Food Microstructure and Fat Content Affect Growth Morphology, Growth Kinetics, and Preferred Phase for Cell Growth of Listeria monocytogenes in Fish-Based Model Systems

Listeria monocytogenes is one of the most hazardous foodborne pathogens due to the high fatality rate of the disease (i.e., listeriosis). In this study, the growth behavior of L. monocytogenes was investigated at a microscopic scale in food model systems that mimic processed fish products (e.g., fish paté and fish soup), and the results were related to macroscopic growth parameters. Many studies have previously focused on the food microstructural influence on microbial growth. The novelty of this work lies in (i) the microscopic investigation of products with a complex composition and/or structure using confocal laser scanning microscopy and (ii) the direct link to the macroscopic level. Growth behavior (i.e., concerning bacterial growth morphology and preferred phase for growth) was more complex than assumed in common macroscopic studies. Consequently, the effectiveness of industrial antimicrobial food preservation technologies (e.g., thermal processing) might be overestimated for certain products, which may have critical food safety implications.

ABSTRACT

Food microstructure significantly affects microbial growth dynamics, but knowledge concerning the exact influencing mechanisms at a microscopic scale is limited. The food microstructural influence on Listeria monocytogenes (green fluorescent protein strain) growth at 10°C in fish-based food model systems was investigated by confocal laser scanning microscopy. The model systems had different microstructures, i.e., liquid, xanthan (high-viscosity liquid), aqueous gel, and emulsion and gelled emulsion systems varying in fat content. Bacteria grew as single cells, small aggregates, and microcolonies of different sizes (based on colony radii [size I, 1.5 to 5.0 μm; size II, 5.0 to 10.0 μm; size III, 10.0 to 15.0 μm; and size IV, ≥15 μm]). In the liquid, small aggregates and size I microcolonies were predominantly present, while size II and III microcolonies were predominant in the xanthan and aqueous gel. Cells in the emulsions and gelled emulsions grew in the aqueous phase and on the fat-water interface. A microbial adhesion to solvent assay demonstrated limited bacterial nonpolar solvent affinities, implying that this behavior was probably not caused by cell surface hydrophobicity. In systems containing 1 and 5% fat, the largest cell volume was mainly represented by size I and II microcolonies, while at 10 and 20% fat a few size IV microcolonies comprised nearly the total cell volume. Microscopic results (concerning, e.g., growth morphology, microcolony size, intercolony distances, and the preferred phase for growth) were related to previously obtained macroscopic growth dynamics in the model systems for an L. monocytogenes strain cocktail, leading to more substantiated explanations for the influence of food microstructural aspects on lag phase duration and growth rate.

IMPORTANCE Listeria monocytogenes is one of the most hazardous foodborne pathogens due to the high fatality rate of the disease (i.e., listeriosis). In this study, the growth behavior of L. monocytogenes was investigated at a microscopic scale in food model systems that mimic processed fish products (e.g., fish paté and fish soup), and the results were related to macroscopic growth parameters. Many studies have previously focused on the food microstructural influence on microbial growth. The novelty of this work lies in (i) the microscopic investigation of products with a complex composition and/or structure using confocal laser scanning microscopy and (ii) the direct link to the macroscopic level. Growth behavior (i.e., concerning bacterial growth morphology and preferred phase for growth) was more complex than assumed in common macroscopic studies. Consequently, the effectiveness of industrial antimicrobial food preservation technologies (e.g., thermal processing) might be overestimated for certain products, which may have critical food safety implications.

INTRODUCTION

In recent years, global fish product consumption has increased significantly (1–3). Fish products are known to be beneficial for human health, being an important source of high-quality proteins, vitamins, minerals, and omega-3 fatty acids (4–6). However, contamination with foodborne pathogens is common in fish products, as illustrated by the percentage of foodborne outbreaks caused by products of this food category, e.g., 5.4% in 2016 (7). The bacterium Listeria monocytogenes, which causes listeriosis, has been detected in fish products on a regular basis since 1987 (7–9). Listeriosis is an illness with a mortality rate of more than 20% (10), with clinical features ranging from mild influenza-like illness to invasive diseases such as meningitis and meningoencephalitis (11).

In predictive microbiology, the effect of food processing, distribution, and storage operations on microbiological safety is evaluated by means of mathematical models that describe microbial responses to environmental conditions (12, 13). Since predictive models are traditionally developed based on experimental data from homogeneously well-mixed broth media, in essence ignoring food microstructure and composition, model accuracy for the behavior of microorganisms in more structured food products is often limited (14–16). Food microstructure encompasses the spatial arrangement of the various structural elements (e.g., water and oil droplets, gas cells, particles, granules, strands, crystals, micelles, and interfaces) of a food product and their interactions (17). Microbial dynamics are affected by a plethora of food microstructural aspects, e.g., physical constraints on microbial mobility (18–20), the presence of fat in the food matrix (21, 22), the nature of the food matrix (i.e., viscous or gelled) (22), and diffusion of oxygen, water, nutrients, preservatives, and metabolites (23–27).

One approach that allows inclusion of the food microstructural influence into predictive models, is to conduct microbiological experiments in food model systems with various microstructures (28–31). Wilson et al. (26) defined five categories of food microstructures: liquids, emulsions, aqueous gels, gelled emulsions, and surfaces. Based on this classification, Baka et al. (29) investigated the influence of food microstructure on growth dynamics of L. monocytogenes at suboptimal temperatures using model systems based on processed fish products. However, apart from the variation in microstructure among these model systems, there was also variation in compositional and physicochemical factors. These unwanted variations were caused by the presence or absence of fat and gelling agents in some of the systems, a consequence of developing representative model systems for each microstructure. For this reason, Verheyen et al. (31) developed model systems with various microstructures, among which the microstructural effect was isolated by means of minimal variation in compositional and physicochemical aspects. The set of model systems consisted of three viscous systems and two gelled systems, i.e., (i) a liquid system; (ii) xanthan, a more viscous liquid system containing a small concentration of xanthan gum; (iii) an emulsion; (iv) an aqueous gel; and (v) a gelled emulsion, respectively. These model systems were used to investigate the effect of food microstructure on growth dynamics of L. monocytogenes at suboptimal temperatures, i.e., 4 and 10°C (22). The growth morphology of the cells (i.e., planktonic cells, submerged colonies, or surface colonies), the nature of the food matrix (i.e., viscous or gelled), and the presence of fat droplets were reported to exert a significant influence on the parameters of the growth model of Baranyi and Roberts (32). Since the study of Verheyen et al. (22) relied solely on macroscopic growth experiments, the underlying mechanisms have mostly been left unraveled. More specifically, apart from a visual inspection during the macroscopic growth experiments, a detailed investigation of the growth morphology in which L. monocytogenes appeared in the different model systems was not conducted. Quantification of colony sizes could lead to more insight in the observed differences in macroscopic growth dynamics. Colonies can either be classified as microcolonies (i.e., radius < 200 μm) or macrocolonies (i.e., radius > 200 μm). While microcolony growth largely resembles planktonic growth, macrocolony growth is slower due to the presence of pH gradients and diffusion limitations around and inside the colonies (33). In addition, single cells can also cluster together and form small aggregates (i.e., radius < 1.5 μm) which cannot be considered full-fledged microcolonies (33–37). Another finding of the study was the growth-promoting effect of a small percentage of fat droplets in the model system matrix for which the causes remained unknown. More fundamental research toward these phenomena at a microscopic scale will lead to increased insight into the influence of food microstructure on microbial growth dynamics.

While food products generally consist of different phases, most microbiological studies are only conducted at a macroscopic scale, ignoring heterogeneity. In order to characterize the behavior of microorganisms in a complex food product, more advanced microscale measurement techniques are therefore necessary (38). Confocal laser scanning microscopy (CLSM) is a nondestructive technique which has several advantages compared to conventional light microscopy, e.g., the applicability of fluorescent probes to stain and visualize different components, the possibility of using relatively thick samples due to the removal of out-of-focus light, and the possibility of creating three-dimensional (3D) images by using a sequence of optical sections at different sample heights (39).

The aim of this study was to investigate the effect of food microstructure on L. monocytogenes growth dynamics at the microscopic level and relate the obtained results to findings at the macroscopic level. In order to compare microscopic and macroscopic observations, the bacteria were grown inside fish-based food model systems at 10°C, analogous to the macroscopic growth experiments conducted by Verheyen et al. (22). Model system composition was based on processed fish products (e.g., fish soup, surimi, and fish paté), and the microstructure was simulated by including the major food microstructural aspects of those products (e.g., a visco-elastic matrix or fat droplets). While a cocktail of three L. monocytogenes strains isolated from fish-based food products was used in the above-mentioned macroscopic study, a green fluorescent protein (GFP) L. monocytogenes strain was used in the present study in order to facilitate CLSM experiments. Confocal images were used to study the growth morphology of the cells in each model system, and the growth morphology was characterized by means of the number and volume distribution of single cells, small aggregates, and microcolonies with various sizes. For the emulsion and gelled emulsion model systems, the preferred phase (i.e., aqueous phase, fat phase, or the interface) for L. monocytogenes growth was investigated using systems with various fat levels (i.e., 1, 5, 10, and 20%). As a possible explanation for the affinity of the cells for a certain phase, hydrophobicity of the cells was quantified using the MATS (microbial adhesion to solvents) assay (40). The MATS assay was conducted for both the GFP strain and the L. monocytogenes strain cocktail used for the macroscopic growth experiments of Verheyen et al. (22), enabling an improved comparison of microscopic and macroscopic results.

RESULTS

Confocal laser scanning microscopy.

CLSM images were used to visualize the growth behavior of the selected L. monocytogenes (GFP) strain in the different model systems after 14 days of incubation at 10°C. Bacterial cells were visualized in green and fat droplets (if relevant) in orange. Cell cluster sizes were quantified and subsequently classified in six categories, i.e., single cells, small aggregates, and microcolonies of four different sizes. Since interpreting linear size parameters is more straightforward than interpreting squared or cubic size parameters (41), cell cluster size was expressed in terms of the equivalent spherical radius (r_s) based on the measured cluster volumes. This method is similar to the protocol of Jung and Lee (42), in which the equivalent circular colony radius was calculated based on the colony surface. Microcolonies were defined as cell clusters for which r_s ≥ 1.5 μm and were further divided into four different size categories: size I (1.5 μm ≤ r_s < 5.0 μm), size II (5.0 μm ≤ r_s < 10.0 μm), size III (10.0 μm ≤ r_s < 15.0 μm), and size IV (r_s ≥ 15 μm). Microcolonies with sizes ranging between 1.5 and 200 μm are reported in the literature (33); the four microcolony size subcategories in the present study were defined based on the experimental microcolony sizes (as computed from the CLSM images) to enable a balanced microcolony size distribution. Since L. monocytogenes cells are rod-shaped, measuring approximately 0.5 to 2.0 μm in length and 0.4 to 0.5 μm in width (43), cell clusters for which r_s < 1.5 μm were further categorized based on the height of a cylinder with equivalent volume and a diameter of 0.5 μm (i.e., the largest possible width of a single rod-shaped cell). Clusters for which this cylindrical height was ≤2 μm were categorized as single cells, while larger clusters were categorized as small aggregates. For the different model systems, Fig. 1 and 2 show the size distributions of the cell clusters in the aforementioned categories by means of the number and volume distributions, respectively.

FIG 1.

Number distribution (%) of single cells, small aggregates, and microcolonies with different sizes for the GFP L. monocytogenes strain after 14 days of growth at 10°C in 11 model systems, i.e., liquid, xanthan, aqueous gel, emulsions (Em) with four different fat contents, and gelled emulsions (GE) with four different fat contents. Microcolony sizes (μm) are expressed in terms of the equivalent spherical radius (r_s).

FIG 2.

Volume distribution (%) of single cells, small aggregates, and microcolonies with different sizes for the GFP L. monocytogenes strain after 14 days of growth at 10°C in 11 model systems, i.e., liquid, xanthan, aqueous gel, emulsions (Em) with four different fat contents, and gelled emulsions (GE) with four different fat contents. Microcolony sizes (μm) are expressed in terms of the equivalent spherical radius (r_s).

Figure 3 represents the distribution of the selected GFP L. monocytogenes strain in the three different model systems without fat, i.e., liquid, xanthan, and aqueous gel. Figures 1 and 2 illustrate that L. monocytogenes mainly grew as small aggregates and microcolonies in these model systems. In the liquid model system (Fig. 3A), L. monocytogenes grew mainly as small aggregates and size I microcolonies. While the number of small aggregates was higher than the number of microcolonies, most of the volume was taken in by the microcolonies. In the xanthan system (Fig. 3B), a large number of small aggregates and size I microcolonies were present. However, the two larger microcolonies of size II (i.e., r_s of 8.1) and size III (i.e., r_s of 13.9 μm) accounted for 95% of the total volume of L. monocytogenes in xanthan. In the aqueous gel (Fig. 3C), the cells were, in absolute numbers, rather equally divided between small aggregates and size I, II, and III microcolonies, while most of the cell volume was represented by size II and III microcolonies.

FIG 3.

Growth of L. monocytogenes in the liquid (A), xanthan (B), and aqueous gel (C) model systems. Cells are depicted in green.

Figure 4 illustrates the growth behavior of L. monocytogenes in the emulsion and gelled emulsion model system containing 1% fat. In both systems, fat droplets with a diameter of ∼1 μm were present, and L. monocytogenes grew in the space among these fat droplets. However, the green and yellow areas that were observed outside the orange areas indicated that the bacterial cells also grew around the fat droplets on the fat-water interface. In the emulsion containing 1% fat (Fig. 4A), the cells mainly grew as small aggregates and size I microcolonies, with the latter representing the largest cell volume. In the gelled emulsion containing 1% fat (Fig. 4B), a similar growth behavior was observed, although size II microcolonies also represented 11% of the cell volume.

FIG 4.

Growth of L. monocytogenes in the emulsion (A) and gelled emulsion (B) model systems with a fat content of 1%. Cells are depicted in green, while fat droplets are depicted in orange. Yellow areas represent L. monocytogenes growth on the fat-water interphase.

The growth of the GFP L. monocytogenes strain in the model systems with higher fat content (i.e., 5, 10, and 20%) is illustrated in Fig. 5 and 6 for the emulsions and gelled emulsions, respectively. With increasing fat content, bacterial growth on the fat-water interface was dominant over growth in the aqueous phase among the fat droplets. Concerning the number and volume distributions of cells, the growth behavior was relatively similar in emulsions and gelled emulsions with equal fat contents. In systems containing 5% fat, most cell clusters appeared as small aggregates and size I microcolonies, with the latter category representing the largest volume percentage. In the emulsions and gelled emulsions containing 10 and 20% fat, small aggregates and size I microcolonies were the most prominent in absolute numbers, while smaller percentages of size II and IV microcolonies were also seen. However, these size IV microcolonies (i.e., r_s of 30.6, 39.4, 16.3, 27.5, and 37.2 μm) represented between 90 and 100% of the total cell volume. These relatively large microcolonies seem to have been formed by the connection of microcolonies on different fat droplets, as can be observed in Fig. 5 and 6 (see panels B and C).

FIG 5.

Growth of L. monocytogenes in the emulsion model systems with different fat contents, i.e., 5% (A), 10% (B), and 20% (C). Cells are depicted in green, while fat droplets are depicted in orange. Yellow areas represent L. monocytogenes growth on the fat-water interphase.

FIG 6.

Growth of L. monocytogenes in the gelled emulsion model systems with different fat contents, i.e., 5% (A), 10% (B), and 20% (C). Cells are depicted in green, while fat droplets are depicted in orange. Yellow areas represent L. monocytogenes growth on the fat-water interphase.

MATS assay.

Table 1 shows the results of the MATS assay for the GFP strain, the strain cocktail used for the macroscopic growth experiments by Verheyen et al. (22), and the three separate strains of the cocktail (i.e., LMG 23773, LMG 23774, and LMG 26484). In general, the affinities for the polar solvent (i.e., diethyl ether) were higher than for the nonpolar solvent (hexane). The affinity for diethyl ether was significantly higher for LMG 23773 than for the other strains. The highest affinities for hexane were observed for LMG 23773 and the GFP strain. For each strain, the affinity to the polar solvent was also higher than the affinity to the nonpolar solvent.

TABLE 1.

Affinity to diethyl ether and hexane of the four different tested Listeria monocytogenes strains and a strain cocktail consisting of LMG 23773, LMG 23774, and LMG 26484 according to the MATS assay

Strain or strain combination	Mean affinity (%) ± SEMa
	Affdiethyl ether	Affhexane
LMG 23773	53.66 ± 9.11B,b	8.64 ± 2.35B,a
LMG 23774	34.25 ± 5.69A,b	−5.54 ± 2.38A,a
LMG 26484	29.73 ± 5.45A,b	0.52 ± 4.56A,B,a
GFP strain	30.16 ± 3.06A,b	10.83 ± 6.05B,a
Strain cocktail	30.25 ± 5.94A,b	−2.34 ± 4.41A,a

^aAmong the four different strains and the strain cocktail, solvent affinity values bearing different uppercase letters are significantly different (P ≤ 0.05). For each strain, affinity values to the different solvents bearing different lowercase letters are significantly different (P ≤ 0.05).

DISCUSSION

Growth morphology.

Verheyen et al. (22, 31) made a number of assumptions concerning the growth morphology of the L. monocytogenes strain cocktail (consisting of LMG 23773, LMG 23774, and LMG 26484) in the investigated fish-based model systems with various microstructures, i.e., liquid, xanthan, aqueous gel, emulsion (1% fat), and gelled emulsion (1% fat). First of all, it was assumed that L. monocytogenes grew as single cells in the liquid system, although potential cell sedimentation due to the static nature of the growth experiments was also suggested. Second, visual inspection during the macroscopic growth experiments indicated the occurrence of colony growth in the xanthan model system, probably caused by the higher viscosity in comparison to the liquid system. Since the viscosities of the xanthan and emulsion model system containing 1% fat were rather similar, it was assumed that colony growth would also be present in the emulsion model system. Furthermore, it was assumed that L. monocytogenes grew as colonies in the aqueous gel and the gelled emulsion containing 1% fat. It is important to mention that the distinction between micro- or macrocolonies could not be made based on the macroscopic growth experiments. The present study shows that colony growth in the model systems could in fact be classified as microcolony growth.

The assumption of the predominant presence of single cells in the liquid system (Fig. 3A) was not confirmed in the present study, since L. monocytogenes mainly grew as small aggregates and size I microcolonies (i.e., 1.5 μm ≤ r_s < 5 μm). In this regard, bacteria are known to form small aggregates and more dense clusters when grown in liquid systems, especially at static conditions (44). The sedimentation of cells during the 14 days of incubation at 10°C is a plausible explanation for the presence of small aggregates and microcolonies (45, 46). For the xanthan system (Fig. 3B) and the emulsion containing 1% fat (Fig. 4A), the assumption of microcolony growth was confirmed, since the largest cell volume was represented by microcolonies. However, the situation was more complex than assumed, since a large number of small aggregates and some single cells were also present in these two systems. Furthermore, microcolonies grew to significantly larger sizes in the xanthan system than in the emulsion system, indicating that, even at a low fat content of 1%, microcolony size is constrained by the presence of fat droplets. This finding contradicts previous studies on bacterial growth in oil-in-water emulsions for which the main conclusion was that planktonic growth is predominant in emulsions with fat content lower than 80% (21, 26). Not only did the cells grow as small aggregates and small microcolonies, but their colony size was also limited by a fat content significantly lower than 80%. For the aqueous gel (Fig. 3C), the assumption of microcolony growth was also mostly confirmed. Although a substantial number of small aggregates was also detected in the system, most of the cell volume was represented by microcolonies. In the gelled emulsion containing 1% fat (Fig. 4B), the assumption of microcolony growth was also mostly confirmed, again in addition to a large number of small aggregates which only represented a limited percentage of the total cell volume. Microcolonies in the gelled emulsion were generally smaller than in the aqueous gel, probably due to the space limitations caused by the presence of the fat droplets.

Intercolony distances of L. monocytogenes microcolonies can also be investigated in Fig. 3 and 6. The inoculation level of the growth experiments conducted in the present study and by Verheyen et al. (22) was 10² CFU/ml. This low inoculation level has been reported to lead to growth of large microcolonies, far apart from each other (i.e., 1.5 to 5.0 mm) with no intercolony interactions (47). Intercolony distances of the aforementioned order of magnitude could be present among the larger microcolonies in the xanthan system, since only one size III microcolony (i.e., 13.9 μm) was visible in Fig. 3B, implying that more distant large microcolonies could be located at 1.5 to 5.0 mm of the visible microcolony. The absence of these larger intercolony distances in the other model systems was probably related to the limited mobility of the bacterial cells in comparison to the xanthan system. Possible causes for this limited mobility include (i) sedimentation of cells in the liquid system, (ii) immobilization of cells in the aqueous gel and gelled emulsion, and (iii) the presence of fat droplets in the emulsion and gelled emulsion.

In general, the growth morphologies of L. monocytogenes in the different model systems, as assumed by Verheyen et al. (22, 31) for the macroscopic growth experiments, were more simplistic than those observed in the microscopic images in the present study. L. monocytogenes often appeared as a combination of single cells, small aggregates and microcolonies varying in size, in contrast to the more simple classification that was previously assumed, i.e., growth of single cells in the liquid system and submerged microcolony growth in the xanthan, emulsion system, aqueous gel, and gelled emulsion system.

Preferred phase for cell growth.

Verheyen et al. (22, 31) assumed that the aqueous phase was the preferred phase for cell growth in the emulsion and gelled emulsion systems. However, Fig. 4 and 6 illustrate that L. monocytogenes showed a preference for growth around the fat droplets on the fat-water interface, a trend which became more evident in systems with higher fat content (i.e., 5, 10, and 20%). Although previous studies have reported that bacteria grow exclusively in the aqueous phase of oil-in-water emulsions (see, for example, references 21, 48, and 49), some bacteria have been reported to have a preference for the fat-water interface in emulsion systems, e.g., demulsifying bacteria such as Alcaligenes sp. strain S-XJ-1 (50–52), and different bacteria in Emmental cheese (53). Therefore, the preference of L. monocytogenes to grow on the fat-water interface, as observed in this study, is not a totally isolated case. Under certain conditions, bacteria can adhere to oil droplets if their cell surface is (partially) hydrophobic or exhibits specific adherence features such as pili, fimbriae, and flagella (54, 55). L. monocytogenes cells are known to possess flagella at temperatures below 30°C (56, 57), promoting adhesion to inert solid surfaces such as polystyrene and stainless steel (58, 59). However, flagellum-induced L. monocytogenes adhesion to fat droplets has, to the best of our knowledge, thus far not been reported. Therefore, cell surface hydrophobicity was investigated (i.e., by means of the MATS assay) as a possible driving force behind the preference of L. monocytogenes to grow around the fat droplets in the present study. Since cell surface hydrophobicity of L. monocytogenes is strain dependent (60), the MATS assay was conducted for the selected GFP L. monocytogenes strain, the L. monocytogenes strain cocktail and for the three separate strains of the cocktail used in the macroscopic growth experiments (22) in order to check the transferability of the findings to the macroscopic scale.

No statistically significant differences were observed between affinities to the polar solvent of the GFP strain and the strain cocktail (and each separate strain of the cocktail except LMG 23773), while the affinity to the nonpolar solvent was significantly higher for the GFP strain than for the strain cocktail. However, these statistical differences for the affinity to the nonpolar solvent were mainly due to the negative value obtained for the strain cocktail. The occurrence of negative numbers was caused by small measurement variances (i.e., the optical density of the mixed sample being slightly higher than the optical density of the original cell suspension), meaning that negative values can be assumed to be equal to zero. In addition, only one of the three strains of the strain cocktail (i.e., LMG 23774) exhibited a significantly lower affinity to the nonpolar solvent than the GFP strain, also due to the negative value which was obtained for this strain. Hence, it is reasonable to assume that both the polar and nonpolar affinity of the GFP strain and the strain cocktail were similar.

The adhesion of the different strains ranged approximately from 30 to 50% for the polar solvent and from –5 to 11% for the nonpolar solvent. The combination of both a polar and nonpolar affinity for the investigated strains could explain the tendency of the cells to grow on the fat-water interface. This would mean that the partial affinity to the nonpolar fat phase starts to play a more important role when a decreased growth space is available in the aqueous phase (i.e., in systems with a higher fat content). However, in other studies (60–63), L. monocytogenes strains exhibited considerably higher affinities to nonpolar solvents (i.e., up to 96%) than in the present study. In addition, significantly higher affinities to polar solvents than to nonpolar solvents were observed in those studies, an opposite trend compared to the present study. Nevertheless, while cell surfaces in the aforementioned studies exhibited rather hydrophobic properties, the cells still adhered preferably to polar surfaces (e.g., stainless steel). Even though the comparison of cell surface hydrophobicity among different studies is not straightforward (due to the influence of, e.g., the physiological state of the cells, nutrient concentration, growth temperature, and growth phase [62–65]), it can be suggested that mechanisms other than cell surface hydrophobicity were more dominant causes for the preferred growth around the fat droplets in the present study. Future studies could focus on elucidating the exact causing mechanisms of the phenomenon by investigating, e.g., gene expression of L. monocytogenes in the presence of fat droplets, bacterial motility, and the presence or absence of specific adherence features such as pili, fimbriae, and flagella.

Comparison to macroscopic growth experiments.

Verheyen et al. (22) investigated the influence of food microstructure on the growth dynamics of the L. monocytogenes strain cocktail at 4 and 10°C at a macroscopic scale, using the liquid, xanthan, aqueous gel, emulsion (1% fat), and gelled emulsion (1% fat) model systems. An overview of macroscopic growth parameters (i.e., the lag phase λ and the maximum specific growth rate μ_max) obtained in the different model systems for growth at 4 and 10°C is provided in Table 2. Since the main objective of the macroscopic study was to isolate the microstructural effect on growth dynamics, macroscopic growth parameters could only be effectively compared among model systems which only differed in the form of a single isolated microstructural aspect. In this regard, a comparison of planktonic cells in the liquid system and submerged microcolonies in the xanthan system demonstrated that submerged microcolonies of L. monocytogenes grew faster (i.e., similar λ, higher μ_max) than planktonic cells, at least at static conditions (i.e., cultures which were not shaken). Furthermore, growth was faster (i.e., similar λ, higher μ_max) in viscous systems than in gelled systems, as illustrated by the higher μ_max in the xanthan system compared to the aqueous gel and in the emulsion system compared to gelled emulsion. Finally, fat droplets promoted growth (i.e., shorter λ, higher μ_max) at 4°C, illustrated by comparing growth in the xanthan system and the emulsion, and in the aqueous gel and the gelled emulsion. The results from the present study can be used to explain some of the findings from these macroscopic growth experiments, although possible differences in growth behavior between the L. monocytogenes strain cocktail and the GFP strain should be taken into account. In addition, assumptions made in the macroscopic study concerning L. monocytogenes growth morphology in the different model systems were proven too simplistic, as demonstrated above under “Growth morphology.” The complex behavior concerning the preferred phase for cell growth in the emulsion and gelled emulsion systems, as has discussed above in “Preferred phase for cell growth,” could also not be taken into account during the macroscopic growth experiments. Hence, the conclusions from Verheyen et al. (22) concerning the influence of bacterial growth morphology and the presence of fat droplets on L. monocytogenes growth dynamics should be interpreted critically.

TABLE 2.

Statistical analysis of macroscopic growth parameters (i.e., lag-phase λ and maximum specific growth rate μ_max) for the L. monocytogenes strain cocktail consisting of LMG 23773, LMG 23774, and LMG 26484^a

Model system	Mean λ (h) ± SEM		Mean μmax (1/h) ± SEM
	4°C	10°C	4°C	10°C
Liquid	124.2 ± 6.3C	14.0 ± 7.2A	0.029 ± 0.001B	0.091 ± 0.004B
Xanthan	89.5 ± 12.6B	16.2 ± 4.2A	0.029 ± 0.001B	0.101 ± 0.003C
Aqueous gel	93.3 ± 11.0B	28.5 ± 6.2A	0.026 ± 0.001A	0.074 ± 0.003A
Emulsion (1%)	45.0 ± 10.0A	19.2 ± 6.4A	0.031 ± 0.001C	0.094 ± 0.004B,C
Gelled emulsion (1%)	53.6 ± 13.7A	21.1 ± 7.2A	0.029 ± 0.001B	0.079 ± 0.003A

^aData were calculated according to the Baranyi and Roberts (32) model. For the different model systems at the same temperature, parameter values bearing different uppercase letters are significantly different (P ≤ 0.05). These values were adapted from Verheyen et al. (22).

In order to investigate the influence of L. monocytogenes growth morphology on microbial dynamics, macroscopic growth parameters in the liquid and xanthan system were compared. At 4°C, no significant differences in μ_max were observed between the two systems, while λ was longer in the liquid system. At 10°C, the maximum specific growth rate μ_max was higher in the xanthan system, while no significant differences were observed in λ. It was suggested that cells in the liquid model system might have sedimented due to the static nature (i.e., the tubes were not shaken during incubation) of the experiments. Therefore, oxygen availability would be lower for the cells in the liquid than in the xanthan system (45, 46). This assumption of sedimentation could be valid, since the present study shows that the number of small aggregates and small (i.e., size I) microcolonies in the liquid system was considerably higher than the number of single cells. Nevertheless, since the numbers of single cells were similar in the liquid and xanthan systems, differences in macroscopic growth parameters were probably mainly caused by the higher viscosity of the xanthan system, rather than by differences in bacterial growth morphology (i.e., between single cells and microcolonies).

The influence of the nature of the food matrix (i.e., viscous or gelled) on growth dynamics was investigated by comparing macroscopic growth parameters among (i) the xanthan system and the aqueous gel and (ii) the emulsion and the gelled emulsion containing 1% fat. A higher μ_max was observed in viscous systems than in gelled systems at 4 and 10°C, which could be explained by the enhanced nutrient, oxygen, and metabolite diffusion in the viscous systems. Based on the results of the present study, the difference in separation distance between the microcolonies in the viscous and gelled systems could be another possible explanation for the differences in μ_max, at least when comparing the xanthan system and the aqueous gel. Figure 3 illustrates that size III microcolonies (i.e., 10.0 μm ≤ r_s < 15.0 μm) in the aqueous gel were situated more closely together than those in the xanthan system. Since colony interactions from close spatial distribution of colonies occur up to separation distances of 1,400 to 2,000 μm (66, 67), the smaller separation distance between the microcolonies in the aqueous gel might also be an explanation for the higher μ_max in the xanthan system. Single cells, small aggregates, and size I and II microcolonies, however, were also located close to each other and to the size III microcolonies in the xanthan system, possibly also resulting in local depletion of nutrients and oxygen. In addition, the growth behavior of colonies depends on the colony size. Microcolony growth largely resembles planktonic growth, while macrocolony growth is slower than planktonic growth due to the presence of pH gradients and diffusion limitations around and inside the colonies (33). Since no macrocolonies were observed in any of the model systems, enhanced nutrient, oxygen, and metabolite diffusion in the viscous systems compared to the gelled systems remains the most probable explanation for the higher μ_max in the viscous systems.

The influence of fat droplets on L. monocytogenes growth was investigated by comparing macroscopic growth parameters between (i) the xanthan system and the emulsion containing 1% fat and (ii) the aqueous gel and the gelled emulsion containing 1% fat. The results showed that the presence of fat droplets was beneficial for the growth of L. monocytogenes (i.e., shorter λ and higher μ_max), although only at 4°C. Therefore, it was suggested that the presence of fat acts as a cryoprotective agent for L. monocytogenes growth, as concluded by Baka et al. (29). This behavior might be explained by the tendency of the cells to grow around the fat droplets, as can be observed in Fig. 4. Figures 5 and 6 illustrate that the affinity of the cells for the fat droplets seems to increase with increasing fat content. In general, a complex relationship between L. monocytogenes growth temperature and fat presence has been reported in the literature (22, 29), which could also be related to the preferred phase for cell growth. Future studies could combine macroscopic growth experiments and CLSM to investigate L. monocytogenes at different temperatures in emulsion and/or gelled emulsion systems with different fat content in order to get more insight into the cell growth on the fat-water interface and the resulting influence on macroscopic growth parameters. Similar to the concluding remarks of the “Preferred phase for cell growth” section, the influence of bacterial motility and the presence or absence of flagella on macroscopic growth parameters could also be investigated.

The findings of the present study entail significant implications for the microbial safety of processed fish-based food products, and food safety in general. In literature, assumptions concerning microbial growth morphology tend to be rather simplistic, as three different situations are normally distinguished based on the specific food microstructure, i.e., (i) planktonic growth in liquid products, (ii) submerged colony growth in gelled products, and (iii) surface colony growth on food surfaces (26, 68). In the present study, it was demonstrated that this classification does not always adequately describe real microbial behavior, not even in products with a homogeneous microstructure. In liquid products (e.g., the liquid and xanthan model systems in this study), a combination of single cells, small aggregates, and microcolonies can be present, with the distribution of the bacteria over this spectrum probably being dependent on the viscosity and potential shaking of the product. While the presence of small aggregates and microcolonies exerts no significant influence on microbial growth dynamics (33), microbial inactivation treatments (e.g., thermal inactivation, cold atmospheric plasma, antimicrobial compounds) are often less effective when such cell clusters are present in foods (69–71). As a consequence, the inactivation efficiency of preservation processes designed for liquid/viscous food products could be lower than estimated when cells do not exclusively grow in planktonic form. With the model systems in the present study being based on processed fish-based food products, such products containing a viscous aqueous phase (e.g., fish soup or certain fish curries) are potentially affected by the aforementioned consequences. In addition, bacteria growing on the fat-water interface (i.e., around fat droplets) in emulsion or gelled emulsion type food products could exhibit an increased growth potential and inactivation resistance compared to bacteria which solely grow in the aqueous phase of those products. With the applicability of the model systems of the present study in mind, these risks are especially relevant for processed fish products containing 1 to 20% fat (e.g., fish paté or fish sausage) (72–74).

Conclusions.

Microscopic (CLSM) growth experiments in fish-based food model systems with different microstructures (i.e., liquid, xanthan, aqueous gel, emulsion, and gelled emulsion) revealed that the growth morphology and the preferred phase for cell growth of L. monocytogenes were more complex than commonly assumed in macroscopic growth studies. Bacteria appeared as a combination of single cells, small aggregates, and microcolonies of different sizes, with the distribution over these categories being dependent on specific microstructural aspects of the respective model systems. This observation contradicts the traditional classification of planktonic growth in liquid/viscous systems and submerged colony growth in gelled systems. In emulsion and gelled emulsion systems, L. monocytogenes not only exclusively grew in the aqueous phase but also grew around the fat droplets at the fat-water interface, a trend which became more evident with increasing fat content. This preference for the fat-water interface most probably was not caused by a hydrophobic cell surface of the Listeria strains used, and the phenomenon should be further elucidated in future studies. Previously suggested causes for differences in microbial growth parameters (i.e., the lag phase duration λ and the maximum specific growth rate μ_max), based on macroscopic growth experiments, were validated or rejected by means of observations at the microscopic level concerning, e.g., growth morphology, microcolony size, intercolony separation distances, and the preferred phase for cell growth. The occurrence of microcolony growth in liquid/viscous foods on the one hand and growth on the fat-water interface in (gelled) emulsion type foods on the other could entail significant food safety implications. Under these conditions, pathogens potentially exhibit increased resistance to common food preservation techniques (e.g., thermal inactivation). Hence, this study demonstrated that combining experiments at the micro- and macroscales could be beneficial for the acquirement of increased insight into the food microstructural influence on microbial dynamics, concurrently leading to improved food safety.

MATERIALS AND METHODS

Microorganism and preculture conditions.

The GFP L. monocytogenes ScottA strain harboring the plasmid pNF8 (75) was kindly donated by Tine Rask Licht (National Food Institute, Technical University of Denmark). In order to maintain the structural stability of the constructed fluorescent plasmids, 10 μg/ml of erythromycin (Sigma-Aldrich, MO) and 100 μg/ml of nalidixic acid (Sigma-Aldrich, MO) were added to all growth media. Stock cultures were stored in Microbank (Pro-Lab Diagnostics, Ontario, Canada) at –80°C. One Microbank bead was transferred to 20 ml of brain heart infusion broth (BHI; VWR International, Leuven, Belgium) in a 50-ml Erlenmeyer flask, followed by incubation at 30°C for 24 h at static conditions. Afterward, 20 μl of the stationary-phase culture was inoculated into 20 ml of fresh BHI, followed by incubation for 24 h under the same conditions, resulting in stationary-phase cultures with a cell density of ∼10⁹ CFU/ml.

L. monocytogenes strains LMG 23773, LMG 23774 (both isolated from smoked salmon), and LMG 26484 (isolated from tuna salad) were acquired from the BCCM/LMG bacterial collection (Ghent University, Ghent, Belgium). Stock cultures were stored at –80°C in a mixture of 80% (vol/vol) BHI broth and 20% (vol/vol) glycerol (Acros Organics, NJ). For each strain, fresh purity plates were prepared by spreading a loopful of the stock culture onto a BHI agar plate (1.4% [wt/vol], agar technical No3 [Oxoid, Ltd., Basingstoke, UK]). After incubation at 30°C for 24 h, one colony from each purity plate was transferred to separate Erlenmeyer flasks containing 20 ml of BHI, after which the same procedure as for the GFP strain was followed. To prepare the strain cocktail, 10 ml from each culture (i.e., one of each strain) was collected under aseptic conditions and mixed, leading to a stationary-phase mixed culture with a cell density of approximately 10⁹ CFU/ml.

Model system preparation and inoculation.

Fish-based model systems with different microstructures were prepared according to the protocol of Verheyen et al. (31). The composition of the model systems was based on processed fish products (e.g., fish soup, surimi, and fish paté), while major food microstructural aspects of such products were also included (e.g., a visco-elastic matrix or fat droplets). A more detailed description of the model systems, as well as a detailed preparation protocol, is provided by Verheyen et al. (31). Briefly, the model systems were classified into five categories, i.e., liquid, xanthan (a more viscous liquid system containing a small concentration of xanthan gum), emulsion (oil-in-water), aqueous gel, and gelled emulsion. The liquid, xanthan, and emulsion systems were classified as viscous systems, while the aqueous gel and gelled emulsions were classified as gelled systems. In order to study the effect of fat content, emulsion and gelled emulsion systems containing different concentrations of sunflower oil were used, i.e., 1, 5, 10, and 20%. Prior to the pasteurization step (i.e., heating for 2 h at 80°C while being continuously stirred at 400 rpm) of the fat solutions (as described in Verheyen et al. [31]), Nile Red (Sigma-Aldrich, St. Louis, MO) was added to the fat solutions in powdered form to a concentration of 3 μg/g of fat. After the addition of 10 μg/ml of erythromycin and 100 μg/ml of nalidixic acid, model systems were homogeneously inoculated with the GFP L. monocytogenes strain to a cell density of 10² CFU/ml, using the inoculation procedure as described in Verheyen et al. (22). Inoculated systems were distributed over 35-mm-diameter glass bottom dishes with a 27-mm glass viewing area (Nunc; Thermo Fisher Scientific, Waltham, MA) suitable for confocal image analysis (4 ml per dish). Prior to CLSM imaging, model systems were incubated at 10°C for 14 days, resulting in early-stationary-phase cells with a cell density of approximately 10⁸ to 10⁹ CFU/ml (22).

Confocal laser scanning microscopy image acquisition.

Microscopic 3D images were recorded using the z-series dissection function of an A1R confocal laser scanning microscope (Nikon, Tokyo, Japan) at a 60× magnification (water immersion objective). Excitation wavelengths were 408 and 561 nm for the GFP strain and Nile Red, respectively. The recorded emission ranges were 500 to 550 nm and 570 to 620 nm. Images were processed using NIS-Elements C imaging software (Nikon, Tokyo, Japan). All experiments were independently performed in duplicate, and multiple images were taken for each experiment. The images that were chosen were those that were the most representative and clear for the observed phenomena.

Cell cluster size determination.

BioImageXD software (76) was used to calculate the volume of cell clusters (i.e., single cells, small aggregates, and microcolonies) on the confocal images. 3D images were constructed in the software by importing separate TIFF-files for each z-slice of the CLSM images. Noise was filtered using the “mean” function, with x, y, and z values of 3. Images were (manually) thresholded in order to acquire similar cell cluster distributions as for the original CLSM images. All green areas were separated into segmented objects with identifying colors using the “connected component labeling” function, and the volume of the segmented objects was quantified. The equivalent spherical radius (r_s) of all objects was calculated as the radius of a sphere with a volume equal to the object. Similarly, the equivalent cylindrical height (h_c) of the object was calculated, assuming that the object was a cylinder with a diameter of 0.5 μm. Objects were classified as microcolonies (r_s ≥ 1.5 μm), small aggregates (r_s < 1.5 μm and h_c > 2 μm), and single cells (r_s < 1.5 μm and h_c ≤ 2 μm). Microcolonies were further classified in four different size categories: size I (1.5 μm ≤ r_s < 5.0 μm), size II (5.0 μm ≤ r_s < 10.0 μm), size III (10.0 μm ≤ r_s < 15.0 μm), and size IV (r_s ≥ 15 μm). These size subcategory ranges were defined based on the experimental microcolony sizes computed from the CLSM images in order to enable a balanced microcolony size distribution. For each model system, number and volume distributions were calculated to quantify the distribution of cell clusters over the six different categories.

MATS assay.

The MATS assay was performed based on the protocols of Bellon-Fontaine et al. (40) and Takahashi et al. (60). Hydrophobicity of bacterial surfaces was assessed by comparing affinities to diethyl ether (i.e., a polar solvent) and hexane (i.e., a nonpolar solvent having intermolecular attraction comparable to that of diethyl ether). Precultures of the four different L. monocytogenes strains (i.e., LMG 23773, LMG 23774, LMG 26484, and the GFP strain) and the strain cocktail consisting of LMG 23773, LMG 23774, and LMG 26484 were prepared and grown as described in “Microorganism and preculture conditions.” The cells were washed twice with a NaCl solution of 0.90% (wt/vol), centrifuging at 18,500 × g for 10 min at 4°C. In order to reach an initial optical density of approximately 0.400 at 400 nm, the cells were 2.5-fold diluted with the NaCl solution. After we measured the initial (i.e., before mixing) optical density (A₀) of the diluted cell suspensions, 0.400 ml of each diethyl ether (Acros Organics, Geel, Belgium) or hexane (Acros Organics) was added to separate 2.400-ml aliquots of cell suspension. All mixtures were left to stand at room temperature for 10 min and subsequently vortex mixed for 1 min. The mixtures were again left to stand at room temperature for 20 min to allow phase separation of the aqueous and solvent phases, after which the optical density of the aqueous phase was measured. The affinity to each solvent was calculated as follows:

$${\text{Aff}}_{\text{solvent}}=100\left(1-\frac{A}{A_0}\right)$$

Aff_solvent is the affinity to a certain solvent, A is the optical density of the aqueous phase after mixing and settling, and A₀ is the optical density of the cell suspension before mixing. All optical densities were measured at 400 nm in a multiwell plate, using a VersaMax tunable microplate reader (Molecular Devices, Wokingham, UK). To each well, 250 μl of (aqueous) cell suspension was added. All experiments were performed independently in duplicate.

Statistical analysis.

Significant differences between the solvent affinities of the different L. monocytogenes strains and the strain cocktail were determined using analysis of variance (single variance) test at a 95.0% confidence level (α = 0.05). The Fisher least significant difference test was used to distinguish which means were significantly different from others. Standardized skewness and standardized kurtosis were used to assess whether data sets came from normal distributions. Analyses were performed using the Statgraphics Centurion 18 package (Statistical Graphics, WA). Test statistics were regarded as significant when P ≤ 0.05.