Sow wisely!

Ordered seeding and poissoner – a high-precision technique of quantitative microbiologyJournal "Medicine of the XXI century" No. 2-2008, pp. 92-97

N. N. Khromov-Borisov ¹, Jenifer Saffi ², João A. P. Henriques ^{21 Department of Physiology of the Medical Faculty of St. Petersburg State University.}

Email: Nikita.KhromovBorisov@gmail.com2Centro de Biotecnologia da Universidade Federal do
Rio Grande do Sul, Porto Alegre, RS, Brasil. E-mail: jenifer@orion.ufrgs.
br and pegas@cbiot.ufrgs.br

Introduction

The quality of the raw data is one of the fundamental and desirable requirements of Good Laboratory Practice (GLP) and Good Statistical Practice (GSP). In quantitative microbiology, this means that the counts of cells and (or) colonies must obey the "ideal" Poisson distribution [1]. Since the classical works of Student [2] and Fischer et al. [3], it has become obvious that under "ideal conditions" cell counts performed using a hemocytometer (Goryaev chamber) or colony counts on Petri dishes vary according to the Poisson distribution. When these conditions are met, the calculated average value is a direct measure of the density in the studied microbial population, and such estimates are the most accurate. "Any significant deviation from the theoretical distribution [Poisson–NH distribution] is a signal that the estimate of the average value may be completely unreliable" [3].

Modern studies show that counts of microbial colonies on Petri dishes produced, for example, in genetic toxicology (Eames test) [4-6] or in food and drug quality control [7] too often (in more than 50% of cases) exhibit significant extra-Poisson variation with excessively increased dispersion (overdispersion). This phenomenon remains a serious problem in quantitative microbiology. Analytical statisticians are trying to solve this problem by inventing sophisticated probabilistic-statistical models and appropriate criteria for identifying this phenomenon. However, the introduction of additional parameters into such models complicates data analysis and reduces its efficiency and reliability (see, for example, papers [5-8] and references therein). The problem is so serious that German microbiologists and statisticians proposed to introduce non-Poisson models into the national GLP standards for microbiological (epidemiological) control of food and drug safety and quality [7].

One of the main reasons for this may be the low accuracy of traditional methods of sowing microbial cells in Petri dishes. Traditionally , the following two seeding methods are the most common:

Small volumes (drops) of a cellular suspension of microbial cells are introduced using a pipette (dispenser) to the surface of the medium in a Petri dish and distributed as evenly as possible over this surface using a curved glass stick (spatula).

The layer of the main nutrient agar is filled with a thin layer of molten semi-liquid agar with a suspension of microbial cells [9].

In order to provide such "ideal conditions" that ensure almost the same size of microbial colonies or Poisson distribution for their numbers, a special seeding technique called "ordered seeding" and a device called a "poissoner" were invented in 1973 [10, 11]. They are currently used in several laboratories in Russia, the USA, the UK and Brazil [10-24]. They are most popular with molecular geneticists who need accurate estimates of mutation rates, recombination, transformation, survival, etc.

For a long time, this method and device were only briefly described in English in a review on methods for measuring the frequency of spontaneous mutations in yeast, published in 1978 [13,14]. The remaining descriptions were published in Russian - in three theses [10-12], in the PhD thesis of N.N. Khromov–Borisov [15] and in S.G. Inge-Vechtomov's textbook "Genetics with the basics of breeding" ([16], pp. 299-302). The method is presented in lectures and demonstrated in practical classes on the course "General Genetics" to students of the Biology and Soil Faculty of St. Petersburg State University and in the corresponding Methodological Manual provided with a CD with illustrations ([17], p. 34 and slides 5 and 6 for Lecture 11 on CD). The applications described earlier mainly concerned the use of concentrated cell suspensions and the registration of rare events (with frequencies of 10-6), such as colonies of secondary revertants or recombinants on special selective media. A revised more detailed protocol, an optimized poissoner design and some new applications of them were published in English in 1999 [18] and then in 2002 in the form of an article entitled "Perfectorderplating: principleandapplications" (authors: N. N. Khromov-Borisov, J. Saffi, J.A. P. Henriques) in an international peer-reviewed online electronic magazine TTO – Technical Tips Online. However, this publication suffered a sad fate: two years later, Elsevier publishing house ceased to exist without warning, obviously violating the copyright of the authors.

We found it expedient to translate this article into Russian in order to make the method and adaptation described in it available to a wide range of domestic microbiologists. (The authors have received permission from Elsevier Publishing House to translate a non-existent article from a non-existent journal).

The principle of ordered seeding and its application

The principle of ordered seeding is that small drops of a suspension of microbial cells of equal volume are placed on the surface of the agar medium in a Petri dish at equal distances from each other. This arrangement of droplets provides uniform ecological and physiological conditions for cell growth and, as a result, the same size of colonies or an ideal Poisson distribution for their number.

Generation of colonies of the same size

Very small (volume not exceeding 1 ml and diameter not exceeding 1 mm) drops of a thick suspension of cells (~ 10 ^{6-10 8} cells / ml) are placed on the agar surface at the nodes of a regular trigonal lattice (like a pattern of honeycombs with an additional dot in the middle of each cell). This can be done either manually or with the help of a special poissoner inoculator with pins pointed at the end. In the first variant, you can use any device such as a micropipette or a drawing pen ("radish") and a stencil placed under a Petri dish with a diagram of the correct trigonal lattice. After incubation, small spots containing thousands of cells grow into colonies of almost the same size and shape (Fig. 1). When sowing, damage to the surface of the agar should be avoided, otherwise the shape of the colonies will be disturbed.

Fig.1. The principle of ordered seeding: application to the study of colony growth.
Very small drops (with a volume of less than 1 ml and a diameter of no more than 1 mm) of a thick suspension of yeast cells (~10 ^{6-10 8} cells / ml) were placed on the surface of the agar medium using a drawing pen ("radish") according to the nodes of the correct trigonal lattice (in accordance with the stencil placed under the cup).
After growth under standard incubation conditions, colonies are obtained that are the same in size and shape.
The variation in the diameter of the colonies on the cup is exceptionally small: the relative standard error is about 1%.

Poissoner

The poissoner is a microbiological inoculator, the design of which provides Poisson distribution for the number of colonies on Petri dishes. It is made of stainless metal and looks like a stamp (or brush) with cylindrical pins (teeth). Pins can be made of rivets (diameter ~ 2.0 mm and length 1.5‑2.0 cm) with carefully aligned ends. They are mounted on a metal base at equal distances from each other in such a way as to completely fill the inside of the Petri dish. The recommended optimal number of pins is 163 or 187 with a distance between them of 3.0 mm (Fig. 2).

Fig. 2. Poissoner: general view (a) and one of its working positions – face down (b).

The poissoner can be sterilized in the flame of an alcohol lamp, having previously moistened its working part (the ends of the pins) with alcohol. At the same time, overheating of its metal parts should be avoided. For safety reasons, its handle is made of fire-resistant material with low thermal conductivity (usually heat-resistant plastic).

The poissoner can be used in two working positions: "face down" (Fig. 2b) and "face up" as in the classical method of prints (not shown). The first position is used for fast work when it is necessary to sow a lot of Petri dishes without particularly high accuracy. The inverted position is preferred to achieve high accuracy. Accordingly, the end of the handle is made flat so that the inverted poissoner stands firmly on the table.

Generating a Poisson distribution for the number of colonies using a poissoner

A 20-25 ml suspension of microbial cells is poured into a sterile Petri dish and the working part (pins) of a sterile poissoner are dipped into it. The suspension is thoroughly mixed by rotating the poissoner, and then transferred to a Petri dish with agar culture medium. Before that, it should be monitored whether all pins have hemispherical drops of suspension formed at the ends. If not all pins carry such drops, then it is worth cleaning the poissoner from hydrophobic contaminants with alcohol or detergent. The preferred working positions are: the "face up" position for the poissoner and the "face (agar surface) down" position for the Petri dish. The Petri dish is carefully placed on the working part of the poissoner and gently pressed with fingers so that small round dents (traces) from the ends of the pins form on the agar surface, but the surface of the agar would not be damaged. Such traces are clearly visible to the naked eye, but, unfortunately, are poorly distinguishable in photographs (Fig. 3 and 4). Through the transparent bottom of the cup, make sure that each pin leaves a trace with a drop. Carefully remove the cup in such a way as to prevent the droplets from merging.

Fig. 3. The arrangement of colonies generated by the poissoner (red cups) and the traditional method of sowing with a pipette and a spatula (green cups).
Two pairs of cups with approximately equal number of colonies are presented: ~440 colonies in the first column and ~300 in the second.

Fig. 4. Reproducibility of the results generated by the poissoner.
With the help of a poissoner, the same cell suspension was seeded independently by two performers (three Petri dishes for each – green and red).
The variability of the number of colonies is very small: 360-390 colonies per cup.

Data analysis

After incubation, the number of traces (dents) is counted from poissoner pins without colonies, with one colony, with two, three, etc., and present the data in the form of a sample distribution table (Table 1). This procedure is repeated with other Petri dishes. For the obtained sample distributions, a statistical check of agreement with the Poisson distribution is carried out.

Table 1. Distributions of the number of yeast colonies on 10 Petri dishes generated by Poissoner,
and their comparison with the distribution of the number of colonies obtained by the traditional seeding method

[a] P-values for the exact variance criterion of agreement with the Poisson distribution [30]. For all 10 cups (as well as for the total number of colonies, P = 0.97), deviations from the Poisson distribution are statistically insignificant. The Poisson distributions generated by Poissoner are statistically homogeneous and reproducible on all 10 cups: P=0.69 is obtained as an estimate of the P-value for the exact Fisher criterion (based on 5×10 ⁶ randomizations) for the 10×10 conjugacy table using the StatXact‑4 program.
[b] The total number of colonies on Petri dishes generated by Poissoner is also statistically homogeneous, i.e. consistent with the Poisson distribution: Pp = 0.66 according to the Volodin-Bolshev criterion [31-33].
[c] These data were obtained using the same cell suspension, but seeded by the traditional method. Their heterogeneity is statistically extremely significant due to extra-Poisson variability: Pt = 8.3×10-33 according to the Volodin-Bolshev criterion [31-33].

Several statistical procedures were used to verify the agreement of the observed variation in the number of colonies on each trace and (or) on each Petri dish with the expected Poisson distribution [26-33]. In particular, the exact and Monte Carlo versions of the Fisher variance criterion (variance index) were used. For this purpose, a program created and kindly provided by Papworth was used [30]. The agreement of the observed variation in the total number of colonies on Petri dishes generated by the poissoner and the traditional seeding method with the Poisson distribution was checked using the Volodin-Bolshev elegant criterion most adequate for this purpose [31-33].

Advantages of the method of ordered seeding

When using the ordered seeding method to measure colony growth, the variability of colony diameter is exceptionally low: the relative measurement error is about 1% (Fig. 1). This accuracy is comparable to the accuracy of good experiments in physics or engineering.

The method of ordered seeding allows you to achieve high accuracy when calculating the number of colonies. Compared with traditional seeding procedures using a pipette and a spatula, when using a poissoner, colonies are formed that are much better separated from each other and their location is much more uniform (Fig. 3 and 4). The method allows to obtain several hundred (500 or even more) non-overlapping colonies on one Petri dish. As a result, a large saving of consumables is achieved: it is enough to sow 2-3 cups to estimate the average values with a relative error of about 5%.

Ordered seeding generates an "ideal" Poisson distribution with high reproducibility: colony counts on several Petri dishes are statistically homogeneous (Tables 1, Figures 4 and 5) compared with heterogeneous (with increased variance) counts of colonies obtained from the same cell suspension by the traditional seeding method (Table. 1) and compared with the data published in the literature (Fig.5) [4-8].

Fig. 5. Visualization of agreement with the Poisson distribution for published data [4] (A) and data generated by Poissoner (C) and their reproducibility (B).

Notation: The abscissa axis in Fig. A: variants of Petri dishes with the observed number of colonies on them are i; the ordinate axis is the number of such variants of cups – N _i.
The abscissa axes in Fig. B and C: variants of traces (dents) from poissoner pins on agar medium in a Petri dish by the number of colonies on them – i;
the ordinate axis in Fig. B – the number of such variants of traces – n _i on each of their 10 cups;
the ordinate axis in Fig. C is the total number of such variants of traces on all 10 cups - N _i.
Squares and solid lines connecting them in Fig. A and B represent the observed distributions, and triangles and dotted lines represent the theoretically expected Poisson distribution.
By various symbols in Fig. C indicates the data for each of the 10 Petri dishes.

Using a poissoner makes it possible to check statistically how well the observed distribution of the number of colonies agrees with the theoretical Poisson distribution, because each trace of a pin (dent) can be considered as a micro-Petri dish (Table 1). In this laboratory, several hundred Petri dishes were sown using this technique, and no statistically significant deviation from the Poisson distribution was ever detected. Moreover, in cases where the number of colonies on the cup turns out to be so large that some of them merge and it is difficult to distinguish them, then a quick method of estimating the average values can be used [14,20].

Orderly seeding is faster, simpler and more economical than traditional seeding techniques. It is a good alternative to conventional seeding methods. With the help of a poissoner, one Petri dish can be sown in a second, while sowing one cup by the traditional method can take tens of seconds (sometimes even 1-2 minutes). Additional savings result from the fact that sterilization of the poissoner is very cheap compared to the use of both disposable cups, spatulas and tips for dispensers, and reusable consumables that require lengthy sterilization procedures and high energy costs.

Thus, ordered seeding can be used in many microbiological laboratories. It is particularly important in such areas of experimental and applied microbiology in which GLP and GSP are required, such as genetics and selection of microorganisms, genetic toxicology, biotechnology, epidemiological, clinical, pharmaceutical, agricultural and industrial microbiology, microbiological toxicology and microbiological quality and safety control of food and medicines. It can better ensure safety when working with pathogenic microbes. Disadvantages of the method of ordered seeding

Like other laboratory techniques, the method of ordered seeding requires some training skill from the experimenter. Some psychological aspects are also obvious. For example, for an experimenter who is accustomed to the traditional technique of seeding with a pipette and a spatula, at first it may seem unusual and inconvenient to have to work with relatively large volumes of cell suspensions (20-25 ml) placed not in a test tube, but in a Petri dish. Nevertheless, almost all the techniques of miniaturization in microbiology can be easily adapted to the use of a poissoner.

Materials and equipment

In this work, Saccharomyces cerevisiae yeast, a haploid strain N123 used as a wild-type strain, was used in the GENOTOX laboratory at the Center of Biotechnology of the University of Rio Grande do Sul (Porto Alegre, Brazil). His genotype: MATa his1. For its cultivation, a standard complete YEPD medium for yeast was used [25].

The program C.A.MAN was used for graphical visualization [29]. To achieve a higher quality, all graphs were redrawn using the Prizm-3 program. To check the uniformity within each data set, which are conjugacy tables, the exact and Monte Carlo criteria implemented in the StatXact-4 software package were used. Hewlett-Packard ScanJet4 scanner and corresponding software were used to capture images of Petri dishes.

This article has been peer reviewed.

Thanks

The authors thank D. Papworth, A. Masuda, I. daSilvaVazJr., D. Sveshnikov, A. Borodich, I. Rozhdestvensky and K. Kvitko for their help in the work. The work was supported by Brazilian research Foundations: CNPq, CAPES, FAPERGS and GENOTOX – LaboratóriodeGenotoxicidade (UFRGS), as well as the Swedish WallenbergFoundation.

Literature

1. Khromov-Borisov N.N. and Henriques J.A.P. Good Statistics Practice (GSP) in genetic toxicology // Mutat. Res., 1998. 405:97-108.
2. Student. On the error of counting with haemocytometer // Biometrika, 1907. 5:351-360.
3. Fisher R.A. et al. The accuracy of the plating method of estimating the density of bacterial populations. With particular references to the use of Thornton's agar medium with soil samples // Ann. Appl. Biol., 1922. 9:325-359.
4. Linhardt M.S. et al. In vitro information system for collection and analysis of experimental data // J. Environ. Pathol. Toxicol., 1980. 4:1-21.
5. Margolin B.H. et al. Statistical analysis of the Ames Salmonella/microsome test // Proc. Natl. Acad. Sci. U. S. A., 1981. 78:3779-3783.
6. Bogen K.T. Applicability of alternative models of revertant variance to Ames-test data for 121 mutagenic carcinogens // Mutat. Res., 1994. 322:265-273.
7. Voss B. et al. A multinomial model for the quality control of colony counting procedures // Biometr. J., 2000. 42:263-278.
8. Kim B.S. and Margolin B.H. Statistical methods for the Ames Salmonella assay: a review // Mutat. Res., 1999. 436:113-122.
9. Eaton, A.D. et al., (eds.) Standard Methods for Examination of Water and Wastewater (19th edn.). - Washington, DC, USA: American Public Health Association, 1995.
10. Khromov-Borisov N.N. Ordered sowing as a method of studying the growth and mutation of microbes / Mechanisms of biological processes (Tr. III conf. of young specialists, December 26-29, 1972, Leningrad). - L.: Publishing House of LSU, 1973. – S. 41.
11. Khromov-Borisov N.N. The method of ordered seeding for the study of growth and mutations in microorganisms / Abstracts of the Conference on the genetics of industrial microorganisms (December 10-14, 1973). - Tsakhkadzor: USSR Academy of Sciences, 1973. – p. 44.
12. Khromov-Borisov N.N. et al. The use of the method of ordered seeding for the study of intragenic mitotic recombination in yeast-saccharomycetes / Theses of the All-Russian Conference on the genetic basis of selection of industrial microorganisms (December 20-25, 1975, Minsk). – Minsk: USSR Academy of Sciences, 1975. - pp. 105-106.
13. von Borstel R.C. Measuring spontaneous mutation rates in yeast // Methods Cell Biol., 1978. 20:1-20.
14. Khromov-Borisov N.N. Biometrical aspects of measuring mutation rates // Methods Cell Biol., 1978. 20:20-24.
15. Khromov-Borisov N.N. Study of the mutagenic effect of methoxyamine and 6‑hydroxylamionopurine on yeast Saccharomyce cerevisiae. Author's thesis. cand. diss. L.: Leningr. state University, 1976.
16. Inge-Vechtomov S.G. Genetics with the basics of breeding. - Moscow: Higher School, 1989. - pp. 299-302.
17. General genetics. Methodical manual / Edited by S. G. Inge-Vechtomov. – St. Petersburg: Publishing House N-L, 2007. – 124 p. + CD.
18. Khromov-Borisov N.N., Saffi J., Henriques J.A.P. POP (perfect order plating) and “poissoner” – new precise technique and device for microbiological toxicology studies / Anais da II Reunião Nacional de Microbiologia Aplicada ao Meio Ambiente (Proceedings of the Second National Meeting of Microbiology Applied to the Environment). Florianopolis (Brazil), 1999.
19. Gordenin D.A. et al. Yeast mutants with increased bacterial transposon Tn5 excision // Yeast, 1991. 7:37-50.
20. Morrison A. et al. Pathway correcting DNA replication errors in Saccharomyces cerevisiae // EMBO J., 1993. 12:1467-1473.
21. Shcherbakova P.V. et al. Base analog 6-N-hydroxylaminopurine mutagenesis in the yeast Saccharomyces cerevisiae is controlled by replicative DNA polymerases // Mutat. Res., 1996. 369:33-44.
22. Fedorova I.V. et al. The yeast HSM3 gene acts in one of the mismatch repair pathways // Genetics, 1998. 148:963-973.
23. Fedorova I.V. et al. The yeast HSM3 gene is involved in DNA mismatch repair in slowly dividing cells // Genetics, 2000. 154: 495-496.
24. Kozmin S.G. et al. Hypersensitivity of Escherichia coli Delta(uvrB-bio) mutants to 6-hydroxylaminopurine and other base analogs is due to a defect in molybdenum cofactor biosynthesis // J. Bacteriol., 2000. 182:3361-3367.
25. Evans, I.H. (ed.) Yeast Protocols: Methods in Cell and Molecular Biology. - Totowa, NJ, USA: Humana Press, 1996.
26. Merkle W. Poisson goodness-of-fit tests for radiation-induced chromosome aberrations // Int. J. Radiat. Biol., 1981. 40:585-692.
27. Gütler N. and Henze N. Recent and classical goodness-of-fit tests for the Poisson distribution // J. Statist. Plann. Inference, 2000. 90:207-225.
28. Karlis D. and Xekalaki E. A simulation comparison of several procedures for testing the Poisson assumption // The Statistician, 2000. 49:355-382.
29. Böhning D. Computer-assisted Analysis of Mixtures and Applications: Meta-analysis, Disease Mapping and Other. - Boca Raton, FL, USA: Chapman & Hall/CRC, 1999.
30. Papworth D.G. Exact tests of fit for a Poisson distribution // Computing, 1983. 31:33-45.
31. Volodin I.N. On the distinction between Poya and Poisson distributions when a large number of small samples are available // Theory. ver. approx., 1965. 10:335-338.
32. Bolshev L.N. On the characterization of the Poisson distribution and its statistical applications // Theor. ver. prim., 1964. 10:446-456. Reproduced in the book: Bolshevik L.N. Selected works. Probability theory and Mathematical statistics. – M.: Nauka, 1987. – 286 p., pp. 153-165.
33. Volodin I.N., Ioffe V.M. Probabilistic injury models and the distribution of the number of accidents at industrial enterprises. In the book: Safety issues. – M.: VTSNIIOT VTSPS, 1973. – pp. 5-23.

Portal "Eternal youth" www.vechnayamolodost.ru06.04.2009