PERSISTERS AND MUTATION RATE AT EXPONENTIAL GROWTH

O FE R F RI D M AN

HE B R EW U N I VE R S I T Y

Abbreviated Title: Persisters and Mutation Rate Word Count*: 1206 Figures and Tables: 4 Last Date of Revision: 22 October 2007

INTRODUCTION Over the last years new antibacterial research programmes are being stunted by the fast evolving pace of antibiotic resistance (Leeb, 2004). This process produces a need for understanding the mechanism of antibiotic resistance. Persister cells, which were not simply antibiotic-resistant mutants, inevitably survived and it was proposed that these were dormant, non-dividing cells (Lewis, 2007). Preliminary results from an experiment that is beyond the scope of this paper are showing a correlation between antibiotic resistance and persistency phenomena. These experiments were using two types of bacteria (E. coli) producing different amounts of persister cells, MGY and MGYA7†. MGY produces a low amount of persister cells, while MGYA7 produces a much higher amount. In order to test whether the correlation observed between persistence and resistance is due to a higher mutation rate in MGYA7, and not directly linked to its high persistence, we compared the mutation rates of MGYA7 and MGY. For that, we used a known protocol (Luria, S. E., Delbruck, M., 1943) that enables measuring the mutation rate in a culture of exponentially growing bacteria. In those conditions, a low number of persisters is present in the culture (I Keren, N Kaldalu, A Spoering, Y Wang, K Lewis, 2004). In this paper, we have shown that both strains MGYA7 have the same mutation rate. This result rules out that the correlation observed between persistence and resistance is due to an increase in the mutation rate of exponentially growing cells.

* †

Word Count is of all pages, including Title Page, Abstract, References, Figures and Tables MGYA7 is a mutation of MGY producing more persister cells.† MGYA7

Fridman 2007 Persisters and Mutation Rate

MATERIALS AND METHODS The inherent difficulty in measuring the mutation rate of bacteria has been described in (Luria, S. E., Delbruck, M., 1943). When one attempts to measure directly the total number of mutants in a cultures, large variations are observed, as seen in figure 2 (more then two orders of magnitude variation). Therefore, counting the mutant bacteria CFU’s by just plating them will require many repetitions in order to estimate the mutation rate. We used the P0 approach (Luria, S. E., Delbruck, M., 1943), which provides much more accurate results, as can be seen in “Monte Carlo” simulation (figure 3).

EXPERIMENTAL DESIGN In this assay, the E. coli. (MGY, MGYA7) population was taken from an exponential growing culture. In order to avoid irregularity, 1k bacteria were placed in each well (we used 0.1 ml of LB in 96 plates). Then, the plates were placed in incubation at 37⁰C, until the number of bacteria reached the expected amount (approximately 108-109/ml). The number of bacteria was checked using OD, and later it was validated by plating and counting the CFUs. At that stage, a standard amount of Riff antibiotic (0.15 [mgr/ml) with 0.1 [ml] LB) was added to each well. Then, the plates were placed again in incubation at 37⁰C, for more then one day, until there was a significant difference between the empty and the full wells. This difference was measured using OD and could be seen with the naked eye. When the results were not significant, plating on Riff plates was used in order to discriminate between wells with or without resistant mutants.

DATA ANALYSIS

Raw data are presented in Table1.

e − µN (ηN ) ). We k! k

The mutation rate can be calculated using the Poisson distribution ( f (k , µN ) = used the probability for no mutation, P0 (µ , N ) = e −ηN hence µ = − full wells.

2

ln(P0 ) where P0 is the fraction of N

Fridman 2007 Persisters and Mutation Rate

RESULTS E. COLI. MGY AND MGYA7 MUTATION RATE This study uses the probability for no mutation to calculate the mutation rate of E. coli. (MGYA7, MGY), assuming a Poisson distribution. Our results show that the mutation rates in MGYA7 and MGY, in an exponential growing environment, are similar. The results show that both populations have a mutation rate of µ = 1.5 ∗ 10 −8 ± 7 ∗ 10 −9 , in agreement with typical reported values for E.coli (D. W. Kerry, J. M. T. Hamilton-Miller and W. Brumfitt, 1975) It may be concluded that no significant difference exists in mutation rates between our two strains , when grown exponentially (see table 1). The variation in the results is a consequence of the random nature of the phenomena. The relative error deriving from the P0 method is ln −1 (m / n) ⋅ ∆m m , while the error caused by the evaluation of the number of bacteria is ∆N N , where N is the number of bacteria, m is the number of empty wells and n is the number of all wells.

DISCUSSION As can be seen from the results, when grown in exponential phase, there is no significant difference in mutation rate between the two genotypes. When dealing with mutation rate, random fluctuation plays a significant part. To test the inherited variation derived from the phenomena, “Monte Carlo” simulation of the experiment was used. The results (shown at figures 3) show the diversion with no sampling error. The standard deviation is only half from the standard deviation in our results. Considering data showing that in exponential growing there is only a very low number of persisters (I Keren, N Kaldalu, A Spoering, Y Wang, K Lewis, 2004), our results may indicate a relation between the existence of persister cells and the high mutation rate among MGYA7 Cells. Further examination is needed to determine whether the standard deviation of the results is enough to explain the preliminary results described here. To enhance precision, the procedure used here can be scaled up by performing more experiments, or adding more wells for the P0 calculation. Assuming that the difference between the genotypes does not result in a change in the mutation rate, we should increase the number of persister cells and then check the mutation rate. If there is a significant difference, this might explain the preliminary results described above, and even entail adjustments in the usage of antibiotics.

3

Fridman 2007 Persisters and Mutation Rate

BIBLIOGRAPHY D. W. Kerry, J. M. T. Hamilton-Miller and W. Brumfitt. (1975). Trimethoprim and rifampicin: in vitro activities separately and in combination. Journal of Antimicrobial Chemotherapy (1), 417-427. I Keren, N Kaldalu, A Spoering, Y Wang, K Lewis. (2004). FEMS Microbiology Letters , 230 (1), 13-18. Leeb, M. (2004). Antibiotics: A shot in the arm. NATURE -LONDON- (7011), 892-893. Lewis, K. (2007). Persister cells, dormancy and infectious disease. Microbiol , 5, 48–56. Luria, S. E., Delbruck, M. (1943). MUTATIONS OF BACTERIA FROM VIRUS SENSITIVITY TO VIRUS RESISTANCE. Genetics , 28, 491-511.

4

Fridman 2007 Persisters and Mutation Rate

FIGURES AND TABLES

TABLE 1 Strain MG MGYA7 MGYA7 MGYA7 MGYA7 MGYA7 MGY mgy mgy mgy

Empty Wells 61 77 89 49 84 49 61 21 86 54

Number of wells 86 89 95 96 96 96 86 95 96 96

Number of bacteria [#/ml] 2.88E+08 5.99E+07 4.09E+07 5.69E+08 6.78E+07 9.42E+08 3.97E+07 1.53E+09 3.40E+07 5.97E+08

Calculated mutation ratio - µ 1.19E-08 2.42E-08 1.59E-08 1.18E-08 1.97E-08 7.14E-09 1.63E-08 9.87E-09 3.23E-08 9.64E-09

5

∆µ/µ 9.5% 17.9% 34.4% 6.1% 17.8% 6.1% 9.5% 6.3% 21.1% 6.4%

Fridman 2007 Persisters and Mutation Rate

FIGURE 1

Figure 1 - the mutation rate versus the number of bacteria/ml at which the experiment took place. No correlation between strain and mutation rate can be seen.

6

Fridman 2007 Persisters and Mutation Rate

FIGURE 2 Sample results in “Monte Carlo” simulation for 1000 experiments using µ=1.5e-008 Number of mutant cells

6

Histogram of number of mutant cells after 25 cycles 300

Number of mutant cells per experiment

10

5

10

250

4

10

200

3

10

150

2

10

100

1

10

50

0

10

Mean: 8.9e+003 Standard Deviation: 1.6e+004

5

10

15 cycle

20

0

25

0

2

4 6 8 log(Number Of Mutant)*10

10

Figure 2-“Monte Carlo” simulation of 1000 experiments, showing the large variation in number of resistant cells. Therefore, counting the mutant bacteria CFU’s by just plating will require many repetitions in order to estimate the mutation rate.

7

12

Fridman 2007 Persisters and Mutation Rate

FIGURE 3 Sample results in “Monte Carlo” simulation for 1000 experiments 100 90 Standard Deviation: 3.3e-009 Mutation Ratio: 1.5e-008

80

Number of Resolts

70 60 50 40 30 20 10 0

0

0.5

1

1.5 2 Calculated Mutation Ratio( µ)

2.5

3

3.5 -8

x 10

Figure 3- “Monte Carlo” simulation of 1000 experiments mimicking our experimental setup, showing the expected results using the P0 method. As can be seen, although the variation is large, the mutation rate calculation is accurate (STD is 22% from the result.)

8