ph taxis microsystems

7/25/2019 PH Taxis Microsystems

1/131SCIENTIFICREPORTS| 5:11403 | DOI: 10.1038/srep11403

www.nature.com/scientificreports

pH-Taxis of Biohybrid

MicrosystemsJiang Zhuang1, Rika Wright Carlsen1,2& Metin Sitti1,3

The last decade has seen an increasing number of studies developing bacteria and other cell-

integrated biohybrid microsystems. However, the highly stochastic motion of these microsystems

severely limits their potential use. Here, we present a method that exploits the pH sensing of

agellated bacteria to realize robust drift control of multi-bacteria propelled microrobots. Under

three specically congured pH gradients, we demonstrate that the microrobots exhibit both

unidirectional and bidirectional pH-tactic behaviors, which are also observed in free-swimming

bacteria. From trajectory analysis, we nd that the swimming direction and speed biases are two

major factors that contribute to their tactic drift motion. The motion analysis of microrobots also

sheds light on the propulsion dynamics of the agellated bacteria as bioactuators. It is expected that

similar driving mechanisms are shared among pH-taxis, chemotaxis, and thermotaxis. By identifying

the mechanism that drives the tactic behavior of bacteria-propelled microsystems, this study opens

up an avenue towards improving the control of biohybrid microsystems. Furthermore, assuming

that it is possible to tune the preferred pH of bioactuators by genetic engineering, these biohybrid

microsystems could potentially be applied to sense the pH gradient induced by cancerous cells in

stagnant uids inside human body and realize targeted drug delivery.

Biohybrid microsystems, which integrate motile microoganisms or cells with engineered unctional syn-thetic materials, have been heavily studied recently because o their potential applications in medicine,bioengineering, and environmental monitoring1. Tere is a particular interest in employing flagellatedbacteria as onboard actuators and sensors in biohybrid systems due to their high motility, strong viability,versatile sensing abilities, and ease o genetic modification212. o utilize the swimming locomotion othese systems in applications such as targeted drug delivery and therapeutics, it is necessary to developand implement reliable control methods. Using various physical steering methods, the control o severaltypes o microrobotic systems at the single agent level has been demonstrated 1315. However, control othese microsystems at the swarm level has been challenging, partly because o the high inherent sto-chasticity o such systems. Te large variability in the motion o biohybrid microsystems results romactors such as the randomness in the assembly process and the inherently stochastic and heterogeneousbehavior o the assembled bioactuators. Some methods have been developed to improve the controlo these microsystems. Koumakis et al.devised an anisotropic substrate to realize targeted delivery obacteria-propelled colloids16, but the method required the use o sophisticated substrate patterns, whichlargely limits its applicability. Varying degrees o tactic behavior has also been reported in studies thatrely on bacterial chemotaxis as a means to control bacteria-propelled microsystems1720. However, thetactic motion o the microrobotic systems has yet to be extensively quantified, and the driving mecha-nism behind the behavior remains unclear to date.

Maintaining an appropriate pH level is vital to the survival o most microoganisms like bacteria, andthey have evolved various sensing and regulatory strategies to adjust their cytoplasmic pH21,22. Flagellatedbacteria such as E. colihave also been ound to exhibit bidirectional pH-tactic behavior2325, i.e., move-ment away rom both strong acidic and alkaline pH environments. Given the pH tactic response o these

1Department of Mechanical Engineering, Carnegie Mellon University, Pittsburgh, PA 15213, USA. 2Department of

Engineering, Robert Morris University, Pittsburgh, PA 15108, USA. 3Max Planck Institute for Intelligent Systems,

Stuttgart 70569, Germany. Correspondence and requests for materials should be addressed to M.S. (email: sitti@

is.mpg.de)

Received: 11 January 2015

Accepted: 06 May 2015

Published: 15 June 2015

OPEN
http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-mailto:[email protected]:[email protected]:[email protected]:[email protected]://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-


2/13

www.nature.com/scientificreports/

2SCIENTIFICREPORTS| 5:11403 | DOI: 10.1038/srep11403

bacterial strains and knowing that cancerous tumors have a lower pH compared to that o peripherynormal tissue26,27, it would be enticing to explore the potential o applying pH-taxis based control omicrorobotic systems or targeted drug delivery applications. o urther explore the easibility o such anapproach, greater insight into the pH-tactic behavior o these microsystems is required.

Here, we present a method that takes advantage o the bacterial sensing o ambient pH to realizerobust drif control o multi-bacteria propelled microsystems. S. marcescensis employed as the bioactua-tor or the microsystem, not only because it is a typical flagellated bacterial strain with high motility andtactic behaviors, but also because o its natural adhesion to negatively-charged, hydrophobic suraces 28,29,which greatly simplifies the assembly process o the microsystem. S. marcescensbacteria swim in liquid

environments by incessant alternation o run and tumble states similar to E. coli30, with an averagetumble rate measured to be around 1.3 s1 29. Teir mean swimming speed can be as high as 47 m/s31. emperature responses and chemotaxis o S. marcescenshave been characterized and have also beenound to resemble those o E. coli29,32. Since a common signaling machinery is suggested or chemotaxis,thermotaxis, and pH-taxis25, it is expected that S. marcescensalso possesses a similar pH-tactic behavioras E. coli.

o perorm a drif control study o the microsystems, we use a microfluidic device to generate threestable pH gradients. Te bidirectional pH-taxis o ree swimming bacteria is observed or the first timeusing the configured pH gradients; tracking o the swimming bacteria allows us to determine that thebacterial pH-tactic motion is mediated by the biased flagellar tumble rates. Ten we study the distribu-tion and motion o S. marcescenspropelled microrobotic swarms. Depending on the applied pH gradi-ent profile, the microrobotic systems are shown to exhibit either bidirectional or unidirectional tacticmotions. Since it is not intuitively clear how a microrobot with multiple bacteria attached in randomdirections can produce the same pH-tactic response as ree-swimming bacteria, we perorm a detailed

analysis on the trajectories o the microrobots, which enable us to determine that two motion bias actorscontribute to the tactic drif velocity o the microrobotic systems.

ResultspH gradient generation. Using a diffusion based microfluidic gradient generator, three stable pHgradient profiles (named Gradient 1, 2 and 3 or convenience) were created in a quiescent fluidic channel,where samples o bacteria and bacteria-propelled microrobots were loaded and tested. Gradient 1 wascreated to study the bidirectional pH-taxis o bacteria and to demonstrate that the bacteria-propelledmicrorobots can be navigated by the ambient pH distribution, while Gradients 2 and 3 were used toquantiy the unidirectional drif o the microrobots and thereore unveil the multi-bacterial driving andsteering mechanism. Details about the gradient generator and visualizations o the pH gradients areincluded in the methods section.

Bacterial bidirectional pH-taxis. Based on recent fluorescence resonance energy transer (FRE)

results as well as mathematical models, it has been proposed that E. coli is capable o taxis away romboth strongly acidic and alkaline environmental conditions, resulting in accumulation o the bacteria atan optimal pH region23,24. However, such bidirectional taxis has never been visualized directly, and theassociated swimming behavior has not been studied. Using Gradient 1 (pH: 6.0-7.6), which is a stable pHgradient that covers the pH transition rom acidic to alkaline, we were able to observe the bidirectionalpH-taxis o S. marcescensdirectly. As shown in Fig. 1(a,b), the bacteria accumulate to orm a band aroundthe center line o the sample channel afer about 1.5 min rom the start o the experiment; the positiono the band corresponds to a pH value slightly above 7.0. From the color chart o the pH indicator, theoptimal pH was ound to correspond to values between 7.0 and 7.3. Te distribution profile shows asharper decrease in bacterial number in the transition rom ambient to alkaline pH than the transitionrom ambient to acidic pH; this is probably due to the drastic pH change at this corresponding location.

Bacteria in an isotropic environment ollow a purely random walk, which generates a uniorm distri-bution o bacteria in a bounded space at steady state; a nonuniorm distribution o incessantly movingbacteria in our sample channel reveals a motion deviating rom a random walk. A biased distribution

o bacteria is ofen seen in bacterial chemotaxis, which has been attributed to a biased tumble rate dis-tribution based on swimming direction. Since FRE results23 indicate that similar signaling pathwaysare employed in both bacterial pH-taxis and chemotaxis, it is reasonable to expect that the bandeddistribution o S. marcescensunder a pH gradient is also a result o a biased tumble rate. By tracking thebacterial swimming direction and detecting the number o tumble events in two rectangular regions withpH values below and above the optimal value, we find that the tumble rate distribution is significantlybiased and is dependent on the swimming direction (Fig. S1). For both regions, the average tumble rate(based on ~2000 trajectories) o the bacteria is ound to be substantially lower when swimming towardthe optimal pH (~1.0 s1) than when swimming toward the opposite direction (~1.5 s1). Our resultscorroborate the reported bidirectional pH-taxis signaling pathway model24and indicate a resemblancebetween E. coliand S. marcescensin terms o pH-tactic behavior.

Bacteria-propelled microrobotic system controlled by pH gradient. We abricated the biohy-brid robots by randomly attaching multiple bacteria onto 3m diameter polystyrene beads. Fluorescentstaining enabled the simultaneous visualization o the attached bacteria and bead (Fig. 2 (a)). Te mean
http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-


3/13



and standard deviation o the bacterial attachment number were determined to be 9.0 3.4, based onexaminations o 20 microrobots randomly picked rom the whole population. Te position and orien-tation o the attached bacteria varied significantly rom bead to bead, which is an expected result basedon the spontaneous attachment o bacteria during the assembly process. Fig. 2(b) shows a simplifieddepiction o bacteria attached to a bead; the orce vectors represent the instantaneous net orce and

torque exerted on the bead by the attached bacteria.By exposing a large number o microrobots to Gradient 1, we demonstrate drif control o the micro-robots. As shown in Fig. 3(a), the microrobots were initially uniormly distributed. Over time, the uni-orm distribution evolves into a dense band o microrobots located around the centerline o the channel.Tis steady state distribution is achieved afer about 6 minutes. Te most probable location o the micro-robots at steady state, which is shown in Fig. 3(b), coincides with that o ree swimming bacteria (theoptimal pH value). A side by side comparison between the distribution profile o ree swimming bacteriaand microrobots indicates a high degree o resemblance; in both distributions, there is a sharp decreaseon the alkaline side o the optimal pH.

While the bidirectional pH-taxis shows the versatility o bacterial pH-taxis, it is preerable to studythe bias actors that drive the tactic behavior under unidirectional taxis. Tereore, two more pH gradi-ents, Gradient 2 (pH: 3.85.4 ) and Gradient 3 (pH: 8.29.8) (see more details in the methods section),were created to achieve the unidirectional drif control o the microrobotic system. It took approximately10 minutes or the majority o microrobots to accumulate on one side o the channel in the two cases

studied: taxis away rom a more acidic condition (Fig. 4(a)) and taxis away rom a more alkaline condi-tion (Fig. 4(b)). Using image processing, the ycomponent o the center o mass (COM-y, see details inmethods section) o the microrobotic system can be computed at different time points. In Fig. 4(c,d), thedrif behavior is shown to be highly consistent among the three tested samples or the two cases. Becausethe microrobots eventually accumulate at the wall o the device, they effectively dont drif anymore butare still taken into account when calculating the average COM-y. Tis results in an artificial decrease inthe slope o the COM-yover time as shown in Fig.4. Tus, the time derivative o the COM-yproducesan underestimation o the actual pH-taxis drif velocity o the microrobotic system; a more accuratecharacterization o the pH-taxis drif velocity is obtained in the ollowing section via analysis o theswimming trajectories o the microrobots.

Trajectory analysis. o understand how a microrobot, which consists o a microsphere propelledby a group o randomly oriented bacteria, is endowed with pH-taxis capabilities, we tracked individualmicrorobots subjected to unidirectional pH-taxis. Motions o slowly moving microrobots are susceptible

Figure 1. Bacterial bidirectional pH-taxis. (a) A phase contrast image o the ree swimming S. marcescens

at steady state (stabilized afer around 1.5 min) under Gradient 1, where the black and white dots are cellbodies o bacteria. (b) Probability distribution o bacteria position extracted rom multiple images at steadystate.
http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-


4/13



to ambient flows (0.41 0.21m/s); thus we only included the trajectories o microrobots with meanspeed greater than 4m/s (~10 times the ambient flow speed) in our analysis. In addition, trajectoriesthat came within 50m (~15 times the body length o the microrobot) o the walls were removed romthe analysis since these microrobots were subjected to wall effects. Using these two criteria, over 900trajectories (see Fig. S2 or sample trajectories) were collected and analyzed or each swarm sampleundergoing unidirectional taxis.

Heading distribution. Te heading direction o all the trajectories (rom three swarm samples) weredetermined on a rame-by-rame basis. By counting the number o rames where the swimming direction(i.e. orientation angle) was within a defined angle interval, the probability distributions o the swimmingheading could be generated. Te distributions also represent the portion o time that the microrobotsspend swimming in each direction. Substantial biases in heading distributions are observed in Fig. 5(a,b).Te least probable swimming directions correspond with the unavorable conditions and the most prob-able directions correspond with the avorable conditions. Te heading distribution provides a quantifi-cation o the swimming angle preerence o the microrobots; however, it is not intuitively clear how theheading bias is achieved. Our ollowing study on the swimming direction reversing rate along they-axis(pH gradient direction) sheds light on the answer.

Direction reversing rate. In Fig. 5(c), we study the motion bias o the microrobots with respectto their heading directions. Since the y component o the swimming velocity determines whether the

microrobot is swimming towards the optimal pH region or not, we classified the headings o all ramesinto two groups in terms o their y direction: heading towards the optimal pH (shaded in red) andheading away rom the optimal pH (shaded in gray). Te direction reversing rate o each heading groupis simply defined as the total number o y-direction switchings (rom +yto yor vice versa) observedin that group divided by the total number o rames o that heading group. As shown in Fig. 5(c,d),across three different samples, the direction reversing rates or microrobots moving towards avored pHregions are consistently smaller than or those moving towards unavored pH regions. In other words,the orientation o the bacteria propelled microrobots is more persistent and less likely to change whenthey move towards the avored pH regions; this yields a larger portion o time spent moving towards anoptimal pH region, as revealed by the heading bias in Fig. 5(a,b).

Swimming speed. In addition to the heading bias, there is also a bias in the mean swimming speedwith respect to the swimming direction (Fig. 5(e,)); namely, a higher speed is observed when mov-ing toward the avored pH region. Due to the low Reynolds number o the swimming motion, the

Figure 2. Bacteria-propelled biohybrid microrobotic system. (a) Sample fluorescent images o themicrorobots, which are composed o multiple attached bacteria (yellow-green) and a spherical polystyrenebead (red). (b) At each moment, the applied orces on the bead rom the attached bacteria can berepresented by a net orce Fpand a net torque pvector.
http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-


5/13



Figure 3. Bidirectional drif control o the microrobotic system. (a) Fluorescent images o themicrorobots (green dots) show the evolution o the microrobot distribution in a fixed ocal plane over time.Te height (y-dimension) o each image covers the ull channel width. Te inset on the lef side o each

panel indicates the intensity profile o the rame along they-axis. Te color bars on the right hand sideindicate the ambient pH gradient profile (Gradient 1). (b) Probability distribution o the microrobot positionacross the width o the sample channel, where each distribution is averaged rom 200 fluorescent images

taken around the corresponding time point.


6/13



Figure 4. Unidirectional drif control o the microrobotic system. Fluorescent images show the migrationo the microrobotic system away rom more acidic (a) and more alkaline (b) conditions in the sample

channels, where Gradients 2 and 3 were applied, respectively. Te inset on the lef side o each panel in (a)and (b) indicates the intensity profile o the rame along they-axis. Plots o the variations o the COM-yothe microrobotic system over time: away rom more acidic (c) and away rom more alkaline regions (d). Te

results or three different system samples are shown or each case.


7/13



Figure 5. Biases in swimming heading, reversing rate (alongy-axis), and 2D average swimming speed under

two unidirectional pH-taxis cases: away rom a more acidic condition (a,c,e) and away rom a more alkalinecondition (b,d,). (a,b) Probability distributions o the swimming heading in 2D. (c,d) Heading directionreversing rate with respect to the heading direction along the y-axis. (e,) Mean 2D swimming speed with

respect to the heading direction along the y-axis. Te pink color indicates results rom motion towardsavored pH regions while the gray color shows the corresponding results rom motion towards unavoredpH regions.


8/13



hydrodynamics o the microrobot can be described by Stokes law, Fd=6Rv, where Fd is the fluidicdrag orce, is the dynamic viscosity o the fluid medium, R is the radius o the spherical bead, and vis the instantaneous translational speed o the microrobot. Te fluidic drag orce Fd is balanced by theinstantaneous resultant propulsive orce (Fp in Fig. 2) produced by the attached bacteria. From the lin-ear relationship between Fpand v, we can conclude that on average the attached bacteria exert a higherpropulsive orce on the microrobot when it move towards a avored pH environment.

Drift velocity. In the chemotaxis or pH-taxis o ree swimming bacteria, the heading bias is the onlyactor that contributes to the tactic drif. However, in microrobotic systems, both a swimming speed biasand a heading bias contribute to the drif velocity. o evaluate the dependency o drif velocity on thesetwo actors, we derive a drif velocity equation based on a one-dimensional (1D) biased random walkscenario (see methods section or derivation),

V v1

1 1drit D1

=

+ ( )

where and are the coefficients o the swimming speed bias and heading time bias, respectively, and

v D1 is the mean 1D swimming speed. Te drif velocity o the microrobots in the gradient direction canbe readily computed rom our measurements on the heading bias, speed bias, and the mean swimmingspeed along the pH gradient direction. Te average headings (the dashed arcs in Fig. 5) are used to

evaluate the average heading time bias (see the methods section). Te drif velocities o the two cases(moving away rom more acidic pH regions and moving away rom more alkaline pH regions) werecalculated to be similar, both o which are around 0.5m/s. We urther examined the relative contribu-tions o the heading bias and speed bias to the overall drif velocity: the heading bias contributes to ~75%o the total drif velocity while the speed bias contributes to ~25% o the total. Tis indicates that, inaddition to the heading bias, the speed bias is an essential mechanism o the microrobots tactic motion;this is a departure rom the mechanisms known to cause the biased random walk observed in ree swim-ming bacteria under pH-taxis or chemotaxis.

Dependence on swimming speed. Since a wide variance in the swimming speeds o the microro-bots was observed, it is meaningul to inquire about the potential influence o the absolute swimmingspeed on the motion bias. o analyze the dependence on swimming speed, the captured trajectories oreach unidirectional pH-taxis case were divided into groups based on their mean instantaneous speed,as shown along the x-axis in Fig. 6. Te relative reversing rate bias ((rr+)/(r+r+), where r+ and

Figure 6. Dependences o motion biases on the average swimming speed. (a) Swimming speed bias with

respect to the mean swimming speed. (b) Relative reversing rate bias with respect to the mean swimmingspeed. Te horizontal error bars indicate the standard deviation o mean speeds or the trajectories groupedwithin a given speed interval (45, 56, 67, 78, 89, and 9m/s). Te vertical error bars in (a) denotethe standard deviation o the speed bias or the trajectories alling within the corresponding speed intervals.

In both (a) and (b), each speed interval has over 80 sample trajectories measured.
http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-


9/13



rare reversing rates towards and away rom optimal pH, respectively) quantifies the dependence o thedrif velocity on the direction reversing rate33. Both speed and relative reversing rate biases increase withincreasing mean swimming speed, and this trend holds true or both cases (away rom acid, away rombase). Since the speed bias and reversing rate bias (or heading bias) are actors that contribute to a biasedrandom walk, it can be concluded that the microrobot exhibits a stronger tactic motion when a higherswimming speed is achieved.

DiscussionWe have studied the pH-tactic behavior o a large number o bacteria-propelled microrobots in a micro-

fluidic channel with a stable spatial pH gradient. It has been demonstrated that the spatial pH gradientcan effectively and consistently generate drif motion in the microrobotic system. o ully understand thebiased motion o the microrobots and the mechanisms that produce the biased motion, we tracked indi-vidual microrobots and analyzed their trajectories. For ree swimming bacteria, the tumble rate is biasedwith respect to swimming direction, yielding a heading bias, which in turn generates a pH-tactic orchemotactic drif velocity. However, unlike the clear run and tumble switching pattern o ree swimmingbacteria, the motion o bacteria-propelled microrobots can be described as an incessant translation withgradual changes in heading direction. From a trajectory analysis, the drif velocity o the microroboticsystem under a stable pH gradient is ound to result rom two actors, namely, the heading bias and theswimming speed bias.

o explain how the heading bias is produced in the microrobotic system, we must understand theeffect o pH on the flagellar tumbling rate. Free swimming bacteria increase their flagellar tumbling rateswhen moving towards unavored pH regions or when sensing unavored temporal pH changes. Sincethe assembly o bacteria onto micro polystyrene beads relies on physical adherence, we do not expect

undamental changes to the chemical sensing machinery o the bacteria afer integration with the beads.Tereore, when the microrobot moves toward unavored pH regions, the average flagellar tumbling rateo the attached bacteria tends to increase, and this introduces more disturbances to the motion o themicrorobot by requently changing the applied orces and torques. As a result, compared with motiontowards avored pH regions, the microrobot maintains less consistency in its swimming direction whenmoving towards an unavorable pH region; this leads to the reversing rate bias and hence the headingbias.

As we already discussed, the swimming speed bias reveals a bias in the propulsive orce on the micro-robot. Presumably, when the microrobot moves towards a avorable pH, less flagella are in a tumblestate than when compared to moving towards an unavorable pH. However, detailed observation o theflagella when the bacteria is attached to the microrobot is essential to ully understand the driving mech-anisms. Te dependencies o the motion bias on the swimming speed is potentially due to the act thatbacteria sense the temporal change o the ambient pH34; since the spatial pH profile is constant, it is theswimming velocity that determines the temporal pH gradient seen by the bacterial receptors. Tereore,

the enhanced motion bias at higher swimming speeds could be explained by a stronger temporal pHgradient being sensed.Te work o Hu and u indicates that a common biochemical signaling pathway is responsible or

different kinds o bacterial taxis behaviors, including pH-taxis, chemotaxis, and thermotaxis25. It is highlypossible that appropriate chemical gradients and temperature gradients could also enable effective drifcontrol o bacteria-propelled microrobotic systems with a similar design; the two methods should adhereto the same driving mechanisms that we have ound in pH-taxis based control o biohybrid microrobots.o apply biohybrid microrobots in applications in bioengineering and medicine, reliable and efficientcontrol o the microrobot at the swarm scale is a critical step. Our demonstrations o robust drif controlusing pH gradients expand the current scope o bacteria-propelled microrobotic control methods. Teavailability o a pH gradient or ease o deploying a gradient in the workspace will highly depend on thespecific application. Tis work suggests the potential easibility o applying pH-taxis, chemotaxis, andthermotaxis as control methods, and the appropriate method o control can be chosen based on thespecific application.

MethodsBacteria and growth conditions. Serratia marcescens (ACC 274, American ype CultureCollection, Manassas, VA) was initially cultured to exponential growth phase in a nutrient broth (25 gDico LB Miller Broth and 1 L deionized (DI) water, pH 7.0) on a shaker at 37 C or 3.54 hours. Tenan aliquot o 2.0L o the liquid culture was transered to an agar plate (25 g Dico LB Miller Broth, 6 gBacto Agar, 5 g glucose, 1 L de-ionized water), ollowed by an incubation o the agar plate at 30 C or1620 hours. Afer the culturing period, bacteria on the leading edge o the colony was either suspendedin motility buffer or use in the bacterial pH-taxis study or suspended in a bead solution or the abri-cation o the biohybrid microrobots.

Microrobot fabrication. Te microrobots were abricated by randomly attaching bacteria to 3 mdiameter fluorescent polystyrene beads (=1.05 g/cm3, Fisher Scientific, Inc.). o enable natural attach-ment between the bacteria and beads, the original coating o the beads was removed by alternately ultra-sonicating the beads in deionized (DI) water or isopropyl alcohol (IPA, 50%) or a total o five cycles;
http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-


10/13


1 0SCIENTIFICREPORTS| 5:11403 | DOI: 10.1038/srep11403

residual IPA in the bead solution was removed by three more ultrasonication cycles with DI water. Te

washed beads were soaked in motility buffer at a volume concentration o 0.05%. Te microrobots wereassembled by placing an aliquot o 2.5 L bead solution onto the leading edge o the bacteria colony onthe agar plate and gently pipetting 35 times to mix the bacteria and beads sufficiently. Te solutionwas collected back immediately and incubated at room temperature or 5 minutes, allowing or ran-dom attachment o the bacteria to the beads. Ten, the solution was diluted by adding 40 L o Percoll(=1.13 g/cm3, Sigma-Aldrich, St. Louis, MO) and 57.5 L o motility buffer to the solution. Percollwas added to increase the density o the fluid, thereby making the microrobots neutrally buoyant. Tefinal solution was urther diluted to achieve an appropriate concentration or vision tracking o themicrorobots.

Microuidic setup fabrication. Te microfluidic concentration gradient generator was assembledrom a molded hydrogel chip containing the channel eatures35. o mold the hydrogel chip, a mastermold o the channel patterns was abricated by a standard sof lithography method. o increase the chan-nel height, two layers o photoresist were used. Te hydrogel chips were molded by pouring 4% (weight

ratio) hot agarose (Eiken Chemical Co.) solution onto the silicon master mold, where the channel pat-terns were surrounded by polydimethylsiloxane (PDMS) enclosures. Afer the agarose gels were cured,the outlets o the source and sink channels were punched into the gel. Subsequently, the sample solutionto be tested was careully pipetted into the sample (middle) channel ensuring that there was neitheroverflow nor much vacant space lef in the sample channel. Te channel-patterned side o the agarose gelwas covered with a cover slip immediately afer loading the sample solution. o complete the assemblyo the gradient generator, the agarose gel chip (including the diffusion section, a PDMS enclosure and acover slip) was sandwiched between two acrylic panels as shown in Fig. 7(a).

pH gradient generation and visualization. Unlike the generation o pH gradients by electrolysis36,a diffusion-based method can eliminate the electrical field induced effects on the motion o the micro-robot. Tereore, a flow-ree diffusion based gradient generator design35,37was applied to abricate thepH gradient generator. As shown in Fig. 7(a,b), the gradient generator consists o three parallel channels,namely the sample channel and two side channels. Te width o the channels is 500 m and the height

Figure 7. (a) Configuration o the three-channel diffusion-based pH gradient generator (side view). (b) opview o the three parallel channels and the three gradient profiles visualized in situby three appropriate pHindicators, where bright lines in the color profiles indicates the channel walls. Te pH gradient in the sample

channel was generated by pumping two fluids with different constant pH values into the side channels,while the sample channel was completely closed and quiescent. Te pH indicators used to visualize the threegradients rom lef to right were Bromothymol Blue (sensitive pH 6.07.6, with color transitioning rom

yellow to blue), Bromocresol Green (sensitive pH 3.85.4, with color transitioning rom yellow to blue) andCresolphthalein (sensitive pH 8.29.8, with color transitioning rom colorless to purple).
http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-


11/13



is around 200m. Te channels are separated rom each other by two 250 m wide agarose gel ridges. Aconstant flow o fluid was pumped through the outlets o the two side channels, whereas the outlets o thesample channel were sealed to decrease undesired drif flows. A programmable syringe pump (BraintreeScientific Inc.) was used to pump fluid with different pH values into the two side channels at a flow rateo 5L/min. Generation o a linear concentration gradient in the device has been ully calibrated andreported in our previous work32.

o create solutions with different pH values, either HCl or NaOH solution was added to motilitybuffer (0.01 M KH2PO4, 0.067 M NaCl, 10

4M EDA, pH =7.0). Tree stable pH gradients were gener-ated by pumping motility buffer with different values o pH into the two side channels o our device. o

veriy the existence o a stable pH gradient in the sample channel, the same concentration o an appro-priate pH indicator was added to all three channels. Te pH gradients were visualized in situ and thepH transition in the sample channel was determined, as shown in Fig. 7(b). Based on the indicator colorcharts and the resulting color profiles, the ollowing pH ranges were measured: Gradient 1 maintains apH range between 6.0 (bottom) and 7.6 (top) in the sample channel, and a neutral pH region lies closeto the center line o the sample channel; Gradient 2 maintains a pH range between 3.8 (bottom) and 5.4(top), creating a more acidic environment along the bottom; the color profile o Gradient 3 indicates apH range between 8.2 (bottom) and 9.8 (top), showing that the top is more alkaline. Te pH gradientsgenerated by the diffusion o ions were verified to be stable by the constant color profiles o the indica-tors afer the estimated diffusion time. Since the time scale o ionization and recombination is negligiblecompared to that o diffusion, the diffusion time o H+(0.9 min) and OH(1.6 min) across the two sidechannels was used to characterize the stabilization time o the pH gradients. In addition, we can treatthe ionization o the sample solution to be quasistatic at each moment, i.e., there was no electric fieldinduced by the diffusion o ions.

Imaging and tracking. Te samples were imaged using a 10x (fluorescence imaging o microrobots)or 40x (phase contrast imaging o bacteria) objective in an inverted microscope (Zeiss Axio Observer100). Video data were captured at 88 ps (bacteria) and 5 ps (microrobots) by a digital camera (QICAM,QImaging) attached to the microscope. Both the captured bacteria and microrobots were located ar away(10 body lengths) rom any channel walls to eliminate wall effects on their motion. wo-dimensional(2D) motion o the microrobots was obtained rom the video data using an in-house tracking programdeveloped in MALAB (R2012a, Te MathWorks, Inc, Natick, MA). 2D bacterial swimming motion wastracked and analyzed with the methods described in our previous publications.29,32.

o calculate the COM-yo the microrobotic system or a given imaging rame, all the microrobots inthe rame were detected by image processing and their 2D positions were determined. Ten COM-ywascalculated by averaging the ypositions across all microrobots: = =y yCOM n i

ni

11 , where n is the total

number o microrobots.

pH-taxis drift velocity model. A particles movement along one direction, namely the y-axis, ismodeled to determine the actors that can contribute to the drif velocity o a system exhibiting an inher-ent biased random walk. Assuming that the particle maintains different mean speeds when movingtowards +yand ydirections, denoted by v+yand vy, respectively; the particle can switch its directiono motion in a way such that the portion o time it spends moving towards the +y direction, t+y, isdifferent rom the time it spends moving towards the ydirection, ty. It is straightorward to describethe mean speed v D1 and the drif velocity Vdrif(with +ybe the deault direction) o the particle.

vv t v t

t t 2D

y y y y

y y1 =

+

+ ( )

+ +

+

Vv t v t

t t 3drit

y y y y

y y

=

+ ( )

+ +

+

In Eq. (3), the drif velocity is essentially caused by the bias in swimming speed and the bias in thetime spent in moving in a given direction. o include the two biasing actors, we define two coefficients:the coefficient o speed bias, =vy/v+y, and the coefficient o orientation bias, =ty/t+y. Substitutingthese coefficients into the two equations above, Vdrifcan be expressed in terms o the mean speed v D1 ,

V v1

1 4drit D1

=

+.

( )

References1. Carlsen, . W. & Sitti, M. Bio-hybrid cell-based actuators or microsystems. Small10,38313851 (2014).2. Martel, S., remblay, C. C., Ngaeng, S. & Langlois, G. Controlled manipulation and actuation o micro-objects with magnetotactic

bacteria.Appl. Phys. Lett.89,233904 (2006).
http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-


12/13



3. Hiratsua, Y., Miyata, M., ada, . & Uyeda, . Q. P. A microrotary motor powered by bacteria. Proc. Natl. Acad. Sci. USA103,1361813623 (2006).

4. Beham, B. & Sitti, M. Bacterial flagella-based propulsion and on/off motion control o microscale objects.Appl. Phys. Lett.90,023902 (2007).

5. Sitti, M. Miniature Devices: Voyage o the microrobots. Nature458,11211122 (2009).6. Angelani, L., Leonardo, . D. & uocco, G. Sel-starting micromotors in a bacterial bath. Phys. ev. Lett.102,048104 (2009).7. Soolov, A., Apodaca, M. M., Grzybowsi, B. A. & Aranson, I. S. Swimming bacteria power microscopic gears. Proc. Natl. Acad.

Sci. USA107,969974 (2010).8. Leonardo, . D. et al.Bacterial ratchet motors. Proc. Natl. Acad. Sci. USA107,95419545 (2010).9. Fernandes, ., Zuniga, M., Sassine, F. ., araoy, M. & Gracias, D. H. Enabling cargo-carrying bacteria via surace attachment

and triggered release. Small7,588592 (2011).

10. ojima, M., Zhang, Z., Naajima, M. & Fuuda, . High efficiency motility o bacteria-driven liposome with raf domain bindingmethod. Biomed. Microdevices14,10271032 (2012).

11. Ma, Q. et al.Construction and operation o a microrobot based on magnetotactic bacteria in a microfluidic chip. Biomicrofluidics6,24107 (2012).

12. Edwards, M. ., Carlsen, . W. & Sitti, M. Near and ar-wall effects on the three-dimensional motion o bacteria-drivenmicrobeads.Appl. Phys. Lett.102,143701 (2013).

13. Steager, E. B. et al.Electroinetic and optical control o bacterial microrobots.J. Micromech. Microeng.21,035001 (2011).14. Magdanz, V., Sanchez, S. & Schmidt, O. G. Development o a sperm-flagella driven micro-bio-robot. Adv. Mat. 25, 65816588

(2013).15. Carlsen, . W., Edwards, M. ., Zhuang, J., Pacoret, C. & Sitti, M. Magnetic steering control o multi-cellular bio-hybrid

microswimmers. Lab Chip14,38503859 (2014).16. oumais, N., Lepore, A., Maggi, C. & Leonardo, . D. argeted delivery o colloids by swimming bacteria. Nature Comm. 4,

2588 (2013).17. raor, M., Sahari, A. & Beham, B. Computational and experimental study o chemotaxis o an ensemble o bacteria attached

to a microbead. Phys. ev. E84,16 (2011).18. im, D., Liu, A., Diller, E. & Sitti, M. Chemotactic steering o bacteria propelled microbeads. Biomed. Microdevices14, 1009

1017 (2012).

19. Par, D. et al. Motility analysis o bacteria-based microrobot (bacteriobot) using chemical gradient microchamber. Biotech.Bioeng.111,134143 (2014).

20. Sahari, A., raore, M. A., Schar, E. B. & Beham, B. Directed transport o bacteria-based drug delivery vehicles: bacterialchemotaxis dominates particle shape. Biomed. Microdevices16,717725 (2014).

21. Booth, I. . egulation o cytoplasmic pH in bacteria.Microbio. ev.49,359378 (1985).22. rulwich, . A., Sachs, G. & Padan, E. Molecular aspects o bacterial ph sensing and homeostasis. Nature ev. Microbio. 9,

330343 (2011).23. Yang Y. & Sourji, V. Opposite responses by different chemoreceptors set a tunable preerence point in escherichia colipH taxis.

Mol. Microbio.86,14821489 (2012).24. Hu, B. & u, Y. Precision sensing by two opposing gradient sensors: How does escherichia colifind its preerred pH level? Biophys.

J105,276285 (2013).25. Hu, B. & u, Y. Behaviors and strategies o bacterial navigation in chemical and nonchemical gradients. PLOS Comp. Bio.10,

DOI: 10.1371/journal.pcbi.1003672 (2014).26. annoc, I. F. & otin, D. Acid pH in tumors and its potential or therapeutic exploitation. Cancer es.49,43734384 (1989).27. Griffiths, J. . Are cancer cells acidic? British J. Cancer64,425427 1991.28. Dicson, J. S. & oohmaraie, M. Cell surace charge characteristics and their relationship to bacterial attachment to meat

suraces.App. Env. Microbio.55,832836 1989.

29. Edwards, M. ., Carlsen, . W., Zhuang, J. & Sitti, M. Swimming motility characterization o Serratia marcescens. J. Micro-Bioobotics9,4760 (2014).

30. Berg, H. C. & Brown, D. A. Chemotaxis in Escherichia coli analysed by three- dimensional tracing. Nature239,500504 (1972).31. Darnton, N., urner, L., Breuer, . & Berg, H. C. Moving fluid with bacterial carpets. Biophys. J. 86,18631870 (2004).32. Zhuang, J. et al.Analytical modeling and experimental characterization o chemotaxis in serratia marcescens. Phys. ev. E89,

052704 (2014).33. ivero, M. A., ranquillo, . ., Buettner, H. M. & Lauffenburger, D. A. ransport models or chemotactic cell populations based

on individual cell behavior. Chem. Eng. Sci.44,28812897 (1989).34. Brown, D. A. & Berg, H. C. emporal stimulation o chemotaxis in escherichia coli. Proc. Nat. Acad. Sci. USA71, 13881392

(1974).35. Cheng, S. Y. et al.A hydrogel-based microfluidic device or the studies o directed cell migration. Lab Chip7,763769 (2007).36. He, B., Wang, Z., Liu, C., Li, Y. & Shen, . Swimming behavior analysis based on bacterial chemotaxis in solution. J. Bionic Eng.

9,315321 (2012).37. Diao, J. et al.A three-channel microfluidic device or generating static linear gradients and its application to the quantitative

analysis o bacterial chemotaxis. Lab Chip6,381388 (2006).

AcknowledgmentsTis work was supported by the National Science Foundation Cyberphysical Systems program (grantnumber: CNS-1135850). Any opinions, findings, and conclusions or recommendations expressed inthis material are those o the authors and do not necessarily reflect the views o the National ScienceFoundation.

Author ContributionsAll authors developed the idea and designed the experiments. J. Z. perormed the experiments. Allauthors wrote and edited the manuscript.

Additional InformationSupplementary inormationaccompanies this paper at http://www.nature.com/srep

Competing financial interests: Te authors declare no competing financial interests.
http://www.nature.com/srephttp://www.nature.com/srep


13/13


How to cite this article: Zhuang, J. et al.pH-axis o Biohybrid Microsystems. Sci. Rep.5, 11403; doi:10.1038/srep11403 (2015).

Tis work is licensed under a Creative Commons Attribution 4.0 International License. Teimages or other third party material in this article are included in the articles Creative Com-

mons license, unless indicated otherwise in the credit line; i the material is not included under theCreative Commons license, users will need to obtain permission rom the license holder to reproducethe material. o view a copy o this license, visit http://creativecommons.org/licenses/by/4.0/
http://creativecommons.org/licenses/by/4.0/http://creativecommons.org/licenses/by/4.0/

ph taxis microsystems

Documents