E. coli cells overgrown by and embedded within V. cholerae clusters are shielded from phage exposure

V. cholerae N16961 (serogroup O1 El Tor) and E. coli AR3110 were engineered to constitutively produce the fluorescent proteins mKO-κ and mKate2, respectively, such that they could be distinguished by fluorescence microscopy. We note that the strain background of V. cholerae that we use here, N16961, does not antagonize E. coli via Type VI secretion activity in culture conditions used for this study, which are detailed below [60,61]. The 2 species were inoculated at a 2:1 ratio of V. cholerae and E. coli into microfluidic devices bonded to glass coverslips, allowed to attach to the glass surface for 45 min, and then incubated under continuous flow of M9 minimal medium with 0.5% glucose for 48 h. Within this time frame, biofilms begin to form; however, monoculture E. coli biofilms have not yet produced sufficient curli matrix to prevent phage entry. This time frame was established by prior work and confirmed in our own experiments described below [55,59]. T7 phages were then introduced to the system continuously at 10⁴ per μL for 16 h; this strain of T7 contains a reporter construct causing infected hosts to produce sfGFP prior to lysis [55]. Changes in E. coli abundance and localization in the chamber were tracked and compared to those in equivalent biofilms without phage introduction.

Prior to phage introduction, we noted considerable variation in biofilm structure and composition across the glass substrata of our flow devices. Depending on the initial surface distribution of V. cholerae and E. coli, different regions of the devices contained cell groups of E. coli mostly on its own, locally mixed with V. cholerae, or occasionally embedded in the bottom layers of highly packed, V. cholerae-dominated clusters. Shortly after phage introduction, most E. coli cells growing on their own quickly began reporting infection and then lysed (Fig 1A and S1 Movie). Over the next 16 h, E. coli cells embedded on the bottom layers of V. cholerae-dominated cell groups largely survived phage exposure, with scattered singleton E. coli cells elsewhere in the chambers. These single cells persisted for as long as we continued to track the system (up to 144 h) but did not appear to be actively replicating. After 16 h in the dual species biofilms, waves of T7 infection could be seen proceeding partially into groups of E. coli embedded within V. cholerae biofilms, but a fraction of E. coli most often survived (Fig 1A).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. E. coli embedded within V. cholerae cell groups can evade exposure to phages in the surrounding medium.

(A) Time-lapse series of a dual culture biofilm of E. coli (yellow) and V. cholerae (purple), undergoing T7 phage exposure (infected E. coli cells reporting in cyan/white). The biofilm was grown for 48 h prior to continuous phage introduction thereafter. Time points noted in the upper right of each panel represent time since phage introduction was started. (B) The neighborhood biovolume fraction (biovolume fraction within a 6 μm around each segmented bacterium) of the merged biovolumes of both V. cholerae and E. coli for the first time point in panel A. (C) Mean V. cholerae fluorescence signal found around E. coli cells in biofilms with and without phage exposure (Mann–Whitney U test with n = 9). (D) E. coli biovolume normalized to biovolume prior to the introduction of phage in dual culture with V. cholerae and monoculture controls (Mann–Whitney U test with n = 9, n = 3). (E) Total biovolume of E. coli in dual culture and monoculture control biofilms with and without phage exposure at equivalent time points (Mann–Whitney U tests with n = 8, n = 8, n = 6, n = 7 from left to right). The data underlying this figure can be found in S1 Data.

https://doi.org/10.1371/journal.pbio.3001913.g001

To determine if the remaining E. coli survived in dual species biofilms because of de novo evolution of resistance to T7, we performed runs of this experiment after which all E. coli cells in the chamber were dispersed by agitation and tested for T7 resistance (see Methods). The frequency of T7 resistance in the surviving E. coli population was 10⁻⁵, roughly the same as the frequency of resistance prior to the introduction of T7 phages [62]. This outcome shows that there was little or no substantive population compositional shift due to selection for de novo phage resistance (S1C Fig). This is not particularly surprising, as the host and phage population sizes—and, most importantly, the extent of movement and contact events between hosts and phages—are dramatically lower in these experiments compared to those that are typical in well-mixed batch culture. Rather, these experiments suggest that T7-susceptible E. coli survives phage introduction in our biofilm culture conditions by avoiding exposure to them entirely when embedded in groups of V. cholerae. We confirmed that this outcome is specific to the biofilm context by replicating the same experiment in shaken liquid culture beginning with the same cell inoculum and phages introduced at equivalent multiplicity of infection (see Methods). In liquid co-culture conditions with V. cholerae, T7-susceptible E. coli gained no protection against phage exposure and infection (S1D Fig). In the biofilm context, the delay between the start of biofilm growth and phage introduction was important for the experimental outcome; if phages were introduced from the beginning of biofilm growth, rather than 48 h after biofilm growth, then the extent of E. coli protection was all but eliminated (S2 Fig).

Our observations above suggested that in dual-species culture conditions, the majority of E. coli that survive phage introduction are the cell groups that have been overgrown and enveloped within the bottom layers of expanding, densely packed V. cholerae clusters. To test this idea quantitatively [16,63], we segmented and merged the cell volumes of E. coli and V. cholerae to calculate the joint neighborhood cell packing density for the 2 species throughout the imaged 3D space (Fig 1B). By visual inspection, regions in which E. coli survived contained a majority of V. cholerae and had relatively high cell packing (biovolume fraction >0.9), in comparison with other regions where E. coli tended to die of phage exposure and cell packing was lower (biovolume fraction = 0.3 to 0.6). We next measured the spatial association of V. cholerae with E. coli to see how this may change in the presence versus the absence of phage exposure. For this measurement, we segmented the E. coli population away from background, and then measured V. cholerae fluorescence in direct proximity within 2 μm of E. coli throughout all replicates with or without phages introduced. Compared to control experiments with no phages (Fig 1C), V. cholerae fluorescence was indeed significantly elevated in close proximity to E. coli after phage exposure, representing the surviving, protected portion of the E. coli population embedded within groups of V. cholerae. This protection effect could be replicated when introducing λ phages instead of T7 phages (S3 Fig), and in a parallel study, we show that the same effect occurs under predation by the bacterium B. bacteriovorus [64].

At the scale of the entire chamber community, E. coli showed higher survival rate in co-culture with V. cholerae than in monoculture on its own (Fig 1D). In absolute terms, the total population size of E. coli after phage exposure in co-culture with V. cholerae was not statistically different from the surviving population size after phage exposure in E. coli monoculture (Fig 1E). This result occurred because E. coli total abundance prior to phage introduction is lower in co-culture with V. cholerae, with which it is competing for space and nutrient resources, but due to embedding of many E. coli cells within V. cholerae clusters, their per-cell survival rate against phage exposure is substantially higher relative to a E. coli monoculture condition (Fig 1D). So, on short time scales after phage introduction (10 h), there is a significant increase in survival rate for E. coli growing in co-culture with V. cholerae, but not yet a significant difference in the absolute abundance of surviving E. coli relative to monoculture conditions. However, our experiments later in the paper demonstrate that on longer time scales (100+ h), the E. coli that survive phage exposure in co-culture within V. cholerae colonies maintain positive net growth and recover from the initial population decline, whereas the surviving E. coli cells from monoculture biofilms do not recover. Before elaborating on this point with longer time scale experiments detailed below, we first turn to the biofilm architectural mechanisms in co-culture biofilms of E. coli and V. cholerae that are responsible for the observations reported in Fig 1.

Protection within V. cholerae cell clusters depends on their packing structure

After demonstrating that E. coli cells have reduced exposure to phages when embedded in clusters of V. cholerae, we explored the biofilm architectural features needed for protection to occur. As we have found previously that the extent of V. cholerae cell–cell packing can influence transport of phages and bacteria through biofilms, our first hypothesis based on prior work was that the high-density cell packing of V. cholerae biofilms was important for this protection mechanism [11,16,41]. Our other hypothesis, not mutually exclusive, was that phages may be sequestered away from E. coli by irreversible attachment to the surface of V. cholerae cells in close proximity. To distinguish between these mechanisms, or to estimate their relative contribution to E. coli protection within V. cholerae clusters, we performed new experiments manipulating V. cholerae cell packing in co-culture with E. coli and assessing the degree of attachment and neutralization of T7 phages on the surface of V. cholerae.

To alter V. cholerae cell packing structure, we performed co-culture experiments similar to those in the previous section, but using a strain of V. cholerae (denoted ΔrbmA) with a clean deletion of the rbmA locus. This strain cannot produce the matrix protein RbmA, which is not essential for biofilm formation but is necessary for the tight cell packing that is characteristic of mature V. cholerae biofilms (S4 Fig) [44,47,65]. Biofilms without RbmA, in contrast with those of wild type (WT), can be invaded into their interior by planktonic competitor bacteria as well as predatory bacteria such as B. bacteriovorus [11,16]. If the high cell packing to which RbmA contributes is important to the protection of E. coli from phage exposure, we expect that in co-culture biofilms with V. cholerae ΔrbmA, E. coli will be more exposed to T7 phage predation and show different population dynamics relative to control co-cultures with WT V. cholerae.

We grew E. coli and V. cholerae ΔrbmA in biofilm co-culture, introduced T7 phages after 48 h as above, and found that the E. coli grown in the presence of V. cholerae ΔrbmA does not exhibit population recovery after phage introduction as it does in co-culture with V. cholerae WT (Fig 2A). This outcome suggests that E. coli does not gain protection from phage exposure amidst V. cholerae ΔrbmA, and that the cell packing architecture of V. cholerae WT is in fact important for this protection effect. If E. coli is protected within V. cholerae WT clusters, but not within ΔrbmA clusters, then in a triculture experiment of E. coli, V. cholerae ΔrbmA, and V. cholerae WT, we expect a statistical shift of E. coli spatial association toward WT V. cholerae after introducing phages as E. coli associated with ΔrbmA V. cholerae are more often killed. We performed this triculture experiment, measuring the average distance between E. coli and V. cholerae WT, and that between E. coli and V. cholerae ΔrbmA, before and after phage introduction. Without the addition of phages into the triculture condition, E. coli cells are just as likely to be associated with WT V. cholerae (median distance: 0.88 μm) as they are with ΔrbmA V. cholerae (median distance: 0.69 μm) (Fig 2B and 2C). When phages are introduced, the remaining E. coli were significantly closer to WT V. cholerae (median distance: 0.59 μm) than they were to ΔrbmA V. cholerae (median distance: 3.46 μm) (Fig 2B).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. E. coli evasion of phages within V. cholerae biofilms depends on the high cell–cell packing produced by WT V. cholerae.

(A) E. coli biovolume over time in dual culture conditions with either V. cholerae WT or V. cholerae ΔrbmA (n = 7, n = 8, n = 3–6, n = 3–8 from top to bottom in legend). (B) Average distance between E. coli cells and either V. cholerae WT or ΔrbmA in a triculture condition with or without phage exposure (Wilcoxon paired comparison tests with n = 9). (C, D) Representative images from the triculture condition with E. coli (yellow), V. cholerae WT (purple), and V. cholerae ΔrbmA (cyan) (C) without phage exposure and (D) after phage exposure. (E) PFU recovered after incubation of starting T7 phage inoculum with either no bacteria, V. cholerae, E. coli WT, or E. coli ΔtrxA over a 60-min time course. E. coli ΔtrxA allows for T7 phage attachment and genome ejection, but not for phage replication. Each trajectory shows the data for 1 run of each treatment (n = 3 for each treatment, giving 3 traces per treatment). (F) The neighborhood biovolume fraction of the merged biovolumes of both V. cholerae genotypes and E. coli from panel (D). The data underlying this figure can be found in S1 Data. PFU, plaque-forming unit; WT, wild type.

https://doi.org/10.1371/journal.pbio.3001913.g002

The experiments above indicate that the packing architecture of V. cholerae WT biofilms is important for phage exposure protection of E. coli within them, as E. coli gains little if any protection from phage exposure in proximity to loosely packed V. cholerae ΔrbmA. These data do not exclude the possibility that this difference is due in part to sequestration of phages by attachment to V. cholerae cells, which could occur more often in WT clusters with higher density of available V. cholerae cell surface relative to clusters of the ΔrbmA strain. To help assess whether sequestration of phages by direct attachment to V. cholerae cell surface was important, we incubated V. cholerae, E. coli, and ΔtrxA E. coli with T7 phages in shaken liquid culture, tracking the ability to recover T7 phages every 5 min for 1 h (Fig 2E). In addition to a blank media control, the ΔtrxA E. coli strain was included because this strain can adsorb phages normally but undergoes abortive infection, preventing phage amplification [66]. As expected, with E. coli ΔtrxA incubation, T7 PFU recovery steadily decreased until saturation at 1 h. Incubated with T7-sensitive E. coli WT, T7 PFU recovery initially decreased as phage adsorption occurred, followed by a rapid increase as new phages were released. Another round of latency and amplification then occurred before the 1-h stop time. Incubated with V. cholerae, no change in T7 PFU recovery was observed, which was identical to the blank media control for the duration of the experiment. These data suggest that T7 phages are not sequestered by adsorption to the V. cholerae cell surface.

Though T7 phages do not appear to adhere to V. cholerae cell surface in planktonic culture, within biofilm cell clusters V. cholerae surface properties may differ, and they are also embedded in matrix polysaccharide and protein components. With this in mind, we also performed experiments with fluorescently labeled T7 phages in biofilm growth conditions to determine if T7 is sequestered to V. cholerae cell groups in this context. Labeled phages were introduced to V. cholerae and E. coli dual culture biofilms and tracked over time, and we found that they localize strongly to unprotected E. coli cells and not to V. cholerae (S5 Fig). When V. cholerae monoculture biofilms were grown in flow devices with labeled phages added continuously in the media, we saw no accumulation of phages along the outer surface of V. cholerae cell clusters (S6 Fig). We found occasional phages within V. cholerae cell groups along the basal glass substrata, but not in the rest of their interior volume (S6 Fig). As phages were added from the beginning of biofilm growth onward in this experiment, the results suggest that V. cholerae biofilm colonies expanded over the top of initially glass-attached phages, rather than phages diffusing through biofilms to the basal layer.

Taken all together, the data in the experiments above suggest that E. coli is unexposed to phages within WT V. cholerae biofilms due to their architectural features, with minimal if any sequestration of phages by direct adsorption to the surface of V. cholerae cells.

Cohabitation with V. cholerae alters E. coli matrix production

As noted in the first Results section, E. coli accumulates less quickly in co-culture with V. cholerae than it does on its own, owing to competition for limited space and resources. Previous work has shown that, in monoculture, E. coli biofilms can protect themselves against phages once they begin to produce curli matrix proteins, which interrupt phage binding on the single-cell scale and contribute to biofilm architecture that blocks phage diffusion on the collective cell scale [55]. Curli production does not usually start until several days after beginning E. coli biofilm growth in microfluidic culture conditions [55], and we wondered if growing together with V. cholerae in dual culture might delay or disrupt curli formation. We note again that in the experiments in previous sections, biofilms were cultivated for too short a time for E. coli to begin producing curli matrix even in monoculture conditions. Here, we explored whether co-culture with V. cholerae impacts curli production on longer time scales, when E. coli on its own would ordinarily be able to protect itself against phage exposure via curli production.

If curli production is reduced or disrupted by growth with V. cholerae as a competitor, we would expect no difference in phage exposure survival between E. coli WT and a strain lacking curli matrix in co-culture with V. cholerae. To explore this possibility, E. coli WT and an isogenic curli null deletion strain (denoted ΔcsgA) were grown either on their own or in co-culture with V. cholerae for 96 h. This cultivation period is twice as long as is normally required for monoculture E. coli WT biofilms to produce curli and block phage diffusion. Biofilms were imaged at 96 h, exposed to phages at 10⁴ per μL under 0.1 μL/min flow for 16 h, and then imaged again to document population sizes of WT and ΔcsgA before and after phage introduction. As expected, the E. coli WT monoculture biofilms had the highest level of survival, with some replicates showing net increases in population size after the 16-h phage treatment. E. coli ΔcsgA monoculture biofilms, lacking any protection mechanism against phage exposure, had the lowest level of survival. In contrast, when in co-culture with V. cholerae, E. coli WT and ΔcsgA (Fig 3A) showed no substantial difference in survival to phage exposure (pairwise test not significant with Bonferroni correction), suggesting that curli production is no longer necessary for T7 exposure protection for WT E. coli in this context.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. E. coli biofilms’ normal production of curli matrix protein is interrupted in co-culture with V. cholerae to the extent that phage protection is no longer provided by E. coli biofilm matrix.

(A) E. coli biovolume normalized to biovolume prior to phage introduction in dual culture and monoculture conditions for both E. coli WT and E. coli ΔcsgA (Mann–Whitney U tests with n = 7, n = 12). (B) Total E. coli biovolume with and without phage treatments at equivalent time points (Mann–Whitney U tests with n = 12). In these experiments, in contrast with Fig 1E, biofilms were grown for longer periods before phage addition such that E. coli WT on its own could produce protective curli matrix prior to phage addition. (C) Frequency distribution of csgBAC transcriptional reporter fluorescence around E. coli in monoculture and dual culture conditions. (D) Frequency distribution of curli immunofluorescence intensity in proximity to E. coli in monoculture and dual culture conditions. (E) Dual culture conditions of E. coli (yellow) and V. cholerae (purple) before phage exposure (top) and after 16 h of continuous phage exposure (bottom). (F) Monoculture conditions of E. coli before phage exposure (top) and after phage exposure (bottom). The data underlying this figure can be found in S1 Data.

https://doi.org/10.1371/journal.pbio.3001913.g003

To assess why curli-based phage protection was no longer operating for E. coli even in co-culture biofilms that had grown over 96 h, we repeated the experiments above with an E. coli WT strain harboring reporter fusions for monitoring csgBAC transcription and curli protein production. The transcriptional reporter was made previously by introducing mKate2 in single copy on the chromosome within the csgBAC operon encoding 2 subunits of curli fiber protein (CsgB baseplate and CsgA primary curli monomer) and CsgC, which inhibits improper aggregation of CsgA monomers [55]. The protein production reporter was also made previously by introducing a 6x-His fusion tag to csgA, which allowed for in situ immunostaining of curli fibers produced by E. coli during growth in monoculture and co-culture with V. cholerae. As noted previously, the total population size of E. coli in biofilms with V. cholerae is lower than that found in monoculture (Fig 3B and 3E and 3F). On a per cell basis over the entire chambers, csgBAC transcription and curli immunostaining were significantly higher for E. coli growing alone versus E. coli growing in co-culture with V. cholerae (Fig 3C and 3D). These patterns manifested at the scale of the whole chamber; on a smaller spatial scale, E. coli distance from V. cholerae in co-culture was not correlated with curli production (S7 Fig).

Overall, these results suggest that E. coli curli production is substantially reduced when growing together with V. cholerae. It is not clear exactly why this is the case, but we speculate here that co-culture with V. cholerae alters one or a combination of nutrient availability, microenvironment osmolarity, and envelope stress experienced by E. coli, all of which influence the regulation of curli production [52]. The reduction in curli production may in turn contribute to the loss of curli-based protection against T7 phages even after long incubation periods over which E. coli normally develops curli-based phage protection on its own in monoculture. Together with the previous section, our experiments here also indicate that while E. coli has a lower ability to protect itself via curli matrix production when in co-culture, it can avoid phage exposure altogether when it been overgrown by and embedded within V. cholerae colonies.

Ecological consequences of joint interspecific competition and phage exposure

Our results thus far suggest complex ecological dynamics in which E. coli suffers a fitness reduction in spatial competition with V. cholerae, but on the other hand, E. coli gains a protective fitness benefit against phage exposure when embedded in the highly packed biofilm cell clusters that V. cholerae produces. It is still not clear, though, whether this protection is lasting under prolonged phage exposure, or whether E. coli remains viable within V. cholerae clusters despite being packed into their bottom-most cell layers. To characterize these population dynamics more thoroughly, we performed new experiments in which E. coli and V. cholerae were inoculated alone or together and grown for 48 h, followed by either continuous phage exposure or no phage exposure for an additional 96 h (Fig 4). Note that this experimental regime is such that even in monoculture, E. coli will not have produced sufficient curli to block phage diffusion at the onset of phage influx into the biofilm chambers [55].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. Population dynamics of E. coli (yellow) and V. cholerae (purple) in monoculture and dual culture conditions, where biofilms grew for 48 h prior to phage exposure, and phage exposure was applied continuously for 96 h thereafter.

(A, B) Representative images from time course imaging of (A) E. coli monoculture and (B) co-culture with V. cholerae. (C, D) Magnification of E. coli phage infection (reporting in cyan/white) within a cluster embedded in a larger colony of V. cholerae at (C) 96 h and (D) 120 h. Expanded fields of view in (C) and (D) are denoted by checked white boxes in panel B. (E) E. coli population dynamics in monoculture and in co-culture with V. cholerae, with and without T7 phage exposure from 48 h onward (n = 7, n = 8, n = 6–7, n = 6–8 from top to bottom in the legend). Note that the data in the gray circles and squares through 120 h are repeated from the gray data in Fig 2A. (F) V. cholerae population dynamics in monoculture and in co-culture with E. coli, with and without T7 phage exposure from 48 h onward (n = 4, n = 4, n = 4–8, n = 4–8 from top to bottom in the legend). The data underlying this figure can be found in S1 Data.

https://doi.org/10.1371/journal.pbio.3001913.g004

The population dynamics of E. coli and V. cholerae without T7 phage addition confirm our earlier suggestion that this interaction is competitive by default; E. coli population size is reduced in co-culture relative to when growing on its own (Fig 4E; yellow versus gray square trajectories). V. cholerae total productivity is also reduced in co-culture with E. coli relative to when growing on its own (Fig 4F), though overall it outcompetes E. coli by a substantial margin (S8 Fig). This result was driven by V. cholerae biofilm clusters expanding more rapidly and robustly in lateral and vertical space, displacing some neighboring E. coli and overgrowing other E. coli cell groups along the glass surface (Fig 4B). Under prolonged phage exposure, however, these same enveloped clusters remain mostly protected from phage killing. We did observe occasional E. coli deaths within trapped clusters, shown via the T7 infection reporter (Fig 4C and 4D), but overall, the E. coli cell groups maintained positive net growth and expanded laterally as the overlaid V. cholerae biofilms expanded as well (Fig 4E). Based on our earlier experiments, we suspect that progeny phages released from these occasional infection events within protected clusters were mostly trapped in place, and a sufficient impedance to phage diffusion allows for long-term survival of phage-susceptible hosts in close proximity [55,67–69]. Though we have linked V. cholerae cell packing to this phage diffusion limitation, the exact biophysical explanation for limited phage diffusion is an important future question. We speculate here that high density packing of V. cholerae, combined with the biochemical properties of its matrix and sequestration of phages to trapped debris from lysed E. coli, all contribute to the strongly impeded diffusion of T7 phages from initial sites of infection and amplification.

Given that monoculture E. coli biofilms have a lower cell packing density than V. cholerae biofilms (S9 Fig), and that the inclusions of E. coli within the V. cholerae biofilms continue expanding through time, we were curious to see if E. coli cell groups trapped within V. cholerae biofilms interrupted their highly packed structure. We assessed this question by calculating their 2 species’ joint neighborhood cell packing, finding it to be stable over time and indistinguishable from what V. cholerae produces on its own. This suggests that the V. cholerae biofilm architecture, once it has been initiated by V. cholerae cells growing together, can drive cell groups of other species trapped within them into high packing orientations that do not disrupt the overall structure (S9 Fig) [40,43,70,71]. As we explored in a parallel study on B. bacteriovorus predation in dual species biofilms, co-culture with E. coli only disrupts V. cholerae architecture when cells of both species begin dividing directly adjacent to each other from the outset of biofilm growth [64].

From an ecological point of view, the net result of these architectural details is that if phage exposure occurs before E. coli is able to produce protective curli matrix, E. coli has higher absolute fitness in co-culture with V. cholerae—with which it is otherwise competing—than it does on its own (Fig 4E, yellow versus gray circles). The same process by which E. coli is overgrown and enveloped by expanding V. cholerae biofilms, which in the absence of phages reduces E. coli population growth relative to monoculture, protects E. coli from near total population collapse when phages are present. Since V. cholerae still has somewhat reduced absolute fitness in co-culture with E. coli compared to monoculture (regardless of phage addition: Figs 4F and S8), this interaction can be characterized as E. coli parasitizing or exploiting V. cholerae biofilm structure and gaining some protection at their expense in the presence of E. coli-targeting phages.

Strains

All V. cholerae strains used in this study are strain N16961 (serogroup O1 El Tor) and derivatives (Table 1). The fluorescent protein expression construct insertions and ΔrbmA deletion mutants were made here and previously using standard allelic exchange [41,85]. E. coli strains are all AR3110 and its derivatives. AR3110 was derived from the K-12 strain W3110, for which cellulose production is disrupted by a polar stop codon mutation in bcsQ. This stop codon was corrected in AR3110 to yield a strain that produces the full complement of E. coli biofilm matrix components including cellulose and curli protein [48]. Like other K-12 derivatives, the E. coli AR3110 parental strain lacks O-antigen and is susceptible to T7 phages. AR3110 derivatives were produced via lambda red recombination or through allelic exchange. Briefly, primers encoding regions of homology to the host genome were used to amplify fluorescent protein expression constructs fused by SOE PCR to Kan^R or Cm^R resistance cassettes. These PCR products were used to knock in the fluorescent marker and selected for with the respective resistance cassette [55,59]. Recombinant T7 phages were created previously using T7select415-1 phage display system [55]. Recombinant λ phages were generously provided by Lanying Zeng’s group, which created them by infecting λDam cI₈₅₇ bor::KanR phages on LE392 (permissive host) with plasmid pBR322-λD-mTurquoise2/mNeongreen-E for recombination, and further selection for fluorescent plaques [86].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Bacterial strains and reagents used in this study.

https://doi.org/10.1371/journal.pbio.3001913.t001

Microfluidic flow device fabrication

Microfluidic chambers are produced by casting poly-dimethysiloxane (PDMS; Dow Chemical Company, SYLGARD 184, cat. # 04019862) onto preexisting chamber molds (a diagram of the chamber design used in this study is provided in S10 Fig). The resulting PDMS blocks were cut to size, hole-punched for inlet and outlet channels, and then bonded to #1.5 glass coverslips using plasma cleaning preparation of the PDMS and glass coverslips (Azer Scientific, cat. # 1152260). Between the inlet and outlet port areas, the internal space of the chambers in which biofilms were cultivated measured 5,000 μm × 500 μm × 70 μm (LxWxH). Segments of inlet tubing (Cole Parmer PTFE #30, cat. # 06417–11) attached to 27Gx1/2 needles (BD Precision, cat. # 305109) on 1 mL syringes (Brandzig, cat. #CMD2583) were plumbed into chamber inlets, and the syringes were driven by Harvard Apparatus Pico Plus Elite syringe pumps (Harvard Apparatus, cat. # 70–4506). Tubing from chamber outlet channels was fed to effluent collection dishes.

Biofilm culture conditions

Overnight cultures of V. cholerae and E. coli were inoculated into the microfluidic chambers at a ratio of 2:1. To achieve comparable amounts of biofilm growth in the ΔrbmA V. cholerae, WT V. cholerae, and E. coli triculture experiments, chambers were inoculated with a ratio of 6:3:1, respectively. After a 45-min incubation period without flow to allow for surface attachment, M9 minimal media with 0.5% glucose continuously flowed into the chamber at a rate of 0.1 μL/min. For experiments with longer term incubation to promote curli production by E. coli prior to phage introduction, and to discourage biofilm overgrowth of the chamber, chambers were incubated with M9 minimal media with 0.25% glucose.

For the immunostaining of curli, a new AR3110 strain harboring a translational 6xHis tag fused to csgA, which encodes the monomer for curli production, was stained with Anti-6X His Epitope Tag (Rabbit) antibody conjugated to Dylight 405 (Rockland Immunochemicals, cat. # 600-446-382) added to the media at a concentration of 0.1 μg/mL for the entirety of the experiment. Prior work has shown that addition of the 6xHis tag to CsgA does not interrupt its function by any measures tested [55]. For experiments investigating the effects of phage exposure, pairwise comparisons were made between flow devices of equal age that had phages added or did not have phages added, as the control for these experiments was the dynamics of co-culture growth without the presence of phage. For experiments investigating the disruption of curli matrix production, comparisons were made between flow devices before they experience phage exposure and after they experience phage exposure, as the control for these experiments were E. coli monoculture biofilms surviving exposure from phages. All experiments were carried out at room temperature.

Phage propagation, staining, and introduction to biofilms

T7 phages were produced by growing sensitive E. coli to OD₆₀₀ = 0.4 in M9 minimal media with 0.5% glucose, before adding an aliquot of T7 phage and incubating until the bacterial cultures were cleared. Phages were quantified using standard plaquing techniques and back diluted to 10⁴/μL in M9 with 0.5% glucose. To visualize phage infection, we used a previously constructed T7 strain that induces sfGFP production by the host prior to lysis; to visualize phages directly, we stained T7 phages with Alexa Fluor 633 NHS Ester (Thermo Fisher Scientific, cat. # A20005) using the Phage on Tap protocol [87]. For experiments in which capsid-labeled phages were introduced to biofilms, the biofilms were grown for 48 h as described above, followed by continuous labeled phage addition at 5 × 10⁶/μL for the remainder of the experiment. λ phages were produced by growing lysogenic E. coli to an OD₆₀₀ = 0.2 in M9 minimal media with 0.5% glucose, then heat shocked at 42°C for 20 min, and then incubated at 37°C until visible lysis occurred. For experiments with short phage exposure, phages were continuously introduced for 16 h. For experiments with extended phage exposure (Fig 4), phages were continuously added for 96 h. For the nascent biofilm phage exposures, phages were introduced into the chamber immediately following the initial attachment step. For the labeled phage time lapse, phages were added for a total of 20 h.

Biofilm dispersal and detection of de novo T7-resistance mutants from biofilm culture

V. cholerae and E. coli dual culture biofilms grown for 48 h and then treated with phages for 16 h were dispersed by removing the tubing from the microfluidic device and vigorously pipetting 100 μL of M9 media and air bubbles back and forth between the inlet and outlet ports. This was done to ensure maximal removal of all cells in the chamber in order to capture accurate measurements of total E. coli and de novo T7-resistant E. coli. To determine cell viability and phage sensitivity, this 100 μL volume containing dispersed biofilm cells was serially diluted and plated on LB+50 μg/mL kanamycin plates for total E. coli counts and, in parallel, on LB+50 μg/mL kanamycin plates saturated with T7 phages to determine de novo T7-resistant counts. Kanamycin was added to selectively plate for E. coli and kill V. cholerae, as all E. coli strains carried a kanamycin resistance cassette on the chromosome from insertion of their constitutive fluorescent protein production constructs [55].

Dual liquid culture phage assay

Approximately 5 mL liquid dual cultures of room temperature M9 minimal media with 0.5% glucose were inoculated 2:1 with V. cholerae and E. coli each normalized to OD₆₀₀ = 0.15. A total of 5 μL samples of these dual cultures were taken at regular time intervals, serially diluted, and plated onto LB+50 μg/mL kanamycin plates to obtain the CFU count for E. coli. At an OD₆₀₀ measurement of 1.5 (15 h), 6.5 × 10⁷ T7 phages were introduced to the dual cultures, as this is the estimated MOI for phages introduced in the biofilm condition. Note that MOI (multiplicity of infection) estimation is less straightforward in biofilm culture, as many phages introduced by flow do not contact host cells and pass out of the chambers in the liquid effluent.

Phage adsorption assay

Bacterial cultures of E. coli, E. coli ΔtrxA, and V. cholerae were grown and back-diluted to an OD₆₀₀ = 0.5 in LB medium. T7 phages were added to a final concentration of 5 × 10⁴ phages per μL. Cultures were incubated at 37°C on an orbital shaking platform, and 500 μL aliquots were taken every 5 min, passed through a 0.2-μm filter, and stored on ice until the end of the experiment. The filtration step served to exclude any bacterial cells, and any phages that were attached to them, allowing us to measure free phages remaining in the liquid medium. Flow-through samples were then serially diluted and plated for PFUs.

Microscopy and image analysis

All imaging was performed using a Zeiss 880 line-scanning confocal microscope, using a 40x/1.2 N.A. water objective or a 10x/.4 N.A. water objective. The 6xHis Tag Antibody Dylight 405 that was used to stain 6xHis-tagged curli polymers was excited with a 405 laser line. The mTFP protein that V. cholerae produces in experiments investigating curli transcription was excited with the 458 laser line. The sfGFP protein produced by the T7 infection reporter construct and the mNeonGreen capsid label of the λ phages were both excited (in separate experiments) with a 488 laser line. The mKO-κ protein that V. cholerae expresses constitutively and by E. coli that reports csgBAC transcription was excited with a 543 laser line (in separate experiments). The mKate2 protein that E. coli expresses constitutively and upon csgBAC transcription, and the mRuby2 protein that ΔcsgA E. coli constitutively expresses was excited with a 594 laser line (in separate experiments). The Alexa Fluor 633 conjugated onto the capsid of labeled T7 phage virions was excited with a 633 laser line. For each chamber in each experiment, multiple independent locations were chosen within each biofilm chamber and averaged to give 1 measurement for a given chamber in the case of whole-biofilm measurements. Prior to export, images were processed by constrained iterative deconvolution in ZEN blue.

Replication, quantification, and statistics

Replication is reported for each experiment individually in the legends of all of the figures. The reported n for each figure panel refers to biological replicates. One biological replicate was defined as the averaged outcome for measurements across a single microfluidic flow chamber inoculated from independent overnight culture preparations. Biological replicates for the core biofilm microscopy experiments in the study were performed across 2 to 3 weeks with independent microfluidic chambers. Technical replicates were separate z-stacks captured at randomized locations throughout a given flow chamber; measurements from these technical replicates were averaged to calculate the value for the biological replicate corresponding to that flow chamber. All biofilm image quantification was performed within the BiofilmQ framework [63]. For 3D grid-based measurements detailing microscale architecture, segmented microbial volumes were divided into a 3D grid with each node 0.8 μm on a side. Joint neighborhood cell packing measurements merged the biovolume of all bacteria within a sample and calculated the local biovolume fraction within 6 μm of each segmented bacterial volume within each grid cube [16]. For experiments using λ phages, infection was measured by calculating a Mander’s overlap coefficient between E. coli cells and λ phages. Mann–Whitney U tests with the Bonferroni correction were used for pairwise comparisons. We chose nonparametric comparison tests because they are relatively conservative and because the assumptions required for parametric tests could not consistently be assessed for our data.