J Clin Pathol 1998;51:623–628 623

Measurement of of , J Clin Pathol: first published as 10.1136/jcp.51.8.623 on 1 August 1998. Downloaded from , and coli by real time computer tracking using the Hobson BacTracker

Q N Karim, R P H Logan, J Puels, A Karnholz, M L Worku

Abstract , and the slightly slower growth of Aims—(1) To make precise measurements motile flagellate cells would ensure the loss of and comparisons of various aspects of the motile phenotype if it did not confer some motility of three gastrointestinal patho- substantial benefit to the cell, as explained by gens, Helicobacter pylori, Campylobacter Armitage.1 Microbial motility can be assessed jejuni, and , in log phase microscopically by hanging drop, by phase growth; (2) to provide background infor- contrast and dark ground illumination, or mation on motility data to study the influ- macroscopically, by growth in semisolid agar, ence of pH, viscosity, and chemotactic swarming on surface of solid agar, or by the factors, thereby gaining a better under- crossing of a filter paper bridge over a trench standing of bacterial pathogenesis. cut in solid agar. Methods—Computer image processing These methods distinguish motile from non- technology and phase contrast micros- motile and assess motility only subjec- copy (Hobson BacTracker) were used to tively, though giving the impression that some measure several indices of bacterial mo- bacteria move faster than others. The methods tility in real time. Ten clinical isolates of are not precise enough to follow transient or each species in log phase liquid culture minor changes in motility. Video recording of were studied. motile bacteria has provided a means of quan- Results—C jejuni moved fastest, with a tifying motility. The video recorded tapes have median curvilinear velocity (CLV) of 38.76 been replayed, and motility measured manually µm/s (range 29.08 to 52.82). Next was H on the screen by planimeter and stop watch, as pylori, median CLV 25.02 µm/s (range described by Ferrero and Lee.2 Such manual 12.07 to 29.07). E coli was the slowest, measurements are cumbersome and diYcult to median CLV 12.73 µm/s (range 8.20 to do on large numbers of bacteria. http://jcp.bmj.com/ 18.04). The straight line velocities showed These problems can be overcome by the similar trends. Measurement of track lin- Hobson BacTracker (Hobson Tracking Sys- earity (TL) showed that C jejuni moved tem, SheYeld, UK), which allows precise and the straightest (TL 60.3%), H pylori objective measurements of motility. It is a moved in wide circles (TL 28.7%), and E measuring system that incorporates unique coli showed spinning movement without “blob and track” image processing technology, Department of much linear displacement (TL 18.3%). analysing bacterial movement in real time. The , Imperial There were significant diVerences in these tracker comprises a phase contrast microscope on September 30, 2021 by guest. Protected copyright. College of Science, three variables between the species stud- connected to a video camera. The camera is Technology and ied, but no significant diVerences in linked to a video recorder, which is connected Medicine at St Mary’s measurements of time and frequency of to a computer that tracks 120 moving bacteria Hospital, London W2, halts between movement runs. on screen simultaneously and continuously. UK Q N Karim Conclusions—The BacTracker provides a The results are displayed as histograms or trail J Puels useful technical aid for measuring several draws. This system records several indices of A Karnholz indices of objectively, motility including direction, curvature rates, M L Worku reproducibly, and precisely, which is diY- curvilinear velocity, and straight line velocity, cult to achieve without computer assist- which can be measured accurately, objectively, Division of ance. Accurate quantification of motility and reproducibly so that comparisons can be Gastroenterology, Queens Medical provides a basis for studying the factors made under diVerent experimental conditions. Centre, Nottingham, which influence bacterial motility. It can Recordings of movement can be stored in the UK provide phenotypic measurements of the computer and can be downloaded for more R P H Logan eVect of flagellar depletion. detailed statistical analysis. (J Clin Pathol 1998;51:623–628) Real time computer tracking has been used Correspondence to: to assess the motility of Rhodobacter sphaeroides, Dr Q N Karim, Senior Keywords: bacterial motility measurement; computer Lecturer, Department of image processing technology; bacterial tracking Rhodospirullum rubrum, and Microbiology, Imperial typhimurium.3 The aim of our present study was College of Science Technology and Medicine at to assess the motility of Helicobacter pylori (H St Mary’s Hospital Medical For many bacteria, motility confers a survival pylori), Campylobacter jejuni (C jejuni), and School, Norfolk Place, advantage by permitting migration towards a Escherichia coli (E coli). These are major patho- London W2 1PG, UK. favourable microenvironment, or away from an gens infecting the upper and lower gastrointes- Accepted for publication unfavourable one. The operation of motility tinal tract, in which motility is acknowledged to 16 April 1998 machinery is very energy expensive for micro- be a factor.2 4–6 Motility measure- 624 Karim, Logan, Puels, et al

pears from view or moves out of the analysis window (fig 1). J Clin Pathol: first published as 10.1136/jcp.51.8.623 on 1 August 1998. Downloaded from Stop—A stop occurs when the speed of the bacterial cell falls below the stop speed defined in the setting screen. This was set at the speed shown by Brownian movement of dead bacte- ria. Run—A run is the track between two stops (fig 1). Curvilinear velocity (CLV)—This is the length of a track divided by the time taken to travel it. It is calculated by summing the incre- mental distances moved in each frame along the sampled path and divided by the total time for the track. It is measured for tracks (total path length) and for runs (incremental path lengths between two stops) in µm/s (fig 2). Straight line velocity (SLV)—This is calcu- lated by measuring the straight line distance between the start and end point of the track and dividing by the time taken to travel it. It is measured in µm/s (fig 2). Track linearity percentage (TL%)—This is the ratio of the straight line velocity to curvilinear velocity × 100 (SLV/CLV(100)). For a bacte- Figure 1 Quantities analysed by the BacTracker. rium that runs straight this value is 100%. The ments of these bacteria in log phase growth can more curved the route taken by the bacterium form the basis for studying and comparing the greater will be the curvilinear velocity, and changes that take place because of diVerent the lower the straight line velocity and the value environmental factors, such as phases of of track linearity percentage. For a bacterium growth, pH, temperature, and the presence of that spins around a point the straight line agents. These measurements can velocity may be so small compared with the be used to study and help us to curvilinear velocity that the value may ap- understand this aspect of bacterial physiology proach 0%. Thus, depending on the straight- and pathogenesis. ness or curvature of the path of the bacterium, the value of track linearity will be between 100% and 0%. This is shown diagrammatically Methods in fig 2. http://jcp.bmj.com/ THE HOBSON BACTRACKER Curvature rate (CVRT+/s)—This is meas- The BacTracker separates the moving parts of ured using the incremental sum of change in the image from the static parts. The moving angle as the bacterium changes direction for object is then filtered and a threshold imposed the length of the track. It includes the sign to to produce coloured “blobs” which are super- reflect the direction of change. For each track imposed on the moving objects on the black point the signed change in angle from the pre- and white video image. Once the moving vious track point is measured. The sum of the objects have been identified the BacTracker has signed change of angle is accumulated along on September 30, 2021 by guest. Protected copyright. the coordinates of every moving object in the the track. At the end of the track the sum of the frame and with this information various meas- signed angle change is divided by the time for urements can be made. The BacTracker is a the track to give a value for the curvature rate in real time image processing system where degrees per second. For bacteria with net anti- motility measurements are carried out as they clockwise movement the value is positive, and are occurring—the computer measures con- for a bacteria with net clockwise the value is tinuously and does not have to stop, calculate, negative. This is shown as direction change and measure as in snapshot measurements (DRCH) (fig 1). made over a short period of time. Various Stop time (STTM)—This is the time of a measurements are displayed on the computer defined stop between two adjacent runs. The screen, such as summary graphs and histo- average of all the stop time is displayed in sec- grams, while the tracking screen shows the cells onds (fig 1). as they are tracked. Thus the system can be Stop frequency (STFRQ)—This is a measure validated while watching in real time. The of how often the cell stops. The time is video images are processed at 25 Hz in PAL measured from the start of a run through to the format, or 30 Hz or 6 Hz in NTSC North end of the following stop or the start of a new American format. run. This time is divided into 1 to give a frequency in Hz or times/s. A value is displayed VARIABLES MEASURED as an average of each track. The following variables were measured by the Limits—When bacteria travel at a varying BacTracker: velocity, the BacTracker measures the median Track—A track is the path travelled by a velocity and the histogram shows the distribu- moving cell. It is measured from the point of tion of bacteria with diVerent velocities. By detection by the computer until the cell disap- setting the “limits” values to a high or low Measurement of motility by real time computer tracking 625

velocity one can include or exclude the fast or microaerophilic environment using gas packs slow moving bacteria and preferentially study (Campypak, BBL, Maryland, USA) in gas jars J Clin Pathol: first published as 10.1136/jcp.51.8.623 on 1 August 1998. Downloaded from the fast or the slow bacteria in a mixed popu- (Oxoid, Basingstoke, UK). The colonies were lation. identified by typical appearance on Gram film and . The H pylori were isolated by OF BACTERIA plating the antral biopsies on Columbia agar We used 10 isolates of each species of E coli, H (BBL) with 7% horse (TSCL, Bucking- pylori, and C jejuni, obtained from specimens ham, UK) made selective by adding received for diagnostic purposes. Reference supplements of 20 mg/l , 2 mg/l strains H pylori NCTC 11637, E coli NCTC amphotericin B, and 3 mg/l vancomycin. The 10413, and a clinical isolate of C jejuni used as a plates were incubated for 72 hours at 37°Cina laboratory control were also included. The E coli microaerophilic environment using Campypaks strains were isolated from patients with bacter- as above. Only one subculture was made from aemia or urinary tract . They were plates that showed heavy confluent growth on identified by standard laboratory procedures primary culture. This subculture was incubated using the API 20 E system (API, Lyon, France). for 48 hours as above and used for motility The C jejuni were isolated by culturing stool on studies. Only the first subculture was used for Skirrow’s media7 at 42°C for 24 hours in a this study. This was to avoid repeated subcul- http://jcp.bmj.com/ on September 30, 2021 by guest. Protected copyright.

Figure 2 Left panels: computer tracking of 5 second trail draws in real time of (A) C jejuni, (B) H pylori, and (C) E coli. Right panels: Diagrammatic representation of relation between straight line velocity (SLV),curvilinear velocity (CLV), and track linearity (TL%). Track linearity values: (A) 83%; (B) 44%; (C) 14%. 626 Karim, Logan, Puels, et al

Table 1 Comparison of diVerent indices of mobility of E coli, H pylori, and C jejuni. Ten isolates were examined for each

species J Clin Pathol: first published as 10.1136/jcp.51.8.623 on 1 August 1998. Downloaded from

Measurement E coli (p Value) H pylori (p Value) C jejuni

Curvilinear velocity (µm/s) 12.73 (8.20 to 18.04) 25.02 (12.47 to 29.07) 38.76 (29.08 to 52.82) 0.0036 0.0008

Straight line velocity (µm/s) 2.27 (0.16 to 9.15) 7.2 (1.05 to 14.31) 23.41 (8.00 to 26.00) 0.0113 0.0022

Track linearity (%) 18.3 28.7 60.3 0.0173 0.0376

Curvature rate (°/s) +117.8 +157.2 +251.2 0.1212 0.0140

Stop frequency/s (Hz) 1.05 (0.29 to 1.33) 1.25 (0.82 to 1.61) 1.07 (0.89 to 1.27) 0.0890 0.0890

Stop time (s) 0.44 (0.29 to 0.65) 0.67 (0.26 to 0.53) 3.05 (0.24 to 0.41) 0.256 0.9097

Values are medians (range).

tures as many strains of H pylori became coccoid eter of 100 µm (Camlab VD/5010 050, and non-motile after three to four subcultures. Cambridge, UK) and observed under phase contrast microscope at 40× magnification on a BACTERIAL PREPARATION stage slide heater (MS100 Linkam Scientific Growth curves were determined in liquid Instruments, Surrey, UK) at 37°C. The ends of cultures to find the inoculum to be used and the the slides were sealed with vinyl plastic putty to period of incubation required to attain the log avoid drifts during measurements of motility. phase of growth. Our preliminary experiments The BacTracker was set to measure 100 tracks showed that H pylori moved fastest in their log for each bacterium, and calibrated for 40× phase,8 although this may not be the case with E magnification for the motility measurements. coli, as explained in the discussion. E coli and C Brownian movements were excluded by track- jejuni were inoculated in peptone water, and H ing and measuring formalin killed bacteria; this pylori in brain–heart infusion (BHI) broth corresponded to a curvilinear velocity of up to enriched with newborn calf serum and the anti- 5.0 µm/s and a straight line velocity of up to 2.0 biotic supplements detailed above. The peptone µm/s. water inoculated with C jejuni and BHI broth inoculated with H pylori were distributed in 5 STATISTICS ml volumes in 60 ml plastic flat bottomed screw The results were analysed by the Minitab capped containers. The caps were loosened and statistical package. The distributions of the the cultures incubated in an upright position various indices of motility were non- http://jcp.bmj.com/

without agitation in an atmosphere of 8% CO2 parametric, so the Mann–Whitney U test was in air at 37°C for 48 hours for C jejuni and H used to determine statistical significance. pylori. The peptone water inoculated with E coli was incubated overnight at 37°C in ordinary Results atmosphere. Immediately after inoculation and Of the three species studied, C jejuni moved at timed intervals, optical densities and viable fastest with a median curvilinear velocity of counts were performed for each of the isolates 38.76 µm/s, H pylori were next with a median of on September 30, 2021 by guest. Protected copyright. for the period of incubation. All the optical 25.02 µm/s, and E coli were slowest with a densities were measured at 540 nm. median of 12.37 µm/s (table 1). Using the Plate cultures of the respective bacteria were Mann–Whitney U test, the diVerences in used for inoculating bacterial suspensions for motility were statistically significant between E motility measurements. This was to ensure coli, H pylori, and C jejuni. The straight line purity of culture at the point of inoculation to velocity also showed the same ranking, the the liquid media. The dilutions of the bacterial median values being 23.41 µm/s for C jejuni, suspensions were adjusted on the basis of their 7.20 µm/s for H pylori, and 2.27 µm/s for E coli. optical densities to correspond to 106 colony These diVerences were statistically significant forming units (CFU)/ml, as determined by the between the three species (table 1). growth curve experiments; 500 µl of this Assessment of tracking linearity showed that inoculum were added to 5 ml of the corre- C jejuni moved the straightest, with the highest sponding liquid cultures and incubated as TL value of 60.3%. H pylori moved in wide cir- described above. Motility was measured after cles with a TL value of 28.7%, and E coli had a three hours for E coli and after 24 hours for C spinning movement, with the lowest value TL jejuni and H pylori, when they had attained the value (18.3%). This is shown in their logarithmic phase following static incubation. representative trail draws (fig 2). This was confirmed by measuring serial optical The comparative values and distribution of density values and viable counts at timed inter- curvilinear velocity, straight line velocity, and vals during this incubation. tracking linearity of the individual strains of the three genera are shown in fig 3. The corre- MICROSCOPY sponding curvature rate, reflecting the signed Each of the liquid cultures was drawn into rec- change in angle, was +117.8° for E coli, tangular capillary microslides of internal diam- +157.2° for H pylori, and +251.2° for C jejuni. Measurement of motility by real time computer tracking 627

60 coli showed the most variation in motility measurements. Such strain variation was ob- J Clin Pathol: first published as 10.1136/jcp.51.8.623 on 1 August 1998. Downloaded from 50 served less with H pylori, which was more aVected by the phase of growth. C jejuni 40 showed the least strain variation as far as 30 motility was concerned. The relevance of these variations in terms of bacterial physiology and

CLV ( µ m/s) CLV 20 disease potential has to be explored. The 10 motility of each of the three species was studied in one medium only; further studies in different 0 media are required to examine the influence of E coli H pylori C jejuni media constituents on bacterial motility. 30 The Hobson BacTracker allows precise, objective measurements of motility, where the 25 data can be stored and analysed. Its particular advantage is its ability to study large numbers of 20 bacteria in a very short time, which facilitates 15 statistical analysis. In a typical sample it is pos- sible to measure 1000 bacteria in approximately 10 40 seconds.9 The technique is not prone to sub- SLV ( µ m/s) SLV 5 jective errors, is reproducible, and allows many diVerent variables to be studied simultaneously. 0 This is not possible without computer aid, which makes a valuable contribution to the E coli H pylori C jejuni measurement of bacterial motility. However, the benefit of these various measurements 70 needs to be evaluated and not all may be 60 relevant to a particular set of experiments. The 50 limitations are that, although individual bacte- ria are tracked, in any population the final result 40 is a mean or a median value of all the bacteria 30 tracked. This is therefore dominated by the TL (%) measurements contributed by the majority of 20 the population. and one has to depend on the 10 histograms to see the distribution of individual 0 measurements of the bacterium in relation to E coli H pylori C jejuni the whole population. However, one can selec- Figure 3 Distribution of the median values of curvilinear tively study the faster or slower bacteria by pre- http://jcp.bmj.com/ velocity (CLV),straight line velocity (SLV),and track setting the “limits” of individual measurements. linearity (TL) for each of the 10 strains of E coli, H pylori, and C jejuni. Filled circles are reference strains. To study a single bacterium, then, the specimen has to be so diluted that only one bacterium is The diVerence between H pylori and C jejuni visible in the field. was statistically significant but not the diVer- The ease of measuring bacterial motility ence between E coli and H pylori (table 1). The enabled us to study the comparative motility of overall movement of all three bacterial species these three genera that infect the gastrointesti- was positive or anticlockwise. nal tract. We saw diVerent behaviour in the on September 30, 2021 by guest. Protected copyright. The median stop frequency was similar for movement of these organisms in their log phase the three groups, being 1.05 for E coli, 1.25 for of growth in liquid culture. From conventional H pylori, and 1.07 for C jejuni; the diVerences in motility studies we know that bacteria are these values were not statistically significant motile, and routine microscopy gives the between the three species. The median stop impression that some are more motile than times for the corresponding groups were 0.44 others, but the impression is subjective and seconds for E coli, 0.67 seconds for H pylori, cannot easily be quantified. We observed and and 3.05 seconds for C jejuni. There was no quantified diVerences in the motility of the tree statistical diVerence in these variables between bacterial species, C jejuni being much faster the three species (table 1). than H pylori, which in turn are faster than E coli. The measurements were performed in the Discussion log phase, where many bacteria move quickly. Accurate measurements are an important This is supported by published work showing requirement for studying bacterial motility. that in general the best flagellation occurs in Only with accurate measurements can tran- the logarithmic phase,10 and that H pylori pro- sient or subtle changes in motility under the duces 20 times more flagellar in the influence of chemicals and varied growth con- logarithmic phase.11 However, this is not the ditions be studied. These factors warrant case for E coli, which is known to control further investigation, and the convenience of expression of flagella in part by catabolite quantifying motility should aid such studies. repression, and motility is generally lower in Precise measurements are also needed for the logarithmic phase. This may be the measuring bacterial chemotaxis. Bacteria vary explanation of the observation in our experi- in their motility in diVerent phases of growth.8 ment that C jejuni and H pylori moved much Among the 10 strains of each species studied, E faster in the logarithmic phase than E coli. Fur- 628 Karim, Logan, Puels, et al

ther studies are required to establish the This study was supported by grants received from the University

of London and Abbott laboratories. We are indebted to Profes- J Clin Pathol: first published as 10.1136/jcp.51.8.623 on 1 August 1998. Downloaded from relation of motility with growth phase in other sor G Hobson and H Roe for their help in setting up the bacterial species. E coli also showed more spin- BacTracker program, to Dawn Beaumont who helped with the ning movements than translational move- statistics, to Mr Jon White for medical illustrations, and to DrJHBaronandDrJJMisiwicz for their constant support ments, whereas the H pylori generally moved in encouragement and for providing us with clinical material. wide circles and C jejuni movement was straighter. We do not know the significance of these dif- 1 Armitage J P. Behavioural responses in bacteria. Annu Rev Physiol 1992;54:638–714. ferent patterns of movements. Since subtle 2 Ferrero RL, Lee A. Motility of Campylobacter jejuni in a changes can be measured continuously over a viscous environment: comparison with conventional rod- shaped bacteria. J Gen Microbiol 1988;134:53–9. long period, the technique described should be 3 Poole PS, Sinclair DR, Armitage JP. Real time computer useful in studying chemotaxis. It will also pro- tracking of free-swimming and tethered rotating cells. Anal vide a phenotypic basis for the study of mutants Biochem 1988;175:52–8. 4 Eaton KA, Morgan DR, Krakowa S. Motility as a factor in where the flagellar have been partially or the colonisation of gnotobiotic piglets by Helicobacter completely deleted, and the eVect quantified in pylori. J Med Microbiol 1992;37:123–7. terms of alteration of motility. We have 5 Eaton KA, Morgan DR, Krakowa S. Campylobacter pylori virulence factors in gnotobiotic piglets. Infect Immun 1989; observed lack of motility of aflagellate mutants 57:1119–25. of proteus species compared with the actively 6 Burke DA, Axon AT. Adhesive E coli in inflammatory bowel motile wild type in previous experiments. Our disease and infective diarrhoea. BMJ 1998;297:102–4. 7 Skirrow’s campylobacter media. In: Mackie and McCart- preliminary studies have shown that, while dif- ney, eds. Practical medical microbiology, 14th ed. Edinburgh: ferent genera show similar changes in motility Churchill Livingstone, 1996:444. 8 Mandelstam J, McQuillen K, Dawes I, eds. The biochemistry on exposure to varying pH, their motility of , 3rd ed, section 2: Cells and populations. patterns vary widely in the face of altered Oxford: Blackwell Scientific Publications, 0000:15–18. viscosity and in the diVerent phases of growth. 9 Karim QN, Ashton L, Sidebotham RL, et al. Measurement of Helicobacter pylori motility in diVerent growth phases in We are carrying out further work on these conventional broth cultures by real time computer tracking aspects. Some species showed greater strain [abstr]. J Med Microbiol 1996;44:X. variation, which could have clinical signifi- 10 Kodaka H, Armfield AY, Lombard GL, et al. Practical pro- cedure for demonstrating bacterial flagella. J Clin Microbiol cance. These bacteria need to be studied under 1982;16:948–52. diVerent experimental conditions to evaluate 11 Josnhans C, Labigne A, Suerbaum S. Reporter gene analy- sis shows that expression of both H pylori flagellins is the significance of their varied motility in terms dependent upon the growth phase [abstr]. Gut 1995; of their physiology and pathogenesis. 37(suppl 1):264. http://jcp.bmj.com/ on September 30, 2021 by guest. Protected copyright.