Rapid activation of dormant type IV pili enables a dispersal–infection tradeoff in environments with fluctuating nutrients

6 Bacteria in fluctuating environments must balance the high metabolic costs of motility 7 against risks of bacteriophage predation and immune clearance. While flagellar trade-off 8 mechanisms are well -documented, regulation of type IV pilus (T4P) activity during 9 environmental transitions remains unclear. We show that Pseudomonas aeruginosa uses 10 an energy-dependent idling strategy to synchronize T4P -mediated surface motility with 11 nutrient availability. In nutrient -depleted stationary phase, T4P transcription and protein 12 levels remain constant, pre-assembled machines persist at the cell pole, yet cells produce 13 only sparse, truncated pili that extend and retract slowly. Using a single -cell ATP 14 biosensor, we show that T4P dynamics respond directly to cellular adenylate energy 15 charge. Carbon source addition rapidly elevates intracellular ATP, reactivating pre-16 assembled T4P within minutes. This bypasses de novo protein synthesis, restoring pilus 17 number, length, and extension/retraction rates. This rapid response drives opportunistic 18 biofilm dispersal but , at the same time, creates an immediate tradeoff: reactivated T4P 19 restore susceptibility to pilus-specific phages upon nutrient upshift. Thus, energetic gating 20 of T4P enables P. aeruginosa to minimize exposure to phages during starvation while 21 remaining poised for rapid reactivation. Importantly, T4P promote resistance to 22 opsonization and phagocytosis by macrophages and neutrophils. Upon nutrient upshift, 23 full T4P activity therefore supports dispersal and host colonization while conferring 24 immune protection, revealing a fundamental dispersal –infection tradeoff at the host –25 microbe interface in fluctuating environments such as the lung and gut. 26 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted April 24, 2026. ; https://doi.org/10.64898/2026.04.23.720421doi: bioRxiv preprint

27 Type IV pili (T4P) are dynamic, retractable appendages that serve as key virulence factors 28 in numerous clinically important pathogens, including Pseudomonas , Neisseria, and 29 Vibrio species [ 1-5]. These extracellular filaments mediate critical functions such as 30 surface attachment, twitching motility, biofilm formation, and DNA uptake through 31 repeated cycles of extension and retraction powered by dedicated ATPases [6-10]. Upon 32 surface attachment, retraction of a surface bound pilus generates strong mechanical force 33 that pulls the cell forward in a process known as twitching motility [ 6-10]. The same 34 mechanical interactions also enable surface sensing that regulates cyclic AMP and cyclic 35 di-GMP (c-di-GMP) signaling pathways, thereby coupling physical contact to downstream 36 virulence programs [11-14]. 37 T4P play essential roles in biofilm development in vitro and during infection [ 15-17]. 38 Although mutants lacking pilus assembly or retraction show clear defects in biofilm 39 formation, the precise molecular mechanisms by which dynamic T4P cycles shape biofilm 40 architecture remain incompletely understood. Paradoxically, functional T4P also render 41 cells vulnerable: many bacteriophages exploit T4P as primary receptors, using pilus 42 retraction as an entry route into the cell [ 18-22]. In response, some phages encode 43 proteins that disrupt T4P function to block superinfection [23-26]. 44 Bacteria frequently deploy multiple specialized T4P systems tailored to distinct 45 environmental contexts. Vibrio cholerae, for example, expresses at least four distinct 46 systems: the toxin-coregulated pilus (TCP), competence pili (CP), the mannose-sensitive 47 hemagglutinin (MSHA) pilus, and the chitin-regulated pilus system (ChiRP), each 48 optimized for specific functions such as host colonization, natural transformation, or 49 surface attachment [1, 5]. In striking contrast, the opportunistic pathogen Pseudomonas 50 aeruginosa relies on a single, highly versatile T4P system to mediate multiple behaviors 51 simultaneously, making it an ideal model for dissecting how one machinery balances 52 tradeoffs between conflicting demands [27]. 53 The P. aeruginosa T4P machine spans the cell envelope and consists of a core assembly 54 complex (PilMNOPQ), an inner-membrane platform (PilC), and three motor ATPases: PilB 55 for pilus extension and PilT/PilU for retraction [1, 27-30]. The dynamic interaction of these 56 motors with the machine complex polymerizes and depolymerizes PilA subunits from the 57 inner membrane into the pilus fiber and controls the number and length of surface pili [4, 58 31-36]. 59 Bacterial habitats are inherently fluctuating. Nutrient availability shifts rapidly in liquid 60 environments due to flow and convection, in the host gut as digest transits, and on fomites 61 during transmission between surfaces [37-42]. While T4P gene expression is known to 62 respond to nutrient availability [37, 43-48] and growth-phase transitions [49-52], far less 63 is understood about how such fluctuations directly alter T4P dynamics (e.g., pilus number, 64 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted April 24, 2026. ; https://doi.org/10.64898/2026.04.23.720421doi: bioRxiv preprint length, and activity) and, in turn, influence T4P -dependent behaviors such as biofilm 65 formation or dispersal and phage susceptibility. This gap is particularly consequential 66 because T4P are central to the pathogenesis of many bacteria, yet most studies have 67 focused on transcriptional or second- messenger regulation rather than direct physical 68 constraints imposed by cellular energy status. 69 Here, we address how energy-limiting conditions during nutrient transitions reshape T4P 70 dynamics in P. aeruginosa and thereby modulate two key virulence traits: biofilm 71 formation/dispersal and susceptibility to pilus -dependent bacteriophages. Together, our 72 findings reveal how direct coupling of T4P activity to adenylate energy charge allows 73 bacteria to remain "cloaked" from phage exploitation during starvation while remaining 74 poised for opportunistic dispersal when conditions improve , highlighting a fundamental 75 tradeoff between motility, dispersal, and predator avoidance in fluctuating environments. 76 77

78 Pilus extension is rapidly activated upon back-dilution of stationary phase cells 79 To better understand how T4P dynamics change when cells transition between growth 80 phases, for example due to nutrient shifts, we thought to initially compare the activity of 81 T4P in stationary phase (SP) cells to exponential phase (EP) cells using Cysteine-82 maleimide based fluorescent labeling of the major pilin PilA to visualize pili [31, 32, 53]. 83 Consistent with prior reports, EP cells made several active pili on most of their poles [32] 84 (Figure 1A, Supplementary Movie 1) . In stark contrast, the activity of T4P in SP cells 85 appeared much reduced with most cells making no or only few short pili compared to EP 86 cells (Figure 1A, Supplementary Movie 2). To verify this observation, we quantified the 87 number of cells that extended at least one active pilus fiber in a 30 second time window 88 (Figure 1B). Interestingly, and in support of our observation, we found that only ~ 15 % of 89 SP cells make active pili. This ratio increased to almost 100 % within 30 minutes of 90 transitioning the cells to fresh LB medium and remained constant over a five -hour 91 incubation period simulating the transition from SP to EP (OD600 ~ 1.0). 92 93 T4P machines and motors are present in stationary phase cells 94 The observation that cells activate pilus extension in just 30 minutes after back-dilution of 95 SP cells into fresh growth medium made us wonder if SP cells have functional TFP 96 machines assembled or if transcription of T4P genes is required to be rapidly activated 97 after nutrient addition to generate machines that can extend functional pilus fibers. To test 98 this question, we first quantified the expression of several key T4P operons using single 99 cell fluorescent transcriptional reporters (Figure 1C-F). These reporters were made based 100 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted April 24, 2026. ; https://doi.org/10.64898/2026.04.23.720421doi: bioRxiv preprint on the established PaQa reporter that uses a constitutive rpoD::mKate2 as a control for 101 cell growth, metabolism, and fluorophore degradation, and flipped the promoter of the 102 PPaQa::yfp component in this reporter with the promoters for the major pilin pilA, the 103 extension motor pilB, and the core machine complex operon pilMNOPQ. To our surprise, 104 transcription of all three genes/operons remained approximately similar between SP cells, 105 and cells after one and two hours of back-dilution. To test if our reporter constructs worked 106 properly, we confirmed that the transcriptional activity of the pilA promoter is elevated in 107 a pilA background and diminished in a pilSR background, as expected [54] (Figure 1D). 108 Next, we tested if the T4P components are present in S P cells using fluorescent fusions 109 PilB::mScarlet-I3 and PilO::mCherry to image cells. In support that SP cells have T4P 110 assembly complexes present, we found bright fluorescence at the pole of SP and EP cells 111 for both fusion proteins ( Figure 1G). When we quantified changes in protein levels as a 112 function of cell growth by measuring the total brightness of individual cell poles, we did 113 not observe biologically significant changes between SP cells and the first two hours after 114 back-dilution into fresh medium ( Figure 1H-I). This supports our result that transcription 115 of the T4P genes does not change significantly during transition from SP to EP and 116 demonstrates that SP cells have T4P complexes assembled at the pole, but that these 117 complexes are not extending pilus fibers. 118 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted April 24, 2026. ; https://doi.org/10.64898/2026.04.23.720421doi: bioRxiv preprint 119 Figure 1. T4P machines are present in stationary phase cells and a re rapidly activated in fresh 120 medium. A) Time series of pilus dynamics in stationary phase (SP) and exponential phase (EP) cells. B) 121 The fraction of cells that made pili in a 30 second time window of three biological and four technical 122 replicates each. C-F) Transcriptional activity of the major T4P operons for the major pilin pilA (C,D), the 123 extension motor pilB (E), and the machine scaffold operon pilMNOPQ (E) as a function of growth time 124 (C,E,F) and mutant controls (D) for 50 cells of three biological replicates each. G) Images of protein fusions 125 to the extension motor pilB::mScarlet-I3 and the scaffold protein pilO::mCherry in stationary phase cells. 126 H,I) Quantification of total protein levels in individual cells for the fusions show in G for 50 cells of three 127 biological replicates each. Statistical analysis: Statistical significance was tested using a bootstrapped 128 (10,000 iterations) non-parametric Wilcoxon-Mann-Whitney two-sample rank test: (NS) not significant, P > 129 0.05; (*) P < 0.05; (**) P < 0.01; (***) P < 0.001; (****) P < 0.0001. Box plots: Boxes represent median and 130 interquartile range (25th–75th percentiles). 131 G) 100 80 60 40 20 0 543210.5SP Growth time (hr) Fraction of piliated cells (%) **** PpilMNOPQ::yfp / PrpoD::mKate 4 3 2 1 0 21SP Growth time (hr) PpilMNOPQ F) PpilA::yfp / PrpoD::mKate 4 3 2 1 0 21SP Growth time (hr) PpilA C) NS NS PpilB::yfp / PrpoD::mKate 4 3 2 1 0 21SP Growth time (hr) PpilB E) NS NS NS NS D) PpilA::yfp / PrpoD::mKate 12 6 4 2 0 8 10 **** *** PpilA control Stationary phaseExponential phase 0 sec 5 sec 10 sec 0 sec 5 sec 10 sec A) B) pilO::mCherry pilB::mScarlet 0.1 1 10 0.2 0.4 0.6 2 4 6 20 0.8 8 SP 1 2 Growth time (hr) PilB protein levels (a.u.) NS * 0.1 1 10 0.2 0.4 0.6 2 4 6 20 0.8 8 SP 1 2 Growth time (hr) PilO protein levels (a.u.) NS NSH) I) (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted April 24, 2026. ; https://doi.org/10.64898/2026.04.23.720421doi: bioRxiv preprint T4P become more active as cells recover from stationary phase 132 To understand how the T4P machines that are present in SP cells are rapidly activated to 133 extend pilus fibers, we sought to analyze the entire dynamic extension and retraction 134 cycles of individual pilus filaments in detail. When we compared the number of pilus fibers 135 that individual cells made in a 30 second time window, we found that the majority of active 136 cells made only one pilus, with few exceptions that made two or three pili (Figure 2A). 137 Pilus activity rapidly increased to an average of two to three pili per cell 30 minutes after 138 back-dilution and to a broad distribution of two to five pili after two hours ( Figure 2B-C). 139 Similarly, we saw the median length of pili increase from ~200 nm (SP cells) to ~900 nm 140 (two hours back-diluted) and the maximum length increase about 4-fold from less than 1 141 µm to more than 3 µm (Figure 2D). Interestingly, while the median and maximum lengths 142 increased gradually during the first two hours post back -dilution, they also decreased 143 gradually between two and five hours post back-dilution, but remained significantly above 144 SP levels (see Supplementary Figure 1). 145 146 Pili of stationary phase cells extend and retract much slower than exponential 147 phase cells 148 The observation that pilus length changes dramatically as cells transition between 149 different growth phases made us examine the rate of monomer incorporation into the 150 fiber, given by the extension rate of individual pili (Figure 2E). SP cells extended pili slowly 151 with a median of ~ 80 nm/s. This value steadily increased to ~ 200 nm/s at the two hour 152 mark. This behavior is consistent with our observations for the change in pilus fiber length 153 and suggests that pili are shorter in SP cells because they extend more slowly. 154 Interestingly, we observed a similar trend for the retraction rate of individual pili, that 155 increased from ~ 180 nm/s (SP) to 400 nm/s (2 hrs) (Figure 2F). Quantification of the 156 relative change in extension and retraction rates between SP and 0.5 hours and 0.5 hours 157 and 2 hours of growth shows that these rates both change about two- fold in the first 30 158 minutes and relatively little afterwards ( Figure 2G-H). Since these processes are 159 facilitated by ATPases (PilB and PilT), this suggests that SP cells might extend pili more 160 slowly due to nutrient limitations, specifically due to limitation of the cellular ATP pool. 161 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted April 24, 2026. ; https://doi.org/10.64898/2026.04.23.720421doi: bioRxiv preprint 162 Figure 2. Pilus extension and retraction rates are rapidly upshifted within 30 minutes in fresh 163 medium.. A-C) Histograms of how many pili individual cells extended in stationary phase (SP) cells (A), 164 and for cells incubated in fresh medium for 30 minutes (B) and 2 hours (C) for 50 cells of three biological 165 replicates each. D) The maximum length of all individual pili extended by 50 cells of three biological 166 replicates each. E,F) The extension rates (E) and retraction rates (F) of 50 individual pili of three biological 167 replicates each. G,H) Relative fold change of the extensions rate (G) and retraction rate (H) between 168 different growth times. Statistical analysis: Statistical significance was tested using a bootstrapped 169 (10,000 iterations) non-parametric Wilcoxon-Mann-Whitney two-sample rank test: (NS) not significant, P > 170 0.05; (*) P < 0.05; (**) P < 0.01; (***) P < 0.001; (****) P < 0.0001. Box plots: Boxes represent median and 171 interquartile range (25th–75th percentiles). 172 100 80 60 40 20 0 Fraction of cells (%) 6421 3 5 7 Number of pili per cell 100 80 60 40 20 0 Fraction of cells (%) 6421 3 5 7 Number of pili per cell 100 80 60 40 20 0 Fraction of cells (%) 6421 3 5 7 Number of pili per cell Maximum length of pili (nm) 100 1000 40 60 80 200 400 600 800 2000 4000 6000 2.00.5SP Growth time (hr) 2.00.5SP Growth time (hr) Retraction rate (nm/s) 40 60 80 100 200 400 600 800 1000 2000 **** NS **** Extension rate (nm/s) 40 60 80 100 200 400 600 800 1000 2000 2.00.5SP Growth time (hr) **** **** **** **** **** **** A) B) C) D) E) F) G) H) Stationary phase (SP) 0.5 hr incubation 2 hr incubation 2.0 1.5 1.0 0.5 0.0Extension rate fold change (hr-1) 2.0 1.5 1.0 0.5 0.0Retraction rate fold change (hr-1) (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted April 24, 2026. ; https://doi.org/10.64898/2026.04.23.720421doi: bioRxiv preprint Stationary phase cells are ATP limited and energy supplementation rapidly 173 activates dormant T4P machines 174 We sought to test this hypothesis that SP cells make fewer and shorter pili because they 175 are ATP limited by supplementing SP cells with different carbon sources directly added to 176 spent LB (overnight culture). In support of the ATP limitation hypothesis, supplementation 177 with either glucose, lactate, or succinate all increased the ratio of cells that made pili within 178 20 minutes to ~60% (Figure 3A). Consistent with the idea that the availability of these 179 carbon sources increased cellular ATP levels, individual deletions of the three importers 180 oprB, dctPQM, or lldD did not show a similar increase in the ration of T4P active cells. 181 Similarly, the median and maximum lengths of pili increased from SP levels (180 nm) to 182 levels similar of cells 30 min post back-dilution into fresh LB (~400 nm) in the same time 183 frame that was dependent on the presence of the specific carbon source importer (Figure 184 3B). Together this indicates that the availability and import of nutrients rapidly triggers the 185 activation of pilus extension. 186 To determine if the supplementation of nutrients indeed yields changes in cellular ATP 187 levels that could explain our pilus dynamics observation, we employed a fluorescent ATP 188 reporter to measure ATP levels in single cells [55] (Figure 3C). In EP cells, the brightness 189 of the ATP biosensor was homogenous between individual cells and cells mostly 190 appeared dim, indicative of high ATP levels. In contrast, in SP cells, we observed a large 191 heterogeneity between cells, with most cells expressing a bright biosensor (indicative of 192 low ATP levels) but some cells remained a dim sensor. This is indicative of different ATP 193 levels in a population of SP cells. We specifically quantified the ATP biosensor brightness 194 and compared their distributions between SP cells and cells grown for one and two hours 195 post back-dilution. (Figure 3D) These data confirm our initial assessment that SP cells 196 are energy depleted and cellular ATP levels are increased ~ 3-fold when transitioning to 197 EP. Importantly, a catalytically dead biosensor control did not change its brightness 198 (Figure 3E). Together, this suggests that SP cells make slowly extending pili because 199 they are ATP limited, and that supplementation of fresh nutrients activate pili because 200 ATP levels are refueled rapidly. 201 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted April 24, 2026. ; https://doi.org/10.64898/2026.04.23.720421doi: bioRxiv preprint 202 Figure 3. Carbon source supplementation activates T4P by increasing ATP production . A ,B) 203 Supplementation of stationary phase (SP) cells with different carbon sources increase the fraction of pilus 204 producing cells (A) and pilus length (B) within 20 minutes, similar to transitioning cells to fresh rich growth 205 medium (LB), for three biological replicates and 4 technical replicates each (A) or 50 pili for three biological 206 replicates each (B). C) Images of ATP biosensor activity in individual SP and exponential phase (EP) cells. 207 D,E) ATP biosensor activity in individual SP cells and cells grown for one hour in fresh LB medium (D) and 208 control of a catalytically dead sensor version (E). Statistical analysis: Statistical significance was tested 209 using a bootstrapped (10,000 iterations) non- parametric Wilcoxon- Mann-Whitney two- sample rank test: 210 (NS) not significant, P > 0.05; (*) P < 0.05; (**) P < 0.01; (***) P < 0.001; (****) P < 0.0001. Box plots: Boxes 211 represent median and interquartile range (25th–75th percentiles). 212 213 Rapid activation of dormant T4P activates biofilm dissemination and increases 214 phage susceptibility 215 Next, we sought to test if rapid activation of pilus dynamics in SP cells has a functional 216 consequence for the behavior of cells. Our findings indicate that T4P are dormant in 217 nutrient depleted environments and activated quickly when nutrients become available 218 EP SP ATP level (a.u.) high low Stationary phaseExponential phase C) Phase Contrast ATP biosensor 1hrSP 0.1 0.2 0.4 0.6 0.8 1 2 4 6 8 10 1hrSP 0.1 0.2 0.4 0.6 0.8 1 2 4 6 8 10ATP biosensor activity (a.u.) ATP biosensor activity (a.u.) ATP biosensor ControlD) E) **** NS lldD WT dctPQM WT oprB WT WT WT SP 0.5 hr Glucose Succinate LactateLB 20 min Strain: Carbon source: Growth: Fraction of piliated cells (%) A) lldD WT dctPQM WT oprB WT WT WT SP 0.5 hr Glucose Succinate LactateLB 20 min Strain: Carbon source: Growth: 60 80 100 200 400 600 8001000 2000 4000 Maximum length of pili (nm) **** **** **** **** ** ** ** NS NS NS NS* 120 100 80 60 40 20 0 *** *** *** *** ** ** ** NS NS NS NS*** B) (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted April 24, 2026. ; https://doi.org/10.64898/2026.04.23.720421doi: bioRxiv preprint again. This situation resembles liquid environments where fluid flow changes nutrient 219 availability continuously. Here, cells accumulate into biofilms when nutrients are poor and 220 disseminate to opportunistically explore the surroundings when conditions are favorable. 221 We wondered if the activation of T4P supports this behavior and analyzed the dispersal 222 of cells from biofilms after we added glucose to biofilms in SP. We first grew biofilms in a 223 standard biofilm assay on pegs submerged into growth medium in 96 well plate format 224 using WT cells and a pilus null mutant ( pilA) that is unable to assemble pilus filaments 225 [56]. After biofilms had formed and cells reached SP, we treated the peg-associated 226 biofilms either with glucose or a vehicle control (MOPS buffer), stained the residual biofilm 227 with crystal violet, and subsequently dissolved and quantified the dye in a plate reader by 228 OD600 measurement. Consistent with previous research, we observed that both WT and 229 pilA were able to form robust biofilms (untreated) (Figure 4A). After treatment of WT cells, 230 the amount of residual biofilm was significantly larger (~50%) in the vehicle control 231 compared to the g lucose treated environment , and the amount of biofilm dropped 232 significantly compared to the untreated condition in both cases (Figure 4B). In support of 233 our hypothesis that this difference is due to the activation of T4P, we found no significant 234 difference between both treatments in the pilA mutant. Comparing WT cells to pilA , we 235 did see a large decrease in residual biofilm with and without treatment, indicating that 236 even low levels of pilus activity (as in SP cells) significantly support dispersal of cells from 237 mature biofilms and that more active T4P increase dispersal. 238 This result that the activation of pili supports the dissemination of biofilms posed the 239 question if this creates a tradeoff with infection. As T4P become more dynamic and extend 240 longer filaments, they might also become more susceptible to bacteriophages. We tested 241 this idea by monitoring cell growth of SP cells after 30 minutes of glucose incubation (to 242 activate pili) that were then subjected to the pilus specific phage JBD68 for two minutes 243 (to infect cells). Both uninfected cells and the pilA mutant grew to SP at high OD ~ 1.2, 244 as expected (Figure 4C). Infected WT cells displayed reduced growth due to cell lysis in 245 a glucose-dependent manner. The no glucose control showed only a small reduction in 246 growth (due to low activity of T4P) while the g lucose treated cells had a substantial 247 reduction in growth by ~ 70% compared to the uninfected control ( Figure 4D). This 248 demonstrates that the increase in activity of T4P exposes cells to the risk of phage 249 infection (Figure 5). 250 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted April 24, 2026. ; https://doi.org/10.64898/2026.04.23.720421doi: bioRxiv preprint 251 Figure 4. Activation of dormant T4P by nutrient upshift triggers biofilm dispersal and increases 252 susceptibility to phage infection . A) Biofilm biomass before treatment (treatment: none) and after 253 treatment with (+) and without ( -) glucose in WT cells capable of extending pili and a pilA mutant control 254 unable to extend pili. B) Dispersal measured by the decrease in biofilm biomass relative to the no treatment 255 control. C) Growth of WT cells (un-)infected with T4P specific phage JBD68 with (+) and without (-) glucose 256 and a pilA control. D) Infectivity measured by the decrease in cell growth at the last time point relative to 257 the uninfected control. Statistical analysis: Statistical significance was tested using a bootstrapped 258 (10,000 iterations) non-parametric Wilcoxon-Mann-Whitney two-sample rank test: (NS) not significant, P > 259 0.05; (*) P < 0.05; (**) P < 0.01; (***) P < 0.001; (****) P < 0.0001. Box plots: Boxes represent median and 260 interquartile range (25th–75th percentiles). Error bars: represent the mean and standard deviation. 261 262

263 Bacteria use motility to opportunistically explore environments, for example to find nutrient 264 patches or to colonize new niches in fluctuating host environments like the gut or lung . 265 This behavior comes at the expense of substantial metabolic costs and increased risks 266 of immune clearance or phage infection. How bacteria strike this balance has been well 267 worked out for f lagella-based motility. Construction and operation of flagella consume 268 roughly 10% of a cell’s total energy budget in E. coli and up to ~40% in some taxa, [57]. 269 In nutrient-limited or starvation conditions, bacteria dynamically adjust motility investment 270 in proportion to anticipated chemotactic benefit, while marine copiotrophs display a clear 271 risk–reward dichotomy: some strains sustain costly swimming for days by converting 272 2.0 1.5 1.0 0.5 0.0 Biofilm (OD600) -+NoneTreatment: Strain: -+None WTWT WT pilApilA pilA NS NS NS** **** **** **** C) A) B) 1.4 1.2 1.0 0.8 0.6 0.4 0.2 0.0 Cell growth (OD600) Glucose (+) Glucose (-) Uninfected (+) Glucose (+) WT WT WT pilA 121086420 1197531 Time post infection (hr) 100 80 60 40 20 0 Dispersal (%) -+Treatment: Strain: -+ WT WT pilA pilA 100 80 60 40 20 0 -+Treatment: Strain: + WT WT pilA Infectivity (%) D) (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted April 24, 2026. ; https://doi.org/10.64898/2026.04.23.720421doi: bioRxiv preprint biomass into energy to forage for new patches (limokinetic), whereas others rapidly arrest 273 motility to conserve resources ( limostatic) [ 58, 59]. Active flagella also heighten 274 susceptibility to phages, as mutations conferring phage resistance frequently reduce 275 motility through regulatory pathways such as the Rcs phosphorelay [60]. 276 Here, we show that Pseudomonas aeruginosa employs an energy -dependent “idling” 277 strategy for type IV pili (T4P)-mediated surface motility that parallels these flagellar trade-278 offs while using a distinct mechanistic solution. In nutrient-depleted stationary phase, T4P 279 transcription and protein levels (PilA, PilB, PilO) remain largely constant and pre-280 assembled machines persist at the cell pole, yet cells produce only sparse, truncated pili 281 that extend and retract slowly (~80 nm/s extension rate). Supplementation of carbon 282 sources (glucose, lactate, or succinate) directly to spent medium rapidly elevates 283 intracellular ATP levels, as monitored by a fluorescent biosensor, and activates pilus 284 dynamics within minutes. This restores the fraction of active cells, increases pilus number 285 and length, and accelerates extension (~80 to ~200 nm/s) and retraction rates without the 286 need for de novo protein synthesis. Consequently, nutrient upshift triggers quick dispersal 287 of cells from mature biofilms, evidenced by significantly lower residual biofilm biomass in 288 WT cultures compared with pilA mutants after glucose addition, including greater baseline 289 dispersal even in vehicle controls. However, this rapid activation also heightens 290 susceptibility to pilus-specific phages like JBD68 in a glucose-dependent manner (~70% 291 growth reduction in activated WT cells), highlighting a tradeoff between motility and 292 infection (Figure 5). 293 This energy-gated idling of T4P may be particularly relevant in host environments such 294 as the lung, where nutrient availability fluctuates locally and intact pili help P. aeruginosa 295 resist opsonization and clearance by alveolar macrophages and neutrophils [61, 62]. By 296 maintaining low T4P activity under nutrient stress, cells could minimize unnecessary 297 surface exposure while poised for rapid reactivation that supports dispersal and 298 colonization when conditions improve (Figure 5). 299 This T4P idling strategy shares important functional parallels with flagellar motility. Both 300 systems couple dispersal benefit (biofilm escape or nutrient -finding) with an immediate 301 infection cost, while minimizing unnecessary energy expenditure during starvation. The 302 primary mechanistic difference lies in the mode of control: flagellar systems frequently 303 rely on transcriptional regulation, growth- rate sensing, or strain- specific behavioral 304 choices, whereas T4P activity here is closely linked to cellular energy status in a manner 305 consistent with direct energetic gating of the extension (PilB) and retraction (PilT) motors. 306 This low- latency, post -translational response may be particularly advantageous for 307 surface-associated lifestyles in competitive or flow-variable environments [38, 63, 64]. 308 Our biofilm dispersal data further suggest that even low -level T4P activity in stationary 309 phase contributes to reduced biofilm robustness in WT versus pilA cells. While extensive 310 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted April 24, 2026. ; https://doi.org/10.64898/2026.04.23.720421doi: bioRxiv preprint work has addressed T4P roles in early surface colonization and macroscopic biofilm 311 patterning (e.g., via cell arrangement and matrix interactions), how dynamic T4P 312 modulation shapes local architecture or mechanical properties within established biofilms 313 remains less explored [15, 65-70]. Recent studies show that P . aeruginosa biofilms exhibit 314 highly ordered cellular arrangements, including vertical striations of lengthwise- packed 315 cells across much of the biofilm depth, and that mutations altering surface structures 316 (including those affecting pili) can disrupt this microstructure, influencing local metabolic 317 activity and permeability. Whether dynamic, energy -gated changes in T4P activity fine-318 tune microscopic cohesion, cell packing, or mechanical integrity in mature biofilms 319 warrants further investigation. 320 Overall, these findings highlight a conserved ecological logic across motility systems: 321 costly appendages are kept in a poised but throttled state during nutrient stress to 322 minimize metabolic burden and infection risk, then rapidly deployed when conditions 323 improve. By linking T4P dynamics directly to adenylate energy charge, P . aeruginosa 324 achieves near -instantaneous biofilm escape without heavy biosynthetic costs. This 325 strategy could enable opportunistic pathogens to navigate the complex risk –reward 326 decisions required for persistence and dissemination at the host –microbe interface [71, 327 72]. Future experiments will test whether similar energetic idling occurs in other T4P - or 328 flagella-bearing species. This strategy may represent a widespread adaptation that allows 329 bacteria to protect the population under starvation while exploiting transient opportunities 330 for dispersal and colonization. 331 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted April 24, 2026. ; https://doi.org/10.64898/2026.04.23.720421doi: bioRxiv preprint 332 Figure 5. Schematic of how reactivation of dormant T4P upon nutrient upshift presents a tradeoff 333 between dispersal and predation by the host or bacteriophages in the environment. 334 Phage infection Low nutrient condition Sparse and truncated pili. High nutrient condition Many full-length pili. Phages Biofilm dispersal Biofilm Environment Host Prone to immune clearance Immune evasion Cloaked from predation but sessile Exploration but prone to predation Nutrient upshift Dormant T4P machines activated within minutes (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted April 24, 2026. ; https://doi.org/10.64898/2026.04.23.720421doi: bioRxiv preprint

335 We would like to thank Sayak Muckhopadhyay and Pushkar Lele for the recommendation 336 of the ATP biosensor and stimulating discussion and Josh Brehm and Joe Sorg for 337 suggesting and lending the Cerillo plate reader. We would like to thank the entire Koch 338 lab for stimulating discussion and the Department of Biology at Texas A&M for its 339 supportive environment. 340 This work was supported by grant R35GM155280 from the National Institute of Health 341 and startup funds from the College of Arts and Sciences, Division of Research, and 342 Department of Biology at Texas A&M University to M.D.K. 343 344 Author contributions 345 A.O.Y and M.D.K. designed research and wrote the manuscript. A.O.Y and Z.K.M 346 provided reagents, genetic constructs, and performed experiments. A.O.Y , Z.K.M., and 347 M.D.K. analyzed data. 348 349 Competing Interest Statement 350 The authors declare no competing finical interests. 351 352

and Protocols 353 Strains and Growth Conditions. Information on cloning, plasmids, and primers used in 354 this study as well as detailed methods on individual assays can be found in the 355 Supplementary Information (SI) Methods and Protocols. 356 P. aeruginosa PAO1 was grown in liquid lysogeny broth (LB) Miller (Difco) and EZ rich 357 defined medium (Teknova) in a floor shaker, on LB Miller agar (1.5 % Bacto Agar), on 358 Vogel-Bonner minimal medium (VBMM) agar (200 mg/l MgSO4 7H2O, 2 g/l citric acid, 10 359 g/l K2HPO4, 3.5 g/l NaNH4HPO4 4 H2O, and 1.5% agar), and on no-salt LB (NSLB) agar 360 (10 g/l tryptone, 5 g/l yeast extract, and 1.5% agar) at 30 °C (for cloning) or at 37 °C. 361 Escherichia coli S17 was grown in liquid LB Miller (Difco) in a floor shaker and on LB 362 Miller agar (1.5 % Bacto Agar) at 30 °C (for cloning, SI Appendix) or at 37 °C. Antibiotics 363 were used at the following concentrations: 200 μg/mL carbenicillin in liquid (300 μ g/mL 364 on plates) or 10 μg/mL gentamycin in liquid (30 μ g/mL on plates) in liquid for P. 365 aeruginosa, and 100 μg/mL carbenicillin in liquid (100 μ g/mL on plates) or 30 μ g/mL 366 gentamycin in liquid (30 μg/mL on plates) for E. coli. 367 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted April 24, 2026. ; https://doi.org/10.64898/2026.04.23.720421doi: bioRxiv preprint

368 1. Ellison, C.K., G.B. Whitfield, and Y . V . Brun, Type IV Pili: dynamic bacterial 369 nanomachines. FEMS Microbiol Rev, 2022. 46(2). 370 2. Hoppe, J., et al., PilY1 Promotes Legionella pneumophila Infection of Human Lung 371 Tissue Explants and Contributes to Bacterial Adhesion, Host Cell Invasion, and 372 Twitching Motility. Front Cell Infect Microbiol, 2017. 7: p. 63. 373 3. Floyd, K.A., et al., c-di-GMP modulates type IV MSHA pilus retraction and surface 374 attachment in Vibrio cholerae. Nature Communications, 2020. 11(1): p. 1549. 375 4. Ellison, C.K., et al., A bifunctional ATPase drives tad pilus extension and retraction. 376 Science Advances, 2019. 5(12): p. eaay2591. 377 5. Ellison, C.K., et al., Subcellular localization of type IV pili regulates bacterial 378 multicellular development. Nature Communications, 2022. 13(1): p. 6334. 379 6. Henrichsen, J., Twitching motility. Annu Rev Microbiol, 1983. 37: p. 81-93. 380 7. Siryaporn, A., et al., Colonization, competition, and dispersal of pathogens in fluid 381 flow networks. Current Biology, 2015. 25(9): p. 1201-1207. 382 8. Shen, Y ., et al., Flow directs surface-attached bacteria to twitch upstream. Biophys J, 383 2012. 103(1): p. 146-51. 384 9. Oliveira, N.M., K.R. Foster, and W.M. Durham, Single-cell twitching chemotaxis in 385 developing biofilms. Proceedings of the National Academy of Sciences, 2016. 386 113(23): p. 6532-6537. 387 10. Jin, F ., et al., Bacteria use type-IV pili to slingshot on surfaces. Proceedings of the 388 National Academy of Sciences, 2011. 108(31): p. 12617-12622. 389 11. Koch, M.D., et al., Pseudomonas aeruginosa distinguishes surfaces by 390 stiffness using retraction of type IV pili. Proceedings of the National Academy of 391 Sciences, 2022. 119(20): p. e2119434119. 392 12. Luo, Y ., et al., A Hierarchical Cascade of Second Messengers Regulates 393 Pseudomonas aeruginosa Surface Behaviors. mBio, 2015. 6(1): p. 394 10.1128/mbio.02456-14. 395 13. Persat, A., et al., Type IV pili mechanochemically regulate virulence factors in 396 Pseudomonas aeruginosa. Proc Natl Acad Sci U S A, 2015. 112(24): p. 7563-8. 397 14. Siryaporn, A., et al., Surface attachment induces Pseudomonas aeruginosa 398 virulence. Proceedings of the National Academy of Sciences, 2014. 111(47): p. 399 16860-16865. 400 15. Klausen, M., et al., Biofilm formation by Pseudomonas aeruginosa wild type, flagella 401 and type IV pili mutants. Mol Microbiol, 2003. 48(6): p. 1511-24. 402 16. Haagensen, J.A.J., et al., Differentiation and Distribution of Colistin - and Sodium 403 Dodecyl Sulfate-Tolerant Cells in Pseudomonas aeruginosa Biofilms. Journal 404 of Bacteriology, 2007. 189(1): p. 28-37. 405 17. Chiang, P . and L.L. Burrows, Biofilm formation by hyperpiliated mutants of 406 Pseudomonas aeruginosa. J Bacteriol, 2003. 185(7): p. 2374-8. 407 18. Meng, R., et al., Structural basis of Acinetobacter type IV pili targeting by an RNA virus. 408 Nature Communications, 2024. 15(1): p. 2746. 409 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted April 24, 2026. ; https://doi.org/10.64898/2026.04.23.720421doi: bioRxiv preprint 19. Wang, Y ., et al., Structural mechanisms of Tad pilus assembly and its interaction with 410 an RNA virus. Science Advances, 2024. 10(18): p. eadl4450. 411 20. Harvey, H., et al., Pseudomonas aeruginosa defends against phages through type IV 412 pilus glycosylation. Nature Microbiology, 2018. 3(1): p. 47-52. 413 21. Bae, H. -W., et al., Pilin regions that select for the small RNA phages in 414 Pseudomonas aeruginosa type IV pilus. Journal of Virology, 2025. 99(4): p. 415 e01949-24. 416 22. Chung, I.-Y ., et al., A phage protein that inhibits the bacterial ATPase required for type 417 IV pilus assembly. Proceedings of the National Academy of Sciences, 2014. 111(31): 418 p. 11503-11508. 419 23. Hendrix, H., et al., PlzR regulates type IV pili assembly in Pseudomonas aeruginosa 420 via PilZ binding. Nature Communications, 2024. 15(1): p. 8717. 421 24. Taylor, V .L., et al., Prophages block cell surface receptors to preserve their viral 422 progeny. Nature, 2025. 644(8078): p. 1049-1057. 423 25. Schmidt, A.K., et al., A Filamentous Bacteriophage Protein Inhibits Type IV Pili To 424 Prevent Superinfection of Pseudomonas aeruginosa. mBio, 2022. 13(1): p. e02441 -425 21. 426 26. Thongchol, J., et al., Removal of Pseudomonas type IV pili by a small RNA virus. 427 Science, 2024. 384(6691): p. eadl0635. 428 27. Burrows, L.L., Pseudomonas aeruginosa twitching motility: type IV pili in action. Annu 429 Rev Microbiol, 2012. 66: p. 493-520. 430 28. Roberge, N.A. and L.L. Burrows, Building permits— control of type IV pilus assembly 431 by PilB and its cofactors. Journal of Bacteriology, 2024. 206(12): p. e00359-24. 432 29. Guo, S., et al., PilY1 regulates the dynamic architecture of the type IV pilus machine 433 in Pseudomonas aeruginosa. Nature Communications, 2024. 15(1): p. 9382. 434 30. Chang, Y .-W., et al., Architecture of the Type IVA Pilus Machine. Biophysical Journal, 435 2016. 110(3): p. 468a-469a. 436 31. Ellison, C.K., et al., Real-time microscopy and physical perturbation of bacterial pili 437 using maleimide-conjugated molecules. Nature Protocols, 2019. 14(6): p. 1803-1819. 438 32. Koch, M.D., et al., Competitive binding of independent extension and retraction 439 motors explains the quantitative dynamics of type IV pili. Proceedings of the National 440 Academy of Sciences, 2021. 118(8): p. e2014926118. 441 33. Talà, L., et al., Pseudomonas aeruginosa orchestrates twitching motility by sequential 442 control of type IV pili movements. Nature Microbiology, 2019. 4(5): p. 774-780. 443 34. Kühn, M.J., et al., Two antagonistic response regulators control Pseudomonas 444 aeruginosa polarization during mechanotaxis. Embo j, 2023. 42(7): p. e112165. 445 35. Chlebek, J.L., et al., PilT and PilU are homohexameric ATPases that coordinate to 446 retract type IVa pili. PLOS Genetics, 2019. 15(10): p. e1008448. 447 36. Teipen, A.E., et al., Structural modeling reveals the mechanism of motor ATPase 448 coordination during type IV pilus retraction. bioRxiv, 2025: p. 2025.10.30.685630. 449 37. Padron, G.C., et al., Shear flow promotes bacterial growth and shapes spatial 450 gradients by rapidly replenishing scarce nutrients. mBio. 0(0): p. e03446-25. 451 38. Shuppara, A.M., et al., Shear flow patterns antimicrobial gradients across bacterial 452 populations. Science Advances, 2025. 11(11): p. eads5005. 453 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted April 24, 2026. ; https://doi.org/10.64898/2026.04.23.720421doi: bioRxiv preprint 39. Gilbert, J.A., et al., Current understanding of the human microbiome. Nature 454 Medicine, 2018. 24(4): p. 392-400. 455 40. Kuziel, G.A. and S. Rakoff -Nahoum, The gut microbiome. Curr Biol, 2022. 32(6): p. 456 R257-r264. 457 41. Stephens, B., et al., Microbial Exchange via Fomites and Implications for Human 458 Health. Curr Pollut Rep, 2019. 5(4): p. 198-213. 459 42. Lopez, G.U., et al., Transfer efficiency of bacteria and viruses from porous and 460 nonporous fomites to fingers under different relative humidity conditions. Appl 461 Environ Microbiol, 2013. 79(18): p. 5728-34. 462 43. Krebs, S.J. and R.K. Taylor, Nutrient-dependent, rapid transition of Vibrio cholerae to 463 coccoid morphology and expression of the toxin co -regulated pilus in this form. 464 Microbiology, 2011. 157(10): p. 2942-2953. 465 44. Ni, L., et al., Bacteria differently deploy type-IV pili on surfaces to adapt to nutrient 466 availability. npj Biofilms and Microbiomes, 2016. 2(1): p. 15029. 467 45. Edén, C.S. and H.A. Hansson, Escherichia coli pili as possible mediators of 468 attachment to human urinary tract epithelial cells. Infection and Immunity, 1978. 469 21(1): p. 229-237. 470 46. Mokrzan, E.M., T.J. Johnson, and L.O. Bakaletz, Expression of the Nontypeable 471 Haemophilus influenzae Type IV Pilus Is Stimulated by Coculture with Host 472 Respiratory Tract Epithelial Cells. Infection and Immunity, 2019. 87(12): p. 473 10.1128/iai.00704-19. 474 47. Chirakadavil, J.B., et al., Calcium signals natural transformation in 475 Acinetobacter baumannii. bioRxiv, 2026: p. 2026.02.23.707608. 476 48. Denise, R., S.S. Abby, and E.P .C. Rocha, Diversification of the type IV filament 477 superfamily into machines for adhesion, protein secretion, DNA uptake, and motility. 478 PLoS Biol, 2019. 17(7): p. e3000390. 479 49. Bradley, D.E., A study of pili on Pseudomonas aeruginosa. Genetical Research, 1972. 480 19(1): p. 39-51. 481 50. Vesel, N. and M. Blokesch, Pilus Production in Acinetobacter baumannii Is Growth 482 Phase Dependent and Essential for Natural Transformation. Journal of Bacteriology, 483 2021. 203(8): p. 10.1128/jb.00034-21. 484 51. Song, X.-M., et al., The growth phase-dependent regulation of the pilus locus genes 485 by two -component system TCS08 in Streptococcus pneumoniae. Microbial 486 Pathogenesis, 2009. 46(1): p. 28-35. 487 52. Salzer, R., et al., Environmental factors affecting the expression of type IV pilus genes 488 as well as piliation of Thermus thermophilus. FEMS Microbiology Letters, 2014. 489 357(1): p. 56-62. 490 53. Ellison, C.K., et al., Obstruction of pilus retraction stimulates bacterial surface 491 sensing. Science, 2017. 358(6362): p. 535-538. 492 54. Kilmury, S.L. and L.L. Burrows, Type IV pilins regulate their own expression via direct 493 intramembrane interactions with the sensor kinase PilS. Proceedings of the National 494 Academy of Sciences, 2016. 113(21): p. 6017-6022. 495 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted April 24, 2026. ; https://doi.org/10.64898/2026.04.23.720421doi: bioRxiv preprint 55. Yaginuma, H., et al., Diversity in ATP concentrations in a single bacterial cell 496 population revealed by quantitative single -cell imaging. Scientific Reports, 2014. 497 4(1): p. 6522. 498 56. Marko, V .A., et al., Pseudomonas aeruginosa type IV minor pilins and PilY1 regulate 499 virulence by modulating FimS-AlgR activity. PLoS Pathog, 2018. 14(5): p. e1007074. 500 57. Schavemaker, P .E. and M. Lynch, Flagellar energy costs across the tree of life. elife, 501 2022. 11: p. e77266. 502 58. Ni, B., et al., Growth-rate dependent resource investment in bacterial motile behavior 503 quantitatively follows potential benefit of chemotaxis. Proceedings of the National 504 Academy of Sciences, 2020. 117(1): p. 595-601. 505 59. Keegstra, J.M., et al., Risk–reward trade- off during carbon starvation generates 506 dichotomy in motility endurance among marine bacteria. Nature Microbiology, 2025. 507 10(6): p. 1393-1403. 508 60. Burmeister, A.R., H. Tewatia, and C. Skinner, A tradeoff between bacteriophage 509 resistance and bacterial motility is mediated by the Rcs phosphorelay in Escherichia 510 coli. Microbiology, 2024. 170(8): p. 001491. 511 61. Tan, R.M., et al., Type iv pilus of pseudomonasaeruginosa confers resistance to 512 antimicrobial activities of the pulmonary surfactant protein-a. Journal of innate 513 immunity, 2014. 6(2): p. 227-239. 514 62. Tan, R.M., et al., Type IV pilus glycosylation mediates resistance of Pseudomonas 515 aeruginosa to opsonic activities of the pulmonary surfactant protein A. Infection and 516 immunity, 2015. 83(4): p. 1339-1346. 517 63. Padron, G.C., et al., Bacteria in Fluid Flow. Journal of bacteriology, 2023. 205(4): p. 518 e00400-22. 519 64. Padron, G.C., et al., Shear flow promotes bacterial growth and shapes spatial 520 gradients by rapidly replenishing scarce nutrients. mBio, 2026: p. e03446-25. 521 65. Dayton, H., et al., Cellular arrangement impacts metabolic activity and antibiotic 522 tolerance in Pseudomonas aeruginosa biofilms. Plos Biology, 2024. 22(2): p. 523 e3002205. 524 66. Gomez, S., et al., Substrate stiffness impacts early biofilm formation by modulating 525 Pseudomonas aeruginosa twitching motility. Elife, 2023. 12: p. e81112. 526 67. Nijjer, J., et al., Mechanical forces drive a reorientation cascade leading to biofilm self-527 patterning. Nature communications, 2021. 12(1): p. 6632. 528 68. Huang, X., et al., Vibrio cholerae biofilms use modular adhesins with glycan-targeting 529 and nonspecific surface binding domains for colonization. Nature communications, 530 2023. 14(1): p. 2104. 531 69. Hartmann, R., et al., Emergence of three-dimensional order and structure in growing 532 biofilms. Nature physics, 2019. 15(3): p. 251-256. 533 70. Vidakovic, L., et al., Biofilm formation on human immune cells is a multicellular 534 predation strategy of Vibrio cholerae. Cell, 2023. 186(12): p. 2690-2704. e20. 535 71. Sahu, A. and R. Ruhal, Immune system dynamics in response to Pseudomonas 536 aeruginosa biofilms. npj Biofilms and Microbiomes, 2025. 11(1): p. 104. 537 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted April 24, 2026. ; https://doi.org/10.64898/2026.04.23.720421doi: bioRxiv preprint 72. Swart, A.L., et al., Pseudomonas aeruginosa breaches respiratory epithelia through 538 goblet cell invasion in a microtissue model. Nature Microbiology, 2024. 9(7): p. 1725-539 1737. 540 541 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted April 24, 2026. ; https://doi.org/10.64898/2026.04.23.720421doi: bioRxiv preprint