Scientific
Publications - Work Done by Microbiology Reader Bioscreen C
Journal of Applied Microbiology, 1999, Apr,
86(4), 689-694
An investigation into the differences between the Bioscreen and traditional plate count disinfectant test methods
Lambert RJ and van der Ouderaa ML
ABSTRACT
Investigations of biocide efficacy by automated methods
involving optical density measurements, e.g. using the recently published
'Bioscreen' method, are complicated by the fact that a poor correlation often
exists between the log reductions obtained using the automated method vs those
obtained by the traditional plate count methods. It was hypothesized that the
differences observed between the two methods were due to the level of cell
injury, which was masked by the optical density methods but which was recognized
by the plate counts. Comparisons of log reductions following a disinfection test
always showed the Bioscreen method to be overestimating the log reductions with
respect to the plate counts. A correlation between colony size on the plates and
the 'Bioscreen' results for a fixed disinfectant concentration and contact time
was found using Global Imaging software. The results obtained also suggested
that the observed colony area was dependent on the amount of injury incurred by
a microbe during the disinfection process. A mathematical model of injury was
developed which predicted the observed differences between the Bioscreen and the
traditional plate method. The model further suggested a possible reason for
biocidal lags.
INTRODUCTION
Busta & Jezeski (1963) reported lower thermal death times with
Staphylococcus aureus when enumerated on media containing added NaCl rather than
on normal media. They concluded that the salt caused an adverse effect on
injured organisms. The lower death times were due to injured organisms being
unable to repair and grow quickly enough to be counted on the plates, leading to
the conclusion that more were killed than was in fact the case. This so-called
lag phase, which is the time required to repair injury before growth occurs, has
been the subject of many investigations as the minimization of lag has large
commercial benefits (e.g. Stephens et al. 1997). Before the release of
foodstuffs from the warehouse into the market place, microbiological testing
occurs. An enrichment stage is essential to allow injured bacteria to repair.
The faster the repair can be effected, the quicker the test becomes, and the
lower the costs of holding back the product.
The length of the delay in colony formation on agar plates, i.e.
the time of lag, is dependent on the severity of the [heat] stress given
(Kaufman et al. 1959; Jackson & Woodbine 1963; Anagnostopoulos et al. 1966;
Payne 1978). The greatest variation in lags was observed with populations which
had undergone the greatest level of stress. Under such conditions, lag times of
up to 70 h were recorded (Mackey & Derrick 1984). It was considered that the
injured population was a heterogeneous one, with individuals exhibiting varying
levels of injury (Van Schothorst & Van Leusden 1975; Mackey & Derrick 1982).
However, methods required to measure the variation within a population were not
forthcoming. The lag time observed for a population was considered to be the
time for the least injured organism to repair and divide. Recently, a method was
described which used an automated growth analyser to examine the lags of single
heat-injured Salmonella cells (Stephens et al. 1997); the results showed the
broad distribution of lags expected.
Recently, a method using a Bioscreen microbiological growth
analyser for disinfectant testing was reported (Lambert et al. 1998). However,
the results obtained differed from those of the traditional plate count method.
A hypothesis that the differences between the methods was due to cell injury,
which was masked by the Bioscreen method but visible in the plate counts, was
suggested. This difference needed to be fully explained to prove that the
Bioscreen method was effective and accurate and could be compared to the plate
count results.
MATERIALS AND METHODS
Preparation of bacterial suspensions
Staphylococcus aureus ATCC no. 6538 (Staph. aureus) was grown
overnight in 80 ml Tryptone Soya Broth (TSB; Oxoid). The flask was incubated at
30 °C and shaken continuously. The resulting mixture was divided into four
universal tubes and centrifuged at 512 g (4000 rev min
1;
Sigma model 3K-1) at 15 °C for 10 min. The supernatant fluid was discarded and
the resulting pellets were pooled and re-suspended in 9 ml 0·1% peptone water.
Suspension tests
The method has been described elsewhere (Lambert et al. 1998) but
briefly, 20 ml test disinfectant at each required concentration and 20 ml
distilled water (as a control) were dispensed into universal containers. A 200 
l
volume of the Staph. aureus suspension was added to give approximately 1 
108
ml
1
in each contact tube. A clock was immediately started. At the required contact
times of 3, 6, 9, 12, 18, 24, 0 and 36 min, the universal container was shaken
and 1 ml of each solution was transferred into 9 ml Universal Quenching Agent
(Lambert et al. 1998) and shaken again. The quenching agent inhibits any
further action by the disinfectant. The quenched solution was then either used
in the Bioscreen, or was plated out.
Global imaging
Global Lab Image is an image processing and analysis software package for
Microsoft® WindowsTM. This software allowed the
distribution in colony size (pixel areas) to be observed over time for each
concentration of disinfectant used. The colonies on the plates were stained with
tetranitro-blue-tetrazolium (TNBT; Cambrian Chemicals Ltd), which gave the
colonies a dark brown to black colour which increased their contrast
significantly.
A water-treated control was placed on each plate, hence reducing
plate-to-plate variability with respect to external controls. Differences in
area between the test colonies and the water controls were expressed as an
average of the water control colony area. Graphs of percentage areas over time
were constructed for each concentration.
Mathematical models
RESULTS
Bioscreen vs traditional plate method
The observed differences in the calculation of log reductions between the
Bioscreen and the traditional plate count methods (TPM) for the phenol
disinfection of Staph. aureus is shown in Fig. 1. All points lie below
the equivalence line which suggested prima faci that the Bioscreen method
gave a higher log reduction in numbers than that obtained from the plates. The
most logical conclusion was that the Bioscreen method failed to detect all the
living cells capable of forming colonies on an agar plate.
A similar result is shown in Fig. 2 where phenethyl alcohol (PeA) has been
used to disinfect the Staph. aureus culture. In both cases, the trend
appears to be for the gradient of the experimental curve to become parallel to
that of the equivalence line. This suggests that after an initial lag with
respect to the Bioscreen, both techniques record the same rate of disinfection.
It is argued that the horizontal difference is due to the numbers of injured
organisms which are being counted on the plates but which are absent from the
Bioscreen calculations.
Direct correlation between Bioscreen and TPM
Our hypothesis is that the Bioscreen measures the healthiest cells. On a
plate, the healthiest cells are those which produce the largest colonies because
after a disinfection experiment, the parent cell has spent the least amount of
time repairing any injuries before replication starts. By enumerating only those
colonies equivalent in size to the water controls, a direct correlation with the
log reductions from the Bioscreen could be obtained (Fig. 3).
Distribution of injury
The plates obtained from a series of disinfection tests were imaged using
Global Imaging software. At first, colony size was grouped into one of three
classes, i.e. small, medium or large. The large colonies had an area comparable
with that of the water controls. Figure 4 shows the changes in these classes as
the time of disinfection increases. Clearly, there is a move to smaller colony
size with time, with the percentage of medium-sized colonies first increasing
and then decreasing in number, and with the largest, and therefore the
healthiest, colonies reducing in number with time. However, the classification
was rather arbitrary. By using the value for the pixel area of the colony, this
problem was circumvented as the software produced a complete data set for all
colonies, e.g. pixel area, lengths of axes or eccentricities of the colonies;
this allowed double colonies to be used in the calculations. Figures 5 and 6
show the changes in colony pixel area with time for the disinfection of
Staph. aureus with phenol (0·85%) and phenethyl alcohol (1·5%). The same
general trend towards smaller colonies with time of disinfection can be seen.
It appears that the variation in colony size of phenol-disinfected Staph.
aureus vs PeA disinfection correlates with the size of the difference
between the log reductions of the Bioscreen vs the log reductions from
the TPM (Figs 1 and 2). The phenol disinfection shows a large difference between
the two techniques, and a larger degree of variation in colony size with time
than that observed for the PeA disinfection. This may be reflecting, in some
way, mechanism of action, phenol giving rise to a greater number of injured
states than PeA. However, in general, as time proceeds fewer cells remain
totally healthy as the disinfectant has had a longer time to act upon the
microbe, increasing its chance of becoming injured.
Mathematical model of injury
The basis of the hypothesis under test is that the Bioscreen measures the
healthiest cells in a population following disinfection whereas plates give a
measure of the number of viable cells, both injured and healthy. From this, a
basic mathematical model can be constructed as follows:
In this equation, an inoculum of uninjured microbes, A1, becomes
injured, A2, through the action of a biocide at a rate given by k1;
the injured bacteria then become non-viable, P, after a further dose of biocide
at a rate given by k2. The first assumption is that there is no
possibility of recovery during the disinfection process and that there is only
one distinct 'injured' population.
The differential equations obtained are easily solved and the values of A1,
A2 and P with respect to time are given by the expressions described
in the methods section with the equations for models with two and three levels
of injured sub-populations. Using this simple model, a hypothetical population
was constructed and disinfected. Figure 7 gives the numbers of uninjured,
injured and dead cells with respect to time.
The model predicts that as the number of uninjured cells falls with time, the
number of injured cells increases and the number of microbes becoming non-viable
begins to increase. As the time of disinfectant contact increases, the number
injured reaches a peak and then begins to decay as fewer cells are left
uninjured and more cells become non-viable. The model appears to be reflecting
the experimental observations.
For comparison with the Bioscreen and plate data, the Bioscreen data can be
likened (for simplicity) to the uninjured population and the plates to the total
number of viable cells, i.e. the sum of the injured and non-injured. Figure 8
shows a modelled comparison between the Bioscreen and TPM. This is a
particularly useful result as it reflects what is observed (compare with Figs 1
and 2). The trend is for the modelled data to become parallel to the equivalence
line, as found experimentally.
The model suggests that for values of k2 ] k1, the
smaller the difference between the equivalence line and the observed. This is
simply stating that if k2 is large, then as soon as the microbe
becomes injured, it is killed. When k2 is comparable in size to k1,
the model suggests that a difference will be observed. However, if k2 < k1,
then the gradient of the modelled line will not reach 1. It is often suggested
that an injured microbe will be more prone to a biocidal agent than one which is
uninjured. This model and the experimental observations suggests that this may
indeed be the case.
DISCUSSION
Bacterial cells that have been injured as a result of a disinfection test,
but which are still viable, will need time to recover before they can start to
divide. While these cells are undergoing repair, the healthy cells will be
dividing geometrically. As a result, injured cells in a suspension will not
significantly affect the optical density relative to the more healthy cells,
which may have undergone several divisions before the more injured cells begin
replication, i.e. their growth will be masked by the more extensive growth of
the healthy cells.
When injured cells are spread onto an agar plate, they will be able to repair
and begin to divide. As long as they have divided enough times for the colony to
become visible on the plate, they will be included in a calculation of a log
reduction. The plate count method of calculating log reductions involves
counting each colony but does not take into account the size of the colony. It
has been observed that colony areas, after a disinfection test, show a large
variation in size, and the average area grows smaller with disinfectant contact
time. We have suggested that the smaller colonies originated from injured cells
that needed to recover before multiplying, and therefore did not have as much
time to reach the size of the healthy colonies. These colonies will be counted
as long as they are visible to the naked eye; therefore, there will appear to be
more growth on the plates compared with the Bioscreen method where the more
injured cells are masked and do not appear to be present.
The distribution of colony size appears to be dependent on the level of
disinfectant used and the time of exposure. As the bacteria in these experiments
were all from the same strain and are therefore capable of replicating at the
same rate, any differences in the colony size, which is related to growth rate,
must be due to the level of injury of the parent cell; the smaller the colony
the more injured was the original parent cell. This is direct experimental
confirmation that the longer the applied stress, the higher the degree of
injury.
The Bioscreen method enumerates the healthiest cells in a population. These
cells may not necessarily be from an uninjured population; experimental evidence
(Figs 4, 5 and 6) suggests they are not always totally uninjured.
Mathematical models based on mechanistic hypotheses are useful tools with
which to direct experiments, or at least to suggest whether the proposed
mechanism has any claim in reality. Unfortunately, many models have little or no
experimental evidence with which to correlate the observables. The mathematical
model described here does fit with the experimental evidence, although this does
not mean that it has a true physical significance. However, the comparison
between the Bioscreen and the plates, and the variation of the colony size with
time, does suggest that such a mechanism may be in operation. A model proposed
by Prokop & Humphrey (1970) to account for aspects of spore death kinetics
suggested a sensitive intermediate population, i.e. with respect to model 1, k1 ] k2.
The observations given in this work suggest the opposite may be true, at least
for vegetative bacteria, as plots of the Bioscreen vs TPM appear to be
tending towards a gradient of 1, which would not occur if k1 > k2.
The model described here, and the experimental evidence, suggests that an
injured bacterium is disinfected more easily than an uninjured one.
A possible explanation as to why plots of Bioscreen log reduction, log Rb,
vs TPM log reductions, log Rp, tend to a gradient of 1 can be
found using the evidence of the model given here and the model of Wickramanayake
& Sproul (1988). The number of uninjured microbes falls according to log linear
kinetics, i.e. log Rb = k1 t. The total viable
shows a lag because of an injured sub-population, but the overall rate of
disinfection is governed by the slowest step, k1, by definition. For
a given level of disinfectant Wickramanayake and Sproul showed that log Rp
= k1 t -C, where C is the observed lag. The difference,
log Rb- log Rp = C. Thus, a constant separation between
the Bioscreen (log Rb) and the plates (log Rp) should be
found, as is the case.
The model suggests that the rate of disinfection is given, as expected, by
the rate determining step, i.e. the slowest step (k1). However, by
the method adopted here it would suggest that the Bioscreen should not describe
lags, but we know this to be false. Plots of log reduction against time using
the Bioscreen do show the presence of lags (Lambert et al. 1998).
Although model 1 with one injured sub-population cannot deal with this
situation, the other models, which allow for a larger number of injured
sub-populations, can. For example, using the model with three injured
sub-populations, if we assume that the first two populations can fully recover,
then lags become a natural consequence of this model.
FIGURES
Fig. 1 Comparison of log
reductions between the traditional plate count method and Bioscreen for the p...
Fig. 2 Comparison of log
reductions between the traditional plate count method and Bioscreen for the p...
Fig. 3 Comparison of the
traditional plate count method (enumerating largest colonies only) and Bioscr...
Fig. 4 Distribution of
Staphylococcus aureus colony size with respect to disinfection contact time.
Di...
Fig. 5 Distribution of
Staphylococcus aureus colony size with respect to disinfection contact time.
Di...
Fig. 6 Distribution of
Staphylococcus aureus colony size with respect to disinfection contact time.
Di...
Fig. 7 Plot of the numbers
of microbes undergoing a modelled disinfection process. (
),
Uninjured; (
),
...
Fig. 8 Comparison of the
modelled traditional plate count method and Bioscreen data using the paramete...
REFERENCES
• Anagnostopoulos, G.D., Seamen, A. & Woodbine, M. 1966 An attempt to
increase the heat resistance of Microbacterium lacticum. Journal of Applied
Bacteriology, 29, 207-212.
• Busta, F.F. & Jezeski, J.J. 1963 Effect of sodium chloride concentration in
an agar medium on growth of heat-shocked Staphylococcus aureus. Applied
Microbiology, 11, 404-407.
• Jackson, H. & Woodbine, M. 1963 The effect of sublethal heat treatment on
the growth of Staphylcoccus aureus. Journal of Applied Bacteriology, 26,
152-158.
• Kaufman, O.W., Harmon, L.G., Pailthorp, O.C. & Pflug, I.J. 1959 Effect of
heat treatment on the growth of surviving cells. Journal of Bacteriology, 78,
834-838.
• Lambert, R.J., Johnston, M.D. & Simons, E.-A. 1998 Disinfectant testing:
use of the Bioscreen microbiological growth analyser for laboratory biocide
screening. Letters in Applied Microbiology., 26, 288-292.
• Mackey, B.M. & Derrick, C.M. 1982 The effect of sublethal injury by
heating, freezing, drying and gamma-radiation on the duration of the lag phase
of Salmonella typhimurium. Journal of Applied Bacteriology, 53, 243-251.
• Mackey, B.M. & Derrick, C.M. 1984 Conductance measurements of the lag phase
of injured Salmonella typhimurium. Journal of Applied Bacteriology, 57, 299-308.
• Payne, J. 1978 Damage and recovery in Streptococci. In: Streptococci, SAB
Symposium. Series, 7, ed. Skinner, F.A. & Quesnel, L.B, pp. 349-369. London:
Academic Press.
• Prokop, A. & Humphrey, A.E. 1970 Kinetics of disinfection. In: Disinfection
(ed. Bernarde, M.A), pp. 61-83. New York: Marcel Dekker.
• Stephens, P.J., Joynson, J.A., Davies, K.W., Holbrook, R., LappinScott,
H.M. & Humphrey, T.J. 1997 The use of an automated growth analyser to measure
recovery times of single heat-injured Salmonella cells. Journal of Applied
Microbiology, 83, 445-455.
• Van Schothorst, M. & Van Leusden, F.M. 1975 Further studies on the
isolation of injured salmonellae from foods. Zentralblatt Fur Bakteriologie,
Parasitenkunde, Infektionskrankheiten und Hygiene. Abt. 1, Originale A, 230,
186-191.
• Wickramanayake, G.B. & Sproul, O.J. 1988 Ozone concentration and
temperature effects on disinfection kinetics. Ozone Science and Engineering, 10,
123-135.
------- 8< -------
(Full Text online)
