COMMUNICATION TO THE EDITOR

Correlating Single Cell Motility With PopulationGrowth Dynamics for Flagellated Bacteria

Sucheta Arora, Vidya Bhat, Aditya Mittal

Department of Biochemical Engineering & Biotechnology,

Indian Institute of Technology Delhi, Hauz Khas, New Delhi 110016, India;

telephone: 91 11 26591052; fax: 91 11 26582282; e-mail: aditya@dbeb.iitd.ac.in

Received 20 September 2006; accepted 29 January 2007

Published online 1 February 2007 in Wiley InterScience (www.interscience.wiley.com

). DOI 10.1002/bit.21372

ABSTRACT: Many bacteria used for biotechnological appli-cations are naturally motile. Their ‘‘bio-nanopropeller’’driven movement allows searching for better environmentsin a process called chemotaxis. Since bacteria are extremelysmall in size compared to the bulk fluid volumes in bior-eactors, single cell motility is not considered to influencebioreactor operations. However, with increasing interest inlocalized fluid flow inside reactors, it is important to askwhether individual motility characteristics of bacteria areimportant in bioreactor operations. The first step in thisdirection is to try to correlate single cell measurements withpopulation data of motile bacteria in a bioreactor. Thus, weobserved the motility behavior of individual bacterial cells,using video microscopy with 33 ms time resolution, as afunction of population growth dynamics of batch cultures inshake flasks. While observing the motility behavior of themost intensively studied bacteria, Escherichia coli, we findthat overall bacterial motility decreases with progression ofthe growth curve. Remarkably, this is due to a decrease in aspecific motility behavior called ‘‘running’’. Our results notonly have direct implications on biofilm formations, but alsoprovide a new direction in bioprocess design research high-lighting the role of individual bacterial cell motility as animportant parameter.

Biotechnol. Bioeng. 2007;97: 1644–1649.

� 2007 Wiley Periodicals, Inc.

KEYWORDS: motility; growth curve; nanopropeller;chemotaxis; bioprocess design; video microscopy

Introduction

Industrial production of several important metabolites isachieved through growth of bacterial cultures in bioreactors(Glazer and Nikaido, 1995). Growth of bacteria is dependenton several factors including environmental parametersapart from the nutritional status of the culture medium.Bioreactors are designed and operated to ensure homo-geneity in the system with the aim of maintaining identicalgrowth conditions for each cell (Shuler and Kargi, 2002).

Correspondence to: A. Mittal

1644 Biotechnology and Bioengineering, Vol. 97, No. 6, August 15, 2007

Although the operational conditions are aimed at com-pletely homogenizing the bulk fluid in the bioreactors,formation of localized fluid flow patterns is inevitable(Ranade, 2002).

Many bacteria that are used for production of biotech-nological products are naturally motile. For example,Escherichia coli, through the course of evolution, hasdeveloped an ability to move in search of favorableenvironments by using its own natural motor–propellerassembly—the flagellar motor system (Berg, 2003; Kojimaand Blair, 2004). Bacterial cell movement is enabled byflagellar proteins on their surfaces that act as nanopropellersto direct the bacteria towards better living environments in aprocess called chemotaxis (Wadhams and Armitage, 2004).This nanopropeller system is embedded in the bacterial cellmembrane and driven by ion gradients across the cellmembrane (Gabel and Berg, 2003; Sowa et al., 2005).

Since the bacteria are extremely small in size, compared tothe bulk fluid volumes handled in bioreactors, motility ofindividual bacterial cells is not considered to influencebioreactor operations. However, with increasing interest inlocalized fluid flow inside reactors, along with the bulk flow,it is important to ask the question whether individualmotility characteristics of bacteria are significant inbioreactor operations. Motion of a submerged body in afluid depends on the ability of its propeller to overcomemainly two forces: the inertial and the viscous forces. Formacroscopic bodies the inertial forces dominate. However,for microscopic bodies inertial forces are negligible and theviscous forces are the primary constraints in their motility(Behkam and Sitti, 2005). In principle, bacterial motilitymight appear to be non-relevant; for large-scale bioreactors,however, its relevance (or lack of) in bioreactor operations isnot proven either way. Thus, it is important to ask whetherbacterial motility is important in bioprocesses. It might beespecially important in operational conditions of bioreac-

� 2007 Wiley Periodicals, Inc.

tors which deal with high viscosity media and low agitationrates to avoid shear stress on cells.

Bacterial motility is also an extremely importantparameter biofilm formation (Wood et al., 2006) andwould certainly be relevant in microfluidic devices. Whilechemotactic motility has been intensely studied frommicrobiological (Berg, 2003; Kojima and Blair, 2004) andbiophysical (Behkam and Sitti, 2005; Berg and Brown, 1972;Brown and Berg, 1974; DiLuzio et al., 2005; Lauga et al.,2006) perspectives in the last three decades, correlatingsingle cell motility with population studies (in bioreactorsystems) of motile bacteria is still an unexplored territory(but see Staropoli and Alon, 2000). In this work we hope tofill this gap by providing the first experimental evidenceon the possible variations in motion characteristics ofindividual bacterial cells as a function of the growth cycle ofthe motile bacteria in shake flask bioreactors.

Figure 1. Growth curve for batch culture of bacteria. Growth is expressed in

absorbance of the sample at 600 nm and shown by solid circles representing

mean�SD of three independent experiments. The smooth curve is drawn to visually

guide through different stages of the growth curve, approximately demarcated by

arrows into lag phase, log phase; and stationary phase.

Materials and Methods

E. coli strain JM109 was used for this study. The strainJM109 was chosen since it has been reported to be a fairlymotile strain of E. coli compared to other strains (Woodet al., 2006). Initial studies were also carried out with E. colistrain DH5-a. However, the motility was not quantifiablefor most experiments (less than 1%). The lack ofquantifiable motility is in agreement with the resultsobtained by Wood et al. (2006). Nutrient broth of analyticalgrade was obtained from HiMedia Laboratories Pvt. Ltd(Mumbai, India).

For inoculum preparation, bacterial cells were grown toexponential phase in 50 mL conical flasks, until the turbidityreached 1 at 600 nm (approximately 4 h). These cultureswere then inoculated into 250 mL of final culture volumes tobe aerobically grown in 500 mL shake flasks. Specifically, 250mL of freshly prepared Nutrient Broth media (HiMediaLaboratories Pvt. Ltd., Mumbai, India) was inoculated with10% of the inoculum culture in a 500 mL conical flask andgrown at 378C, 200 rpm in an incubator-shaker (ORBITEKAID Electronics, Chennai, India) for 9 h. Samples foranalysis were collected using pipettes under asepticconditions every half an hour, throughout the growthperiod i.e., 9 h. The culture flask was restored in its originalconditions as soon as possible.

Thirty-three millisecond (ms) resolution video micro-scopy to record bacterial motility was done as describedearlier (Gupta et al., 2006; Sharma et al., 2007). Briefly, foreach sample, the collected sample was transferred to a glassslide immediately (within a few seconds), covered with acover slip and digital videos of bacterial motion were re-corded under the 100� oil immersion objective of Motic B1upright microscope (MOTIC Microscopes) equipped with adigital camera. The camera is capable of acquiring images atthe rate of 30 frames per second (i.e., 33 ms time resolution)using the software Motic Images Plus 2.0 ML. Note that ourmotility recordings were done within 30–60 s of the time

when the sample was extracted from the culture flask toeliminate the possibility of any change in motility of bacteriadue to change in its surrounding environment (fromshaking inside a flask to that on a glass slide). It must benoted that the microscope stage temperature was notmaintained at the same temperature as that of the shakeflasks (378C). However, there was no large change inenvironment upon going from 378C to ambient (micro-scope stage), which was at 27–288C at all times. Whateversmall changes were experienced during the transfer of theculture, were expected to be normalized over all theexperiments.

Simultaneously, an aliquot of each sample was stored at48C for optical density measurement at 600 nm, to getthe growth kinetics, using a Manual Simultaneous Spec-trophotometer (SPEKOL-1200 Analytik Jena AG, Jena,Germany), which was carried out at the end of theexperiment. The experiments were done in triplicates; witheach video recording �20–200 bacterial cells simultaneously(less number of cells was essentially seen only in the veryinitial phases of growth due to very low number of cells). Inone of the experiments the optical density of samples wasmeasured immediately when the samples were taken outfrom the flask. This was done to check if there was anyvariation in the results due to storage of samples. Growthcurve shown in Figure 1 confirmed that the absorbance wasindependent of whether it was measured for fresh samples orit was measured after a few hours of storage at 48C. Videoswere then analyzed for motility data of individual cells byextracting the frames (images) with 33 ms resolution.

Arora et al.: Single Cell Motility in Bioreactor Operations 1645

Biotechnology and Bioengineering. DOI 10.1002/bit

Figure 2. Motility characteristics of single bacterial cells studied via 33 ms resolution video microscopy. If (A) is considered to be an image taken at time t, (B) shows the same

field at time¼ tþ 33 ms, and (C) shows the same field at time¼ tþ 66 ms. The bacterial cell shown in the rectangle exhibits running motion relative to a fixed point in the field of view

marked by the arrow and the encircled bacterial cell shows tumbling motion.

Results

Experiments were carried out with E. coli strains DH5-a andJM109. However, the motility was not quantifiable forexperiments done with the former strain (less than 1%). Thelack of quantifiable motility for the E. coli strain DH5-a is inagreement with the results obtained by Wood et al. (2006).Therefore, all the results are reported for the E. coli strainJM109 only.

Figure 1 shows the bacterial growth curve obtained by us.It also represents the growth curve of a typical batch cultureof bacteria in a reactor (Bailey and Ollis, 1986). The variousstages of culture growth in a bioreactor are shown in thefigure as the lag phase, the log phase and the stationaryphase. Note that that our experiments were carried out up tothe onset of stationary phase only and Figure 1 does notinclude the growth curve of bacteria in the death phase. Theabove is due to the facts that (a) we assume that the motilitybehavior of the bacterial cells at the onset of the stationaryphase represents the behavior for the remaining stationary

Figure 3. Percentage of bacteria that are motile in a batch culture as a function of gr

was calculated by counting the total bacterial cells in the field of view and then finding the n

(B) are shown to visually guide through the negative correlation (see text for details). Re

1646 Biotechnology and Bioengineering, Vol. 97, No. 6, August 15, 2007

phase and (b) death phase is not relevant for the purposes ofthis study.

We also measured the number of viable bacterial cells at afew points, by the CFU (colony forming units) assay to ensurethe basic features of the batch culture growth curve. The CFUnumbers correlated well with the absorbance data shown inFigure 1 (not shown), thus confirming that up to the time ofour measurements, we did not have a significant number ofdead cells contributing to our batch culture growth kineticsdata. Note that absorbance at 600 nm, as shown in Figure 1, isa standard measure representing estimates for cell numbers inE. coli batch cultures (Bailey and Ollis, 1986; Olfosson et al.,2003), and for the rest of this work, we will use this absorbanceto compare motility characteristics.

Individual bacterial cells could be essentially classifiedinto four categories: non-motile, runners (that swam inlinear trajectories for several decades of microns), tumblers(that tumbled or rotated at a somewhat fixed position) andtumbling-runners (that displayed short runs with tumblingbehavior). Figure 2 shows three successive frames of a video

owth in a batch culture w.r.t. time (A), and, absorbance at 600 nm (B). The percentage

umber of motile bacteria amongst them. The smooth curve in (A) and the smooth line in

sults are expressed as mean� SD of three independent triplicate experiments.

DOI 10.1002/bit

Figure 4. Percentage of runners as a function of growth in a batch culture w.r.t. time (A), and, absorbance at 600 nm (B). The percentage was calculated by counting the total

motile bacterial cells in the field of view and then finding the number of runners amongst them. The smooth curve in (A) and the smooth line in (B) are shown to visually guide through

the negative correlation (see text for details). Results are expressed as mean� SD of three independent triplicate experiments in (A) and only the mean is shown in (B).

recording for a field in a sample. Relative to a fixed point inthe field of view (marked by an arrow), the bacterial cellenclosed in a square is clearly seen to be a runner in theupward direction from A to C. In contrast, the encircledbacterium is a tumbler.

Figure 5. Percentage of tumblers (A and B) and tumbling-runners (C and D) as a funct

D). The percentage was calculated by counting the total motile bacterial cells in the field of

smooth line in (B) and (D) are shown to visually guide through a lack of correlation (see

experiments in (A and C) and only the mean is shown in (B and D).

Having recorded motility of bacteria at each point of thegrowth curve shown in Figure 1, we decided to investigatethe motility characteristics of bacteria as a function ofthe growth curve. Figure 3 shows that the percentage ofmotile bacteria (percentage of bacteria showing any

ion of growth in a batch culture w.r.t. time (A and C), and, absorbance at 600 nm (B and

view and then finding the number of tumblers and tumbling-runners amongst them. The

text for details). Results are expressed as mean�SD of three independent triplicate

Arora et al.: Single Cell Motility in Bioreactor Operations 1647

Biotechnology and Bioengineering. DOI 10.1002/bit

motility), tend to decrease with both time (Fig. 3A) andincreasing absorbance (Fig. 3B).

These results were statistically corroborated by the factthat fairly strong negative correlation coefficients wereobtained between absorbance and percentage of motilebacteria (i.e. for data shown in Fig. 3B) in the growth curve.While the overall average percent of bacteria that weremotile (i.e. averages of triplicates) showed a correlationcoefficient (CC) value of �0.58 against absorbance, theaverage CC, when considering individual experiments, wasfound to be �0.42� 0.18. Note that investigating the latterwas important, since experiment to experiment variationsresult in large standard deviations as seen in Figure 3B.

Was there a specific kind of motile bacteria that weredecreasing as growth progresses in a growth curve, or allkinds of individual bacterial motion decrease with time andcell numbers in a batch culture? To answer this question wefirst plotted percentage of runners out of the total motilebacteria (and not total bacteria in the field of view as inFig. 3), against time and absorbance as shown in Figure 4.

We observed a strong decline in the number of runnerswith the progression of the batch culture growth (Fig. 4).The statistical strength of our observations was shown bystrong negative CCs between absorbance and percentage ofrunners (i.e. for data shown in Fig. 4B) in the growth curve.While the overall average percent of runners had a very highCC value of �0.70 against absorbance, the average CC, whenconsidering individual experiments, was found to be�0.58� 0.06. These results certainly indicate that abilityof motile bacteria to ‘‘run’’ sharply decreases with thegrowth curve in a batch culture.

Next we plotted percentage of tumblers and tumbling-runners out of the total motile bacteria (as in Fig. 4), againsttime and absorbance as shown in Figure 5. We observed thatpercentage of tumblers and tumbling-runners are indepen-dent of the progression of the batch culture growth (Fig. 5).Our observations were confirmed by very low CCs betweenabsorbance and percentage of tumblers and tumbling-runners (i.e. for data shown in Figure 5B and D respectively)in the growth curve. While the overall average percent oftumblers and tumbling-runners had CC values of 0.19 and�0.02 respectively against absorbance, the average CC, whenconsidering individual experiments, was found to be0.31� 0.24 for tumblers and �0.10� 0.63 for tumbling-runners. These results certainly indicate that ability of motilebacteria to exhibit tumbling or combined tumbling-runningbehavior does not change with the growth curve in a batchculture.

Discussion and Conclusions

Bioprocess engineering assumes that the growth curve for abatch culture represents identical bacterial cells multiplyinginside a bioreactor. Are they identical indeed? We clearlydemonstrate that this is not the case with a physico-chemicalparameter of bacterial motility. The E. coli strain JM109 was

1648 Biotechnology and Bioengineering, Vol. 97, No. 6, August 15, 2007

used in our studies, since this strain has been shown to beone of the most motile strains (Wood et al., 2006). Whilethere is no set standard assay for establishing motility ofbacterial strains, it has been studied using either videomicroscopy (Berg and Brown, 1972; Diluzio et al., 2005;Gupta et al., 2006) or measuring the spread of bacterial‘‘halo’’ on agar plates (Wood et al., 2006). Considering thefact that the motility of bacteria in fluid would be differentthan on an agar plate, we used video microscopy to studybacterial motility. The bacteria used throughout experi-mentation should naturally move towards a favorablenutrient environment (chemotaxis). Therefore, the naturalexpectation during a batch culture would be that the overallbacterial population should include more motile bacteriawith advancing growth due to overall depletion of nutrientsin the culture broth.

We report a finding exactly opposite to the one expectedabove, that the percentage of motile bacteria decreases withincreasing population in a batch culture. Remarkably, thisdecrease is due to a significant decrease in a specific motilitybehavior of bacteria call ‘‘running.’’ This, on the one hand isin complete contrast to the expected behavior in which onewould expect an increase in motility with progression of thegrowth curve due to constant reduction in nutrient levels ofthe culture. On the other hand, it could simply be a result ofincreasing population of bacterial cells leading to individualcells impeding the ‘‘running’’ motion for each other. If thelatter was the case, then one would expect it only beyond themid-log phase of the batch culture, when the bacterial cellconcentration starts becoming substantial, which is clearlynot the case. Thus, while we do not know at this stage as towhy motility decreases with the progression of the growthcurve, it does raise the possibility that bacterial motility canbe utilized as an important parameter in bioprocessoperations, apart from the normally accepted parametersof aeration, agitation and other physiological factors. Whilecell synchronization on a biochemical level was recognizedto be of utmost importance in bioprocess engineering manyyears ago (Bailey and Ollis, 1986), we present the firstevidence that the physical parameter of cell motility may beequally important. Staropoli and Alon (2000) have reportedfor E. coli K-12 strain RP437 that run speeds of bacterialcells peak in mid-exponential phase and tumbling frequencyincreases over the course of growth in a batch culture in 125mL flasks. However, their elegant experiments were notdesigned to investigate motility of fresh cultures, i.e.substantial time was provided for biological adaptation tothe bacterial samples to undergo chemotaxis driven motilityunder various conditions while observing under themicroscope. Thus, their results were quite reasonable interms of the expected outcome of the experiments. We, onthe other hand, seem to have stumbled upon counter-intuitive results, highlighting the importance of microbialmotility in bioreactor operations, which opens a new line ofthinking that is testable for other strains of motile bacteria.As of today, motility of microbial cells is not a parameterconsidered in bioreactor design and operation. By learning

DOI 10.1002/bit

from different motility behaviors exhibited by bacterial cellsin different phases of the growth curve, it may be possible toenhance the bioprocess productivity by incorporatingstrategies to normalize chemotaxis driven motility overthe complete growth profile during an operation.

References

Bailey JE, Ollis DF. 1986. Biochemical Engineering Fundamentals, 2nd ed.

(International). Singapore: McGraw Hill Book Co.

Behkam B, Sitti M. 2005. Modelling and testing of a biomimetic flagellar

propulsion method for microscale biomedical swimming robots. Pro-

ceedings of the 2005 IEEE/ASME International Conference on

Advanced Intelligent Mechatronics, 37–42.

Berg HC. 2003. The rotary motor of bacterial flagella. Ann Rev Biochem

72:19–54.

Berg HC, Brown DA. 1972. Chemotaxis in Escherichia coli analysed by

three-dimensional tracking. Nature 239:500–504

Brown DA, Berg HC. 1974. Temporal stimulation of chemotaxis in Escher-

ichia coli. Proc Natl Acad Sci USA 71:1388–1392.

Diluzio WR, Turner L, Mayer M, Garstecki P, Weibel DB, Berg HC,

Whitesides GM. 2005. Escherichia coli swim on the right hand side.

Nature 435:1271–1274.

Gabel CV, Berg HC. 2003. The speed of the flagellar rotary motor of

Escherichia coli varies linearly with protonmotive force. Proc Natl Acad

Sci USA 100:8748–8751.

Glazer AN, Nikaido H. 1995. Microbial Biotechnology: Fundamentals of

Applied Microbiology. New York: W.H. Freeman and Company.

Gupta R, Sharma M, Mittal A. 2006. Effects of membrane tension in

nanopropeller driven bacterial motion. J Nanosci Nanotechnol 6:

3854–3862.

Kojima S, Blair DF. 2004. The bacterial flagellar motor; structure and

function of a complex molecular machine. Intl Rev Cytol 233:93–

134.

Lauga E, DiLuzio WR, Whitesides GM, Stone HA. 2006. Swimming in

circles: motion of bacteria near solid boundaries. Biophys J 90:400–

412.

Olfosson AC, Hermansson M, Elwing H. 2003. N-Acetyl-L-cysteine affects

growth, extracellular polysaccharide formation and bacterial biofilm

formation on solid surfaces. Appl Environ Microbiol 69(8):4814–

4822.

Ranade VV. 2002. Computational Flow Modelling for Chemical Reactor

Engineering. Process Systems Engineering, Vol. 5. San Diego: Academic

Press.

Sharma M, Naresh M, Mittal A. 2007. Morphological changes in magne-

totactic bacteria in presence of magnetic fields. J Biomed Nanotechnol

3:1–6.

Shuler ML, Kargi F. 2002. Bioprocess Engineering—Basic Concepts. 2nd

ed. Singapore: Pearson Education.

Sowa Y, Rowe AD, Leake MC, Yakushi T, Homma M, Ishijima A, Berry RM.

2005. Direct observation of steps in rotation of the bacterial flagellar

motor. Nature 437:916–919.

Staropoli JF, Alon U. 2000. Computerized analysis of chemotaxis at

different stages of bacterial growth. Biophys J 78:513–519.

Wadhams GH, Armitage JP. 2004. Making sense of it all: bacterial chemo-

taxis. Nat Rev Mol Cell Biol 5:1024–1037.

Wood TK, Barrios AFG, Herzberg M, Lee J. 2006. Motility influences

biofilm architecture in Escherichia coli. Appl Microbiol Biotechnol

72:361–367.

Arora et al.: Single Cell Motility in Bioreactor Operations 1649

Biotechnology and Bioengineering. DOI 10.1002/bit