A TURBIDIMETRIC METHOD OF FOLLOWING CELL. inhibition of respiration coincident with inhibition of multiplication. The

Transcription

1 A TURBIDIMETRIC METHOD OF FOLLOWING CELL MULTIPLICATION WITHIN WARBURG FLASKS Department of Bacteriology, School of Medicine, University of Pennsylvania, Philadelphia, Pa. Received for publication February 19, 1945 Recent trends in research employing multiplying cells like bacteria and yeast have led to the use of respirometers such as the Warburg. Experiments are frequently made with cell inhibitors so that a correlation is sought between inhibition of respiration coincident with inhibition of multiplication. The inhibition of respiration can easily be followed by taking periodic manometer readings during the course of the experiment and comparing respiratory values of the inhibited cells with those of the controls. Following the inhibition of division is not so simple, since all the various methods for estimating cell population available so far require opening the particular manometer, thus removing that particular manometer from the experiment. The methods available include chamber count, plate count, dilution count, and turbidimetric measurement. In order to correlate inhibition of respiration with inhibition of division over a period of time, or in fact to determine either one alone in any single instance, it is prerequisite to know with some degree of accuracy the growth curves for the cell population during the period for which the inhibition is to be calculated. This has required running as many manometers for each inhibitor concentration or control as there are to be points on the growth curve. For example, if five points are desired to construct the curve, five manometers must be started, one being removed from the experiment at each time interval and the cell population determined by one of the above-mentioned methods. This usually makes experiments difficult because of the large number of manometers required if several inhibitor concentrations are run simultaneously. Furthermore, a definite error is introduced when the respiration and division are not correlated for the identical cell population. This paper describes a turbidimetric method of following the multiplication of a cell suspension within a respirometer flask, thus circumventing the particular difficulties described above. APPARATUS The special flask was made from an ordinary Warburg flask having a single side arm. The side arm was removed and replaced by a short length of glass tubing shrunk on a square steel rod;' the end was then sealed off (figure 1). The turbidimeter was constructed by building a Bradley 20 B barrier layer cell into a small unit as shown in figure 2. Sufficient sensitivity for turbidity measurements can only be obtained if a There is no reason why this side arm should not be round. 31

2 32 to6v FIG. 1. SPECIAL FLASK 6v lamp and housing....barrier laiqer elernent..blue filter Arm of f lask inserte4i FIG. 2. PHOTOCELL UNIT ere Downloaded from Rot± B2 PHOTO single throw switcb K, = Single pole. K= Galvanoroeter ker4 G = Galvanometer P = StAdent potentiometer B, = 1'5 V &rrj cell Ba= 6V storage battertj R= 950oQ R2= 35 -a PHOTO= photoelectric cell FIG. 3. WIRING DIAGRAM FOR MEASURNG OUTPUT OF PHOTOCELL UNIT blue or green ifiter is used, since maximum light absorption is obtained in this range. The authors used a double thickness of flattened, smoothed, blue on December 16, 2018 by guest

3 TURBIDIMETRIC METHOD OF FOLLOWING CELL MULTIPLICATION cellophane. The output of the photocell was measured by a simple potentiometric setup employing apparatus found in most laboratories (figure 3). The procedure followed in using this setup is as follows: The photocell lamp is turned on several minutes before the time of the first measurement to allow 33 output of the barrier layer cell to reach its maximum and constancy of output. P is then arbitrarily set to some value (550 in these experiments), K1 closed, K2 closed, and R adjusted until the galvanometer (G) deflection is zero. K2 is IBook 41) N Si 600-' ROO Klett-Summerson Units FIG. 4. CORRELATION BETWEEN TURBIDITY READINGS BY THE Two TYPES OF APPARATUS ON MILK SUSPENSIONS A single experiment; solid line = correlation line by the method of least squares; dotted lines = dispersion lines at a distance from the correlation line equal to S. = standard error of estimate = 6.9. then released and the flask's side arm, filled with the suspension the turbidity of which is to be determined, is inserted into the photocell unit. K2 is then closed, and P adjusted to zero deflection of G. The reading of P now corresponds to the turbidity. The entire procedure of taking the manometer and flask out of the water bath, taking the turbidity measurement, and replacing the manometer and flask in the bath can easily be performed in well under a minute. If several minutes will elapse before another reading is to be taken, K1 is opened to avoid unnecessary drain on B1. It is to be noted that by this method the actual output of the photocell is not determined, but this is of no importance since

4 34 only relative turbidities are to be measured. Drifts in output of B1 and BE2 over periods of time exceeding any single determination are compensated for by the initial adjustments always made at the beginning of each determination (adjusting R2 so that G deflection is zero when photocell is at full flush and with P set at the arbitrary reading, which, as stated, was in these experiments 550). EXPERIMENTAL This study is concerned with comparing results using the special apparatus described with those obtained with the Klett-Summerson colorimeter, and is not an attempt to establish the value of the turbidimetric method. The latter is something which must be determined by the experimenter for his every individual problem, because results published by various workers comparing the turbidimetric method with other methods such as the plate count have varied considerably (Kohn and Harris, 1941; Libby, 1940). In any particular problem it is mandatory that the method used in determining cell populations give sufficiently accurate results for the interpretations which are to be based upon them. Milk Suspensions Figure 4 shows the correlation between readings by the two types of apparatus on milk suspensions. The degree of correlation is very high, the correlation coefficient being in this particular case Other single experiments gave similar values. It is seen that the standard error of estimate of Klett values from the special flask values is 6.9. Thismeans that areadingof 650,forexample, on the special apparatus is equivalent to a true Klett reading between 151 and 179,95 times out of Actually, from the graph of figure 4 the Klett equivalent of a reading of 650 on the special flask would be taken as 165, thus the error 95 times out of 100 would be less than +14 Klett units. Both methods of measurement have their own inherent errors in measuring the turbidity of any particular suspension. For example, repeated readings (30) on the same milk suspension of a Klett turbidity with a mean of gave a standard deviation on the Klett of 0.62, and on the special apparatus a standard deviation equivalent to 1.47 Klett units. This means that, in this vicinity of turbidity at least, in order for two turbidity readings to be considered significantly different (correctly so at least 95 times out of 100) they must in the case 2B2 can be a 6-volt storage battery or, if available, a constant voltage transformer. Four li-volt dry cells used in a series do not give a sufficiently constant current source. One very efficient type of design obviates the necessity of having such a constant current source. In this type the photocell would be split, one half receiving illumination through the turbid suspension, the other from the same light source unhindered. The two photocell components are balanced against each other and thus variation in light source is canceled out. The Klett-Summerson apparatus is so designed. 3 It must be remembered that these values apply only to the apparatus as set up with the arbitrary zero point on P set at 550. The accuracy, however, should not be materially affected by a change in the zero point.

5 TURBIDIMETRIC METHOD OF FOLLOWING CELL MULTIPLICATION 35 -o 0 0,e5o 150~~ 200-' Time in irnintes FIG. 5. GRowTH of Staphylococcus aureus AS MEASURED By KLETT-SummERsoN COLORIMETER (OPEN CIRCLES) AND THE, SPECIAL FLASK (CLOSED CIRCLES) I --I I I I I *50 S O~~~~~~~./ O 1S Tirmein minttes3 FIG.6. GROWTH Or Staphylococcus aureusas MEASURED BYTEmLETT-Su1MER80N COLORIMETER (OPEN CIRCLES) AND THE, SPECIAL; FLASK (CLOSED CIRCLES) 4 >~~~~ Ttme in minraaes FIG. 6. GROWTH OF Staphylococcus aureus AS MEASURED BY THEn LETT-SUMMERSON COLORIMETER (OPEN CIRCLSES) AND THE SPECIAL FLASK (CLOSED CIRCLES)

6 36 of the Klett turbidimeter differ by 1.8 units, and in the case of the special apparatus, the equivalent of 4.2 Klett units. Bacterial Suspension A strain of Staphylococcus aureus was used in these experiments and was subcultured daily. Six experiments were set up, each including the special flask and 10 usual Warburg flasks. In each flask were placed 0.2 ml of a salinephosphate suspension of Staphylococcus aureus, 2.0 ml of saline-phosphate buffer of ph 7.2, 0.5 ml of 1 per cent neopeptone, and 0.5 ml of 2 per cent glucose. The inner cup contained 0.3 ml of 20 per cent KOH to absorb the CO2 given off. The initial turbidity was the same in all flasks, as measured by the Klett equipment. The shaking rate was 130 per minute to and fro through an excursion of 6.5 cm. The water bath was kept at a constant temperature of 37.5 C. Optimal oxygenation was being obtained, as was indicated by the fact that no change in oxygen consumption occurred either when the shaking rate was varied or when the size of the initial bacterial inoculum was decreased. The oxygen consumption and turbidity in the special flask were read every 15 minutes. The last turbidity, at the end of 21 hours, was also read on the Klett apparatus. Oxygen consumption was read every 15 minutes in all the Warburg flasks, and one flask was removed every half hour for the purpose of determining the turbidity of the contents on the Klett equipment. The remaining five flasks were also removed at the end of 24 hours and all turbidities were taken. Growth results obtained from the special flask and Warburg flasks, one removed every half hour, are plotted in figures 5 and 6 for two of the six experiments.4 These are typical of all six experiments. It is seen that the points obtained by the special flask lie on a very smooth curve, whereas the points obtained by Klett measurements on the replicates are scattered to a considerably greater degree. As stated earlier, one source of error in using the replicate method for obtaining growth curves and respiratory rates of growing bacteria is the inevitable difference in values between replicates. The coefficient of variation of turbidity between 30 replicates at the end of a 24-hour period, as measured by the Klett colorimeter, was 3.3 per cent. The coefficient of variation of total oxygen consumption between the same replicates at the end of this period was 3.9 per 4 Growth here is reported as Klett turbidity units. It is assumed that turbidity varies in direct proportion to the mass of bacteria. It is not the purpose of this paper to test this relationship but merely to compare two methods of measuring turbidity; therefore, whether this relationship is linear or not is immaterial. However, in studies where Qos values are determined, it is highly advisable that the whole range of turbidimetric readings be standardized against bacterial nitrogen. In order to put the growth curves obtained on the same grid, the special flask turbidity units were transposed to equivalent Klett units by means of a correlation graph drawn up similarly to figure 4, but constructed from points obtained with suspensions of the bacteria used in these experiments instead of milk.

7 TURBIDIMETRIC METHOD OF FOLLOWING CELL MULTIPLICATION cent. This means that 95 times out of 100 when the difference of turbidity between two flasks at the end of 21 hours exceeds 9.3 per cent, the difference is not from error of random sampling, hence such differences may be considered significant; in the case of total respiration the values would have to differ by 11.0 per cent. DISCUSSION In order to get some idea of the relative accuracy of the two methods, the 02 consumed per turbidity unit per 1 hour was calculated for each method with its standard deviation. Since six experiments were run, each with five 1-hour periods, 30 values for each method could be calculated, a sample sufficiently large for statistical analysis. A word must be said with regard to the calculation of 02 consumed per turbidity unit per 1 hour. Let us take a hypothetical case. Suppose at the beginning of the 1-hour period the turbidity reading were 100 and at the end of the 1-hour period the turbidity were 200. If the 02 consumed during this time were 300 mm3, the number of bacteria which consumed this amount of 02 certainly would not correspond to the turbidity of 200, because the only time during the i hour that this number was present was at the very end. Obviously, therefore, if the 02 consumed per bacterium (or other standard unit) is desired, the 02 consumed for the 1-hour period must be divided by the average number of bacteria existing during that period of time. If the growth line is a curve and its equation is known, then the average for the time period can be found by calculus.5 If the line is straight, then the average is the mid-point of that segment of the line corresponding to the particular period of time. In calculating the 02 consumed per turbidity unit per 1 hour in the special flask, the average of the turbidity units for each 1-hour period was taken as the 4-hour turbidity reading. Very little error is introduced by this assumption because the points are only 1 hour apart and because the growth curve does not deviate to a great degree from a straight line. In the case of the curves drawn from the Klett readings, the average was taken as the mid-point of a line drawn between the reading at the beginning of the 1-hour and the reading at the end. It is admitted that in actual practice one would not connect scattered points by broken straight lines, but rather would attempt to draw a smooth curve which would be a good approximation. These calculations, however, are being made to reveal the error in the method, therefore the hypothesis is made that 5 Equations of growth curves could only be obtained with considerable difficulty. In cases of growth lines which are considerably curved the following relatively simple method can be employed to arrive at an approximation of the average number of bacteria existing during any period of time. The growth curve is plotted on graph paper with small squares. The number of squares underneath the growth curve between the limits of the time period for which the value is to be calculated are counted, and this number is divided by the number of squares corresponding to the length of the time period. This is an approximation of the average number of squares corresponding to the turbidity units for that particular period of time. The value desired is obtained merely by multiplying this number of squares by the number of turbidity units represented by each square. 37

8 930 each point is a true one. The 02 consumed per turbidity unit per j hour in the special flask had a coefficient of variation of 9.0 per cent, and for the Mett measurements on replicates a coefficient of variation of per cent. Thus a difference of 25 per cent would have to exist between two respiratory values by the special flask method and 35 per cent by the Mlett method before the difference could be considered significant (correctly at least 95 times out of 100). Naturally, this turbidimetric method involving the use of the special flask is not without its limitations and will not be suitable for all types of experiments. For example, it cannot be used if it is impossible to obtain a uniform suspension within the side arm merely by shaking. Some organisms, such as pneumococci, streptococci, and staphylococci may form a thick and slimy growth, others may agglutinate; and in such cases the contents must be removed and broken up before a turbidity measurement can be taken. SUMMARY A new method is described whereby turbidity measurements can be made on suspensions within a modified Warburg respirometer flask. The growth of cells such as bacteria and yeast can thus be followed simultaneously with respiratory measurements, both measurements being made on identical cell populations. Experiments and statistical analysis of them show that this method is at least as accurate as the method requig replicates, the turbidities of which are read on a colorimeter such as the Klett-Summerson. Considerable work is saved by making replicates unnecessary. REFERENCES KOUN, H. I., AND HARRIS, J. S On the mode of action of the sulfonamides. I. Action on Eacherichia coli. J. Pharmacol., 73, LIBBY, R. L The activity of chemotherapeutic agents. J. Bact., 40,