ssRNA phage penetration triggers detachment of the F-

Laith Harba,b, Karthik Chamakuraa,b, Pratick Kharac, Peter J. Christiec, Ry Younga,b, and Lanying Zenga,b,1

aDepartment of Biochemistry and Biophysics, Texas A&M University, College Station, TX 77843; bCenter for Phage Technology, Texas A&M University, College Station, TX 77843; and cDepartment of Microbiology and Molecular Genetics, The University of Texas Health Science Center at Houston, Houston, TX 77030

Edited by Thomas J. Silhavy, Princeton University, Princeton, NJ, and approved August 27, 2020 (received for review June 9, 2020) Although the F-specific ssRNA phage MS2 has long had paradigm of the gRNA from the and its penetration into the cell status, little is known about penetration of the genomic RNA (gRNA) (13), a reduction in cellular nucleoside triphosphates (14), and into the cell. The phage initially binds to the F-pilus using its matu- cleavage of the 44-kDa Mat into two peptides that remain as- ration protein (Mat), and then the Mat-bound gRNA is released from sociated with the cell (15). the viral capsid and somehow crosses the bacterial envelope into The role of F-pili in the transfer of gRNA from the virion into the cytoplasm. To address the mechanics of this process, we fluo- the cell is mysterious. Two prevailing models account for the rescently labeled the ssRNA phage MS2 to track F-pilus dynamics mechanism of transport by F-pili. In the “pilus conduction” during infection. We discovered that ssRNA phage infection triggers model initially proposed by Brinton in 1965 (16), the pilus- the release of F-pili from host cells, and that higher multiplicity of docked phage delivers its gRNA into the central channel of infection (MOI) correlates with detachment of longer F-pili. We also the F-pilus for conveyance to the cell interior. Although there is report that entry of gRNA into the host cytoplasm requires the little direct evidence for this model, the pilus lumen contains – F- encoded coupling protein, TraD, which is located at the basic residues that potentially interact with the negatively cytoplasmic entrance of the F-encoded type IV secretion system charged gRNA (17). In the “pilus retraction” model, the pilus- (T4SS). However, TraD is not essential for pilus detachment, indicat- docked phage gains access to the bacterial cell surface through ing that detachment is triggered by an early step of MS2 engage- retraction of the pilus, whereupon the Mat-gRNA complex ment with the F-pilus or T4SS. We propose a multistep model in passes across the cell envelope through the T4SS or another MICROBIOLOGY which the ssRNA phage binds to the F-pilus and through pilus re- route (18). Although not reported for F-pili, RNA phage parti- traction engages with the distal end of the T4SS channel at the cell cles have been shown to accumulate at the base of pili elaborated surface. Continued pilus retraction pulls the Mat-gRNA complex out by IncP and IncC during infection, suggesting that pilus of the virion into the T4SS channel, causing a torsional stress that retraction has occurred (12, 19, 20). More support for this model breaks the mature F-pilus at the cell surface. We propose that P. aeruginosa phage-induced disruptions of F-pilus dynamics provides a selective comes from the finding that cells infected with the advantage for infecting phages and thus may be prevalent among ssRNA phage PP7 had a 50% reduction in overall type IV pilus the phages specific for retractile pili. lengths and that phage particles were often seen at the bases of pili, as observed on electron microscopy (21). In the absence of ssRNA phage MS2 | F-pilus | pili retraction | pili detachment | firm experimental support for either model, how the Mat-gRNA genomic RNA entry complex passes through the bacterial envelope into the cell in- terior remains unknown. ingle-stranded (ss) RNA phages have been identified for Sdiverse bacterial hosts, including (1, 2), Significance Pseudomonas aeruginosa (3, 4), Caulobacter crescentus (5), and Acinetobacter spp. (6). All known ssRNA phages infect their Many pathogenic bacteria utilize dynamic appendages called hosts by initially binding to retractile pili, such as the conjugative pili to facilitate important functions, such as transfer and pili associated with conjugative DNA transfer and the type IV motility. These processes contribute to the spread of antibiotic pili (which are not phylogenetically related to conjugative pili) resistance and support persistence in bacterial infec- mediating DNA uptake and twitching motility (7). Most of our tions, making them prime targets for therapeutic purposes. The knowledge regarding RNA phage biology stems from studies of single-stranded (ss) RNA all use dynamic pili to the ssRNA phages that infect enteric bacteria carrying the F sex facilitate infection; however, key mechanistic details describing factor plasmids (8). These plasmids code for type IV secretion how the pilus promotes viral entry into the host cell have remained elusive. Here we used fluorescence microscopy to systems (T4SSs) and associated F-pili, which are hollow, filamen- uncover a telling phenomenon associated with ssRNA phage tous, and dynamic appendages that extend and retract to initiate infection: initial penetration of the viral payload causes donor–recipient cell contacts during (9). breakage of host pili. This provides a selective advantage for The infection process has been best studied with the ssRNA the infecting phage, and thus this phenomenon may be phage MS2 and its derivatives. MS2 has a ∼3.5-kb genomic RNA widespread among other pilus-specific phage systems. (gRNA) that encodes four proteins: the maturation protein

(Mat) used for host recognition, the coat protein (Coat) that Author contributions: L.H. and L.Z. designed research; L.H., K.C., and P.K. performed re- forms the capsid, the protein (L), and the replicase (Rep), search; L.H. and L.Z. analyzed data; and L.H., K.C., P.K., P.J.C., R.Y., and L.Z. wrote the viral subunit of the RNA-dependent RNA replicase. The the paper. virions are ∼27 nm in diameter (10) and consist of 178 copies of The authors declare no competing interest. Coat encapsidating the gRNA, which is bound to a single copy of This article is a PNAS Direct Submission. Mat (11). Infection is initiated when MS2 adsorbs to the side of Published under the PNAS license. the F-pilus via the Mat protein. The binding of MS2 to piliated 1To whom correspondence may be addressed. Email: [email protected]. cells at 4 °C or free F-pili at 4 to 37 °C is reversible and does not This article contains supporting information online at https://www.pnas.org/lookup/suppl/ lead to loss of infectivity in the viral particles (12); however, at doi:10.1073/pnas.2011901117/-/DCSupplemental. 37 °C, the interaction of MS2 with piliated cells leads to release

www.pnas.org/cgi/doi/10.1073/pnas.2011901117 PNAS Latest Articles | 1of8 Downloaded by guest on September 29, 2021 In this study, we used fluorescently labeled MS2 phage to are inhibited on ice (27, 28), the addition of MS2-GFP to characterize F-pilus dynamics during ssRNA phage infection. quantify pili does not promote further infection. Strikingly, we discovered that MS2 triggers the release of F-pili by Remarkably, the MS2-infected culture showed a marked in- a mechanism requiring retraction. Our findings strongly support crease in the amount of free F-pili in the media (i.e., F-pili not the “pilus retraction” model and suggest a novel mechanism for attached to a cell) compared with the buffer-only control culture. superinfection exclusion resulting from ssRNA phage infection. As shown in Fig. 2B, the frequency of detached pili increased with time and reached a plateau after ∼10 min (∼45%), which is Results when gRNA penetration is essentially complete (29). This sug- Detection of F-pili by Fluorescent Labeling of MS2 Particles. Previ- gests that F-pilus detachment is related to MS2 gRNA entry into ously, ssRNA phage particles labeled with residue-specific fluo- the cell. Notably, we did not see any increase in detached pili in rescent dyes have been used to identify F-pili by fluorescence the uninfected culture, suggesting that the binding of high microscopy (22, 23). Although this method produces labeled numbers of MS2-GFP to the pili does not lead to shearing or virions, we noted a significant titer loss of 50% after labeling, breakage during imaging. possibly due to the labeling of Mat by the dye (23). To avoid To determine whether F-pilus detachment is MS2-specific, we perturbing the binding interface of the phage particles, we tagged tested whether other F-specific phages induce detachment. Like β MS2 virions with fluorescent proteins to illuminate F-pili, as was MS2, the ssRNA phage Q (an Allolevivirus) was seen to bind to previously done for labeling of phage lambda and P1 (24, 25). the side of the F-pilus. In contrast, ssDNA phage M13 was bound Cells harboring a plasmid encoding the MS2 Coat fused to exclusively to the pilus tip. Similarly to MS2, we observed an β sfGFP (Fig. 1A) were infected with MS2 and subsequently in- increase in detached pili when cells were infected with Q , al- though ultimately only ∼30% of the total pili were detached duced for production of the Coat-sfGFP fusion protein. During β assembly, the phage incorporates fluorescent Coat proteins, (Fig. 2B). In contrast to MS2 and Q , infection by M13 yielded a resulting in mosaic particles containing both native and GFP- similar basal level of pilus detachment as the uninfected control. 13 These results suggest that phage-triggered F-pilus detachment is tagged Coat (Fig. 1B). We typically obtain titers of ∼5 × 10 a phenomenon associated only with infection by ssRNA phages, pfu/mL for both the GFP-tagged and WT purifications. Purified or at least phages that bind to the sides of the F-pilus. fluorescent phages (herein designated MS2-GFP) are easily de- tected by fluorescence microscopy (Fig. 1C). Analysis of the fluorescence intensity of the MS2-GFP foci showed a uniform F-Pilus Retraction Is Required for Phage-Triggered Detachment. We distribution of the GFP signal, suggesting that the majority of the next examined how phage MOI affects the frequency and char- foci represent one population of MS2-GFP (Fig. 1D). When acteristics of F-pilus detachment. Cells were infected with MS2 mixed with cells harboring the F-plasmid, MS2-GFP specifically at MOIs of 0, 0.5, 5, and 10, and the numbers of detached pili bound to and illuminated F-pili (Fig. 1E). MS2 does not bind to were quantitated at 10 min postinfection, which is when pilus detachment reaches a plateau. Notably, the frequency of de- cells that lack F-pili (26), and, accordingly, cells carrying a de- ∼ letion of the F-pilin gene traA were not fluorescent (SI Appendix, tached pili increased from 5% without infection (i.e., MOI of 0) to ∼25% at an MOI of 0.5, and further increased and pla- Fig. S1). teaued at ∼50% at MOIs of 5 and 10 (Fig. 2C). The observed correlation between phage MOI and level of pilus detachment Early Infection by ssRNA Phages Leads to F-Pilus Detachment. Ob- provides further evidence that phage entry triggers F-pilus de- serving pili directly after phage infection is difficult and tradi- tachment. tionally has required the use of electron microscopy (14, 21). We next exploited differences in MOIs to test whether F-pilus Here we used the fluorescent phage to quantitatively evaluate retraction is required for phage-triggered detachment. Because the effect of MS2 binding on F-pilus behavior in liquid culture MS2 can bind along the length of the pilus [22], at a low MOI, (Fig. 2A). Separate cultures of E. coli HfrH were treated with phage will bind sparsely. In contrast, at high MOI, phages will MS2 at a multiplicity of infection (MOI) of 5, or buffer only, and saturate the length of the pilus and, importantly, have a greater incubated at 37 °C. Infections were performed without shaking to likelihood of binding near the base of the pilus. We reasoned prevent pilus shearing. Aliquots were removed at different time that if F-pilus retraction is required for phage-triggered de- points, dispensed into prechilled tubes, and allowed to equili- tachment, infection at higher MOIs should enrich for longer brate on ice prior to the addition of excess MS2-GFP at MOI = species of detached pili compared with infection at lower MOIs. 100 to visualize the F-pili. Since F-pilus outgrowth and retraction Indeed, we observed a shift in the distribution toward longer

Fig. 1. Fluorescent capsid labeling of MS2 virions to detect F-pili. (A) The MS2 Coat protein was fused to sfGFP and cloned downstream of an inducible lac

promoter (Plac) on the pZE12 vector. (B) Schematic of MS2-GFP assembly. Cells carrying the Coat-sfGFP plasmid are infected with WT MS2 (RNA in red; not drawn to scale), then induced with IPTG to produce fluorescent Coat proteins. Coat-sfGFP is incorporated along with WT Coat proteins to form the mosaic capsid. (C) Purified MS2-GFP particles imaged under the fluorescence microscope. (D) Intensity distribution of MS2-GFP signal (1,826 spots). (E) Strain HfrH incubated with MS2-GFP. The fluorescent virions bind along the F-pili and enable their detection by proxy.

2of8 | www.pnas.org/cgi/doi/10.1073/pnas.2011901117 Harb et al. Downloaded by guest on September 29, 2021 MICROBIOLOGY

Fig. 2. Infection by ssRNA phages leads to pilus detachment. (A) Schematic of pilus enumeration assay. A phage suspension or equivalent buffer is added to a static culture of HfrH. Small aliquots are removed at different time points and dispensed into prechilled tubes. A surplus of MS2-GFP is then added to the chilled aliquots, and the samples are imaged by fluorescence microscopy. The representative image depicts detached pili and cell-associated pili. (B) Frequency of pili detachment over time. Number of pili: MS2, 4,701; Qβ, 3,070; M13, 3,730; HfrH control, 2,814. The percentage of detached pili is the number of detached pili divided by the number of total pili (detached and cell-associated). (C) MOI effect on pilus detachment. Number of pili per MOI: 0, 1,072; 0.5, 1,170; 5, 1,130; 10, 1,123. HfrH cells were infected with different MOIs of MS2, and samples were collected at 10 min. Note that MOI = 0 means uninfected. (D) Length distribution of detached pili at different MOIs. Number of detached pili per MOI: 0, 71; 0.5, 300; 5, 380; 10, 351. Detached pili from MS2 infected cultures at 10 min postinfection were measured for length. Average lengths of detached pili per MOI: 0, 2.32 μm; 0.5, 1.64 μm; 5, 2.16 μm; 10, 2.19 μm. Error bars denote SD. For B, on points where error bars are not visible, SD is <1%. Each experiment was done at least twice.

lengths of detached pili with increasing MOI (Fig. 2D). The First, we asked whether MS2 is capable of ejecting its gRNA average detached pilus length increased from 1.64 μm at an MOI on binding to F-pili elaborated by ΔtraD mutant cells. When of 0.5 to 2.16 μm at an MOI of 5 and 2.19 μm at an MOI of 10. ssRNA phages bind to F-pilus receptors, they undergo a process The detached pili resulting from MS2 infection are generally termed “eclipse,” which involves release of the capsid from the shorter than the cell-associated pili in the uninfected culture, phage particle, leaving the RNA susceptible to exogeneous which had an average length of 2.66 μm(SI Appendix, Fig. S2). RNase (13). Therefore, phage eclipse can be assessed through These findings strongly indicate that MS2 causes detachment of the addition of exogenous RNase to a culture before infection F-pili in the process of retraction. and then the use of a plaque assay to determine whether there is a reduction in plaque-forming units (PFUs) of phage recovered from the culture supernatant (33). MS2-Resistant Mutant ΔtraD Inhibits gRNA Entry. Having gathered We assayed for MS2 gRNA ejection on binding of strains evidence suggesting that F-pilus retraction is required for MS2- harboring pOX38, a fully functional variant of the F-plasmid Flac, triggered detachment, we next asked whether subsequent steps and an isogenic strain in which traD was cleanly deleted from of phage infection are also required. Studies of MS2 and related – pOX38 (34). Our studies confirmed that the strain harboring phages (e.g., R17) have shown that the F-plasmid encoded T4SS pOX38ΔtraD elaborated WT levels of F-pili as monitored by is required for phage entry, implying that the phage crosses the MS2-GFP phage binding (SI Appendix, Fig. S3), but failed to cell envelope via the T4SS channel (30). However, discriminating conjugatively transfer the F-plasmid through the F-encoded between the requirements of the F-pilus receptor and elabora- T4SS in mating assays (SI Appendix, Table S1). Using the tion of the T4SS channel for phage infection and pilus detach- phage eclipse assay, we determined that MS2 infection of the ment is challenging, because mutations of T4SS components ΔtraD mutant in the presence of RNase resulted in a decrease in almost invariably abolish pilus production (31). One prominent viable phage particles recovered from the culture supernatant exception is that mutation of TraD, an ATPase situated at the over time, with PFU becoming undetectable by 10 min (Fig. 3A). channel entrance that couples the F-plasmid substrate to the It is important to note that no loss of phage viability was reported T4SS for conjugative transfer, blocks infection by the MS2- when intact phage particles were incubated with RNase, indi- related phage f2 without affecting F-pilus biogenesis or phage cating that the loss of infectivity requires phage eclipse (33). This adsorption (32). Thus, we explored the contribution of TraD to decrease in PFU counts closely resembled that observed with MS2 infection and MS2-triggered detachment of F-pili. MS2 infection of the RNase-treated parental strain harboring

Harb et al. PNAS Latest Articles | 3of8 Downloaded by guest on September 29, 2021 native pOX38 (Fig. 3A). Thus, MS2 commits to infection by to undergo phage eclipse on binding of F-pili elaborated by releasing its gRNA on binding to F-pili elaborated by pOX38- strains harboring either pOX38 or the ΔtraD variant, the phage carrying WT and ΔtraD mutant cells. RNA enters the cytoplasm of pOX38-carrying cells only. We The phage eclipse assay monitors release of gRNA from its conclude that TraD is required for passage of the MS2 gRNA capsid but does not provide an indication of whether gRNA across the cell envelope or entry into the cytoplasm. enters the bacterial cytoplasm. To determine whether the ΔtraD mutation affects gRNA entry into the cell, we used single- MS2 Triggers Detachment of F-pili Elaborated by ΔtraD Mutant Cells. molecule fluorescent in situ hybridization (smFISH) to directly We have shown that MS2-triggered detachment of F-pili re- detect MS2 RNA in the cell. As shown in Fig. 3B, with probes quires pilus retraction (Fig. 2). We next asked whether pilus designed to bind to the first half of the MS2 , we visu- detachment also requires phage entry. In view of our evidence alized MS2 RNA in pOX38-carrying cells. Single fluorescent foci Δ corresponding to MS2 RNA were observed within 5 min of in- the traD mutation permits extracellular stages of infection (e.g., fection, larger clusters of fluorescence were seen at 10 min phage binding and eclipse) but blocks phage entry, we tested Δ postinfection, and the signal filled the cell at 20 min postinfec- whether phage infection triggers the release of F-pili from traD tion, indicative of phage RNA replication. In striking contrast, mutant cells. Interestingly, reminiscent of our findings with HfrH no signal was detected on MS2 infection of ΔtraD cells, even cells, infection of the ΔtraD mutant also induced pilus detach- after 10 min postinfection. By quantifying the percentage of in- ment. At 10 min postinfection, MS2 triggered F-pilus detach- fected pOX38-carrying cells (i.e., cells containing a detectable ment from the ΔtraD mutant, albeit at a lower frequency of MS2 RNA signal over time), we found that the MS2 entry period ∼20% compared with ∼40% for infection of pOX38-carrying concludes by 10 min (Fig. 3C). Thus, despite the capacity of MS2 cells (Fig. 3D).

Fig. 3. MS2 resistant mutant ΔtraD inhibits MS2 gRNA entry while retaining pili detachment by MS2 infection. (A) Loss of MS2 titer over time during in- fection of pOX38 or ΔtraD cells in the presence of RNase. (B) Representative images of MS2 RNA in infected cells as detected by smFISH (MOI = 0.5). The yellow signal (6-TAMRA) is indicative of MS2 RNA. (Top) pOX38 cells at different time points postinfection. (Bottom) ΔtraD cells. (C) Percentage of infected cells (i.e., cells containing MS2 RNA compared with total cells) at different time points postinfection. Number of cells: pOX38, 4,515; ΔtraD,3,511.(D) Frequency of pili detachment over time when MS2 interacts with ΔtraD cells and pOX38. The experiment was performed as in Fig. 2. Number of pili: pOX38, 1,424; ΔtraD, 1,669; ΔtraD(−), 2,450; pOX38(−), 2,088. Minus symbols indicate the uninfected controls. (E) Length distribution of detached pili from samples in D at 10 min postinfection. Number of detached pili per sample and average lengths: pOX38, 118, 1.99 μm; ΔtraD, 115, 2.31 μm; ΔtraD(−), 17, 2.68 μm; pOX38(−), 52, 2.92 μm. Error bars denote SD. Where error bars are not visible, SD is <2%. Each experiment was done at least twice.

4of8 | www.pnas.org/cgi/doi/10.1073/pnas.2011901117 Harb et al. Downloaded by guest on September 29, 2021 We also measured the lengths of the detached pili found in protofilament junction might cause a torsional stress that breaks supernatants of strains harboring pOX38 and the ΔtraD mutant the mature pilus from the protofilament. In the latter pathway, plasmid (Fig. 3E). The distribution of pili from the MS2-infected after detachment of the mature pilus, further retraction of the ΔtraD mutant was similar to that from the infected pOX38- protofilament would effectively pull Mat and associated gRNA carrying strain, with average lengths of 2.31 μm and 1.99 μm, from the virion and into the T4SS channel (Fig. 4C). The empty respectively. The difference in length could be attributed to the capsid, lacking any ability to bind to the cell, is then released Δ increased length of F-pili elaborated by traD cells; the average (Fig. 4D). Δ μ length of pili bound to traD cells was 2.65 m, compared with Even relieved of the distal pilus and the viral capsid, the μ 2.40 m for pili bound to pOX38-carrying cells (SI Appendix, pilin-Mat-gRNA complex still represents a formidable cargo to Fig. S4). traverse the T4SS channel, given the megadalton size of the Taken together, these findings indicate that first, F-pili elab- Δ gRNA. Unless the channel expands during this process, its di- orated by the traD mutant retract and, second, F-pilus retrac- ameter is too small to accommodate the numerous secondary tion is necessary for MS2-triggered pilus detachment. Finally, structures that occupy most of the gRNA. It is also possible that MS2 triggers F-pilus detachment before entry of gRNA into the the energy driving pilus retraction (e.g., ATP hydrolysis or the cytoplasm. proton motive force) suffices to extract the Mat-gRNA complex Discussion from the virion and denature the secondary and tertiary structure – In this study, we explored the fate of F-pili after infection with of the gRNA. Other than the requisite Mat pilin contacts for pilus-specific phages. We discovered that the ssRNA phages gRNA entry, we do not know whether Mat or the capsid exterior MS2 and Qβ trigger detachment of F-pili from the bacterial cell also forms specific contacts with the external surface of the surface early during the infection process. Both the degree of channel. Even if this were so, these additional interactions are detachment and the length of detached pili were correlated with not required for phage infection, because the Mat-gRNA com- the MOI, supporting the notion that pilus retraction is required plex itself is infectious (36). Consequently, the strength of the for phage-triggered detachment. We also found that deletion of Mat–pilin interaction is the driving factor for successful infec- the TraD coupling protein blocks entry of MS2 gRNA into the tion, and it may also specifically underlie the dependence of host cell cytoplasm but does not affect the initiating steps of phage MS2, but not of Qβ, on the TraD T4CP for infection. Previous binding or “eclipse,” or of phage-triggered pilus detachment. work has shown that MS2 adsorbs much more readily to F+ cells Our findings support a conclusion that MS2 and, likely other than Qβ (37), presumably reflecting differences in binding in- ssRNA phages such as Qβ, induce detachment of the F-pilus at teractions between the two Mat proteins with F-pilin. Thus, MICROBIOLOGY “ ” a step of the infection process subsequent to the eclipse TraD may be needed to detach the MatMS2 from the pilin at the phase but before entry of gRNA into the cell. We propose the cytoplasmic interface of channel, but not for the comparatively model shown in Fig. 4 to account for our findings, and further weaker MatQβ–pilin interaction. suggest that ssRNA phage-triggered detachment of F-pilus We also visualized gRNA entry into the bacterial cytoplasm receptors represents a previously undescribed mechanism of over time using smFISH. At early postinfection times, we de- superinfection exclusion. tected one or a few foci, possibly reflecting sites of phage pen- ssRNA phages initiate infection by binding to the sides of the etration across the cell envelope and entry of the gRNA into the F-pilus (Fig. 4A), which has been shown to dynamically extend cytoplasm. Between 10 and 20 min postinfection, phage repli- and retract (9). During retraction, bound phages are drawn close cation ensues, resulting in a proliferation of phage nucleic acid to the surface of the cell (Fig. 4B). To explain how phages might throughout the cell. Interestingly, in pOX38-carrying cells, we trigger F-pilus detachment during the retraction phase, it is first important to summarize recent structural findings for the did not observe a significant increase in the cell-associated pili per cell following the 10-min entry period. In contrast, cells F-encoded T4SS and associated F-pilus obtained by in situ cryo- Δ electron tomography (cryoET). Importantly, these studies harboring the pOX38 traD variant, or cells not exposed to established that the diameter of the T4SS channel is only ∼3.5 phage, continue to elaborate F-pili over time, well beyond the nm, considerably less than the width of the mature F-pilus (∼8.5 10-min time frame required for phage entry into pOX38-carrying nm). To reconcile this finding with early evidence that F-pili cells (SI Appendix, Fig. S3). Coupled with our finding that ΔtraD assemble and retract by reiterative rounds of extraction and cells are protected from infection by MS2, these findings suggest reincorporation of TraA pilin monomers from and into an inner that phage infection blocks synthesis of F-pili, possibly due to the membrane pilin pool, it was proposed that F-pili nucleate in two stages. First, TraA pilins are extracted from the inner membrane to build a thin protofilament, which extends through the central channel of the T4SS. Second, as the protofilament reaches the cell surface, it packs into the helical array visualized for the mature pilus. This two-stage assembly reaction proceeds until receipt of a signal for retraction, whereupon the process reverses and ultimately shunts pilin monomers back into the inner membrane. Interestingly, it was also observed that the walls of the F-pilus at the junction with the T4SS channel were thinner than in the mature helical fiber. Thus, the protofilament–mature pilus structural transition might represent a point of weakness that could account for the observed susceptibility of F-pili to breakage by shear forces (35). Fig. 4. Model for pilus detachment-facilitated gRNA entry of ssRNA phage In the context of this model for F-pilus biogenesis, we propose MS2. (A) The ssRNA phage (not to scale) binds to the side of the pilus (or- ange) using the Mat (magenta). (B) Pilus retraction brings the ssRNA phage one of two outcomes for pilus-bound phage on encountering the – to the cell surface and in proximity to the basal body. (C) Continued re- pilus T4SS channel junction during pilus retraction. Further traction forces the Mat-gRNA (red) complex into the distal end of the basal retraction might generate a force that causes disruption of the body, causing breakage of the pilus at the point of entry. (D) Passage Mat –pilin contacts and thus release of the virions. Alternatively, through the basal occurs, and entry is facilitated by TraD (green). (Figure contact of the comparatively bulky phage with the pilus– adapted with permission from ref. 35.)

Harb et al. PNAS Latest Articles | 5of8 Downloaded by guest on September 29, 2021 ∼ energy burden exacted on the host cells, as evidenced by a reduction an OD600 of 0.4 before being diluted into prewarmed LB containing 4 mM β in cellular nucleotide triphosphatases that accompanies infection. CaCl2 to an OD600 of 0.1. Afterward, purified WT MS2, Q , or M13 was added In summary, we have identified two mechanisms by which to the culture at an MOI of 5. Infections were performed at 37 °C without ssRNA phages can infect F-plasmid–carrying cells but in the shaking to minimize pilus agitation. Samples were removed at regular time process render their hosts less susceptible to superinfection by points using cut pipette tips and dispensed into prechilled tubes and allowed to equilibrate on ice. A small volume of MS2-GFP equating to an MOI of 100 other phages that are dependent on F-pili for attachment. First, was added to the samples and left on ice for 20 min to allow phage infection triggers detachment of the F-pilus receptors. This ef- adsorption to the pili. fectively prevents other male-specific phages from infecting the Samples were prepared for microscopy by spotting 1 μL of sample onto a same host cell, and it coincidentally generates a pool of F-pilus large coverslip (No. 1, 24 × 50 mm; Thermo Fisher Scientific) and gently “decoys” in the milieu to which phage bind nonproductively, overlaying a small, 1-mm-thick 1.5% agarose pad dissolved in PBS on the thereby reducing the numbers of phage particles available for sample before placing it under the microscope. Z stacks of 300 nm were used infection. Second, entry of phage into host cells imposes a block in the GFP channel to precisely measure pili. Cells and pili were imaged on on further F-pilus assembly, again rendering host cells less sus- multiple stage positions (between 10 and 20 per sample) in phase (100 ms ceptible to superinfection. In view of the selective advantage exposure to detect cells) and GFP (100 ms to detect pili) channels. The images accompanying superinfection exclusion, we propose that one or were analyzed using NIS-Elements software (Nikon) for cell and pili counts both mechanisms may be prevalent in ssRNA phage systems that and length measurements. are dependent on retractile conjugative or type IV pili for infection. Construction of pOX38ΔtraD. E. coli strain HME45 carrying pOX38 was used to generate complete deletion of traD by recombineering (39). In brief, a FRT- Materials and Methods KanR-FRT cassette from pKD13 (40) was PCR-amplified with primers to carry Bacterial Strains. The strains used in this study are summarized in SI Appendix, flanking 5′ and 3′ sequences of 41 and 50 bases with homology to the up- Table S2.AllE. coli strains were grown in LB medium containing appropriate stream and downstream regions of traD. HME45(pOX38) was induced for antibiotics at standard concentrations: ampicillin, 100 μg/mL; carbenicillin, expression of the λ red-gam genes by growth at 42 °C for 15 min. The FRT- 100 μg/mL; tetracycline, 10 μg/mL; rifampicin, 50 μg/mL; kanamycin, 50 μg/ KanR-FRT amplicon was introduced by electroporation, and transformants mL; and chloramphenicol, 10 μg/mL. Cells were grown in baffled flasks at were selected by plating on LB agar plates with kanamycin. Plasmid pPK31

37 °C with shaking (225 rpm) unless indicated otherwise. expressing traD from the PBAD promoter was constructed by PCR ampli- fication of traD using primers listed in SI Appendix,TableS2and pOX38 as a template, digestion of the PCR product with NheI and HindIII, and in- Construction of Coat-sfGFP. Plasmid pBAD33 MS2 Coat-GS -sfGFP (cat araC 15 troduction of the resulting restriction fragment into similarly digested Para::coat-gs15-sfGFP) encoding a Coat-GlySer15–linked super-folder Green plasmid pBAD24. pPK31 was then introduced by transformation into Florescent Protein was constructed by cloning the synthetic DNA fragment Δ R (g-block_MS2Coat_sfGFP; IDT) into pBAD33 between the KpnI and HindIII HME45 carrying pOX38 traD::FRT-Kan -FRT for conjugative transfer of sites. The MS2 coat gene was codon-optimized to reduce the probability the pOX38 variant into MC4100 carrying pCP20 (40) expressing the Flp of recombination with the infectious MS2 cDNA clones. To construct the recombinase. Transconjugants were streaked on LB agar and grown at fusion construct, the MS2 Coat (YP_009640125.1) was codon-optimized 42 °C overnight to induce recombinase expression for excision of the FRT- R using the codon optimization tool (idtdna.com/CodonOpt), and the fi- Kan -FRT cassette; the temperature shift also cured temperature-sensitive r s s nal translated product differed by a single substitution pCP20. Individual colonies were screened for Tet , Chl ,andKan to Δ Δ (A125V). A DNA sequence encoding a 15-aa “GGGGSGGGGSGGGGS” identify a strain harboring pOX38 traD.The traD mutation was con- linker and sfGFP was added in frame with the codon-optimized 130 firmed by PCR amplification across the deletion junction and sequencing

codons of MS2 coat to construct the coat-gs15-sfGFP fusion and finally to of the PCR fragment. facilitate cloning, restriction sites for KpnI and HindIII were added up- stream and downstream, respectively. The cloned construct was verified Conjugation Assay. Donor and recipient strains were grown overnight in LB by Sanger sequencing (Eton Bioscience) with primers KC30 and KC31. broth with appropriate antibiotics. Strains were then diluted 1:50 in fresh The insert from this plasmid was amplified by PCR using primers LH1 and antibiotic-free LB broth and incubated with shaking for 3 h at 37 °C. For LH2 that introduce KpnI and XbaI sites upstream and downstream of the induction of traD, the overnight donor strain was subcultured 1:50 in LB product, respectively. The product was subcloned into pZE12 for induc- broth for 1.5 h, induced with arabinose (0.2% final concentration), and tion with isopropyl β-D-1-thiogalactopyranoside (IPTG) and confirmed by grown for another 1.5 h at 37 °C. Donor and recipient cultures were mixed in sequencing. a 1:1 ratio and incubated at 37 °C for 1.5 h without shaking. The mating mixtures were then serially diluted and plated onto LB agar containing Preparation of MS2 Lysates. A single colony of ER2738 was grown in 5 mL of LB antibiotics selective for transconjugants (TCs) and donors (D). The frequency with appropriate antibiotics overnight at 37 °C. The overnight culture was of DNA transfer was calculated as the number of TCs per donor (TCs/D). diluted 1:100-fold in LB containing 4 mM CaCl2 with antibiotics and grown at Experiments were repeated three times in triplicate, and the results of a ∼ 37 °C until OD600 0.4 and infected with MS2 at an MOI of 20 and allowed to representative experiment are presented. propagate for 5 h. Then 1% chloroform was added, and the cells were centrifuged at 8,000 rpm for 30 min. The GFP-tagged MS2 was prepared in the same way using LZ2619 cells with 1 mM IPTG treatment at 10 min Phage Sensitivity Assay. Fifty microliters of overnight-grown cells were spread postinfection. Purification of the phages was achieved using ammonium on LB agar plates containing appropriate antibiotics and 0.2% arabinose if μ 9 sulfate precipitation and CsCl gradient centrifugation (38). In brief, 280 g per necessary for induction of traD. After drying, 2 L of MS2 (10 pfu/mL) and 11 liter of ammonium sulfate was added to the lysates in small increments, M13 (10 pfu/mL) phages were spotted at different locations on the plate followed by chilling at 4 °C for 4 h. The precipitates were sedimented by and incubated overnight at 37 °C. Plates were then examined for plaque centrifugation at 8,000 rpm for 1 h and resuspended in MS2 buffer (150 mM formation indicative of phage sensitivity. NaCl, 5 mM EDTA, and 50 mM Tris pH 7.5). The suspensions were dialyzed in – 3.5 10 kDa molecular weight cutoff cassettes three times at 2-h intervals RNA Ejection Assay. MS2 ejection was measured essentially as described with MS2 buffer at 4 °C, followed by treatment with DNase at 10 U/mL for previously (33). In brief, piliated cells were diluted 1:100 from overnight 1 h at 4 °C. A clearing spin was performed, after which the phage prepa- cultures into LB containing 4 mM CaCl2. Once the OD600 reached 0.4, cells rations were mixed with 0.55 g CsCl per gram of phage solution to produce a were diluted to an OD600 of 0.1 in prechilled LB containing 4 mM CaCl2 and final density of 1.38. The suspensions were centrifuged at 45,000 rpm for 100 μg/mL RNase A. Purified MS2 was added to this culture at an MOI of 24 h at 4 °C. The bands were collected and dialyzed against MS2 buffer, and 0.01, followed by incubation on ice for 20 min. Then the cultures were the purified preparations were stored at 4 °C. moved to a 37 °C shaking water bath to initiate infection. At regular time points after phage addition, samples were removed and diluted 100-fold Pilus Detachment Assay. Overnight cultures of piliated cells were diluted into prechilled SM buffer. The titer was then determined for these samples

1:1,000 in LB containing antibiotics and 4 mM CaCl2. The cells were grown to following standard protocol.

6of8 | www.pnas.org/cgi/doi/10.1073/pnas.2011901117 Harb et al. Downloaded by guest on September 29, 2021 RNA smFISH. A set of 48 probes each ∼20 nt long targeting the 5′ half of the formamide, 2× SSC), incubated at room temperature for 5 min, and then MS2 genome with at least a 2-nt spacing between each probe was designed centrifuged again. For hybridization, the cells were resuspended in 25 μLof and synthesized using the Stellaris Probe Designer (Biosearch Technolo- hybridization solution (40% [wt/vol] formamide, 2× SSC, 1 mg/mL E. coli gies); the sequences are listed in SI Appendix, Table S3).The probes were tRNA, 2 mM ribonucleoside-vanadyl complex, and 0.2 mg/mL BSA) contain- designed with a 3′ mdC(TEG-amino) modification to enable labeling with ing the probe mixture at a final concentration of 1 μM. The samples were 6-carboxytetramethylrhodamine, succinimidyl ester (6-TAMRA; Thermo Fisher incubated overnight at 30 °C, protected from light. Afterward, samples were Scientific, catalog no. C6123) following previously described protocols (41). In washed three times by incubating the cells in wash solution for 30 min at μ μ brief, we pooled 7.5 L of each of the oligo solutions (48 oligo in total, 100 M 30 °C. After the third washing, 10 μg/mL DAPI was added to the wash so- each) and added 40 μL of 1 M sodium bicarbonate, pH 8.5. We then added lution and used to stain the E. coli DNA. Cells were resuspended in 2 × SSC dye solution (1 mg of 6-TAMRA dissolved in 2.5 μL of DMSO and 25 μLof and prepared for imaging. Cells were imaged in phase (100 ms exposure) 0.1 M sodium bicarbonate, pH 9.0) to the probe mixture and incubated the and in the Cy3 channel to detect 6-TAMRA signals (50-ms exposure). Z-stacks mixture protected from light overnight at 37 °C. were taken at 300-nm intervals. The next day, we mixed 47 μL of 3 M sodium acetate (pH 5.2) into the solution, added 1,180 μL of 100% ethanol, and incubated the sample at −80 °C for 3 h to precipitate the oligos. The oligos were then centrifuged Fluorescence Microscopy Imaging. Imaging was performed primarily on a and washed twice more by dissolving the pellet in 45 μL of DEPC-treated Nikon Eclipse Ti2 inverted epifluorescence microscope using a 100× objective water, 5 μL of 3 M sodium acetate, pH 5.2, and 125 μL of 100% ethanol. (Plan Apochromat, NA 1.45, oil immersion) at room temperature and ac- μ × After the washing steps, the probes were resuspended in 250 Lof1 TE, quired using a cooled EMCCD camera with mask (Princeton Instruments). × resulting in a 10 probe stock solution. The probe solution was diluted with Occasionally, imaging was performed on a Nikon Eclipse Ti inverted epi- × × 1 TE to make the 1 probe solution on use, and the labeling efficiency of fluorescence microscope using a 100 × objective (Plan Fluo, NA 1.40, oil the probes was measured at 86% using a NanoDrop One UV-Vis Spectro- immersion) with a 2.5× TV relay lens, and acquired using a cooled EMCCD photometer (Thermo Fisher Scientific). camera (IXON 897; Andor). Cells were imaged under the phase-contrast and The probes were then used to detect MS2 RNA during infection using the with standard fluorescent filter cubes: GFP filter (Nikon 96363) and Cy3 filter following protocol. Overnight cultures of piliated cells were diluted 1:1,000 in (Nikon 96323). When presenting microscopy images in figures, uniform LB containing 4 mM CaCl2 and appropriate antibiotics. The cells were grown contrast settings were applied for each separate channel throughout the at 37 °C with shaking to an OD600 of 0.4, then diluted into prechilled LB entire figure. containing 4 mM CaCl2 to an OD600 of 0.1. MS2 at a MOI of 0.5 was added, and the culture was incubated on ice for 20 min to allow for adsorption to pili. As a negative control, the same volume of MS2 buffer was added to the Data Availability. The data supporting the findings are provided in the main culture. The cultures were moved to a 37 °C shaking water bath to initiate text and SI Appendix. All experimental materials are available upon request. infection. At regular time points, an aliquot of the cells was centrifuged at

× × MICROBIOLOGY 4500 g for 5 min and resuspended in 1 PBS containing 3.7% formalde- ACKNOWLEDGMENTS. We thank members of the L.Z., R.Y., and Junjie hyde. At this point, the samples were further processed following the Zhang laboratories at Texas A&M University for technical assistance and established protocol (41). The cells were fixed for 30 min on a nutator at support; Shivangi Patel and Junho Lee for assistance with laboratory room temperature before being washed three times in 1× PBS to remove techniques; and Christopher Hayes for the gift of strains. This research was excess formaldehyde. The cells were then treated with 70% ethanol supported by NSF Grant 1902392 and Texas A&M University X-Grant 290386 and nutated for 1 h to permeabilize the cells for probe penetration. The cells to the L.Z. laboratory, NIH Grants R01 GM27099 and R35 GM136396 to the were centrifuged and resuspended in wash solution (40% [wt/vol] R.Y. laboratory, and NIH Grant R35 GM131892 to the P.J.C. laboratory.

1. K. V. Gorzelnik et al., Asymmetric cryo-EM structure of the canonical allolevivirus Qβ 16. C. C. Brinton Jr., The structure, function, synthesis and genetic control of bacterial pili reveals a single maturation protein and the genomic ssRNA in situ. Proc. Natl. Acad. and a molecular model for DNA and RNA transport in gram- negative bacteria. Trans. Sci. U.S.A. 113, 11519–11524 (2016). N. Y. Acad. Sci. 27, 1003–1054 (1965). 2. S. Kannoly, Y. Shao, I. N. Wang, Rethinking the evolution of single-stranded RNA 17. T. R. D. Costa et al., Structure of the bacterial sex F pilus reveals an assembly (ssRNA) bacteriophages based on genomic sequences and characterizations of two of a stoichiometric protein-phospholipid complex. Cell 166,1436–1444.e10 – R-plasmid-dependent ssRNA phages, C-1, and Hgal1. J. Bacteriol. 194,5073 5079 (2016). (2012). 18. D. Arutyunov, L. S. Frost, F conjugation: Back to the beginning. Plasmid 70,18–32 3. D. E. Bradley, The structure and infective process of a Pseudomonas aeruginosa (2013). – containing ribonucleic acid. J. Gen. Microbiol. 45,83 96 (1966). 19. R. O’Callaghan, R. Bradley, W. Paranchych, The effect of M13 phage infection upon 4. C. Pourcel, C. Midoux, G. Vergnaud, L. Latino, A carrier state is established in Pseu- the F pili of E. coli. Virology 54, 220–229 (1973). domonas aeruginosa by phage LeviOr01, a newly isolated ssRNA levivirus. J. Gen. 20. D. E. Bradley, J. Whelan, Escherichia coli tolQ mutants are resistant to filamentous – Virol. 98, 2181 2189 (2017). bacteriophages that adsorb to the tips, not the shafts, of conjugative pili. J. Gen. 5. J. M. Schmidt, R. Y. Stanier, Isolation and characterization of bacteriophages active Microbiol. 135, 1857–1863 (1989). against stalked bacteria. J. Gen. Microbiol. 39,95–107 (1965). 21. D. E. Bradley, Shortening of Pseudomonas aeruginosa pili after RNA-phage adsorp- 6. J. Klovins, G. P. Overbeek, S. H. E. van den Worm, H. W. Ackermann, J. van Duin, tion. J. Gen. Microbiol. 72, 303–319 (1972). Nucleotide sequence of a ssRNA phage from acinetobacter: Kinship to coliphages. 22. C. K. Biebricher, E. M. Düker, Light-microscopic visualization of F and type 1 pili. J. Gen. Virol. 83, 1523–1533 (2002). J. Gen. Microbiol. 130, 941–949 (1984). 7. J. Rumnieks, K. Tars, Diversity of pili-specific bacteriophages: Genome sequence of 23. K. Daehnel, R. Harris, L. Maddera, P. Silverman, Fluorescence assays for F-pili and their IncM plasmid-dependent RNA phage M. BMC Microbiol. 12, 277 (2012). application. Microbiology (Reading) 151, 3541–3548 (2005). 8. T. D. Lawley, W. A. Klimke, M. J. Gubbins, L. S. Frost, F factor conjugation is a true type 24. L. Zeng et al., Decision making at a subcellular level determines the outcome of IV secretion system. FEMS Microbiol. Lett. 224,1–15 (2003). bacteriophage infection. Cell 141, 682–691 (2010). 9. M. Clarke, L. Maddera, R. L. Harris, P. M. Silverman, F-pili dynamics by live-cell im- 25. K. Zhang, R. Young, L. Zeng, Bacteriophage P1 does not show spatial preference aging. Proc. Natl. Acad. Sci. U.S.A. 105, 17978–17981 (2008). when infecting Escherichia coli. Virology 542,1–7 (2020). 10. P. G. Stockley, N. J. Stonehouse, K. Valegård, Molecular mechanism of RNA phage 26. D. Moore, B. A. Sowa, K. Ippen-Ihler, The effect of tra mutations on the synthesis of morphogenesis. Int. J. Biochem. 26, 1249–1260 (1994). – 11. R. I. Koning et al., Asymmetric cryo-EM reconstruction of phage MS2 reveals genome the F-pilin membrane polypeptide. Mol. Gen. Genet. 184, 260 264 (1981). – structure in situ. Nat. Commun. 7, 12524 (2016). 27. C. P. Novotny, P. Fives-Taylor, Retraction of F pili. J. Bacteriol. 117, 1306 1311 12. W. Paranchych, P. M. Krahn, R. D. Bradley, Stages in phage R17 infection. Virology 41, (1974). 465–473 (1970). 28. C. C. Brinton Jr., P. Gemski Jr., J. Carnahan, A new type of bacterial pilus genetically 13. R. C. Valentine, H. Wedel, The extracellular stages of RNA bacteriophage infection. controlled by the fertility factor of E. coli K 12 and its role in chromosome transfer. Biochem. Biophys. Res. Commun. 21, 106–112 (1965). Proc. Natl. Acad. Sci. U.S.A. 52, 776–783 (1964). 14. W. Paranchych, S. K. Ainsworth, A. J. Dick, P. M. Krahn, Stages in phage R17 infection. 29. C. C. Brinton Jr., H. Beer, “The interaction of male-specific bacteriophages with F Pili” V. Phage eclipse and the role of F pili. Virology 45, 615–628 (1971). in The Molecular Biology of , J. S. Colter, W. Paranchych, Eds., (Academic Press, 15. P. M. Krahn, R. J. O’Callaghan, W. Paranchych, Stages in phage R17 infection. 1967), pp. 251–289. VI. Injection of A protein and RNA into the host cell. Virology 47, 628–637 30. S. Lang et al., An activation domain of plasmid R1 TraI protein delineates stages of (1972). gene transfer initiation. Mol. Microbiol. 82, 1071–1085 (2011).

Harb et al. PNAS Latest Articles | 7of8 Downloaded by guest on September 29, 2021 31. N. Willetts, M. Achtman, Genetic analysis of transfer by the Escherichia coli sex 36. T. Shiba, T. Miyake, New type of infectious complex of E. coli RNA phage. Nature 254, factor F, using P1 transductional complementation. J. Bacteriol. 110,843–851 157–158 (1975). (1972). 37. J. Manchak, K. G. Anthony, L. S. Frost, Mutational analysis of F-pilin reveals domains 32. M. Achtman, N. Willetts, A. J. Clark, Beginning a genetic analysis of conjugational for pilus assembly, phage infection and DNA transfer. Mol. Microbiol. 43, 195–205 transfer determined by the F factor in Escherichia coli by isolation and characteriza- (2002). tion of transfer-deficient mutants. J. Bacteriol. 106, 529–538 (1971). 38. J. H. Strauss Jr., R. L. Sinsheimer, Purification and properties of bacteriophage MS2 33. R. E. Danziger, W. Paranchych, Stages in phage R17 infection. 3. Energy require- and of its ribonucleic acid. J. Mol. Biol. 7,43–54 (1963). ments for the F-pili mediated eclipse of viral infectivity. Virology 40, 554–564 39. T. Baba et al., Construction of Escherichia coli K-12 in-frame, single-gene knockout (1970). mutants: The keio collection. Mol. Syst. Biol. 2, 2006.0008 (2006). 34. J. E. Bidlack, P. M. Silverman, An active type IV secretion system encoded by the F 40. K. A. Datsenko, B. L. Wanner, One-step inactivation of chromosomal genes in Es- plasmid sensitizes Escherichia coli to bile salts. J. Bacteriol. 186, 5202–5209 cherichia coli K-12 using PCR products. Proc. Natl. Acad. Sci. U.S.A. 97, 6640–6645 (2004). (2000). 35. B. Hu, P. Khara, P. J. Christie, Structural bases for F plasmid conjugation and F 41. S. O. Skinner, L. A. Sepúlveda, H. Xu, I. Golding, Measuring mRNA copy number in pilusbiogenesisinEscherichiacoli.Proc. Natl. Acad. Sci. U.S.A. 116, 14222–14227 individual Escherichia coli cells using single-molecule fluorescent in situ hybridization. (2019). Nat. Protoc. 8, 1100–1113 (2013).

8of8 | www.pnas.org/cgi/doi/10.1073/pnas.2011901117 Harb et al. Downloaded by guest on September 29, 2021