determination of yeast viability by means of fluorescence microscopy

DETERMINATION OF YEAST VIABILITY BY MEANS OF
FLUORESCENCE MICROSCOPY AND IMAGE ANALYSIS
Ing. Petra Jeřábková
Supervisor: doc. Ing. Oldřich Zmeškal, CSc.
Institute of Physical and Applied Chemistry, Faculty of Chemistry, Brno University of
Technology, Purkyňova 118, 612 00 Brno, e-mail: [email protected]
1. INTRODUCTION
In a condition assessment of microbiological populations, the important factor is a
representation of live and dead individuals that it is possible to differentiate by means of
fluorescent labelling. A direct microscopic count is usually used for direct determination of
the cell number in counting chambers (by Thom, Bürker). The number of cells of microorganisms can frequently be determined indirectly with the assistance of cultivation when it is
assumed that one colony will grow up from a viable cell and these colonies will be counted
after that. It is possible to use an image analysis for the detection of defined objects
(e.g. yeasts) without the necessity to count them.
In this work, the wavelet transformation at a fractal analysis of an image, the so-called box
counting method, was utilised. The wavelet transformation (or the Haar one) makes the
calculation of squares of different sizes of a laid mesh more effective with the box counting
method provided that a square area is being analysed. An analysed image with a size that is
given by the power of a definite number (2i) is transformed into the space of spectra. On this
spectra, the filter of a low pass type with the size of square 2i, where i = 1, 2, 4… n – 1 is
gradually applied and the number of black, white and black and white squares is determined
in images that arise from the inverse transformation in the same way as with a box counting
method. It is possible to substitute the process of transformation and inverse transformations
of an image by the addition of neighbouring foursome pixels and to analyse a relevant
sequence of these pictures. Three fractal dimensions of the black and white area and their
interface (DBBW, DBWB, DBW) and fractal measures (KBBW, KBWB, KBW) will again be the result.
When analysing a black and white picture that may be obtained from a coloured picture by the
process called thresholding, a method identical with a box counting method is described [1].
With the assistance of the fractal analysis, characteristic data about an analysed structure,
namely the fractal dimension D and the fractal measure K are determined. These parameters
can be used for picture ordering evaluation as well as e.g. for specifying the number of
defined objects or for specifying their radius [2].
Pictures for the analysis were obtained by means of fluorescence microscopy because this
method is quick and simple. Results are obtained within minutes, whereas classic cultivation
methods require the minimum of 24 hours.
2. EXPERIMENTAL PART
The fractal analysis was used for specifying the number and dimension of images of live
and heat-killed yeast cells. The fluorescent dye of acridine orange (AO) was used for the
simultaneous distinction of live and dead cells (dead cells give out red or orange fluorescence
and live cells yellow-green; 180 µM AO, pH 6). Fluorescein diacetate (FDA) was used for the
distinction of live cells (greenish fluorescence; 6 mM FDA, pH 7.2). With the assistance of
Sborník soutěže Studentské tvůrčí činnosti Student 2006 a doktorské soutěže O cenu děkana 2005 a 2006
Sekce DSP 2005, strana 115
fluorescent labelling it is possible to threshold either dead or live cells on a black colour in
one picture. Images of cells were thresholded by the intensity or coloured components of the
RGB space. To detect the convenient thresholding, it is possible to use the fractal spectrum,
i.e. fractal dimension dependence on the intensity or selected RGB component, which is
accessible in the HarFA software as a tool referred to as Fractal Analysis – Range [3]. The
number of cells x and their radius r were determined provided that images of cells were of a
circular shape, similar in size and distinguishable on the background according to relations
x=
2
N BW
,
4π ( N B + N BW )
r=
(N B + N BW )ε 2
πx
where NB is the number of black boxes, NBW is the number of black and white boxes with the
size of ε × ε pixels. The resulting cell number x is derived from the value x, which is the
maximum from the calculated values for different sizes of the mesh. The maximum value is
selected because the fractal structure is bordered most conveniently for the given mesh size.
Yeasts from Culture Collection of Yeasts – CCY, the Institute of Chemistry, Slovak
Academy of Sciences in Bratislava and from the Faculty of Chemical and Food Technology
of the Slovak University of Technology in Bratislava were used for the image analysis. Yeast
Saccharomyces cerevisiae FV1 and yeast Saccharomyces cerevisiae subsp. cerevisiae CCY
21-4-102 have ellipsoidal and spherical cells with width about 2.6–6.4 µm and with length of
3.7–9.7 µm. Yeast Kloeckera apiculata CCY 25-6-22 has the shape of a lemon and size of
(1.5–5) µm × (2.5–11) µm. Vegetative cells of yeast Hansenula anomala CCY 38-1-30 are
spherical or ellipsoidal with size of 2–4 and 2–6 µm [4].
Yeasts that were cultivated on an inclined agar by the room temperature were, then,
inoculated into a liquid culture medium (glucose peptone yeast extract medium) where they
were cultivated for a minimum of 24 hours again by the room temperature under aerobic
conditions. After that, a half of the volume of the culture medium with cells was exposed to
95°C for 3–6 hours (according to the yeast species). A drop of live and dead cells mixture was
put on a slide and a drop of fluorescent dye was added. Microscopical preparation was
observed immediately at staining by fluorescein diacetate and after 30 minutes at staining by
acridine orange with 40× objective.
The number of cells was evaluated in 35 visual fields. A suitable neutral density filter
(ND4, ND8, ND16), which decreases the intensity of the excitation light, was inserted
between the excitation light (mercury lamp) and excitation filter, when required. For the
recording of pictures, the combination of the epifluorescence microscope Nikon Eclipse E200
with the filter cube and a digital camera Nikon Coolpix5400 with the resolution of 2592 ×
1944 was used. For the storage of pictures, the software Lucia Net 1.16.6. was used. The
obtained pictures with the size of 1024 × 1024 were analysed by means of the software
HarFA [3].
Sborník soutěže Studentské tvůrčí činnosti Student 2006 a doktorské soutěže O cenu děkana 2005 a 2006
Sekce DSP 2005, strana 116
Table 1
Average numbers of cells from 35 pictures of sample of different yeast species that were
obtained by fractal analysis
AO
FDA
Yeast
Dead
Average
Live
Average
Total
species
cells
error
cells
error
number
error
number
(%)
number
(%)
of cells
(%)
5.42
8.05
6.60
8.66
11.97
6.10
11.40
4.02
4.19
7.23
7.00
4.85
11.19
4.61
8.96
3.74
4.94
6.61
10.34
6.94
14.90
5.66
10.53
5.93
10.17
8.44
10.38
6.92
20.57
5.22
8.05
8.11
Saccharomyces
cerevisiae
Saccharomyces
cerevisiae FV1
Kloeckera
apiculata
Hansenula
anomala
Average Live cells
number
Average
error
(%)
Table 2
Average radii of cells from 35 pictures of sample of different yeast species that were obtained
by fractal analysis
AO
Yeast
Dead cells
Live cells
radius
Saccharomyces
cerevisiae
Saccharomyces
cerevisiae FV1
species
Kloeckera
FDA
Live cells
radius
Live and dead cells
radius
3.25
4.12
3.74
3.78
3.47
3.70
3.55
3.75
2.88
2.72
2.96
2.62
2.61
2.65
2.72
3.17
radius
apiculata
Hansenula
anomala
3. CONCLUSION
The fluorescence microscopy in the connection with the image analysis seems to be a
suitable method for the counting of yeast cells. For the staining of dead and live cells
simultaneously, the acridine orange dye is suitable, for the staining of live cells it is the
fluorescein diacetate. When counting live and dead cells of different yeast species by means
of the fractal analysis it was determined that the average error of determination of the yeast
number from 35 pictures is less than 10 % with the number of cells up to 20 in a picture
Sborník soutěže Studentské tvůrčí činnosti Student 2006 a doktorské soutěže O cenu děkana 2005 a 2006
Sekce DSP 2005, strana 117
(Table 1). This error can also be caused by an unequal size of cells and shapes, differences in
colouring of the individual cells and quality of the recorded picture. If cells are killed by heat,
the cytoplasm is sometimes stained greenly as well, which complicates the thresholding
process with cells that were stained by the acridine orange. Budding cells were counted as one
cell identically as in the direct determination of the cell number whereas if a daughter cell
carried another bud still not separated from a mother cell, it was counted as two.
Obtained radiuses of cells image provided that they have a circular shape are listed in
Table 2 and are within the range of 2.61–4.12 µm. The sizes of the radii fall within values that
are listed in literature or, in some case, are slightly greater [4].
4. REFERENCES
[1] Tománková, K. Studium vlastností hrubě disperzních soustav pomocí metod obrazové
analýzy. Brno, 2004, 42 s. Diploma thesis, Faculty of Chemistry, Brno University of
Technology. Supervisor: Oldřich Zmeškal.
[2] Zmeškal O., Sedlák O., Nežádal M.: Metody obrazové analýzy dat, Digital Imaging in
Biology and Medicine, Czech Academy of Science České Budějovice: May 13., 2002, pp.
34 - 43, ISBN 80-901250-8-5.
[3] Zmeškal O., Nežádal M., Buchníček M., Bžatek T.: HarFA 5.0, Harmonic and Fractal
Image Analyzer, http://www.fch.vutbr.cz/lectures/imagesci, Brno, 2003.
[4] Kocková-Kratochvílová, A.: Taxonómia kvasinek a kvasinkovitých mikroorganizmov.
1. vyd. Bratislava: Alfa, 1990. ISBN 80-05-00644-6.
Sborník soutěže Studentské tvůrčí činnosti Student 2006 a doktorské soutěže O cenu děkana 2005 a 2006
Sekce DSP 2005, strana 118