Shigella sonnei Does Not Use Amoebae as Protective Hosts

ABSTRACT Shigella flexneri and Shigella sonnei bacteria cause the majority of all shigellosis cases worldwide. However, their distributions differ, with S. sonnei predominating in middle- and high-income countries and S. flexneri predominating in low-income countries. One proposed explanation for the continued range expansion of S. sonnei is that it can survive in amoebae, which could provide a protective environment for the bacteria. In this study, we demonstrate that while both S. sonnei and S. flexneri can survive coculture with the free-living amoebae Acanthamoebae castellanii, bacterial growth is predominantly extracellular. All isolates of Shigella were degraded following phagocytosis by A. castellanii, unlike those of Legionella pneumophila, which can replicate intracellularly. Our data suggest that S. sonnei is not able to use amoebae as a protective host to enhance environmental survival. Therefore, alternative explanations for S. sonnei emergence need to be considered. IMPORTANCE The distribution of Shigella species closely mirrors a country's socioeconomic conditions. With the transition of many populous nations from low- to middle-income countries, S. sonnei infections have emerged as a major public health issue. Understanding why S. sonnei infections are resistant to improvements in living conditions is key to developing methods to reduce exposure to this pathogen. We show that free-living amoebae are not likely to be environmental hosts of S. sonnei, as all Shigella strains tested were phagocytosed and degraded by amoebae. Therefore, alternative scenarios are required to explain the emergence and persistence of S. sonnei infections.

as environmental hosts to protect it from water sanitation measures that are implemented in transitional countries (10). Amoebae are free-living organisms found in a variety of water sources, such as swimming pools and lakes, as well as in soil and dust. Importantly, they have even been found in chlorinated public water sources in developed countries (11). They are able to tolerate harsh and changing conditions, making them a good host for a variety of bacteria (12). Legionella pneumophila is the most well-known bacterium known to utilize amoebae as protective hosts, but Campylobacter jejuni, Salmonella enterica serovar Typhimurium, and Vibrio cholerae have also been shown to survive intracellularly in amoebae (13)(14)(15)16). However, some bacteria which were initially described as surviving in amoebae have later been shown to grow extracellularly, potentially through saprophytic growth on dead amoebae or amoeba waste (17)(18)(19)(20).
Previous work has suggested that S. sonnei can survive in amoebae for extended periods of time. S. sonnei, Shigella dysenteriae, and S. flexneri were all found to be phagocytosed by Acanthamoebae castellanii; however, only S. sonnei and S. dysenteriae appeared to survive and replicate in the cytosol (21).
Here, we explore the hypothesis that amoebae can act as an environmental reservoir for S. sonnei. Although S. sonnei is phagocytosed by amoebae, we found no evidence that S. sonnei is able to survive and replicate in the cytosol of A. castellanii.

RESULTS
Shigella cells survive extended coculture with amoebae. Consistent with previous research, we observed that strains of S. sonnei and S. flexneri were able to survive in coculture with A. castellanii over 18 days at 22°C. We used two S. flexneri serotypes (strain M90T, serotype 5a, and strain 2457T, serotype 2a) and two S. sonnei isolates (the commonly used 53G strain and a recent clinical isolate, H140860381, here referred to as 381). All strains remained culturable at 10 6 to 10 7 CFU/ml over the 18 days (Fig. 1A). The amoebae were also maintained at ca. 5 ϫ 10 4 cells/ml throughout this time period (Fig. 1B). These data indicate that Shigella species can survive extended coculture in the presence of amoebae but give no information as to whether the bacteria are residing within amoebae and potentially utilizing the amoebae as an environmental reservoir.
We determined the intracellular bacterial numbers by taking samples at the indicated time points, treating with gentamicin to kill extracellular bacteria, and lysing the amoebae prior to CFU determination (Fig. 1C). This analysis revealed that all bacterial strains could be recovered intracellularly at all time points. However, fewer intracellular bacteria were recovered at the later time points. No difference was observed in intracellular bacterial numbers between the S. flexneri and S. sonnei strains at any time point.
The intracellular bacteria observed in this assay could be recently phagocytosed bacteria that had not yet been degraded, or bacteria that had established an intracellular niche and were surviving and/or replicating. We sought to examine these possibilities further.
All Shigella strains are phagocytosed by A. castellanii. To determine the efficiency of phagocytosis of Shigella strains, amoebae incubated in low-nutrient medium (Page's modified Neff's amoeba saline [PAS]) at 22°C were allowed to phagocytose bacteria for 1 h, which was followed by 1 h gentamicin treatment to kill extracellular bacteria. The numbers of total bacteria (prior to gentamicin treatment) were similar for all strains and indicated that, on average, 5 ϫ 10 6 CFU/ml, or approximately 50 bacteria/amoeba, were present ( Fig. 2A). Following gentamicin treatment, recoverable CFU decreased by 4 log, indicating that the majority of the bacteria were extracellular or were rapidly degraded by the amoebae during the gentamicin incubation (Fig. 2B). There was no significant difference in the numbers of amoeba-associated bacteria or phagocytosed bacteria between nonpathogenic Escherichia coli strain MG1655 and the Shigella strains, or between the Shigella species.
We repeated the experiments in high-nutrient medium (peptone-yeast-glucose medium [PYG]) to increase the rate of phagocytosis by amoebae. As anticipated, significantly higher numbers of phagocytosed bacteria could be enumerated than in low-nutrient medium (Fig. 2C). On average, there were 0.5 intracellular bacteria/ amoeba. However, we again saw no difference between the phagocytosis rates for S. flexneri or S. sonnei strains. All Shigella strains showed a small, nonsignificant trend of increased phagocytosis by amoebae compared to that for nonpathogenic E. coli. Microscopic analysis of differentially stained bacteria confirmed the presence of intracellular bacteria for both S. sonnei and S. flexneri (Fig. 2D). Extracellular bacteria can be seen adhered to the plastic, rather than to the amoebae, suggesting that the amoebae efficiently phagocytose all bacteria contacted.
Shigella does not survive intracellularly in A. castellanii. We tested the intracellular survival of Shigella by CFU determination at 1 h, 4 h, and 20 h. Cell counts for all strains decreased over this time frame similarly to those of the negative control of nonpathogenic E. coli, indicating they were efficiently digested by A. castellanii. The same trend of reduced intracellular numbers over time was observed in both high- (Fig.  3A) and low-nutrient media (Fig. 3B), with a 2-log decrease between 1 h and 20 h. Due to the low number of phagocytosed bacteria in low-nutrient medium, by 20 h all strains were below the limit of detection, unlike in high-nutrient medium, where approximately 1,000 bacteria/sample were still recoverable.
To confirm that the amoebae were capable of facilitating intracellular growth, L. pneumophila was used as a positive control. These experiments were conducted at 30°C, as this is the optimal temperature for L. pneumophila growth (Fig. 3C). As expected, the numbers of culturable wild-type L. pneumophila cells increased over the 20 h of incubation, whereas cell counts for all Shigella strains again decreased, in line with those of nonpathogenic E. coli.
To investigate the intracellular fate of bacteria, we observed the association of bacteria with amoebae stained with neutral red (Fig. 3D). Neutral red preferentially accumulates in lysosomes, due to their relative acidity (22,23). S. flexneri, S. sonnei, and E. coli were all observed to have neutral red structures accumulating around the intracellular bacteria and to be infiltrated with neutral red, suggesting they were being digested by the amoebae (24). This provides a visual confirmation of the intracellular killing observed in the preceding assays by bacterial enumeration.
The T3SS does not alter Shigella interaction with amoebae. Considering the importance of the type 3 secretion system (T3SS) for Shigella virulence, the intracellular survival assays were repeated at 37°C, the temperature at which the T3SS is active and effector proteins are translocated (25). Shigella bacteria induce different T3SSdependent outcomes depending on the cell type infected; in epithelial cells, vacuolar escape and intracellular replication, and in macrophages, vacuolar escape and pyroptosis.
If the T3SS facilitated intracellular survival within amoebae, we would expect to see increased intracellular bacterial counts at 4 h and 20 h during incubation at 37°C. Instead, we saw a decrease in viable intracellular bacteria numbers, similar to those at 22°C and 30°C, suggesting that an active T3SS could not facilitate intracellular survival in amoebae (Fig. 4A). To further investigate the involvement of the T3SS, the intracellular survival of T3SS mutants was determined. Again, no difference in intracellular survival between wild-type and T3SS mutants was found (Fig. 4B), indicating that the T3SS was not altering the interaction of Shigella with amoebae.
It was previously reported that S. flexneri used its T3SS to kill amoebae (21,26). Having not seen an effect of the T3SS (Fig. 4A and B), or significant amoeba death upon long-term exposure to S. flexneri (Fig. 1C), we decided to investigate amoeba cell death further by using a propidium iodide (PI) assay to measure the membrane integrity of the amoebae (Fig. 4C). At all temperatures analyzed, there were no significant differ-ences in PI levels between amoebae infected with the negative controls (E. coli MG1655 and Shigella T3SS mutants) and any of the wild-type Shigella strains. Therefore, in our assays neither S. flexneri or S. sonnei were able to induce cell death in amoebae.
S. sonnei is not released by amoebae. V. cholerae was recently shown to resist intracellular killing by A. castellanii, and at low frequency it can be released intact by the amoebae (16). While we have seen no evidence of Shigella resisting intracellular killing, we questioned whether a small number of bacteria were being released from the amoebae. Following killing of extracellular bacteria with gentamicin treatment, fresh medium with no gentamicin was added to cells, and the cell supernatant was harvested after 4 h (Fig. 4D). Low numbers of Shigella were recovered from the supernatant, and these could potentially be a source of Shigella for infection. However, there was no significant difference in bacterial release between S. sonnei and S. flexneri strains, and indeed, no significantly increased numbers of bacteria were released compared to those for nonpathogenic E. coli, indicating this is not a Shigella-or S. sonnei-specific mechanism for dispersal.
Intracellular Shigella bacteria are not more infectious. Amoebae have been proposed to act as "training grounds" for intracellular pathogens, adapting them to an intracellular lifestyle (27). While this is considered a long-term adaptation, we ques- tioned whether it facilitated short-term infectivity as well. We therefore tested whether bacteria harvested from amoebae were more proficient at invading or replicating within mammalian cells. We found the bacteria harvested from amoebae were less able to invade and survive in mammalian cells ( Fig. 5A and B). These findings support the previous conclusion that Shigella bacteria are being degraded by the amoebae, rather than adapting and surviving.

DISCUSSION
The frequency of S. sonnei isolation directly correlates with per capita gross domestic product (GDP) (28). The underlying reason(s) for this association is not understood, although a number of hypotheses have been proposed, one of which is that S. sonnei uses amoebae as a protective host (10). We show here that S. sonnei has no survival advantage in amoebae compared to S. flexneri, or indeed, compared to nonpathogenic E. coli. Both Shigella species were able to survive in long term coculture assays in low-nutrient medium suggesting that, like Listeria monocytogenes, Shigella species can utilize amoeba debris for nutritional requirements (17). However, this offers no explanation for why S. flexneri, but not S. sonnei, levels of infection are reduced in areas where living conditions and water sanitation are improved.
Having disproven one hypothesis explaining the emergence of S. sonnei, it remains to experimentally test additional hypotheses. One popular suggestion is that exposure to unsanitized water in developing countries can result in Plesiomonas shigelloides infection and hence in natural immunity against S. sonnei. P. shigelloides serotype O17 The intracellular survival was also determined for S. flexneri M90T and S. sonnei 53G T3SS mutants and compared to those of parental strains. (C) Propidium iodide uptake was used to measure membrane integrity in amoebae exposed to the indicated strains for 3 h. All values are a percentage of the maximum PI uptake calculated for amoebae exposed to 0.25% sodium deoxycholate for 10 min. (D) The indicated bacterial strains were incubated with A. castellanii for 1 h and then washed and treated with gentamicin for 1 h to kill extracellular bacteria. Fresh medium without gentamicin was then added for a further 3 h, after which time the supernatant was harvested and the number of released bacteria determined by colony counting. For all assays, two-way ANOVA indicated no significant differences between any strains at any time point. Mean and a standard deviation of 3 repeats are shown. has a lipopolysaccharide O-antigen identical to that of S. sonnei (10). People living in areas with good water sanitation would therefore have reduced exposure to P. shigelloides, and hence reduced cross-protection against S. sonnei. This hypothesis is difficult to prove without widespread serological data from countries with high S. flexneri versus high S. sonnei infection rates. However, it also suggests additional differences regarding the transmission of S. sonnei, as the reduced exposure to P. shigellelloides and S. flexneri through improved water quality does not extend to S. sonnei. This could be explained by the suggestion that S. sonnei is spread directly from person to person or that S. sonnei has an increased ability to acquire antibiotic resistance. Acquisition of antimicrobial resistance has clearly aided the spread and establishment of particular S. sonnei isolates (29). However, epidemiological data indicate that S. flexneri and S. sonnei isolates have similar resistance profiles (30,31), suggesting antibiotic resistance alone does not explain the altered transmission.
S. sonnei possesses multiple antibacterial mechanisms. The majority of clinical isolates produce colicins (29,32,33), which are effective against a narrow phylogenetic range of bacteria. While S. flexneri cells are also reported to produce bacteriocins (34), there are few studies indicating the prevalence or identity of bacteriocins in S. flexneri clinical isolates. S. sonnei has also recently been shown to have a functional type 6 secretion system (T6SS), which provides a niche-specific competitive advantage for S. sonnei over S. flexneri (35). Therefore, the success of S. sonnei may be explained by a combination of these factors potentially altering colonization dynamics and facilitating person-to-person spread.
We have demonstrated that amoebae are not a protective host for S. sonnei and that alternative explanations for the rising rates of S. sonnei infection in transitional countries require further investigation. Now that it is well established that S. sonnei possesses unique pathogenic traits (35)(36)(37), considerable work is required to understand the differences in virulence and transmission of S. sonnei in comparison to those of S. flexneri.

MATERIALS AND METHODS
Bacterial strains and growth. Isolates of Shigella (Table 1) were plated on trypticase soya agar (TSA) ϩ 0.01% Congo red to identify those with a large virulence plasmid (LVP) (38). Colonies were inoculated in trypticase soya broth (TSB) and incubated overnight at 37°C and 200 rpm. The overnight culture was diluted 1:100 in TSB and incubated until an optical density at 600 nm (OD 600 ) of 0.5 was reached. Bacteria were washed in phosphate-buffered saline (PBS), resuspended in the appropriate medium, and added to cells.
Legionella cells were plated on buffered charcoal-yeast extract (CYE) agar plates for 3 days at 37°C. Bacterial colonies were diluted to OD 600 ϭ 0.1 in ACES [N-(2-acetamido)-2-aminoethanesulfonic acid] yeast extract (AYE) broth and incubated for 21 h at 37°C and 200 rpm.  HeLa cell infection. HeLa cells were seeded in 96-well plates at 1 ϫ 10 5 cells/ml 24 h prior to infection. A. castellanii was seeded at 10 7 cells/flask in T75 flasks in PAS medium, infected with Shigella at an MOI of 100, and incubated at 22°C. After 24 h, amoebae were detached, centrifuged at 500 ϫ g for 5 min, and resuspended in 150 g/ml gentamicin in PAS for 1 h. Amoebae were then washed with PAS and lysed with 0.25% sodium deoxycholate. Released bacteria were centrifuged, washed, and resuspended in DMEM. Broth-cultured Shigella cells were prepared as described above and resuspended in DMEM. Prior to infection, medium was replaced with serum-free DMEM, and HeLa cells then infected at an MOI of 100 with Shigella cells released from amoebae or grown in broth at 37°C or 22°C. Cells were centrifuged at 600 ϫ g for 10 min and incubated for 30 min at 37°C and 5% CO 2 . Medium was replaced with 150 g/ml gentamicin in supplemented DMEM and incubated for a further 1 h or 3 h. At these time points, cells were washed with PBS and lysed with 0.5% Triton X-100. Serial dilutions were performed and plated to calculate intracellular CFU.