Detection of active pathogenic bacteria under stress conditions using lytic and specific phage

Salmonella enterica subspecies enterica serovar Typhimurium is a widespread foodborne and waterborne pathogenic bacterium (Bin et al. 2013). This species of bacteria is generally assessed using conventional methods based on the determination of viable and cultivable bacteria on selective media. However, under rush conditions and to survive, S. typhi, as other species of bacteria, could enter into the viable but nonculturable (VBNC) state (Ducret et al. 2014).

Bacteria in the VBNC state are unable to grow and to form colonies on the routine bacteriological media but remain alive and retain metabolic activity (Créach et al. 2003). This survival strategy adapted by bacteria under stress conditions such as starvation, disinfection, UV irradiation, etc. is characterized by significant modifications of bacterial morphology, metabolism, and possibly genomic structure (Arana et al. 2007). Compared with dead cells, VBNC cells have an intact membrane and conserve gene expression (Oliver et al. 2005).

VBNC bacteria can cause false-negative results and, as a consequence, can pose a significant and potential health risk (Oliver 2010). Indeed, a variety of human pathogens that enter the VBNC state can resuscitate and therefore start their metabolic activity, with the potential to initiate diseases under favorable conditions (Maalej et al. 2004). Moreover, Oliver (2010) suggested that antibiotics, which are highly active on viable and culturable cells, do not necessarily act on VBNC cells. This resistance to antibiotics supports the potential pathogenicity of these bacteria mainly after resuscitation.

For instance, the survival of pathogenic bacteria in VBNC state in treated water can represent a serious problem mainly when we reuse the water for agriculture or animal husbandry. Consequently, the inactivation of 99.99% of indicator bacteria after water disinfection cannot reflect the real quality of treated water (Besnard et al. 2002). Indeed, the use of traditional, culture-based microbiological analyses might lead to an underestimation or a misinterpretation of microbial water quality. Thus, to assess the effectiveness of water treatment processes and to control water safety, it would be pertinent to develop reliable methods for the detection of active-VBNC cells.

There are many non-culture techniques for the detection of VBNC bacteria. In recent years, a new differential staining assay, the BacLight Live/Dead assay, has been developed to determine if cell membranes of non-culturable bacteria remained intact after exposure to the respective treatment conditions (Fakruddin et al. 2013). Furthermore, for VBNC bacteria investigation we can also use molecular biology tools such as hybridization probes or the specific amplification of target DNA and fluorescent in situ hybridization (FISH). However, these procedures have lower sensitivity and cannot distinguish live and dead cells. In contrast, the determination of gene expression under experimental conditions and the reverse transcriptase polymerase chain reaction (RT-PCR) are more able to discriminate between culturable and non- culturable forms of target bacteria (Lleò et al. 2000)

All these detection methods need sophisticated and costly equipment. In this paper, we purpose the development of reliable methods for the detection of active-VBNC bacteria using a lytic and specific phage as a biosensor.

Phages have been used in different fields such as in wastewater treatment. For example, a cocktail of phages was used to enhance the physical water treatment process, UVC radiation (Ben said et al. 2010). Additionally, in the food industry, phages were investigated to eliminate pathogens in raw food and fresh food or to reduce food spoilage bacteria (Sulakvelidze 2013). Moreover, lytic phages were also used in medicine to irradiate antibiotic multi-resistant bacteria such as Pseudomonas aeruginosa in the case of cystic fibrosis (Cafora et al. 2019).

The study of the relationship between phages and host cells under stress conditions has highlighted the flexibility of the phage's replication cycle with the physiological state of host cells (Ducret et al. 2014). This positive interaction can be applied to check the viability of bacteria under unfavorable conditions.

The purpose of this study was to develop reliable and low cost methods to control the viability of starved bacteria. To achieve this goal, a specific phage with a short lytic cycle was investigated as a potential means to control the presence of active-VBNC S. typhi under starvation conditions. This method was based on the determination of phage infectivity parameters with the existence of potential active bacteria.

Salmonella typhi obtained from the American Type Culture Collection (ATCC 560) was used as a host for the isolation, propagation, and characterization of the bacteriophage named ST560Ø.

ST560Ø was isolated from wastewater and characterized based on its host range, and evaluated for its potential to detect viable bacteria.

Wastewater samples (30 ml) were cleared by filtration (0.22 μm pore size, Millipore, MA, USA) and incubated for 16 h. in the presence of exponentially growing S. typhi ATCC 560 cells at 37 °C. Cultures that showed visible lysis were pelleted by centrifugation at 4 °C and 8,000 × g for 20 min. Bacterial cell debris was removed by filtration (0.22 μm filters), and the presence of phages was verified by plating 100 μl of the filtrate on lawns of S. typhi using a standard soft agar overlay method (Carlson 2005).

For this agar overlay method, 100 μl phage solutions were added to 200 μl of an overnight culture of a susceptible S. typhi strain and mixed with liquid soft agar (50 °C). This mixture was spread on solid agar media, incubated overnight at 37 °C. The presence of phage was determined by the presence of visible zones of lysis or plaques.

The preparation of the bacteria was done according to Oliver et al. (2005). Briefly, bacterial cell cultures in the logarithmic phase were therefore harvested by centrifugation at 5000 rpm for 10 min at 4 °C and washed twice with autoclaved saline water (1 M NaCl solution, 0.9% bw). Then, the washed cells were filtered through a 0.22 μm Millipore membrane filter and resuspended in the sterilized water at a final density of 10⁶ CFU/ml.

The microcosm water system at a final concentration of approximately 10⁶ CFU/mL was used to starve S. typhi cells. Samples were then left in the dark at room temperature (10–90 days). To test the presence of the culturable bacteria, different samples were plated on nutrient agar plates at 37 °C for 18 h. All these studies were performed in triplicate.

After incubation (2 h at 37 °C) of Salmonella-Specific Phage (10⁵PFU) in the presence of increased density of host cells, with the host cells (0–10⁶ CFU/ml) kept in starvation condition (minimum broth culture), the cell-phage mixtures were filter sterilized using a 0.45 μm syringe filter to obtain free phage in the filtrate, and the supernatants were diluted in order to determine the titration of the phage.

The phage titer was determined by using a standard double-layer agar plate. The phage titration experiment was determined. To verify the sensitivity and the reproducibility of the assay, all phage titration experiments were done three times.

The presence of VBNC pathogenic bacteria in treated water may be considered an immediate public health risk. To resolve this problem, it's important to develop new, reliable and low-cost methods to detect the presence of potential active bacteria in treated water.

The present work aimed to detect VBNC- starved bacteria (S. typhi) before their resuscitation using a specific and a lytic phage. The modeling of the phage growth cycle with target host cells under experimental conditions was allowed to detect the presence of potential active bacteria underestimated by using conventional methods.

To control the existence of active-VBNC bacteria in oligotrophic water, starved bacteria with different concentrations (C_n: from 10¹ to10⁶ UFC/ml)) were mixed with specific and lytic phage.

The tested phage has a short lytic replication cycle with a latent period termination of about 20 min or less and phage-progeny release of about 45 to 50 min. The biosensor was inoculated in oligotrophic water samples with an initial phage titer (P₀) equal to 10⁵ PFU/ml. After contact time (2 h), the concentration of free phage titers (P_n) was determined for each water sample.

The analysis of results was based on the comparison of the free phage concentration (P_n) to the initial bio-inoculated phage titer (P₀), and on the determination of the phage's kinetic parameters using mathematical models such as the phage amplification rate (P_n/P₀).

Figure 1 shows a difference in phage amplification rate or burst size with the density of starved bacteria. In the mixtures including phage host cells, with an initial concentration equal to 10¹ and 10² CFU/ml, the phage amplification rates were shown to be augmented by respectively 10² and 10¹ PFU/ml compared to the initial bio-inoculated phage titer (P₀). The increase of phage amplification rate despite the VBNC state of the target bacteria was related to the positive interaction between the lytic phage and the starved bacteria. This increase of phage amplification confirms the achievement of the burst and the release of mature and infectious phages by starved-VBNC bacteria at low density. Thus, the lytic phage can be used as a sensor to control and detect the presence of active bacteria in an oligotrophic water sample undetectable in the usual media.

The decrease of phage amplification rate and the delay of burst in the mixtures including a higher density of starved cells can be explained by a different hypothesis, as follows:

However, under stress conditions, the major of phage kinetic parameters were changed. For example, a period of host selection and extracellular diffusion of phage progeny to a host cell (t_e) can be extended, allowing the disturbance of all stages of the phage life cycle (You et al. 2002). This presumption can be explained by the destabilization of the phage growth cycle in relationship with the bacterial physiological state marked by the depletion of the host growth rate. Besides, Ben Said et al. (2010) suggested that the timing of phage-induced host cell lysis may be related to the host quantity and mainly quality. Indeed, to survive, phages were able to react and change their life cycle as a response to stress and as an adaptation strategy to maintain the chance of survival (Mudgal et al. 2006; Golec et al. 2011).

Moreover, in contact with slow-growing bacteria under starved conditions, the phage cycle development may be stopped (Łoś et al. 2003). To surmount this trouble, lysogenization is a strategy that can be adopted by some groups of bacteriophages. Accordingly, when host bacteria were perturbed by stress such as starvation, the lytic phage can adopt this strategy to persist. Consequently, the genome of the phage can be maintained in the host cell until the bacterium's growth resumes, and then a burst of progeny phage is produced. This maintenance of the phage genome in the cell, without its replication, is called pseudolysogeny (Łoś et al. 2003). It is known that the latent period is defined by the timing of phage-induced host cell lysis, which is typically under the control of a phage protein complex known as a holing (Abedon et al. 2001). The suspension of burst and the release of mature progeny phages with an extension of a latent period can be considered as a beneficial strategy to ensure the persistence of the phage and keep the bacterial host viable. In the same context, Erez et al. (2016) suggested that phages use a small-molecule communication system to coordinate lysis–lysogeny decisions.

The depletion of free phage released in contact with a high density of starved bacteria could also be attributed to morphological change of the host cells. Indeed, Zeng et al. (2013) showed that the VBNC cells were smaller than the normal viable and culturable bacteria and tend to assemble as a response to stress conditions. In addition, Marsh et al. (1998) suggested that after starvation, and due to the bacteria's resistance to lysis, we must add a lysozyme into the DNA extraction method to allow more efficient DNA extraction. Consequently, the modification of bacterial shape after starvation could directly influence the phage cycle, mainly the latent period.

Additionally, we cannot neglect the role of bacteria in delaying the achievement of the phage cycle. Indeed, the phage-host interactions can be controlled by quorum sensing (QS). Indeed, new knowledge continues to be revealed by the important role of quorum sensing in regulating phage susceptibility (Kamran et al. 2014; Saucedo-Mora et al. 2017).

QS is a bacterial cell-cell communication process that involves the production, detection, and response to extracellular signaling molecules called autoinducers (Rutherford & Bassler 2012). This communication system is often used to coordinate social behaviors such as virulence and biofilm formation across the population (Popat et al. 2015). In addition, Lazazzera (2000) suggested that under starvation conditions, quorum-sensing pathways converge with starvation-sensing pathways to regulate cell entry into the stationary phase.

Saucedo-Mora et al. (2017) supposed that quorum sensing could additionally control the susceptibility to the phage and, consequently, this system may serve as a strategy to protect bacteria from phage infection (Hoyland-Kroghsbo et al. 2013; Kamran et al. 2014). This hypothesis can explain the fact that when we infect a high density of starved bacteria with a lytic phage (C_n equal to 10³–10⁶ CFU/ml) we cannot appreciate the real phage amplification rate due to the phage infectivity quenching by host cells.

To resume, in contact with a high density of starved-host cells, the phage needs more time to select a host cell and to modulate its growth cycle according to the physiological state of the bacteria. This supplemental time was added to the latent period causing the production of a small burst of progeny phage.

These hypotheses can limit the use of the phage infectivity rate to estimate the viability of a high density of VBNC bacteria under stress conditions within a short contact time. However, we cannot ignore the ability of phage to interact with their potential active-host cells. Indeed, the reduction of free phage titers compared to the initial bio-inoculated titer (P₀) reveals the ability of ST-phages to recognize and to attach to their host cells and probably convert their standard cycle to persist and overcome hostile conditions with a probability of burst and release of new infectious phages. Consequently, the decrease of phage amplification rate was not related to the absence of active bacteria in a water sample, but, on the contrary, the reduction of free phages can reveal the existence of potential intracellular mature phages that we can call ‘engaged phages’. This type of phage can be released after an extension of the latent period.

By homology, the ‘engaged’ phages can adopt the same survival strategy as a stressed bacteria, namely the VBNC state. Phages under hostile conditions can hide their lysis ability and maintain the intracellular mature progeny non-bursting state to increase the probability of survival and persistence.

Figure 2 shows the prediction of the amount of intracellular-mature phage or ‘engaged phages’ using Equation (4). These ‘engaged’ phages could be used as tools for the detection of active but non-cultivable bacteria underestimated by using classic assessment methods. Indeed, after a short contact time (2 h), the decrease of free phages (compared to the initially injected phage titer P₀) can report the existence of ‘engaged phages’ not yet released by a putative active bacteria.

As a complement to the first part of this study, and to understand the impact of starved cell density on phage infectivity, and mainly to highlight the capacity of starved-VBNC host cells to ensure the phage lytic cycle, salmonella-specific phage growth parameters were determined after prolongation of contact time 18 h (P’_n) with a different concentration of potential starved host cells (C_n).

After an extension of contact time (18 h), Figure 3 shown an increase of the amount of free phage (P’_n) in the mixture of phage-starved hosts mainly, with an initial bacterial concentration higher than 10² CFU/ml.

The difference between (P_n) and (P’_n) can reflect the degree of phage inhibition amplification and release exerted by host cells after a short contact time with phage/target cells under stress conditions (2 h). As explained in the first part of this study, after 2 h of interaction with phage-starved hosts, the specific phage can establish contact with the target bacteria via absorption without burst and release of progeny infectious phages by switching of the phage lytic cycle to peudolysogenic cycle as a response to stress conditions. In addition, the later burst size can be related to the host cells quenching controlled by the quorum-sensing system, mainly for a high density of stressed bacteria.

The increase of free phages titers after an extension of contact time confirms, as a result, the existence of intracellular mature not yet released phage (P_e) by a potential active bacteria and demonstrating, consequently, the ability of potential starved-VBNC bacteria to achieve the phage cycle and release new and infectious phages. The intracellular ‘engaged phages’ required an extension of the latent period until the host bacteria removal of the phage quenching to achieve the virus life cycle. Subsequently, the phage lysis step delay and the stretching of the phage cycle were directly correlated with the physiological state and the adaptation strategy of viable host bacteria to surmount stress conditions.

Figure 4 shows that the amount of calculated free phages (Pcal) determined according to Equation (5) was slightly higher than the experimental phage titers, especially for the samples including an increasing density of potential starved host cells (Figure 4). This difference is related to the density of infected host cells, in which phages were able to infect host cells without the probability of the production of virions by bursting. The inhibition or quenching of the phage cycle which is revealed by the decrease in the level of phage amplification can report the impact of starvation on bacterial viability and activity.

In fact, under stress conditions, the bacterial suspension can include a variety of bacterial states: viable and culturable bacteria, dead bacteria, VBNC bacteria without a probability of resuscitation, VBNC bacteria with a probability of recovery and retention of their virulence factors. This last type can be qualified as active-VBNC bacteria that can be able to initiate disease following their resuscitation to the actively metabolizing state. Despite this heterogeneity, phages were able to select a potential ‘good quality’ of host cells that can allow the phage cycle with a high probability of release of mature and infectious virus.

The putative active host cell groups (ß) and instantly permissive bacteria (S) that can keep the viability and the permissiveness to the phage could also maintain the metabolic activity and the expression of their virulence factor, being thus a reservoir and representing a public health hazard. Therefore, the phage infectivity rate can mimic bacterial activity and the ability of target microorganisms to establish many mechanisms to survive in a hostile environment.

The monitoring of phage amplification rates in the presence of a variable concentration of host cells under stress conditions has revealed a fluctuation of phage cycle (by an increase or decrease) concerning the lysis time and burst size. This fluctuation can be translated into data allowing the detection of active but non-culturable bacteria in the analyzed sample.

Accordingly, as a perspective on this work, the phage amplification rates determined in relation with target bacteria under stress conditions can be converted to bacterial activity signals at different levels with thresholds using software.

It would be pertinent to take into consideration, after water disinfection, the switching of certain species of bacteria such as Salmonella typhi to the VBNC state. Thus, it is important to develop reliable, simple, sensitive and low-cost techniques to control bacteria's survival and viability in the final product (water, foods, etc.).

Under stress conditions (starvation, water disinfection, etc.), the bacterial suspension can include a variety of bacterial viability states, mainly VBNC cells undetectable using classic methods based on viability and cultivability in the usual media. Despite this heterogeneity, phages were able to select a potential ‘good quality’ of host cells that can allow the phage cycle with a high probability of achieving their life cycle and releasing a mature and infectious virus. Accordingly, the phage infectivity rate can be used as a biosensor to detect active bacteria that are not detectable by conventional methods.

The results showed that the phage cycle was modulated by the density of active bacteria and the viability rate of target bacteria. Indeed, the phage can detect the presence of active bacteria within 2 hours when the starved-VBNC bacterial density was lower than 10² potential active bacteria/ml. However, when the density of VBNC bacteria exceeds 10²cells/ml the extension of the phage latent period proved to be indispensable in permitting the stabilization of the phage cycle and detecting the presence of potential active bacteria.

The use of the lytic phage as a biosensor, for a sensitive control test to detect the presence of potentially viable and active bacteria non-detectable by conventional methods, can be considered as a simple and not costly method to detect the viability state of target host cells, mainly pathogenic cells or fecal indicator bacteria. Thus, this method can be applied to monitor microbiological water quality (after disinfection or storage) or to verify the safety of agri-food products. Furthermore, this method can be used to complement conventional techniques to avoid hazard to public health, particularly related to the resuscitation of active-non culturable cells, especially for pathogenic bacteria.

This work is supported by CERTE contract programs funded by the Ministry of Higher Education and Scientific Research of Tunisia.