Stress Physiology of Lactic Acid Bacteria

Common Stresses Encountered by LAB in Their Ecological NichesLAB are confronted with both abiotic and biotic stresses. Abiotic stresses are mainly those arising during food production and the manipulation of starter or probiotic cultures, while biotic stresses are encountered in the host or in complex ecosystems. In several instances, the type of elementary stress condition is the same regardless of the origin, be it technological or biological. The stresses that have been studied in LAB to date, along with some important characteristics, are presented below.

In the case of LAB, acid stress is a self-imposed stress. Lactic acid is the major end product of sugar fermentation and is virtually used as an antimicrobial agent against competing microorganisms. LAB are relatively acid tolerant, but the accumulation of lactic acid ultimately influences their physiology (17). It is not uncommon for LAB to cease to grow due to autoacidification rather than depletion of nutrients, while prolonged exposure to acidic conditions usually results in cell death. Acid stress is also relevant for probiotic LAB. In the GIT, the stomach retains a low pH between 2.0 and 4.0 via the production of HCl. The gastric juice also contains digestive enzymes (e.g., pepsin) that may additionally damage cells. Even though probiotics can be protected from the gastric environment by appropriate encapsulation, stomach transit with minimal loss of viability is still considered an important probiotic property (12). Likewise, pathogenic LAB are confronted with low pHs, e.g., in macrophages after phagocytosis. In brief, low pH damages both the cell wall and the cell membrane, thus influencing ΔpH and the membrane potential. Acidification of the cytosol is genotoxic and results in the denaturation of proteins. Overall metabolism is affected, which leads to energy depletion and cell death.

Research of bacterial responses to high temperatures led to the discovery of a set of heat-inducible proteins known as heat shock proteins (HSPs), which are employed to counteract the pleiotropic effects of heat stress (18). Major HSPs partake in the repair and turnover of damaged proteins. Heat stress is commonly encountered by many LAB. During food fermentation, high temperatures (>60°C) are used for pasteurization of raw materials. The indigenous LAB population has to cope with pasteurization, particularly in spontaneous fermentations. In fermentations where LAB are added as starters after the pasteurization step, they may be exposed to reheating steps that are necessary for the production of specific foods (e.g., in several types of cheese). In contrast to food-related LAB, commensals and pathogenic LAB encounter less drastic temperature fluctuations due to thermoregulation of the host. Still, fever may be considered a defense mechanism relying on heat stress to combat pathogens.

Increased osmolarity is an important hurdle used in the production of numerous fermented foods. As many food spoilage and pathogenic microorganisms are rather sensitive to high osmotic pressures compared to LAB, NaCl is usually added to aid the indigenous or starter LAB in initiating and taking over the fermentation process. Osmotic stress decreases the positive turgor of bacterial cells as a result of dehydration. Under such conditions, cells either produce or import small molecules, called osmolytes (e.g., glycine betaine, choline, or proline), to balance the difference between intracellular and extracellular osmolarities to allow rehydration through membrane-associated channels (19). Even though variations in osmolarity may exist among the different niches within a host, osmotic stress is generally not acknowledged as a major stress for commensal and pathogenic LAB.

Low temperatures are used for storing raw materials and foods to prevent spoilage. Also, LAB starter or probiotic cultures are most frequently stored in a frozen or freeze-dried form, with appropriate cryoprotectants included to increase viability (20). Cold stress leads to the induction of a set of proteins called cold shock proteins (CSPs) (21). All CSPs belong to one family of closely related low-molecular-weight proteins. They can bind to single-stranded nucleic acids and resolve secondary structures formed at low temperatures. CSPs are thus considered to support transcription and translation under cold stress. Cold temperatures above freezing may lead to growth arrest of LAB, but such conditions do not abruptly provoke cell death. In fact, many LAB can be stored at low temperatures (>0°C) for several days. In contrast, freezing of LAB cultures influences survival in a strain-dependent manner. As in the case of osmotic stress, cold stress is not commonly encountered by commensal and pathogenic LAB.

Although LAB are typically microaerophiles lacking a functional respiratory chain and catalases, several species are aerotolerant. Notwithstanding this, LAB are susceptible to aerobic conditions during food production and in the host. O₂ metabolism by LAB can also lead to the production of reactive oxygen species (ROS), and some strains can produce copious amounts of H₂O₂. They can also be exposed directly to ROS, e.g., during the oxidative burst in cells of the immune system. Oxidative stress influences the redox potential of the cell by affecting many enzymatic reactions. ROS are highly reactive moieties that can damage all major macromolecules of the cell, including proteins, DNA, and lipids. LAB possess a number of mechanisms for the detoxification of ROS. Nevertheless, several LAB have been shown to undergo respiration if heme and/or menaquinones are supplied exogenously (22). The implications of respiration in LAB are still a matter of investigation, but it is clearly an additional defense against oxidative stress.

Starvation, as a characteristic stress for free-living microorganisms, has been studied to some extent in LAB, although LAB reside in highly nutritious environments in which depletion of a nutrient rarely becomes the limiting factor for growth. It is generally acknowledged that starvation may be induced indirectly in LAB, as a side effect of another stress. A typical example is the starvation caused by lactic acid autoacidification through interference of the low pH with the action of transporters in the cytoplasmic membrane and the consequent abolishment of nutrient uptake. Sugar starvation is important from a technological perspective because under these conditions, food-related LAB start to catabolize amino acids as an alternative carbon/energy source, resulting in the production of aroma compounds. Dairy LAB may convert branched-chain amino acids (BCAAs) to volatile branched-chain fatty acids (BCFAs) for ATP synthesis (23).

The cell envelope is the physical barrier separating the cell from its environment and, as such, the first line of defense against environmental perturbations. Changes in chemical composition of both the cell wall and the cell membrane triggered by stress have been shown to aid in cell survival. Maintaining the integrity of the cell wall under stress conditions is a matter of life or death for bacteria. The cell envelope is also a major cellular organelle with several physiological functions. It is thus not surprising that the cell envelope is the direct target of a multitude of antimicrobials (24). Over the past years, it has been demonstrated that LAB, like other bacteria, closely monitor the integrity of the cell envelope and that specialized repair mechanisms are induced in case of damage (25).

Apart from the stress conditions mentioned above that are relevant to the majority of LAB, there are stresses that may be faced exclusively by a limited number of species. Ethanol stress is of particular importance mainly for Oenococcus oeni, which performs malolactic fermentation (MLF) during wine making. There are also stresses that have gained momentum rather recently, such as metal stress (26). During acidification, LAB may cause the solubilization of metals, which may be toxic depending on their concentration. Moreover, there are stresses that have not been investigated in LAB as thoroughly as in other bacteria (e.g., DNA damage) and stresses that have not been considered physiologically relevant to the lifestyle of LAB (e.g., hypo-osmotic or alkaline stress).

In the majority of studies of the stress physiology of LAB, strains are exposed to only a single stressor to better dissect the physiological and molecular mechanisms underpinning the responses. In reality, LAB reside mostly in multistress environments that are fairly nutritious to compensate for their auxotrophies. The fastidious nature of LAB is often perceived as a vulnerability that may be associated with a diminished tolerance to environmental stresses. This assumption is far from the truth, and although no LAB has been characterized formally as an extremophile, several species/strains can tolerate or even grow in harsh environments. During transit through the GIT, several Lactobacillus spp. survive the low pH of the stomach (pH 2.0 to 4.0) and the subsequent exposure to bile salts and pancreatic juice in the duodenum (27). Lactobacillus spp. have also been found in the stomach microbiome (28). Lactobacillus suebicus isolated from fruit mash is able to grow at pH 3.0 and in the presence of 14% ethanol (29). O. oeni strains have been reported to proliferate in the presence of 13% ethanol at pH 3.2 and 18°C (30). Tetragenococcus spp. can survive and grow in salt at concentrations of up to 25% (wt/vol) (31), while Leuconostoc gelidum isolated from chilled products can grow at low temperatures, even at 1°C (32). In conclusion, there is ample evidence that LAB may be particularly robust bacteria.

Experimental Context for the Study of LAB Stress PhysiologyA number of experimental approaches for the study of bacterial stress physiology have been adopted over the years. In practice, a stress is defined as a condition that results in reduced bacterial cell growth or survival. Growth and survival assays require an a priori determination of the optimal conditions to be used as controls for comparison. It has been argued that choosing the optimal conditions is arbitrary, since it is impossible to determine them experimentally without making at least some assumptions (e.g., suitability of the growth medium) (33). A more general definition of stress has been suggested to circumvent this discrepancy (33). In current terms, stress can be considered any transition of a bacterial cell from one condition to another that causes alterations to the cell's genome, transcriptome, proteome, and/or metabolome leading to reduced growth or survival potential. This definition of stress applies without the need to determine any “optimal” conditions.

Under stress, cells try to adapt by appropriate molecular responses in an attempt to ameliorate the negative effects and restore growth or survival potential. These adaptive or stress responses are the main focus of LAB stress physiology research. Bacteria continually monitor changes in their environment and respond whenever necessary (see below). Stress responses have been correlated with specific phenotypes so that they can be induced in a controllable and reproducible manner.

The phenotype appearing most frequently in the literature is that of the adapted cell. While adaptation is a general term describing the effort of an organism to resist and persist under stress, it has also been associated with a specific experimental setup in which cells are transiently exposed to mild nonlethal stress conditions that result in increased survival after a subsequent lethal challenge to the same stress. This type of adaptive response is mostly triggered rapidly, in the first minutes or hours of exposure to the mild stress. Two main variations of this basic experiment exist. First, cells may be left under suboptimal conditions much longer than needed to induce adaptation. This adaptive procedure is probably best described as habituation, a term used only scarcely for LAB (34). The molecular mechanisms underlying transient adaptation and habituation to a specific stress may overlap to a degree, but they are not completely identical (34–36). This may explain a number of contradictory results reported for some LAB species, since the two responses have sometimes been considered identical (37, 38). Alternatively, cells are transiently adapted to a stress and the lethal challenge is performed with a different stress. This treatment often results in increased survival, a phenomenon known as cross-protection. The exact combination of the two stresses that leads to cross-protection is species or even subspecies dependent. Cross-protection suggests an induction of molecular mechanisms during exposure to the first stressor that protects cells from the subsequent lethal challenge, and it may be of particular importance for LAB, as they are often exposed sequentially to a variety of stresses.

There are also phenotypes displaying a generalized resistance to many stresses at the same time. One such phenotype is observed when cells enter stationary phase. Cells enter stationary phase due to the exhaustion of nutrients and/or the accumulation of toxic metabolic products in their environment during growth. The environmental conditions of stationary phase become so stressful that the death rate of cells is accelerated (34, 39). Transition from the exponential to the stationary phase is accompanied by the induction of multiple regulons, resulting in the ability to cope with a number of different stresses. This phenotype increases the likelihood of survival until growth conditions are reestablished, and it is especially important for LAB, which, unlike several other Gram-positive bacteria, are unable to form spores. Another multistress resistance phenotype has been shown for cells in biofilms. Such cells are more resilient than planktonic cells. Some bacterial species are naturally prone to forming biofilms, but biofilm formation may also be stress induced. In fact, early research with oral LAB, such as Streptococcus mutans, allowed the resistance phenotype of biofilms to be established (40). Biofilms are relevant for all types of LAB, including food-related ones (41), probiotic LAB (42), commensals (43), and pathogens (40). Interestingly, it has been demonstrated that during stationary-phase, biofilm formation, starvation, and other stressful conditions, some bacteria are viable but not culturable (VBNC). In this distinct physiological state, cells are metabolically active and multistress resistant but have lost the ability to proliferate. VBNC cells may resuscitate under specific conditions or in the presence of specific resuscitating molecules. Entry into the VBNC state is considered an adaptive strategy for long-term survival (44). A number of studies have addressed the VBNC state in LAB. However, since injured cells (see below) and VBNC cells may have similar phenotypes, it is not clear whether LAB truly exhibit the VBNC adaptive response. Finally, persister cells are subpopulations of multiple-antibiotic-resistant cells. In contrast to resistant cells that can grow in the presence of antibiotics, persister cells can resist lethal doses of antibiotics in the absence of growth. The persister cell phenotype has in a few cases been investigated in pathogenic LAB (for example, see references 45 and 46). Only recently was it suggested that the VBNC and persister physiological states may be closely related, with both employing dormancy as the main mode of stress resistance (47). The development and physiology of dormancy in LAB are poorly understood and surely deserve further investigation.

All adaptive responses described above rely basically on epigenetic mechanisms, since regrowth of adapted cells under optimal conditions abolishes any resistance phenotype. Even though it may take a number of generations before the resistance fades away, the fitness of the population ultimately returns to basal levels. Interestingly, stress-induced mutations can also occur; cells exposed to certain stresses can enter a hypermutable state, thereby increasing the diversity of a clonal population, which may lead to new genotypes allowing survival under conditions otherwise detrimental to the wild type (48). Increased mutation rates can be achieved by error-prone DNA polymerases, errors during transcription and/or translation, and activation of mobilizable elements. There is some evidence that adaptive mutations occur in LAB as a response to certain stresses (49, 50).

Adaptive responses, whether epigenetic or not, are the ultimate means to ensure bacterial survival under stress. Compelling evidence suggests that such responses are not inert but are characterized by some degree of plasticity. A number of studies show that diverse stress-resistant phenotypes may protect against the same stressor despite the fact that the molecular mechanisms involved may differ slightly or even in a major way (35, 36). Cells can apparently deploy multiple and overlapping resistance mechanisms to prevent or repair damage and achieve maximal survival.

Moving from the Population to the Single CellA fundamental problem inherently linked to the study of bacterial stress physiology is how to determine survival. The most practical methods rely on culturability, which imposes a binary logic to the assessment of survival (33). Cells are pronounced alive or dead depending on whether they are able to proliferate in a specific medium and under certain conditions. In reality, only culturable cells can be counted accurately on the basis of growth, while the number of dead cells is deduced indirectly from the untreated control population. However, there is at least one additional physiological state after exposure of a population of cells to stress, namely, that of the “injured” cell (51). Injured cells are metabolically active but have suffered damage to a degree that causes a transient or even permanent loss of proliferation capacity (34). They sometimes require extended recovery times or can recover only under special conditions. Injured cells may in a broad sense be considered VBNC, but they are not the result of an adaptive response. The assays to determine the presence or number of injured and VBNC cells are often the same. The simplest procedure is to use a fluorescent probe to reveal metabolic activity in the absence of culturability. A variety of probes are commonly used in in situ viability tests to measure, e.g., metabolic activity, membrane potential, replication, and membrane integrity (51, 52). Such probes have been coupled successfully with fluorescence microscopy or flow cytometry to assess the viability of stressed LAB. Novel quantitative PCR (qPCR)-based methods are also being developed. It is becoming increasingly evident that there is heterogeneity among live and injured cells. The more multiplex a viability assay is, for instance, by employing an increasing number of probes, the more subpopulations can be identified (51). Subpopulations determined by in situ assays can be characterized further by various culturability tests after cell sorting (34). In situ assays are appealing for industrial applications because they might offer very fast, nearly real-time monitoring of the physiological state of cells. Another important advantage of in situ assays is that they can be applied at the single-cell level. In the majority of studies, stress responses of LAB have been assessed only at the population level. Such strategies demonstrate the involvement of molecular mechanisms in an averaging manner, while the presence of any subpopulation(s) remains undetected. The coexistence of live, injured, and dead cells after lethal challenge is an indirect indication of an inherent heterogeneity in the original population. This population heterogeneity among bacterial clones has been attributed to asynchronous progression through the cell cycle, differences in cell age, mutations, variations in microenvironment conditions, and stochastic phenomena (52). Monitoring the kinetics of adaptation at the single-cell level has also revealed that the response is acquired on a cell-by-cell basis (34, 53), while not all cells are able to adapt within the same time frame or to the same extent (34). Current developments in “-omic” technologies are expected to allow for in-depth study of the stress physiology of single cells that is the basis of any stress response.

Two-Component SystemsTCSs are signal transduction pathways typically consisting of a usually membrane-bound sensor histidine kinase (HK) and a cytoplasmic response regulator (RR). HKs and RRs are modular proteins containing homologous domains, namely, a kinase domain and an H box in HKs and a receptor domain in RR, all of which are involved in the phosphotransfer reaction. They also contain heterologous sensory (HKs) and signaling (RRs) domains, which are involved in the reception of a specific stimulus and the delivery of the corresponding response, respectively. In general, detection of a specific stimulus triggers HK autophosphorylation on a conserved His residue and the subsequent transfer of the phosphate group to the receptor domain of its cognate RRs. Phosphorylation of the RR modulates its activity, which in most cases involves transcriptional regulation (55). Dephosphorylation of RRs is carried out by auxiliary phosphatases or, often, by the cognate HKs (56, 57). The final output response results from a balance between kinase and phosphatase activities.

TCS complements vary widely in LAB (58, 59). A clear correlation between genome size or lifestyle and the number of TCSs cannot be established, although, generally, species with the largest genomes encode the largest numbers of TCSs (58). Lactococci encode relatively few TCSs, ranging from 7 in Lactococcus lactis (Lc. lactis) strains IL1403 and MG1363 to 10 in Lc. lactis KF147 (59). The numbers of TCSs in streptococci vary from 8 in Streptococcus thermophilus LMD-9 to 31 in Streptococcus pyogenes MGAS8232 or S. pyogenes MGAS315 (59). In contrast to those in other LAB, orphan HKs and RRs are relatively frequent in streptococci.

The involvement of TCSs in stress responses in LAB has been evidenced mainly by phenotypic analyses of TCS-defective mutants (60–67). Inactivation of homologous TCSs often results in different phenotypes, suggesting that they have different physiological roles. For example, inactivation of the rrp-31 hpk-31 (LSA0277-LSA0278) system of Lactobacillus sakei led to premature arrest of growth under reference conditions (MRS broth at 30°C), poor growth at high temperature (39°C), sensitivity to heat shock, aeration, and H₂O₂, and a higher resistance against vancomycin (60). In contrast, inactivation of the Lactobacillus casei (Lb. casei) BL23 homolog TCS01 resulted in normal growth in MRS broth at 37°C and sensitivity to acid and vancomycin (61). In many cases, it is not clear whether the observed effects correspond to a specific stress response or to physiological changes that result in an altered ability to respond to stress. The few examples in which control by TCSs of the response to specific stressors has been established mostly concern those involved in the cell envelope stress response. These are dealt with further below, in Protecting the Cell Envelope.

One-Component SystemsAlthough the roles of TCSs in sensing and signaling have been studied extensively, OCSs are actually much more abundant in prokaryotes (54). OCSs are proteins containing sensory and signaling domains; they lack HK and receptor domains and can be identified by amino acid conservation in their DNA-binding domains and by different conserved motifs (54). Twenty families of major prokaryotic OCSs have been recognized so far (68). With the exception of the MetJ family, they are all represented in LAB. Knowledge about the involvement of OCSs in LAB stress responses is still scant, but a number of systems have been characterized, especially those involved in resistance against cationic antimicrobial peptides (CAMPs) (see Protecting the Cell Envelope), in oxidative stress, and in metal homeostasis and resistance.

Thermosensors in LABBacteria possess two main routes to sense a sudden temperature change and to transmit the information. First is the evolutionarily conserved response to the heat-induced accumulation of denatured proteins, and the second is the direct sensing of temperature changes through primary thermosensory structures, such as DNA, RNA, proteins, or lipids, which either have a direct effect or lead to the activation of signal transduction pathways (114). The Lc. lactis CtsR regulator, a winged helix-turn-helix dimeric DNA-binding protein, has been shown to function as a thermosensor (115, 116) (Fig. 1). CtsR regulates the expression of clp and other HSP genes in LAB (117). It has been demonstrated for several CtsR proteins, including that of Lc. lactis, that their activity depends on the temperature, as they bind to DNA with a higher affinity at lower temperatures. The temperature sensitivity of CtsR is adapted to the specific living conditions of different low-GC Gram-positive bacteria (115). The thermosensor region is a highly conserved tetraglycine loop within the winged helix-turn-helix domain. This conserved region, which possesses high conformational entropy and therefore displays decreased thermostability, senses specific temperature shifts and regulates gene activity, whereas the more flexible regions of CtsR are responsible for adaptation to host-specific temperatures (115).

The Stringent Response: the Ribosome as a SensorAccumulation of the alarmone (p)ppGpp in the cell triggers the stringent response (SR), a highly conserved bacterial stress response originally defined as a response to amino acid starvation but nowadays recognized as being triggered by a wide range of environmental stress conditions (118) (Fig. 1). The SR induces large-scale transcriptional alterations that ultimately lead to a physiological shift to a nongrowth state. In most Firmicutes, up to three genes code for (p)ppGpp synthetases (Rels): the bifunctional RelA/SpoT homolog, also named RelA or Rel (RSH), and the small RelQ and RelP proteins, with only a (p)ppGpp synthetase domain (119). An evolutionary analysis of the RSH superfamily showed that members of the Lactobacillales encode anywhere from one (O. oeni) to four (one RSH and three small synthetases) RSH proteins in some streptococci (120).

The SR has been implicated in acid stress resistance in Lc. lactis (121) and Lb. casei (122) and in a number of stress conditions in E. faecalis (123, 124), S. mutans (125), and S. pneumoniae (126). A number of studies have shown that RSH is mainly involved in the classical SR, whereas the small synthetases may play different roles. In E. faecalis, RelQ apparently operates only in maintaining baseline levels of (p)ppGpp during homeostatic growth (124). There is also evidence indicating that RelP and RelQ may have distinct and specialized functions in S. mutans (127).

It remains to be established in full detail how (p)ppGpp modulates cell physiology in LAB. In B. subtilis, (p)ppGpp affects the transcription of rRNA genes by reducing the availability of the initiating nucleotide GTP (128), and a more recent study showed that GTP is a limiting factor for the growth rate (129): GTP levels become detrimental to growth when they reach a certain threshold. (p)ppGpp is crucial for maintaining GTP homeostasis (129). Interestingly, growth inhibition by GTP stress also occurs in E. faecalis cells unable to produce (p)ppGpp, suggesting that this phenomenon might also be conserved among LAB (130).

The mechanisms of the regulation of expression of (p)ppGpp-producing enzymes in LAB are largely unknown. In S. mutans, relP is cotranscribed with the relRS TCS, whose inactivation results in a significant reduction in the basal level of (p)ppGpp. This indicates that RelRS is involved in the regulation of (p)ppGpp metabolism (127). The environmental signals to which RelRS responds have not been determined, although it has been suggested that RelRS may sense oxidative stressors or by-products of oxidative metabolism (131). Expression of relPRS is regulated by the MarR family transcriptional regulator RcrR, which is involved in stress tolerance and competence, although the signals to which RcrR responds remain unidentified (131, 132). There is some evidence suggesting that nucleotide pools may be involved in the modulation of (p)ppGpp production. Inactivation of the purine nucleoside phosphorylase DeoD in S. thermophilus resulted in a thermotolerant phenotype which correlated with an increased ppGpp content in this mutant (133).

The SOS Response in LAB: DNA as a SensorThe SOS response is usually induced by damage to DNA or by a stop of replication resulting in exposure of single-stranded DNA (ssDNA). Triggering of the response is regulated by the concerted actions of the repressor LexA and the activator RecA, resulting in the induction of expression of proteins involved in DNA repair (134). The SOS response has received little attention in LAB, with the exception of Streptococcaceae. Some streptococci can elicit a DNA damage response mediated by the regulator HdiR, which represses its own gene and that of the DNA polymerase UmuC, a protein involved in the SOS response in other organisms (135). RecA catalyzes the self-cleavage of HdiR, although an additional cleavage of the N-terminal fragment of HdiR by ClpP was required for induction of HdiR-repressed genes (135). A homologous system was subsequently characterized in Streptococcus uberis, in which a gene cluster consisting of hdiR, umuC, and two uncharacterized genes was identified (49). Expression of this cluster is under the control of HdiR and is induced by DNA damage (49). A recent study showed that a LexA-like transcriptional regulator of S. mutans (SMU.2027) is induced in response to heat or DNA damage and through the CSP-ComDE quorum-sensing (QS) pathway (45). In contrast to the typical SOS response, activation of S. mutans LexA leads to the formation of persister cells (45).

In streptococci, such as S. pneumoniae, antibiotics causing DNA damage induce competence in a RecA-dependent way, but the mechanism remained unknown (136). It now appears that it is caused by the chromosomal location of early competence genes. Slager et al. (137) showed that antibiotics targeting DNA replication cause replication to stall while initiation of DNA replication continues. This results in a higher copy number of genes close to the replication origin, including the comCDE operon. The increase in gene dosage is sufficient to trigger the competent state (137). The origin-proximal location of early com genes thus constitutes a sensory mechanism that enables cells to activate competence in response to antibiotics interfering with DNA replication (137). In contrast, the SOS response mediated by HdiR and competence are antagonistic processes in S. thermophilus (138). Intriguingly, competence in this organism is regulated by the comRS genes, which are located far from the origin of replication (137).

Cyclic Nucleotides as Second Messengers in LABSeveral cyclic nucleotides (cyclic AMP [cAMP], cyclic GMP [cGMP], cyclic di-GMP [c-di-GMP], cyclic di-AMP [c-di-AMP], and cyclic AMP-GMP) play key roles in the regulation of bacterial cell physiology (139, 140). Very little is known about their role in LAB physiology. In recent years, c-di-AMP has been identified as a major second messenger (reviewed in reference 141). c-di-AMP is synthetized from two molecules of ATP by diadenylyl cyclases (DACs), and it is degraded to pApA or AMP by phosphodiesterases (PDEs) (142, 143).

Among LAB, the presence and synthesis of c-di-AMP were first described for S. pyogenes (144). Inactivation of a gene encoding a DAC homolog in S. thermophilus, ossG, resulted in a methyl viologen-sensitive phenotype, suggesting the involvement of this gene in the oxidative stress response of this organism (144, 145). On the other hand, inactivation of the Lc. lactis PDE-encoding gene gdpP resulted in increased heat resistance and salt hypersensitivity (146). Results obtained so far suggest that c-di-AMP is important in LAB stress responses, but much remains to be unraveled about its role in LAB.

Quorum SensingBacterial quorum sensing regulates a number of cellular processes, including biofilm development, conjugation, competence, bacteriocin production, and pathogenesis. Furthermore, evidence suggests that it also plays a role in stress responses in oral streptococci, where production of competence-stimulating peptides is stimulated by stress conditions (147). Acidic conditions or exposure to spectinomycin increased the expression of the competence-stimulating peptide-encoding gene comC in S. mutans (148). Mutants of this organism impaired in the signaling pathway of the competence-stimulating peptides produced a significantly smaller number of persister cells following acid challenge, amino acid starvation, and/or oxidative stress (149).

Two kinds of signaling molecules have been identified in quorum-sensing systems in LAB: the autoinducer AI-2 and small pheromone peptides. AI-2 is a furanosyl borate diester produced and recognized by a wide variety of bacteria (for a review, see reference 150). AI-2 is produced from S-ribosylhomocysteine by the S-ribosylhomocysteine lyase LuxS, which is present in most Lactobacillales species, and a number of studies have shown that many LAB respond to AI-2 (see the references in references 150 and 151). Inactivation of luxS in S. pyogenes resulted in increased acid tolerance (152), while AI-2 has been shown to influence biofilm formation in other streptococci, such as Streptococcus intermedius (153), S. gordonii, and Streptococcus oralis (154). It still remains to be established clearly, however, whether AI-2 plays a relevant role in stress responses.

The role of small pheromone peptides or autoinducing peptides (AIs) in intercellular communication of LAB is better characterized. AIs can be detected at the cell surface by specific TCSs or recognized by intracellular receptors after internalization. An example of a pheromone peptide is the previously discussed CSP regulating competence in S. pneumoniae. S. mutans uses a different CSP and a paralogous TCS system to regulate both competence and production of bacteriocins (155). The second type of pheromone peptides is exemplified by the ComRS system of streptococci from the bovis, pyogenes, and salivarius groups (156, 157). The involvement of competence quorum-sensing systems in stress responses is discussed above.

Metabolism of Carbon Sources and Energy ProductionUnder environmental stress conditions, LAB change metabolic and energy fluxes, modify the rate of growth, and adapt the metabolism of carbon sources to the new environment by modifying the synthesis of enzymes and metabolites (161). Environmental stresses inhibit the glycolytic pathway of Lc. lactis and decrease the synthesis of biochemical energy (17). Similar metabolic perturbations are common in other LAB. Consequently, the ability of LAB to efficiently transport and metabolize carbohydrates and other carbon sources, such as malate and citrate, under environmental stress conditions is crucial for growth and persistence.

Metabolism of NitrogenThe metabolism of nitrogen, which includes proteolytic and urease systems, plays a pivotal role in LAB growth and survival. The regulation of nitrogen metabolism occurs mainly via GlnR and CodY (196, 197). A gene for GlnR is present in all sequenced genomes of LAB, while CodY is present only in strains belonging to the Lactococcus, Streptococcus, and Enterococcus genera. Under conditions of high nitrogen availability, GlnR controls the import of nitrogen-containing compounds and the synthesis of intracellular NH₃. On the other hand, CodY interacts with its BCAA coeffector to control amino acid biosynthesis, transport systems, and peptidases.

Preventing Macromolecules from Being DamagedSome stresses are imposed on cells through specific molecules, such as protons, ROS, metal ions, and salts (such as NaCl). LAB have developed various mechanisms to directly counteract the negative effects of these molecules (Fig. 4). Evidence suggests that other stress-causing events, such as irradiation, starvation, and highly/lowly osmotic environments, are indirectly manifested through some of the aforementioned molecules, especially ROS. Consequently, toxic molecules are not always associated with a single type of stress.

• Open in new tab
• Download powerpoint

FIG 4

Schematic representation of the main molecular mechanisms preventing macromolecules from being damaged in LAB. SOD, superoxide dismutase.

The Cell Envelope of LABLAB have cell envelopes that are typical of Gram-positive bacteria. They consist of a cytoplasmic membrane and a cell wall that may additionally be surrounded by an external polysaccharide capsule and/or a proteinaceous S layer. The cell wall itself comprises a PG sacculus that is usually decorated with (lipo)teichoic acids, polysaccharides, and proteins (for a review, see reference 358). The cell wall provides a point of (non)covalent attachment of important extracellular enzymes and external structures, such as pili and flagella (359). It also participates in processes of adhesion, cell aggregation, and cell division (360). Most importantly for this review, the cell envelope acts as a diffusion barrier between the cell and its surroundings. It is crucial for maintaining the integrity and shape of the cell and for resisting changes in osmotic pressure.

Mechanosensitive ChannelsWhen bacteria are faced with a sudden drop of the external osmotic pressure, they use mechanosensitive channels (MSC) to counteract the effects. The MSC act as emergency release valves, allowing the expulsion of osmolytes and water. MSC were discovered in bacteria more than 25 years ago (419), and the two major MSC were named after their E. coli equivalents, i.e., MscS (small conductance) and MscL (large conductance). MSC share an organization of a C-terminal cytoplasmic domain and an N-terminal transmembrane domain formed by symmetrical helices that can change between open and closed states (420). Although their mechanism of sensing is still unclear, it is well known that MSC can sense changes in the tension of the lipid bilayer of the cytoplasmic membrane (421). Most microbes appear to possess members of one or both families of bacterial MSC (422). MSC in LAB have been studied only for Lc. lactis (423). Although Lc. lactis has genes encoding both MscS (yncB) and MscL (mscL) channels, only the latter is functional and is used as the main mechanosensitive solute release system to protect cells under conditions of hypo-osmotic stress (423) (Fig. 1).

Stress Associated with Technological ProductionProbiotics are normally grown to high numbers before undergoing a drying process to produce a high-cell-density dried powder formulation. The stresses encountered during this processing include extremes in temperature, from very high (up to 200°C) during spray-drying to very low (down to −196°C) during freeze-drying and storage. Such extreme temperatures can affect membrane fluidity and compromise cellular integrity and basic cell processes, such as ribosomal function, protein folding, and enzymatic activity (27). Some species of lactobacilli are relatively thermotolerant; for example, Lb. paracasei 8R2 has been shown to survive thermization (60°C, 5 min) of cheese milk (with 90% survival) (453, 454). The effect of heat shock and the induction of a stress response have been studied in a variety of lactobacilli (see above). In general, the resistance to stress is higher when cells are previously exposed and adapted to a sublethal treatment with a homologous stress. For example, enhanced heat tolerance was obtained when Lb. rhamnosus GG was preexposed to 60°C for 20 min (451). Similarly, treatment of Lb. paracasei NFBC 338 at 52°C for 15 min increased the survival of the strain 700-fold in reconstituted skim milk and 18-fold during spray-drying (27, 455). This improvement was associated with induction of the chaperone GroEL. Heat stress was previously shown to enhance the expression of the heat shock genes hsp18.5, hsp19.3, and hsp18.55, carried on the genome of Lb. plantarum NC8 (339, 456). As mentioned earlier, CSPs are induced as a result of cold shock pretreatment, and they have been associated with the stabilization of mRNA (457). Three induced CSPs were previously identified in Lb. plantarum following cold shock treatment, namely, CspL, CspP, and CspC (457).

In addition to temperature stress, spray-drying also exposes cells to osmotic stress, dehydration, and oxidative stress (27). A sudden increase in the environmental osmolarity (hyperosmotic stress) results in the movement of water from the cell to the outside, causing a detrimental loss of cell turgor pressure and changing the intracellular solute concentration and cell volume, which ultimately can seriously affect cell viability (458). Probiotic bacteria can develop significant physiological changes to respond to osmotic stress conditions. Expression of the htrA gene increased 8-fold in Lb. helveticus CNRZ32 in response to exposure to NaCl (160). Similarly, the classical GroES/GroEL chaperone was positively regulated in osmotically stressed Lb. rhamnosus HN001 cells (459). Probiotic LAB can also adapt to hyperosmotic conditions by accumulating compatible solutes, such as betaine, carnitine, and trehalose (160, 458, 459).

Probiotic LAB may use enzymes (NADH oxidase, NADH peroxidase, and SOD) or nonenzymatic compounds (Mn²⁺, ascorbate, tocopherols, and glutathione) to reduce or eliminate the deleterious effects of oxygen radicals (16, 160). For example, Lb. delbrueckii subsp. bulgaricus ATCC 11842 can reduce oxygen to H₂O₂ with an NADH-dependent oxidase to eliminate or lower the level of oxygen (460). The same was found for Lb. plantarum ATCC 8014 through the oxidation of NADH (160). In a study on Lb. helveticus CNBL 1156, the fatty acid composition of the cell membrane changed in response to oxidative stress (461). This change was explained in terms of an increased activity of the oxygen-consuming fatty acid desaturase system, which serves to reduce free radical damage in the cell.

Stress Associated with Intestinal TransitThe principal stresses encountered by probiotic bacteria during passage through the stomach and upper intestinal tract include the low pH in the stomach as well as the detergent-like properties of bile in the duodenum. Exposure to these stresses causes damage to the cell envelope, DNA, and proteins (16). The survival of probiotic lactobacilli in low-pH environments can be increased hugely (several orders of magnitude) in the presence of metabolizable sugars, such as glucose (462). It is thought that glucose, by providing ATP to F-ATPase, enables proton exclusion and hence protection (see above). Other systems to regulate internal pH also exist; for example, Lb. reuteri uses both the GAD system and the ADI pathway to combat acid stress (463, 464). Similarly, amino acid decarboxylation combined with an amino acid antiporter leads to the biochemical consumption of protons, thus contributing to acid tolerance during growth (218, 465). The exposure of Lb. delbrueckii cells to acidic conditions prevailing in the GIT (pH 3.5 for 60 min) was shown to result in increased expression of the clpP, clpE, clpL, and clpX genes. As detailed above, the Clp chaperones actively promote refolding or degradation of damaged proteins (466).

Acid stress can also cause molecular changes in the cell surface. Moreover, bacteria are able to change their membrane lipid composition in response to external stresses. The composition of the cytoplasmic membrane can have a dramatic effect on the ability of probiotic bacteria to survive stresses ranging from heat to low pH. For example, the membranes of acid-adapted cells of Lb. casei ATCC 334 and Lb. johnsonii NCC 533 showed a significant increase in the ratio of saturated to unsaturated membrane fatty acids (122, 467). In the latter strain, the membrane fatty acid composition (and in particular the saturation level) is associated with changes in cell survival during heat or acid shock.

Probiotic LAB can also sense and develop resistance mechanisms against bile salts. For example, an eight-gene operon encoding a TCS, a transporter, an oxidoreductase, and four hypothetical proteins has been implicated in sensing bile salt in Lb. acidophilus (468). Likewise, the genes encoding multidrug transporters driving extrusion of bile salts are upregulated in Lb. casei and Lb. acidophilus during exposure to subinhibitory concentrations of bile (235). Indeed, mutations in the lr1584 gene (encoding an MDR transporter) impaired the ability of Lb. reuteri ATCC 55730 to reinitiate growth in the presence of bile and to survive in the human small intestine (469). Analogous results in terms of growth and survival were obtained when the LBA0552, LBA1429, LBA1446, and LBA1679 transporters were deleted in Lb. acidophilus NCFM (470). Exposure of Lb. delbrueckii to 0.1% bile salts increased the expression of clpP and clpE, which, as we have seen above, are involved in repair/degradation of damaged proteins (466). This is in agreement with the observation that mutations in lr1864 (clpL) in Lb. reuteri were associated with reduced survival in the presence of bile (469). Some probiotic LAB can efficiently express a range of bile salt hydrolases (BSH) that confer protection against bile via bile salt deconjugation (471). The cellular response toward bile of the well-documented probiotic Lb. rhamnosus GG showed that it can also respond by changing cell envelope-related functions, such as altering pathways affecting fatty acid composition, cell surface charge, and thickness of the EPS layer (236). Production of external EPS layers can have a significant protective effect against bile and other harsh environmental conditions. Indeed, strains producing EPS exhibit significantly more growth in the presence of 0.3% bile than strains that do not (472). Although intrinsic bile tolerance appears to be strain dependent, probiotic strains can be made to adapt progressively to the presence of bile salts by subculture in gradually increasing concentrations of bile (473).

In considering probiotics, it is also important that stationary-phase cells are generally more resistant than log-phase cells to various stressors, since many stress resistance mechanisms are switched on when active growth ceases. This is an important consideration, since probiotic LAB are generally prepared from stationary-phase cultures.

Increasing the Stress Resistance of Probiotic LABUnderstanding the complex stress response networks of probiotic bacteria has allowed the development of a variety of methods to improve their technological and gastrointestinal performance to potentially produce “superfit” bacteria. Some of the approaches aim at improving culture viability under the conditions of spray-drying, a procedure increasingly employed to concentrate and preserve particular probiotic cultures used in functional foods and health supplements. Exposure of Lb. johnsonii NCC 533 to acid stress significantly increased (P < 0.05) the oleic acid content in the membrane when the growth medium was supplemented with linoleic or linolenic acid, indicating that saturation of the membrane fatty acids occurred during acid stress (467). Thus, supplementation of the growth medium with saturated fatty acids is likely to increase the acid and heat tolerance. Presumably, this effect is due to increased rigidity of the cytoplasmic membrane. Overexpression of HSPs or CSPs in probiotic LAB has also proved successful. The overexpression of the GroES/GroEL chaperone in Lb. paracasei increased the strain's ability to better survive both spray-drying and freeze-drying (27). Similarly, overexpression of HSPs in Lb. plantarum improved bacterial growth at high (37°C or 40°C) and low (12°C) incubation temperatures (474). Overexpression of CspL, CspP, or CspC improved the performance during temperature downshift of this strain (475). Interestingly, heterologous expression of the BilE transporter in Lc. lactis resulted in a 2.5-log enhanced resistance of the modified strains to 1% porcine bile (27). Other authors have emphasized the importance of preconditioning cells with homologous stressors, particularly heat shock prior to spray-drying (453). Bile-adapted strains usually display cross-resistance to a variety of other stress factors. Finally, culture reconstitution conditions should be optimized for each probiotic strain in order to achieve accurate viable probiotic numbers from dried cultures (476).

General Stress Responses and VirulenceAs already discussed above, general stress responses are triggered by different stress conditions that result in the accumulation of misfolded proteins, inducing DnaK, GroEL, ClpP, and HtrA. The genes encoding the ClpP peptidase and the Clp ATPase partners are by far the best-studied streptococcal stress genes in terms of virulence potential. The virulence of S. pneumoniae ΔclpP strains was attenuated in mice, regardless of the administration method (477–479). The core genome of S. pneumoniae encodes four Clp ATPases (ClpC, ClpE, ClpL, and ClpX), although ClpL does not have the recognition tripeptide responsible for the interaction with ClpP; thus, it functions mainly as a chaperone. For an S. pneumoniae mutant lacking ClpE, virulence was strongly reduced in a mouse peritonitis model (480). Transcriptomic and proteomic analyses indicated that ClpE affects pneumococcal pathogenesis by modulating the expression of important virulence determinants and metabolism-related factors (480). Despite participating in several important physiological processes (481, 482), ClpC is apparently not required for pneumococcal virulence (477, 479), while inactivation of clpX is lethal (303). Although ClpL is required for growth at 43°C and for penicillin tolerance (478, 483), an S. pneumoniae ΔclpL mutant showed a level of virulence similar to that of the parental strain in a murine intraperitoneal infection model (478). Systematic deletion of clpP and the five Clp ATPase genes (clpB, clpC, clpE, clpL, and clpX) in the dental pathogen S. mutans revealed that the ClpXP proteolytic system modulates the expression of several virulence-related traits (304, 484) and that the infectivity of strains lacking clpP or clpX is slightly reduced in a rat model of oral colonization (304). Inactivation of the remaining Clp ATPases did not affect tooth colonization, but the role of the S. mutans Clp proteins in caries development or in infective endocarditis remains to be evaluated further. Virulence of a mutant GBS, S. agalactiae lacking clpP, was not attenuated in mice inoculated intravenously (485); the roles of ClpP and associated Clp ATPases in GAS virulence remain to be established. The involvement of Clp ATPases has not been studied extensively in enterococci. E. faecalis carries genes for ClpP, ClpB, ClpC, ClpE, and ClpX (214). Deletion of clpB reduced the ability of E. faecalis to acquire thermotolerance and for pathogenesis as assessed in Galleria mellonella larvae (308).

Another highly conserved bacterial protein associated with the pathogenic potential of streptococcal species is HtrA. Deletion of htrA in S. pneumoniae significantly reduced nasopharyngeal colonization (486) and virulence in mouse models of pneumonia and bacteremia (487, 488). The 50% lethal dose (LD₅₀) for an S. pyogenes htrA mutant in mice was 35-fold higher than that for the parent strain (489). A subsequent study revealed that HtrA is responsible for processing secreted virulence factors, such as the SpeB cysteine protease and streptolysin S (490). However, an HtrA-deficient GAS mutant was not attenuated in a murine model of subcutaneous infection (490).

Additional general stress proteins have been implicated in stress responses and virulence of LAB. The trigger factor protein, a ribosome-associated molecular chaperone functioning in cooperation with the DnaK and GroEL chaperones, has been associated with stress tolerance in different streptococcal species (491–493). For the zoonotic pathogen S. suis, a trigger factor-deficient strain was attenuated in a mouse peritonitis model (493). Two general stress proteins, Gls24 and GlsB, have been identified in E. faecalis (494). Deletion of gls24 led to decreased lethality in a mouse peritonitis model and to sensitivity to bile salt. Deletion of glsB did not influence virulence but resulted in sensitivity to bile salt. Similar conclusions concerning the involvement of Gls24 and GlsB in pathogenicity were also reached in a mouse endocarditis model (495). Two paralogous gls24-glsB loci (gls33-glsB and gls20-glsB1) were identified in E. faecium. Only deletion of both loci resulted in reduced pathogenicity, while sensitivity to bile stress was observed in the double mutant and each of the single mutants (496).

Global Transcriptional Regulators and VirulenceGlobal transcriptional regulators, such as quorum sensing-activated regulators and serine-threonine protein kinases, which regulate a variety of cellular processes, including stress survival, have also been shown to mediate streptococcal virulence. The universal LuxS QS system, which produces the furanosyl borate diester AI-2, regulates a large panel of genes in a variety of microorganisms. By luxS inactivation in GAS, it was demonstrated that loss of AI-2 signaling leads to increased acid tolerance, increased cell invasion, and increased intracellular survival (152, 497). However, the exact role of the LuxS/AI-2 system in GAS pathogenesis remains to be elucidated. An S. pneumoniae luxS mutant could colonize the nasopharynx of mice as efficiently as its parent, but it was impaired in the ability to disseminate to distant sites, such as the lungs, and was less virulent in a peritonitis mouse model (208). Inactivation of the luxS gene in S. suis increased tolerance to H₂O₂ but negatively affected capsule production and adherence to epithelial cells (498). As a result, the luxS mutant was dramatically attenuated in a piglet model (498).

Transcriptome studies have highlighted the importance of several TCSs of GAS in coordinating the expression of genes involved in stress adaptation and virulence (499–502). In particular, the TCS CovSR was shown to regulate approximately 15% of the genome of S. pyogenes and is required for growth under general stress conditions (503). In a murine model of nasopharyngeal colonization, the CovSR system was shown to be required for S. pyogenes for optimal infection and transmission from the nasopharynx (504). Likewise, S. pyogenes covR mutants caused significantly less invasive disease and death in mice than those seen with the wild-type strain, although the local lesions produced by the covR mutants were more severe and purulent (505). A GBS mutant of the CovSR ortholog, also known as CsrS/CsrR, displayed impaired viability in human serum and attenuated virulence in murine infection models (506, 507). It has been hypothesized that the translocation of GBS from the acidic milieu of the vagina to neonatal tissues with a neutral pH signals the conversion from a colonizing to an invasive phenotype in a CsrRS-dependent manner (508). The CiaHR TCS of S. pneumoniae regulates many genes involved in competence development and virulence, including htrA (see above) (486). Virulence of an S. pneumoniae ciaR mutant in mice was significantly attenuated (509). As described above, HtrA has been implicated in virulence by influencing the ability of pneumococci to colonize the nasopharynx of rats (486, 487). Notably, the reduced virulence of the ciaR mutant could be restored by increasing the expression of HtrA (509). The S. suis CiaHR ortholog was also shown to contribute to virulence in mice and pigs (510).

In addition to TCS signaling systems, an alternative regulatory pathway involving reversible phosphorylation is controlled by the eukaryote-type serine/threonine kinase (Stk). Stk of S. pyogenes activates genes involved in osmoregulation, fatty acid and cell wall synthesis, and virulence (511). An S. pyogenes stk deletion mutant was more sensitive to penicillin and osmotic stress than the wild-type strain, and Stk was required for optimal virulence (511, 512). An S. suis stk mutant exhibited reduced tolerance to a multitude of stresses and showed attenuated virulence in animal models (513).

Oxidative Stress and VirulenceBacteria are constantly exposed to oxidative stress of either an intracellular or extracellular origin. Of particular relevance to bacterial virulence is the organism's ability to evade the oxidative onslaught of phagocytic immune cells by using a combination of reducing enzymes, ROS scavengers, and protein and DNA repair enzymes. Thus, it is not surprising that the virulence of strains with defective oxidative stress responses is often attenuated or totally abolished. Streptococci and enterococci are generally considered unable to carry out oxidative phosphorylation. Instead, the bulk of oxygen metabolism in streptococci is due to the NADH oxidase Nox, which reduces oxygen, one electron at a time, to water through the oxidation of NADH to NAD⁺. If not fully reduced by Nox, environmental oxygen can form toxic superoxide and H₂O₂. Strains of GBS and pneumococcus lacking Nox were unable to grow under vigorous agitation conditions and showed attenuated virulence in animal models (514, 515). An E. faecium Tn917 transposon insertion mutant of nox showed a 98% reduction of NADH oxidase activity (516). The mutated strain was defective in nematode killing as a result of less H₂O₂ production than that in the wild type. The possible involvement in the observed phenotype of the gene downstream of nox through a polar effect could not be ruled out given the inability to perform a complementation analysis of nox.

The genome of GBS encodes a cytochrome bd oxidase (CydABCD), and the organism can engage in respiratory metabolism when the environment provides quinone and heme (517). Inactivation of cydA negatively affected GBS growth in blood and strongly attenuated virulence in a neonatal rat sepsis model (517). The animal pathogen S. uberis also contains cytochrome bd oxidase; since it lacks quinone and heme biosynthesis pathways, it is thought to use a similar strategy to activate respiration. E. faecalis also carries a cytochrome bd-type respiratory oxidase (518), but its involvement in virulence has not been investigated.

Streptococci lack catalase but encode several other detoxification enzymes, including SOD and up to three peroxidases. For GBS, S. pneumoniae, and S. suis, sod mutants were more susceptible to oxidative stresses and showed attenuated virulence in different murine infection models (519–521). The contributions of alkylhydroperoxidase (AhpCF) and thiol peroxidase (Tpx) to streptococcal virulence have been assessed only for S. pneumoniae. Inactivation of tpx impaired pneumococcal virulence in mice infected intranasally but not in those in which the bacteria were administered directly into the bloodstream (522). These results most likely can be explained by the differences in oxygen tension in the nasopharynx (high) and blood (low). Loss of alkylhydroperoxidase activity attenuated pneumococcal virulence in cutaneous, pneumonia, and bacteremia models (523). The third characterized streptococcal peroxidase, glutathione peroxidase (GpoA), is essential for GAS pathogenesis in several murine models that mimic suppurative diseases but not in a zebrafish streptococcal myositis model characterized by the absence of inflammation (524). In contrast to streptococci, E. faecalis contains the heme-dependent catalase KatA, which is activated by heme taken up from the environment (259); its implication for virulence has not yet been investigated. Deletion of E. faecalis sodA attenuated resistance to H₂O₂ and survival in mouse macrophages (254) and microglia (525). E. faecalis also contains three peroxidases: Npr, AhpCF, and Tpx (248). All three are involved in resistance to H₂O₂, to various degrees, while survival in murine macrophages was mostly affected in the Δtpx mutant and the peroxidase triple mutant. Two additional antioxidant repair enzymes, the methionine sulfoxide reductases MsrA and MsrB, have been characterized for E. faecalis (526). Mutants carrying a deletion in the respective genes exhibited sensitivity to H₂O₂ and attenuated virulence in systemic and urinary tract infection models.

In addition to reducing enzymes and molecular scavengers, the DNA-binding protein Dpr (also known as MrgA in GAS) is another major player in oxidative stress responses. Dpr provides protection against oxidative stress by binding to iron and thereby preventing the Fenton reaction. An S. pneumoniae Δdpr strain displayed a reduced colonization ability and was more rapidly cleared from the nasopharynxes of infected animals (527). Despite the in vitro hypersensitivity to peroxide (528), loss of Dpr did not affect GAS virulence in zebrafish or in different murine infection models (529).

Two ubiquitous LAB oxidative stress regulators have been shown specifically to be required for streptococcal and enterococcal virulence. The PerR metalloregulator, also involved in iron homeostasis, has been studied extensively in GAS. Virulence of perR mutants was significantly attenuated in a variety of murine models as well as in a baboon model of GAS pharyngitis (529–531). Likewise, the pathogenicity of an S. suis perR mutant was attenuated in a mouse peritonitis model (532). Deletion of perR in E. faecalis increased resistance to H₂O₂ but did not cause major alterations in the expression profile of eight genes related to oxidative stress (94). The perR mutant survived as well as the wild type in murine macrophages, but it was less lethal in a mouse peritonitis model. These findings indicate that although PerR plays a role in the virulence of E. faecalis, its involvement in the oxidative stress response of this bacterium may be different from that in B. subtilis (94). E. faecalis HypR was identified as a regulator of the oxidative stress response and as a virulence factor (533). It is a transcriptional activator of several genes during oxidative stress, including ahpCF, tpx, ef3270 (encoding a glutathione reductase), sodA, and katA (248, 534). An E. faecalis hypR mutant was highly sensitive to H₂O₂ and showed a reduced ability to survive in macrophages and a decreased lethality in a mouse peritonitis model (533, 534). Recent studies have shown that streptococcal genomes encode two copies of the redox-sensing Spx regulator, designated SpxA1 and SpxA2. In S. mutans, loss of spxA1, spxA2, or both attenuated virulence in the G. mellonella invertebrate model (346), while the ability to colonize the teeth of rats was significantly impaired in the ΔspxA1 mutant and the double mutant but not in the ΔspxA2 mutant (346). SpxA1 and SpxA2 were shown to modulate stress tolerance in S. suis, and loss of spxA1 or spxA2 significantly reduced the ability of S. suis to survive in pig blood and to disseminate to different tissues (347). In a rabbit model of infective endocarditis, a Streptococcus sanguinis ΔspxA1 strain exhibited an approximately 5-fold reduction in competitiveness in a competitive index assay (348). Although SpxA2 has also been implicated in the stress tolerance of S. sanguinis, its relevance in vivo has not been investigated. E. faecalis carries only one spx gene, and when it was deleted, the mutant strain showed defective growth under aerobic conditions and sensitivity to oxidative stress agents (535). The deletion mutant was also susceptible to killing in murine macrophages and was less efficient at colonization and dissemination in a murine peritonitis model. AsrR (antibiotic and stress response regulator), an oxidative sensing regulator of the MarR family, was recently described for E. faecium (536). The AsrR regulon consists of genes involved in pathogenesis, antibiotic resistance, oxidative stress, and adaptive responses. An E. faecium ΔasrR strain showed greater persistence in G. mellonella colonization and mouse systemic infection models but a lower survival rate in murine macrophages. Deletion in E. faecalis of the gene encoding the PrfA-like regulator Ers (enterococcal regulator of survival) resulted in sensitivity to H₂O₂ and decreased survival in an in vivo-in vitro macrophage infection model (537).

Acid Stress and VirulencePathogenic streptococci must be able to withstand the low pHs encountered upon the formation of necrotic lesions or abscesses. In addition, cariogenic streptococci, such as S. mutans, need to cope with sustained acidification of their environment as a result of the production of lactic acid from fermentable sugars by LAB residing in plaque. Because of the direct linkage between low-pH survival and dental caries and the ease of genetic manipulation, S. mutans has become the bacterial paradigm for LAB acid stress responses (538). Notwithstanding this, the number of investigations that directly associate S. mutans acid stress survival pathways with caries formation is still rather limited. In addition to the aforementioned in vivo studies with Clp chaperones/proteases and the Spx oxidative stress regulators, which are also important for acid stress survival (304, 346), several caries studies were carried out using S. mutans strains lacking different TCS regulators or the PknB serine/threonine kinase (539–541). An S. mutans fabM mutant did not produce unsaturated membrane fatty acids, was extremely acid sensitive, was poorly transmitted from host to host, and exhibited fewer and less severe carious lesions than those caused by the parent strain (542, 543).

Starvation and VirulencePathogenic LAB often have to face severe starvation or intermittent nutrient availability, depending on the infection site. As mentioned above, LAB have developed sophisticated regulatory networks that combine transcriptional regulation with allosteric modulation of enzymatic activities to coordinate nutrient biosynthesis and acquisition. They utilize nutrient-sensing regulators controlling the expression of genes involved in adaptation to starvation as well as those encoding bona fide virulence factors, such as toxins, proteases, and adhesins. The roles in streptococcal virulence of the key nutrient stress-responsive regulators CcpA and CodY and of the (p)ppGpp alarmone have been tested in in vivo models. The ccpA mutants of the GAS S. pneumoniae and S. suis were significantly less virulent than their parents in different mouse models of infection (544–548). To date, studies on the roles of CodY and (p)ppGpp in streptococcal virulence are restricted to S. pneumoniae. A CodY-deficient strain poorly colonized the nasopharynx of mice, a finding that was supported by its diminished ability to adhere to nasopharyngeal cells in vitro (549). An S. pneumoniae mutant lacking Rel, the major (p)ppGpp synthetase activating the stringent response, was severely attenuated and displayed a significantly altered course of disease progression in a murine pneumonia model (126). Deletion of ccpA in E. faecium also resulted in attenuated virulence in a rat endocarditis model (550). The levels of (p)ppGpp in E. faecalis correlated with stress responses, antibiotic resistance, and virulence (123, 124, 130). An E. faecalis mutant lacking Rel (RelA), the main one of the two (p)ppGpp synthetases in this organism (see above), exhibited altered behavior with respect to different stress conditions and antibiotic resistance, but its pathogenicity was similar to that of the wild type (123, 124). Only the double relAQ mutant showed diminished pathogenic potential in Caenorhabditis elegans and G. mellonella invertebrate models and impaired growth and survival in a rabbit subdermal abscess model (124, 551, 552).

Phage InductionActivation of the so-called SOS response causes induction of several lambdoid lysogens (560). Extrinsic stress factors, such as ROS, UV radiation, and various chemicals, all induce DNA damage and DNA replication arrest, thereby instigating the SOS response and, consequently, prophage excision and lytic activation. Much of the current literature on phage induction relates to phage lambda: when RecA becomes activated, it triggers the SOS response and catalyzes autoproteolysis of the lambda CI repressor, thereby causing a switch from a lysogenic to a lytic life cycle (561). As in lambda, genomes of temperate LAB phages carry a genetic switch region encompassing divergently oriented promoters that are controlled by analogs of the CI and Cro proteins (562–566).

Further factors that influence lysogenic maintenance or phage induction in LAB include various environmental stresses, such as pH, oxidative stress, temperature fluctuations, osmotic pressure, and low nutrient concentration, as well as the presence of toxic compounds, such as fluoroquinolones. The effects of the latter compounds were studied in Streptococcus spp., where they were shown to induce phages via an SOS-dependent, i.e., RecA-mediated, route due to a response to a topoisomerase IV-fluoroquinolone complex (567).

The effects of extracytoplasmic stress on prophage induction are, however, not well understood, especially for polylysogenic hosts (568). In this context, for phage LC3, which infects Lc. lactis strain 3107, it was shown that a decrease in pH caused a reduction in the level of phage induction (569). It has been postulated that changes in the intracellular pH affect the stability and binding features of LexA and CI repressors and thus influence the transcription of their target genes (570). In contrast, various other stressors (e.g., exposure to high temperature or high osmotic pressure or being in late stationary phase) did not provoke phage induction in Lc. lactis (571), suggesting that the control system that keeps lactococcal prophages in their lysogenic state is very tight and that stressors which commonly occur in the industrial environment of Lc. lactis do not influence phage induction (571).

Stress Mediated by Phage InfectionWhen a phage particle infects a bacterial cell, it causes major changes in the host transcriptome that are considered phage infection stresses. In this context, phage infection of Lc. lactis IL1403 is perceived as a membrane stress by the host, triggering adjustments in the cell envelope (572). This is believed to be orchestrated by the action of membrane-associated, phage shock-mediated protein C-like homologs, the global regulator SpxB, and the two-component regulatory system CesSR (572). Analysis of the host transcriptome response to infection by the lactococcal phages Tuc2009 and c2 revealed the induction of genes involved in metabolic flux and energy production, in cell wall modification, and in the conversion of ribonucleotides to deoxyribonucleotides (572, 573).

The Genomic AccelerationIn 1995, the first genomic sequence of the free-living microbe Haemophilus influenzae was published (574), launching the genomic era of molecular microbiology. Although the initial focus of microbial genome sequencing was centered on pathogenic bacteria, advances in high-throughput and low-cost sequencing technologies over the past decade have expanded the genomic perspective to a large variety of microbes relevant to biotechnology as well, including LAB. The first LAB genomes were published in 2001 (575). The 2.4-Mbp Lc. lactis subsp. lactis strain IL1403 genome was predicted to encode 2,310 proteins. The observed close relatedness of Lc. lactis to the streptococcal genus was further supported by the determination of the genome of the yogurt bacterium S. thermophilus (576). A comparison of the genomes of representatives of the Lactobacillus genus, Lb. plantarum (213) and Lb. johnsonii (319), exemplified the high degree of diversity within this genus (577), while the genome of Lb. acidophilus (201) underlined the relatively high relatedness of subgroups within this genus, i.e., the “acidophilus complex.”

Comparative genomics of LAB took a strong step forward by the comparative analysis of nine genome sequences of various LAB species, including the definition of the so-called LaCOGs, which represent an LAB-specific refinement of the existing categories of orthologous genes (COGs). The distributions of LaCOGs among the genomes of these nine LAB allowed analysis of their genome-wide evolutionary relatedness (6). Comparative genomics also enabled the comparative analysis of specific subsets of functions, e.g., the secretome, encompassing all proteins exported to the cell surface and thus playing a direct role in interaction with the environment (578–580) and contributing significantly to stress tolerance (see below).

These species comparisons were followed by comparative genomic analyses of strains within a species, initially employing comparative genome hybridization (CGH) approaches for different LAB. At present, it is quite common to determine genome sequences for multiple strains of an LAB species (581–583). As a consequence, the GenBank database contains more than 100 complete LAB genomes and an even larger number of partial genome sequences, representing LAB isolated from a variety of environmental sources and encompassing more than 50 LAB species. This wealth of information requires accurate and effective bioinformatic mining strategies to evaluate the relatedness and evolutionary relationships of strains, but also to analyze specific gene categories, such as those associated with stress responses and robustness. Notably, a variety of biotechnologically important functions are known to be encoded on plasmids (584–587), underlining that the evaluation of the genetic repertoire of LAB should pay appropriate attention to these extrachromosomal genetic elements.

Comparative Genomics of the LAB StressomeThe original LaCOG assignments enabled effective mining of LAB genomes for genes associated with stress response functions (i.e., the LAB stressome) and their regulators (588). Most LAB encode the typical HrcA regulator that acts as a repressor of the class I HSPs, encompassing the universally conserved chaperonin functions (GroELS, DnaJK, and GrpE) but in various LAB also including additional chaperones, such as HtpX and HSP20 (IbpA). As mentioned above, a notable exception is O. oeni, which appears to lack an hrcA gene, and class I HSP regulation in this species has been proposed to be controlled by complex regulatory networks involving multiple regulators (589). The involvement of CtsR in regulation of class III stress proteins, including the Clp proteases and related functions, is predicted for many LAB, although ctsR appears to be absent in several LAB genomes, such as those of L. mesenteroides and “acidophilus complex” lactobacilli. Genes for the class III-associated regulon Clp proteases (ClpC, -X, -Q, and -P) are present in most LAB genomes, whereas the presence of other Clp-encoding genes (clpYQ; also designated hslUV) is more variable (588). Protection against oxidative stress in LAB involves a variety of functions that are highly variable among the LAB species and strains (including many oxidases, peroxidases, catalases, and superoxide dismutases [see above]). Nevertheless, a central role in oxidative stress tolerance has been proposed for the strongly conserved thioredoxin system (TrxA and TrxB) (588, 590, 591). Many environmental stress conditions can lead to DNA damage, and DNA damage responses also appear to be strongly conserved among LAB, including functions associated with homologous recombination and double-strand-break repair (RecABDFJNOR, RuvAB, and Ssb), the homology-independent facilitator complex (GyrAB and TopA), and the pathway involved in base excision repair (Mfd, UvrABCD, and Xth), although the endonuclease IV (Nfo) that is important in this pathway is absent in many species (588). Notably, the highly conserved MutS and MutL functions that are involved in DNA mismatch repair appear to be lacking in O. oeni, which also lacks the related RecQ-encoding gene. The latter gene is also absent in S. thermophilus, which has been suggested to be related to the high frequency of pseudogenes and function loss observed in this species (250, 576). Notably, this typical genome decay is a characteristic that S. thermophilus shares with its protocooperative partner in yogurt fermentations, i.e., Lb. delbrueckii subsp. bulgaricus, and it has been proposed to be a consequence of their extensive adaptation to the nutrient-rich milk environment (320).

These examples illustrate the acceleration of our understanding that is promoted by (comparative) genomics, identifying both conserved and different mechanisms of stress responses in LAB that contribute to their robustness and survival under the challenging conditions encountered in industrial applications. However, an important restriction of these approaches is that genomic sequences provide no more than a blueprint of the encoded capacities. Experimental verification of predicted functions in stress responses is required to establish the roles of gene regulatory networks and the effector functions they control in stress tolerance and robustness.

Functional Genomic Approaches To Unravel the LAB StressomeDNA microarrays have been employed intensively to identify genetic factors regulated during stress exposure of LAB. The majority of studies have been performed under laboratory conditions in which a single stressor is applied. Recently, these transcriptome studies have been complemented with several proteome studies. Specific studies also report the responses to multiple individual stresses or combined stresses in single strains or several strains of an LAB species. Such multiple “stressomics” approaches are highly valuable for comprehensive stress response analyses, as they not only identify the genes directly involved in robustness and/or stress survival but also can reveal the regulatory networks involved. Moreover, the availability of several transcriptome profiles for a strain grown under different (stress-related) conditions (592, 593) allows identification of genes that are coregulated. This information can then be used to reconstruct gene regulatory (stress) networks and regulons (351). An alternative approach to identifying stress regulons is the transcriptome analysis of regulator mutants, with sometimes highly revealing results. For example, individual mutations of the regulator genes hrcA and ctsR of Lb. plantarum WCFS1 (see above) led to loss of regulation of their known regulon members but also significantly changed the expression of a variety of genes associated with transport and binding of sugars and other compounds, primary metabolism, transcription regulation, capsular polysaccharide biosynthesis, and fatty acid metabolism. These findings are in good agreement with global proteome analyses of Lb. plantarum WCFS1 and its ΔctsR derivative under optimal and heat stress conditions (594). An even more pleiotropic effect on the genome-wide transcription profile was observed in a strain deficient for both hrcA and ctsR, which illustrates the complexity of and cross talk between the gene regulatory networks influenced by these stress-associated regulators (318).

Since HrcA and CtsR act as repressors of transcription of class I and class III stress responses, respectively, the deletion of their genes might be assumed to result in strains that are more stress tolerant. Indeed, a higher stress robustness of ctsR deletion mutant derivatives has been reported for several LAB. For example, compared to their cognate wild-type strains, an S. thermophilus ctsR mutant was more tolerant to heat stress during exponential growth (595), and a ctsR-deficient Lb. plantarum strain could resist higher levels of oxidative stress (318). On the other hand, mutation of ctsR in S. thermophilus increased osmotic and oxidative stress sensitivity (595, 596), while an Lb. plantarum ctsR mutant was more sensitive than its parent to ethanol and heat stresses (318). These results clearly illustrate the unpredictability of the impact in LAB of deregulating stress responses, especially multiple stress regulons at the same time. This is most likely related to the pleiotropic effects on gene transcription in these (multiple) regulator mutants, which may affect many phenotypes other than stress tolerance (318).

The unpredictable nature of the stress robustness phenotype with enhanced stress regulon expression (e.g., as seen in the hrcA and ctsR mutants) is in apparent contradiction with the observation that preexposure to specific sublethal stress conditions can induce adaptation responses that protect against various subsequent severe stresses (451). These cross-protective effects are probably related to similarities in the molecular damage caused by different stresses to DNA, protein, or cell envelope components. This assumption implies that cross-protection depends on the induction of generic damage protection and repair functions, such as those encoded within the class I and class III stress regulons. For example, the adaptation of Lc. lactis to sublethal acid stress induced improved robustness under subsequent heat, ethanol, oxidative, and osmolality stresses (597). Cross-protective stress responses can be employed to design preadaptation approaches for the production of robust LAB with improved survival and functional properties under industrial application conditions. Genomics-based approaches have started to shed light on the genes and regulatory mechanisms involved and may thereby accelerate the exploitation of such preadaptation strategies.

Near-zero growth rates can be induced by carbohydrate-limited cultivation in a retentostat (an adjusted chemostat that retains biomass by use of a cross-flow filter in the effluent line). This setup forces microbes into a state where cellular physiology and metabolic energy are reoriented from growth-associated processes toward those dedicated to maintenance (598, 599). Retentostat studies have been reported for LAB, for example, Lc. lactis (600) and Lb. plantarum (601). Transcriptome studies of Lc. lactis adaptation to retentostat conditions not only confirmed that the expression of metabolic genes was adjusted to sustain zero-growth physiology (600, 602) but also revealed the induction of general stress responses, including derepression of the HrcA and CtsR regulons. These results showed that a tight correlation exists in this organism between growth rate and stress robustness (603). Stress exposure experiments employing severe heat, acid, and oxidative stress treatments confirmed that near-zero-growth cultures of Lc. lactis displayed strongly enhanced stress robustness compared to faster-growing cells. Moreover, mathematical modeling defined quantitative relationships between growth rate, stress robustness, and expression of stress genes (603). This work illustrated the congruency of the intrinsic stress regulation network in Lc. lactis with the generic evolutionary trade-off between growth and survival (604). In this context, it is important that protein turnover is considered one of the most important metabolic costs in bacterial cells, including Lc. lactis (605), and that generic stress responses play important roles in the capacity of cells to maintain or restore the functionality of cellular components, for example, by the refolding capacity of chaperonin complexes, such as GroELS and DnaKJ. This consideration explains the biological relevance of stress response induction in near-zero-growth cultures of Lc. lactis, which are entirely focused on maintenance-associated processes and benefit from these stress responses by reducing protein turnover costs. Although the activation of general stress responses under near-zero-growth conditions appears to be conserved among some microorganisms, it appeared to be absent in others, e.g., Lb. plantarum. In the latter, predominantly SOS responses are induced, which may enhance mutation rates in these microbes rather than their intrinsic robustness (606). These recent studies illustrate how an improved (genomics-driven) understanding of the regulatory networks involved in stress response regulation can help in the design of industrial approaches to harness the intrinsic adaptive responses of LAB for enhanced industrial robustness.

Applications of Functional Genomic Technologies In SituThe data described above were generated with simplified laboratory models that do not fully mimic the physicochemical complexity encountered during industrial fermentation and processing (607, 608) or the multitude of stresses encountered in the GIT (609). Therefore, in situ transcriptome and proteome profiling studies are valuable, complementary approaches to identify factors contributing to the functionality of LAB under these circumstances. A proteomic approach was used to unravel, e.g., the adaptive behavior of S. thermophilus during milk fermentation. That study revealed a strong regulation of sugar metabolism pathways and induction of the transport and biosynthetic pathways for sulfur-containing amino acids (610). Despite the insight generated, the study did not reflect the most predominant industrial application of S. thermophilus, namely, coculture with Lb. delbrueckii subsp. bulgaricus during yogurt production. Therefore, the same group also reported on the S. thermophilus gene expression profiles in milk during coculture with Lb. delbrueckii (611), whereas a complementary transcriptome study proposed that formic and folic acid production by S. thermophilus and the proteolytic capacity of Lb. delbrueckii provide the metabolic dependencies that stabilize this classic industrial coculture fermentation system (612). Analogously, recent studies reported on the global transcriptome and proteome profiles of Lc. lactis and Lb. helveticus during growth in milk (613, 614). The response of probiotics to GIT conditions is also highly relevant. Notably, in situ transcriptomes of Lb. johnsonii (615) and Lb. plantarum (616, 617) residing in the GIT have been reported. These studies illustrate the complex molecular adaptations of these LAB to the many physicochemical challenges encountered during mammalian intestinal tract transit. Many of the elicited transcriptional responses were associated with metabolic flexibility and energy metabolism, emphasizing the competitive nature of this habitat that is densely colonized with microbes. Also, significant modulations of cell envelope-associated functions were seen, illustrating the importance of cell surface architecture in this challenging niche.

Experimental evolution strategies are very well suited to investigate the molecular adaptations enabling the enhancement of complex phenotypes. Low-cost next-generation sequencing has drastically boosted the downstream genomic analysis of experimentally evolved variants. Experimental evolution has been used to study the adaptation of LAB to novel or altered environmental conditions. An elegant example is the adaptation of the plant isolate Lc. lactis KF147 to growth in milk. Accumulation of mutations that enhanced the capacity of the variants to harvest amino acids from milk was observed, while other mutations suppressed the broad-spectrum carbohydrate utilization capacity that is typical for plant isolates of this species. Since the transcriptome signatures of the milk-adapted variants were significantly more similar to those of typical Lc. lactis dairy strains, these adaptations were referred to as domestication signatures, based on the assumption that historically, initial milk fermentations involved plant-derived strains (618). Experimental evolution strategies to enhance LAB stress robustness have also been reported. For example, Lc. lactis variants with enhanced heat resistance contained single amino acid substitutions in a membrane-bound stress signaling protein of the GdpP family that was predicted to exhibit c-di-AMP-specific phosphodiesterase activity, and these elicited altered stress responses in these variants. Notably, the heat-resistant variants displayed hypersensitivity to salt stress compared to the wild-type strain, which may illustrate the existence of an evolutionary trade-off between these two phenotypic traits (146) but may also be a consequence of the specific experimental setup employed for variant selection (619). An experimental evolution approach to adapt Lb. plantarum to the complex conditions of the mammalian GIT has also been reported (620). Repeated exposure to the mouse intestinal tract by three consecutive rounds of (re)feeding and isolation of colonies that displayed the longest persistence in the system allowed isolation of independent intestine-adapted variants with a >5-fold prolonged gut residence time compared to that of the original strain. The variants had accumulated mutations that could predominantly be assigned to functions associated with energy metabolism and the synthesis of cell envelope components (620). This study corroborates the importance of the latter functions in survival and persistence in this complex habitat deduced from in situ transcriptome signatures of wild-type strains (see above). Intriguingly, cell envelope-associated functions have also been reported to be the main effectors driving the strain-specific host interactions that are likely to support the health benefits of probiotic products, exemplifying the dualistic nature of these functions in the context of probiotic functionality (621–623).

Gene Function Discovery by Phenotype Diversity Mining StrategiesAlthough stressome-related functions appear to be quite conserved among strains of a species and even among species (see above), the stress robustness of strains of a species is quite variable. This phenotypic variation has a major impact on survival and performance under industrial conditions. It is not restricted to stress tolerance but encompasses many other aspects of in situ functionality of LAB. For example, different Lc. lactis strains have different effects on flavor formation in cheese production (624, 625). Phenotypic strain variation is at least partially caused by gene content differences, which implies that mining the data from genotype-phenotype matching (GPM) approaches may reveal novel gene functions. The validity of the GPM approach for LAB was first illustrated by the discovery in Lb. plantarum of the gene encoding the mannose-specific adhesin (Msa), which was proposed to play a role in the probiotic competitive exclusion of enterotoxic E. coli in the human intestine (626). Subsequent comparative analysis of the msa gene and the Msa protein in various strains of this species revealed strain-specific domain compositions of Msa, which were correlated with strain-specific mannose adhesion capacities (627). Advanced GPM approaches based on the Random Forest algorithm for correlation analysis (628, 629) identified or verified the roles of several genes in sugar metabolism in Lc. lactis (628) and various metabolic capacities in Lb. plantarum (630). However, while GPM approaches employing strain-specific heat and oxidative stress robustness phenotype variations in Lc. lactis or intestinal robustness phenotype variations among strains of Lb. plantarum did identify genes with correlated gene presence-absence patterns, subsequent mutational analyses did not verify their function in the respective phenotypes (631, 632).

Clearly, many phenotypes may not be determined by the “simple” presence or absence of one or more genes but may depend on strain-specific levels of expression of conserved genes. An example of such transcriptional diversity leading to phenotypic diversity has been described for Lc. lactis, where the activity levels of five enzymes involved in flavor formation were shown to be subject to highly strain-specific transcriptional regulation (633). The results suggest that transcript level diversity is the predominant driver of the observed variation of this industrially relevant phenotype (633). Analogously, because many genes involved in stress tolerance may be conserved among strains of a species, strain-specific diversity of this phenotype most likely also depends on the relative expression levels of genes rather than their presence or absence. This notion was recently confirmed using the two model systems described above, i.e., heat and oxidative stress robustness in Lc. lactis MG1363 (593) and intestinal robustness in Lb. plantarum WCFS1 (592, 634). Both studies combined differential fermentation conditions with transcriptome profiling and stress tolerance analyses to identify robustness effector molecules on the basis of transcriptome-phenotype matching (TPM). Fermentation conditions were identified that induced increased or reduced stress tolerance levels compared to those under reference fermentation conditions. For example, aerobic fermentation of Lc. lactis MG1363 stimulated enhanced heat stress tolerance, whereas fermentation at elevated temperatures enhanced the survival of the cultures under oxidative stress conditions (593). Increased NaCl concentrations during fermentation induced a reduction of intestinal stress tolerance in Lb. plantarum WCFS1, while growth at a low pH enhanced the survival when the strain was exposed to stomach conditions (i.e., pH 2.5) (634). TPM using random forest-based correlation analysis allowed identification of transcripts that are quantitatively correlated with enhanced or suppressed stress tolerance, suggesting a role of the encoded functions in the studied phenotypes. For example, elevated expression of the metC-cysK operon, involved in sulfur-containing amino acid metabolism in Lc. lactis MG1363, correlated with increased oxidative stress tolerance. Subsequent experiments using cysteine-free medium, known to enhance metC-cysK expression (635, 636), confirmed a role for these genes and the corresponding metabolic pathway in oxidative stress robustness (593). Similarly, TPM of the Lb. plantarum WCFS1 data sets uncovered several candidate genes whose expression levels were predicted to contribute to GIT survival. This TPM-predicted role could be confirmed for three of the candidate genes by robustness analysis of mutants of these genes. The identified intestinal robustness factors included Pbp2A and a Na⁺/H⁺ antiporter (NapA3) that not only contributed to overall intestinal stress robustness but also played a role in bile acid and osmotic stress tolerance in Lb. plantarum WCFS1. The third robustness gene encoded a transcriptional regulator of the AraC family that was shown to affect the expression of cell envelope-associated genes in Lb. plantarum WCFS1, in particular the expression of a capsular polysaccharide synthesis gene cluster. Thus, all three identified robustness effectors affected cell envelope-associated functions, supporting the importance of this functional category in stress robustness, particularly for intestine-related environmental conditions, such as low pH (stomach) and bile acid exposure (small intestine) (634).