Resuscitation of Salmonella enterica Serovar Typhimurium and Enterohemorrhagic Escherichia coli from the Viable but Nonculturable State by Heat-Stable Enterobacterial Autoinducer

Abstract

Salmonella enterica serovar Typhimurium and enterohemorrhagic Escherichia coli were stressed by prolonged incubation in water microcosms until it was no longer possible to observe colony formation when samples were plated on nonselective medium. Overnight incubation of samples in nutrient-rich broth medium supplemented with growth factors, however, allowed resuscitation of stressed and viable but nonculturable cells so that subsequent plating yielded observable colonies for significantly extended periods of time. The growth factors were (i) the trihydroxamate siderophore ferrioxamine E (for Salmonella only), (ii) the commercially available antioxidant Oxyrase, and (iii) the heat-stable autoinducer of growth secreted by enterobacterial species in response to norepinephrine. Analysis of water microcosms with the Bioscreen C apparatus confirmed that these supplements enhanced recovery of cells in stressed populations; enterobacterial autoinducer was the most effective, promoting resuscitation in populations that were so heavily stressed that ferrioxamine E or Oxyrase had no effect. Similar results were observed in Bioscreen analysis of bacterial populations stressed by heating. Patterns of resuscitation of S. enterica serovar Typhimurium rpoS mutants from water microcosms and heat stress were qualitatively similar, suggesting that the general stress response controlled by the σ^s subunit of RNA polymerase plays no role in autoinducer-dependent resuscitation. Enterobacterial autoinducer also resuscitated stressed populations of Citrobacter freundii and Enterobacter agglomerans.

Salmonella species and enterohemorrhagic Escherichia coli (EHEC), in particular serovar O157:H7, are important food-borne pathogens that represent an increasingly significant public health issue in industrialized countries. The problem, at least in part, is that these organisms can persist for long periods in the environment in a heavily stressed state known variously, and often contentiously, as viable but nonculturable (VNC) (15, 17, 21, 26-28) or not immediately culturable (8). These heavily stressed microorganisms show only very weak metabolic activity, often at the very limits of detection, and they lose the ability to form colonies on nonselective plating media or to grow in nonselective broth media. Nevertheless, in the case of nonculturable populations of pathogenic bacteria in the environment (soil or water, etc.) or associated with bacteriological spoilage of human foods and animal feeds, they may still be capable of causing disease if ingested by a susceptible animal host (8, 26). The important questions are how such cells can be resuscitated to aid in vitro identification of potential pathogens and what role they play in the various habitats in which they exist and in the pathogenesis of infectious disease.

We previously demonstrated that the microbial trihydroxamate siderophore ferrioxamine E was able to resuscitate stressed Salmonella enterica serovar Typhimurium strains in soil for periods of up to 2 years and in water microcosms for more than 4 months (19). This process was shown to be dependent on siderophore uptake, since neither Salmonella mutants deficient in ferrioxamine uptake nor strains of E. coli that naturally cannot use ferrioxamines as sources of iron were resuscitated from stressed states by ferrioxamine E (19). We proposed that resuscitation by ferrioxamine E is due to its ability to sequester intracellular iron in a form that reduces the risk of generating damaging oxygen radicals that would otherwise kill bacterial cells emerging from stressed states (19). Consistent with this proposal, we demonstrate here that the commercially available antioxidant Oxyrase, an oxygen radical-destroying enzyme prepared from E. coli, is also able to resuscitate stressed populations of S. enterica serovar Typhimurium and EHEC.

The aim of the work described in this paper was to assess the ability of a heat-stable autoinducer (AI) of growth, which is secreted by a number of enterobacterial species in the presence of the mammalian neuroendocrine hormone norepinephrine (NE) (3, 12), to resuscitate stressed bacteria. The ability of NE and other catecholamines to promote bacterial growth in vitro in a medium designed to reflect the harsh environments infectious bacteria may encounter in vivo is well established (3, 11-13). In the case of many gram-negative species, such growth is accompanied by production of AI, which can itself stimulate bacterial growth under stress conditions in the absence of NE and also the production of more AI (3, 12). Characterization of the activity produced by several enterobacterial species suggested a family of structurally closely related but functionally identical molecules (3) that we refer to in the generic sense as enterobacterial AI. Here we demonstrate that enterobacterial AI is able to resuscitate heavily stressed populations of a number of strains of S. enterica serovar Typhimurium, EHEC, and other pathogenic species, and we propose the routine use of AI for more effective screening for enterobacterial pathogens in environmental and food samples.

MATERIALS AND METHODS

Bacterial strains.

Of the species reported in our previous study of NE-induced AI (3), the strain that reproducibly produced the highest AI levels was one that was originally reported by the Leicester Public Health Laboratory Service to be Hafnia alvei on the basis of API tests. However, in subsequent independent tests with the Bactid identification system (Centers for Disease Control, Atlanta, Ga.), which in our hands is more reliable, this strain was identified as Yersinia ruckeri. In addition, it was positive in a pyrase test, which is characteristic of Yersinia spp., but negative in a highly sensitive test for Hafnia. For technical convenience we chose this strain as the routine source of heat-stable enterobacterial AI, and for consistency all of the experiments reported in this paper were performed with batches of AI prepared from this strain. It should be noted, however, that AI prepared from cultures of various strains of E. coli, S. enterica serovar Typhimurium, and H. alvei gave qualitatively similar data in all experiments in which they were used. Details of the Salmonella and E. coli strains tested for resuscitation by AI are given in Table 1. The sources of strains of other species tested are indicated elsewhere in the text.

TABLE 1.

Salmonella and E. coli strains tested for resuscitation

Species and strain no.	Characteristics	Source and/or referencesa
S. enterica serovar     Typhimurium
ATCC 14028	Wild type	ATCC
ATCC 14028-1s	ATCC 14028, streptomycin resistant	20
UMR1	ATCC 14028-1s, nalidixic acid resistant	20
SF1005	rpoS::pRR10 (ΔtrfA Ampr)	SGSC
MAE40	UMR1 rpoS::pRR10 (ΔtrfA Ampr)	U. Römling (20)
SL1344	hisG46	7, 19
TA2700	entA fhuC	J. B. Neilands (16, 19)
E. coli
NCTC10418	Wild type	NCTC
EDL933	O157:H7; Shiga toxins 1 and 2	A. D. O'Brien
97-04281	O157:H7; Shiga toxins 1 and 2	RKI
97-04951	08:H−; Shiga toxins 1 and 2	RKI

^aATCC, American Type Culture Collection, Manassas, Va; SGSC, Salmonella Genetic Stock Centre, Calgary, Canada; NCTC, National Collection of Type Cultures, Colindale, United Kingdom; RKI, Collection of the Robert Koch Institute, Wernigerode, Germany.

AI preparation.

SAPI medium containing 30% (vol/vol) adult bovine serum and 50 μM l-(−)-NE (Sigma Chemical Co.) as previously described (3, 12) was used for the preparation of enterobacterial heat-stable AI. SAPI is a variation of the standard American Petroleum Institute medium and contains 6.25 mM NH₄NO₃, 1.84 mM KH₂PO₄, 3.35 mM KCl, 1.01 mm MgSO₄, and 2.77 mM glucose (pH 7.5). Y. ruckeri was inoculated at approximately 100 to 1,000 CFU/ml in serum-SAPI medium and incubated overnight at 37°C in a humidified 5% CO₂ atmosphere. Bacteria were pelleted by centrifugation (6,000 × g for 15 min), and the culture supernatants were filter sterilized and stored at −20°C until required. Sterility was checked by plating samples on sheep blood agar and incubating at 37°C for 48 h under aerobic and anaerobic conditions. Sterile preparations were serially diluted in fresh sterile SAPI medium, and samples of each dilution were added at 5% (vol/vol) to serum-SAPI medium inoculated with 100 to 1,000 CFU of an indicator E. coli strain per ml (4). The dilution that promoted bacterial growth after overnight incubation at 37°C to an optical density at 620 nm of 0.4, which represents 10⁸ CFU/ml, was used as a supplement to liquid growth media as described below. NE, SAPI, and bovine serum had no significant effect when used as supplements, either individually or combined.

Purified enterobacterial AI is available as Bacxell (BioNutrix LLC, Minneapolis, Minn.) (International Patent Application no. WO 98/53047 [May 1998] and United Kingdom Patent Application no. 0120120 [August 2001]).

Water microcosms.

Water microcosms were set up as previously described (19). Briefly, 1.5-liter batches of sterile double-distilled water were inoculated with saline-washed bacterial growth from fresh tryptic soy agar (TSA) (BD Heidelberg, Heidelberg, Germany) cultures at 10⁵ to 10⁶ CFU/ml and stored at room temperature under normal laboratory light conditions. Samples were taken at intervals for viable count assays, growth in liquid culture, or Bioscreen C analysis as appropriate (see below). Water microcosm experiments were typically of several months' duration in the case of the gram-negative species tested, but gram-positive organisms died within a few days under these conditions.

Colony counts.

Culturable counts of samples from water microcosms were measured by plating serially diluted samples in triplicate on TSA containing 0.1% (wt/vol) sodium pyruvate to minimize further selective stress by the medium. In the case of S. enterica serovar Typhimurium microcosms, recovered colonies were checked by subculturing onto Galle-Chrysoidin-glycerol (GCG) agar (SIFIN, Berlin, Germany) and by agglutination with omnivalent Salmonella serum. For E. coli microcosms, colonies on GCG agar and BCM E. coli O157:H7(+) selective agar medium (Biosynth AG, Staad, Switzerland) were confirmed serologically. Other species of the Enterobacteriaceae, Pseudomonas aeruginosa, Aeromonas hydrophila, and Burkholderia cepacia were checked by subculturing onto GCG agar. Gram-positive bacteria were characterized by typical growth on sheep blood agar.

Growth in liquid media.

When colonies were no longer observed by direct viable counting of 0.1- and 0.5-ml samples, 60-ml samples of water microcosms were inoculated into 90 ml of 1.67-fold-concentrated buffered peptone water (BPW) (Oxoid) in the case of Salmonella strains or phosphate-buffered tryptic soy broth (p-TSB) (containing 1.5 mg of K₂HPO₄/ml) for the other bacterial species tested. Experiments to test the effects of supplementation were performed only if no growth had occurred in this mixture after overnight incubation. Ferrioxamine E (Novartis AG, Basel, Switzerland) was added at a final concentration of 50 ng/ml, Oxyrase (Oxyrase Inc., Mansfield, Ohio) was added at 0.2 U/ml as defined in the manufacturer's technical bulletin, and enterobacterial heat-stable AI (prepared as described above) was added as a 1% (vol/vol) supplement. Suspensions were incubated with rotary shaking at 37°C, except in the case of P. aeruginosa cultures, which were incubated at 30°C. Bacteria were inoculated onto TSA, and growth was checked biochemically, serologically, or morphologically as described above.

Heat stress.

Freshly cultivated bacteria from TSA plates were suspended in phosphate-buffered saline to a density of approximately 10⁸ to 10⁹ CFU/ml as estimated by turbidity measurements. Viable cell counts at the beginning of the stress period were enumerated by the most-probable-number technique (7a). Bacterial suspensions were incubated in a shaking water bath (GFL1086; Gesellschaft für Labortechnik mbH, Burgwedel, Germany) at the temperatures and for the times indicated in Table 2.

TABLE 2.

Reactivation and resuscitation of Salmonella and E. coli strains from water microcosms and following heat stress

Strain	Stress conditions	No. of wells with growth (lag phase, h)
		Reactivation (unsupplemented)	Resuscitation by supplementation with:
			Ferrioxamine E	Oxyrase	Enterobacterial AI
ATCC 14028a	Water microcosm	6 (14-17)	8 (11-21)	3 (11-17)	13 (<5-18)
ATCC 14028-1s	Water microcosm	0	0	0	10 (5-19)
MAE40	Water microcosm	9 (10-17)	13 (8.5-14.5)	13 (10-13.5)	25 (7-13)
SF1005	Water microcosm	1 (11)	3 (11-17.5)	1 (13)	14 (5-14)
NCTC10418	53°C, 50 min	7 (9-14)	NDb	2 (7.5-10.5)	11 (7-16)
97-04281	53°C, 35 min	0	ND	0	5 (14-19)
UMR1	48°C, 3 h + 46°C, 19 h	0	ND	ND	11 (8.5-14.5)
MAE40	48°C, 3 h + 46°C, 19 h	2 (12)	ND	ND	8 (9-18)

^aSame data as in Fig. 2.

^bND, not done.

Bioscreen analysis.

Three-hundred-microliter volumes of BPW or p-TSB as appropriate, with or without supplementation as described above, were applied to each of 25 wells of a 100-well microtiter plate (i.e., to provide 25 replicates of each experimental condition). Ten-microliter aliquots of stressed cell suspensions (from water microcosms or after heat stress) were inoculated into each well, and bacterial growth was monitored in a Bioscreen C apparatus (Labsystems, Helsinki, Finland). Optical densities at 620 nm values were measured at 20-min intervals (sample plates were shaken for 10 s before each measurement) over a period of 24 to 48 h at 37°C, or at 30°C in the case of P. aeruginosa. The measurements were monitored by the computer program DOS of the Bioscreen C apparatus and transferred to a Windows 98 platform for graphical presentation. In the graphs shown, each line represents growth in a single well; however, for reasons having to do with the operating system of the computer that records the data, one line in each panel is markedly thicker than all of the others. Bacteria from two wells of each incubation regimen in which resuscitation had occurred were streaked onto nonselective agar, and their identities confirmed biochemically, serologically, or morphologically as described above.

Reproducibility.

Since the effects of stress are essentially random at the level of the individual cell, each experiment generated a unique population that responded to resuscitation quantitatively differently from other stressed populations even if apparently identical stress conditions were used. Thus, all data shown in this paper are representative of at least three replicate experiments in which qualitatively similar results were obtained.

RESULTS

Reactivation and resuscitation of stressed S. enterica serovar Typhimurium from water microcosms.

Water microcosms were inoculated at 10⁵ to 10⁶ CFU/ml and sampled at intervals to monitor the decline in numbers of viable bacteria recoverable by direct plating on TSA. Results were variable from experiment to experiment for any individual strain and between strains, presumably reflecting natural variation in bacterial populations undergoing stress and damage, as noted in Material and Methods. In the case of wild-type S. enterica serovar Typhimurium strains, viable cells were no longer detectable by plating after periods ranging from about 1 month to almost 4 months (Fig. 1a). Typically, S. enterica serovar Typhimurium produced small irregular colonies on the plating medium towards the end of this period, but subculture of these colonies on TSA gave normal-sized smooth colonies after one or two passages. Stressed cells that were no longer detectable by direct plating could be reactivated by overnight incubation of microcosm samples at 37°C in BPW, and were therefore detectable for an additional period of 11 to 18 days. We distinguish experimentally between microcosms containing stressed cells that can be reactivated by a period of incubation in nutrient-rich broth medium, such as BPW or p-TSB, and those in which stressed cells can be resuscitated only by supplementation with particular growth factors. As we reported previously (19), addition of ferrioxamine E to BPW resulted in prolonged periods of resuscitation of S. enterica serovar Typhimurium from highly stressed populations beyond the time when incubation in BPW alone no longer resulted in reactivation. Resuscitation by ferrioxamine E is dependent on siderophore uptake, since recovery of the fhuC mutant TA2700 (16, 19), which is unable to transport ferrioxamines, was not enhanced by the presence of ferrioxamine E. Furthermore, for all strains of Salmonella that we have tested, including TA2700, supplementation of BPW with enterobacterial AI resulted in resuscitation of cells in stressed populations for longer periods of time than did ferrioxamine E supplementation (Fig. 1a).

FIG. 1.

Histograms showing end points (in days) for recovery of Salmonella and EHEC strains without supplementation or in the presence of ferrioxamine E, Oxyrase, or AI. (a) S. enterica serovar Typhimurium strains ATCC 14028, SL1344, TA2700, and UMR1 were assayed by direct plating (white bars) or following incubation in BPW (light grey bars), in BPW supplemented with ferrioxamine E (dark grey bars), or in BPW supplemented with AI (black bars). (b) E. coli O157: H7 strains EDL933 and 97-04281 were assayed by direct plating (white bars) or following incubation in p-TSB (light grey bars), in p-TSB supplemented with Oxyrase (dark grey bars), or in p-TSB supplemented with AI (black bars).

Reactivation and resuscitation of stressed EHEC from water microcosms.

The viability of EHEC strains of serotype O157:H7 in water microcosms also declined with time, so that at about 3 months no growth was detectable by direct plating (Fig. 1b). Here again, small irregular colonies were observed on the plating medium just before this time. Incubation of microcosm samples in nutrient-rich p-TSB enabled reactivation of stressed cells for additional periods of up to a week. E. coli cannot use ferrioxamines to acquire iron, and so stressed EHEC cells cannot be resuscitated by supplementation with ferrioxamine E as in the case of Salmonella. However, supplementation of p-TSB with the biological antioxidant Oxyrase resuscitated highly stressed EHEC beyond the time that incubation in unsupplemented p-TSB no longer gave visible growth (Fig. 1b). Moreover, supplementation with enterobacterial AI enabled resuscitation of stressed E. coli O157:H7 cells for significantly longer periods, in one experiment for as much as an additional month. In an experiment with EHEC O8:H⁻ strain 97-04951, growth was detectable following AI supplementation of p-TSB for up to 455 days, compared with only 88 days in the absence of supplementation (data not shown). In all of these experiments, EHEC bacteria recovered after resuscitation in AI-supplemented p-TSB still produced Shiga toxins (data not shown).

Bioscreen analysis of resuscitation of stressed bacteria from water microcosms.

Because the development of stress in water microcosms is an essentially random process, prolonged incubation presumably leads to progressive changes in the proportions of (i) stressed cells that can be reactivated in liquid culture, (ii) VNC cells that require a growth supplement (ferrioxamine E, Oxyrase, or enterobacterial AI) for resuscitation, and (iii) dead cells. In fact, it is not possible to predict from experiment to experiment whether a particular population sample contains any bacteria capable of reactivation or resuscitation. To analyze stressed populations in greater detail, we used the Bioscreen C apparatus to monitor the effects of various supplements on 25 replicate subcultures of a population simultaneously in real time. In an attempt to standardize the experimental approach, we tested water microcosms at the time that direct plating just failed to detect colony-forming bacteria, but we used quantitative data from the last day on which colonies were detectable to adjust the notional inoculum size (usually by concentrating samples of microcosm) to 0.3 viable bacterium per well of the Bioscreen apparatus. Figure 2 illustrates the effects of supplementation on cells of S. enterica serovar Typhimurium ATCC 14028 stressed to the extent that growth was observed in 6 of the 25 wells containing unsupplemented BPW. In this particular experiment, only a few cells in the microcosm were resuscitated by ferrioxamine E (8 of 25 wells) or Oxyrase (3 of 25 wells), although it should be noted that the lag phases of some of these subcultures were significantly shorter than those in unsupplemented subcultures (11 h, compared with 14 h or more in the absence of supplementation). However, 13 of the 25 replicate samples supplemented with enterobacterial AI showed resuscitation, some, remarkably, with lag phases of as short as 4 to 7 h (Fig. 2; Table 2). Note that variation between lines representing a particular treatment presumably indicates natural, random variation within the population of stressed bacteria, with longer lag phases and lower exponential growth rates presumably being due to greater levels of stress or damage in the cells of each inoculum. Moreover, variation between experiments presumably also represents natural, random variation in the development of stress in different microcosms even when these were set up identically. Increasing stress presumably leads to conditions in which dead cells predominate in the population. In an experiment with S. enterica serovar Typhimurium strain ATCC 14028-1s, the level of stress was such that neither the wells with unsupplemented BPW nor the wells supplemented with ferrioxamine E or Oxyrase showed any evidence of reactivation or resuscitation (Table 2). Enterobacterial AI, however, stimulated growth in 10 of the 25 wells with lag phases ranging from 5 to 19 h. The use of combinations of ferrioxamine E or Oxyrase with AI did not significantly increase the level of resuscitation observed with AI alone (data not shown).

FIG. 2.

Resuscitation of S. enterica serovar Typhimurium ATCC 14028 from water microcosms by supplementation of BPW with AI, Oxyrase, or ferrioxamine E. Results for the same unsupplemented BPW wells (grey lines, growth in 6 of 25 wells) are shown in all three panels. Growth in supplemented BPW is shown as black lines. (a) Supplementation with AI (growth in 13 of 25 wells); (b) supplementation with Oxyrase (3 of 25 wells); (c) supplementation with ferrioxamine E (8 of 25 wells). OD, optical density.

To determine whether RpoS plays a role in resuscitation, we tested microcosms of S. enterica serovar Typhimurium strains carrying a mutation in the rpoS gene. These strains lost the ability to form colonies more rapidly that wild-type strains (in 2 weeks as opposed to several months), but resuscitation patterns with ferrioxamine E, Oxyrase, or enterobacterial AI were remarkably similar to those observed for wild-type strains. Table 2 shows the effects of supplementation on stressed cells of rpoS mutants MAE40 and SF1005 in water microcosms; AI resulted in growth in more wells and with markedly shorter lag phases than ferrioxamine E or Oxyrase.

Bioscreen analysis of resuscitation of heat-stressed bacteria.

While water microcosms are good models for the stresses that pathogenic bacteria may encounter in the external environment, survival from heat stress is an important factor in the etiology of food-borne illness due to faulty food processing and poor kitchen hygiene. Table 2 illustrates an experiment in which incubation of E. coli strain NCTC10418 at 53°C for 50 min resulted in a population from which 7 of 25 subculture wells contained cells that could be reactivated by incubation in p-TSB. In this case, supplementation with Oxyrase or enterobacterial AI promoted resuscitation in 2 and 11 wells, respectively. In an experiment with another EHEC strain, 97-04281, heated at 53°C for 35 min (Table 2), the level of stress was so great that no growth was observed in any of the unsupplemented p-TSB wells or in any of the wells supplemented with Oxyrase. Supplementation with AI, however, allowed resuscitation in five wells from this highly heat-stressed population. Similar data were obtained with a heat-stressed population of S. enterica serovar Typhimurium strain UMR1 in which more prolonged periods of heating were used; in this case enterobacterial AI resulted in resuscitation of stressed cells in 11 of the 25 wells (Table 2).

AI also resuscitated heat-stressed cells of the Salmonella rpoS mutant MAE40 with a pattern similar to those observed for resuscitation of water microcosms (Table 2). Lag phases of just 9 h were observed in some wells, compared with 12 h in unsupplemented cultures. Enterobacterial AI also effectively resuscitated heat-stressed populations of wild-type Citrobacter freundii strains 98-01367 (from the collection of the Robert Koch Institute) and ATCC 8090 and of Enterobacter agglomerans ATCC 13020, but it had no measurable effects (in terms either of numbers of wells or of lag phases of resuscitated cultures) on Proteus vulgaris 718/96, Proteus mirabilis NM12, Providencia rettgeri NM19, Providencia stuartii 20137, and Morganella morganii SBK3 (all from the Robert Koch Institute collection); on A. hydrophila ATCC 7966; or on P. aeruginosa DSM27853 and B. cepacia DSM7288 (both from the Deutsche Sammlung für Mikroorganismen und Zellkulturen, Braunschweig, Germany) (data not shown). Moreover, AI was not able to resuscitate the gram-positive organisms Staphylococcus aureus NCTC6571 and Staphylococcus epidermidis CCM2124 (from the Culture Collection of Microorganisms, Brno, Czech Republic), Enterococcus faecalis ATCC 29212, and Listeria monocytogenes NCTC7073 (data not shown).

DISCUSSION

It is axiomatic that only a tiny fraction of the bacterial species inhabiting the biosphere have so far been discovered. Moreover, among the minority of bacteria that have been discovered, more than 90% are as yet nonculturable and can be detected only by molecular techniques based on probes for 16S and 23S rRNAs or on determination of mRNA, either by reverse transcriptase PCR (23) or by fluorescence techniques such as in situ hybridization, microradiography, epifluorescence microscopy, and flow cytometry (2, 14). Even among the minority of species that are normally considered to be readily culturable, environmental stresses of various kinds drive populations of microorganisms towards a state in which increasing proportions cease to be culturable by the use of known culture media and conditions. In the case of pathogenic species of bacteria, this so-called VNC state (8, 15, 21, 26-28) is a potentially dangerous public health problem, particularly because stressed cells are apparently more virulent that well-fed bacteria (5, 26). Bacteria may be damaged by a wide variety of stress conditions, including nutrient starvation, oxygen radicals, heat, freezing temperatures, changes in pH, near-UV radiation, and osmotic pressure. Adaptive networks have evolved in bacteria to overcome the challenges of rapidly changing environments and to permit survival under conditions of stress. The important practical question is how to monitor highly stressed cells in the environment and the food chain.

In this paper we report the ability of ferrioxamine E, Oxyrase, and enterobacterial AI to resuscitate bacterial cells in populations stressed by prolonged incubation in water or by heat treatment. Two independent approaches were used to assess resuscitation. In one case, microcosm samples taken at intervals were cultivated in nutrient-rich medium with or without the supplements. This is an “all-or-nothing” approach that simply detects the presence of any recoverable cells in the population (theoretically a single such cell would be sufficient), but it is useful as an indicator of the presence of cells that may, given the right set of environmental parameters, go on to cause disease in susceptible individuals, human or animal, that ingest them. The second approach was to use the Bioscreen to analyze the effects of supplementation simultaneously in 25 subpopulations of a stressed population. This method enables comparisons to be made between populations at various arbitrarily defined levels of stress, from relatively mild (in which reactivation occurred in most or all of the subpopulations even in the absence of supplements) to severe (in which some subpopulations contained VNC cells but most contained only dead cells).

The important practical point to note is that despite differences in readout, the data obtained from the two approaches were entirely consistent. Thus, we confirmed our previous observation that ferrioxamine E can resuscitate wild-type S. enterica serovar Typhimurium from water microcosms and extended it by demonstrating a similar effect on heat-stressed populations. Since Salmonella strains of subspecies I, II, and IIIb, which include more than 98% of all clinical isolates, possess the high-affinity ferrioxamine uptake and utilization system (9), supplementation with ferrioxamine E can be considered to be semiselective for detection of Salmonella spp. in stressed populations (19). Ferrioxamine B functions similarly but slightly less effectively than ferrioxamine E (data not shown). Oxyrase was similarly able to resuscitate stressed S. enterica serovar Typhimurium strains and was also effective on EHEC populations both from water microcosms and after heat stress (24). Oxyrase functions by destroying oxygen radicals in growing cells, and so these observations are consistent with our proposal that ferrioxamine E acts by preventing damaging oxygen radicals from killing recovering cells. However, while Oxyrase shows promise as an effective growth supplement for rapid recovery of stressed cells from various contaminated materials, unlike ferrioxamine E it would be expected to work nonselectively on both pathogenic and nonpathogenic bacteria in a population and so would not be useful in situations where the detection of particular pathogens is required.

More effective than either ferrioxamine E or Oxyrase at resuscitating bacteria from heavily stressed populations was the enterobacterial AI produced in serum-SAPI medium in the presence of NE (12). This medium was designed to mimic the kind of stressful environment an infectious microorganism might encounter in vivo, and indeed most bacterial species fail to thrive in it unless NE or other catecholamines are present to facilitate iron removal and uptake from serum transferrin (3, 4). Enterobacterial AI also promotes bacterial growth in serum-SAPI; the mechanism is as yet unknown but is undoubtedly different from that of catecholamines. Thus, growth of clinical isolates of 17 gram-negative and 6 gram-positive bacterial species in serum-SAPI medium was stimulated by supplementation with AI preparations from several enterobacterial species (3). Note that in these previous experiments, serum-SAPI medium was inoculated with bacteria from fresh, nonstressed starting cultures. The present study, by contrast, suggests that the ability of enterobacterial AI to resuscitate heavily stressed populations is restricted to S. enterica serovar Typhimurium, EHEC, C. freundii, and E. agglomerans. AI had no measurable effect on individual representative strains of eight other gram-negative species (P. vulgaris, P. mirabilis, P. rettgeri, P. stuartii, M. morganii, A. hydrophila, P. aeruginosa, and B. cepacia) and four gram-positive species (S. aureus, S. epidermidis, E. faecalis, and L. monocytogenes) that we tested.

The mechanism of action of AI is not yet known. One possibility is that it functions as a quorum sensor for stressed bacteria; the structure of AI is still under investigation, but it is known not to be a homoserine lactone (25), nor does it have properties similar to those of the recently characterized autoinducer-2 molecule (22). Another possibility is that AI interacts with components of bacterial global stress responses. A number of bacterial virulence factors are known to be activated during the transition from exponential- to stationary-phase growth in response to induction of rpoS (18, 26). Moreover, bacteria in stationary phase are more thermotolerant, more resistant to oxidative stress and acidic conditions, and better equipped to survive both osmotic stress and starvation due to differential production of products such as catalase, glycogen, and heat shock and cold shock proteins (1). The σ^s subunit of RNA polymerase (RpoS) is the master regulator of more than 35 genes involved in the general stress response in S. enterica serovar Typhimurium, E. coli, and other enteric bacteria (1, 5, 6, 10). Levels of RpoS are low in rapidly growing cells that have not been exposed to any particular stress but are induced in response to a variety of diverse environmental stresses (6). Inactivation of RpoS renders cells much more sensitive to a variety of stresses, as seen by the more rapid decline in cell counts in water microcosms compared with wild-type strains of S. enterica serovar Typhimurium. The qualitative pattern of resuscitation of S. enterica serovar Typhimurium rpoS mutants by ferrioxamines, Oxyrase, or AI, however, was essentially the same as with wild-type strains, indicating that the mechanism(s) of resuscitation is independent of RpoS. Similarly, RNA polymerase subunit RpoB, which is not involved in the stress response, appears not to be involved in recovery from the VNC state, since an S. enterica serovar Typhimurium rpoB mutant showed the same pattern of resuscitation as wild-type strains (data not shown). In order to learn more about the mechanism of recovery from stress, we are currently performing a detailed analysis of reproductive, deenergized, depolarized, and dead cells in stressed and resuscitated populations by using various fluorescent dyes measured by flow cytometry (14).

Meanwhile, from the practical point of view, we define the process of resuscitation as the ability of VNC cells to grow on nonselective agar only following incubation in nutrient-rich broth medium containing supplements such as ferrioxamine E (if appropriate), Oxyrase, or enterobacterial AI. Our data demonstrate that AI is able to promote significant resuscitation of populations that were so heavily stressed that the other supplements were ineffective. Moreover, EHEC cells resuscitated by AI still retained the capacity to produce Shiga toxins (in the case of one serotype O8:H⁻ strain, after a period of 15 months in a water microcosm). We propose that enterobacterial AI has important benefits, compared with unsupplemented preenrichment media, for improved semiselective isolation of pathogenic Salmonella spp. and E. coli O157:H7 (and other EHEC) from contaminated environments and foods.