Shear force enhances adhesion of Pseudomonas aeruginosa by counteracting pilus-driven surface departure

To characterize how fluid flow affects P. aeruginosa surface interactions, we performed single-cell imaging of bacteria colonizing surfaces. To generate fluid flow, we custom-fabricated microfluidic devices, which were then connected to a precisely controlled syringe pump (Fig. 1A). As surface arrival is the first step in bacterial surface colonization, we imaged and quantified wild-type (WT) P. aeruginosa PA14 cells arriving on a glass surface. For this experiment, microfluidic channels started with no cells and were imaged as cells were introduced into the channel in flow. Specifically, cells were loaded into a syringe at a mid-log concentration and were introduced at a shear rate (a measure of fluid flow) of 800 s⁻¹. As cells flowed into the channel, many cells arrived on the surface and remained attached (Fig. 1A). We quantified the cell arrival rate based on the number of cells that arrived and remained attached to the surface for at least 2 s (Fig. 1B). Over the 60-s experiment, a significant number of WT cells arrived and began to accumulate on the surface.

How do type IV pili influence P. aeruginosa surface arrival? As type IV pili are known to affect attachment in many bacteria including P. aeruginosa (23, 24, 37), we hypothesized that type IV pili would have an important role in surface arrival in our microfluidic assay. To test the role of type IV pili, we generated a ΔpilA mutant (which lacks pili) and a ΔpilTU mutant (which lacks typical pilus retraction) (Fig. 1C). To quantify the length of pili and number of pili per cell, we introduced a cysteine point mutation (T51C) in the major type IV pilus protein PilA. Then, we used a thiol-reactive Alexa488 maleimide dye to fluorescently label type IV pili (38, 39). Using this approach, we observed that WT cells make approximately 1 to 2 pili per 30 s that are typically 0.5 μm long and ΔpilTU cells make approximately 1 to 3 pili that are typically 0.8 μm long (SI Appendix, Fig. S1). Next, we measured surface arrivals in the ΔpilA and ΔpilTU strains and compared their surface arrival rate to wild type (Fig. 1B). The surface arrival rate of ΔpilA cells was approximately 50% lower than wild type, supporting our hypothesis that type IV pili are important for surface arrival. In contrast, the surface arrival rate of ΔpilTU cells was not statistically different than wild type. Together, these results suggest that the presence of type IV pili contribute to surface arrival.

To measure the effect of type IV pili on surface colonization, we flowed WT cells into empty microfluidic channels at a shear rate of 800 s⁻¹ over 30 min (Fig. 1D). Then, we imaged and quantified surface colonization, which we defined as the number of cells that accumulated on the surface. We observed that WT cells accumulated quickly for the first 15 min of the experiment, and then the total number of cells began to plateau (Fig. 1D). While the dynamics of early surface colonization are complex, we note that type IV pili and cyclic adenosine monophosphate signaling have been previously shown to affect surface engagement (40). To test the role of type IV pili on surface colonization, we repeated the experiment with ΔpilA and ΔpilTU strains. While ΔpilA cells colonized the surface approximately 50% less than wild type, ΔpilTU cells colonized the surface approximately 75% more than wild type.

To confirm that the ΔpilA and ΔpilTU mutations were linked to the surface colonization phenotypes, we reintroduced pilA and pilTU into the genomes of the respective mutants and repeated our experiment. Reintroduction of pilA or pilTU returned the mutant phenotypes to WT levels, indicating that mutations in pilA and pilTU were causative of the surface colonization phenotypes (SI Appendix, Figs. S2 and S3). As the pilA mutation only reduced surface colonization by 50%, we wondered whether the polysaccharide Pel was responsible for the residual 50% (41). To test the role of Pel on surface colonization, we generated ΔpelF and ΔpilA ΔpelF mutants and compared their surface colonization to their respective parental strains. We observed that ΔpelF mutations did not have mutant phenotypes, indicating that polysaccharide Pel was not responsible for surface colonization in our experiments (SI Appendix, Fig. S4). While the reduction in colonization observed in the ΔpilA mutant could be explained by the reduction in surface arrival rate (Fig. 1B), the mechanism underlying the enhancement of surface colonization in the ΔpilTU mutant remained a mystery.

As surface colonization is affected by surface arrival and departure, the enhanced colonization phenotype of the ΔpilTU mutant could potentially be explained by less surface departures. To test whether ΔpilTU cells depart the surface less than wild type, we forcibly removed cells from the surface using flow. We reasoned that if ΔpilTU cells experience less departures, they would be harder to remove from the surface with flow. Fluid flow exerts a shear stress on cells that is proportional to the shear rate and the solution viscosity (21) (Fig. 2A). The shear force that cells experience is proportional to the shear stress and their surface area (21) (Fig. 2A). To quantify the shear force required to remove WT cells from the surface, we measured cell departure over a period of 1 min while simultaneously applying various flow treatments. Specifically, we altered shear rate by changing the flow rate and altered solution viscosity by adding the viscous agent Ficoll to our media. While Ficoll has been used as a viscous agent in multiple biological studies (21, 22) and does not appear to have biological effects, we sought to be careful in our analysis by independently modulating shear rate and viscosity. At a shear rate of 4,000 s⁻¹ without Ficoll, approximately 25% of WT cells departed the surface (Fig. 2B). In contrast, at a shear rate of 40,000 s⁻¹ without Ficoll, approximately 95% of WT cells departed the surface (Fig. 2B). As confirmation that shear force drives surface departure, flow at a shear rate of 4,000 s⁻¹ with 15% Ficoll [which increases solution viscosity 10x (21)] removed approximately 90% of WT cells from the surface (Fig. 2B). Together, these experiments demonstrate that a shear force between 10 and 100 pN is required to remove WT cells from a glass surface.

To precisely test how flow removes cells from the surface, we subjected WT, ΔpilA, and ΔpilTU cells to a range of shear forces. In host environments such as the bloodstream, urinary tract, and lungs, P. aeruginosa cells typically experience shear forces in the range of 0.1 to 10 pN (43–45) (Fig. 2C). When subjected to shear forces in the host-relevant regime, WT, ΔpilA, and ΔpilTU cells all remained attached to the surface throughout our experiment (Fig. 2C). However, at higher shear forces, flow removed cells. Specifically, a shear force of 100 pN was required to remove the majority of WT and ΔpilTU cells and a shear force of 20 pN was required to remove the majority of ΔpilA cells. As we found no difference between WT and ΔpilTU cells, the mystery of how ΔpilTU cells colonized the surface to greater levels (Fig. 1D) remained unanswered. Our observation that very high shear forces were required to remove cells made us question whether forcing cells to leave the surface was mechanistically different than allowing cells to depart the surface on their own.

To examine how cells depart the surface on their own, we measured the residence times of cells in a host-relevant flow regime. We defined cell residence time as the time between surface arrival and surface departure (Fig. 3A). We quantified the residence times of cells exposed to a shear force of 2 pN as they naturally arrived and departed from the surface (Fig. 3A). During the experiment, 83% of WT cells had a residence time of less than 2 min and 6% had a residence time greater than 10 min (Fig. 3B). In contrast, ΔpilA cells had significantly longer residence times. Specifically, 48% of ΔpilA cells resided for less than 2 min, while 42% resided for greater than 10 min (Fig. 3B). The residence times of ΔpilTU cells were also significantly longer than WT cells, as 72% of ΔpilTU resided for less than 2 min and 22% of ΔpilTU cells resided for greater than 10 min (Fig. 3B). Together, our results support the conclusion that type IV pili promote surface departure in a host-relevant flow regime. The conclusion that type IV pili promote surface departure provides a rational basis for why ΔpilTU cells exhibit an increase in surface colonization (Fig. 1D). Furthermore, our observation that type IV pili promote surface departure in a host-relevant flow regime highlights the importance of studying bacteria in mechanically realistic environments.

How do type IV pili promote surface departure? A previous report (34) proposed that type IV pili tilt cells off a surface, which subsequently facilitates surface departure. That report focused on P. aeruginosa PAO1, a commonly studied isolate of P. aeruginosa. As our study is focused on P. aeruginosa PA14, the other most commonly studied isolate of P. aeruginosa, we explored whether surface orientation was altered in our experiments. While the previous study examined bacteria in flow cells, the shear forces used were not in the range of the shear forces in the bloodstream (44), urinary tract (43), and lungs (45). To examine how type IV pili promote surface departure in our experiments, we quantified the cell surface orientation of WT, ΔpilA, and ΔpilTU cells while being simultaneously subjected to host-relevant shear force. To simplify our analyses, we classified cells into three categories: horizontal, tilting, and vertical (Fig. 4A). By carefully measuring the observed length of individual cells in different orientations, we used trigonometry to estimate that we classified cells as tilting if their cell-surface angle was between 15 and 55 degrees (SI Appendix, Fig. S5). While WT cells were approximately evenly distributed between these three categories, ΔpilA and ΔpilTU cells were mostly horizontal (Fig. 4A). Highlighting that type IV pilus retraction promotes cell tilting, 36% of WT cells were classified as vertical, while 0% of both ΔpilA and ΔpilTU cells were vertical (Fig. 4A). Thus, our results reinforce previous observations (34) and provide new understanding of how type IV pilus retraction tilts cells away from the surface to promote departure in host-relevant flow environments.

During our experiments, we noticed that ΔpilA and ΔpilTU mutants exhibited different surface behaviors. For example, the residence times of ΔpilA cells were longer than the residence times of ΔpilTU cells (Fig. 3B). Additionally, the percentage of ΔpilA cells that exhibited a horizontal surface orientation was greater than observed with ΔpilTU cells (Fig. 4A). Based on these observations, we hypothesized that the presence of pili (independent of retraction) changes the behavior of cells on the surface. To test how type IV pili affect the physical interaction between cells and surfaces, we tracked individual cells in flow and measured their mean squared displacement (MSD). MSD represents a physical measurement of how individual cells move relative to their original position over time and the slope of MSD trajectories indicate the type of motion exhibited by cells. While MSD slope of 1 represents diffusive motion, a slope >1 is indicative of superdiffusive motion and a slope <1 is indicative of sub-diffusive motion. As WT cells are capable of twitching motility driven by type IV pilus retraction, we hypothesized that WT cells would have an MSD slope greater than 1. Our results supported this hypothesis, as WT cells had an MSD slope ~1.1 (Fig. 4B and SI Appendix, Fig. S6).

Based on differences we observed between ΔpilA and ΔpilTU cells, we hypothesized that ΔpilTU cells would have a higher MSD than ΔpilA cells. In support of that hypothesis, our experiment revealed that MSD slopes for ΔpilTU cells were significantly higher than for ΔpilA cells (Fig. 4B). We interpret this result to indicate that ΔpilTU cells are adhered to the surface via pili and thus have more freedom of motion than ΔpilA cells. This interpretation is consistent with our results that demonstrate ΔpilTU cells have slightly more and slightly longer pili than WT cells (SI Appendix, Fig. S1). Together, our results support the conclusions that the presence of pili (independent of retraction) increase the motion of cells on the surface and pilus retraction further increases surface motion.

How does shear force affect the surface residence time of P. aeruginosa? A previous study (36) reported the counter-intuitive observation that increasing flow intensity increases surface residence time of P. aeruginosa PA14. To confirm this observation, we quantified surface residence time of WT cells exposed to shear rates of 160 s⁻¹ or 1,600 s⁻¹ (SI Appendix, Fig. S7). In support of the previous study, we observed that only 6% of WT cells resided for greater than 10 min at 160 s⁻¹ (Fig. 5B and SI Appendix, Fig. S7). In contrast, when subjected to a shear rate of 1,600 s⁻¹, 37% of WT cells resided for greater than 10 min (SI Appendix, Fig. S7). The previous study suggested that the shear force associated with flow was responsible for the increase in surface residence time. As shear force is proportional to shear rate and solution viscosity, we explicitly tested the hypothesis that shear force increases residence time by using the viscous agent Ficoll. To ensure that our flow treatments were host-relevant and allowed for cells to adhere efficiently to the surface, we used shear forces of 0.4 pN and 4 pN. At these shear forces, there are a similar number of cells (Fig. 2C and SI Appendix, Fig. S8), which allowed for us to carefully examine the effect of shear force on residence time. When cells were exposed to a shear rate of 160 s⁻¹ without Ficoll [shear force ~ 0.4 pN (21)], 76% of WT cells resided for less than 2 min (Fig. 5B and SI Appendix, Fig. S7). However, at a shear rate of 160 s⁻¹ with 15% Ficoll [shear force ~ 4 pN (21)], only 15% of WT cells resided for less than 2 min (Fig. 5B). Collectively, our results explicitly demonstrate that increasing shear force generates the counter-intuitive outcome of enhancing surface adhesion.

While the mechanism by which shear force enhances adhesion of P. aeruginosa is unknown, there are multiple theoretical explanations for this phenomenon. For example, shear-enhanced adhesion of E. coli is mediated by the adhesin FimH, which uses a catch bond that resembles a finger trap. It is also possible that shear force could enhance adhesion by changing the geometric orientation of rod-shaped cells to increase surface contact. Based on our results, we favored this explanation and hypothesized that shear force enhances adhesion by promoting horizontal cell attachment. To test how flow intensity affects surface orientation, we quantified the cell surface orientation of WT cells exposed to shear rates of 160 s⁻¹ or 1,600 s⁻¹ (SI Appendix, Fig. S9). We observed that WT cells exposed to a shear rate of 160 s⁻¹ were approximately evenly distributed between our three orientation categories (Fig. 5C and SI Appendix, Fig. S9). However, the majority of WT cells exposed to a shear rate of 1,600 s⁻¹ were found in the tilting or horizontal orientation (SI Appendix, Fig. S9). To explicitly test how shear force affects surface orientation, we altered solution viscosity by using the viscous agent Ficoll. When exposed to a shear rate of 160 s⁻¹ with 15% Ficoll, we found that WT cells were predominantly in the tilting and horizontal orientations (Fig. 5C).

Based on our observations that shear force and type IV pilus retraction can affect cell surface orientation, we wanted to explore whether the cell-surface angle was dynamic. To test whether shear force can tip cells over, we imaged individual cells over time before and after flow was applied. Consistent with the hypothesis that flow can tip cells over, we observed that cells become more horizontal in the first 15 s after flow was applied (SI Appendix, Fig. S10). To test whether cells stand back up after flow stops, we imaged individual cells over time before and after flow was halted. Consistent with the hypothesis that cells can stand back up, we observed that cells become more vertical in the first 30 s after flow was halted (SI Appendix, Fig. S10). To provide a mathematical basis for these observations, we estimated the flow-generated forces experienced by cells in the vertical and horizontal orientations. Based on the fact that flow velocity increases as you move away from the channel wall, our calculations indicate that a vertical cell will experience approximately 3.75 times more force than a horizontal cell (SI Appendix, Fig. S11). Thus, flow should reorient cells from a vertical to horizontal as we observed in our microfluidic experiments. Together, our experimental results and calculations indicate that the force associated with flow tips cell over, promotes horizontal cell orientation, and decreases the frequency of surface departure.

To precisely examine how shear force affects cell surface association, we tracked individual cells exposed to different flow conditions. First, we aimed to test whether increasing shear rate would lower the MSD of cells, hypothesizing that more flow would result in firmer surface association. In support of our first hypothesis, cells exposed to a shear rate of 1,600 s⁻¹ had a lower MSD than cells exposed to shear rate of 160 s⁻¹ (SI Appendix, Figs. S9 and S12). Second, we aimed to test whether increasing solution viscosity would lower the MSD of cells, hypothesizing that more shear force would result in firmer surface association. In support of our second hypothesis, cells exposed to a shear force of 4 pN had a lower MSD than cells exposed to shear force of 0.4 pN (SI Appendix, Figs. S12 and S13). Together, these experiments provide evidence that shear force restricts the motion of cells on a surface, which likely promotes P. aeruginosa colonization in host-relevant flow regimes.

Our experiments reveal how shear force and type IV pili coordinate P. aeruginosa surface departure. Using a ΔpilA mutant, we showed that type IV pili have an important role in cell surface arrival and surface colonization in flow (Fig. 1). Additionally, we demonstrated that type IV pili have an important role in cell surface departure in host-relevant flow regimes (Fig. 3). Furthermore, we discovered that ΔpilTU cells (which lack pilus retraction) are better than wild type at colonizing a surface (Fig. 1), which can be explained by the increased residence time of ΔpilTU cells (Fig. 3). By independently modifying shear rate and viscosity, we determined that while very high shear forces promote surface departure (Fig. 2), host-relevant shear forces prevent surface departure (Fig. 5). Through precise cell tracking, we established that the cell surface orientation angle and cell MSD are associated with increased surface departure. Together, our results highlight how physical and biological factors combine to drive surface departure of the human pathogen P. aeruginosa.

In recent years, it has become clear that type IV pili impact bacterial surface behavior in a multitude of ways. Originally, type IV pili were recognized as important factors driving bacterial surface attachment (23, 24, 37). However, recent reports have highlighted how type IV pili also play major roles in surface sensing and the regulation of intracellular signaling, which can affect bacterial behavior over short and long timescales (38, 40, 46–48). Additionally, other recent reports have suggested that type IV pili also have an important role in bacterial surface departure (32–34). The data we present here support that idea that type IV pili drive both surface arrival and departure in flow. As type IV pili extend away from the cell body and make direct contact with the surface, it is intuitive to understand how they promote surface arrival. However, the mechanism by which type IV pili drive surface departure may be less obvious. Our results support the conclusion that type IV pilus retraction leads to the movement and tilting of cells off the surface, which increases the likelihood of their complete dissociation from the surface. As P. aeruginosa cells employ multiple pili at once, it is likely that the coordinated retraction of multiple pili is required to fully tilt cells off the surface. Conceptually, this process is similar to twitching motility, which can use multiple pili to “twitch” and generate “slingshot” motion on surfaces (49). Thus, future biophysical and cell biological investigation of how type IV pilus retraction controls surface interactions in complex environments is sure to uncover interesting and important mechanistic insight.

Intuitively, one would expect that shear force removes adherent cells from a surface. However, we show that shear force can enhance surface adhesion. Conceptually, there are multiple mechanisms that could allow for force-enhanced adhesion (7–15). For example, a surface protein could utilize a conformational change similar to a children’s toy known as a finger trap which shrinks the circumference of a tube when pulling force is applied. In fact, the FimH adhesin from E. coli uses a finger trap–like mechanism to strengthen its adhesive properties when subjected to shear force (11). Another mechanism that could allow for force-enhanced adhesion involves reorientation and establishment of new adhesive contacts (11). The data we have presented suggests that this mechanism underlies the force-enhanced adhesion of P. aeruginosa. Specifically, our data show that shear force reorients cells with respect to the surface (Fig. 5), creating a situation where new adhesive contacts can form. Our MSD data shows decreased surface motion when cells are reoriented by shear force (Figs. 4 and 5), supporting the interpretation that shear force enhances the association between cells and the surface. Based on findings presented here, it is clear that the human pathogen P. aeruginosa optimizes surface association in response to host-relevant shear force, which potentially explains its impressive ability to colonize a diverse array of host niches.

How much shear force do cells experience during infection? For an organism like P. aeruginosa which infects many sites of the human body, there is a wide range of shear forces it may experience. For example, the bloodstream typically generates shear forces ranging from 0.1 pN to 10 pN (44), the lung typically generates shear forces ranging from 1 pN to 5 pN (45), and the urinary tract typically generates shear forces ranging from 0.1 pN to 5 pN (43). As P. aeruginosa cells are likely to experience 0.1 to 10 pN of shear force during host colonization, it is logical to hypothesize that they have evolved mechanisms to optimize surface association when exposed to host-relevant shear forces. Our data strongly support this hypothesis, as WT cells adhere well when subjected to shear forces between 0.1 and 10 pN and are only removed from the surface when subjected to 100 pN (Fig. 2). Based off multiple pieces of evidence, our central conclusion is that P. aeruginosa optimizes surface adhesion to promote surface colonization in flow-rich host environments. In support of our central conclusion, orthogonal approaches (50) have recently generated a comprehensive mathematical model that explains how P. aeruginosa uses type IV pili to optimize the adhesion-migration trade-off during colonization. Thus, there is a growing body of evidence (50) that P. aeruginosa couples type IV pilus dynamics and cell geometry to optimize surface adhesion and cell migration in fluctuating mechanical environments.

We thank Anuradha Sharma, Alex Shuppara, Gilberto Padron, Nick Martin, Lisa Wiltbank, Vada Becker, Ben Bratton, Dan Kearns, Ankur Dalia, Paola Mera, Howard Stone, and Thomas Kehl-Fie for helpful discussions and comments on the manuscript. This work was supported by funding from the European Research Council through a starting grant for B.S. (BacForce, g.a. no. 852585). This work was also supported by start-up funds from the University of Illinois at Urbana-Champaign and grant K22AI151263 from the NIH to J.E.S.