To identify genes and pathways regulated by 1132, we carried out genome-wide transcriptional profiling by RNA sequencing of VIR-O Δ1132 vs. wild-type VIR-O, which revealed a total of 74 differentially regulated genes in VIR-O Δ1132 (greater than 2-fold change, p value less than 0.05) ( Supplementary Table 1 ). These results showed that 1132 impacts transcription of a variety of genes involved in regulation, metabolism, protein synthesis, and possibly the cell stress response. Genes that encode ribosomal proteins, RNA polymerase, translation initiation factors, and both transcriptional and translational elongation factors are among genes that are upregulated in VIR-O Δ1132. On the other hand, several genes that are downregulated in the absence of 1132 encode a variety of enzymes involved in oxidative stress protection including catalases, peroxidases and others that interact with glutathione.

Deletion of 1132 results in loss of secretion of the quorum sensing signal 3-OH-C₁₂-HSL (AHL) ( Figure 1B ). This is based on the inability to activate an Agrobacterium tumefaciens traG::lacZ fusion when grown on a soft agar lawn containing this biosensor strain and X-Gal (Niu et al., 2008; Paulk Tierney and Rather, 2019). We utilized a Tn7 transposon system (Ducas-Mowchun et al., 2019a) to provide single-copy complementation of 1132 (VIR-O Δ1132-Tn7/1132), which restored AHL secretion. This restoration is lost again if the Tn7 copy of 1132 is disrupted ( Figure 1B ). The loss of AHL secretion also occurs in the AV-T Δ1132 mutant ( Figure 1C ).

Considering our previously published findings that 1132 overexpression activates the abaI-abaR system (Perez-Varela et al., 2020), we hypothesized that loss of AHL secretion was due to downregulation of abaI when 1132 is deleted. Surprisingly, our RNA sequencing analysis indicated wild-type levels of abaI expression in VIR-O Δ1132, which we confirmed by qRT-PCR ( Supplementary Figure 1 ). This result suggested that the mutant cells synthesize AHL, but do not secrete it. To investigate this possibility, we conducted an assay utilizing the A. tumefaciens biosensor in which 10% SDS was added to a well at the center of the plate ( Supplementary Figure 2 ). Cultures of wild-type VIR-O, an abaI mutant (VIR-O abaI::T26), and VIR-O Δ1132 were then added to the plate in lines going toward the SDS-containing well. As expected, we saw that the wild-type VIR-O cells uniformly activate the biosensor. However, the VIR-O Δ1132 cells show activation of the biosensor only at the point where the VIR-O Δ1132 cells have been lysed by the SDS in the presence of non-lysed biosensor cells, which confirmed our hypothesis that AHL is synthesized but cannot exit the cell. Since there is no activation by an abaI mutant near the SDS, the activating signal in the 1132 mutant is 3-OH-C₁₂-HSL and not a released metabolite. These results indicate a more complicated role for 1132 in quorum sensing than simple regulation of abaI, which is further considered in the Discussion section.

We previously reported that overexpression of 1132 increases surface-associated motility 3.8-fold and that this effect requires both abaI and the ABUW_3766-ABUW_3773 operon (Perez-Varela et al., 2020). As expected, both VIR-O Δ1132 and AV-T Δ1132 demonstrate a significant decrease in motility compared to their wild-type counterparts, both which are complemented by the single-copy chromosomal insertion of 1132 ( Figure 1D , only VIR-O results shown).

The 1132 deletion behaves in a manner opposite to 1132 overexpression with respect to motility, but the mechanism for this is unclear. As previously mentioned, abaI mRNA levels are unaffected by 1132 deletion, and although the RNA sequencing data shows a 1.5-fold downregulation of abaR, there are no transcriptional differences in the ABUW_3766-ABUW_3773 operon ( Supplementary Table 1 ). We also considered the possibility that AHL itself acts as a surfactant to some extent and that the loss of AHL secretion could reduce motility. To test this, we constructed a double mutant of 1132 and abaR—to control for quorum sensing-directed motility—and measured motility on 0.3% Eiken agar plates containing 1 μM synthetic 3-OH-C₁₂-HSL. However, motility was similar between the solvent control (38.75 mm ± 0.83) and the plate with AHL added (41.0 ± 1.73) p = 0.11.

VIR-O Δ1132 exhibited lower levels of colony sectoring than observed for the wild-type VIR-O on 0.5X LB agar ( Figure 2A ). This lack of sectoring indicated a decreased rate of switching to the AV-T variant, and we subsequently quantified a 16.4-fold decrease in the switching frequency ( Figure 2B ). To confirm these effects were due to 1132 deletion, we provided single-copy complementation of 1132 using the Tn7 transposon system. This strain exhibited the wild-type phenotype of colony sectoring and restored VIR-O to AV-T switching to wild-type levels ( Figure 2B ).

We further noted that AV-T Δ1132 colonies are more opaque than wild-type AV-T colonies, although AV-T Δ1132 is still translucent compared to the VIR-O Δ1132 colonies ( Figures 3A, B ). Single-copy chromosomal complementation of the AV-T Δ1132 mutant restored wild-type levels of translucence ( Figure 3A ). We first hypothesized that the increased opacity indicated hyper-switching from AV-T to VIR-O within the colony; however, the rate of switching was the same as the wild-type AV-T ( Supplementary Figure 3 ). We then considered that the AV-T Δ1132 mutant’s increased opacity may be due to increased levels of capsule.

Previous work revealed that wild-type VIR-O cells exhibit a 2-fold increase in capsule compared with the wild-type AV-T (Chin et al., 2018). We utilized a Percoll density gradient, a method that was recently described as a method to separate A. baumannii strains by capsule level (Kon et al., 2020), to compare capsular polysaccharide levels in AV-T Δ1132 to wild-type VIR-O and AV-T cells. As seen in Figure 3C , the gradient is able to distinguish between capsule levels of the wild-type VIR-O and AV-T variants, with the AV-T cells migrating significantly further than the VIR-O cells. Intriguingly, AV-T Δ1132 cells exhibited a high degree of heterogeneity and occupied a space in the gradient layer that was intermediate to that of the wild-type VIR-O and AV-T. We interpret this result as indication that deletion of 1132 causes a dysregulation and increase of capsule in the AV-T state, which imparts the more opaque appearance of AV-T Δ1132.

To confirm the effect of 1132 deletion on capsule, we next carried out TEM imaging of cell samples of wild-type VIR-O, wild-type AV-T, and AV-T Δ1132 stained with Ruthenium red ( Figure 3D ). We used ImageJ to measure capsule width in 100 cells per strain with 3 measurements taken per cell, which revealed average capsule widths of 0.142 µM ± 0.010, 0.072 µM ± 0.008, and 0.092 µM ± 0.008 in VIR-O, AV-T, and AV-T Δ1132, respectively ( Figure 3E ). The resulting ratio of these values is 1.96:1.00:1.28, which demonstrates the 2-fold difference in capsule previously seen in the VIR-O vs. AV-T (Chin et al., 2018) and confirms a 28% increase in capsule in the AV-T background when 1132 is deleted. The difference between the wild-type AV-T and AV-T Δ1132 is highly significant at p < 0.0001 as determined by a Student’s two-tailed t test.

In light of these results, we considered that, in the AV-T state, 1132 may regulate the K locus genes that encode the proteins largely responsible for the biosynthesis and export of capsular polysaccharide (CPS). We assessed representative genes from the K locus—manB, galE, ABUW_3818, ABUW_3820, ABUW_3821, ABUW_3830, wza, wzb, and wzc—by qRT-PCR across three sets of samples ( Supplementary Figure 4 ). These results indicated that 1132 did not transcriptionally regulate the K locus genes.

Capsule is a known virulence factor in A. baumannii, and presumably contributes to virulent phenotype of the VIR-O variant relative to AV-T (Russo et al., 2010; Chin et al., 2018; Singh et al., 2018; Tipton et al., 2018; Talyansky et al., 2021). The intermediate capsular levels of AV-T Δ1132 therefore suggested that this deletion may increase virulence. Before initiating virulence studies, we first tested the growth rates of the AV-T and AV-T Δ1132 strains and found no significant differences under laboratory conditions ( Supplementary Figure 5A . Using the Galleria mellonella (waxworm) model of infection, a modest, but statistically significant increase in AV-T Δ1132 virulence was observed ( Figure 4A ). Waxworms injected with the wild-type AV-T control showed a 23.3% survival rate after five days, while only 10% of those injected with AV-T Δ1132 survived. Complementation of AV-T Δ1132 with the single-copy chromosomal insertion of 1132 (AV-T Δ1132/Tn7-1132) reversed the increase in virulence, bringing the rate of survival back up to 30%.

We next examined the virulence of AV-T Δ1132 in a mouse pneumonia model of infection. In three experiments, mice were intranasally inoculated with 1 x 10⁸ CFU/mL of VIR-O (n=10), AV-T (n=17), or AV-T Δ1132 (n=17). At 24 hours post-inoculation, lungs were harvested, and the CFU/g for each tissue was calculated where a significant increase in CFU/g in AV-T Δ1132 was observed compared to wild-type AV-T in the lungs ( Figure 4B ). Both strains exhibited a similar number of VIR-O variants recovered from the lungs (approximately 0.1%). The highly virulent VIR-O variant was unaffected by the Δ1132 mutation in both a Galleria mellonella waxworm model and in a mouse pneumonia model ( Supplementary Figures 5B, C ) and both strains exhibited similar growth rates in-vitro ( Supplementary Figure 5A ).

All experiments were conducted using the clinical isolate AB5075. For all experiments, cultures were started from frozen glycerol stocks containing at least 99.5% of the desired colony variant (VIR-O or AV-T). Cultures were grown in either broth or on solid media containing LB (10 g tryptone, 5 g yeast extract, and 5 g NaCl per liter) and 1.5% agar (Difco) (“1X” LB agar) or 0.8% agar (“0.5X” LB agar).

Cultures were plated or streaked on 0.5X LB agar and viewed under a dissecting microscope, illuminated from underneath at an oblique angle. Colonies must be viewed at high colony density (at least 100 colonies per plate) to effectively distinguish VIR-O and AV-T colonies.

Plates containing the Agrobacterium tumefaciens traG::lacZ biosensor and X-Gal were prepared as described in Paulk Tierney and Rather (2019). AB5075 cell cultures were grown in 2 mL LB to OD₆₀₀ ~ 0.3, and 1 µL was spotted onto the soft agar lawn. Plates were incubated at 28°C overnight.

Dilutions were plated from frozen pure stocks of the strains to be assessed, and plates were incubated at 37°C for a set number of growth hours. High density plates were used to verify the purity of the stock. Isolated colonies were extracted from the plate by cutting out a section of agar underlying the colony and then resuspended and dilutions plated. The percent variant was then determined for each set of resuspended colonies from each strain. All experiments were performed twice with three colonies each for a total of six colonies.

Cell cultures for competent cells were grown in 2 mL LB to OD₆₀₀ ~ 0.5 (mid-log phase) at 37°C, shaking from frozen stocks. Cultures were pelleted and washed twice in sterile dH₂O, then resuspended to accommodate a volume of approximately 50 µL of cell per electroporation. Plasmid minipreps or ligation products mixed with cells were added to 2 mm cuvettes and electroporated at 2.5 kV. Cells were recovered in 1 mL LB at 37°C, shaking for one hour, then plated onto media containing selective antibiotics.

In-frame deletion of ABUW_1132 was generated using sucrose counterselection and a suicide vector containing the sacB marker (pEX18Tc) using methods previously described (Hoang et al., 1998). Briefly, PCR amplification (Phusion polymerase, Thermo-Fisher Scientific) was used to amplify the 2-kb regions upstream and downstream of the gene to be deleted, gel purified (UltraClean 15 DNA Purification Kit, MO BIO Laboratories), and ligated (Fast-Link DNA Ligation Kit, Epicentre Biotechnologies). The 4-kb ligation product was PCR amplified, gel purified, and ligated into the pEX18Tc vector MCS. The product vector was verified by sequencing and then transformed into competent E. coli Transformax EC100D cells (Epicentre Biotechnologies) by electroporation. The resulting suicide vector was confirmed by PCR and transformed by electroporation into AB5075, grown to OD₆₀₀ of 0.5 in 2-mL LB and washed twice in 10% glycerol. Transformants were plated on 1X LB agar plus Tetracycline (5-µg/mL) to yield single-crossover mutants. Counterselection was carried out at room temperature on 1X LB plates with 10% sucrose and no NaCl, and colonies were screened by PCR for the deletion. Primer sequences are recorded in Supplementary Table 2 .

To construct the Δ1132, abaR::T26 double mutant, genomic DNA was purified from an abaR::T26 mutant obtained from the University of Washington AB5075 transposon mutant library. Genomic DNA was then electroporated into VIR-O Δ1132. Transformants were selected on 1X LB plus Tetracycline (5-µg/mL) and confirmed by PCR.

We utilized a Tn7-based single-copy insertion element system (Ducas-Mowchun et al., 2019a; Ducas-Mowchun et al., 2019b) to reintroduce 1132 into the VIR-O Δ1132 and AV-T Δ1132 deletion mutants. We opted to use a segment of 1132 that includes a large portion of the up and downstream regions (GenBank Accession NZ_CP008706.1:1152309-1156236). This is the portion of the genome contained within the original 1132-containing fragment isolated from a high-copy AB5075 chromosomal library during the screen that led to our first discovery of 1132 (Perez-Varela et al., 2020), and previous experiments showed that this fragment allows for optimal expression of 1132. A modified version of this construct containing a T26 transposon (tetracycline) in 1132 had also been made to confirm loss of the phenotypes presumably caused by 1132 overexpression. This was made by digestion of the plasmid with PmlI, which cuts once near the beginning of the 1132 ORF, and re-ligation with a PCR-amplified T26 transposon. The original 1132-containing fragment and the modified version with 1132::T26 were excised from these plasmids by digestion with XbaI, which flanks the site into which the library was cloned, and gel purified to be used in construction of the suicide vector.

The pUC18T-mini-Tn7T-Apr-LAC construct was digested with SpeI and ligated with the 1132- or 1132::T26-containing fragment (Fast-Link DNA Ligation Kit, Epicentre Biotechnologies). The ligation product was transformed into competent E. coli Transformax EC100D cells (Epicentre Biotechnologies) by electroporation. The resulting constructs—pUC18T-mini-Tn7T-Apr-LAC/1132 and pUC18T-mini-Tn7T-Apr-LAC/1132::T26—were confirmed by PCR and sequencing then transformed into pure cultures of VIR-O Δ1132 and AV-T Δ1132 by electroporation along with the helper plasmid pTNS2. Transformant colonies were screened by PCR for insertion into the attTn7 site and those containing the correct insertion were again verified by sequencing. Wild-type Tn7 and Δ1132-Tn7 control strains were also made in both VIR-O and AV-T backgrounds using the same methods with an empty pUC18T-mini-Tn7T-Apr-LAC construct. Primer sequences are recorded in Supplementary Table 2 .

Strains for qRT-PCR analysis in Supplementary Figure 1 were prepared for RNA isolation by growing them from pure frozen stocks to OD₆₀₀ ~ 0.15 then plating 100 µL on a 1X LB plate. Plates were incubated at 37°C for 6 hours, then pooled with 3 mL ice cold LB. Resuspensions were normalized to within OD₆₀₀ of 0.01, then 1 mL of the resuspension was pelleted, flash frozen in an ethanol-dry ice bath and stored at -80°C. The resuspension cultures were streaked on 0.5X LB to ensure sample purity.

Strains for qRT-PCR analysis in Supplementary Figure 4 were grown up to match the growth conditions of the cells prepared for electron microscopy (see below). Samples were grown from pure frozen stocks in overnight cultures of 2 mL LB and were then shaken at 37°C to the same OD₆₀₀ of 0.3. 20 µL of each culture was then plated as a line onto a 0.5X plate, allowed to incubate at 37°C for 4 hours, and then harvested off the plate. Scraped cells were resuspended to an OD₆₀₀ of 0.6 in 2 mL LB. 1 mL of cells was pelleted, flash frozen in an ethanol-dry ice bath, then stored at -80°C. The resuspension cultures were streaked on 0.5X LB to ensure sample purity.

RNA was prepared using the Epicentre MasterPure RNA Purification kit according to the manufacturer’s protocols. The resulting nucleic acid product was purified to remove DNA contamination using the Invitrogen TURBO DNA-free kit according to the manufacturer’s instructions. Following quantification of RNA concentration using a NanoDrop ND-1000 spectrophotometer, cDNA was prepared from 1 µg of RNA using the High-Capacity cDNA Reverse Transcription Kit by Applied Biosystems and subsequently diluted 1:10 in nuclease-free water. qRT-PCR experiments were carried out on a Bio-Rad CFX Connect Real-Time PCR Detection System using iQ SYBR Green Supermix reverse transcriptase from Bio-Rad. RNA purity was confirmed by qRT-PCR through the inclusion of template controls made without reverse transcriptase. qRT-PCR data was analyzed by the delta-delta Ct method (2^–∆∆Ct) with comparison to 16S as an internal control. This method was carried out for three biological replicates. Each biological replicate had three technical replicates for each primer set. Primer sequences are recorded in Supplementary Table 2 .

Three independent sets of RNA were prepared from each of AB5075 strains VIR-O and VIR-O Δ1132. Cultures were grown in 2 mL LB from pure frozen stocks to OD₆₀₀ ~ 0.5 (shaking, 37°C) and were streaked on 0.5X LB to ensure purity of samples. 1 mL of cells was pelleted and flash frozen in a dry ice bath before being stored at -80°C. RNA was prepared using the Epicentre MasterPure RNA Purification kit according to the manufacturer’s protocols.

RNA quality control, sequencing, and analysis was carried out by the Yerkes Non-Human Primate Genomics Core at Emory University. RNA quantity and quality assessments were carried out using a Thermo Nanodrop2000 and Agilent 2100 Bioanalyzer, respectively. RNA sequencing was carried out using an Illumina HiSeq 3000, and reads were normalized and mapped using Cufflinks software. The RNA-Seq data discussed in this publication have been deposited in NCBI’s Gene Expression Omnibus and are accessible through GEO Series accession number GSE185730 (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE185730).

Method slightly modified from Kon et al. (2020). Cell cultures of VIR-O, AV-T, and AV-T Δ1132 were grown in 2 mL LB from pure frozen stocks to OD₆₀₀ ~ 0.6 (shaking, 37°C). For each strain, 600 µL of cells was pelleted and washed in PBS 1X, then resuspended in 100 µL PBS 1X. Percoll solutions were prepared by first mixing 9 parts Percoll (Sigma Aldrich Cat. P4937) and 1 part 1.5 M NaCl, which was further diluted using 0.15 M NaCl to make 40% and 50% Percoll solutions. For each cell sample to be tested, 1 mL of 40% Percoll was added to a 12x75 mm glass tube (Fisherbrand 14958C), and 1 mL 50% Percoll was gently layered underneath using a 1cc syringe and a 1.5 in. 23G needle (BD 305194). 100 µL of cells were gently layered at the top of the gradient. Samples were spun at 3,000 xg for 30 minutes at room temperature and then visually assessed. This protocol was repeated 6 times.

Ruthenium red-lysine fixation and cell staining and subsequent transmission electron microscopy were performed by the Robert P. Apkarian Integrated Electron Microscopy Core at Emory University using previously described techniques (Fassel et al., 1997; Fassel et al.,1998; Beaussart et al., 2014). Samples were provided to the EM Core as follows: cells grown from pure frozen stocks in overnight cultures were shaken at 37°C until OD ~ 0.30, and 20 µL of each culture was plated as a line onto the same 0.5X LB plate. Plates were then incubated at 37°C for 4 hours and stored overnight at 4°C before being transported to the EM Core.

Capsule widths in the electron micrographs were measured by ImageJ 1.53e (Rasband, W.S., ImageJ, U. S. National Institutes of Health, Bethesda, Maryland, USA, https://imagej.nih.gov/ij/, 1997-2018). Three measurements were taken for 100 cells of each strain at the same magnification, which were averaged and then converted to micrometers.

Waxworms were purchased from Speedy Worm (www.speedyworm.com). Cultures of AV-T, AV-T Δ1132, and AV-T Δ1132/Tn7-1132 were grown in 2 mL LB from pure frozen stocks to OD₆₀₀ ~ 0.5 (shaking, 37°C). 5 µL of a LB control, AV-T, AV-T Δ1132, or AV-T Δ1132/Tn7-1132 was injected into the hemolymph of a larvae of mass 150-200 mg (n=30 per condition; approximately 1.2 x 10⁶ CFU were administered per strain). Larvae were then housed in petri dishes in a humidified incubator at 37°C for 5 days. Each day, dead larvae were removed and surviving larvae counted.

Approximately 1 × 10⁸ CFU were administered per mouse for infections to quantify the bacterial load. For mouse infections, overnight standing bacterial cultures at room temperature were sub-cultured in LB broth and grown at 37°C with shaking to an OD600 ~0.15, washed and re-suspended in PBS. Each mouse was inoculated intranasally with 50 µL of bacteria. Mice were anesthetized with isoflurane immediately prior to intranasal inoculation. At 24 hours, the mice were sacrificed and the lungs were harvested, homogenized, and ten-fold serial dilutions plated for CFU on 0.5X LB plates. The mouse strain used was C57BL/6J (females at 8-10 weeks of age) from Jackson Laboratories (JAX stock #000664). Experiments were carried out under the Institutional Animal Care and Use Committee guidelines.

Statistics were performed using GraphPad Prism 9.2.0 software for Windows (GraphPad Software, San Diego, California USA, www.graphpad.com). The following statistical tests were utilized: (1) Mann-Whitney test for the mouse experiments (two-tailed), two-sample switching assays (two-tailed), motility assays (two-tailed) and qRT-PCR experiments (two-tailed or multiple, as indicated); (2) Log-Rank (Mantel-Cox) test for the Galleria mellonella experiments; (3) Welch’s ANOVA analysis for three-sample switching assays; and (4) Student’s two-tailed t test for capsule width measurements from the TEM micrographs. The Mann-Whitney test was utilized instead of a Student’s t test to allow for the possibility or reality of a non-Gaussian distribution. Welch’s ANOVA was chosen over standard ANOVA to allow for unequal variance between samples.