Bacterial hydrophilins promote pathogen desiccation tolerance

Summary

Acinetobacter baumannii is a leading cause of hospital-acquired infections, where outbreaks are driven by its ability to persist on surfaces in a desiccated state. Here, we show that A. baumannii causes more virulent pneumonia following desiccation and profile the genetic requirements for desiccation. We find that desiccation tolerance is enhanced upon disruption of Lon protease, which targets unfolded and aggregated proteins for degradation. Notably, two bacterial hydrophilins, DtpA and DtpB, are transcriptionally upregulated in Δlon via the two component regulator, BfmR. These proteins, both hydrophilic and intrinsically disordered, promote desiccation tolerance in A. baumannii. Additionally, recombinant DtpA protects purified enzymes from inactivation and improves the desiccation tolerance of a probiotic bacterium when heterologously expressed. These results demonstrate a connection between environmental persistence and pathogenicity in A. baumannii, provide insight into mechanisms of extreme desiccation tolerance, and reveal potential applications for bacterial hydrophilins in preservation of protein- and live bacteria-based pharmaceuticals.

Graphical Abstract

eTOC blurb

Hospital surfaces are a major reservoir for transmission of Acinetobacter baumannii. Green et al determine that the virulence of A. baumannii is enhanced following drying, and discover that desiccation tolerance is influenced by Lon protease through the modulation of the predicted intrinsically disordered hydrophilin proteins DtpA and DtpB.

Introduction

Hospital-acquired infections present an urgent threat to human health. These infections prolong hospital stays, increase morbidity and mortality, and escalate healthcare costs. Amplifying all of these outcomes is the rising incidence of antibiotic resistance among healthcare associated pathogens (Peleg and Hooper, 2010). Despite infection control efforts, hospital surfaces remain a major source of pathogen spread (Dancer, 2014). The multi-drug resistant gram-negative pathogen Acinetobacter baumannii is a leading cause of nosocomial pneumonias, soft-tissue and bloodstream infections, and is found ubiquitously on hospital surfaces, equipment, personnel, and indwelling devices (Raro et al., 2017; Wang et al., 2017). Contamination of hospital units by A. baumannii has been associated with nosocomial outbreaks, suggesting that hospital surfaces serve as a reservoir for its transmission (Aygün et al., 2002). Moreover, the COVID-19 pandemic has caused disruptions to infection control practices within healthcare settings, leading to surges in A. baumannii cases and underscoring the potential for continued outbreaks with this pathogen (Gottesman et al., 2021).

Bacteria tolerate an onslaught of stresses during persistence on environmental surfaces, including water loss due to extended periods of dryness, a process termed desiccation (Wendt et al., 1997). Survival times ranging from weeks to several months of desiccation have been reported for clinical isolates of A. baumannii, suggesting that this organism is broadly adapted to withstand this stress (Wendt et al., 1997). While recent studies have identified some factors promoting this desiccation phenotype (Aranda et al., 2011; Boll et al., 2015; Farrow et al., 2018, 2020; Tipton et al., 2018; Wang et al., 2020; Zeidler and Müller, 2019a), the mechanisms underlying the extreme desiccation tolerance of A. baumannii and the impact of desiccation on pathogen transmission and infection severity remain largely unexplored.

Here, we show that A. baumannii causes more virulent infections following rehydration from desiccation and report the discovery of two bacterial hydrophilins, the predicted intrinsically disordered proteins (IDPs), DtpA and DtpB, which promote desiccation tolerance in this organism. We show that recombinant DtpA can protect purified enzymes from desiccation and heat inactivation and determine that heterologous expression of this protein enhances the desiccation tolerance of a probiotic bacterium, suggesting a function for bacterial hydrophilins in the prevention of protein denaturation or aggregation during desiccating conditions. Additionally, we find that DtpA expression is controlled by a conserved Lon protease, revealing a key regulatory mechanism utilized by A. baumannii to tolerate environmental stress.

Results

A. baumannii exhibits extreme desiccation tolerance.

To examine the effects of desiccation on physiology and pathogenesis in A. baumannii, bacteria were applied to a polystyrene surface, desiccated, and incubated at ambient temperature and 20–30% relative humidity. Intermittently, bacteria were rehydrated and viability relative to the starting population was determined. Remarkably, A. baumannii grown to early-stationary phase retained nearly 100% viability after several days of desiccation, and a fraction of the starting population remained viable after 7 months of desiccation. (Figure 1A). Consistent with previous reports (Wang et al., 2020), this tolerance was growth-stage dependent, as cultures desiccated from exponential phase rapidly lost viability following drying (Figure S1A). The desiccation tolerance of A. baumannii after 28 days was markedly higher than that of several other gram-negative bacterial species examined, including the pathogens Klebsiella pneumoniae and Pseudomonas aeruginosa (Figure 1B), suggesting that A. baumannii is well-adapted to survive prolonged periods of drying. During persistence on surfaces, bacteria are subjected to both dehydration and nutrient restriction. However, A. baumannii was no more resistant to starvation than the other gram-negative species assessed (Figure S1B-C), indicating a specific advantage for this bacterium in tolerating water loss. Moreover, recent clinical isolates of A. baumannii were nearly 10-fold more resistant to desiccation relative to the laboratory-adapted strain ATCC 17978, suggesting that environmental surfaces may exert a selective pressure on strains circulating in hospitals (Figure 1B).

Figure 1. Infections with desiccated A. baumannii are associated with higher bacterial loads.

(A) Survival of early-stationary phase A. baumannii after desiccation for the indicated time points. N=3–6. (B) Survival of early-stationary phase cultures after 28 days of desiccation. Dashed line represents mean desiccation tolerance of A. baumannii ATCC 17978. N=3–8. (C) CFUs recovered at 36 hpi following infections with desiccated and non-desiccated (control) inocula. N=14–20. (D) Competitive index (C.I.) of desiccated A. baumannii at 36 hpi following co-infection with non-desiccated (control) or desiccated bacteria. N=9–12. (E) H₂O₂ sensitivity of desiccated, non-desiccated (control), or bacteria subcultured from desiccation after 30 minutes of incubation with H₂O₂ at the indicated concentrations. N=3. All data are mean ± SEM. Panel C, D by Mann-Whitney. Panel E by Two-way ANOVA with Dunnett’s multiple comparison post-test on log₁₀-transformed data. *P<.05, **P<.01, ***P<.001, ****P<.0001, ns = not significant.

A. baumannii causes more virulent infection following rehydration from desiccation.

To interrogate the impact of desiccation on A. baumannii pathogenesis, bacteria were desiccated for 48 hours, then rehydrated and inoculated intranasally into mice to model bacterial pneumonia. We observed that mice experienced a trend toward higher levels of mortality following infections with desiccated A. baumannii (Figure S2A), and this mortality was associated with enhanced bacterial burdens in tissues at 36 hours post-infection (Figure 1C). Growth of A. baumannii was not significantly affected following desiccation (Figure S2B). Moreover, bacteria rehydrated from desiccation had an increased competitive index (C.I.) when co-infected with non-desiccated bacteria in a murine pneumonia model (Figure 1D), suggesting that desiccated bacteria cannot provide cross-protection to a non-desiccated inoculum. Thus, we hypothesized that the increased virulence was due to intrinsic differences in stress tolerance following desiccation. A. baumannii encounters host-derived reactive oxygen species (ROS) and withstands killing by phagocytic immune cells during lung infection (Juttukonda et al., 2019; Lee et al., 2020); therefore we assessed the impact of desiccation on resistance to these important immune mechanisms of protection. Desiccated A. baumannii was more resistant to macrophage killing (Figure S2C) and exhibited enhanced resistance to H₂O₂ killing (Figure 1E). Sensitivity to H₂O₂ was regained after 8 hours of growth in rich media (Figure 1E), indicating that the increased resistance observed in desiccated populations is a reversible phenotype likely driven by changes in gene or protein expression. Additionally, desiccated bacteria displayed increased resistance to hypochlorous acid (Figure S2D), a primary component of the neutrophil oxidative burst (Anderson et al., 1999), and were more resistant to heat-induced killing (Figure S2E). Similar ROS resistance and increased tissue colonization were observed following desiccation of a more recent clinical isolate, the multi-drug resistant ABUW 5075 (Figure S2F and G), demonstrating that the mechanisms underlying this phenotype might be conserved across multiple A. baumannii isolates.

An intrinsically disordered protein protects A. baumannii from desiccation.

To uncover the molecular mechanisms driving desiccation tolerance in A. baumannii, a genome-wide transposon sequencing (Tn-seq) screen was performed (Figure 2A). This screen identified gene products influencing survival after short (2 days) and sustained (14 days) periods of desiccation (Figure S3A, Table S1). These included factors with known or predicted roles in DNA damage repair (xseA, ruvA), oxidative stress resistance (ahpC, msrA), transcriptional regulation (bfmR, ACX60_13660), protein homeostasis (dnaJ, clpX), and stress resistance (copC, cspA). Additionally, among these factors was the conserved Lon protease, a member of the AAA+ family of proteases, which targets unfolded and aggregated proteins for degradation (Gur, 2013). Cells experience toxic protein unfolding and aggregation as a result of drying (Tapia and Koshland, 2014; Wang et al., 2020). Paradoxically, mutants harboring transposon insertions in lon were enriched following 2 (Z score=4.91) and 14 (Z score=1.47) days of desiccation (Table S1), and a lon deletion mutant (Δlon) was highly resistant to desiccation (Figure 2B, Figure S3B). Additionally, Δlon was more resistant to oxidative stress (Figure S3C), suggesting that factors stabilized or upregulated in the absence of lon might promote generalized stress responses in A. baumannii.

Figure 2. An intrinsically disordered protein promotes desiccation tolerance.

(A) Schematic of Tn-seq experiment. (B) Survival following desiccation for the indicated time points. N=9. (C) Insoluble protein fractions were isolated from bacterial cultures. Arrow indicates enrichment of 46-kD protein species. (D) Prediction of naturally disordered regions in DtpA. (E) Overlaid CD spectra of ¹⁵N-labeled DtpA from 30°C-50°C. (F) Survival following desiccation for the indicated time points. N=15–30. All data are mean ± SEM. Panel B by by Two-way ANOVA with Sidak’s multiple comparison post-test on log₁₀-transformed data. Panel F by by Two-way ANOVA relative to WT with Tukey’s multiple comparison post-test on log₁₀-transformed data. *P<.05, **P<.01, ***P<.001,****P<.0001, ns = not significant.

To determine whether aggregated or unfolded proteins are present in Δlon, insoluble proteins were isolated from WT and Δlon cultures. A 46-kD protein species accumulated in the insoluble fraction of Δlon (Figure 2C). Proteomics mapped this protein to the locus ACX60_11190, which we named Desiccation Tolerance Protein A (DtpA), due to its association with a desiccation tolerance phenotype. Aside from a short KGG domain sequence at the N-terminus of the protein, the amino acid sequence of DtpA is almost completely dominated by repeats of charged, hydrophilic residues (Figure S3D-E), and is predicted to encode an intrinsically disordered protein (IDP) (Figure 2D). Circular dichroism analysis of purified DtpA confirmed a lack of secondary structure, a feature of IDPs (Figure 2E). Additionally, domain prediction analysis showed a series of 13 predicted AT-HOOK binding motifs across the intrinsically disordered portion of DtpA (Figure S3E). These motifs, which have been primarily characterized in eukaryotic organisms, have been associated with DNA-binding proteins including chromatin (Aravind and Landsman, 1998).

Highly charged and hydrophilic IDPs are associated with desiccation tolerance in tardigrades, nematodes, yeast, and plant seeds, where these “hydrophilin” proteins are proposed to serve as molecular shields that protect proteins and other cellular components from drying-induced damage (Boothby et al., 2017; Chakrabortee et al., 2010; Dure and Chlan, 1981; Kim et al., 2018). While insertions within dtpA were not present in the transposon library screened at the 14-day timepoint, results from our Tn-seq screen indicated that mutants with transposon insertions in dtpA were selected against after 2 days of desiccation (Table S1). Therefore, this locus was deleted in the WT and Δlon backgrounds and the absence of DtpA accumulation in the insoluble fraction of a ΔlonΔdtpA strain was confirmed (Figure S3F). Consistent with the phenotype observed in the Tn-seq screen, deletion of dtpA decreased the desiccation tolerance of a WT strain (Figure 2F, Figure S3G), demonstrating that A. baumannii can utilize IDPs to survive drying. Intriguingly, while deletion of dtpA reduced the desiccation tolerance of a Δlon strain, the ΔlonΔdtpA strain remained more desiccation tolerant than a ΔdtpA mutant after 28 days (Figure 2F), suggesting that Lon might affect the expression or stability of additional factors that promote desiccation tolerance in A. baumannii.

dtpA expression is regulated by a Lon and BfmR-dependent transcriptional circuit.

In addition to its role in the destruction of unfolded proteins, Lon regulates stress response mechanisms through the degradation of DNA binding proteins, thereby promoting regulatory responses that can facilitate adaptation to proteotoxic stress (Jonas et al., 2013; Lu et al., 2013). To determine whether Lon controls dtpA transcription, dtpA transcript levels were quantified in Δlon. We observed that dtpA transcript levels were elevated more than 200-fold in Δlon relative to WT (Figure 3A). Additionally, through the use of a transcriptional reporter fusion of the dtpA promoter to the luciferase operon, we observed that dtpA transcription is regulated by growth phase, with expression peaking during the transition to stationary phase, as well as during late-stationary phase (Figure 3B), corresponding to the increased desiccation tolerance observed after stationary phase growth (Figure S1A). These data suggest that Lon-mediated regulation of dtpA is at least partially indirect, likely occurring through the degradation of one or more transcriptional regulators. Moreover, we observed transcriptional induction of dtpA following 4 hours of desiccation (Figure 3C), demonstrating that expression of this factor is increased in response to drying.

Figure 3. dtpA expression is regulated by a Lon- and BfmR- dependent transcriptional circuit.

(A) dtpA transcript levels were quantified after 8 hours of growth. Median is shown. N=6. (B) WT or Δlon A. baumannii strains harboring transcriptional reporter fusions of the dtpA promoter fused to the luciferase operon luxABCDE were cultured for the indicated time points and luminescence and optical density were determined. Dashed lines indicate OD₆₀₀ while solid lines report luminescence relative to OD₆₀₀. Mean± SEM is shown. N=9 and data are representative of two independent experiments. (C) WT A. baumannii was cultured overnight and subjected to desiccation. dtpA transcripts were quantified at time zero and after 4 hours of desiccation. Median is shown. N=6. (D) WT, Δlon, and Δlon suppressor (bfmR*) strains were streaked on LB agar. (E) Survival of WT, Δlon, Δlon suppressor (bfmR*), and ΔlonΔbfmR strains was monitored after 7 days of desiccation. Mean± SEM is shown. N=8–12. (F) Insoluble protein fractions were isolated after 8 hours of growth. (G) Strains were cultured for 8 hours and dtpA transcript levels were determined. N=8–9. Median is shown. (H) Strains harboring transcriptional reporter fusions of the dtpA promoter fused to luxABCDE were cultured for the indicated time points and luminescence relative to OD₆₀₀ was determined. Panel on right indicates mean ± SEM of luminescence relative to OD₆₀₀ at the indicated time points. N=3–7 and data are representative of two independent experiments. Panel A, C by Mann-Whitney. Panel E by One-Way Anova with Tukey’s mulitple comparison post-test on log₁₀-transformed data. Panel G by Kruskal-Wallis with Dunn’s multiple comparison post-test. Panel H by Two-Way Anova with Dunnett’s multiple comparison post-test. Panel H by *P<.05, **P<.01, ***P<.001,****P<.0001, ns = not significant.

During routine culturing of Δlon, a colony variant displaying an altered, translucent morphology was recovered (Figure 3D). Whole genome sequencing of the variant revealed an A43E substitution in the receiver domain of BfmR (Figure S4A), the response regulator of the BfmRS two-component system (Tomaras et al., 2008). This variant (Δlon bfmR*) exhibited high sensitivity to desiccation (Figure 3E) and ROS (Figure S4B) and produced less DtpA (Figure 3F). An in-frame deletion of bfmR reproduced these findings (Figure 3E, Figure S4B), suggesting that deletion of bfmR is sufficient to reverse Lon-dependent increases in stress tolerance. Consistent with previously published reports (Farrow et al., 2018), we found that deletion of bfmR attenuated desiccation survival in A. baumannii (Figure 3E) while loss of bfmS, which negatively regulates BfmR, enhanced overall survival (Figure S4C). We therefore hypothesized that BfmR controls activation of dtpA transcription. Supporting this hypothesis, deletion of bfmR reversed dtpA upregulation in the Δlon mutant after 8 hours of growth (Figure 3G), indicating that BfmR-dependent activation of dtpA is enhanced in the absence of Lon. However, dtpA expression was still heightened in the Δlon bfmR* and ΔlonΔbfmR strains relative to WT in late-stationary phase (Figure 3H, Figure S4C), suggesting that an additional regulatory protein might also be responsible for the heightened expression of dtpA in the absence of Lon protease.

A second hydrophilin promotes A. baumannii desiccation tolerance.

Despite sharing biochemical features with eukaryotic hydrophilins, the full-length sequence of DtpA is only conserved within the 6 members of the pathogenic Acinetobacter calcoaceticus-baumannii complex (Figure 4A), where it is encoded within a cluster of genes with predicted stress response functions, including the catalase katE, a predicted oxidoreductase, and a predicted redox family protein (Figure 4A). Curiously, we noted that a gene present within this cluster, ACX60_11220, encodes a protein consisting of a series of repeating hydrophilic and charged amino acid residues (Figure 4B). This protein, which we named DtpB, was predicted to be intrinsically disordered across its entire coding sequence, and was also conserved only within the Acinetobacter calcoaceticus-baumannii complex (Figure 4A,C). Additionally, we observed that dtpB expression was upregulated in a Δlon mutant, and that a bfmR mutation reversed this upregulation (Figure 4D), suggesting a shared mechanism of regulatory control with dtpA. Likely due to its small size (222 bp), mutants harboring transposon insertions in dtpB were not present in the transposon library assessed in our Tn-seq screen. However, because other organisms, including tardigrades and plants, employ multiple IDPs to tolerate desiccation, we reasoned that DtpB might serve as a second hydrophilin protein in A. baumannii. Supporting this hypothesis, ΔdtpB and ΔdtpAΔdtpB mutants were sensitive to desiccation relative to WT A. baumannii (Figure 4E). Together, these data suggest that A. baumannii utilizes multiple hydrophilin proteins to shield against desiccation-induced damage. Moreover, the observation that the ΔlonΔdtpAΔdtpB mutant retains appreciable desiccation tolerance suggests the existance of remaining as-yet-unidentified factors involved in mediating this phenotype.

Figure 4. A second hydrophilin, DtpB, contributes to desiccation tolerance.

(A) Genetic alignments of Acinetobacter species that are predicted to contain orthologs of dtpA and adjacent loci. The numbers on each gene correspond to amino acid similarity, based on alignment to the representative A. baumannii genes. (B) Amino acid sequence of DtpB. Pink text denotes hydrophilic residues, purple text denotes residues with positive charge, and blue text denotes residues with negative charge. (C) Prediction of naturally disordered regions in DtpB. (D) dtpB transcript levels were quantified after 8 hours of growth. N=8–9. Median is shown. (E) Survival following desiccation for the indicated time points. N=12–18. Data are mean ± SEM. Panel D by Kruskal-Wallis with Dunn’s multiple comparison post-test. Panel E by Two-Way ANOVA relative to WT with Tukey’s multiple comparison post-test on log₁₀-transformed data (comparisons with WT and Δlon are shown). *P<.05, **P<.01, ***P<.001,****P<.0001, ns = not significant.

In addition to increased desiccation tolerance, a Δlon mutant displays increased tolerance to H₂O₂ (Figure S3C). While deletion of dtpA resulted in a minor reduction in the oxidative stress tolerance of a Δlon mutant (Figure S5A), we hypothesized that upregulation of additional factors in the absence of Lon might also contribute to this phenotype. Supporting this hypothesis, a Δlon strain exhibited increased catalase activity (Figure S5B), corresponding with a four-fold increase in transcription of the catalase katE, encoded by ACX60_11205 (Figure S5C). Consistent with previous findings that showed BfmR-dependent regulation of katE expression (Farrow et al., 2018; Palethorpe et al., 2021), we observed that deletion of bfmR reversed katE upregulation in the Δlon background (Figure S5C), indicating that BfmR activity is required for increased expression of this factor in the absence of Lon. Deletion of katE reversed the enhanced H₂O₂ resistance of Δlon (Figure S5D), implying that upregulation of katE might account for the heightened oxidative stress tolerance of the Δlon strain. Additionally, we found that katE transcription was induced after 4 hours of desiccation (Figure S5E), and observed a significant reduction in the desiccation tolerance of a ΔkatE mutant after 14 days (Figure S5F), suggesting that tolerance to oxidative stress is a critical factor for overcoming desiccation in A. baumannii. However, a ΔlonΔkatE mutant remained resistant to desiccation (Figure S5F), supporting the conclusion that additional factors produced in the absence of lon, including dtpA and dtpB, contribute to dessication tolerance in A. baumannii.

Expression of dtpA and dtpB is enhanced in a recent clinical isolate.

We observed that recent clinical isolates of A. baumannii exhibit increased desiccation tolerance relative to the lab-adapted strain ATCC 17978 (Figure 1B), which was originally isolated from a meningitis-infected patient in 1951 (Smith et al., 2007). This observation, coupled with our characterization of DtpA and DtpB, motivated the hypothesis that the increased desiccation tolerance of clinically circulating strains could be attributed to higher expression of dtpA and dtpB in these bacteria. To address this hypothesis, dtpA and dtpB transcript levels were measured in the recent clinical isolate ABUW 5075, a highly virulent and multi-drug resistant strain isolated from an osteomyelitis infection in 2008 (Jacobs et al., 2014). Consistent with our hypothesis, we observed that relative to ATCC 17978, dtpA and dtpB transcript levels were elevated more than 2,000-fold in ABUW 5075 (Figure 5A-B). Moreover, ABUW 5075 mutants harboring transposon insertions in dtpA and dtpB were defective at tolerating desiccation relative to WT ABUW 5075 (Figure 5C), suggesting a vital function for these factors in protecting this strain from desiccation. Together, these results suggest that increased expression of hydrophilin proteins might contribute to the long-term persistence of A. baumannii on hospital surfaces.

Figure 5. A recent clinical isolate expresses increased levels of dtpA and dtpB.

(A-B) dtpA and dtpB transcript levels were quantified after 8 hours of growth. N=5–6. Median is shown. (C) Survival of ABUW 5075 WT, tn::lon, tn::dtpA, and tn::dtpB strains was monitored after 7, 14, and 28 days of desiccation. N=9. Data are mean ± SEM. A-B by Mann-Whitney. Panel C by Two-Way ANOVA with Dunnett’s multiple comparison post-test on log₁₀-transformed data. *P<.05, **P<.01, ***P<.001,****P<.0001, ns = not significant.

DtpA is sufficient for desiccation tolerance.

It has been postulated that the hydrophilins of plants and tardigrades protect cells from proteotoxic stress by forming molecular shields that surround proteins and inhibit their unfolding or aggregation, thereby facilitating protection from environmental extremes such as salinity, drought, desiccation, freezing, and heat (Janis et al., 2018). Therefore, we hypothesized that one mechanism by which hydrophilins protect A. baumannii from desiccation might be the shielding of enzymes from drying-induced damage. To interrogate this hypothesis, the activity of recombinant β-lactamase was assessed after drying in the presence and absence of purified DtpA. We observed that β-lactamase enzymatic activity was preserved when desiccated in the presence of DtpA, which provided more protection than a positive control protein, bovine serum albumin (BSA) (Figure 6A), a homolog of human serum albumin, a U.S. Food and Drug Administration approved stabilizing agent (Bosse et al., 2005). Additionally, DtpA protected β-lactamase from heat-inactivation (Figure 6B), suggesting a function for DtpA in safeguarding proteins against denaturation or aggregation.

Figure 6. DtpA is sufficient to protect proteins from drying and heterologous expression enhances the desiccation tolerance of probiotic E. coli.

(A) TEM β-lactamase was dehydrated and rehydrated in the presence of recombinant protein at the indicated molar ratios (protein:TEM). β-lactamase activity was assessed following rehydration and normalized to the activity of untreated protein. N=4–8. (B) TEM β-lactamase heat-inactivated in the presence of recombinant protein at the indicated molar ratios (protein:TEM). β-lactamase activity was assessed following heat inactivation and normalized to the activity of untreated protein. N=4–14. (C) Survival of EcN after 7 days of desiccation. N=15–18. (D-E) EcN strains were desiccated for 7 days, rehydrated, and surviving bacteria were intragastrically inoculated into streptomycin-treated mice. Presence of EcN in feces was determined at 8 hpi. n.d.=not detected. N=9–10. All data are mean ± SEM. Panel A-B by One-way ANOVA with Dunnett’s multiple comparison post-test relative to vehicle. Panel C by Two-way ANOVA with Sidak’s multiple comparison test on log₁₀-transformed data. Panel E by Fisher’s Exact test. *P<.05, **P< .01, ***P< .001,****P< .0001, ns = not significant, n.d. = not detected.

The desiccation tolerance provided by DtpA prompted the hypothesis that heterologous expression of this protein could protect other organisms from dehydration. The use of live-bacterial probiotic supplements poses a promising advancement in the treatment of gut-related inflammatory and infectious diseases (Sanders et al., 2019). However, the viability of bacterial probiotic supplements is often limited due to the requirement for storage and transport of products in a dry state (Meng et al., 2008). To determine whether expression of a bacterial hydrophilin could improve the shelf-life of probiotic bacteria, dtpA was expressed in the E. coli Nissle (EcN) probiotic strain (Sonnenborn and Schulze, 2009), and EcN was subsequently desiccated for 7 days, rehydrated, and delivered to mice via oral gavage. Expression of dtpA increased the desiccation tolerance of EcN (Figure 6C), and therefore dramatically enhanced recovery of EcN from mice (Figure 6D-E), highlighting the potential utility for DtpA in stabilizing probiotic supplements prior to human consumption.

In summary, we show that A. baumannii can utilize IDPs to survive drying, and that expression of these factors is modulated by Lon protease. This work provides insight into the mechanisms of extreme desiccation tolerance in A. baumannii, and poses significant translational potential, as disinfectants designed to target the expression, stability, or function of these proteins may provide a means to decontaminate A. baumannii from hospital surfaces. Additionally, we show that DtpA can protect purified enzymes and heterologous organisms from desiccation, highlighting the potential utility for bacterial hydrophilins in stabilizing diverse biological materials, including probiotics, vaccines, and other biologics.

Discussion

There are rare examples spanning every Kingdom of life of organisms that have evolved strategies to withstand virtually complete loss of water for extended periods of time, despite the cellular damages incurred as a consequence of desiccation. While some mechanisms of desiccation tolerance have been described in bacteria, these studies have been largely focused on environmental microbes adapted to tolerate dry climates, and little is known about the mechanisms by which bacterial pathogens survive dehydration or the impact that desiccation has on their transmission and virulence (Lebre et al., 2017). The gram-negative bacterium A. baumannii is an emerging pathogen transmitted predominantly on hospital surfaces, where it persists in a desiccated state (Wong et al., 2017). While the extreme desiccation tolerance of A. baumannii has been previously documented and is a known contributor to hospital transmission (Zeidler and Müller, 2019b), the molecular mechanisms underlying this phenomenon are not well understood. Here, we report the discovery of factors influencing the desiccation tolerance of A. baumannii, including the predicted intrinsically disordered proteins (IDPs) DtpA and DtpB.

Although the role of IDPs in mediating the desiccation tolerance of bacteria is not well understood, the function of desiccation-related IDPs has been reported in a number of eukaryotic organisms (Boothby and Pielak, 2017; Chakrabortee et al., 2012). Desiccation-related IDPs were first discovered in plants, which accumulate intrinsically disordered Late embryogenic abundance (LEA) proteins in response to drying conditions (Dure and Chlan, 1981). LEA proteins are predominantly comprised of hydrophilic residues, charged amino acids, small amino acids like glycine and serine, and a high proportion of disorder-promoting residues (Hincha and Thalhammer, 2012). It is thought that these proteins protect from proteotoxic stress by forming glass-like matrices within cells that physically prevent proteins from denaturing or forming aggregates. Alternatively, or additionally, it is possible that during desiccation, highly hydrophilic IDPs could form hydrogen bonds with cellular components, thereby replacing water and preventing protein denaturation, aggregation, and membrane fusion (Boothby and Pielak, 2017; Chakrabortee et al., 2012).

Desiccation related IDPs, often referred to as hydrophilin proteins, have since been discovered in other eukaryotic organisms, including yeast, nematodes, and tardigrades (Boothby and Pielak, 2017; Chakrabortee et al., 2007; Kim et al., 2018). Despite the their mechanistic similarities, tardigrade-specific IDPs and LEA proteins seem to have evolved independently, highlighting the utility of this class of proteins in protecting cellular matter across multiple Kingdoms of life (Boothby et al., 2017). In this study, we show that two bacterial hydrophilins, DtpA and DtpB, promote the desiccation tolerance of A. baumannii (Figure 2, 4). Additionally, we demonstrate that recombinant DtpA can protect enzymes from desiccation and heat-induced inactivation (Figure 6), indicating that this protein might function to shield against enzyme unfolding or inactivation during times of proteotoxic stress.

Like the eukaryotic hydrophilins, both DtpA and DtpB are predicted to be intrinsically disordered and hydrophilic in nature (Figure 2, 4). The sequence of DtpA in particular is dominated by charged amino acid residues, with 27% of the 411 amino acid protein having a positive charge and 31% having a negative charge. While the full-length sequences of DtpA and DtpB are only conserved within the Acinetobacter calcoaceticus-baumannii complex, a 50 amino acid region within the N terminus of DtpA shares sequence identity with proteins encoded by a diverse array of organisms, including other Proteobacteria as well as some Cyanobacteria, Archaea, and species of plants and fungi (Figure S6). This region of DtpA contains a highly characteristic sequence motif, KGG (Figure S3E), which is found in some species of bacteria, as well as in plants and lower eukaryotes. Intriguingly, while their functions have not been characterized, KGG motifs are repeated in some LEA proteins (Tunnacliffe and Wise, 2007), lending further support to the functional role of DtpA as a hydrophilin, and suggesting a potential shared evolutionary history between plant and bacterial hydrophilin proteins. Additionally, domain prediction analysis of the DtpA sequence showed a series of 13 AT-HOOK binding motifs within the intrinsically disordered region of the protein (Figure S3E). AT-HOOK binding motifs have been largely characterized in eukaryotic organisms, where they bind to the minor groove of AT-rich DNA and have been associated with known functional domains in chromatin- and DNA-binding proteins (Aravind and Landsman, 1998). The presence of these motifs in DtpA raises the interesting possibility that DtpA might also interact with DNA to protect from desiccation-induced damage. Additionally, our finding that A. baumannii expresses at least two hydrophilin proteins is consistent with studies of hydrophilins in plants and tardigrades, which express numerous IDPs during desiccating conditions, implying the potential for functional redundancy between these factors (Boothby et al., 2017; Hincha and Thalhammer, 2012) and raising the possibility that A. baumannii might encode one or more additional, yet-to-be identified hydrophilin proteins, thus opening an exciting area of future research in this organism.

We have determined that dtpA and dtpB expression is modulated by a conserved Lon protease. Our initial observation that mutations in lon caused increased desiccation tolerance (Figure 2) was surprising, as Lon protease is associated with maintaining protein quality control during conditions of proteotoxic stress, through the degradation of misfolded or damaged proteins (Mahmoud and Chien, 2018). However, the enhanced abundance of dtpA and dtpB in a Δlon strain (Figure 2,4) suggests that increased production of these factors could at least partially account for the desiccation resistance observed in this mutant. Because both DtpA and DtpB are predicted to be disordered in nature, it is possible that these proteins could serve as direct proteolytic targets of Lon. However, we observed that transcription of both dtpA and dtpB was highly upregulated in Δlon (Figure 3–4), demonstrating that Lon at least partially regulates DtpA and DtpB abundance indirectly, likely through the degradation of one or more transcriptional regulatory proteins. This finding is consistent with other published reports showing that, in addition to targeting unfolded proteins for destruction, Lon mediates the degradation of natively folded substrates, including DNA binding proteins required for regulating a transcriptional response to proteotoxic stress (Jonas et al., 2013; Kubik et al., 2012; Kunová et al., 2017; Lu et al., 2013; Wright et al., 1996). Moreover, we observed that a triple ΔlonΔdtpAΔdtpB mutant remains resistant to desiccation (Figure 4E), implying the possibility that additional proteins stabilized or upregulated in the absence of Lon play important roles in facilitating desiccation tolerance. Supporting this, a catalase encoded by katE which is encoded within the same genomic region as dtpA and dtpB was found to promote desiccation tolerance (Figure S5F) and exhibited increased expression in the absence of lon (Figure S5C).

While the mechanisms by which Lon protease could facilitate a transcriptional response to desiccation are unknown, recent work in Caulobacter crescentus has shown that accumulation of unfolded or aggregated proteins during proteotoxic stress can serve as a critical sensing mechanism that allows for the “titration” of Lon by unfolded substrates, resulting in the stabilization of regulatory proteins normally targeted by this enzyme (Zeinert et al., 2020). It is therefore intriguing to hypothesize that, in addition to its protein quality control functions, the Lon protease of A. baumannii might play an important role in coordinating a regulatory response to conditions that induce proteotoxic stress, including desiccation. Supporting this hypothesis, during the course of our investigation, a study was published corroborating the observation that Δlon is more resistant to desiccation (Wang et al., 2020). This report found that protein aggregates are formed in A. baumannii in response to desiccation and, intriguingly, observed that induction of proteotoxic stress with a ribosome-targeting antibiotic prior to desiccation increases the desiccation tolerance of a WT strain (Wang et al., 2020). While this study attributed the enhanced desiccation tolerance of Δlon to the formation of protein aggregates in this mutant, our work demonstrates that the increased expression of dtpA and dtpB is at least partially responsible for this phenotype. Moreover, the finding that treatment with a ribosome targeting antibiotic enhances desiccation tolerance raises the possibility that unfolded or aggregated proteins could titrate available pools of Lon, thereby allowing for the accumulation of regulatory protein (s) responsible for controlling the transcription of dtpA and dtpB.

While it is unknown which regulatory protein(s) accumulate in the absence of Lon, this work demonstrates that Lon-dependent transcriptional regulation might occur through the modulation of the BfmRS two component system, which regulates the expression of more than 1,800 genes (Geisinger et al., 2018), and has been previously implicated in a number of vital stress response and infection-related functions, including biofilm formation (Tomaras et al., 2008), intrinsic antibiotic resistance (Geisinger et al., 2018), survival in mammalian hosts (Wang et al., 2014), as well as in promoting desiccation tolerance (Farrow et al., 2018). Through genetic suppressor analysis, we showed that mutation of bfmR suppresses the enhanced desiccation and oxidative stress tolerance of the Δlon mutant (Figure 3E, Figure S4B). Moreover, in early-stationary phase, we observed that deletion of bfmR reverses the transcriptional upregulation of dtpA and dtpB observed in Δlon (Figure 3G, Figure 4D). These data are consistent with a model where BfmR serves as a direct proteolytic substrate of Lon; however it is also possible that these two factors may converge on dtpA and dtpB regulation indirectly, perhaps through modulation of a shared regulatory target. Additionally, we found that a ΔlonΔbfmR mutant exhibits intermediate levels of dtpA expression in late-stationary phase relative to each individual mutant (Figure 3H), suggesting the existence of an as-yet unidentified BfmR-independent regulation controlling dtpA transcription.

In addition to the initial characterization of DtpA and DtpB, in this work we report the observation that A. baumannii rehydrated from desiccation causes more virulent infections than liquid-cultured bacteria (Figure 1). This finding is consistent with a growing body of literature documenting the link between environmental stress resistance and virulence in A. baumannii. For example, mutants of A. baumannii unable to resist environmental stresses such as disinfectants and antibiotics are avirulent in animal models of infection, suggesting that mechanisms used to resist environmental stress are also beneficial in the context of the host immune response (Chin et al., 2018; Gebhardt et al., 2015; Roux et al., 2015; Tipton et al., 2018). Additionally, exposure of A. baumannii to antibiotics or ethanol prior to infection enhances its virulence, indicating that phenotypic adaptations to stresses in the environment might increase the ability of this pathogen to cause disease (Gandhi et al., 2014; Geisinger and Isberg, 2015). In this study, we observed increased resistance to a number of environmental and host-derived stressors, including the oxidants H₂O₂ and HOCl, as well as heat shock (Figure 1, Figure S2) after desiccation. Consistent with this, A. baumannii exhibited transcriptional induction of katE (Figure S5E) and dtpA (Figure 3C) following desiccation, raising the possibility that upregulation of these factors and others could contribute to the enhanced stress-tolerance and virulence displayed following desiccation. Moreover, it has been previously reported that activation of the BfmRS system by sublethal concentrations of certain ribosome-targeting antibiotics enhances the virulence of A. baumannii in a murine infection model (Geisinger and Isberg, 2015). Given the central role for BfmRS in surviving desiccation, it possible that a similar regulatory circuit may become activated in response to drying. Together, these results highlight the importance of incorporating the use of desiccated bacteria into animal and in vitro studies of A. baumannii pathogenesis, as these models represent an important and clinically relevant mode of transmission.

Finally, we report that heterologous expression of dtpA in a probiotic bacterium can increase its desiccation tolerance (Figure 6). These results are consistent with previous reports that heterologous expression of tardigrade hydrophilins protects lab-domesticated strains of E. coli from desiccation (Boothby et al., 2017). By adapting this approach to a probiotic bacterium, the commensal organism E. coli Nissle, we demonstrate a potential utility of expressing hydrophilin proteins to preserve the shelf-life of commercially distributed probiotics, which are frequently packaged into capsules and distributed at room temperature (Meng et al., 2008). We also show that DtpA increases gut colonization by a dehydrated version of EcN, supporting the idea that DtpA expression may increase the efficacy of probiotic therapies.

In summary, this work highlights a connection between desiccation tolerance and pathogenicity in A. baumannii. These findings have significant clinical relevance, as studies modeling the virulence of desiccated A. baumannii might better mimic the primary mode of transmission for this pathogen. Additionally, we establish a class of intrinsically disordered A. baumannii hydrophilin proteins, and show that these proteins are required for desiccation tolerance. These data support a role for IDPs in bacterial stress tolerance and reveal potential applications for bacterial hydrophilins to preserve activity of protein- and live bacteria-based pharmaceuticals that require desiccation for long-term storage and transport.

STAR Methods

RESOURCE AVAILABILITY

Lead Contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Eric Skaar (eric.skaar@vumc.org).

Materials availability

All materials generated in this study are available from the lead contact upon request.

Data and code availability

• Raw sequencing files obtained from Tn-seq experiments are available in the National Center for Biotechnology Information (NCBI) Gene Expression Omnibus (GEO) under accession number: GSE198004. Whole genome sequencing data is available in the NCBI sequence read archive (SRA) under BioProject: PRJNA813302.

• This paper does not report original code.

• Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

EXPERIMENTAL MODEL AND SUBJECT DETAILS

Animal experiments

Mouse experiments were performed using female 8–12 week old C57BL/6 (Jackson Laboratories stock no. 000664) mice. Animals were maintained at the Vanderbilt University Medical Center (VUMC) Animal Facilities, with a 12 h light-dark cycle and food and water provided ad libitum. For experimental endpoints, animals were humanely euthanized. All animal experiments were approved and performed in compliance with the Institutional Animal Care and Use Committee (IACUC) of Vanderbilt University (protocol number M1900043–00) and conform to policies and guidelines established by VUMC, the Animal Welfare Act, the National Institutes of Health, and the American Veterinary Medical Association.

Bacterial strains and culture conditions

The strains used in this study are described in Table S2. Strains were cultured in lysogeny broth (LB) or on LB agar plates at 37°C. Unless otherwise noted, antibiotics were used at the following concentrations: kanamycin (Km) (Sigma), 40 μg/mL; carbenicillin (Carb) (Fisher), 75 μg/mL; chloramphenicol (Cm) (Fisher), 15 μg/mL; and tetracycline (Tet) (Alfa Aesar), 10 μg/mL.

Cell lines and primary cell cultures

To generate bone marrow-derived macrophages (BMMφs), bone marrow was isolated from the tibias and femurs of mice C57BL/6J (Jackson Laboratories stock no. 000664), cells were isolated using Lymphocyte Separation Medium (Cedarlane Laboratories), and differentiated with BMMφ differentiation media containing Dulbecco’s Modified Eagle Medium containing 10% (vol/vol) fetal bovine serum (FBS) and L929 cell (European Collection of Cell Cultures) supernatant.

METHOD DETAILS

Desiccation tolerance assays.

With the exception of heterologous dtpA expression experiments, all desiccation tolerance assays were performed as follows. For desiccations with mid-exponential phase bacteria (Figure S1), single colonies were inoculated into LB medium and grown overnight at 37°C with shaking at 180 rpm, then diluted 1:1000 into fresh medium and grown for 3.5 hours. For all other desiccations, single colonies were inoculated into LB medium and grown for 8 hours at 37°C with shaking at 180 rpm. Bacteria were then centrifuged, washed twice in PBS, and resuspended in PBS at a concentration of 1×10¹¹ CFU/mL. A volume of 10 μL was then spread sterilely across the surface of the well of a 24-well polystyrene plate. A portion of the starting population was serially diluted and plated on LB agar to determine the input CFU. Plates were incubated at 20–30% relative humidity and ambient temperature. At defined intervals, desiccated bacteria were rehydrated in PBS, serially diluted, and plated on LB agar to quantify surviving bacteria.

Starvation assays.

Single colonies were inoculated into LB medium and grown for 8 hours at 37°C with shaking at 180 rpm. Bacteria were then centrifuged, washed twice in PBS, and resuspended in PBS at a concentration of 1×10¹¹ CFU/mL. A volume of 10 μL was then spread sterilely across the surface of 1X PBS agar embedded in the well of a 24-well plate. A portion of the starting population was serially diluted and plated on LB agar to determine the input CFU. Plates were incubated at ambient humidity and ambient temperature. At defined intervals, bacteria were resuspended from PBS agar, serially diluted, and plated on LB agar to quantify surviving bacteria.

A. baumannii mouse infections.

C57Bl/6 female mice were purchased from Jackson Laboratories and infected at 8–12 weeks of age. Two days prior to infection, single colonies were inoculated into LB medium and grown for 8 hours at 37°C with shaking at 180 rpm. Bacteria were then centrifuged, washed twice in PBS, and resuspended in PBS at a concentration of 3×10¹¹ CFU/mL. A volume of 100 μL was then spread sterilely across the surface of a 100 mm polystyrene petri dish. Petri dishes were incubated at 20–30% relative humidity and ambient temperature for 48 hours. On the day of infection, single colonies were inoculated into LB medium and grown for 8 hours at 37°C with shaking at 180 rpm. Bacteria were then harvested, washed twice in PBS, and resuspended in PBS at a concentration 1×10¹⁰ CFU/mL. Desiccated bacteria were rehydrated in PBS, washed twice in PBS, and resuspended at an equivalent optical density to liquid-grown inocula for a final concentration 1×10¹⁰ CFU/mL. Bacterial inocula were serially diluted and spot-plated to confirm equivalent bacterial concentrations between groups.

For competitive infections, suspensions of WT A. baumannii were desiccated as described above, and combined at a 1:1 ratio with a liquid culture of A. baumannii containing a transposon conferring kanamycin resistance (Km^R) inserted in a locus with no impact on in vivo fitness (ACX60_00265::himar1). In parallel, liquid-grown WT kanamycin-sensitive (Km^S) and Km^R A. baumannii strains were prepared as described above and combined at a 1:1 ratio. Bacteria were serially diluted and spot-plated onto both LB agar and LB Km 40 agar to confirm equivalent bacterial populations in the inocula.

Prior to infections, mice were anesthetized by intraperitoneal injection of 2,2,2 tribromoethanol diluted in PBS. Anesthetized mice were infected intranasally with 4×10⁸ CFU of the inoculum in a 40 μL volume. Infections proceeded for 24 (ABUW_5075) or 36 (ATCC 17978) hours, during which time survival was monitored. Mice were euthanized by forced CO₂ inhalation and lungs, livers, and spleens were sterilely harvested and placed on ice. Organs were homogenized and serial dilutions of homogenized tissues were spot plated onto LB agar for enumeration. For competitive infections, dilutions were plated onto both LB agar and LB Km 40 agar to quantify each population, and competitive index was determined by dividing the ratio of Km^S/ Km^R bacteria in the output to input pools.

Stress assays.

Two days prior to experiments, single colonies were inoculated into LB medium and grown for 8 hours at 37°C with shaking at 180 rpm. Bacteria were then harvested, washed twice in PBS, and resuspended in PBS at a concentration of 2×10¹¹ CFU/mL. A volume of 50 μL (containing 1×10¹⁰ CFU) was then spread sterilely across to the surface of a 60 mm polystyrene petri dish. Petri dishes were incubated at 20–30% relative humidity and ambient temperature for 48 hours. On the day of experimentation, single colonies were inoculated into LB medium and grown for 8 hours at 37°C with shaking at 180 rpm. Bacteria were then harvested, washed twice in PBS, and resuspended in PBS at a concentration of 1×10⁸ CFU/mL. Desiccated bacteria were rehydrated in PBS, washed twice in PBS, and resuspended at an equivalent optical density to liquid-grown inocula for a final concentration 1×10⁸ CFU/mL. For growth curves, bacteria were diluted 1:100 into fresh LB media and OD₆₀₀ was measured at 15-minute intervals for 15 hours using a plate reader (BioTek). For H₂O₂ and HOCl killing assays, bacterial suspensions were incubated with a range of oxidant concentrations, as indicated in figure legends. Following 30 minutes of incubation, bacterial cultures were serially diluted in PBS and dilutions were spot-plated onto LB agar for CFU enumeration. For samples subcultured from desiccation prior to H₂O₂ treatment, bacteria desiccated for 48 hours were rehydrated, diluted 1:1000 into LB media, and incubated for 8 hours at 37°C with shaking at 180 rpm. Samples were then subjected to H₂O₂ killing as described above. For heat tolerance assays, bacterial suspensions were incubated at 50°C for 30 minutes, then serially diluted in PBS and spot plated on LB agar for CFU enumeration. Parallel samples sere incubated at room temperature for 30 minutes, and % survival was determined by dividing CFU recovered after heating to the room temperature control.

Macrophage isolation and infections.

To differentiate bone marrow microphages (BMMφs), we generated L929 cell (European Collection of Cell Cultures) supernatant by plating 2.5×10⁵ cells into a T150 flask with 50 μL of D10 media (Dulbecco’s Modified Eagle Medium + 10% (vol/vol) fetal bovine serum (FBS)) and incubating for 12 days (37 °C, 5% CO₂). After 12 days, supernatant was collected and filtered. Single-cell suspensions of bone marrow were prepared from the tibias and femurs of mice. Mononuclear cells were isolated by using Lymphocyte Separation Medium (Cedarlane Laboratories), plated in a 60-mm Petri dish with 6 mL of BMMφ differentiation media (D10 media with 10% (vol/vol) L-cell supernatant), and cultured overnight (37 °C, 5% CO₂). Nonadherent cells were plated into non-tissue culture-treated 100-mm petri dishes (1 mL cells per Petri dish) with 7 mL of fresh BMMφ differentiation media. To promote BMMφ differentiation, cells were incubated for 6 days (37 °C, 5% CO₂) with an additional 5 mL of BMMφ differentiation media being added on day 4. The resulting BMMφs were removed from the dish by washing with ice cold PBS. BMMφ cultures were 98% CD11b+, I-Alo and B7.2lo. BMMφs were rested on ice for at least 30 min in D10 media and then transferred to ultra-low cluster round bottom 96-well plates (Costar) in D10 media and incubated (37 °C, 5% CO₂) for 2 hours prior to experiments (200,000 BMMφs per well). Two days prior to infections, single colonies were inoculated into LB medium and grown for 8 hours at 37°C with shaking at 180 rpm. Bacteria were then harvested, washed twice in PBS, and resuspended in PBS at a concentration of 2×10¹¹ CFU/mL. A volume of 50 μL (containing 1×10¹⁰ CFU) was then spread sterilely across the surface of a 60 mm polystyrene petri dish. Petri dishes were incubated at 20–30% relative humidity and ambient temperature for 48 hours. One day prior to experimentation, single colonies were inoculated into LB medium and grown overnight at 37°C with shaking at 180 rpm. Overnight cultures and desiccated bacteria were then harvested, washed twice in PBS, and resuspended in PBS. Bacteria suspensions were normalized to an equivalent OD₆₀₀, briefly opsonized in non-heat inactivated serum for 30 min on ice, diluted in PBS, and added to the BMMφs at a MOI of 1:10 (20,000 CFU per 200,000 BMMφs). After 1 hour, samples were serially diluted and spot-plated onto LB agar.

Transposon library generation.

The vector pJNW684 (Wang et al., 2014) containing a Himar1 transposon was used to generate the transposon mutant pool as follows. To perform transposition, SM10γpir containing pJNW684 was mated with WT A. baumannii ATCC 17978. Prior to mating, bacterial cultures were grown overnight at 37°C with shaking at 180 rpm. Strains were washed three times in 1X PBS, and SM10γpir (pJNW684) was mixed with WT A. baumannii at a 2:1 ratio. Mixtures were transferred to LB agar plates and matings were allowed to proceed for 4 hours at 37°C. Bacteria were then resuspended from plates and a fraction of the mating was spread across the surface a LB agar plate containing Km 40 and Cm 15. A library pool containing approximately 90,000 colonies was then scraped off plates, resuspended in a medium containing 20% glycerol, and stored at −80°C.

Tn-seq screen.

Transposon library aliquots were inoculated into 10 mL LB medium (three biological replicates per condition) and grown for 8 hours at 37°C with shaking at 180 rpm. Bacteria were then harvested, washed twice in PBS, and resuspended in PBS at a concentration of 3×10¹¹ CFU/mL. A volume of 100 μL (containing 3×10¹⁰ CFU) per sample was spread sterilely across the surface of a 100 mm polystyrene petri dish. Petri dishes were incubated at ambient humidity and temperature. After 2 and 14 days of desiccation, bacteria were rehydrated in 1 mL of PBS. For 2-day samples, 5 μl of rehydrated bacteria were diluted into 10 mL of LB. For 14-day samples, 50 μl of rehydrated bacteria were diluted into 10 mL of LB. Cultures were grown for 4 hours at 37°C with shaking at 180 rpm to enrich for surviving transposon mutants. In parallel, input samples were prepared by inoculating transposon library aliquots into 10 mL LB medium and growing for 8 hours at 37°C with shaking at 180 rpm. Following outgrowth, bacterial cultures were pelleted and stored at −80°C.

Tn-seq library preparation and analysis.

DNA libraries were prepared for sequencing using the homopolymer tail-mediated ligation PCR technique (Lazinski and Camilli, 2013). Genomic DNA (gDNA) was extracted from bacterial pellets using the Qiagen DNeasy Blood and Tissue kit according to manufacturer’s instructions. gDNA was sheared by sonication using the Covaris LE220 instrument to generate 350 bp fragments. Sheared DNA was treated with terminal deoxytransferase to generate a 3’ poly C-tail sequence, and two rounds of nested PCR were employed to amplify transposon junction regions. These products were multiplexed using 8-bp indexing primers and sequenced on the Illumina HiSeq 2500 at Tufts University Core Facility. Following sequencing, reads were trimmed, filtered for quality, and mapped to the ATCC A. baumannii 17978 accession NZ_CP012004. A “Dval” score was assigned to each gene in each library pool, representing the aggregate number of reads for all transposon insertions within a gene in a given library sample, divided by the total number of predicted reads for that gene based on its size and the total number of reads obtained for the library pool. Output Dval scores were normalized to input Dval scores to calculate a fitness score for each gene in each respective condition. Fitness scores were Log₂-transformed and an average Log₂ fitness score was calculated for each gene in each condition analyzed (2 and 14 days of desiccation). A Z-score was calculated that represents the number of standard deviations from the mean for each respective gene (Table S1).

Strain generation.

The strains and plasmids used in this study are listed in Table S2 and primers are listed in Table S3. Δlon, ΔdtpA, ΔlonΔdtpA, ΔdtpB, ΔdtpAΔdtpB, ΔlonΔdtpB, ΔlonΔdtpAΔdtpB, ΔbfmR, ΔlonΔbfmR, ΔbfmS, ΔkatE, and ΔlonΔkatE mutants were created via allelic exchange as follows. For generation of the Δlon, ΔbfmR, ΔbfmS and ΔkatE mutants, approximately 1,000 bp of DNA in both the 5’ and 3’ flanking regions surrounding targeted genes was amplified using A. baumannii genomic DNA as a PCR template. The kanamycin resistance gene aphA was amplified by PCR from the vector pUCK1. These products were cloned into the pFLp2 vector using HiFi Assembly (New England Biolabs). The resulting pFLp2 constructs were then introduced into WT A. baumannii by tri-parental conjugation using an E. coli HB101 strain containing the helper plasmid pRK2013. Matings were plated onto LB Carb 75 Cm 15 agar to select for strains containing the integrated plasmid. Strains were then plated onto agar containing 10% sucrose to select for clones that had resolved the integrated plasmid and resulting sucrose-resistant colonies were patched onto LB Kan 40 to screen for the loss of target gene and replacement with aphA. Deletion of loci was confirmed by multiple PCRs using both A. baumannii and aphA-specific primers. For construction of the ΔdtpA and ΔlonΔdtpA mutants, 1,000 bp of DNA in both the 5’ and 3’ flanking regions surrounding dtpA was amplified by PCR. The tetracycline resistance gene tetA was amplified using genomic DNA isolated from the tetracycline-resistant Acinetobacter baumannii isolate AB0057 as a PCR template. Products were cloned into pFLp2 and the resulting plasmid was introduced into WT and Δlon by conjugation as described above. Transconjugants were selected for by plating matings onto LB Carb 75 Cm 15 agar, and the replacement of dtpA with tetA was selected for and confirmed as described above. For generation of ΔdtpB, ΔdtpAΔdtpB, ΔlonΔdtpB, and ΔlonΔdtpAΔdtpB mutants, 1,000 bp of DNA in both the 5’ and 3’ flanking regions surrounding dtpB was amplified by PCR and products were cloned into pFLp2 and the resulting plasmid was introduced into WT, Δlon, ΔdtpA, and ΔlonΔdtpA strains by conjugation as described above. Transconjugants were selected for by plating matings onto LB Carb 75 Cm 15 agar, and the unmarked deletion of dtpB was selected for by sucrose selection as described above, and confirmed by PCR. For lon and dtpA complementation vectors, open reading frames (ORFs) were amplified by PCR, and HiFi Assembly was used to clone products into the pWH1266 vector downstream of a constitutive A. baumannii rpsA promoter (P_rpsA). To generate the P_dtpA-luciferase reporter plasmid, an approximately 400-bp segment of the dtpA promoter was PCR amplified, and Hifi Assembly was used to clone into the p.luxABCDE.MU368.tet. plasmid (Juttukonda et al., 2016) upstream of the luxABCDE operon. To generate the DtpA purification construct, the dtpA ORF was amplified, and Hifi Assembly was used to clone into pET15b (Novagen, EMD Millipore) to generate a N-terminal hexahistidine-tagged construct. For expression, the recombinant pET15b-dtpA plasmid was transformed into BL21 (DE3) pRIL cells. To generate an inducible dtpA expression vector in E. coli Nissle, the dtpA ORF was amplified by PCR and HiFi Assembly was used to clone into pTRC99a downstream of the Ptrc promoter. Transposon mutants in the A. baumannii ABUW 5075 strain were obtained from the University of Washington A. baumannii mutant library.

Insoluble protein isolation.

Fresh A. baumannii colonies were inoculated into LB medium and incubated for 8 hours at 37 °C with shaking at 180 rpm. Following shaking, cells were harvested by centrifugation, washed twice with 1X PBS and normalized to an OD₆₀₀ of 1 in 20 mL total volume. Cells were then pelleted and insoluble fractions were isolated as previously described. Briefly, cell pellets were resuspended in 40 μL of Buffer A (10 mM potassium phosphate buffer, pH 6.5, 1 mM EDTA, 20% w/v sucrose, 1 mg/mL lysozyme) and incubated on ice for 30 minutes. 360 μL of Buffer B (10 mM potassium phosphate buffer, pH 6.5, 1 mM EDTA) was then added and lysates were sonicated for 8 cycles with 30 seconds on, 30 seconds off at 50% amplitude. Lysates were centrifuged at 2,000 x g for 15 minutes at 4°C to remove unlysed cells. Supernatants were transferred to a fresh tube and centrifuged at 15,000 x g for 20 minutes at 4°C to pellet insoluble content. An additional wash in Buffer B was performed, then pellets were resuspended in 400 μL of Buffer C (10 mM potassium phosphate buffer, pH 6.5, 1 mM EDTA, 2% w/v IGEPAL [Sigma]) and centrifuged again 15,000 x g for 30 minutes. An additional wash in Buffer C was performed, and pellets were resuspended in 400 μL of Buffer B. Samples were centrifuged at 15,000 x g for 20 minutes at 4°C, then resuspended in 200 μL of Buffer B. A portion of each fraction was analyzed by sodium dodecyl sulphate–polyacrylamide gel electrophoresis (SDS-PAGE).

Mass spectrometry.

A band migrating at 46-kD was excised and sent to the University of South Florida Proteomics Core facility for mass spectrometry analysis. The gel section was minced and destained before being reduced with dithiothreitol (DTT), alkylated with iodoacetamide (IAA), and finally digested with Trypsin/Lys-C overnight at 37˚C. Peptides were extracted using 50/50 acetonitrile (ACN)/H₂O/0.1% formic acid and dried in a vacuum concentrator (Labconco). Peptides were resuspended in 98%H₂O/2%ACN/0.1% formic acid for LC-MS/MS analysis. Experiments were performed in triplicate. Peptides were separated using a 50cm C18 reversed-phase HPLC column (Thermo) on an Ultimate3000 UHPLC (Thermo) with a 120-minute gradient (2–32% acetonitrile with 0.1% formic acid) and analyzed on a hybrid quadrupole-Orbitrap mass spectrometer (Q Exactive Plus, Thermo Fisher Scientific) using data-dependent acquisition in which the top 10 most abundant ions are selected for MS/MS analysis. Raw data files were processed in MaxQuant (www.maxquant.org) and searched against the current Uniprot UP000036024 protein sequence database. Search parameters included constant modification of cysteine by carbamidomethylation and the variable modification, methionine oxidation. Proteins were identified using the filtering criteria of 1% protein and peptide false discovery rate.

Prediction of intrinsically disordered regions.

The DtpA and DtpB amino acid coding sequences were analyzed using the Prediction of Naturally Disordered Regions (PONDR) program to predict intrinsically disordered motifs. The resulting prediction was scored from a scale of −1.0 (ordered) to 1.0 (disordered) and mapped across the protein sequence.

Circular dichroism.

To purify ¹⁵N-labeled DtpA, BL21 (DE3) pRIL harboring pET15b-dtpA was grown in LB with 35 μg/mL chloramphenicol and 74 μg/mL carbenicillin. The small-scale overnight culture was then transferred to unlabeled or ¹⁵N-labeled large scale culture, to which 35 μg/mL chloramphenicol and 74 μg/mL carbenicillin was added and grown at 37 °C until an OD₆₀₀ between 0.6 – 0.8 was reached. Once the OD₆₀₀ was reached, 1 mM IPTG was added into each large-scale culture to induce expression and then continue shaking overnight at 16 °C. Cells were harvested by centrifugation for 20 min at 4000 rpm and resuspended with 25 mL of Lysis buffer (50 mM HEPES pH 8.0, 150 mM NaCl, 20 mM imidazole, 1 mg/mL lysozyme, 1x Protease inhibitor cocktail [Millipore]). Samples were sonicated for a total of 20 min with a 30 s on pulse and a 30 s off pulse. Lysates were centrifuged for 10 min at 13,000 rpm and the resulting pellets were resuspended in 20 mL of wash buffer (50 mM HEPES pH 8.0, 150 mM NaCl, 25 mM imidazole, 1 mM PMSF, 8 M Urea), and centrifuged for 10 min at 13,000 rpm. Low imidazole buffer (50 mM HEPES pH 8.0, 150mM NaCl, 25 mM imidazole 1 mM PMSF) and high imidazole buffer (50 mM HEPES pH 8.0, 150mM NaCl, 500 mM imidazole 1 mM PMSF) were prepared and used to prepare the Akta pure FPLC (GE) for equilibration with a Ni-NTA affinity column. Samples were injected into a 50 mL Superloop and a program was set up with an equilibrium, a gradient wash, and an elution step to collect the peak fraction that showed the protein that still contains the 6xHis tag. Once peak fractions were collected, the protein was added to the dialysis tubing and placed into dialysis buffer (50 mM HEPES pH 8.0, 150 mM NaCl, 5% glycerol, 3M urea) and allowed to stir for 2 hours at 4°C. After 2 hours, dialysis buffer without urea was refreshed (50 mM HEPES pH 8.0 150 mM NaCl, 5% glycerol) and stirred at 4°C. Once protein was purified, ¹⁵N-labeled DtpA was dialyzed into a transparent buffer optimal for CD (25 mM phosphate pH 8.0, 150 mM NaF). The Jasco J-715 CD instrument was booted based on the operating manual before placing the sample into the sample holder (Jasco). ¹⁵N-Labeled DtpA (4.3 μM) was pipetted into a 1 mm glass cuvette and inserted into the sample holder. Temperature setting was set to collect from 30°C - 50°C with 5°C increments. The range of wavelengths collected was from 280 nm to 190 nm with a 0.1 nm data pitch, and a scan speed of 50 nm/min. A total of 5 accumulations were taken with a 10 second delay time. Data that were collected were an average of the 5 accumulations at each temperature and converted from mdeg to molar ellipticity. The data were subtracted by taking the average of the largest positive signal within the range of wavelengths plotted to obtain a baseline near 0 for each run.

RNA isolation.

A. baumannii strains were inoculated from freshly streaked colonies into LB and grown at 37°C with shaking at 180 rpm. At timepoints indicated in figure legends, an equivalent of 1×10¹⁰ bacteria was removed, centrifuged, and resuspended in 1 mL of TRIzol Reagent (Ambion), then stored at −80°C until time of RNA isolation. For isolation of RNA from desiccated bacteria, bacteria were grown for 16 hours, cells were harvested and washed twice in PBS, then resuspended in PBS at a concentration of 2×10¹¹ CFU/mL. A volume of 50 μL (containing 1×10¹⁰ CFU) was then spread sterilely across the surface of a 60 mm polystyrene petri dish. For time zero samples, bacteria were immediately resuspended in 1 mL of TRI Reagent (Sigma) then stored at −80°C until time of RNA isolation. Desiccated samples were incubated at 20–30% relative humidity and ambient temperature for 4 hours, then resuspended in 1 mL of TRI Reagent (Sigma) and stored at −80°C until time of RNA isolation. For RNA isolation, resuspended bacteria were homogenized in a bead beater with Lysing Matrix B beads (MP Biomedical) at a speed of 6 m/s for 45 seconds. Homogenized suspensions were then centrifuged, and the upper phase was collected and mixed with 200 μL of chloroform (Acros Organics) by vigorous shaking for 15 seconds. Samples were incubated at room temperature for 2 minutes, centrifuged at 4°C for 15 minutes, and 400 μL of the upper aqueous phase was collected. RNA was precipitated from the upper aqueous phase by mixing with 400 μL of 70% ethanol (Sigma) and purified using the PureLink RNA kit (Invitrogen), according to manufacturer’s instructions. Purified RNA was eluted in 50 μL of DNase-Free, RNase-Free water (ThermoFisher). DNA contamination was removed using the Turbo DNA-free kit (Invitrogen) according to manufacturer’s instructions. RNA was stored at −20°C.

Quantitative RT-PCR.

cDNA was generated from 1 μg of RNA using the iSCRIPT™ cDNA Synthesis Kit (Bio-Rad) according to manufacturer’s instructions, and subjected to qRT-PCR using iQ SYBR green supermix (Bio-Rad) with the primer pairs listed in Table S3. cDNA template was diluted 1:100 for reactions performed with 16S rRNA, dtpB, and katE primers and left undiluted for dtpA amplification. Amplification was performed on a CFX96 qPCR cycler (Bio-Rad) using a 3-step melt curve program. Threshold cycle (CT) values for each transcript were normalized by 16S rRNA.

dtpA-reporter luminescence assay

For reporter luminescence assays, A. baumannii strains harboring a dtpA reporter luminescence plasmid were grown overnight at 37°C in LB medium. Cultures were back-diluted 1:100 in LB medium, incubated at 37°C for two hours, then back-diluted 1:50 into a black-sided 96-well plate (Corning), and luminescence and optical density were measured using a plate reader at 30-minute intervals (BioTek).

Whole genome sequencing and suppressor analysis.

Genomic DNA (gDNA) was extracted from WT, Δlon, and Δlon suppressor mutant strains using the Qiagen DNeasy Blood and Tissue kit according to manufacturer’s instructions. Purified gDNA was sequenced by MiGS using the Illumina NextSeq 550 platform with 150 Mbs coverage. Adapter sequences were trimmed and genome sequences were mapped to the ATCC A. baumannii 17978 accession NZ_CP012004 and NZ_CP012005 using breseq version 30.0 (Deatherage and Barrick, 2014) with default settings. The spontaneous suppressor mutation uncovered by whole genome sequencing was located in the receiver domain of BfmR and mapped onto the structure of the beryllium fluorinated receiver domain of BfmR (Russo et al., 2016) using PyMOL. The homodimer was aligned with a monomer to display the two dimmer interfaces.

Catalase activity assay

For reporter catalase assays, A. baumannii strains were inoculated from freshly streaked colonies into LB and grown at 37°C for 8 hours with shaking at 180 rpm. Cells were harvested by centrifugation, washed twice with 1X PBS, and normalized to an OD₆₀₀ of 1. Catalase activity was assessed using a Catalase Assay kit (Cayman Chemical) according to manufacturer’s instructions. Prior to experimentation, a fraction of diluted cells was serially diluted and spot-plated to determine CFU/mL. Activity was normalized to the CFU/mL of each sample.

DtpA expression and purification for enzyme protection assays.

DtpA purification was adapted from prior IDP purification protocols as follows. E. coli BL21 (DE3) pRIL containing pET15b-dtpA was grown in LB Carb 75 Cm 34 at 37°C to an OD₆₀₀ of 0.5–1.0 before induction with 1 mM IPTG (Sigma). Following induction, bacteria were maintained at 37°C for an additional 18h. Bacteria were then centrifuged at 8,000 rpm for 20 min, and pellets were stored at –80°C. Prior to purification, pellets were thawed and resuspended in lysis buffer (50 mM HEPES [pH 8], 300 mM NaCl, 20 mM imidazole, 1 mg/mL lysozyme, protease inhibitor cocktail [Sigma]). Cells were lysed by sonication for a total of 18 cycles (30s on, 2 minutes off) at 90% amplitude. Following sonication, lysates were centrifuged at 8,000 rpm for 20 min, and pellets (containing insoluble content) were resuspended in wash buffer (50 mM HEPES [pH 8], 300 mM NaCl, 25 mM imidazole). For batch purification of DtpA, Ni-nitrilotriacetic acid (NTA) (Qiagen) was added to resuspended pellets, incubated for 15 minutes at 4°C with rocking, and centrifuged at 4,000 rpm. Pellets containing NTA-bound protein were washed twice as described above for a total of 20 bed volumes. Pellets were then subjected to sequential washes with buffer containing increasing concentrations of imidazole (100 mM, 150 mM, 200 mM, 250 mM, 300 mM, 500 mM). For each wash step, resuspended pellets were incubated for 15 minutes at 4°C with rocking, transferred to 95°C for 10 minutes to solubilize unbound DtpA, then centrifuged at 4,000 rpm for 10 minutes to pellet NTA resin. Supernatant fractions were collected after each wash and analyzed by SDS-PAGE. DtpA eluted at 300 mM imidazole. Purified protein was added to a dialysis cassette with a 20 kD molecular weight cutoff (Slide-A-Lyzer™ Dialysis Cassette, ThermoFisher), placed into dialysis buffer (50 mM HEPES pH 8, 150 mM NaCl, 5% glycerol), and allowed to stir for 2 hours at 4°C. After 2 hours, dialysis buffer was replaced and stirred overnight at 4°C. Total protein concentration was measured with the Pierce™ BCA Protein Assay Kit (ThermoFisher Scientific) using BSA as standard.

β-lactamase activity assays.

For desiccation protection assays, 50 nM (in 25 μL) recombinant β-lactamase (Abcam) was applied to a well of a 96-well plate and mixed with an equal volume of purified DtpA or BSA (Sigma) at 1:1, 1:2.5, 1:5, and 1:10 molar ratios, or with a vehicle control. Plates were placed uncovered in a 37°C incubator for 16 hours. Desiccated proteins were then rehydrated in 50 μL of 1X PBS and mixed with an equal volume of the colorimetric substrate nitrocefin (500 μg/mL in 1X PBS) (Abcam). Absorbance at 410 nm was measured after 15 minutes of incubation to quantify β-lactamase activity. Parallel experiments were performed using non-desiccated proteins and % activity was determined by dividing the activity of desiccated samples by that of non-desiccated controls. For heat-protection experiments, proteins were mixed as above, covered, and placed in a 65°C incubator for 10 minutes. Fifty μL of nitrocefin was then added to each well and absorbance was measured as described above. Control samples were prepared by measuring the activity of non-heated protein mixtures, and % activity was determined by dividing the activity of heated samples by that of non-heated controls.

Heterologous expression experiments

E. coli Nissle strains expressing dtpA under the trc promoter (pTRC99a-dtpA) or harboring an empty vector were inoculated into LB medium with Carb 75 and incubated for 8 hours at 37°C with shaking at 180 rpm. After 8 hours of growth, dtpA expression was induced with 0.5 mM IPTG or with vehicle treatment (uninduced control). Induction was allowed to proceed for 2 hours at 37°C with shaking at 180 rpm. Following induction, cells were harvested by centrifugation, washed twice in PBS, and resuspended in PBS at a concentration of 1×10¹¹ CFU/mL. A volume of 10 μL was then spread sterilely across to the surface of the well of a 24-well polystyrene plate. A portion of the starting population was serially diluted and plated on LB agar to determine the input CFU. Plates were incubated at 20–30% relative humidity and ambient temperature. A fraction of the starting population was serially diluted and plated on LB agar to determine the input CFU. After 7 days, desiccated bacteria were rehydrated in PBS, serially diluted, and plated on LB agar to quantify surviving bacteria.

E. coli Nissle mouse colonization.

7 days prior to colonization, E. coli Nissle pTRC99a-dtpA and empty vector control strains were inoculated into LB medium with Carb 75 and incubated for 8 hours at 37°C with shaking at 180 rpm. After 8 hours of growth, 0.5 mM IPTG was added to both pTRC99a-dtpA and empty vector strains. Induction was allowed to proceed for 2 hours at 37°C with shaking at 180 rpm. Bacteria were then harvested, washed twice in PBS, and resuspended in PBS at a concentration of 2×10¹¹ CFU/mL. A volume of 50 μL (containing 1×10¹⁰ CFU) was then spread sterilely across to the surface of a 60 mm polystyrene petri dish. Petri dishes were incubated at 20–30% relative humidity and ambient temperature for 7 days. A fraction of the starting population was serially diluted and plated on LB Carb 75 agar to determine the input CFU. Streptomycin (Fisher) (2g/L) was provided in the drinking water of 8–10-week-old female C57Bl/6 mice from Jackson Laboratories for 48 hours, then replaced with normal drinking water for 48 hours. On day of colonization, desiccated bacteria were rehydrated, serially diluted, and spot-plated to determine inoculum titer. Rehydrated surviving bacteria were intragastrically inoculated into mice and fecal pellets were collected after 8 hours. Presence of E. coli Nissle in fecal pellets was determined by plating on LB Carb 75.

QUANTIFICATION AND STATISTICAL ANALYSIS

Raw data were recorded in Microsoft Excel and imported into GraphPad Prism for statistical analysis. Data were analyzed by Student’s t-test, One-way ANOVA, Two-way ANOVA, Mann-Whitney U-test, Log-rank, or Fisher’s Exact test, as indicated in figure legends. Asterisks indicate the statistical significance: *P<.05, **P< .01, ***P< .001,****P< .0001, ns = not significant. N values, definitions of center, and dispersion and precision measurements for each experiment are reported in the figure legends.

Key resources table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
NA	NA	NA
Bacterial and virus strains
For bacterial strains used in this study see Supplementary Table 2	NA	NA
Biological samples
NA	NA	NA
Chemicals, peptides, and recombinant proteins
β-lactamase	Abcam	Cat#ab67672
Bovine Serum Albumin	Sigma	Cat#A6003–25g
Lymphocyte Separation Medium	Cedarlane Laboratories	Cat#CL5031
Dulbecco’s Modified Eagle Medium	Gibco	Cat#11995–065
Fetal Bovine Serum	R&D Systems	Cat#S11150
UltraPure Water	Invitrogen	Cat#10977–023
TRIzol reagent	Ambion	Cat#15596018
Nitrocefin	Abcam	Cat#ab145625
Hydrogen peroxide	Fisher	Cat#H324–500
Kanamycin sulfate	Sigma	Cat#60615–5G
Carbenicillin	Fisher	Cat#BP2648–5
Chloramphenicol	Fisher	Cat#BP904–100
Streptomycin	Fisher	Cat#BP910–50
Tetrocycline hydrochloride	Alfa Aesar	Cat#B21408
NaOCl	Clorox	NA
Critical commercial assays
PureLink RNA kit	Invitrogen	Cat#12183018A
Turbo DNA-free kit	Invitrogen	Cat#AM1907
DNeasy Blood and Tissue kit	Qiagen	Cat#69506
GeneJET PCR Purification Kit	Thermo Scientific	Cat#K0702
GeneJET PCR Plasmid Miniprep Kit	Thermo Scientific	Cat#K0503
NEBuilder HiFi DNA Assembly Master Mix	New England Biolabs	Cat#M5520A
iSCRIPT cDNA Synthesis Kit	Bio-Rad	Cat#1708891
Catalase Assay Kit	Cayman Chemical	Cat#707002
iQ SYBR Green Supermix	Bio-Rad	Cat#1708882
Deposited data
Raw sequencing data generated from Tn-seq experiments	This paper	NCBI GEO: GSE198004
Raw sequencing data generated from WGS of Δlon and Δlon bfmR* strains	This paper	NCBI SRA BioProject: PRJNA813302
Experimental models: Cell lines
L929 cells	European Collection of Cell Cultures	Cat#85011425
Experimental models: Organisms/strains
Mouse: C57BL/6J	The Jackson Laboratory	RRID:IMSR_JAX:000664
Oligonucleotides
For oligonucleotides used in this study see Supplemental Table 3	NA	NA
Recombinant DNA
For plasmids used in this study see Supplemental Table 2	NA	NA
Software and algorithms
Prism 9	GraphPad	https://www.graphpad.com
Canvas X 16	Canvas GFX	https://www.canvasgfx.com
Breseq	Deatherage and Barrick, 2014	version 30.0
Tn-seq analysis http://galaxy.med.tufts.edu/	TUCF Genomics	NA
BioRender	BioRender (2020)	https://www.biorender.com
Other

Highlights.

• A. baumannii displays increased ability to colonize tissues following desiccation

• A Tn-seq screen reveals that mutations in Lon protease increase desiccation tolerance

• The predicted intrinsically disordered proteins DtpA and DtpB are upregulated in Δlon

• Recombinant DtpA protects proteins from desiccation and heat-inactivation