New Blue Pigment Produced by Pantoea agglomerans and Its Production Characteristics at Various Temperatures

Abstract

A bacterium capable of producing a deep blue pigment was isolated from the environment and identified as Pantoea agglomerans. The pigment production characteristics of the bacterium under various conditions were studied. The optimal agar plate ingredients for pigment production by the bacterium were first studied: the optimal ingredients were 5 g/liter glucose, 10 g/liter tryptic soy broth, and 40 g/liter glycerol at pH 6.4. Bacterial cells grew on the agar plate during the incubation, while the pigment spread into the agar plate, meaning that it is water soluble. Pigment production was affected by the initial cell density. Namely, at higher initial cell densities ranging from 10^6.3 to 10^8.2 CFU/cm² on the agar plate, faster pigment production was observed, but no blue pigment was produced at a very high initial density of 10^9.1 CFU/cm². Thus, the cell population of 10^8.2 CFU/cm² was used for subsequent study. Although the bacterium was capable of growing at temperatures above and below 10°C, it could produce the pigment only at temperatures of ≥10°C. Moreover, the pigment production was faster at higher temperatures in the range of 10 to 20°C. Pigment production at various temperature patterns was well described by a new logistic model. These results suggested that the bacterium could be used in the development of a microbial temperature indicator for the low-temperature-storage management of foods and clinical materials. To our knowledge, there is no other P. agglomerans strain capable of producing a blue pigment and the pigment is a new one of microbial origin.

Temperature is one of the most influential of the physical factors that regulate growth and other physiological phenomena of microorganisms. Fresh foods and clinical materials, such as organs for transplantation and blood, are easily spoiled by the growth of contaminants at a particular temperature. Thus, monitoring and recording the temperature history of these products during the period from manufacturing to consumption is very important. Devices capable of assessing the temperature history of food and clinical products, known as time-temperature integrators or indicators (TTIs), have been developed (2, 3, 11, 15, 21). TTIs are categorized as electronic, chemical, and biological (microbial). A digital thermorecorder is an electronic TTI, and the temperature data need to be transformed to microbial growth data with a mathematical model. The mechanisms of chemical and biological TTIs have to be based on irreversible changes and reactions; most of the changes are expressed as color changes or the movement of a boundary (2, 3, 11, 15, 21). Among these TTIs, the main advantage of a microbial TTI is that test microorganisms that have growth characteristics similar to those of food or clinical product contaminants can be selected. The mechanism in most microbial TTIs is an irreversible color change of a chemical chromatic indicator which follows a pH decline due to microbial growth in a medium (2, 21).

On the other hand, many microorganisms are known to produce pigments, including yellow, pink, red, and violet colors. Some pigments have been chemically identified, such as carotenoids, prodigiosin, and pyocyanin (12, 16). Among these pigments, the number of blue/violet pigments is very small. Violacein, which is produced by members of the genera Chromobacterium, Janthinobacterium, and Iodobacter, is a well-known violet pigment (8, 9, 12, 16).

We recently identified a bacterium capable of producing a deep blue pigment. First, we found that the bacterium grew at temperatures above and below 10°C and that it could only produce the pigment at temperatures ≥10°C. There is a possibility that our bacterium may be useful in the development of a microbial TTI. In other words, a TTI equipped with the bacterium, by producing the blue color in the TTI, may be able to indicate spoilage of nonsterile food and clinical materials which may be exposed to extreme temperatures. However, the environmental conditions for pigment production by the bacterium were unknown.

In this study, aiming to incorporate this bacterial strain in a microbial TTI, we assessed the optimal conditions for an agar plate for pigment production by the bacterium. With the agar plate, we then studied pigment production by the bacterium under conditions varying from the initial cell level and temperature.

MATERIALS AND METHODS

Phenotypic and molecular characterization.

Bacteriological examination of the test strain was performed using conventional methods (1). The base sequence of 16S rRNA for the strain was analyzed with a PRISM 310 genetic analyzer (Applied Biosystems, Foster City, CA). The base sequence data obtained were examined for homology with standard bacterial strains using an ABI database (Microseq ID Analysis software, version 2.0; Applied Biosystems).

Cell preparation.

All microbiological procedures were done under sterile conditions. The test strain was isolated on a nutrient agar plate (Nissui Chemicals, Tokyo, Japan) and incubated at 25°C for 2 days. Cells of some well-grown colonies were collected and cultured in 50 ml of tryptic soy broth (TSB) (Becton Dickinson, Sparks, MD) with shaking (80 rpm) at 25°C for 24 h. The culture was washed with the same volume of saline (0.9% sodium chloride) by centrifugation at 4°C and 13,000 × g for 15 min. Cells were then suspended in the same volume of saline. This was the so-called “original suspension” and was used in most of the experiments in this study. Condensed and diluted cell suspensions were analyzed in an experiment with various initial cell densities.

Media for pigment production.

A salt solution consisting of 5 g/liter ammonium sulfate (Wako Pure Chemical Industries, Osaka, Japan), 2 g/liter potassium dihydrogen phosphate (Wako), and 1 g/liter magnesium sulfate heptahydrate (Wako) in deionized, purified water (19) was mixed with certain amounts of d-glucose (Wako) and TSB (Becton Dickinson) as indicated in Results. After the medium was adjusted to the desired pH with 0.1 M hydrochloric acid and mixed with 15 g/liter agar (Nissui), the mixture was autoclaved at 121°C for 15 min, and then 4 ml was poured into each petri dish (34 mm in diameter) (product no. 1008; Becton Dickinson). The agar plates were allowed to solidify and then dried.

Storage study.

The cell suspension prepared as described above was placed on the surface of an agar plate (0.2 ml/plate). The plates were stored in an incubator (SU-221; Espec, Osaka, Japan) for up to 140 h at various temperatures ranging from 8 to 20°C. At regular time intervals, triplicate plates were taken from the incubator. The plates at each time point were used for the color measurement described below and then for cell counts. The cell density on the surface of a plate was measured as follows. The agar (three grams on average) was thoroughly removed from the petri dish and homogenized with buffered saline (97 ml) (Nissui) in a sterile plastic bag with a stomacher for 1 min. After a serial 10-fold dilution of the homogenate with saline, the cell concentration of the dilution was measured with a spread-plate method on nutrient agar plates (Nissui) in duplicate after a 48-h incubation at 25°C. The cell density on the agar plate (CFU/cm²) was then calculated by dividing the whole-cell population on the agar plate by the area of the petri dish (9.07 cm²). Cell growth on the agar plate during the storage period was analyzed with a new logistic model described below (4-6).

Color measurement.

Blue pigment production on each agar plate was measured as the color change of the plate using a Tristimulus colorimeter (CM-3500d; Konica-Minolta, Tokyo, Japan), because the pigment produced diffused into the whole agar plate. The plate color was measured for three plates per time point (two measurements per plate). The color change was expressed as the distance of color with ΔE*, which is expressed as follows (21):

(1)

where L*, a*, and b* are coordinates of the CIE 1976 L*a*b* color space (L* is the lightness of the color, a* is its position between red/magenta and green, and b* is its position between yellow and blue) of a sample and L*₀, a*₀, and b*₀ are those of the control and of the agar plate immediately after inoculation.

Analysis and prediction of pigment production.

Values of ΔE* for samples during storage at constant temperatures were analyzed with a new logistic model that we developed for microbial growth (4-6). The model, which describes a sigmoidal growth curve, was transformed to express the increase in ΔE* by substituting ΔE* for microbial population as follows.

(2)

Here, r is the rate constant of increase in an increase curve of ΔE* and t is time. m and n (>0) are parameters related to the curvature of the deceleration phase and the period of the lag phase, respectively, in the increase curve. ΔE*_max is the maximum ΔE* value, and ΔE*_min corresponds to the initial value, ΔE*₀. ΔE*_min was set to be (1 − 10⁻⁶) × ΔE*₀, similar to the minimum cell population (N_min) in our microbial growth model (4-6). The value for ΔE*₀ is theoretically zero from the definition (equation 1). However, we are not able to mathematically solve equation 2 when ΔE*starts at zero, or ΔE*₀ = 0. This is because the denominator in an equation can never be zero. To avoid this, the initial value for ΔE* in equation 2 was set to be very low, near zero (0.01 in this study). The equation successfully described the increase in ΔE* during an experiment.

The values of ΔE* for samples exposed at a given constant temperature during storage were analyzed with a program equipped with our logistic model (7). The values of r obtained at constant temperatures were then analyzed with a square root model (14).

The value of ΔE* of an agar plate exposed at a dynamic temperature was also predicted with equation 2, similar to microbial growth prediction (4-6). That is, the temperature of an agar plate during storage was monitored at intervals of 30 s with a digital recorder (AM-7002; Anritsu Meter, Tokyo). The measured temperature was transformed into r with the square root model, and then those values of r were input into equation 2 for prediction. The averages for m and n at constant temperatures were used for the parameters in equation 2 (4-6).

Statistical analysis.

The averages and standard deviations (SDs) for ΔE* and the cell density were calculated for each sample with spreadsheet software in Microsoft Excel. A t test was also done with Excel.

RESULTS

Bacteriological identification.

Several characteristics of our bacterium were first studied for identification. The bacterium was a Gram-negative rod that was positive for motility, Voges-Proskauer, catalase, malonate, and utilization of urea and arabinose. It was negative for oxidase, lysine, and ornithine decarboxylases, arginine dihydrolase, indole production, and utilization of lactose. It also produced a yellow pigment on nutrient and standard agar plates. These characteristics were identical to those of Pantoea agglomerans (1, 10). The base sequence analysis of the strain's 16S rRNA showed that it had the highest similarities with Pantoea agglomerans ATCC 27155 (99.20%) and P. agglomerans ATCC 33243 (98.91%), followed by other Pantoea sp. strains. These results suggested that our strain belongs to P. agglomerans.

Optimal agar plate development for pigment production.

The optimal agar plate condition for production of the blue pigment were studied. First, the concentrations of the organic ingredients of the agar plate were studied. The organism was inoculated onto agar plates with various concentrations of glucose (5 to 20 g/liter) and TSB (2.5 to 20 g/liter) and adjusted to pH 6.8. After incubation at 20°C for 14 h, the color change of the agar plate, ΔE*, was measured as an index of pigment production (Fig. 1). At 20 g/liter TSB, the ΔE* values were all high, but the original color of the agar plate prior to inoculation was brownish due to the high broth concentration, which is not ideal for use as a TTI. At other concentrations of glucose and TSB, a combination of 10 g/liter TSB and 5 g/liter glucose generated the highest value (Fig. 1). On agar plates without TSB or glucose, no blue pigment was produced, while bacterial growth was observed.

FIG. 1.

Blue pigment production in agar plates containing various concentrations of glucose and tryptic soy broth. Symbols show means, and error bars show SDs. The asterisk shows the optimal combination of the ingredient concentrations. g/l, grams per liter.

The optimal pH of agar plates with the above-described ingredients was then examined. On agar plates adjusted to various pHs between 5.6 and 7.6, the highest ΔE* value was obtained at pH 6.4 (Fig. 2). Thus, the optimal pH for pigment production was determined to be 6.4.

FIG. 2.

Blue pigment production in the agar plates with various pH values. Cells were incubated at 20°C for 14 h. Closed circles show means, and error bars show SDs.

Since it was reported that glycerol stimulated violacein production by Janthinobacterium (17), the effect on pigment production of glycerol added to the agar plate was studied. The addition of 10 and 40 g/liter of glycerol slightly increased the pigment production at the peak (for both concentrations, t < 0.01) in comparison with that in the control (0 g/liter glycerol), and the blue pigment in the agar plate lasted longer during incubation at higher glycerol concentrations (Fig. 3). At 80 and 160 g/liter glycerol, pigment production was delayed (Fig. 3). These results suggested that a concentration of 40 g/liter would be optimal for pigment production and stability. We finally decided that an agar plate with these ingredients would be optimal for producing the pigment and developing a TTI. This plate was named the G plate and was used for further studies.

FIG. 3.

The effects on pigment production of glycerol supplementation of the agar plates. Symbols show means, and error bars show SDs.

Effect of initial cell density.

The relationship between pigment production and cell growth at various initial cell densities was studied. The initial cell densities examined were 10^6.3, 10^7.2, 10^8.2, and 10^9.1 CFU/cm². Here, the density of 10^8.2 CFU/cm² was that of the original cell suspension. Pigment production was faster at higher initial values of between 10^6.3 and 10^8.2 CFU/cm² (Fig. 4 A). An interesting observation was that no blue pigment was produced at the highest initial concentration (10^9.1 CFU/cm²), but cells developed a yellow color during incubation (Fig. 4A).

FIG. 4.

Pigment production (A) and growth (B) of the bacterium at various initial densities (•, 10^6.3 CFU/cm²; ▪, 10^7.2 CFU/cm²; ▴, 10^8.2 CFU/cm²; ⧫,10^9.1 CFU/cm²). Cells were inoculated onto the G plate and stored at 20°C. Bars show SDs. (C) Pigment production and bacterial growth with an initial population of 10^6.3 CFU/cm². Arrows show the corresponding axes. An open circle shows the point at an ΔE* value of 20. Curves are described with the new logistic model.

Cell growth at the above-listed initial cell densities was also observed. Cell growth at low initial cell densities (10^6.3 and 10^7.2 CFU/cm²) showed sigmoidal curves (Fig. 4B). The increase in the cell population during storage at the highest initial cell density was very small (about 10^0.6 CFU/cm²).

The relationship between pigment production and cell growth was studied. An example of the results at the initial cell density of 10^6.3 CFU/cm² is shown in Fig. 4C. The increase in ΔE* and the cell growth were well described by the new logistic model. It was observed that pigment production began at the late logarithmic phase of cell growth. Let us evaluate a point in the middle of the slope of ΔE* to confirm the relationship between pigment production and cell growth, because there was no distinct logarithmic phase at higher initial cell densities (Fig. 4B). Here, this point was set to be a point with an ΔE* value of 20. The point was located at a density of 10^8.7 CFU/cm², as shown in Fig. 4C. The values at the initial cell densities of 10^7.2 and 10^8.2 CFU/cm² were 10^8.6 and 10^9.0 CFU/cm², respectively. Those cell populations were close to each other.

The initial cell density at the original suspension (10^8.2 CFU/cm²) that presented the fastest pigment production was used for subsequent study.

Pigment production at constant temperatures.

Pigment production by the bacterium at various constant temperatures between 8 and 20°C was studied. A temperature-dependent manner of pigment production was observed (Fig. 5). Pigment production was observed at ≥10°C. The increases in ΔE* at the constant temperatures were well described by the new logistic model, as shown by the data in Fig. 4C (data not shown). After ΔE* reached the maximum value, it gradually decreased at all temperatures (Fig. 5).

FIG. 5.

Pigment production of the bacterium at various temperatures. Cells at 10^8.2 CFU/cm² were inoculated onto the G plate. Symbols show means, and bars show SDs.

An example of production of the blue pigment during storage is shown in Fig. 6. During the storage period, the color of the G plate changed from white (the original color) to pale yellow (Fig. 6, 14 and 17 h) and then the production of blue pigment progressed (Fig. 6, 20 to 29 h). During longer time periods, the blue color became pale and the ΔE* value decreased, but the color could be still observed with the naked eye. Actually, a positive sample with an ΔE* value of >5 was visually confirmed to be blue. On the other hand, the blue pigment produced spread into the agar plate, meaning that the pigment was soluble in water.

FIG. 6.

Color changes of the inoculated G plates during incubation at 16°C. Numbers show the hours of incubation. Plate diameter is 34 mm.

The rate constant of pigment production, r in equation 2, showed a high linearity at constant temperatures, with a regression coefficient of 0.869 by the square root model. The relationship between r and temperature (T) was described with the following model:

(3)

At 8°C, no blue pigment was produced, but the bacterium grew as yellow cells. A small increase in ΔE* (<5) was observed at 8°C during storage (Fig. 5), suggesting that this increase was due to the yellow pigment production.

These temperature characteristics of the production of the blue pigment suggested that the bacterium could be applied to a TTI for samples that need to be kept cool at less than 10°C. An intense blue pigment was produced at up to 30°C, but the bacterium grew as yellow cells at above 30°C. Thus, the temperature range of pigment production by the bacterium was between 10°C and 30°C.

Pigment production at dynamic temperatures.

Pigment production by the bacterium at dynamic temperatures was studied with the new logistic model using the data obtained at constant temperatures, shown in Fig. 5. Here, the values for r in the model were evaluated using equation 3 with the temperatures measured at the time points. The pigment production with the changing temperature was observed (Fig. 7 A and B). The temperature ranges in the experiments were 9.0 to 15.4°C (Fig. 7A) and 9.4 to 19.5°C (Fig. 7B). With the averages of m (0.89) and n (0.39) at the constant temperatures, the prediction for pigment production was much earlier than the observed pigment production; the predicted lag period was much shorter (curves 1 in Fig. 7A and B). Thus, the n values that gave the minimum mean square error between the predicted and the observed ΔE* values were estimated because n is a parameter related to the lag period. These optimal values for n gave much better predictions (curves 2 in Fig. 7A and B). The n values for the data shown in Fig. 7A and B were 0.22 and 0.19, respectively, which were close to each other. Here, the initial ΔE* value was set to be about 1.5; this value compensated for the ΔE* value by the pale yellow color change prior to the blue color change during storage.

FIG. 7.

Pigment production by the bacterium at a dynamic temperature. Closed circles show means, and bars show SDs. The periodic curve is the temperature history of the plate. The sigmoidal curves are curves predicted with the new logistic model. The temperature ranges were 9.0 to 15.4°C (A) and 9.4 to 19.5°C (B). The values of n (see equation 2) are 0.39 (curve 1) and 0.22 (curve 2) in panel A and 0.39 (curve 1) and 0.19 (curve 2) in panel B. Arrows show the corresponding axes.

Similar results were obtained for other dynamic temperature patterns (data not shown). In total, the average for the optimal n value was 0.21 ± 0.033 (mean ± SD; n = 4). Thus, prediction with this (average) value for n may be successful at other dynamic temperatures.

DISCUSSION

Blue pigment production was observed at the late logarithmic phase of growth at an initial density of 10^6.3 CFU/cm² (Fig. 4C). This means that a certain number of cells were needed for pigment production. Similar results have been obtained for other microbial metabolites, like staphylococcal enterotoxin production in milk and prodigiosin production in LB medium (6, 22). Prodigiosin is a red pigment produced by Serratia spp. (12). On the other hand, production of microbial pigments, including prodigiosin and violacein, is known to be controlled by quorum sensing systems (13, 20). In quorum sensing, a bacterial cell can sense the cell density by the accumulation of signaling molecules. When our strain was isolated and grown on a G agar plate, the blue pigment production was observed where there was a high cell density (Fig. 8). These results suggested that the production of the blue pigment may be regulated by quorum sensing.

FIG. 8.

The blue-pigmented region of a G plate on which the bacterium grew. The bacterium was isolated and incubated at 25°C for 3 days. The diameter of the plate is 90 mm. The many gray dots in the photo are shadows of isolated colonies cast on the bottom of the plate.

No blue pigment was produced at the highest initial density (10^9.1 CFU/cm²), which was already over the densities at the midpoints, which ranged from 10^8.6 to 10^9.0 CFU/cm² (Fig. 4A, B, and C). This means that, theoretically, the blue pigment would not be produced at this high initial cell density. During the experiment, bacterial cells grew slightly at the highest initial cell density. This suggests that those cells did not produce any signaling molecules in the quorum sensing system, which stimulates blue pigment production, because the cell density was great enough.

At further, longer periods of storage, the value of ΔE* gradually decreased, as shown by the data in Fig. 3 and 5, and the blue color of the G plate became paler. However, this color change was irreversible; the blue color remained and the ΔE* value was maintained (over 10) for a longer period. Thus, once the blue pigment was produced, we could see it for a long period with the naked eye. This means that the microbe could be applied as a microbial alarm for spoilage of food or clinical materials and that the decrease in ΔE* over long periods would not be a practical problem. However, the results shown in Fig. 5 and 7 suggested that a microbial TTI with our organism should be used for a period within 1 week but not for a longer period.

The new logistic model has succeeded in predicting microbial growth and microbial toxin production at dynamic temperatures (4-6). However, the values for ΔE* at dynamic temperatures predicted with the model were much faster than the measured values (Fig. 7). The amounts of staphylococcal enterotoxin production in milk at dynamic temperatures predicted with the model were also higher than the measured amounts, and thus, we introduced a correction factor in the prediction model (6). The production dynamics of metabolites of microbial cells at dynamic temperatures are generally thought to be very complex to study. In the present study, we could predict values for ΔE* at dynamic temperatures by the change in value of parameter n, but we need to develop a better model for the pigment production in the future. The model might be built on the kinetics of cell population, based on the above discussion on quorum sensing.

Also, when one of the environmental factors, including the ingredients of the agar plate, the storage temperature, and so on, was not good for production of the blue pigment, the bacterium grew into yellow-pigmented cells. Furthermore, the bacterium only produced the blue pigment aerobically; when the bacterium was incubated under stationary conditions in a broth containing the ingredients needed for pigment production, the pigment was only produced at the interface with the surrounding air. These phenomena suggest that the blue pigment is produced under limited conditions.

To our knowledge, no report has been published showing that P. agglomerans strains produce a blue pigment (10, 12, 16). P. agglomerans strains are known to grow into yellow colonies on nutrient agar plates (10), and our strain also did. Although the cells themselves were not blue pigmented on the G plate, the surrounding crowded colonies were clearly pigmented blue, which is an indication of the water-soluble nature of the pigment (Fig. 8).

We studied some characteristics of our pigment. The pigment demonstrated solubility in water but not in organic solvents, such as chloroform, acetone, or acetonitrile; however, when these solvents contained a small amount of water, the pigment could easily dissolve. This is different from other microbial violet and blue pigments, such as violacein and others (8, 9, 12), showing that, to our knowledge, our pigment is a new blue pigment of microbial origin. Of interest, the color of our pigment changed with the pH to pink in an acidic solution but did not change in an alkaline solution. These characteristics of the pigment are similar to those of anthocyanin, a blue pigment in plants (18). Also, our pigment was not stable in water, and the blue color gradually became paler in water. We are currently studying the structure, as well as the physicochemical and biological characteristics, of the pigment. Some pigments are known to have an effect on microorganisms. For example, violacein protects bacterial cells from other Gram-positive bacteria as an antibiotic (12). Our pigment in this study may also have such effects.

Generally, Pantoea spp. are isolated from plants, seeds, fruits, soils, water, and humans and animals (10). Since a microbial TTI is attached to a food or clinical package, the microbe used must not be a pathogen. We are going to study whether or not the P. agglomerans strain in the present study is a pathogen. We should also clarify if the metabolites of the strain, including the pigment, are pathogenic or not.

First, we studied the growth characteristics of violacein-producing strains of the genera Chromobacterium, Janthinobacterium, and Iodobacter to develop a microbial TTI. Several strains of these genera were obtained from the National Institute of Technology and Evaluation, Japan, and then incubated on nutrient agar plates (Nissui) at various temperatures. However, no strain was an appropriate candidate; that is, these strains inoculated on the agar plates were already violet pigmented before being exposed to a given temperature. Also, these strains grew very slowly at 10 to 20°C, and thus, violacein production of these strains on the plates was also very slow. On the other hand, pigment-producing microorganisms other than our strain could be also candidates for microbial TTIs (12, 16). Psychrophiles would be suitable for the low-temperature storage management of food and other items. A study of the characteristics of growth and pigment production of these candidates at various temperatures will be needed, similar to this study of the current strain.