer. As a result, the asymmetric bilayer is a more effective barrier than a phospholipid bilayer would be, and is thus less permeable to lipophilic compounds. If the asymmetry of the outer membrane is perturbed, phospholipids diffuse from the inner leaflet to the outer leaflet, which compromises the barrier function of the outer membrane [3].
Three lipid asymmetry-maintaining systems, primarily studied in Escherichia coli, are the phospholipase A PldA [4], the LPS palmitoyltransferase PagP [5], and the maintenance of lipid asymmetry (Mla) [6] systems. PagP and PldA both remove phospholipids from the outer membrane by destroying the phospholipid. PagP transfers a palmitate residue from the sn-1 position of outer leaflet phospholipids to lipid A to form hepta-acylated LPS, which increases hydrophobic interactions between adjacent LPS molecules [7]. PldA forms an active dimer in the outer membrane in the presence of phospholipids or lyso-phospholipids, then removes the sn-1 and sn-2 fatty acid side chains from the misplaced phospholipids [8]. These fatty acids can act as signal molecules to enhance LPS production, which suggests PldA possesses a secondary function as a sensor for altered membrane homeostasis [9]. In contrast to PagP and PldA, the six-component Mla system, composed of MlaA-F, does not destroy phospholipids in the outer leaflet. Instead, the Mla system is proposed to participate in retrograde transport of phospholipids from the outer leaflet, through the periplasm via MlaC, and back to the inner membrane, where phospholipids are thought to be integrated through the action of the Mla-FEDB complex (Fig 1A; [6]). The outer membrane component of this system, MlaA, was recently crystallized from Klebsiella pneumoniae and Serratia marcescens in complex with the outer membrane β-barrel protein OmpF. The crystal structure revealed that MlaA forms a pore through which phospholipid head groups are able to travel [10]. Loss of MlaA disrupts outer membrane integrity, as demonstrated by increased bacterial sensitivity to a combination of sodium dodecyl sulfate (SDS) and ethylenediaminetetraacetic acid (EDTA), in addition to several antibiotics [6,[10][11][12]. Single knockouts in any component of the Mla system in E. coli appear to mirror the defects caused by the lack of MlaA, and double knockouts of mlaA with any other Mla component exhibit similar sensitivity to single knockouts [6]. With the exception that iron limitation restricts MlaA transcription in vitro and in vivo [13], little is known about regulation of the Mla system.
MlaA is present in pathogenic and non-pathogenic Gram-negative bacteria. Its primary function, therefore, appears to be the maintenance of lipid asymmetry. However, infection studies with different pathogenic bacteria suggest that MlaA possesses divergent moonlighting roles. Originally discovered in Shigella flexneri and named VacJ for virulence associated, chromosome locus J, this protein contributes to the ability of S. flexneri to invade adjacent epithelial cells [14,15]. In Haemophilus influenzae, VacJ plays a role in serum resistance [16], and in Pseudomonas putida and Campylobacter jejuni, MlaA provides protection against oxidative stress [17,18]. Downregulation of VacJ has also been associated with the increased formation of membrane vesicles in H. influenzae and Vibrio cholerae [13]. MlaA/VacJ knockouts in S. flexneri, H. parasuis, and Salmonella enterica Typhimurium exhibit virulence defects [11,14,15]. In contrast, P. aeruginosa deficient in VacJ was significantly more virulent [12].
Phospholipid regulation other than the action of PldA has not been studied in the genus Neisseria, which includes the human pathogens N. meningitidis and N. gonorrhoeae. The Neisserial Mla system has not been characterized. N. gonorrhoeae, the causative agent of gonorrhea, is a worldwide public health threat. The World Health Organization estimates that 78 million new cases are acquired globally every year [19]. Treatment failures with the last effective class of antibiotics have been encountered in several countries, highlighting the necessity of developing new therapeutic interventions [20,21]. We identified MlaA as a therapeutic candidate for gonorrhea in a high-throughput proteomic examination of the CE and naturally released membrane vesicles (MVs). Deletion of this protein, encoded by the ngo2121 open reading frame in N. gonorrhoeae FA1090, resulted in phenotypes that suggested disrupted outer membrane integrity [22]. We subsequently performed a comprehensive phenotypic microarray screen to assess the function of seven proteome-derived gonorrhea vaccine candidates and therapeutic targets, including MlaA, in CE homeostasis. The results revealed an extensive sensitivity phenome in a ΔmlaA mutant, including increased susceptibility to compounds that trigger oxidative stress. Importantly, deletion of mlaA in the highly antibiotic resistant WHO X strain resulted in similar phenotypes to those observed for a ΔmlaA mutant constructed in the FA1090 laboratory strain [23].
In this work, we further characterized the role of MlaA in gonococcal physiology and pathogenesis using different in silico, genetic, proteomic, and in vitro and in vivo assays. The results of our investigations revealed a previously unknown gonococcal virulence pathway and suggest that N. gonorrhoeae may employ MlaA to modulate its CE and MV protein profile and fine-tune its ability to colonize the host.

MlaA conservation
Despite the importance of MlaA in the Gram-negative cell envelope and its different outcomes on pathogenesis, its conservation across bacterial species has not been addressed. Accordingly, we analyzed the similarity at the amino acid level in a diverse range of Gram-negative bacteria in comparison to N. gonorrhoeae MlaA. With the exception of other Neisseria species, the percent identity was relatively low and ranged from 26.15% to 34.53% (Table 1, S1 File). We previously noted that although N. gonorrhoeae MlaA is annotated as a predicted lipoprotein, it lacks the universally conserved cysteine residue required for lipidation and membrane anchoring [23,24]. We were curious whether the lack of a lipoprotein signal peptide is limited to N. gonorrhoeae; therefore, MlaA homologs from Neisseria species and other bacteria in which MlaA/VacJ has been characterized were scrutinized for the presence of a lipoprotein signal peptide. This analysis revealed that homologous proteins in P. aeruginosa, P. putida, C. jejuni, Caulobacter crescentus, and Desulfovibrio vulgaris did not contain the conserved cysteine residue. Additionally, homologs in the closely-related N. meningitidis, N. lactamica, and N. weaveri did not contain a lipoprotein signal peptide (Table 1). We subsequently carried out a secondary search for MlaA without a predicted lipoprotein signal sequence for the presence of a signal peptidase I (SPaseI) motif and constructed a maximum likelihood phylogenetic tree (Fig 2A). Within the tree, SPaseI-cleaved MlaA homologs clustered separately from those containing a lipoprotein signal peptide (SPaseII-cleaved proteins), with the exception of K. pneumoniae and Francisella tularensis, with the latter protein forming an outgroup within the larger cluster. An examination of the level of nucleotide conservation between MlaA homologs in all Neisseria performed using the Neisseria Multi Locus Sequence Typing database (locus identifier NEIS1933) indicated that the majority of the ngo2121 nucleotide sequence represented 90-100% of the alleles, with 461 alleles exhibiting 410 polymorphic sites (Fig 2B). Phylogenetic analyses of amino acid sequences representing MlaA alleles from all Neisseria and from N. gonorrhoeae isolates exclusively showed that MlaA alleles are closely related among Neisseria sp. and within N. gonorrhoeae (S1A and S1B Fig, respectively).
In summary, our examinations of MlaA conservation, both outside and within the Neisseria genus, revealed differences that warranted further investigation into the function of this protein in N. gonorrhoeae.

Genome context and organization of MlaA
To extend our observations of the differences between MlaA in divergent bacteria, we examined the genomic location of mlaA across 7 bacterial species in which MlaA has been investigated. Predictions by biocyc.org suggested that N. gonorrhoeae mlaA is a member of an operon consisting of ngo2120 to ngo2124 (Fig 2C). However, upon closer inspection of the local genome context, mlaA appeared to be a part of a polycistronic operon composed of ngo2116 to ngo2124, primarily because of the small intergenic distances and the lack of predicted promoters between each open reading frame. Importantly, NGO2120, NGO2119, NGO2118, NGO2117, and NGO2116 exhibited homology to the other components of the E. coli Mla system, MlaB, MlaC, MlaD, MlaE, and MlaF, respectively (Fig 2C, Table 1 in S1 Text). The genetic organization surrounding mlaA is shared by N. meningitidis and N. lactamica. In contrast, E. coli, S. marcescens, S. flexneri, K. pneumoniae, P. aeruginosa, and S. enterica enterica serovar Typhimurium mlaA are spatially isolated from the genes encoding the remaining components of the Mla system (S2 Fig). The genomic organization of the Mla system into one operon within the Neisseria genomes suggests regulation of Mla complex expression differs between various bacterial species.

Protein purification and antibody generation
To generate molecular tools for our studies, we set out to purify a soluble, recombinant variant of MlaA. A representative schematic of full-length MlaA is presented in Fig 3A. This protein contains a predicted signal peptide from residues 1 to 20, cleaved by SPaseI, and an ABC transporter Mla domain from residues 25 to 218. Although full-length MlaA was predominantly found in the soluble protein fraction, initial purification attempts with a 6× His-tag or a maltose-binding protein purification handle were unsuccessful due to extensive aggregation of MlaA (S3 Fig). To circumvent this challenge, a truncated version of the protein without the first predicted transmembrane helix, MlaA 120-277 , with a C-terminal 6× His-tag was generated ( Fig 3B) and purified by affinity chromatography, yielding 99% pure MlaA that migrated at 22 kDa, consistent with the predicted size of the engineered protein ( Fig 3C). Rabbit polyclonal antibodies produced against this protein recognized purified MlaA 120-277 -His ( Fig 3D) and a protein band corresponding to the approximate size of native MlaA (29.6 kDa) in whole cell lysates of WT N. gonorrhoeae but not the ΔmlaA mutant ( Fig 3E). In all cell lysates, a major cross-reactive protein with highly variable expression that was not associated with MlaA A phylogenetic tree of MlaA was constructed in MEGA using amino acid sequences of MlaA/VacJ homologs downloaded from NCBI. The Jones-Taylor-Thornton model was used to generate a pairwise distance matrix. Neighbor-Join and BioNJ algorithms were subsequently applied to the matrix to obtain the initial tree for a heuristic search. 500 bootstrap iterations were performed to test the phylogenies. The highest log-likelihood tree is presented. Homologs without lipoprotein signal levels was also observed ( Fig 3E, marked with an asterisk). BLAST searches of the FA1090 genome with the MlaA 120-277 amino acid sequence did not reveal any proteins that could be the cross-reactive band. A titration with isopropyl β-D-1-thiogalactopyranoside (IPTG) to examine induction of MlaA in the complemented strain (ΔmlaA/P lac ::mlaA) constructed previously [23] revealed that maximal expression of MlaA was achieved with 50 mM IPTG ( Fig  3E). However, even with this high level of inducer, expression was not restored to WT levels, indicating that certain phenotypes observed for the mlaA null strain may not be entirely complemented, as we previously observed with hypersensitivity to bile salts [23]. A similar IPTG titration performed with an E. coli strain harboring the pGCC4-ngo2121 complementation plasmid revealed that MlaA expression was readily induced in a heterologous host ( Fig 3E). We independently generated additional complementation strains, one in the FA1090 ΔmlaA  purification with first 119 amino acids removed and a 6 × Histidine tag (His; as a blue rectangle) added to the N-terminus. Schematics are not to scale. (C) Truncated MlaA construct was overexpressed in E. coli and purified by nickel affinity chromatography in the presence of 1% Triton-X 100 detergent. Detergent was subsequently removed by incubation with Bio-Rad Bio-Beads SM-2 resin. To assess purity, 1 μg of protein was subjected to 1D SDS-PAGE and visualized by staining with Colloidal Coomassie Blue G-250. Open arrow indicates migration band of MlaA 120-277 . Migration of a molecular weight marker (in kDa) is indicated to the left of the gel. (D) Immunoblot evaluation of anti-MlaA antiserum. Indicated amounts of purified MlaA 120-277 were separated by SDS-PAGE, transferred to a nitrocellulose membrane, and probed with anti-MlaA antiserum. (E) Equivalent OD 600 units of WT, isogenic ΔmlaA, and either ΔmlaA/P lac ::mlaA or E. coli harboring the pGCC4-ngo2121 complementation plasmid cultured with indicated concentrations of IPTG, were separated by SDS-PAGE, transferred to a nitrocellulose membrane, and probed with anti-MlaA antiserum. Open arrow indicates MlaA. Non-specific cross-reactive band is marked with an asterisk ( � ). OD 600 , optical density at 600 nm; IPTG, isopropyl β-D-thiogalactopyranoside; SDS-PAGE, sodium dodecyl sulfate-polyacrylamide gel electrophoresis.
https://doi.org/10.1371/journal.ppat.1007385.g003 mutant and four in WHO X lacking MlaA [23]. The non-specific cross-reactive band was not observed in the WHO X strain. WT levels of protein expression were not restored in any of the strains (S4A Fig). These results suggest that MlaA may be under the influence of secondary regulation or requires the presence of additional Mla components for stability in N. gonorrhoeae.

Expression of MlaA among gonococcal clinical isolates and within Neisseria
We examined the expression pattern of MlaA in whole cell lysates collected from geographically and temporally distinct clinical isolates [25,26]. Immunoblots indicated that MlaA was expressed by all 38 strains at different levels ( Fig 4A). Additionally, in the case of isolate LG20, MlaA migrated at a lower molecular weight than FA1090 MlaA. Finally, the non-specific band observed in FA1090 lysates (band marked by asterisk in Fig 4A) was highly abundant only in three of the 18 clinical isolates (LGB26, UW07, and UW13), one WHO reference strain (WHO L), FA6140, and three of the four laboratory strains examined (F62, FA19, and 1291).
We further assessed whether antiserum against FA1090 MlaA could recognize homologous proteins from other Neisseria species, as well as more distantly related bacteria. Of the seven Whole cell lysates of different Gram-negative bacteria, including E. coli BL21(DE3); V. cholerae N19691; P. aeruginosa PA01; K. pneumoniae 6069; N. meningitidis MC58; the commensal bacterium N. lactamica NLI83/-01; and the opportunistic pathogen N. weaveri 1032 were subjected to immunoblot analysis. All lysates were standardized by OD 600 values, separated in a 4-15% Tris-glycine gel, and probed with polyclonal rabbit antiserum against MlaA. FA1090 and ΔmlaA were included in blots as positive controls. Open arrow indicates MlaA. Non-specific crossreactivity is marked with an asterisk ( � ). OD 600 , optical density at 600 nm; SDS-PAGE, sodium dodecyl sulfatepolyacrylamide gel electrophoresis. species tested, only N. meningitidis and N. lactamica homologs were detected ( Fig 4B). In summary, these observations provide additional support for the results of our bioinformatic analyses and indicate that MlaA is likely to play a similar role in the CE of clinical isolates collected from different geographical locations, at different points in time, and exhibiting all known antibiotic resistance profiles [25,26]. Further, the expression of MlaA across a range of diverse clinical isolates supports our use of FA1090 as a type strain to study the effects of the loss of MlaA on gonococcal fitness and pathogenesis.

Phenotypic characterization of MlaA in vitro
To gain further insights into the impact of MlaA on gonococcal physiology, we first examined the growth kinetics of the WT and ΔmlaA mutant under standard growth conditions in liquid medium. Neither the ΔmlaA mutant nor the complemented strain displayed a difference in growth compared to WT (Fig 5A), consistent with our previous observations using solid medium and chemically defined Graver-Wade liquid medium [22,23]. An assessment of MlaA abundance in WT whole cell lysates from liquid cultures over time revealed that expression of MlaA was lower during the lag phase, slightly increased during logarithmic growth, reached maximum expression at approximately mid-logarithmic phase, and was maintained at similar levels until stationary phase (Fig 5B), at which point the experiment was terminated to avoid the effects of autolysis [27].

N. gonorrhoeae restricts the MlaA cellular pool during iron deprivation
Iron is a well-characterized regulator of gonococcal gene expression. Numerous genes are iron-repressed and are thus expressed in body sites where iron is limited [28,29]. Therefore, we assessed the expression of N. gonorrhoeae MlaA and its effect on bacterial viability under iron deprivation. Titration with the iron chelator desferal (up to 25 μM) revealed that MlaA production decreased under increasing iron limitation, in contrast to TbpB, a well-characterized protein induced during iron starvation ( [30]; Fig 5C). Antiserum against MtrE, a component of the MtrCDE efflux pump that is not influenced by iron [25,31], was used as a loading control and showed that MtrE expression was unaffected under any of the desferal concentrations tested. Densitometry analysis of MlaA abundance indicated a statistically significant decrease in the MlaA cellular pool under exposure to 15 μM desferal and higher ( Fig 5D).
To examine whether the lack of MlaA affects bacterial viability under various iron concentrations, we monitored the growth of WT and ΔmlaA bacteria over time by enumeration of colony forming units (CFU/mL). Both strains were equivalently viable when exposed to moderate iron starvation (10 μM desferal; Fig 5E, top panel). In contrast, exposure to high iron starvation (25 μM of the iron chelator) resulted in a significant decrease in bacterial viability beginning at 3 h and continuing to the experimental endpoint for both strains, with the ΔmlaA mutant exhibiting a slight, non-statistically significant growth advantage over the WT (Fig 5E,  bottom panel).
Together, these results show that depletion of iron negatively regulates the MlaA cellular pool and suggest that decreased levels of MlaA are favored by the bacteria under iron starvation encountered in the host.

MlaA affects bacterial viability under other in vitro conditions relevant to infection
N. gonorrhoeae is also exposed to conditions other than iron limitation during infection of different niches within the human host. Gonococci proliferate in microaerobic or anaerobic conditions within the female reproductive tract and may also be exposed to serum within  [32]. The recovery of viable ΔmlaA mutant bacteria was statistically significantly higher (~2-fold) than that of the WT for all tested conditions relevant to infection of the host using solid medium with 5 μM desferal (-Fe), 7.5% normal human serum (NHS), or with nitrite under anaerobic conditions (-O 2 ; Fig 5F). As expected from our investigations into iron starvation in liquid medium, immunoblotting analysis indicated no difference in MlaA expression under aerobic exposure to 5 μM desferal. The presence of serum also had no effect on MlaA levels. In contrast, MlaA was elevated under anaerobiosis in comparison to standard growth ( Fig 5G). To examine the interplay between iron starvation and anoxia on MlaA expression, immunoblotting was also performed on bacteria collected from anaerobic conditions with increasing concentrations of desferal. Densitometry analysis indicated a significant difference in the abundance of MlaA during anoxia, as well as during anaerobic exposure to 10 or 20 μM desferal, although MlaA expression was highly variable under each anaerobic condition tested ( Fig 5H). The integral membrane component of the β-barrel assembly machinery (BAM) complex, BamA [25,33], was used as a loading control (Fig 5G), and its expression was not altered under any of the conditions examined [25].

Iron starvation diminishes the MlaA cellular pool independent of Fur levels
In N. gonorrhoeae, the ferric uptake regulator (Fur) protein governs expression of iron homeostasis genes in response to the intracellular pools of this important metal. To dissect the mechanism of MlaA repression during iron deprivation, we employed a conditional Fur knockout, Δfur/P lac ::fur, as Fur is essential in N. gonorrhoeae (S5A Fig, [34]). Initial viability assessment indicated that Δfur/P lac ::fur proliferated identically to WT bacteria in liquid medium were probed with antisera against MlaA, MtrE, and TbpB. ΔmlaA cultured under standard conditions was included as a reference. (D) Densitometry analyses of MlaA using immunoblots from three independent desferal titration experiments shown as a representative blot in panel (C). (E) WT FA1090 and isogenic ΔmlaA were cultured in liquid medium containing either 10 μM (top panel) or 25 μM (bottom panel) desferal for 6 h. At each hour, samples were withdrawn and diluted for CFU/mL enumeration. Both graphs contain growth curves from cultures maintained under standard conditions (blue curves). Statistical significance was assessed by two-way ANOVA using Sidak's multiple comparisons test. (F) WT FA1090 and isogenic knockout ΔmlaA were cultured in liquid medium under standard conditions until OD 600 had at least doubled (~3 h). Cultures were standardized to an OD 600 of 0.2 and dilutions were spotted onto GCB. Plates were prepared as normal (SGC); under iron deprivation (-Fe); supplemented with 7.5% normal human serum (+NHS); or supplemented with 1.2 mM NaNO 2 and cultured anaerobically (-O 2 ). Strains were maintained at 37˚C in 5% CO 2  Cultures were diluted to an OD 600 of 0.05 in sterile PBS and diluted 1000-fold in EMEM. Suspensions were combined with an equal volume of EMEM, NHS, or heat-inactivated NHS and incubated for 1 h in 5% CO 2 at 37˚C. Bacteria from each well were spotted onto GCB plates. CFUs were scored after 20-22 h incubation in 5% CO 2 at 37˚C (p value between WT and ΔmlaA exposed to active NHS, 0.07). (L) Rapidly growing cultures of WT FA1090 and isogenic knockout ΔmlaA were diluted to 10 5 CFU/mL and cultured for 3 h in the presence of liquid medium (SGC), 0.01% acetic acid (vehicle), or 10 μM human defensin (HBD1). Bacteria were serially diluted and spotted onto GCB plates. CFUs were scored after 20-22 h incubation in 5% CO 2 at 37˚C and survival was calculated relative to SGC. n�3; mean ± SEM presented for all experiments; panels D, H, J, and L present values from each replicate; � p < 0.05; OD 600 , optical density at 600 nm; SGC, standard growth conditions; IPTG, isopropyl β-D-thiogalactopyranoside; SDS-PAGE, sodium dodecyl sulfate-polyacrylamide gel electrophoresis; CFU, colony forming unit; GCB, gonococcal base medium; GCBL, gonococcal base liquid medium; PBS, phosphate buffered saline; EMEM, Eagle's minimal essential medium. Immunoblotting ( Fig 5I) coupled with densitometry revealed a statistically significant decrease in MlaA levels under iron starvation, which was not restored with low (10 μM IPTG) or high (100 μM IPTG) Fur induction ( Fig 5J). A non-significant reduction in MlaA expression compared to standard conditions was observed with the lowest Fur level examined during iron repletion, and MlaA abundance trended upward as IPTG was added ( Fig 5J). We also examined TbpB expression as a Fur-regulated control [35]. Our analysis indicated that low Fur expression during iron repletion was insufficient to repress TbpB to the levels observed for the WT under standard conditions (Fig 5I and S5C Fig). Upon iron depletion, TbpB was derepressed in the Δfur/P lac ::fur strain, regardless of Fur induction (Fig 5I), reflecting the inability of the repressor to dimerize and bind to its DNA target in the absence of iron [36]. BamA was used as a loading control and exhibited no alterations under any condition examined. Our quantitative immunoblotting assessments suggest Fur may exert a slight positive effect on MlaA expression under iron replete conditions, but that iron starvation overrules this influence. It is possible that MlaA is regulated by several factors-similar to TbpB, which is controlled by a long noncoding RNA and the MisR response regulator [28,36].

N. gonorrhoeae MlaA does not contribute to serum resistance
Next, we sought to examine the sensitivity of WT, ΔmlaA, or ΔmlaA/P lac ::mlaA bacteria to NHS in an assay using liquid medium [37], as VacJ has been described to contribute to serum resistance in H. influenzae [16]. Enumeration of CFUs after exposing gonococci to 50% NHS for 1 h revealed a significant decrease in viability for all strains compared to the same concentration of heat-inactivated NHS. No significant difference in survival, however, was observed between the WT, ΔmlaA, or ΔmlaA/P lac ::mlaA strains during exposure to active serum ( Fig 5K).

Bacteria lacking MlaA are more sensitive to human defensins
To extend our examination into the possible physiological contribution of MlaA during infection of the host, we exposed WT and ΔmlaA bacteria to human β-defensin 1 (HBD1). Defensins are cationic antimicrobial peptide components of the innate immune system which interfere with membrane integrity to exert their antibacterial effect [38]. Lack of MlaA was associated with a significant 2.5-fold reduction in viability compared to WT after HBD1 treatment ( Fig 5L). N. gonorrhoeae is naturally resistant to HBD1 [39], which is constitutively expressed by epithelial cells [38]. The ΔmlaA mutant's diminished viability upon HBD1 exposure suggests that loss of MlaA results in a cell envelope defect.

Outer membrane integrity is altered in a mlaA null mutant and can be partly rescued by PldA overproduction
Our observation that the ΔmlaA mutant was more susceptible to the membrane perturbing activity of HBD1 is consistent with our previous studies, in which we reported increased sensitivity of ΔmlaA mutants in the FA1090 and WHO X genetic backgrounds to polymyxin B on solid medium [22,23]. Gonococci lacking MlaA were also more susceptible to several antibiotics during growth in chemically defined liquid medium, a sensitivity phenome that suggested an outer membrane defect, rather than general loss of membrane integrity [23]. To further determine how the altered outer membrane integrity in the ΔmlaA mutant translates to antibiotic tolerance, we applied an alternative approach and examined bacterial susceptibility to ten antibiotics with different mechanisms of action using Etest strips ( Table 2). As expected from our previous studies, loss of MlaA resulted in a 2-fold decrease in the minimal inhibitory concentration (MIC) of polymyxin B (ΔmlaA mutant, MIC 32 μg/mL; WT bacteria, MIC 64 μg/mL), ampicillin (0.064 μg/mL, versus 0.125 μg/mL against the WT), and vancomycin (4 μg/mL, versus 8 μg/mL against the WT). All three antimicrobial compounds act by interfering with the barrier function of the outer membrane and the cell wall [40], and vancomycin was selected specifically to serve as a marker for outer membrane permeability, as an intact Gram-negative outer membrane excludes this antibiotic [41]. No differences in the MICs of other antibiotics with different mechanisms of action were observed. Consistent with results in the FA1090 background, deletion of mlaA from the WHO X genome resulted in 16-and 2-fold lower polymyxin B and vancomycin MICs, respectively (Table 2). Additionally, the clinically relevant antibiotics ceftriaxone and cefixime were not more effective against the WHO X ΔmlaA mutant, which provided further evidence that the permeability defect solely affects the outer membrane, as observed for the FA1090 mlaA null strain. On the other hand, the loss of MlaA does not interfere with cell envelope integrity to the extent that periplasmic proteins are released to the extracellular milieu, as no increase in the abundance of DsbA or SurA was observed in ΔmlaA-derived supernatants ( Fig 6A). Additionally, neither DsbA nor SurA was upregulated in whole cell lysates of the ΔmlaA mutant, suggesting that absence of MlaA does not interfere with proper outer membrane protein folding [42][43][44].
Overexpression of the phospholipase PldA rescued SDS/EDTA sensitivity in E. coli ΔmlaC and in ΔmlaA/ΔmlaC knockouts, presumably by removal of phospholipids accumulated at the cell surface [6]. We therefore determined whether PldA overproduction would rescue the antibiotic sensitivity phenotypes observed for the N. gonorrhoeae ΔmlaA mutant. While WT resistance to ampicillin was restored in the ΔmlaA/P lac ::pldA strain, the vancomycin MIC was the same as the ΔmlaA strain. However, when PldA was overexpressed, the strain's polymyxin B sensitivity increased two-and four-fold in comparison to the ΔmlaA mutant and the WT strain, respectively (Table 2). Finally, follow-up agar dilution MIC assessment confirmed the fold decrease in MIC for each strain for polymyxin B and vancomycin but not for ampicillin. Complementation with MlaA was sufficient to restore resistance to polymyxin B but not to   vancomycin ( Table 2 in S1 Text). In all cases, the MICs were higher by agar dilution than by Etest, an effect noted previously [45].

Loss of MlaA results in reduction of gonococcal colony size
Our antibiotic susceptibility testing indicated ΔmlaA exhibited an outer membrane defect that was exacerbated upon overproduction of PldA in the presence of polymyxin B. To examine these effects more closely, viability was assessed for WT, ΔmlaA, ΔmlaA/P lac ::mlaA, and ΔmlaA/P lac ::pldA grown on solid medium supplemented with 800 U/mL (~84 μg/mL) polymyxin B and either 0 or 0.5 mM IPTG for the expression of MlaA or PldA. Visual inspection revealed an apparent decrease in ΔmlaA viability from the -5 to the -3 dilution during exposure to polymyxin B (Fig 6B, top panels). Upon examination with a stereo-microscope, colonies of the ΔmlaA mutant were noticeably smaller than those of the WT, although colony counting revealed no significant difference in CFUs compared to the absence of polymyxin B for the ΔmlaA mutant (106.2% ± 6.8 for WT compared to 73.6% ± 24.5 for ΔmlaA). In the absence of PldA induction, the viability of the ΔmlaA/P lac ::pldA strain was not significantly different from that of the WT when exposed to polymyxin B (54.2% ± 17.42 for ΔmlaA/P lac ::pldA). However, upon pldA overexpression in the ΔmlaA background, survival was significantly lower than WT bacteria (117.2 ± 5.7 for WT; 0.04% ± 0.01 for ΔmlaA/P lac ::pldA), dropping by 99.96% compared to cells cultured in the absence of polymyxin B (Fig 6C). Measurement of colony diameters (Fig 6D, left column) revealed that lack of MlaA resulted in significantly decreased colony size, even in the absence of the antimicrobial peptide (282.7 ± 2.66 μm for WT, 223.2 ± 2.19 μm for ΔmlaA). This phenotype was not reversed in the ΔmlaA mutant by expression of MlaA from a heterologous location (219.1 ± 2.91 μm; Fig 6E, left panel). The difference in colony size between WT and ΔmlaA was aggravated in the presence of polymyxin B (148.6 ± 2.35 μm for WT, 63.28 ± 1.02 μm for ΔmlaA; Fig 6D, right column), and was partially restored in the ΔmlaA/P lac ::mlaA strain (115.4 ± 2.45 μm; Fig 6E, right panel). Analysis of LOS isolated from WT, ΔmlaA, ΔmlaA/P lac ::mlaA, and with PldA at either native levels or overproduced in ΔmlaA/P lac ::pldA bacteria in the presence or absence of polymyxin B revealed that neither loss of MlaA nor overproduction of PldA in the ΔmlaA background resulted in alterations to LOS abundance or migration ( Fig 6F). Likewise, the presence of the antimicrobial peptide had no effect on LOS in any of the strains tested. Thus, the polymyxin B phenotype was not due to LOS defects.
Our experiments provided the first evidence that in N. gonorrhoeae, absence of MlaA affects outer membrane permeability to compounds acting against the cell wall, which can be partly rescued by the action of PldA, and alters colony morphology. Further, the relatively low levels of MlaA present in the ΔmlaA/P lac ::mlaA strain partially alleviated the polymyxin B sensitivity phenotype resulting from this defect (Fig 3E). We conclude that MlaA is more important to the bacteria during exposure to cell envelope stress conditions than under standard laboratory growth.

Gonococcal cell morphology is unaltered in the absence of MlaA
To determine whether the loss of MlaA either alone or combined with PldA expression would alter cell morphology and thus result in increased sensitivity to polymyxin B, we employed transmission electron microscopy. Cells from the ΔmlaA mutant typically appeared identical to WT cells, although the mutant occasionally exhibited ruffled membranes (Fig 7A, top row). Additionally, the ΔmlaA/P lac ::pldA strain was indistinguishable from WT, either in the presence or absence of PldA overexpression (Fig 7A, top row). To determine whether any of the strains exhibited morphological differences in the presence of polymyxin B, we also imaged were separated by low-speed centrifugation and filtration, treated with DNAseI, and precipitated with a pyrogallol red-molybdate-methanol procedure. Precipitated supernatants and whole cell lysates were standardized by the OD 600 of the source culture, separated by SDS-PAGE, and probed with indicated antisera. (B) WT FA1090, isogenic knockout ΔmlaA, complementation strain ΔmlaA/P lac ::mlaA, and PldA overexpression strain ΔmlaA/ P lac ::pldA were cultured aerobically in liquid medium for 3 h, back diluted to an OD 600 of 0.1, cultured 2 h longer, serially diluted, and spotted onto GCB without (left column) or with (right column) polymyxin B (800 U/mL) and either without (top row) or with (bottom row) 0.5 mM IPTG. Dilution spots from each condition were imaged with a Bio-Rad ImageDoc system. (C) CFUs for permissive and restrictive conditions with or without IPTG were counted and relative viability was calculated. Experiment was performed on three separate occasions (mean ± SEM on graph; � p < 0.05), and typical plate images are presented. (D) Representative micrographs from 10 −4 dilution taken with a Zeiss AxioObserver.D1 microscope at 10× magnification 0.25 Phase Contrast 1 of WT (Row 1), ΔmlaA (Row 2), and ΔmlaA/P lac ::mlaA (Row 3). MlaA expression was induced by inclusion of 0.1 mM IPTG in the solid medium for the complementation strain. (E) Images of 10 −4 dilution were also taken at 2.5× magnification 0.06 Phase Contrast 1 and colony diameters were measured with ImageJ software. Colonies were measured for each of two independent experiments for the − polymyxin B condition (WT, n = 548; ΔmlaA, n = 755; ΔmlaA/P lac ::mlaA, n = 664) and for the + polymyxin B condition (WT, n = 836; ΔmlaA, n = 1197; ΔmlaA/P lac ::mlaA, n = 1121; mean ± SEM on graphs; � p < 0.05). (F) Rapidly growing liquid cultures of WT, ΔmlaA, ΔmlaA/ P lac ::mlaA, and ΔmlaA/P lac ::pldA incubated in the presence or absence of polymyxin B were lysed and treated with proteinase K to isolate diplococci collected from cultures supplemented with the antimicrobial peptide. The presence of polymyxin B did not alter the overall morphology of any of the strains examined (Fig 7A,  bottom row). Finally, we occasionally observed numerous blebs on cell surfaces, as shown for WT bacteria cultured without polymyxin B and ΔmlaA bacteria collected from both conditions ( Fig 7A).

MlaA localizes to the cell envelope and naturally released membrane vesicles
We identified MlaA within CE and MV fractions isolated from four different N. gonorrhoeae strains in proteomic investigations using isobaric Tagging for Relative and Absolute Quantification coupled with Mass Spectrometry [22]. To validate these results and to determine to which cellular compartment(s) N. gonorrhoeae MlaA associates, we performed an immunoblotting analysis of subcellular fractions using anti-MlaA antiserum and control antisera against the periplasmic-facing lipoprotein member of the BAM complex, BamD [46]; BamA [25,33]; and the cytoplasmic enzyme Zwf [47][48][49]. This analysis revealed that none of the proteins were detected in the supernatant (SS). MlaA exclusively localized to the CE and MV fractions, similar to BamD. BamA was detected primarily in the CE and MV fractions, but was also found in small amounts in the cytoplasmic/periplasmic fraction, which reflects the presence of five periplasmic polypeptide transport associated domains [25,33]. Finally, as expected [49], Zwf was detected solely in the cytoplasmic/periplasmic fraction (Fig 7B).

Cell envelope and membrane vesicle composition is altered in the ΔmlaA mutant
The presence of membrane blebs on the surface of ΔmlaA cells both under permissive and stress conditions (Fig 7A), combined with the proposed role of MlaA in vesicle biogenesis [13], prompted us to assess whether this mechanism was conserved in N. gonorrhoeae. We examined vesicle formation by WT, ΔmlaA, and ΔmlaA/P lac ::pldA bacteria during standard growth in liquid medium, as well as under polymyxin B exposure. Indeed, quantitation of MVs revealed that the ΔmlaA mutant produced significantly more vesicles than WT under standard conditions (0.80 ± 0.02 mg L -1 OD 600 -1 compared to 0.47 ± 0.03 mg L -1 OD 600 -1 for WT, Fig 7C). Overexpression of PldA partially decreased MV production, although the ΔmlaA/P lac ::pldA strain still released significantly more MVs than WT bacteria (0.58 ± 0.03 mg L -1 OD 600 -1 , Fig 7C). However, in the presence of polymyxin B, all three strains produced significantly more MVs than under non-stress conditions (2.8-, 1.9-, and 2.7-fold increases for WT, ΔmlaA, and ΔmlaA/P lac ::pldA, respectively). No significant differences were observed in MV secretion between strains during polymyxin B exposure (Fig 7C). The hypervesiculation phenotype of the ΔmlaA mutant did not translate into increased biofilm formation (Fig 7D, S6  Fig) despite the fact that gonococcal biofilms are primarily membranous material generated by the release of MVs [50].

MlaA reduces gonococcal fitness in the female mouse lower genital tract
The increased blebbing, sensitivity to antimicrobial peptides, and the differential expression pattern of MlaA during iron deprivation and anaerobiosis led us to test whether the lack of MlaA impacts N. gonorrhoeae survival during infection. We chose to establish a competitive index (CI) as it provides a relative measure of bacterial fitness in comparison to a reference strain (in this case, WT bacteria) that takes the starting inoculum into account (see Materials and methods for calculation). A CI equal to 1 indicates that both strains are comparably fit whereas CI values greater or lower than 1 show that the mutant is able to outcompete the WT strain or displays decreased fitness, respectively. We first established that MlaA does not affect bacterial fitness using in vitro competitive experiments (Fig 8A) then used the female mouse model of gonococcal lower genital tract infection as a measure of in vivo fitness [51,52]. In three biological replicates, each with at least seven mice, we observed a 9.8-to 16.9-fold increase in the competitive index for the ΔmlaA strain. Additionally, only mutant bacteria were isolated from several mice at each time point (Fig 8B). Expectedly, the same phenotype was observed in the ΔmlaA/P lac ::mlaA strain (S7 Fig) due to the inability to restore MlaA expression to WT levels in this complemented strain (Fig 3E).

Quantitative proteomic profiling reveals significant differences in the cell envelope and membrane vesicles upon loss of MlaA
We hypothesized that the enhanced fitness of the ΔmlaA strain may be due to increased MV formation, as well as specific changes to CE and MV subproteomes (Fig 9A). To quantitatively examine these alterations, we utilized Tandem Mass Tag (TMT) 6plex isobaric mass tags to label trypsinized proteins from CE and MV fractions collected from WT and ΔmlaA bacteria on two separate occasions. Proteins were considered differentially expressed in the mutant if the ratio of ΔmlaA protein abundance to WT protein abundance was greater than 1.5-fold (increased in ΔmlaA) or less than 0.67-fold (decreased in ΔmlaA). Differentially expressed proteins in CE and MV fractions are listed in Table 3 and Fig 9C and 9D, and proteomic data are presented in S2 and S3 Files, respectively. All MS data are also available via ProteomeXchange with identifier PXD008673.
Of the 884 CE proteins identified in both replicates (S2 File), 21 and 22 were differentially abundant in the mutant in the first and second experiments, respectively, with 6 common proteins between the two experiments (Fig 9B, left, and Fig 9C). In this group, five proteins showed increased levels and one protein was decreased in the mutant. As expected from our previous proteomic mining of MVs [22], fewer proteins were present in this sub-proteome fraction (568; S3 File). However, the MV protein profile was more dramatically affected by the loss of MlaA than the CE proteome, similar to the qualitative results of coomassie staining (Fig 9A). In the first and second biological replicate, 89 and 81 proteins were differentially expressed in the mlaA null strain, respectively, with 23 proteins in common (Fig 9B, right, and  Fig 9D). Of the common differentially expressed proteins, 21 increased in abundance in the absence of MlaA, and two proteins showed decreased levels. Additionally, the amounts of NGO2119 (MlaC) were augmented in the MVs isolated from ΔmlaA, whereas a corresponding increase was not observed in the CE fraction. Importantly, this result demonstrated that no immediate upstream or downstream effects were introduced to the NGO2116-NGO2124 operon (Fig 2C) during construction of the ΔmlaA mutant.
Subsequently, we performed an immunoblot analysis of CE and MV fractions isolated from WT and ΔmlaA using available antisera against eight outer membrane proteins including MlaA; BamA, D, and E; LptD; MtrE; TamA; and NGO2054 [25,76]. As expected, MlaA was not detected in the CE or MVs of ΔmlaA. None of the other proteins examined were differentially regulated in the ΔmlaA CE or MVs (Fig 9E), which further corroborated the quantitative proteomic investigation (S2 and S3 Files).
One of the proteins that was most highly abundant in the ΔmlaA CE compared to WT was the tRNA adenosine deaminase TadA. This essential protein catalyzes an adenosine-to-inosine transition in RNA. A recent report provided the first evidence of mRNA modification in prokaryotes and demonstrated that the modifications were mediated by TadA, although TadA was originally thought to act exclusively on tRNA. E. coli TadA modifies over 250 proteins, including cell envelope proteins BamA, TamA, and PldA [77]. The tyrosine-to-cysteine transitions induced by TadA modification altered the toxicity of one of the target proteins, the toxin HokB [77], which may suggest that the biological activities of gonococcal TadA substrates are affected by TadA upregulation in the ΔmlaA mutant.
The increased presence of protein products with homology to pilS (NGO10975, NGO11150, and NGO07995; Table 3) in the mutant CE and MVs may indicate that pilin antigenic variation is enhanced in mlaA null bacteria. pilS cassettes are transcriptionally silent loci involved in antigenic variation of PilE through RecA-mediated recombination [54]. Also upregulated in the ΔmlaA CE and MVs was the opacity-associated protein Opa54 (NGO04980; OpaF). Opa proteins interact with members of the carcinoembryonic antigen-related cell adhesion molecule (CEACAM) family of receptors to mediate bacterial attachment to epithelial cells, neutrophils, and monocytes [56][57][58][59].
In addition to pilus-related proteins, several other potential virulence factors within the MVs of the ΔmlaA mutant were elevated. Across the two experiments, the Neisseria adhesin complex protein (ACP) was present in higher quantities in the mutant MVs by an average of 1.7-fold. ACP is involved in meningococcal adhesion to human cells [67] and possesses a secondary function as a lysozyme inhibitor [66]. Originally discovered in a screen for gonococcal hemolysins [68], Gly1ORF1 (NGO0358) was also increased 1.7-fold in the ΔmlaA MVs. The ngo0358 locus is annotated as a putative chitinase. While overexpression of gly1ORF1 has not been studied in N. gonorrhoeae, chitinases play roles in the virulence of other bacterial species by binding to the N-acetyl-glucosamine moiety of glycolipids and glycoproteins [69].
Together, our studies are the first to address the expression and function of MlaA in gonococcal physiology and pathogenesis, revealing that the loss of MlaA not only affects colony morphology, CE and MV subproteomes, resistance to certain antimicrobials, and MV formation, but also increases the fitness of N. gonorrhoeae under in vitro conditions relevant to infection and in the murine female genital tract. Cumulatively, our findings highlighted a new mechanism of N. gonorrhoeae pathogenesis and a better understanding of the function(s) of MlaA in this clinically relevant pathogen.

Discussion
The asymmetry of the Gram-negative outer membrane is responsible for its formidable barrier function. When this asymmetry is disrupted, phospholipids accumulate in the outer leaflet and the cell becomes more vulnerable to both hydrophilic and lipophilic antimicrobial compounds  [3,6]. Thus, bacteria have developed several systems to maintain the asymmetry of the outer membrane. Two systems, PagP and PldA, destroy outer-leaflet phospholipids, while the Mla system removes intact phospholipids and re-integrates them into the inner membrane [6]. MlaA, located within the outer membrane in complex with OmpF, forms a channel through which the head groups of phospholipids are able to travel [10]. Our bioinformatics searches did not identify a PagP homolog in the gonococcal genome, indicating that N. gonorrhoeae relies solely on PldA and the Mla system to maintain phospholipid homeostasis. This aspect of bacterial physiology has not been well studied in Neisseria, and reports on PldA mainly focus on its role during autolysis [78][79][80]. Bioinformatic analyses revealed the existence of two separate classes of MlaA homologs (Fig 2A). N. gonorrhoeae MlaA is a member of the class composed primarily of proteins lacking a lipoprotein signal peptide, with the exception of K. pneumoniae and F. tularensis. In vivo assessments with MlaA (VacJ) knockout strains revealed virulence defects in members of the lipoprotein-containing phylogenetic cluster: S. flexneri, H. parasuis, and S. enterica enterica serovar Typhimurium [11,[13][14][15]. In contrast, P. aeruginosa deficient in VacJ, which lacks a lipoprotein signal peptide, was more virulent than WT bacteria [12]. The opposing results suggest that MlaA may play a different moonlighting role(s) in certain aspects of pathogenesis for bacteria, depending on its association with the CE.
We also provide the first report of differences between the genetic location and organization of mlaA in N. gonorrhoeae, N. meningitidis, and N. lactamica compared to those in other bacteria. In the Neisseria species examined, mlaA appears to be part of a polycistronic operon composed of the other components of the Mla system, while mlaA is physically separated from the other Mla members in E. coli (Fig 2C and S2 Fig). Therefore, Neisseria may employ a different regulation strategy for MlaA. Due to its genomic organization, complementation of individual mla components is technically challenging, if not entirely impossible, without a greater understanding of mla system regulation. This hindrance was illustrated by low levels of MlaA in the ΔmlaA/P lac ::mlaA strain regardless of the high amount of the inducer used ( Fig  3E, S4 Fig). The little amount of MlaA was sufficient to return mutant colonies to nearly WT size during polymyxin B exposure (Fig 6B, 6D and 6E), but was not able to complement the competitive infection phenotype (Fig 8A and 8B; S6 Fig). The difficulty of complementing knockouts in a polycistronic operon, even in a fairly well-studied system in which many of the regulatory elements are already known, has been recognized [81][82][83]. For instance, in the soil bacterium Agrobacterium tumefaciens, a virulence plasmid, pTi, carries an operon composed of 11 virB genes that are each essential for tumor formation in plants [84][85][86]. A complementation study to examine each member of the virB operon revealed additional regulatory requirements for six of the 11 genes, including switching the constitutive (in the A. tumefaciens system) lacZ promoter for the native virB promoter for VirB1 and VirB2, adding 55 or 230 bp of upstream sequence for VirB6 and VirB9, or co-expressing VirB7 and VirB8 on the same complementation plasmid [86]. Mla is a relatively recently discovered protein complex and limited information is available regarding expression of the Mla components, including MlaA [10,13]. To address this gap, we examined MlaA expression patterns throughout the growth of WT N. gonorrhoeae in liquid media, during exposure to host-relevant conditions, and in a panel of diverse Neisseria . We demonstrated that the amount of MlaA was decreased in the absence of iron and increased during anaerobic growth (Fig 5C and 5G). The down-regulation in the absence of iron is consistent with the Fur-dependent vacJ transcription pattern in H. influenzae during in vitro growth and in vivo in a mouse model of nasopharyngeal colonization [13]. While MlaA has not been described as part of the N. gonorrhoeae iron regulon, transcription of MlaF (NGO2116) decreases under iron restriction in a Fur-dependent manner [87], which suggests that the entire Mla system may be under Fur transcriptional control in the gonococcus. Our investigation into the effects of Fur on MlaA suggest that Fur's influence has diminished over the intervening five genes. Free iron concentrations are maintained in the attomolar range (10 −18 M) in bodily fluids to prevent microbial growth, and infection induces lactoferrin secretion to further restrict iron availability [88], including in the reproductive tract. Transcriptomic studies of gonococci isolated from active infections have shown that mlaA is transcribed during cervical infections after recent exposure to an infected male partner, and mlaA transcripts are typically lower during infection than during growth in defined medium [89]. Together, these lines of evidence suggest that the downregulation of mlaA in response to iron restriction, but not the complete abrogation of expression, is part of the infection strategy employed by the gonococcus.
Overexpression of the phospholipase PldA in the ΔmlaA background reduced MV formation but strongly enhanced bacterial vulnerability to polymyxin B (Figs 7C and 6B and 6C, respectively). An identical experiment has not been performed in E. coli. Differences in phospholipid composition between N. gonorrhoeae and E. coli explain our results (Table 4). E. coli membranes contain~87% phosphatidylethanolamine (PE) and 4.7% phosphatidylglycerol (PG) in late-stage cultures [90], while PE and PG comprise 69% and 19% of the membrane of N. gonorrhoeae, respectively [91]. In both species, PldA predominantly destroys PE [79,92]. In contrast, the Mla system does not appear to exhibit a preference for specific phospholipids [93]. In N. gonorrhoeae, only the Mla system and PldA contribute to outer membrane asymmetry as no PagP homolog exists. Thus, in our proposed model, in the absence of MlaA, with native PldA levels, the overall phospholipid composition is largely preserved [13], but the outer membrane will contain phospholipids that have invaded the CE outer leaflet (Fig 1B). PldA dimerization and activation will likely increase upon detection of perturbed phospholipid homeostasis [8], although evidence from E. coli suggests that PldA expression will not be altered in the absence of membrane stress [6]. The enhanced activation of endogenous PldA levels may alter the phospholipid profile slightly. However, upon PldA overproduction in the ΔmlaA mutant, the substrate preference of PldA towards PE will result in a higher abundance of PG within the surface of N. gonorrhoeae (Fig 1C; Table 4). Polymyxin B preferentially targets PG [94,95]. Gonococci overexpressing PldA in the absence of MlaA are therefore more sensitive to the antimicrobial peptide.
Downregulation and deletion of VacJ in H. influenzae and V. cholerae increased MV formation, and accumulation of phospholipids in the outer leaflet has been proposed as a general mechanism of MV biogenesis [13]. Corroborating this suggestion, the N. gonorrhoeae ΔmlaA strain produced significantly more MVs than WT bacteria and PldA overproduction partly reduced their amounts (Fig 7C). However, dramatic increases in MV release for both WT and ΔmlaA bacteria under polymyxin B stress (Fig 7C) provided evidence that invading phospholipids are not the primary factor in MV biogenesis.
Why do we observe increased fitness of the ΔmlaA during competitive infection in the surrogate host? We propose that it is due to the altered CE and MV protein contents (Fig 9A). Quantitative proteomics of these fractions revealed alterations in proteins with significant implications in bacterial physiology and virulence (Fig 9B, 9C and 9D). Among the most upregulated proteins in the ΔmlaA CE was an essential enzyme, TadA, which edits mRNA without changing its expression and recodes protein sequences, potentially affecting their function and cell physiology [77]. Our proteomic profiling revealed a previously unrecognized link between phospholipid homeostasis and RNA editing, which should be examined in future studies, in addition to the role of TadA-mediated protein modification in bacterial pathogenesis. Further, several adhesins were also elevated in the CE and MV fractions of the ΔmlaA mutant (Fig 9C  and 9D, Table 3). Of these, adherence mediated by pili and Opa proteins is host-restricted, although a CEACAM receptor-independent advantage is observed for Opa-positive gonococci in the mouse model [51,96]. Elevated Opa levels in the ΔmlaA mutant may therefore have contributed to the mutant's enhanced fitness. Heightened pilin antigenic variation in the absence of MlaA may also have played a role. However, one or more of the other proteins may contribute more substantially to the increased fitness of the mutant in the mouse model ( Fig  8B), possibly through elevated ACP acting as a lysozyme inhibitor [97,98] within the MVs [56][57][58][59]67]. MVs have been shown to act as decoys against components of the innate immune system, by adsorbing antimicrobial peptides, for example [99], and are often packages for virulence factors [100]. The ΔmlaA mutant appears to exploit both of these functions, as illustrated by the presence of heightened levels of ACP within MVs, which could protect the bacteria from lysozyme attack [66,97]. We have not assessed lysozyme sensitivity in the ΔmlaA mutant, as two gonococcal lysozyme inhibitors, ACP and SliC, are responsible for lysozyme resistance and compensate for each other in vitro [97,98]. However, lack of SliC was sufficient to significantly impact bacterial fitness in the mouse genital tract [97]. Thus, the loss of MlaA results in multifactorial alterations to gonococcal CE and MV protein composition that likely act in concert to enhance the fitness of the ΔmlaA mutant in the mouse model. The primary function of MlaA is in cell envelope biogenesis and homeostasis. However, alterations to the MlaA cellular pool in response to host conditions would modulate the CE and MV protein composition, as well as the amount of blebs released. Our study presents MlaA as an intriguing protein with an inverse effect on gonococcal pathogenesis. We propose that N. gonorrhoeae adjusts virulence through changes in MlaA levels in response to iron availability in the host, as both the human [88] and murine [101] genital tracts are iron-restricted. Downstream effects of MlaA depletion involve alterations to the CE and MV protein composition, including elevated levels of known and potential virulence factors, as well as the possibility of post-translational protein modification. It is exciting to consider that other host/bacterial effectors may also be involved in MlaA-triggered MV biogenesis and in fine-tuning the bacteria-host interaction.

Bacterial strains and growth conditions
For this study, laboratory strain FA1090 [102] and contemporary clinical isolate WHO X [26] were used as wild type (WT) strains. Isogenic null mutants of ngo2121 were previously constructed in both strain backgrounds by replacement of the locus with the kanamycin resistance cassette [22,23], and a complemented strain with ngo2121 under the control of the P lac promoter was also previously constructed in the FA1090 strain background [23]. For this study, we also generated a conditional knockout of the Fur transcriptional regulator, Δfur/P lac ::fur, described below. To analyze expression of MlaA in clinical strains, we used a panel of isolates collected from two public health clinics in Baltimore from 1991 to 1994 (LGB1, LG14, LG20, LG26); the Public Health-Seattle & King County Sexually Transmitted Disease clinic from 2011 to 2013 (UW01-UW13); and the 2016 WHO reference strains [25]. Frozen stocks of N. gonorrhoeae strains stored at -80˚C were streaked onto gonococcal base agar solid medium (GCB, Difco) supplemented with Kellogg's supplements I and II diluted 1:100 and 1:1,000, respectively [25]. After incubation for 18-20 h at 37˚C in a 5% CO 2 atmosphere, colonies were subcultured onto GCB. Transparent piliated cells were used for transformation, while transparent, non-piliated variants were used for all other experiments. Unless otherwise specified, after 18-20 h of subsequent culturing as described above, colonies were collected from solid medium with a polyester-tipped sterile applicator (Puritan) and suspended to a final OD 600 of 0.1 in gonococcal base liquid medium (GCBL) supplemented with 0.042% sodium bicarbonate and Kellogg's supplements as above. Bacteria were then cultured at 37˚C with agitation (220 rpm) for 3 h, diluted to a final OD 600 of 0.1 in supplemented GCBL, and cultured as previously.
E. coli NEB5α was used for genetic manipulations, while E. coli BL21(DE3) was used for heterologous expression of MlaA and DsbA for protein purification. E. coli was maintained at 37˚C in Luria-Bertani (LB, Difco) media or LB agar supplemented with the appropriate antibiotic. Concentrations of antibiotics used are as follows: N. gonorrhoeae kanamycin (40 μg/mL for FA1090, 50 μg/mL for WHO X) and erythromycin (0.5 μg/mL for FA1090, 4 μg/mL for WHO X); E. coli kanamycin (50 μg/mL).
The gene encoding dsbA missing the signal peptide was amplified using primers DsbA-f, GACTCCATGGTGACGGAAGGGGAAGACT and DsbA-r, GACTAAGCTTCGGCTATTT CTGTACAGCAG. The subsequent PCR product was digested with NcoI and HindIII, as indicated, and ligated into similarly cut pRSF-NT to create a tobacco etch virus protease-cleavable C-terminal-6 × His-tagged fusion. PldA was amplified using primers PldA-F, GATCTTAATT AAAAAATGCCGTTTGAAAACCAATATG and PldaA-R, CGGTATTCAGATGCCGTC. PCR product was digested with PacI, ligated into pGCC4 digested with PacI and PmeI (New England Biolabs; PacI restriction site underlined), and transformed into E. coli NEB5α. After sequence verification, pGCC4-PldA was transformed into piliated ΔmlaA as previously [22].
The conditional fur knockout was created using a strategy described previously [25]. An additional copy of the fur gene was initially placed under the control of an IPTG-inducible promoter within the intergenic region between lctP and aspC in the FA1090 chromosome. fur was subsequently replaced in its native chromosomal locus with the nonpolar kanamycin resistance cassette. Specifically, the fur gene was amplified using primers fur-F, GATCTTAATTAA TTACAGGGATATTGAATATTATGGAAAAATTC and fur-R, ACAACAAACCGTCCGGA T. The PCR product was cut with PacI and ligated into PacI/PmeI cut pGCC4 to create pGCC4-fur. The resulting plasmid was used for transformation of FA1090 to create FA1090P lac ::fur. To generate the knockout plasmid, pNEB193 was linearized by PCR using primers pNEB-F, GTTTAAACCTGCAGGCATGCAAG and pNEB-R, TCTAGACTTAATTA AGGATCCGGCG. One thousand bp upstream and downstream of fur were amplified using primers fur-up-F, CGCCGGATCCTTAATTAAGTCTAGATTTCGGCGGCGGCGC; fur-up-R, TCTGAGCGGGACTCCCTGTAAAATAATAGACGCTATAATACGCAATTTCAG; furdown-F, GTGAAGCTAGCTATCCGGACGGTTTGTTGTTCAG; and fur-down-R, AGCTT GCATGCCTGCAGGTTTAAACTTTCCAATTTTATAGTGGATTAAATTTAAACCGGTA CGG. The kanamycin resistance cassette was amplified using primers kan-F, TATTTTACA GGGAGTCCCGCTCAGAAGAACTCG and kan-R, AAACCGTCCGGATAGCTAGCTTC ACGCTGCC. All fragments were purified and assembled using NEBuilder HiFi DNA Assembly Master Mix (NEB). pNEB193-Δfur was linearized with ScaI and used to transform FA1090P lac ::fur. Clones were selected on solid medium supplemented with kanamycin and 0.1 mM IPTG and verified by PCR.
All plasmid insert sequences were confirmed by the Oregon State University Center for Genome Research and Biocomputing.

Protein purification
Truncated, recombinant 6 × His tagged MlaA (MlaA 120-277 ) was expressed in E. coli BL21 (DE3) harboring pET28-MlaA 120-277 . An overnight culture was used to inoculate 1.5 L LB medium supplemented with kanamycin. Protein overproduction was induced with 500 mM IPTG when cultures had reached OD 600~0 .5. After 3 h of growth at 37˚C, cells were pelleted by centrifugation (6,000 rpm for 20 min at 4˚C) and stored at -80˚C until use. Pellets were resuspended in lysis buffer (20 mM Tris pH 8.0, 500 mM NaCl, 10 mM imidazole, 1% Triton-X) in which a Pierce protease inhibitor mini tablet (ThermoFisher Scientific) had been dissolved. Cells were lysed by six passages through a French Press pressure cell at 1,200 psi. Crude lysate was clarified by centrifugation, incubated with DNAseI, and filtered through a 0.45 μm filter. Protein was purified on a Bio-Rad NGC scout system through a Bio-Rad Bio-Scale Mini Nuvia IMAC cartridge (Bio-Rad). Non-specifically bound proteins were washed off the column by 10 column volumes of wash buffer (20 mM Tris pH 8.0, 500 mM NaCl, 40 mM imidazole, 1% Triton-X). Protein was eluted over a 40-250 mM imidazole gradient and fractions were monitored for protein purity by SDS-PAGE. Triton was removed from protein solution by 2 h of end-over-end incubation with Bio-Rad Bio-Beads SM-2 adsorbent polystyrene beads (Bio-Rad) at room temperature. Glycerol was added to a final concentration of 10% and protein was stored at -80˚C.
For the overproduction of DsbA, 1 L of LB supplemented with kanamycin was inoculated with E. coli BL21(DE3) carrying pRSF-dsbA. Protein overproduction was induced by the addition of 1 mM IPTG when the culture reached an OD 600 of about 0.5. After 3 h of growth at 37˚C, bacterial cells were pelleted by centrifugation at 6,000 × g for 20 min at 4˚C and cell pellets were stored at -80˚C until used. Recombinant DsbA was purified under native conditions using the NGC Scout Chromatography system (Bio-Rad). Bacterial cells were resuspended in lysis buffer (20 mM Tris pH 8.0, 500 mM NaCl, 10 mM imidazole) and lysed by five passages through a French pressure cell at 1,200 psi. Cell lysates were clarified by centrifugation and loaded onto a Bio-Scale Mini Profinity IMAC cartridge (Bio-Rad). Loosely bound proteins were removed with 10 column volumes of wash buffer (20 mM Tris pH 8.0, 500 mM NaCl, 40 mM imidazole) and DsbA was eluted with a 40-250 mM imidazole gradient. To remove the 6 × His tag the eluted DsbA was incubated overnight at 4˚C with TEV protease at a 1:20 ratio. The proteins were concentrated to 5 mL using a Vivaspin 20 centrifuge concentrator (GE HealthCare). Proteins were subjected to size exclusion chromatography using HiLoad 16/600 Superdex 75 pg (GE HealthCare) and phosphate buffered saline (PBS) as running buffer. Fractions containing DsbA were concentrated using a Vivaspin 20 centrifuge concentrator (GE HealthCare).

Antisera preparation
Polyclonal rabbit antisera were prepared by Pacific Immunology using about 2 mg of purified MlaA 120-277 or DsbA. A 13-week antibody production protocol was approved by IACUC Animal Protocol #1 and the National Institute of Health Animal Welfare Assurance Program (#A4182-01), and was performed in a certified animal facility (USDA 93-R-283).

Exposure to host-relevant conditions
FA1090 (WT) and ΔmlaA bacteria cultured as above were collected from rapidly-growing liquid cultures, diluted to OD 600 = 0.2, serially diluted, and spotted on GCB plates supplemented with 7.5% NHS or 5 μM of the iron chelator deferoxamine mesylate (desferal). Inoculated plates were incubated at 37˚C in a 5% CO 2 environment for 22 h or anaerobically at 37˚C in the presence of 1.2 mM sodium nitrite as a terminal electron acceptor, and either 0, 5, 10, 20, or 25 μM desferal for immunoblotting, for 48 h. The anaerobic environment was generated using BD GasPaks (BD) in an anaerobic jar. Colonies were counted for colony forming unit (CFU)/mL enumeration. Bacteria were collected from plates for immunoblotting analysis.
Desferal titration was performed in liquid media. Non-piliated FA1090 WT or ΔmlaA were suspended to OD 600 = 0.1 in GCBL supplemented with Kellogg's supplement I and sodium bicarbonate, but without Kellogg's supplement II. Strains were cultured at 37˚C for 3 h with agitation (220 rpm). Cultures were diluted to OD 600 = 0.1 in media supplemented with Kellogg's supplement I and sodium bicarbonate and split. Daughter cultures were supplemented either with Kellogg's supplement II or Desferal (5 to 25 μM). Suspensions were cultured at 37˚C with agitation for 6 h. Every hour, including 0 h, samples of each culture were serially diluted. Five microliters of each dilution were spotted onto a GCB plate, which was subsequently cultured at 37˚C for 18-20 h at 5% CO 2 . Colonies were counted for CFU/mL enumeration after incubation. Samples of WT and ΔmlaA liquid cultures with different desferal concentrations were taken at the 6 h timepoint and subjected to immunoblot analysis.

MlaA Fur regulation assessment
MlaA regulation by Fur was evaluated using the conditional Fur knockout strain Δfur/P lac ::fur. Initial liquid cultures of WT FA1090, ΔmlaA, and Δfur/P lac ::fur were prepared without Kellogg's supplement II. Fur expression was induced by the addition of 0.1 mM IPTG. After three hours of growth, Δfur/P lac ::fur cultures were washed by centrifugation at 5,000 × g for 5 minutes to remove IPTG, the supernatant was decanted, and bacteria were resuspended in fresh medium without IPTG. All strains were back diluted to an OD 600 of 0.1 and supplemented with either Kellogg's supplement II or 25 μM desferal. Fur expression was induced at different levels by the addition of 10, 50, or 100 μM IPTG under both iron-deplete and-replete conditions. Bacteria were cultured for 6 h at 37˚C with shaking (220 rpm), and growth was monitored every hour by OD 600 measurement. At the experimental endpoint, samples of each culture were collected and centrifuged at 6,000 × g for 10 minutes. The supernatant was discarded and pellets were prepared for SDS-PAGE and immunoblotting analysis.

Serum sensitivity
Assessment of bacterial serum sensitivity was essentially as previously described [37]. Nonpiliated strains as indicated in the text were suspended to OD 600 = 0.1 in supplemented GCBL and cultured for 3 h at 37˚C with shaking (220 rpm). Bacterial cultures were suspended to OD 600 = 0.05 in sterile PBS, then diluted 1:1000 in Eagle's minimal essential medium (EMEM). Seventy microliters of this suspension, corresponding to approximately 10 3 CFU, was added to 70 μL NHS (Quidel, San Diego, CA; Ref# A113), heat-inactivated NHS (inactivated at 56˚C for 30 min), or EMEM alone, for final serum concentrations of 0% or 50% in a 96 well microtiter plate (initial testing indicated that bacterial survival was not altered in the presence of 8%, 10%, 12%, or 24% NHS). Plates were incubated at 37˚C in 5% CO 2 for 1 h, after which 20 μL from each well were spotted onto a GCB plate. CFU/mL were enumerated following 18-20 h incubation as above.
Exposure to human defensin HBD1 (AnaSpec Inc. 510791-9560) was solubilized to 100 μM in 0.01% acetic acid. Rapidly growing liquid cultures of WT and ΔmlaA bacteria were diluted to 10 5 cells/mL, and 45 μL of this suspension were transferred to a 1.5 mL tube. Volumes were adjusted to 50 μL with either GCBL, 0.01% acetic acid (vehicle), or HBD1 (final concentration, 10 μM). Bacteria were incubated for 3 h at 37˚C, serially diluted, and spotted onto GCB. CFUs were enumerated following 22 h incubation, and the relative survival was calculated. Experiments were performed on at least four independent occasions.

Etest antimicrobial sensitivity testing
Antimicrobial susceptibility was assessed by Etests according to manufacturer's instructions. Briefly, non-piliated colonies of FA1090 or WHO X WT, ΔmlaA bacteria in either strain background, or ΔmlaA/P lac ::pldA in the FA1090 knockout background were collected from GCB plates and suspended in brain heart infusion liquid medium to a turbidity equivalent to that of a 0.5 McFarland standard. Suspensions were spread on 150 mm tissue culture dishes containing 50 mL GCB solid medium (~4 mm thick), and test strips were laid on the surface of the plates. 0.5 mM IPTG was added to ΔmlaA/P lac ::pldA plates. MICs were measured after approximately 22 h of incubation as above. Experiments were performed on three occasions and consensus MICs from at least two out of the three replicates are presented.

Sensitivity to polymyxin B
FA1090 WT, ΔmlaA, ΔmlaA/P lac ::mlaA, and ΔmlaA/P lac ::pldA were suspended to OD 600 = 0.1 in GCBL supplemented as above and cultured for 3 h at 37˚C with agitation (220 rpm). Cultures were subsequently diluted to an OD 600 of 0.1 in supplemented GCBL and cultured at 37˚C with agitation for 2 h. Bacteria were serially diluted and spotted on plates with or without 800 U/mL polymyxin and 0, 0.05, 0.1, or 0.5 mM IPTG. Plates were cultured at 37˚C in a 5% CO 2 environment for approximately 18 h. Overall colony morphology was documented with a Chemi-Doc MP (Bio-Rad). Colonies were counted to determine overall survival under polymyxin B exposure. Images of colonies were taken with a Zeiss AxioObserver.D1 microscope at 10× magnification 0.25 Phase Contrast 1 and at 2.5× magnification 0.06 Phase Contrast 1. Colony diameters from images taken at 2.5× magnification were measured using ImageJ software. Results were similar for all induction levels. For clarity, % survival data is presented only for the 0.5 mM IPTG level. Microscopy for the complemented strain was performed with 0.1 mM IPTG.

LOS isolation and silver staining
LOS was isolated from WT, ΔmlaA, ΔmlaA/P lac ::mlaA, and ΔmlaA/P lac ::pldA using a method described previously [49]. Strains were harvested from either standard liquid cultures or cultures containing 100 U/mL polymyxin B by suspension in 1.5 mL GCBL to an OD 600 of 0.2 and centrifugation at 15,000 × g for 1.5 min. The expression of MlaA in the ΔmlaA/P lac ::mlaA strain was induced with 0.5 mM IPTG, and ΔmlaA/P lac ::pldA was cultured either with or without 0.5 mM IPTG, as indicated. Cell pellets were resuspended in 50 μL lysis buffer (2% SDS, 4% β-mercaptoethanol, 10% glycerol, 1 M Tris pH 6.8, and 0.01% bromophenol blue) and boiled at 100˚C for 10 minutes to lyse bacteria. After samples were allowed to cool to room temperature, proteins were digested by proteinase K (25 μg in 10 μL lysis buffer) at 60˚C for 1 h. SDS-PAGE was performed on isolated LOS using an 18% gel, and LOS was visualized by silver staining [49].

Supernatant protein profiling
Membrane leakiness was assessed as previously described [23], using filtered supernatants from mid-logarithmic liquid cultures. Soluble supernatant proteins were precipitated with a pyrogallol red-molybdate-methanol procedure after DNAse I treatment. Membrane vesicles were not separated prior to precipitation. Precipitated proteins, standardized by OD 600 values of the source cultures, were separated by SDS-PAGE and analyzed by immunoblotting. Whole cell lysates of source cultures were analyzed simultaneously.

Electron microscopy
Strains indicated in the text were cultured in GCBL as above, either with or without 100 U/mL polymyxin B, until approximately mid-logarithmic growth (OD 600 of 0.5). The ΔmlaA/P lac :: mlaA and ΔmlaA/P lac ::pldA strains were cultured in the presence (both strains) or absence (ΔmlaA/P lac ::pldA only, as indicated) of 0.5 mM IPTG. Bacteria were collected by centrifugation at 4000 × g for 3 min, washed twice and resuspended with PBS sterilized by filtration through a 0.1 μm filter. 2.5 μL of suspension were spotted onto 300 mesh copper grids and cells were allowed to attach to the grid for 15 min before excess PBS was removed. Cells were negatively stained with phosphotungstic acid and visualized with a FEI Helios NanoLab 650 electron microscope housed at the Oregon State University Electron Microscopy Facility. Experiments were performed at least twice, and representative micrographs are presented.

Subcellular fractionation
C/P, CE, MV, and SS fractions were isolated as described previously [22,25]. Briefly, strains as indicated in the text were suspended in 500 mL GCBL supplemented as above and cultured at 37˚C with agitation (220 rpm) until cultures reached OD 600 of~0.8. Where indicated in the text, cultures were also supplemented with 100 U/mL polymyxin B. PldA overproduction was induced by the addition of 0.5 mM IPTG. Supernatants were separated from bacteria by low speed centrifugation and filtration and treated with DNaseI and protease inhibitors. Naturally released MVs were isolated by high-speed ultracentrifugation. Supernatants of MV isolation contain soluble supernatant proteins, which were precipitated by incubation with 15% trichloroacetic acid at 4˚C for 1 h and centrifugation at 14,000 × g for 20 min at 4˚C. Precipitated proteins were washed with ice-cold acetone and centrifuged at 14,000 × g for 20 min at 4˚C. MV and SS fractions were resuspended in PBS + 0.8% SDS for SDS-PAGE, immunoblotting, and quantitative proteomics analyses as indicated. The CE was isolated by cold sodium carbonate extraction and ultracentrifugation [22,25], and suspended in PBS + 0.8% SDS. Samples of supernatants from CE isolation were collected for the cytoplasmic/periplasmic fraction.

Quantitative proteomics
The quantitative proteomics studies were performed as we described previously [103]. Briefly, CE or MV were isolated from WT and ΔmlaA bacteria when the OD 600 of each culture reached 0.6-0.8, as described above. The total protein amount in each CE and MV fraction was assessed using a Protein Assay Kit (Bio Rad) and were subjected to labeling with TMT6plex reagent (Thermo Scientific) according to manufacturer recommendations. Eighty μg of proteins were placed in a 1.5 mL tube and the volume was adjusted to 100 μL using 100 mM triethylammonium bicarbonate (TEAB) buffer. Proteins were reduced by addition of 11.3 mM tris(2-carboxyethyl)phosphine hydrochloride (TCEP) and incubation at 55˚C for 1 h, and alkylated by addition of 20.1 mM iodoacetamide for 30 min in the dark. Proteins were precipitated overnight in 90% acetone at -20˚C, spun down at 15,000 × g for 10 min at 4˚C and washed once with ice cold acetone. Protein pellets were resuspended in 100 μL of 100 mM TEAB and digested with trypsin at a 1:40 ratio overnight at 37˚C. TMT Reagents were reconstituted in 41 μL of acetonitrile (ACN) and were added to the digested samples. Wild type CE samples were labeled with reagent TMT126 and TMT128 for replicates 1 and 2, respectively. CE proteins isolated from ΔmlaA were labeled with TMT127 and TMT129 for replicates 1 and 2, respectively. To label the MV proteins, TMT128 and TMT130 were used for wild type and TMT129 and 131 for ΔmlaA. The reaction was allowed to proceed for 1 h at room temperature and was quenched by addition of 8 μL of 5% hydroxylamine. Samples were pooled and subsequently fractionated by strong cation exchange (SCX) with a Paradigm (Michrom Biosciences) HPLC using 5 mM potassium phosphate monobasic in 30% ACN/70% water (v/v) pH 2.7 (buffer A) and 5 mM potassium phosphate monobasic in 30% ACN/70% water (v/v) pH 2.7 with 500 mM potassium chloride (buffer B) as the mobile phases. The sample was brought up in 200 μL buffer A. Peptides were separated over 60 min with a 2.1 mm x 100 mm Polysulfoethyl A column (PolyLC) at a 200 μL/min flow rate using the following separation profile: hold 2% B for 5 min, 2% to 8% B in 0.1 min, 8% to 18% B in 14.9 min, 18% to 34% B in 12 min, 34% to 60% B in 18 min, 60% to 98% B in 0.1 min and hold for 10 min. We collected 1 min fractions in 96-well microtiter plates. Twelve pools were generated from 60 fractions, and the pools were dried with a speed vac. Oasis HLB 1cc cartridges were subsequently used to desalt the samples. Cartridges were initially washed with 70% ACN/0.1% trifluoroacetic acid (TFA), then equilibrated with 0.1% TFA. Fractionated samples were hydrated in 0.1% TFA to load onto the cartridge, washed with 0.1% TFA, eluted in 1 mL 70% ACN/0.1% TFA, and dried by vacuum centrifugation.
Desalted samples were subsequently analyzed by LC/ESI MS/MS with a Thermo Scientific Easy-nLC II (Thermo Scientific, Waltham, MA) nano HPLC system coupled to a hybrid Orbitrap Elite ETD (Thermo Scientific, Waltham, MA) mass spectrometer at the Proteomic Core at The Fred Hutchinson Cancer Research Center, Seattle, WA. Samples were further desalted in-line using a reversed-phase trap column (100 μm × 20 mm) packed with Magic C 18 AQ (5 μm 200 Å resin; Michrom Bioresources) and peptides were separated on a reversed phase column directly mounted to the electrospray ion source [75 μm × 250 mm column packed with Magic C 18 AQ resin (5 μm 200 Å resin; Michrom Bioresources)]. Chromatographic separations were performed at a flow rate of 400 nL/min to apply a 90 min gradient of 7% to 35% ACN in 0.1% formic acid. The capillary temperature was set to 300˚C, and a 2750 V spray voltage was applied to the electrospray tip. The Orbitrap Elite instrument's data dependent mode was used to switch automatically between MS survey scans [automatic gain control (AGC) target value 1,000,000; resolution 120,000; and injection time 250 msec] and MS/MS spectra acquisition (AGC target value of 50,000; 15,000 resolution; and injection time 250 msec). From the Fourier-transform full scan, the 15 most intense ions were selected for fragmentation by higher-energy collisional dissociation in the higher-energy C-trap dissociation (HCD) cell. The normalized collision energy was set to 40%. Selected ions were dynamically excluded for 30 sec with a list size of 500 and exclusion mass by mass width +/-10ppm.

Proteomic data analysis
Data analysis was performed using Proteome Discoverer 1.4 (Thermo Scientific, San Jose, CA) and the data were searched against the N. gonorrhoeae FA1090 database (UniProt, downloaded May 11, 2017) with the common Repository of Adventitious Proteins (cRAP,http://www. thegpm.org/crap/) FASTA file. Trypsin was set as the enzyme with 2 mis-cleavages allowed. Variable modifications included TMT6Plex (+229.163 Da) on any N-Terminus, oxidation on methionine (+15.995 Da), carbamidomethyl on cysteine (+57.021 Da), and TMT6Plex on lysine (+229.163 Da). Ten ppm was set as the precursor ion tolerance, and 0.8 Da was established as the fragment ion tolerance.

Experimental design and statistical rationale
All experiments described above were performed using CE and MVs isolated from the N. gonorrhoeae wild type and ΔmlaA mutant in biological duplicates. Data were searched using Sequest HT. All search results were run through Percolator for scoring. Quantification was performed using the canned TMT6plex method through Proteome Discoverer to compare the normalized total reporter ion intensity between WT and ΔmlaA CE or MV fractions for each biological replicate. The mass spectrometry proteomics data have been deposited to the Pro-teomeXchange Consortium via the PRIDE [104] partner repository with the dataset identifier PXD008673. A False Discovery Rate of 1% was applied, and only proteins identified by �1 unique and �2 total peptides with a score � 1 for every detected peptide were included for further analysis. Proteins were considered differentially expressed when the calculated ratios were below 0.67 or above 1.50.

Biofilm characterization
The ability of strains to produce biofilms was assessed using a method adapted from Anderson et al. [105]. Non-piliated bacteria, cultured as above, were collected from plates and suspended to OD 550 = 1.5 in supplemented GCBL. One hundred microliters of this suspension were added to the wells of a 96-well flat-bottomed microtiter plate (Corning #3370). Water was added to all remaining wells to minimize the effects of evaporation. The plate was wrapped in plastic wrap and incubated at 37˚C in a 5% CO 2 environment for 24 h without shaking. After incubation, samples from each well were serially diluted and spotted onto GCB plates for CFU/mL enumeration of viable planktonic bacteria. Planktonic bacteria and spent media were removed and biofilm wells were washed once with PBS. Plates were allowed to dry for~4 h at room temperature, after which 65 μL of 0.1% crystal violet were added to each well and incubated for 15 min at room temperature. Wells were then washed 3 times with PBS and allowed to dry overnight at room temperature. Biofilms were dissolved by the addition of 125 μL 30% acetic acid, incubation at room temperature for 30 min, and shaking on a microplate vortexer. Biofilm mass was measured on a BioTek Synergy HT plate reader (BioTek) at 550 nm.

Competitive infection of the murine lower genital tract
Female BALB/c mice (6 to 8 weeks old; Charles River Laboratories Inc., Wilmington, MA; NCI Frederick strain of inbred BALB/cAnNCr mice, strain code 555) were treated with 0.5 mg of Premarin given two days prior to, the day of, and two days after bacterial inoculation to increase susceptibility to N. gonorrhoeae. Mice were also given antibiotics to suppress the overgrowth of commensal flora that occurs under the influence of estrogen [106]. Groups of mice were inoculated vaginally with similar numbers of WT FA1090 and either isogenic ΔmlaA or ΔmlaA/P lac ::mlaA bacteria (total dose 10 6 CFU; 7 mice/group). Vaginal swabs were collected on days 1, 3, and 5 post-inoculation and suspended in 100 μL GCBL. Swab suspensions and inocula were cultured quantitatively on GCB supplemented with streptomycin (total number of CFUs) and GCB with streptomycin and kanamycin (ΔmlaA, ΔmlaA/P lac ::mlaA; CFU). Results are expressed as the competitive index (CI) using the equation CI = [mutant CFU (output)/wild-type CFU (output)]/[mutant CFU (input)/wild-type CFU (input)]. The limit of detection of 1 CFU was assigned for a strain that was not recovered from an infected mouse. A CI of >1 indicates that the mutant is more fit than the WT strain.

Bioinformatic analyses
The MlaA amino acid sequence was used to search the non-redundant NCBI protein database to search for homologous proteins, and the ClustalOmega online tool was used to align sequences to generate a distance matrix. Signal peptides recognized by Signal Peptidase I or II were predicted with the SignalP 4.1 server (http://www.cbs.dtu.dk/services/SignalP/) or the LipoP 1.0 server (http://www.cbs.dtu.dk/services/LipoP/). MEGA7 software was employed to align sequences with the ClustalW tool for examination of phylogenetic relationships. Subsequently, a maximum likelihood tree was generated in MEGA using the Jones-Taylor-Thornton model to calculate a pairwise distance matrix [107]. To generate the initial tree, Neighbor-Join and BioNJ algorithms were applied to the matrix, and the tree was heuristically searched with the Nearest-Neighbor-Interchange method. Five hundred bootstrap replicates were applied to test the phylogenies, and the lowest log-likelihood tree is presented. The mlaA nucleic acid sequence was used to query the Neisseria Multilocus Sequence typing Database for SNP analysis of the locus (NEIS1933) across the 44,289 isolates deposited as of December 2, 2017. Phylogenetic analyses of MlaA among N. gonorrhoeae and between all Neisseria isolates were performed as above. Amino acid sequences of proteins identified through our quantitative proteomic investigations were used to query the NCBI non-redundant protein database using BLASTp (https://blast.ncbi.nlm.nih.gov/Blast.cgi?PAGE=Proteins) to gain insights into their functions.

Statistical analyses
All statistical analyses were performed using GraphPad Prism software (version 6.0h for Mac OS X) with the exception of proteomic data and animal studies described above. Built-in twoway ANOVA (using Sidak's multiple comparisons test) or unpaired t-test analyses were used to test for statistical significance at p<0.05.

Ethics statement
Animal experiments were conducted at the Uniformed Services University of the Health Sciences (USUHS) according to the guidelines of the Association for the Assessment and Accreditation of Laboratory Animal Care under protocol no. MIC16-488 that was approved by the University's Institutional Animal Care and Use Committee. The USUHS animal facilities meet the housing service and surgical standards set forth in the "Guide for the Care and Use of Laboratory Animals" NIH Publication No. 85-23, and the USU Instruction No. 3203, "Use and Care of Laboratory Animals". Animals are maintained under the supervision of a full-time veterinarian. For all experiments, mice were humanely euthanized by trained personnel upon reaching the study endpoint using a compressed CO 2 gas cylinder in LAM as per the Uniformed Services University (USU) euthanasia guidelines (IACUC policy 13), which follow those established by the 2013 American Veterinary Medical Association Panel on Euthanasia (https://www.usuhs.edu/mps/facilities-resources).
Supplemental Materials and Methods are described in S1 Text.  Table 1 in S1 Text: Amino acid identity of members of the N. gonorrhoeae Mla operon with their E. coli homologs. Table 2 in S1 Text: Agar dilution assessment of WT, ΔmlaA, ΔmlaA/P lac ::mlaA, and ΔmlaA/P lac ::pldA MICs.