An orphan cbb3-type cytochrome oxidase subunit supports Pseudomonas aeruginosa biofilm growth and virulence

Hypoxia is a common challenge faced by bacteria during associations with hosts due in part to the formation of densely packed communities (biofilms). cbb3-type cytochrome c oxidases, which catalyze the terminal step in respiration and have a high affinity for oxygen, have been linked to bacterial pathogenesis. The pseudomonads are unusual in that they often contain multiple full and partial (i.e. ‘orphan’) operons for cbb3-type oxidases and oxidase subunits. Here, we describe a unique role for the orphan catalytic subunit CcoN4 in colony biofilm development and respiration in the opportunistic pathogen Pseudomonas aeruginosa PA14. We also show that CcoN4 contributes to the reduction of phenazines, antibiotics that support redox balancing for cells in biofilms, and to virulence in a Caenorhabditis elegans model of infection. These results highlight the relevance of the colony biofilm model to pathogenicity and underscore the potential of cbb3-type oxidases as therapeutic targets.


INTRODUCTION 14 15
Among the oxidants available for biological reduction, molecular oxygen (O 2 ) provides the 16 highest free energy yield. Since the accumulation of O 2 in the atmosphere between ~2.4-0.54 17 billion years ago (Kirschvink & Kopp 2008;Dietrich, Tice, et al. 2006), organisms that can use it 18 for growth and survival, and tolerate its harmful byproducts, have evolved to exploit this energy 19 and increased in complexity (Knoll & Sperling 2014;Falkowski 2006). At small scales and in 20 crowded environments, rapid consumption of O 2 leads to competition for this resource and has 21 promoted diversification of bacterial and archaeal mechanisms for O 2 reduction that has not The opportunistic pathogen Pseudomonas aeruginosa, a colonizer of both plant and animal hosts 29 (Rahme et al. 1995), has a branched respiratory chain with the potential to reduce O 2 to water 30 using at least five different terminal oxidase complexes: two quinol oxidases (bo 3 and a bd-type 31 cyanide insensitive oxidase) and three cytochrome c oxidases (aa 3 , cbb 3 -1, and cbb 3 -2) ( Figure  32 9 these assays, one strain was labeled with constitutively-expressed YFP so that the strains could 199 be distinguished during enumeration of colony forming units (CFUs). Experiments were 200 performed with the label on each strain to confirm that YFP expression did not affect fitness 201 (Figure 3-figure supplement 1A, B). When competitive fitness was assessed after three days 202 of colony growth (Figure 3A), ∆N4 cells showed a disadvantage, with the wild type 203 outcompeting ∆N4 by a factor of two. This was similar to the disadvantage observed for the 204 ∆N1∆N2 mutant, further suggesting that the orphan subunit CcoN4 plays a significant role in 205 biofilm metabolism. Remarkably, deletion of ccoN4 in mutants already lacking ccoN1 and ccoN2 206 led to a drastic decrease in fitness, with the wild type outcompeting ∆N1∆N2∆N4 by a factor of 207 16. This disadvantage was comparable to that observed for the mutant lacking the full cco 208 operons (∆cco1cco2), underscoring the importance of CcoN4-containing isoforms during biofilm 209 growth. 210

211
To further explore the temporal dynamics of N subunit utilization, we repeated the competition 212 assay, but sampled each day over the course of three days ( Figure 3B). The fitness disadvantage 213 that we had found for strains lacking CcoN1 and CcoN2 was evident after only one day of 214 growth and did not significantly change after that. In contrast, the ∆N4-specific decline in fitness 215 did not occur before the second day. These data suggest that the contributions of the various N 216 subunits to biofilm metabolism differ depending on developmental stage. 217 218 DIC imaging of thin sections from wild-type colonies reveals morphological variation over depth 219 that may result from decreasing O 2 availability (Figure 3-figure supplement 1C). We have 220 previously reported that three-day-old PA14 colony biofilms are hypoxic at depth (Dietrich et al. 221 2013) and that O 2 availability is generally higher in thinner biofilms, such as those formed by a 222 phenazine-null mutant (∆phz). We have proposed that the utilization of phenazines as electron 223 acceptors in wild-type biofilms enables cellular survival in the hypoxic zone and promotes 224 colony growth (Okegbe et al. 2014). The relatively late-onset phenotype of the ∆N4 mutant in 225 the competition assay suggested to us that CcoN4 may play a role in survival during formation of 226 the hypoxic colony subzone and that this zone could arise at a point between one and two days of 227 colony growth. We measured O 2 concentrations in wild-type and ∆phz biofilms at specific time 228 points over development, and found that O 2 declined similarly with depth in both strains ( Figure  229  3D). The rate of increase in height of ∆phz tapered off when a hypoxic zone began to form, 230 consistent with our model that the base does not increase in thickness when electron acceptors 231 (O 2 or phenazines) are not available. Although we cannot pinpoint the exact depth at which the 232 O 2 microsensor leaves the colony base and enters the underlying agar, we can estimate these 233 values based on colony thickness measurements ( Figure 3C). When we measured the thickness 234 of wild-type and ∆phz biofilms over three days of incubation, we found that the values began to 235 diverge between 30 and 48 hours of growth, after the colonies reached ~70 µm in height, which 236 coincides with the depth at which O 2 becomes undetectable. ∆phz colonies reached a maximum 237 thickness of ~80 µm, while wild-type colonies continued to grow to ~150 µm ( Figure 3C). In 238 this context, it is interesting to note that the point of divergence for the increase in wild-type and 239 ∆phz colony thickness corresponds to the point at which CcoN4 becomes important for cell 240 viability in our mixed-strain colony growth experiments ( Figure 3B) Furthermore, recent studies, along with our results, suggest that even within the Cco terminal 250 oxidase complexes, the various N subunits may perform different functions (Hirai et al. 2016). 251 We sought to determine whether differential regulation of cco genes could lead to uneven 252 expression across biofilm subzones. To test this, we engineered reporter strains in which GFP 253 expression is regulated by the cco1, cco2, or ccoN4Q4 promoters. Biofilms of these strains were 254 grown for three days, thin-sectioned, and imaged by fluorescence microscopy. Representative 255 results are shown in the left panel of  Our results indicate that different Cco isoforms may function in specific biofilm subzones, but 277 that CcoN4-containing isoforms could potentially form throughout the biofilm. These data, 278 together with our observation that ∆N4 biofilms exhibit a fitness disadvantage from day two 279 The results shown in Figure 2B implicate CcoN4-containing isoforms in the reduction of TTC, a 286 small molecule that interacts with the respiratory chain (Rich et al. 2001). Similar activities have 287 been demonstrated for phenazines, including the synthetic compound phenazine methosulfate 288 (Nachlas et al. 1960) and those produced naturally by P. aeruginosa (Armstrong & Stewart-Tull 289 1971). Given that CcoN4 and phenazines function to influence morphogenesis at similar stages 290 of biofilm growth (Figures 2A, 3, Figure 2-figure supplement 1, Figure 3- figure  291 supplement 1A, B), we wondered whether the role of CcoN4 in biofilm development was linked 292 to phenazine metabolism. We used a Unisense platinum microelectrode with a 20-30 µm tip to 293 measure the extracellular redox potential in biofilms as a function of depth. This electrode 294 measures the inclination of the sample to donate or accept electrons relative to a Ag/AgCl 295 reference electrode. We found that wild-type colonies showed a decrease in redox potential over 296 depth, indicating an increased ratio of reduced to oxidized phenazines, while the redox potential 297 of ∆phz colonies remained unchanged ( Figure 5A). To confirm that phenazines are the primary 298 determinant of the measured redox potential in the wild type, we grew ∆phz colonies on medium 299 containing phenazine methosulfate (a synthetic compound that resembles the natural phenazines 300 that regulate P. aeruginosa colony morphogenesis (Sakhtah et al. 2016)), and found that these 301 colonies yielded redox profiles similar to those of the wild type (Figure 5-figure supplement 302 1A). Therefore, though the microelectrode we employed is capable of interacting with many 303 redox-active substrates, we found that its signal was primarily determined by phenazines in our 304 system. In addition, while wild-type colonies showed rapid decreases in O 2 availability starting 305 at the surface, the strongest decrease in redox potential was detected after ~50 µm (Figure 5A). 306 These results suggest that the bacteria residing in the biofilm differentially utilize O 2 and 307 phenazines depending on their position and that O 2 is the preferred electron acceptor. 308

309
We hypothesized that one or more of the CcoN subunits encoded by the PA14 genome is 310 required for phenazine reduction and tested this by measuring the redox potential over depth for 311 a series of cco mutants (Figure 5B, top). We saw very little reduction of phenazines in the 312 ∆cco1cco2 colony, suggesting that cbb 3 -type oxidases are required for this activity. In contrast, 313 the mutant lacking the catalytic subunits of Cco1 and Cco2, ∆N1∆N2, showed a redox profile 314 similar to the wild type, indicating that isoforms containing one or both of the orphan CcoN 315 subunits could support phenazine reduction activity. Indeed, although redox profiles obtained for 316 the ∆N1∆N2 and ∆N4 mutants were similar to those obtained for the wild type, the redox profile 317 of the ∆N1∆N2∆N4 mutant recapitulated that of ∆cco1cco2. These results indicate redundancy in 318 the roles of some of the CcoN subunits. Consistent with this, ∆N1∆N4 showed an intermediate 319 defect in phenazine reduction. Extraction and measurement of phenazines released from wild-320 type and cco mutant biofilms showed that variations in redox profiles could not be attributed to 321 differences in phenazine production ( Figure 5-figure supplement 1). 322 Our group has previously shown that a ∆phz mutant compensates for its lack of phenazines by 324 forming thinner colonies, thus limiting the development of the hypoxic subzone seen in the wild 325 type (Dietrich et al. 2013). We therefore hypothesized that mutants unable to reduce phenazines 326 would likewise result in thinner colonies. Indeed, we observed that the cco mutants that lacked 327 phenazine reduction profiles in the top panel of Figure 5B produced biofilms that were 328 significantly thinner than wild-type and comparable to that of the ∆phz mutant ( Figure 5B, which may be linked to defects in phenazine utilization ( Figure 5B). To further examine the 341 relationships between Cco isoforms and redox imbalance in biofilms, we prepared thin sections 342 from two day-old colonies and stained with fluorescein-labeled lectin, which binds preferentially 343 to the Pel polysaccharide component of the matrix (Jennings et al. 2015). Consistent with their 344 similar gross morphologies, the wild-type and ∆N1∆N2 biofilms showed similar patterns of 345 staining, with a faint band of higher intensity at a depth of ~40 µm ( Figure 5C). ∆N4 also 346 showed a similar pattern, with a slightly higher intensity of staining in this band. ∆N1∆N2∆N4 347 and ∆cco1cco2 showed more staining throughout each sample, with wider bands of greater 348 intensity at the ~40 µm point. These data suggest that deletion of the Cco complexes leads to a 349 more reduced biofilm, which induces production of more matrix, and that CcoN4 contributes 350 significantly to maintaining redox homeostasis when O 2 is limiting. type biofilm architecture and respiration (Figures 2A, 2C, and 5C), we hypothesized that it 361 could also contribute to virulence. To test this, we conducted virulence assays using the 362 nematode Caenorhabditis elegans as a host. It has been shown that P. aeruginosa is pathogenic 363 to C. elegans and that the slow killing assay mimics an infection-like killing of C. elegans by the 364 bacterium (Tan et al. 1999). While ∆N1∆N2 killed with wild type-like kinetics, ∆N1∆N2∆N4 and 365 ∆cco1cco2 both showed comparably-impaired killing relative to wild-type PA14 (Figure 6).  In well-mixed liquid cultures, mutants lacking the "orphan" subunits do not show growth defects 385 ( Figure 2C) (Hirai et al. 2016). We were therefore surprised to find that the ∆N4 mutant showed 386 a unique morphotype in a colony biofilm assay (Figure 2A). We have applied this assay 387 extensively in our studies of the mechanisms underlying cellular redox balancing and sensing 388 and noted that the phenotype of ∆N4 was similar to that of mutants with defects in electron 389 shuttling and redox signaling (Dietrich et al. 2013;Okegbe et al. 2017). 390

391
We characterized the effects of a ∆N4 mutation on biofilm physiology through a series of assays. 392 In well-mixed liquid cultures, ∆cco1cco2 shows a growth phenotype similar to that of ∆N1∆N2, 393 suggesting that the subunits of the Cco1 and Cco2 oxidases do not form heterocomplexes with 394 CcoN4 or that such complexes do not contribute to growth under these conditions. Consistent 395 with this, deleting ccoN4 in the ∆N1∆N2 background has no effect on growth. However, in 396 biofilm-based experiments, we found that deleting N4 alone was sufficient to cause an altered 397 morphology phenotype (Figure 2A), and that deleting N4 in either a ∆N1 or a ∆N1∆N2 398 background profoundly affected biofilm physiology. These experiments included quantification 399 of respiratory activity in colonies, in which deletion of CcoN4 led to a significant decrease 400  showed that CcoN4 contributes to the repression of Pel polysaccharide production ( Figure 5C). 405 The overlap in zones of expression between cco1, cco2, and ccoN4Q4 seen in colony thin 406 sections (Figure 4) implies that CcoN4 could form heterocomplexes with Cco1 and Cco2 407 subunits that span the depth of the colony and function to influence the physiology of P. 408 aeruginosa biofilms in these ways. 409

410
Our results suggest that CcoN4 supports O 2 and/or phenazine reduction specifically in biofilms. 411 Using a strain engineered to produce GFP under control of the promoter for ccoN4Q4, we found 412 that this locus is expressed throughout the biofilm depth, suggesting that CcoN4-containing 413 isoforms could contribute to cytochrome c oxidation in both oxic and hypoxic zones (Figure 4). 414 This constitutes a deviation from the previously published observation that these genes are 415 specifically induced in hypoxic liquid cultures when compared to well-aerated ones (Alvarez-416 Ortega & Harwood 2007). We conclude that ccoN4Q4 is uniquely induced by the conditions in 417 the upper portion of the biofilm, where O 2 is available as an electron acceptor. This regulation 418 may contribute to the biofilm-specific role of the CcoN4 subunit. availability. Our results suggest that, in the colony biofilm system, enzyme complexes 431 traditionally considered to be specific to oxygen reduction may contribute to anaerobic survival. 432

433
Because biofilm formation is often associated with colonization of and persistence in hosts, we 434 tested whether ccoN4 contributes to P. aeruginosa pathogenicity in C. elegans. Similar to our 435 observations in biofilm assays, we found that the ∆cco1cco2 mutant displayed a more severe 436 phenotype than the ∆N1∆N2 mutant, suggesting that an orphan subunit can substitute for those 437 encoded by the cco1 and cco2 operons. We also found that deleting ccoN4 in ∆N1∆N2 led to a 438 ∆cco1cco2-like phenotype, suggesting that CcoN4 is the subunit that can play this role ( Figure  439 6). In host microenvironments where O 2 is available, CcoN4-containing isoforms could 440 contribute to its reduction. Additionally, in hypoxic zones, CcoN4-containing isoforms could 441 facilitate the reduction of phenazines, enabling cellular redox balancing. Both of these functions 442 would contribute to persistence of the bacterium within the host. The contributions of the Cco 443 oxidases to P. aeruginosa pathogenicity raise the possibility that compounds interfering with 444 Cco enzyme function could be effective therapies for these infections. Such drugs would be 445 attractive candidates due to their specificity for bacterial respiratory chains and, as such, would 446 not affect the host's endogenous respiratory enzymes. 447

448
Our discovery that an orphan cbb 3 -type oxidase subunit contributes to growth in biofilms further 449 expands the picture of P. aeruginosa's remarkable respiratory flexibility. Beyond modularity at 450 the level of the terminal enzyme complex (e.g., utilization of an aa 3 -vs. a cbb 3 -type oxidase), the 451 activity of P. aeruginosa's respiratory chain is further influenced by substitution of orphan cbb 3 -452 type catalytic subunits for native ones. Utilization of CcoN4-containing isoforms promotes 453 phenazine reduction activity and may influence aerobic respiration in P. aeruginosa biofilms. 454 For the exceptional species that contain orphan cbb 3 -type catalytic subunits, this fine level of 455 control could be particularly advantageous during growth and survival in environments covering 456 a wide range of electron acceptor availability (Cowley et al. 2015). 457 ∆N1∆N2∆N4∆hcn, and ∆cco1cco2∆hcn were diluted 1:50) and grown to mid-exponential phase 485 (OD at 500 nm ≈ 0.5). Ten microliters of subcultures were spotted onto 60 mL of colony 486 morphology medium (1% tryptone, 1% agar containing 40 µg/ml Congo red dye and 20 µg/ml 487 Coomassie blue dye) in a 10 cm x 10 cm x 1.5 cm square Petri dish (LDP D210-16). Plates were 488 incubated for up to five days at 25˚C with > 90% humidity (Percival CU-22L) and imaged daily 489 using a Keyence VHX-1000 digital microscope. Images shown are representative of at least ten 490 biological replicates. 3D images of biofilms were taken on day 5 of development using a 491

MATERIALS AND METHODS
Keyence VR-3100 wide-area 3D measurement system. hcn deletion mutants were imaged using 492 a flatbed scanner (Epson E11000XL-GA) and are representative of at least three biological 493 replicates 494 495 TTC reduction assay. One microliter of overnight cultures (five biological replicates), grown as 496 described above, was spotted onto a 1% tryptone, 1.5% agar plate containing 0.001% (w/v) TTC 497 (2,3,5-triphenyl-tetrazolium chloride [Sigma-Aldrich T8877]) and incubated in the dark at 25˚C 498 for 24 hours. Spots were imaged using a scanner (Epson E11000XL-GA) and TTC reduction, 499 normalized to colony area, was quantified using Adobe Photoshop CS5. Colorless TTC 500 undergoes an irreversible color change to red when reduced. Pixels in the red color range were 501 quantified and normalized to colony area using Photoshop CS5. grown to mid-exponential phase (OD at 500 nm ≈ 0.5). Exact OD at 500 nm values were read in 532 a Spectronic 20D+ spectrophotometer (Thermo Scientific) and cultures were adjusted to the 533 same OD. Adjusted cultures were then mixed in a 1:1 ratio of fluorescent:non-fluorescent cells 534 and ten µl of this mixture were spotted onto colony morphology plates and grown for three days 535 as described above. At specified time points, biofilms were collected, suspended in one ml of 1% 536 tryptone, and homogenized on the "high" setting in a bead mill homogenizer (Omni Bead Ruptor 537 12); day one colonies were homogenized for 35 seconds while days two and three colonies were 538 homogenized for 99 seconds. Homogenized cells were serially diluted and 10 -6 , 10 -7 , and 10 -8 539 dilutions were plated onto 1% tryptone plates and grown overnight at 37 °C. Fluorescent colony 540 counts were determined by imaging plates with a Typhoon FLA7000 fluorescent scanner (GE 541 Healthcare) and percentages of fluorescent vs. non-fluorescent colonies were determined. 542 543

Construction of terminal oxidase reporters. Translational reporter constructs for the Cco1, 544
Cco2, and CcoN4Q4 operons were constructed using primers listed in Table 1. Respective 545 primers were used to amplify promoter regions (500 bp upstream of the operon of interest), 546 adding an SpeI digest site to the 5' end of the promoter and an XhoI digest site to the 3' end of 547 the promoter. Purified PCR products were digested and ligated into the multiple cloning site of 548 the pLD2722 vector, upstream of the gfp sequence. Plasmids were transformed into E. coli strain 549 UQ950, verified by sequencing, and moved into PA14 using biparental conjugation with E. coli 550 strain S17-1. PA14 single recombinants were selected on M9 minimal medium agar plates (47.8 551 mM Na 2 HPO 4 •7H 2 O, 22 mM KH 2 PO 4 , 8.6 mM NaCl, 18.6 mM NH 4 Cl, 1 mM MgSO 4 , 0.1 mM 552 CaCl 2 , 20 mM sodium citrate dihydrate, 1.5% agar) containing 100 µg/ml gentamicin. The 553 plasmid backbone was resolved out of PA14 using Flp-FRT recombination by introduction of the 554 pFLP2 plasmid (Hoang et al. 1998) and selected on M9 minimal medium agar plates containing 555 300 µg/ml carbenicillin and further on LB agar plates without NaCl and modified to contain 10% 556 sucrose. The presence of gfp in the final clones was confirmed by PCR. 557 558 Thin sectioning analyses. Two layers of 1% tryptone with 1% agar were poured to depths of 4.5 559 mm (bottom) and 1.5 mm (top). Overnight precultures were diluted 1:100 (∆N1∆N4, 560 ∆N1∆N2∆N4, ∆cco1cco2 were diluted 1:50) in LB and grown until early-mid exponential phase, 561 for two hours. Then five to ten µL of subculture were spotted onto the top agar layer and 562 colonies were incubated in the dark at 25˚C with >90% humidity [Percival CU-22L]). Colonies 563 were grown for up to three days. At specified time points to be prepared for thin sectioning, for two hours each at 55˚C, then colonies were allowed to polymerize overnight at 4˚C. Tissue 573 processing was performed using an STP120 Tissue Processor (Thermo Fisher Scientific 813150). 574 Trimmed blocks were sectioned in ten µm-thick sections perpendicular to the plane of the colony 575 using an automatic microtome (Thermo Fisher Scientific 905200ER), floated onto water at 45˚C, 576 and collected onto slides. Slides were air-dried overnight, heat-fixed on a hotplate for one hour at 577 45˚C, and rehydrated in the reverse order of processing. Rehydrated colonies were immediately 578 mounted in TRIS-Buffered DAPI:Fluorogel (Electron Microscopy Sciences; Fisher Scientific 579 50-246-93) and overlaid with a coverslip. Differential interference contrast (DIC) and fluorescent 580 confocal images were captured using an LSM700 confocal microscope (Zeiss). Each strain was 581 prepared in this manner in at least biological triplicates. 582 583 Colony thickness measurements. Colonies were prepared for thin sectioning as described 584 above, but growth medium was supplemented with 40 µg/ml Congo Red dye (VWR 585 AAAB24310-14) and 20 µg/ml Coomassie Blue dye (Omnipur; VWR EM-3300). Colony height 586 measurements were obtained from confocal DIC images using Fiji image processing software 587 We thank Rachel Hainline for technical assistance with competition assays, Christopher 651 Beierschmitt for technical assistance with worm pathogenicity assays, and Konstanze Schiessl 652 for help with image analysis and feedback on the manuscript. 653   Biofilm morphologies are representative of more than ten biological replicates. Images were generated using a Keyence digital microscope. Scale bar is 1 cm. Bottom: 3D surface images of the biofilms shown in the top panel. Images were generated using a Keyence wide-area 3D measurement system. Height scale bar: bottom (blue) to top (red) is 0 -0.7 mm for WT, ∆N1∆N2, and ∆N4; 0 -1.5 mm for ∆N1∆N2∆N4 and ∆cco1cco2. (B) TTC reduction by cco mutant colonies after one day of growth. Upon reduction, TTC undergoes an irreversible color change from colorless to red. Bars represent the average, and error bars represent the standard deviation, of individually plotted biological replicates (n = 5). P-values were calculated using unpaired, two-tailed t tests comparing each mutant to wild type (***, P ≤ 0.001; ****, P ≤ 0.0001). For full statistical reporting, refer to Table 4    . We also included draft genomes (highlighted in purple) that contained genes involved in phenazine biosynthesis (phzABCDEFG). The tree revealed four clusters, each being more closely related to one of the four N subunits from PA14, which allowed us to annotate the N subunits accordingly. We next probed all genomes with N subunits for the presence of genes involved in cyanide synthesis (hcnABC) and phenazine biosynthesis (phzABCDEFG). In contrast to a previous claim, we did not find a clear correlation between the presence of CcoN4 and Hcn proteins (Hirai et al. 2016). We note that with the exception of two P.
fluorescens strains, those containing phzABCDEFG operons also contained ccoN4.        Figure 6. CcoN4-containing isoform(s) make unique contributions to PA14 virulence. Slow-killing kinetics of WT, gacA, and various cco mutant strains in the nematode Caenorhabditis elegans. Nearly 100% of the C. elegans population exposed to WT PA14 is killed after four days of exposure to the bacterium, while a mutant lacking GacA, a regulator that controls expression of virulence genes in P. aeruginosa, shows decreased killing, with ~50% of worms alive four days post-exposure. (A) ∆N1∆N2∆N4 and ∆cco1cco2 show comparably attenuated pathogenicity relative to WT. Error bars represent the standard deviation of at least six biological replicates. At 2.25 days post-exposure, significantly less C. elegans were killed by ∆N1∆N2∆N4 than by WT (unpaired two-tailed t test; p = 0.0022). (B) ∆N1∆N2 displays only slightly reduced pathogenicity when compared to WT. At 2.25 days post-exposure, significantly more C. elegans were killed by ∆N1∆N2 than by ∆N1∆N2∆N4 (unpaired two-tailed t test; p = 0.003). For full statistical reporting, refer to Table 4. Error bars represent the standard deviation of at least four biological replicates, each with a starting sample size of 30-35 worms per replicate.