Stress Physiology of Lactic Acid Bacteria

INTRODUCTION

Fermented foods are among the oldest forms of processed foods that have evidently survived into today's modern diet. They are produced during the biotransformation of raw materials into the final product by the action of microorganisms. The vast majority of food biotransformations rely either on ethanol fermentation performed by the yeast Saccharomyces cerevisiae or on lactic acid fermentation performed by a relatively wide range of bacteria called lactic acid bacteria (LAB) (1). LAB were among the first bacteria to be studied because of their involvement in food fermentations and in human health.

In the early days, LAB taxonomy relied on morphological and physiological characteristics. The first technical definition, by Orla-Jensen, recognized LAB as Gram-positive cocci or bacilli that were nonsporulating and nonmotile and had the ability to catabolize sugars mainly into lactic acid (2). These classification criteria led to a broad definition of LAB comprising diverse bacteria. During the 1990s, advances in molecular techniques allowed a more elaborate description of LAB (3, 4). LAB generally have a low GC content (<50 mol%), while some lactobacilli have been reported to reach up to 57 mol% (5). They are Gram-positive, non-spore-forming, microaerophilic or anaerobic bacteria that produce lactic acid as the major end product of sugar fermentation. LAB are typically catalase and cytochrome negative, fastidious, aerotolerant, and acid tolerant. The most common genera of LAB considered to be food related are Lactobacillus, Lactococcus, Streptococcus, Enterococcus, Pediococcus, Leuconostoc, Oenococcus, Tetragenococcus, Carnobacterium, and Weissella. Even though it has been suggested that LAB are a heterogeneous group of bacteria and a universal technical definition may not exist, all the aforementioned genera have been shown to have diverged from a common ancestor. Both 16S rRNA gene and whole-genome phylogenies have revealed that the “core” LAB species form the distinct order Lactobacillales in the class Bacilli of the phylum Firmicutes (3, 4, 6). Based on this observation, a nonphylogenetic approach for defining LAB is rapidly becoming obsolete. For example, bifidobacteria or certain Bacillus species, which exhibit some characteristics in common with LAB, are no longer included in this group sensu stricto. According to the latest review about LAB taxonomy, the order Lactobacillales consists of six families with 38 genera and more than 400 species (7).

LAB starter cultures generate a bacteriostatic or even bactericidal environment for spoilage and pathogenic bacteria by lowering the pH of the food matrix during lactic acid fermentation. LAB fermentation combined with appropriate technological hurdles leads to safe food products with an extended shelf-life. LAB also play an important role in the development of the organoleptic properties of the product. Through their metabolic activities (e.g., lipolysis and proteolysis), LAB produce important aroma and flavor compounds, while they can also contribute to the texture (e.g., by the production of exopolysaccharides [EPS]). The involvement of LAB in food production is far from being unintentional. Food-related LAB are among the very few microorganisms that were domesticated by humans (1, 8). Domestication of LAB started several millennia ago. During this period, LAB genomes were streamlined by genome decay due to adaptation to food environments rich in nutrients (6). This process of reductive evolution resulted in metabolic simplification and in LAB strains with multiple auxotrophies due to the loss of several biosynthetic pathways. On the other hand, gene acquisition events allowed the gain of important technological properties (6). The continuous selective pressure for high-quality products exerted by humans led to the development of today's starters.

However, not all LAB are related to food fermentations. Even a superficial examination of LAB taxonomy reveals that several species are commensals and pathogens. This fact has not always been reflected in the literature. Frequently, food microbiologists and food technologists focused solely on benign LAB, leaving LAB pathogens to clinical microbiologists, and vice versa, irrespective of the underlying phylogenetic relationship. Recent metagenomic data support early observations of LAB being a part of the microbiomes of humans and other animals (9, 10). LAB are known to participate in the intricate balance of the microbiome ecosystem, which can be decisive for both health and disease. Interestingly, in some cases this is achieved by acidification of the environment by LAB in a manner similar to that in food fermentations (11). In addition to their antimicrobial properties, certain LAB can stimulate particular activities of the immune system of the host, prevent diarrheas following antibiotic treatment or viral infections, produce vitamins in situ, or lead to reduced cholesterol levels in blood (12). Thus, several LAB are used as probiotic bacteria, i.e., as “live microorganisms that confer a health benefit on the host when administered in adequate amounts” (13). Note that probiotic LAB strains may originate from food fermentations or might be commensals. Nonetheless, there are several formidable opportunistic LAB pathogens, i.e., commensals that can turn virulent given the right conditions. Such species are found mainly in the Streptococcus and Enterococcus genera (14). Group A streptococci (GAS), group B streptococci (GBS), and Streptococcus pneumoniae can cause invasive and life-threatening infections, while Enterococcus faecalis and Enterococcus faecium have emerged as major causative agents of nosocomial infections (14).

Given the involvement of LAB in food production and health, it is not surprising that they are among the best-studied microorganisms. Over the past 30 to 40 years, the physiology, biochemistry, genetics, and evolution of LAB have been the focus of research in many laboratories around the world. Our knowledge about LAB was revolutionized with the advent of genome sequencing, while recently developed meta-omics technologies allow monitoring of their behavior in complex food and microbiome ecosystems. A field of research that has received much attention early on is the stress physiology of LAB. Like all other microorganisms, LAB are exposed to stressful conditions, but studying the stress responses of the different categories of LAB is important for different reasons.

A key aspect of starter LAB is their robustness during the production and storage of fermented foods. Vulnerability of starters to technological hurdles may influence the fermentation process per se, which can have a serious impact on food quality and safety. Similarly, LAB need to be able to resist technological stresses during preparation of probiotic formulas to maintain high viable counts. After consumption, probiotic LAB must survive the harsh conditions in the gastrointestinal tract (GIT). Commensal LAB have to adapt to the various conditions prevailing in the different niches of the host, while pathogenic LAB need to counteract the host's innate immunity. LAB have also emerged as models for the study of bacterial stress physiology. Available genome sequences indicate that LAB are devoid of a dedicated stress sigma factor, such as σ^{, present in} Bacillus subtilis and many other related Gram-positive bacteria (15). This is a major difference with important implications for gene regulation under stress conditions.

The latest review on the overall stress physiology of LAB was published by van de Guchte et al. in 2002 (16). Here we present a detailed overview of the latest developments in the study of the stress behavior of LAB. After an initial presentation of central concepts of LAB stress physiology developed over the years, we concentrate on the stress defense mechanisms that have been reported to date. These responses are grouped according to their direct participation in preserving cell energy, defending macromolecules, and protecting the cell envelope. Important paradigms of stress-induced responses of probiotic and pathogenic LAB are described separately in order to highlight the relationship of stress to probiotic potential and virulence, respectively. The induction of prophages under a variety of environmental stresses is then discussed. The final part of the review is devoted to systems-based strategies used to characterize the “stressome” of LAB and to engineer LAB strains with improved stress tolerance. Considering the immensity of the relevant literature, only representative and/or key citations are included in this review.

STRESSES, EXPERIMENTAL CONTEXT, AND PHENOTYPES

Common Stresses Encountered by LAB in Their Ecological Niches

LAB 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 Physiology

A 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 Cell

A 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.

SENSING AND SIGNALING STRESSES IN LAB

Bacteria utilize a wide array of sensors to monitor the intracellular and extracellular environments and to regulate the cell physiology to cope with environmental changes (Fig. 1). Signal transduction systems can broadly be divided into two major categories: one-component systems (OCSs) and two-component systems (TCSs) (54). In OCSs, the sensory and output functions are located in the same polypeptide, while they are located on separate polypeptides in TCSs. In addition, other molecules, such as nucleic acids and lipids, can also act as sensors. Some of these, mostly RNA molecules, can elicit a response by themselves, while others transfer signals to protein partners that relay them or elicit the response.

FIG 1

Schematic representation of selected LAB stress sensory systems. HK, histidine kinase; RR, response regulator.

Two-Component Systems

TCSs 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 Systems

Although 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.

Metal stress sensory and signaling mechanisms.

Resistance mechanisms against metals are not well characterized for LAB (reviewed in reference 26), and most available information concerns pathogenic streptococci. In bacteria, three types of metal-sensing transcriptional regulators control gene expression in response to metal ion concentrations: derepressors (ArsR-SmtB, CopY, and CsoR-RcnR families), activators (MerR family), and corepressors (Fur, NikR, and DtxR families) (69).

(i) MarR family members CopY and AdcR/ZitR.

Characterized LAB copper sensors include the CopZ-type copper chaperones and the CopY-type regulators (26). For Enterococcus hirae, it has been proposed that CopZ binds cytoplasmic Cu ^{and transfers it to CopB for transporter-mediated export and to the CopY regulator for signaling (}26) (Fig. 1). CopY is an OCS with an N-terminal DNA-binding domain and a C-terminal metal-binding domain (26). At a low Cu ^{concentration, Zn occupies the metal-binding site of CopY and the protein is bound to its target sequence, thereby inhibiting} cop operon expression. When the Cu ^{concentration increases, Cu-CopZ transfers its copper ion to CopY, displacing Zn from the metal-binding site and resulting in CopY release from the DNA. This allows transcription of the} cop operon (26). The Lc. lactis copper resistance sensing and signaling pathway is possibly similar to that described for E. hirae (26). Lc. lactis CopR, the CopY homolog, controls a regulon of 14 mostly uncharacterized genes (70). Recently, a different copper resistance mechanism was described for S. pneumoniae (71). This organism lacks a CopZ homolog. Instead, the resistance system consists of the membrane-bound Cu chaperone CupA, a copper exporter (CopA), and a CopY-type regulator (71). The copY, cupA, and copA genes are arranged in an operon whose expression was shown to be induced by Cu and repressed by CopY (72). The primary roles of CupA are proposed to be the sequestering of Cu ^{and its transfer to CopA, activities that are essential for copper resistance. If the concentration of Cu exceeds the capacity of CupA, free cytoplasmic Cu may be bound by the repressor CopY, thus relieving repression of} copY, cupA, and copA (71).

Bacteria require zinc, but an excess of zinc has toxic effects. In many bacteria, zinc homeostasis is maintained by the concerted action of pairs of sensors that regulate either the uptake or efflux of Zn ⁽73). Regulation in Lc. lactis of the Zn ^{uptake system ZitSQP is under the control of the repressor ZitR (}74). Purified ZitR is a dimer with up to two zinc ligands per monomer. It specifically binds two intact palindromic operator sites overlapping the −35 and −10 boxes of the zit promoter (75). ZitR requires Zn ^{to bind DNA, and transcriptional analyses have shown that} zitSQP expression is induced only at Zn ^{concentrations below 100 nM, indicating that ZitR acts as a sensor of Zn scarcity (}75). S. pneumoniae encodes a homologous Zn ^{uptake system (AdcABC) that is regulated by a ZitR homolog (AdcR) (}76). AdcR also regulates the expression of the four pneumococcal histidine triad (Pht) proteins, PhtA, PhtB, PhtD, and PhtE (77), involved in surface adhesion, as well as the Zn^{-specific AdcA homolog AdcAII and a Zn-dependent alcohol dehydrogenase (}78, 79). The adc operon is present in most streptococci and has also been studied in Streptococcus gordonii, where it is involved in Mn ^{homeostasis (}80), and in Streptococcus suis, where it regulates the expression of a Zn^{/Mn uptake system, ribosomal genes, and} pht genes (81).

(ii) TetR family sensors and zinc resistance.

TetR proteins constitute a widespread family of OCSs regulating many aspects of bacterial physiology and interacting with a vast array of ligands (68). A small number of TetR family regulators have been characterized for LAB, among which only S. pneumoniae SczA has so far been implicated in a stress response. Resistance against zinc in S. pneumoniae depends mostly on the CzcD transporter, whose expression is under the control of SczA and whose activity is Zn ^{dependent (}82). Despite extensive studies on other TetR regulators, metal-binding TetR regulators are quite uncommon in bacteria and await further characterization (68).

(iii) Regulators of the DtxR/MntR family.

Regulators of the DtxR/MntR family are metal sensors that usually bind Fe ^{or Mn (}69). Cytoplasmic manganese plays a significant role in the protection against oxidative stress in LAB (83). In some streptococci, Mn ^{intake is mediated by the ScaCBA transporter, whose expression is under the control of ScaR, a regulator of the DtxR/MntR family (}84). ScaR represses the expression of scaCBA in the presence of Mn^{. Biochemical studies of the} S. gordonii and S. pneumoniae ScaR proteins have shown that ScaR is a homodimer and contains two metal-binding sites per protomer (85, 86). In S. pneumoniae ScaR, Zn ^{occupies site 1, and although it is required for activation, it keeps ScaR in an inactive state. Activation of DNA-binding activity is accomplished only when Mn occupies the lower-affinity site 2 (}85). Zn ^{can also bind to site 2, resulting in ScaR having a low DNA-binding activity (}85). This effect may partly explain the increase of expression of scaCBA in the presence of toxic levels of Zn ⁽85), although a recent study showed that Zn ^{competitively inhibits Mn uptake and therefore leads to a depletion of intracellular Mn content (}87), thus relieving ScaR repression on scaCBA.

S. mutans SloR regulates the expression of the sloABCR operon, which encodes the Mn ^{and Fe transporter SloABC (}88). SloR represses sloABCR expression only in the presence of Mn ⁽88). Later studies showed that SloR controls a large regulon, acting both as a repressor and as an activator (69, 89). The homologous regulator MtsR of S. pyogenes also controls a large regulon that includes the genes encoding the Mn ^{and Fe transporter MtsABC (}90, 91). MtsR also represses the dnaK operon, suggesting that it mediates the response of dnaK to heat shock (90).

(iv) PerR, a regulator of the peroxide stress response.

Although the oxidative stress response has been researched extensively in LAB, the regulatory mechanisms involved are still largely unknown (92). One of the best-characterized oxidative stress sensors in LAB is PerR, a regulator of the Fur family. Although most members of this family are involved in metal homeostasis, PerR is a sensor of H₂O₂ (93). This has been documented for E. faecalis (94) and a number of streptococcal strains, showing that PerR links peroxide resistance and metal homeostasis in these organisms (95). S. pyogenes PerR regulates the expression of a ferritin-like protein of the MrgA/Dps family (Dpr) and PmtA, a CPx-type metal transporter (95, 96). Dpr binds Fe ^{as well as Co, Cu, Mn, Ni, and Zn (}97), and it plays a key role in aerobic growth of S. mutans (98) and S. suis (99). PmtA is a putative Zn ^{efflux transporter, and its overexpression resulted in a strong induction of the AdcR-regulated genes (see the section on the MarR family, above) (}95). Evidence suggests that PerR regulons may vary in different streptococcal species. For example, expression of the Mn ^{uptake system MntABC is under the control of PerR in} Streptococcus oligofermentans (100).

Serine/threonine/tyrosine kinases.

Regulation of protein activity by serine/threonine/tyrosine phosphorylation was first described for eukaryotes but was later also shown to play a central role in bacteria (101). Bacterial genomes contain eukaryotic-type kinases that are mainly responsible for serine and threonine phosphorylation. Phosphorylation at Ser, Thr, or Tyr residues is not as labile as phosphorylation at His or Asp. Therefore, Ser/Thr/Tyr kinases usually are associated with cognate phosphatases in order to quench signaling cascades (102).

Ser/Thr/Tyr kinases are scarce in LAB, although most of their genomes encode at least one kinase/phosphatase pair (103). A number of these have been characterized for some streptococci and E. faecalis. The S. pneumoniae StkP Ser/Thr kinase and PhpP phosphatase are involved in stress responses, among other physiological processes (104) (Fig. 1). StkP belongs to a conserved group of membrane-anchored Ser/Thr kinases that consist of a cytoplasmic kinase domain and an extracellular C-terminal region composed of several penicillin-binding and Ser/Thr kinase-associated (PASTA) domains. PASTA domains have been proposed to bind peptidoglycan (PG) fragments, which thereby act as signaling molecules (105). Indeed, S. pneumoniae StkP can bind PG subunits and β-lactam antibiotics (106). A recent study revealed that StkP requires the GspB protein for proper localization in the cell septum and for autophosphorylation and subsequent phosphorylation of its substrates (107). The StkP/PhpP pair regulates the activity of the phosphoglucosamine mutase GlmM (108), the cell division protein DivIVA (109), and the Mn^{-dependent inorganic pyrophosphatase PpaC (}109), among others. Interestingly, StkP phosphorylates the orphan RR RitR at the DNA-binding domain in an in vitro assay (110). RitR plays a key role in the response to iron and oxidative stresses by regulating the expression of the peroxide resistance protein Dpr and of iron uptake systems (111). The formation of a complex between DNA-RitR and the StkP phosphatase PhpP contributes to the regulation of RitR activity via a mechanism that remains unclear, to date (110).

The E. faecalis kinase/phosphatase pair IreK/IreP is involved in intrinsic resistance against cephalosporin, as evidenced by the fact that mutants lacking ireK exhibit cephalosporin susceptibility, whereas mutants lacking ireP are hyperresistant (112). A subsequent study suggested that IreK/IreP regulates the phosphorylation state of the IreB protein, which would control a cephalosporin resistance pathway through an undetermined mechanism (113).

Thermosensors in LAB

Bacteria 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 Sensor

Accumulation 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 Sensor

The 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 LAB

Several 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 Sensing

Bacterial 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.

PERTURBATIONS OF METABOLISM AND METABOLIC ADAPTATIONS OF LAB UNDER STRESS CONDITIONS

LAB are subjected to marked metabolic perturbations under environmental stress conditions (Fig. 2 and ​and3).3). As a consequence of stress, cells lower their metabolic activities, which decreases energy production and the generation of a proton motive force (PMF) and alters growth and viability (17, 158). Overall, the responses consist of the selection of alternative fates of pyruvate, the utilization of other carbon sources, the activation of the proteolytic system, and/or the catabolism of free amino acids (FAA) by cells. Metabolic adaptation is crucial for survival because it stimulates the production of additional energy and lowers the stress level, e.g., through alkalization of the cytosol under acidic conditions (159). The extent of metabolic perturbation and the main routes to reprogram pathways responsible for substrate catabolism in order to adapt to environmental stresses vary between LAB species (160).

FIG 2

Schematic representation of changes of the carbohydrate metabolism, glycolysis, and fate of pyruvate in lactic acid bacteria. Colored arrows and enzymes indicate common reactions (black), those mainly induced during fermentation by unstressed cells (blue), and those induced in respiratory and/or environmentally stressed cells (red). LDH, lactate dehydrogenase; ACK, acetate kinase; POX, pyruvate oxidase; PFL, pyruvate formate lyase; PdhABCD, pyruvate dehydrogenase complex; α-ASL, α-acetolactate synthase; AdhE, alcohol dehydrogenase; NOX, NADH oxidase; α-ALD, α-acetolactate decarboxylase; DAR, diacetyl reductase.

FIG 3

Schematic representation of the main free amino acid pathways of lactic acid bacteria induced under acid stress and/or starvation conditions. GABA, γ-aminobutyric acid; ADI, arginine deiminase; cOTC, catabolic ornithine transcarbamoylase; CK, carbamate kinase; AguA, agmatine deiminase; AguB, putrescine carbamoyl transferase; AguC, carbamate kinase; AguD, agmatine/putrescine antiporter; GAD, glutamate decarboxylase; HDC, histidine decarboxylase; AspD, aspartic acid decarboxylase; AspT, aspartate-alanine antiporter; Bcat, α-ketoglutarate-dependent branched-chain aminotransferase; HycDH, hydroxyacid dehydrogenase; KaDH, keto acid dehydrogenase; KDC, 2-keto acid decarboxylase; ADH, alcohol dehydrogenase; AIDH, aldehyde dehydrogenase; PTAC, phosphotransacylase; ACK, acetate kinase.

Metabolism of Carbon Sources and Energy Production

Under 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.

Transport and fermentation pathways of carbohydrates.

LAB express numerous proteins responsible for carbohydrate transport and utilization. Proton-coupled active transport by proteins from the major facilitator superfamily (MFS), the glycoside-pentoside-hexuronide (GPH) superfamily, and the ATP-binding cassette (ABC) superfamily and by group translocators, such as the phosphotransferase system–glucose-fructose-lactose (PTS-GFL) superfamily, is used most frequently (162). LAB modulate the synthesis of specific transporters depending on the type of carbohydrate available (163, 164). Under acid stress conditions, Lb. casei and S. mutans markedly decreased the synthesis of the phosphoenolpyruvate phosphotransferase system (PEP-PTS) for glucose, which is the primary carbohydrate transport system belonging to the PTS-GFL superfamily (165,–167). Under the same conditions, an acid-resistant mutant of Lb. casei showed the highest level of PEP-PTS and of the phosphocarrier protein (HPr). High levels of PEP-PTS may improve the acid resistance. The glucose PTS is upregulated at low pH in Streptococcus sobrinus (168) and Streptococcus macedonicus (34, 36). Under optimal conditions and when cells are growing in glucose-rich media, HPr inhibits PEP-PTSs for carbohydrates other than glucose, preventing their transport into the cell. In addition, glycolytic intermediates (e.g., fructose-1,6-bisphosphate) activate the phosphorylation of HPr at the serine residue at position 46 (169). The resulting P-Ser-HPr interacts with the global transcriptional regulator CcpA, and the complex prevents the catabolism of carbon sources other than glucose, the preferred sugar in most bacteria, by binding to the catabolic repression element (cre) upstream of the responsible genes and shutting them down. Under low-pH conditions, some LAB increase pyruvate kinase activity, which may accelerate the depletion of fructose-1,6-bisphosphate, relieving CcpA repression and allowing the use of alternative carbon sources (170).

Acid stress causes intracellular acidification, which decreases the activity of cytoplasmic enzymes (17). Transcriptomic and proteomic studies have highlighted that many LAB enhance the levels of glycolytic enzymes under acid, thermal, and osmotic stresses, but without increasing the synthesis of lactic acid (171, 172). LAB such as Lactobacillus plantarum, Lactobacillus reuteri, Lactobacillus rhamnosus, and Lc. lactis modify pyruvate metabolism at the expense of lactic acid, and they increase the synthesis of basic compounds (e.g., lysine and diacetyl/acetoin) (173, 174), energy-rich intermediates (such as ATP and NADH) (175), EPS, and/or glycogen (176). The level of lactate dehydrogenase (Ldh), which is responsible for the synthesis of lactic acid from pyruvate, markedly decreases. Pyruvate oxidase and phosphate acetyltransferase, used to synthesize acetyl-coenzyme A (acetyl-CoA), are induced in Lactobacillus delbrueckii subsp. bulgaricus and Lb. rhamnosus under acid stress conditions (161, 170). Acetyl-CoA is rerouted toward the biosynthesis of fatty acids instead of butanoate (161, 170), which may enhance the rigidity and impermeability of the cytoplasmic membrane (177, 178). Changes in pyruvate metabolism were also observed in S. mutans under acid stress (166, 167).

Metabolic adaptations in the presence of oxygen and ROS.

Despite having a fermentative metabolism, several LAB species possess genes that code for a respiratory electron transport chain (179). They harbor the cydABCD operon, encoding a heme-dependent cytochrome with the capacity to generate PMF. Compared to when they are fermenting, LAB display an increase of extracellular pH and biomass under energetically favorable respiration conditions, that is, in the presence of oxygen, heme, and menaquinone (179,–181) (Fig. 2). Lc. lactis growing anaerobically displays a fermentative metabolism in which carbohydrates are converted into (mainly) lactic acid (182). Two molecules of NADH, produced from the oxidation of glyceraldehyde-3-phosphate, are reoxidized through the action of Ldh. The NADH/NAD ^{ratio plays a key role in} Lc. lactis in controlling the shift from homolactic to mixed-acid fermentation (183). Under oxygenic conditions, enhanced expression and activities of NADH oxidase and NADH peroxidase compete with Ldh for NADH molecules (184). Thus, the production of lactic acid is reduced and the glycolytic flux is redirected toward production of acetate, ethanol, acetoin, diacetyl, and CO₂. The consumption of oxygen reduces its cytoplasmic concentration and limits the oxidative stress. Hydrogen peroxide is produced under these circumstances, which may affect the growth rate of Lc. lactis and may ultimately result in cell death. The regulation of respiration in Lc. lactis is mediated via CcpA (185). In glucose-rich media, CcpA represses heme intake and cells grow through fermentation, synthesizing only lactic acid (Fig. 2). As glucose is consumed, heme intake occurs and Lc. lactis respires pyruvate and/or lactic acid. The shift toward the respiratory pathway produces metabolic changes, which mainly involve pyruvate oxidase (POX) activity for converting pyruvate into acetyl-P. The latter compound is used via acetate kinase (ACK), producing acetic acid and ATP. The respiratory metabolism of Lc. lactis results in the irreversible inactivation of pyruvate formate lyase (PFL), impeding the conversion of pyruvate into acetyl-CoA. The level of alcohol dehydrogenase (AdhE), which uses NADH to convert acetyl-CoA into acetaldehyde and ethanol, is reduced. Under such conditions, the pyruvate dehydrogenase complex (PdhABCD), which catalyzes the conversion of pyruvate into acetyl-CoA and NADH, is activated. The combined activity of POX and PdhABCD leads to a decrease of pyruvate available for Ldh. LAB decrease the synthesis of lactic acid and ethanol under oxidative stress. At low pH and a high concentration of carbohydrates, LAB decrease the synthesis of lactic acid, and the excess of pyruvate is metabolized via α-acetolactate synthase (α-ASL) to form α-acetolactate. Under aerobic conditions, α-acetolactate is nonenzymatically decarboxylated to diacetyl. Under limiting oxygen conditions, α-acetolactate is decarboxylated into acetoin via α-acetolactate decarboxylase (α-ALD). Diacetyl reductase (DAR) converts acetoin into diacetyl (186). Depending on the cellular redox state, DAR may also catalyze the reduction of acetoin into butanediol.

Carbohydrate starvation.

Under carbohydrate starvation conditions, LAB such as Lc. lactis lose the ability to form colonies, and they enter into the VBNC state for at least 2 weeks (187). They shift the physiological state by lowering the synthesis of DNA and proteins and achieving the stationary phase of growth. At the beginning of starvation, cells accumulate glycolytic intermediates (PEP, 3-phosphoglycerate, and 2-phosphoglycerate). Further, glycolytic intermediates are metabolized to pyruvate and ATP, and the levels of PEP-PTSs markedly decrease. Under lactose starvation conditions, Lb. casei and Lb. rhamnosus also regulated their glycolytic enzymes differently (176, 188). S. mutans modulated its metabolic activities through transcriptional and enzyme allosteric regulation in order to optimize the flow of carbohydrates and to maintain an optimal energy state (189). An increase of the catabolism of carbon sources (e.g., inositol, glycerol, RNA, lipids, proteins, peptides, and especially FAA) was also seen when LAB were faced with carbohydrate starvation (188). This mechanism allows cells to survive without carbohydrates (187). Under long-term starvation conditions, repression of PEP-PTS transporters for carbohydrates and of glycolytic enzymes is mainly due to derepression of carbon catabolite repression (CCR) mediated by CcpA.

Malolactic fermentation pathway.

Species belonging to the Oenococcus, Pediococcus, Lactococcus, Lactobacillus, and Leuconostoc genera and oral Streptococcus species (190) use the MLF pathway to enhance survival under environmental stress conditions, such as low pH, a high concentration of carbohydrates, or starvation. This pathway involves the decarboxylation of l-malic acid to l-lactic acid and CO₂. The regulatory protein (MleR), malate permease (MleP), and malolactic enzyme (MleS) are the components of the route. The uptake of the dianionic malic acid is coupled to the exit of its decarboxylation monoionic product, lactic acid (malate/lactate antiporter system). Cytosolic H ^{ions are consumed during malic acid decarboxylation, which promotes Δψ and ΔpH gradients over the cell membrane. Therefore, MLF increases the cytoplasmic pH and PMF, which helps the cell to take up other nutrients (}158). The PMF formed via MLF is sufficient to drive ATP synthesis through F-ATPase (F₁F_o-H ^{ATPase or ATP synthase). The CO}₂ produced by MLF is partially used by carbonic anhydrase to form bicarbonate, further increasing the cytosolic pH. Under acidic conditions and/or osmotic stress conditions, Lb. plantarum decreases glucose intake and uses malic acid as a preferred energy source (164, 191, 192). MLF has also a protective role for S. mutans against oxidative stress and starvation (193). Malate and low pH represent inducing signals for the MLF pathway, except in O. oeni.

Metabolism of citrate.

Under acid stress conditions, lactococci and thermophilic lactobacilli metabolize citrate, producing pyruvate. Citrate metabolism requires the following three steps: (i) citrate intake via permease (CitP), (ii) formation of oxaloacetate from pyruvate through the activity of citrate lyase (CL), and (iii) decarboxylation of oxaloacetate into pyruvate via oxaloacetate decarboxylase (OAD). The end products are acetic acid, CO₂, and pyruvate. As described above, the environmental conditions determine the fate of pyruvate. Because of the increased pH in the cytoplasm and extracellularly, LAB take advantage of the cofermentation of citrate with other carbohydrates, such as lactose or glucose. ^{C-assisted nuclear magnetic resonance (NMR) analysis showed that} Lc. lactis possesses a citrate/lactate antiporter (194). As described for the MLF pathway, transport and metabolism of citrate increase the PMF due to the generation of Δψ and ΔpH gradients over the cytoplasmic membrane. The main protective effects of cofermentation differ among LAB. In the presence of glucose and citrate, homofermentative and facultatively heterofermentative species (e.g., Lactococcus sp., Enterococcus sp., and Lb. plantarum) use the glycolytic pathway to form 2 mol each of lactic acid and ATP for every 1 mol of glucose consumed. Each mole of citrate produces pyruvate and, subsequently, acetoin, butanediol, and diacetyl, which protect the cell against acid stress. Furthermore, Lc. lactis and Lb. rhamnosus metabolize citrate and synthesize acetic acid and ATP under carbohydrate starvation conditions (176, 195). The cofermentation of citrate and glucose in obligately heterofermentative LAB follows the pentose phosphate pathway, mainly producing lactic acid, ethanol/acetic acid, and ATP from the consumed glucose. One additional mole each of acetic acid and ATP is synthetized for each mole of citrate metabolized.

Metabolism of Nitrogen

The 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.

The proteolytic system.

Genomes of LAB encode extracellular or cell wall-associated proteinases that hydrolyze environmental proteins into oligopeptides, as well as specific transport systems to take up oligo-, di-, and tripeptides and FAA from the environment. Intracellular oligopeptidases and peptidases hydrolyze oligopeptides and peptides into smaller peptides and FAA. LAB modify proteolysis as a response to environmental stresses. Under acidic conditions, they reduce the abundance of amino acid transporters, and FAA become limiting. Consequently, many lactobacilli increase the activities of peptidases, such as PepN and PepO (161, 165, 170, 171). The proteolytic system of LAB is also influenced by carbon metabolism (196). Lc. lactis increased the levels of endo- and exopeptidases (PepO and PepC) during growth under respiratory compared to fermentative conditions (181). VBNC cells of Lc. lactis increased the expression of pepXP, pepDA, and pepDB, while pepQ and pepV were repressed in comparison to their expression in exponentially growing cells. Much more needs to be uncovered regarding the mechanisms of carbon and nitrogen coregulation. CcpA forms complexes with CodY in B. subtilis, and it would be very interesting to see whether such a regulatory mechanism also occurs in LAB (196). Recently, it was shown that the aminotransferase IlvE of S. mutans is regulated by both CcpA and CodY under acid stress conditions, supporting the hypothesis of carbon and nitrogen coregulation in this species (198).

Metabolism of FAA.

The metabolism of FAA not only contributes to flavor development in fermented foods but also is crucial for LAB stress resistance. It allows the production of precursors for the biosynthesis of amino acids, fatty acids, nucleotides, and vitamins, generates energy under starvation conditions, and increases the intracellular pH (Fig. 3). Deamination and decarboxylation of FAA are the main routes that enhance LAB growth and survival under environmental stress conditions. Deamination leads to the synthesis of α-keto acids and NH₃, the latter of which increases the intra- and extracellular pH, protecting cells against acid stress. α-Keto acids are used to synthesize other amino acids for anabolic purposes or are catabolized into ketones, aldehydes, alcohols, or acids to improve the cellular redox potential and/or to synthesize ATP. Amino acid decarboxylation leads to the conversion of an amino acid into the corresponding amine and CO₂. This reaction is coupled to the exchange between substrate and product (e.g., glutamic acid/γ-aminobutyric acid [GABA] or aspartic acid/alanine), which is mediated by an antiporter. The translocation is driven by the opposite gradients of substrate and product over the membrane, which are maintained by decarboxylation. These combined activities generate a PMF because proton consumption during decarboxylation leads to an increase of the intracellular pH (199). Therefore, the antiporter/decarboxylase system is an indirect proton pump analogous to other mechanisms providing energy and protecting from acidic conditions (200). Several LAB genomes (e.g., that of Lactobacillus acidophilus) encode amino acid decarboxylases, cation transport ATPases, and chaperones involved in pH regulation at the cytoplasmic level (201). When they are in the VBNC and/or carbohydrate starvation state, LAB metabolize BCAAs to synthesize ATP and BCFAs (184). The main FAA pathways induced under environmental stress responses are discussed in the following paragraphs.

(i) ADI pathway.

The arginine deiminase (ADI) pathway consists of three enzymes, namely, arginine deiminase (ADI; encoded by the arcA gene), catabolic ornithine transcarbamoylase (cOTC; encoded by arcB), and carbamate kinase (CK; encoded by arcC), and one membrane transport protein (encoded by arcD) (Fig. 3). All genetic components of this pathway are present in LAB genomes (202). Arginine is degraded through the ADI pathway into ornithine, NH₃, and CO₂. Arginine enters the cytoplasm via an arginine/ornithine antiporter system, without energetic cost. As shown for several LAB species isolated from sourdough, cheese, meat, and wine, this pathway is induced by arginine and/or environmental stresses, such as acidity and starvation (203). Nutrient sources (e.g., carbohydrates), growth rate, pH, and arginine all affect the ADI pathway of S. gordonii (204). The ADI pathway of Streptococcus rattus is subject to substrate induction and catabolite repression (205). The NH₃ produced alkalizes the cytoplasm, thus increasing acid resistance (203, 206). In addition, the extra ATP produced via the ADI pathway enhances cell survival after carbohydrate depletion (202), while the intermediate carbamoyl phosphate is used to synthesize pyrimidines. ADI pathway regulation is connected to carbon metabolism via CCR. The intergenic regions of the arcABC genes have several cis-regulatory element sites, which is suggestive of direct regulation of the ADI pathway through CcpA (202, 207). Members of the CRP/FNR family of transcriptional regulators also seem to exert a positive regulation. The arcABC genes of Lc. lactis are repressed under aerobic and respiratory conditions (180).

(ii) AgDI pathway.

The agmatine deiminase (AgDI) pathway, analogous to the ADI system, consists of three enzymes, i.e., putrescine carbamoyl transferase (AguB), agmatine deiminase (AguA), and carbamate kinase (AguC), and one agmatine/putrescine antiporter (AguD) (Fig. 3). All components are encoded by the single operon aguRBDAC, which has been studied in S. mutans, Lactobacillus brevis, Lactobacillus curvatus, E. faecalis, Lc. lactis, and S. mutans (208,–210). This pathway allows the hydrolysis of agmatine into putrescine, NH₃, CO₂, and ATP and is a response mechanism to acid and starvation stresses. It represents a potential health risk for humans due to the synthesis of putrescine, a biogenic amine. Like the ADI pathway, the AgDI pathway is induced by the substrate (agmatine) and regulated via CCR through mechanisms that are dependent or independent of CcpA (208, 210). The AgDI pathway of S. mutans was also induced in response to heat stress (209).

(iii) GAD pathway.

The glutamate decarboxylase (GAD) pathway consists of GAD, a pyridoxal 5′-phosphate (PLP)-dependent enzyme, and a glutamate/GABA antiporter (Fig. 3). The GAD pathway is encoded by gadB and gadC and is regulated by an activator, GadR (206). GAD catalyzes the decarboxylation of glutamate into GABA, a bioactive compound that acts as a neurotransmitter in humans. The uptake of glutamate and the expulsion of GABA are mediated by the electrogenic antiporter, which, together with the decarboxylation of glutamate, produces PMF. Three decarboxylation-antiporter cycles are needed to synthesize 1 mol of ATP. The cytoplasmic pH increases due to the removal of H ^{ions. Concomitantly, the extracellular pH increases slightly due to the exchange between glutamate and the more alkaline compound GABA. Glutamate and/or environmental stresses (e.g., acidity, osmotic stress, or starvation) induce the GAD pathway in the dairy starters} Lc. lactis, S. thermophilus, and Lb. delbrueckii subsp. bulgaricus, in nonstarter lactic acid bacteria (NSLAB), such as Lb. brevis, Lactobacillus paracasei, and Lb. plantarum, and in lactobacilli isolated from human and animal intestines (211, 212).

(iv) AspD pathway.

The aspartic acid decarboxylase (AspD) pathway catalyzes the β-decarboxylation of aspartic acid with the synthesis of alanine and CO₂ (Fig. 3). The uptake of aspartic acid and the expulsion of alanine are mediated by an electrogenic antiporter, which, together with the decarboxylation reaction, produces PMF (212). This pathway has been described for Lactobacillus sp. M3 (199), Lactobacillus buchneri, and Lb. acidophilus (201). Other LAB, including Lb. plantarum WCFS1 and E. faecalis V583, encode putative aspartic acid decarboxylases (213, 214). The exact mechanism of regulation of the AspD pathway is presently unknown.

(v) HDC.

Histidine decarboxylase (HDC) catalyzes the β-decarboxylation of histidine into histamine and CO₂ (Fig. 3). Genes encoding HDC (hdcA) have been characterized for Lactobacillus sp. 30a, Lb. buchneri B301, and Lactobacillus hilgardii (215,–217). In these organisms, hdcA forms an operon with a downstream gene (hdcB) of unknown function. Lb. buchneri and Lb. hilgardii harbor genes encoding a histidine/histamine antiporter and a histidine aminoacyl-tRNA synthase (hisS), which are located up- and downstream of the hdcAB operon, respectively. HDC has a dual role in protecting cells against acid stress and generating a PMF. HDC activity is regulated via the cytosolic pH. The enzyme is active at low pH, but a neutral or alkaline pH disrupts its substrate-binding site. The survival of Lc. lactis and S. thermophilus is increased under acid stress conditions when the glycolytic pathway is coupled to HDC activity (218).

(vi) Catabolism of BCAAs in VBNC cells.

During starvation, LAB increase the catabolism of the BCAAs isoleucine, leucine, and valine through the activity of α-ketoglutarate-dependent branched-chain aminotransferase (BcaT) (Fig. 3). The presence and activity of BcaT vary greatly among LAB. VBNC cells of Lc. lactis metabolize leucine into isobutyric, propionic, and acetic acids. BCAAs are imported via PMF-dependent mechanisms that generate ATP during short (several days)- and long (several years)-term starvation (184). BcaT and aromatic amino acid aminotransferases (AraT) play a pivotal role in the regulation of the proteolytic system of Lc. lactis, as they control the intracellular pool of BCAAs, which in turn affects CodY activity. BCAAs directly interact with CodY and enhance its affinity for target genes (219). BcaT is repressed by glucose, which confirms that a link exists between carbon and nitrogen regulation.

The urease system.

The hydrolysis of urea enhances survival of LAB under acid stress conditions. Among LAB, the urease system has been studied mainly in S. thermophilus, Streptococcus salivarius, Lactobacillus reuteri, and Lactobacillus fermentum (220,–222). The urease system is quite complex, as it is encoded by ureI followed by structural (ureABC) and accessory (ureEFGD) genes (223). The urease system produces NH₃ and CO₂ from urea, thus protecting cells against acid stress. CO₂ serves in several anaplerotic reactions that are responsible for the biosynthesis of amino acids and nucleic acids. The urease operon of the oral LAB S. salivarius is positively regulated by urea, low pH, and an excess of carbohydrates. The urease system is induced when S. salivarius is growing in a biofilm, which helps to modulate the pH of dental plaque. The expression of the urease operon in S. salivarius 57.I is mediated via CodY (224).

Accumulation of polyphosphate in LAB.

Inorganic polyphosphate (poly-P) is a polymer of phosphoanhydride-linked phosphate residues that can be found as chains of a few to thousands of residues in living organisms (225). In bacteria, the accumulation of poly-P is dependent on the activity of polyphosphate kinase (Ppk; EC 2.7.4.1), which catalyzes the ATP-dependent formation of a phosphoanhydride bond between a poly-P chain and orthophosphate (226). The hydrolysis of poly-P is catalyzed by exopolyphosphatases (Ppx; EC 3.6.1.11) and the guanosine pentaphosphate phosphohydrolase (GppA; EC 3.6.1.40). Lactobacillus organisms, such as Lb. plantarum WCFS1, Lb. rhamnosus GG, Lb. casei BL23 and ATCC 334, and Lb. reuteri DSM20016, contain all three enzymes (Ppk, Ppx, and GppA). The ability of lactobacilli to accumulate poly-P is correlated with the presence of ppk genes in the strains and the availability of phosphate in the medium (227, 228). However, Lb. plantarum strains containing ppk genes also accumulate poly-P in media with low phosphate levels (228). Poly-P has a protective role against oxidative stress in bacteria, where it acts as a chaperone preventing the aggregation of damaged proteins (225). Poly-P protects some LAB, such as Lb. plantarum lacking superoxide dismutase, from toxic oxygen-derived compounds (O₂^{; free radicals from the Fenton reaction) and improves the expulsion of toxic reactive metals (Fe and Cu). The synthesis of poly-P in} Escherichia coli is regulated by hypochlorous acid (229). Alcántara and coworkers (228) showed that disruption of the Lb. caseippk gene resulted in a lack of synthesis and intracellular accumulation of poly-P. Lb. casei BL23 Δppk exhibited decreased growth under high-salt or low-pH conditions and increased sensitivity to oxidative stress compared to those of the wild type (228). Poly-P-accumulating LAB (Lb. casei and Lb. fermentum) have also been isolated from natural whey starters, and a positive role has been postulated for poly-P in acid stress resistance of LAB strains during mozzarella cheese production (230). Poly-P is also involved in the regulation of general stress resistance pathways in different bacteria (225), but more research is required to fully understand its role in LAB (228).

PROTECTION OF MACROMOLECULES IN LAB

Preventing Macromolecules from Being Damaged

Some 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.

FIG 4

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

The F-ATPase proton pump.

Among the many mechanisms that LAB species employ to counteract stress damage, the F-ATPase system deserves attention as a system that protects against protons. The F-ATPase acts in parallel to the above-described metabolic adaptations that result in alkalization of the intracellular and extracellular environments. F-ATPase is a multimeric enzyme that in Gram-positive facultative anaerobes hydrolyzes ATP synthetized during carbohydrate fermentation or FAA catabolism, generating a PMF (Fig. 4). F-ATPase is a reversible enzyme that can also synthesize ATP by using protons that flow from the environment into the cell. Under acidic conditions, LAB use F-ATPase as a proton pump to maintain intracellular pH homeostasis. This reaction requires ATP because the expulsion of protons from the cytoplasm to the outside environment takes place against an increasing proton gradient that is a consequence of the synthesis of lactic acid during carbohydrate fermentation (231). The level of F-ATPase activity depends on proton transport demand, substrate catabolism, and availability of ATP. The F-ATPase systems of Lb. casei and Lb. plantarum have optimum activities at pH values (pH 5.0 to 5.5) lower than those (pH 7.0 to 7.5) of S. thermophilus and Lc. lactis (232). Distinctly aciduric oral streptococci, such as S. mutans GS-5 and E. hirae ATCC 9790, possess an F-ATPase with a lower optimal pH than that of the more acid-sensitive oral streptococci (233). Interestingly, LAB species or strains that possess an F-ATPase with an optimal activity at low pH are highly protected against acid stress. The capacity to pump protons out of the cell under acidic conditions protects important glycolytic enzymes and allows long-term survival (234). Lactobacilli increase the synthesis of one or more of their ATP synthase components (e.g., AtpB, AtpA, AtpG, and AtpD) under acidic conditions. The F-ATPase complex not only is involved in the maintenance of cytoplasmic pH homeostasis but also responds to bile salt exposure (235,–237). It was noticed that in Lb. casei BL23 (235) as well as in Lb. rhamnosus GG (236), the exposure of cells to bile salt caused an enhanced abundance of F-ATPase, though this may have been a direct consequence of PMF dissipation, as reported for other LAB treated in the same manner (238).

Detoxification of ROS in LAB.

Oxygen per se does not damage cells, but during the operation of metabolic pathways it is partially reduced to water, leading to the synthesis of ROS, including superoxide anion radicals, hydroxyl radicals, and H₂O₂. These display a considerable oxidizing potential and are highly toxic to cells (239, 240). ROS toxicity is also exerted through chemical modification and/or inactivation of most types of biomolecules (184). As detailed above, Lc. lactis can use oxygen under certain growth conditions (179). Also, H₂O₂ may be formed during mixed-acid fermentation as a consequence of the activity of NADH oxidases (241). Superoxide anion as well as hydroxyl radicals are produced during the Fenton reaction (242, 243). Metal ions can catalyze certain reactions leading to the formation of ROS, which is in fact considered one of the underlying mechanisms of metal-induced stress. Iron, copper, and other redox-active metal ions exert their effects by stimulating the Fenton reaction.

(i) Cellular damage produced by ROS.

At the molecular level, ROS can react with macromolecules, such as proteins, lipids, and nucleic acids. Superoxide anions have a limited oxidizing potential and may attack compounds such as polyphenols, ascorbate, and catecholamines (239). H₂O₂ can inactivate proteins through the oxidation of cysteinyl residues (240) or can react with cations, such as Fe ^{and Cu, producing hydroxyl radicals by the Fenton reaction (}239, 244). The damage caused by the hydroxyl radical, which is the most destructive oxidative agent, includes strand breakage and chemical modification of DNA (239, 244).

(ii) ROS resistome.

Bacteria such as Lc. lactis, E. coli, and B. subtilis are able to cope with oxidative stress through a number of enzymes that counteract the negative effects of oxidation (245). Notably, induction of genes encoding ROS-protective enzymes is growth phase dependent and confers multistress resistance (184, 246). Among the main players in LAB that help to counteract the deleterious effects of oxidative stress are (i) NADH oxidase/NADH peroxidase, (ii) superoxide dismutase (SOD), (iii) cytochrome d oxidase, (iv) catalase in the presence of necessary cofactors, and (v) nonenzymatic dismutation of H₂O₂ by Mn^.

The most conserved oxidative resistance mechanism in LAB results from the coupling of NADH oxidase and NADH peroxidase (184). Oxygen is initially used for the oxidation of NADH to NAD ^{via the NADH oxidase, a reaction that produces H}₂O₂. H₂O₂ is further reduced to water by the NADH peroxidase. NADH peroxidase activity is rather low in Lc. lactis, and the detoxification of H₂O₂ is incomplete. It was recently proposed that metabolically synthetized pyruvate reacts nonenzymatically with residual H₂O₂ to produce water (247). Multiple enzymes with peroxidase activity have been described for LAB, such as AhpCF, Npr, and Tpx of E. faecalis and AhpCF and Tpx of S. mutans (248).

Two distinct NADH oxidases have been identified in S. mutans. They catalyze either the two-electron reduction of oxygen to H₂O₂ (performed by Nox1) or the four-electron reduction of oxygen to water (AhpCF) and thereby provide a very efficient enzymatic defense against oxidative stress (249). S. thermophilus, on the other hand, does not appear to encode a Nox1 enzyme but uses a single NADH oxidase to reduce oxygen directly to water (250).

Another key player in resistance to oxidative stress is SOD, which eliminates superoxide anion radicals. Bacteria possess different types of SOD enzymes depending on the bivalent metal cofactor with which the enzyme is coupled, i.e., Fe^{, Mn, and/or Cu-Zn. LAB produce mainly Mn-binding SODs (}251). Analysis of an Lc. lactissod mutant clearly established that SOD protects against superoxide anion radicals (251). An alternative strategy to counteract superoxide anion radicals is by the production of high levels of intracellular glutathione (252). Various enzymes that are known to counter oxidative stress, namely, MsrAB, AhpCF, Tpx, Npr, KatA, Sod, and HypR, have been characterized for E. faecalis (94, 248, 253,–255).

E. faecalis possesses a respiratory chain that produces extracellular superoxide, whose production level depends on the availability of hematin as a cofactor for cytochrome bd or on fumarate as a terminal electron acceptor (256, 257). When this pathway is impaired by restricting access to these nutrients, extracellular superoxide is generated by partially reduced demethylmenaquinones upon exposure to oxygen (257). Thus, E. faecalis necessarily expresses multiple antioxidant defense mechanisms, such as NADH peroxidase and manganese-containing SOD (MnSOD).

Bacteria can also cope with oxidative stress through catalase, an enzyme catalyzing the detoxification of H₂O₂. The following three classes of bacterial catalases are currently recognized: monofunctional catalases, catalase-peroxidases, and manganese catalases (pseudocatalases) (258). The monofunctional catalases and the catalase-peroxidases contain heme as a prosthetic group. The monofunctional catalase specified by the genome of E. faecalis has been shown to remove H₂O₂ (259).

In order to retain the low intracellular redox potential required to keep proteins in their reduced form, the presence of glutathione as well as the thioredoxin systems is crucial (260). Members of the genus Streptococcus contain glutathione. They either synthesize glutathione or import it from the medium (261). In addition, streptococci encode a glutathione reductase, which catalyzes the reduction of glutathione disulfide that is produced upon oxidation of glutathione by NADPH (262). The glutathione reductase gene in S. thermophilus is induced in the presence of oxygen (262). The S. thermophilus genome also contains genes encoding two thioredoxins, a thioredoxin reductase, and a single glutaredoxin. This relatively large genetic repertoire emphasizes the importance of such enzymes in oxidative stress resistance (for a review, see reference 250). E. faecalis is able to synthesize glutathione and expresses glutathione reductase (261, 263).

Adaptation is also part of the survival strategy of LAB to cope with oxidative stress. Upon exposure to sublethal concentrations of H₂O₂, Lc. lactis can survive in the presence of otherwise lethal concentrations of this molecule (264). Lb. plantarum Lp80 poxB, encoding pyruvate oxidase, is linked to oxidative stress resistance (265,–267). PoxB utilizes oxygen to convert pyruvate to acetyl-phosphate (Fig. 2), which results in the production of CO₂ and H₂O₂. Detoxification of the latter molecule is subsequently done by NADH peroxidase (268). Notably, PoxB activity is induced by oxygen or H₂O₂ and repressed by glucose (265). In silico analyses of Lb. plantarum Lp80 revealed a strongly conserved putative binding site in the promoter region of poxB for OhR, a regulator repressing the ohrA gene involved in peroxide detoxification. This is suggestive of regulation of poxB transcription in response to oxidative stress.

Other specific mechanisms involved in the adaptation of S. thermophilus to oxidative stress that support the aerotolerance of this bacterium imply the existence of a detoxifying mechanism against H₂O₂. Genome analysis of S. thermophilus revealed the existence of a thiol-peroxidase-encoding gene, which may explain this oxygen-tolerant phenotype (269).

(iii) Modulation of ROS-based stress.

The mechanisms that evolved in LAB to overcome oxidative stress seem to overlap, and thus cells subjected to a specific stress condition can trigger different stress responses (for a review, see reference 184).

The cAMP receptor protein, the fumarate and nitrate reduction regulator (FNR), and FNR-like proteins (Flp) are part of a superfamily of proteins that encompasses structurally related transcription factors implicated in the control of stress responses (270). As mentioned above, such proteins display structural features allowing transmission of environmental and metabolic signals to the cell and which may trigger general and specific responses to various (stressful) physiological conditions (270). In Lb. casei, FNP and Flp exert a key role in the prevention of oxidative injuries by interacting with the 4Fe-4S oxygen-labile cluster (271). The Lc. lactis genome contains the two Flp-encoding genes flpA and flpB. They are located in two paralogous operons in which the proximal genes are predicted to encode metal ion transport systems (272). Targeted mutagenesis experiments showed that the FlpA/FlpB proteins control Zn ^{uptake and various stress resistance mechanisms. It was postulated that} Lc. lactis utilizes Zn ^{to protect protein thiol groups from oxidation and that intracellular levels of this ion may act as a metabolite flux sensor for sensing and control of general stress responses (}273).

Many regulators in streptococci modulate the interplay between oxidative stress and iron homeostasis, virulence, and competence (250). Iron is considered a key element in regulatory proteins sensing redox modifications in the presence of oxygen (274). This is illustrated by an S. thermophilus CNRZ368 mutant lacking the genes encoding an iron ABC transporter; the consequent imbalance in intracellular iron is the cause of an increase in superoxide anion radical sensitivity (145). In fact, control of iron homeostasis is considered to be coupled to ROS protection in streptococci (250). Also, the S. pyogenes Rgg regulator modulates the expression of general stress response proteins (275). Its involvement in the regulation of proteins related to oxidative stress was revealed when the rgg gene was inactivated (275). Inactivation of an Rgg-like protein of S. thermophilus CNRZ368 produced a phenotype similar to that of S. pyogenesrgg (250).

Accumulation of compatible solutes.

Osmotically active compounds serve a dual function in osmoregulation. In addition to maintaining a stable cell turgor, they also exert protective effects on biomolecules in vitro during stressful challenges. Various osmotically active compounds, such as proline and betaine, protect soluble and peripherally membrane-bound proteins from the denaturing effects of modifications in intracellular ionic strength and water availability. Osmotically active compounds can interact with proteins in several different ways, such as preferential exclusion, water replacement, hydration force, and vitrification of sugars (276). Betaine transport activity in LAB, such as Lc. lactis or various lactobacilli, increases with increasing osmotic stress and as a consequence of thermal insults (277, 278). Other osmotically active compounds, such as the disaccharide sucrose, protect Lc. lactis cells against pressure-induced inactivation of vital cellular components (279). The protective effect of ionic solutes has been ascribed to the intracellular accumulation of compatible solutes as a response to osmotic stress. Thus, ionic solutes (e.g., glycine betaine) provide only asymmetric protection in Lc. lactis cells; baroprotection requires a higher concentration of ionic solutes than of disaccharides (279).

(Multi)drug resistance systems.

A large number of bacterial ABC-type drug export systems have been characterized (280). Most of these microbial drug extrusion systems are specific chemical or multidrug resistance (MDR) systems that remove various substances from the cytoplasm or membrane of a cell (281). Most of the currently known MDR systems use the PMF rather than ATP as the driving force and operate as drug/H ^{antiporters (}282). A representative member of the MDR systems is the LmrA transporter of Lc. lactis, which shares both functional and structural features with the human MDR1 P-glycoprotein (283). Notably, LmrA expressed in human or insect cells efficiently exports drugs out of these cells and complements a lack of expression of the human MDR1 P-glycoprotein (284). A homolog of lmrA with highly significant structural and functional similarities to its lactococcal equivalent has been identified in O. oeni (282).

Treating Damaged Macromolecules

Major molecular chaperones and the Clp family of proteins.

In bacteria, as in all living cells, the cellular proteome is in a constant state of flux. Protein quality control, including refolding or degradation of damaged proteins, plays an indispensable role under both stressed and nonstressed conditions. Similarly, the catalog of cellular proteins is continuously being modified to meet the challenges arising from changes in the environment. The synthesis of chaperones and proteases is quickly induced under stress conditions to combat the potentially deleterious aggregation of denatured proteins (Fig. 5). While the task of chaperones is to protect functional proteins and to refold misfolded ones, proteases provide the last line of defense by removing irreversibly damaged proteins, after which the amino acids are recycled. Under changing conditions, proteolysis also removes functional proteins for regulatory purposes or simply because they are no longer needed. It is obvious that the balance between refolding and proteolysis, including the highly specific recognition of substrates, is crucial to avoid wasteful destruction (285, 286).

FIG 5

Schematic representation of the main molecular mechanisms repairing damaged macromolecules in LAB.

Chaperone proteins recognize stretches of hydrophobic amino acids typically exposed by nonnative proteins and bind to these regions to prevent aggregation. They can also actively assist protein folding in an ATP-dependent manner. One well-conserved bacterial chaperone system that can efficiently refold misfolded proteins is that of DnaK (Hsp70). The system consists of DnaK, its cochaperone (DnaJ), and a nucleotide exchange factor (GrpE) (286). Expression analyses conducted at the transcript and/or protein level revealed that production of DnaK is induced in different LAB under diverse stress conditions (287,–290). A deletion in dnaK resulted in a thermosensitive phenotype for Lc. lactis (291) and Streptococcus intermedius (292). DnaK is essential in S. mutans (293). The inability of DnaK proteins of S. intermedius, Tetragenococcus halophilus, and Lc. lactis to complement the function of E. coli DnaK in vivo suggests that functional differences exist between the chaperone systems of LAB and the current models (292, 294). Since only a few reports deal with DnaK structure or biochemical function in LAB, several fundamental questions remain to be clarified. For example, to date, the cochaperone-dependent folding activity of DnaK has not been demonstrated for any LAB. A biochemical study using purified DnaK, DnaJ, and GrpE of halophilic T. halophilus did not reveal cooperation of DnaK with the cochaperones (294).

Another refolding system that is well characterized in model bacteria is the GroEL/GroES (Hsp60) chaperonin, comprised of the 14-mer GroEL protein and its heptameric cofactor GroES. In the absence of ATP, GroEL/GroES forms a macromolecular complex arranged as two stacked heptameric rings that can bind unfolded proteins via hydrophobic interactions (286). GroES is the most abundant protein in B. subtilis, with some 500,000 molecules per cell after 30 min of exposure to heat stress (295). The production of the highly conserved GroEL and GroES proteins is activated under various stress conditions in LAB (287,–290, 296).

The bacterial Clp family of proteins and their yeast homologs Hsp104 and mitochondrial Pim1 protease belong to the HSP100 family of HSPs (297). The Clp machinery is probably the main system for general protein turnover in LAB, as is the case in other low-GC Gram-positive bacteria (298). The Clp family is a subgroup of the AAA ^{ATPase superfamily, whose members are characterized by an α-helical core domain and one or two Walker-type ATP-binding domains. The number and spacing of the ATP-binding sites in the proteins are the foundation for their division into various Clp ATPase families, including ClpA, ClpB, ClpC, ClpE, ClpL, and ClpX (}299). The ATP-dependent ClpP protease is a two-component protease consisting of a clpP-encoded serine peptidase subunit and a Clp ATPase subunit. The central proteolytic core consists of 14 ClpP subunits stacked back to back in two heptameric rings, forming an internal chamber that encapsulates the active sites of the ClpP peptidases (300). The peptidase multimer associates with one or two hexameric rings of Clp ATPases to gain proteolytic activity. The Clp ATPases, together with the proteolytic core, are responsible for substrate recognition, unfolding, and translocation into the ClpP degradation chamber. In the absence of ClpP, threading of substrates through the Clp ATPase ring structures leads to disaggregation and refolding of protein substrates, which are activities typical of molecular chaperones (301). Some Clp ATPases are able to interact with ClpP, while others function as independent chaperones. ClpP-interacting Clp ATPases, which in LAB typically include the ClpX, ClpC, and ClpE proteins, can be distinguished by the presence of a ClpP recognition tripeptide (302). There are some distinctions between the catalogs of Clp ATPases in Gram-negative and Gram-positive bacteria, but ClpP and ClpX appear to be widely conserved. ClpE is absent in E. coli but present in many species of LAB, in conjunction with ClpC, ClpL, and ClpB (302). The ClpL and ClpB proteins lack the ClpP recognition tripeptide and are thus considered to be unable to interact with ClpP. With the exception of clpX, the genes encoding the Clp family of proteins are not essential for most bacteria, including LAB (303), which has facilitated the construction of knockout mutants and the study of the physiological roles of Clp proteins in a variety of LAB species. In contrast, ClpX is not essential in S. mutans (304). Phenotypic studies with mutant strains lacking functional ClpP, ClpL, ClpE, or ClpB revealed the central role of these proteins under stress conditions and, for pathogenic LAB species, in virulence (303, 305,–308). In contrast to the situation in B. subtilis, LAB clpC mutants usually do not exhibit stress-sensitive phenotypes (306, 309). Phenotypic characterization of an S. thermophilusclpC mutant revealed that ClpC plays a role in the development of genetic competence in this organism (310). The alternative sigma factor ComX, which drives the expression of competence genes, was shown in S. thermophilus and S. mutans to be under posttranslational regulation by ClpCP through active degradation (310,–313). This task is performed by ClpEP in S. pneumoniae (314).

Regulation of the major molecular chaperones and the Clp family of proteins.

LAB lack the alternative sigma factor σ^{, a key element in the regulation of the general stress response in low-GC Gram-positive organisms, such as} Staphylococcus aureus, Listeria monocytogenes, and B. subtilis. The primary stress induction of molecular chaperones and Clp proteins in LAB occurs at the level of transcription and is regulated mainly by the transcriptional repressors CtsR and HrcA. HrcA recognizes the highly conserved CIRCE (control of inverted repeat for chaperone expression) operator sequence, while CtsR binds to a heptanucleotide repeat (A/GGTCAAA/T) located in the promoter regions of target genes (315). The B. subtilis CtsR and HrcA regulons are distinct from each other, while in LAB they sometimes overlap. One of the repressors is absent in some LAB species, and a single regulator controls expression of both regulons (315). The expression of S. salivarius and S. thermophilusclpP is under dual regulation by CtsR and HrcA (316). The HrcA and CtsR regulons may partially overlap in several other members of the Streptococcaceae (316, 317). Sequence analysis of Lb. plantarum revealed two CIRCE elements and a CtsR-binding site upstream of hrcA-dnaK-grpE-dnaJ, indicating dual control of this operon by both regulators (318). O. oeni was the first example of a low-GC Gram-positive bacterial species that lacks the widely conserved CIRCE-HrcA regulatory circuit. Its CtsR regulon comprises genes that encode the classical chaperone proteins DnaK, GroEL, and GroES and Clp proteins (317). An example of reductive evolution of stress repressors with a different outcome is provided by Lb. acidophilus, Lb. delbrueckii subsp. bulgaricus, and Lactobacillus johnsonii, which all lack CtsR (201, 319, 320). The HrcA regulon in these species has expanded to include clp genes, in addition to genes encoding the groEL and dnaK operons. The molecular details behind the regulation of HrcA function have not been addressed for LAB. According to studies of other organisms, HrcA is prone to aggregation, a characteristic feature that is also central to its autoregulation (321). HrcA is thought to be released from ribosomes in an inactive form. To become an active DNA-binding protein, it must interact with the GroEL chaperone. Under stress conditions, GroEL is titrated away from HrcA by accumulating denatured proteins, which drives the equilibrium toward aggregated/inactive HrcA (321). Thus, protein stress directly affects HrcA activity. Molecular studies have revealed several differences in the mechanisms of regulation of CtsR function between LAB and other bacteria. The ctsR gene is cotranscribed with clpC in all low-GC Gram-positive bacteria. In members of the Bacillales order, the clpC operon is tetracistronic and consists of genes encoding two modulators of CtsR activity (McsA and McsB) in addition to ctsR and clpC. McsA and McsB are lacking in LAB, and ctsR forms a dicistronic operon with clpC (315). McsB is not involved in the regulation of CtsR activity during heat stress in vivo, and CtsR itself acts as an intrinsic thermosensor whose DNA-binding activity gradually decreases with increasing temperature. The CtsR regulon is also derepressed under several environmental stress conditions other than heat shock, such as acid stress and oxidative damage. This implies that alternative mechanisms for CtsR inactivation must exist. Interestingly, in low-GC Gram-positive bacteria lacking McsA/McsB, such as Lc. lactis, ClpE activity appears to partially substitute for the function of McsA and McsB. Lc. lactis ClpE has been reported to contain a zinc finger domain similar to that of McsA and to undergo a thiol-dependent ClpP-mediated cleavage similar to that of McsA (298, 322). Lc. lactisclpE can complement a B. subtilismcsB mutant in inactivating CtsR during oxidative stress (298). A recent report provided evidence that S. mutans ClpL interacts with CtsR both in vivo and in vitro and, by its chaperone activity, prevents CtsR from aggregating under ambient growth conditions (323).

FtsH, HtrA, and small HSPs (sHSPs).

FtsH is a membrane-bound HSP with dual chaperone-protease activity (Fig. 5). The FtsH protein contains transmembrane segments in the N-terminal region and a main cytosolic region consisting of an AAA ^{ATPase and a Zn metalloprotease domain (}324). The functional roles of FtsH in LAB have not been widely studied. FtsH plays roles in protein quality control during heat shock in LAB, such as Lc. lactis (325), O. oeni (326), and Lb. plantarum (327). The FtsH proteins of Lc. lactis and O. oeni are both able to complement the growth defect of an E. coliftsH mutant at 37°C, indicating that FtsH from these LAB can at least partially functionally replace E. coli FtsH. Inactivation of ftsH resulted in salt-, heat-, and cold-sensitive phenotypes of Lc. lactis (325), whereas reduced salt and heat tolerance as well as diminished biofilm formation were reported for an Lb. plantarumftsH mutant (176). Lb. plantarumftsH was shown to be a novel member of the CtsR regulon (327). The mechanisms controlling ftsH expression in other LAB have not been studied.

HtrA (high temperature requirement protein A), also known as DegP, is involved in degradation of extracellular polypeptides and misfolded proteins (Fig. 5). HtrA contains a chaperone and a proteolytic domain. An Lc. lactishtrA mutant is thermosensitive, and HtrA appears to play a fundamental role in the degradation of abnormal exported proteins (328). The htrA gene has been identified as a heat shock gene in Lactobacillus helveticus (329), and elevated HtrA levels have been detected in Lb. casei in response to phenolic compounds (330). However, the regulatory mechanism behind the stress-inducible expression of htrA in LAB remains to be elucidated.

The members of the sHSP family function as folding chaperones, assisting the protein folding process by stabilizing unfolded or partially folded proteins without actively promoting their remodeling (331). Furthermore, sHSPs appear to play an important role in maintaining membrane integrity, especially under stress conditions (332). The sHSP family is widely conserved, and its members have been identified in several LAB (302). In O. oeni, an sHSP named Lo18 was identified as a major stress protein whose synthesis is greatly increased under various conditions, including ethanol shock (333). Lo18 has been suggested to play a dual role in O. oeni by preventing protein aggregation and stabilizing the cytoplasmic membrane (334).

Proteins induced by cold shock.

A sudden temperature downshift induces a cold shock response, during which growth of cells is first arrested for a period before reinitiation at a new and much reduced rate. The cellular effects of cold shock include the following: (i) a decrease in membrane fluidity, (ii) a reduced efficiency of mRNA translation and transcription through stabilization of secondary structures of DNA and RNA, (iii) ineffective or slow protein folding, and (iv) hindered ribosome function (335). In cold-shocked E. coli, overall transcription and translation are slowed considerably, and a set of approximately 26 cold shock genes is preferentially and transiently expressed (336). Members of the CSP family enhance the expression of cold shock-inducible genes by destabilizing nonproductive secondary structures in mRNA at low temperatures and therefore are often called RNA chaperones (336). While cold and heat shocks are considered highly separate processes in E. coli and B. subtilis (337), recent evidence indicates that these responses are interconnected in LAB (338). Cold shock of S. thermophilus is accompanied by increased production of GroEL and ClpL (296, 307). Furthermore, the clpL knockout mutant of S. thermophilus is cold sensitive (307). Cold shock conditions increase the synthesis of DnaK and GroEL in Leuconostoc mesenteroides (288) and of sHSPs in Lb. plantarum (339). Transcriptional analyses have indicated that clpX is induced by both heat and cold in Lc. lactis (340). The proposed cross talk between cold and heat shock responses in LAB is further supported by the improved cryotolerance seen after heat treatment of Lb. johnsonii and Lc. lactis (341, 342). Similar findings may link acid and cold stress responses in LAB; for Lb. delbrueckii subsp. bulgaricus, an increase in freezing tolerance was observed after acidification of the growth medium (343).

Spx governs the response to oxidative damage of macromolecules.

Spx is a global transcriptional regulator of oxidative stress in several Gram-positive bacteria and plays important roles as both a positive and a negative modulator of gene expression (344). The LAB Spx homolog TrmA (now named SpxA2) was first identified via mutations that alleviate the heat-sensitive phenotype of clpP mutants and recA mutants of Lc. lactis (245, 345). Proteolysis of Spx is mediated by the ClpXP protease system, and Spx accumulates to high levels in clpX or clpP mutants (304). Studies of B. subtilis Spx activity revealed that it interacts directly with the α subunit of the RNA polymerase (RNAP) and thereby controls global transcription initiation, either negatively or positively, by a unique mechanism not involving initial contact between Spx and DNA (344). Negative regulation is mediated by binding of Spx to Tyr263 in the C terminus of the RNAP α subunit. This residue is highly conserved among RNAP α subunits of Gram-positive bacteria, and the binding blocks interaction between other transcriptional activators and RNAP (344). Lc. lactis encodes seven Spx paralogs, among which SpxB has been shown to be involved in the cell wall stress response. DNA microarray work with spx mutants confirmed the function of Spx as a major oxidative stress regulator (346, 347). Streptococci harbor two bona fide Spx regulators, whereas enterococci have only one (346,–349).

Repair of stress-induced DNA damage.

According to the current DNA damage paradigm, cells undergo an inducible process called the SOS response during genotoxic stress. This response triggers the expression of genes involved in DNA replication, repair, and mutagenesis (350). In the case of severe DNA damage, high-fidelity DNA polymerase cannot replicate over the incurred DNA lesions and is replaced by a low-fidelity DNA polymerase(s). The process, called translesion synthesis, is performed mainly by SOS-induced error-prone polymerases of the Y family (Pol IV and Pol V) or the C family (DnaE) (350). The SOS response is repressed under physiological conditions by binding of the LexA repressor protein to operator sites in the promoter regions of SOS genes, including its own gene, lexA (350). While the bacterial DNA damage response is one of the most extensively studied stress responses, it has not been studied widely in LAB. LexA regulons have been identified in some LAB species, such as Lb. plantarum (351) and E. faecalis (352). The uvrA gene, which encodes a nucleotide excision repair protein and is a typical member of LexA regulons, has been characterized for S. mutans (353) and Lb. helveticus (354). An S. mutansuvrA mutant, in addition to being sensitive to DNA damage, was also sensitive to acidic conditions, indicating that UvrA plays a role in the repair of acid-induced DNA damage (353). Other DNA repair activities that have been implicated in the resistance of S. mutans to acidic pH include those of the Smx nuclease (355, 356). The S. pyogenes C family DNA polymerase DnaE was shown to be highly error-prone and produced frameshift and point mutations during replication of both damaged and undamaged DNAs in vitro (357). In S. uberis, the Pol V subunit UmuC is SOS induced and essential for UV-induced mutagenesis (49).

Preventing Macromolecules from Being Damaged

Some 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.

FIG 4

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

The F-ATPase proton pump.

Among the many mechanisms that LAB species employ to counteract stress damage, the F-ATPase system deserves attention as a system that protects against protons. The F-ATPase acts in parallel to the above-described metabolic adaptations that result in alkalization of the intracellular and extracellular environments. F-ATPase is a multimeric enzyme that in Gram-positive facultative anaerobes hydrolyzes ATP synthetized during carbohydrate fermentation or FAA catabolism, generating a PMF (Fig. 4). F-ATPase is a reversible enzyme that can also synthesize ATP by using protons that flow from the environment into the cell. Under acidic conditions, LAB use F-ATPase as a proton pump to maintain intracellular pH homeostasis. This reaction requires ATP because the expulsion of protons from the cytoplasm to the outside environment takes place against an increasing proton gradient that is a consequence of the synthesis of lactic acid during carbohydrate fermentation (231). The level of F-ATPase activity depends on proton transport demand, substrate catabolism, and availability of ATP. The F-ATPase systems of Lb. casei and Lb. plantarum have optimum activities at pH values (pH 5.0 to 5.5) lower than those (pH 7.0 to 7.5) of S. thermophilus and Lc. lactis (232). Distinctly aciduric oral streptococci, such as S. mutans GS-5 and E. hirae ATCC 9790, possess an F-ATPase with a lower optimal pH than that of the more acid-sensitive oral streptococci (233). Interestingly, LAB species or strains that possess an F-ATPase with an optimal activity at low pH are highly protected against acid stress. The capacity to pump protons out of the cell under acidic conditions protects important glycolytic enzymes and allows long-term survival (234). Lactobacilli increase the synthesis of one or more of their ATP synthase components (e.g., AtpB, AtpA, AtpG, and AtpD) under acidic conditions. The F-ATPase complex not only is involved in the maintenance of cytoplasmic pH homeostasis but also responds to bile salt exposure (235,–237). It was noticed that in Lb. casei BL23 (235) as well as in Lb. rhamnosus GG (236), the exposure of cells to bile salt caused an enhanced abundance of F-ATPase, though this may have been a direct consequence of PMF dissipation, as reported for other LAB treated in the same manner (238).

Detoxification of ROS in LAB.

Oxygen per se does not damage cells, but during the operation of metabolic pathways it is partially reduced to water, leading to the synthesis of ROS, including superoxide anion radicals, hydroxyl radicals, and H₂O₂. These display a considerable oxidizing potential and are highly toxic to cells (239, 240). ROS toxicity is also exerted through chemical modification and/or inactivation of most types of biomolecules (184). As detailed above, Lc. lactis can use oxygen under certain growth conditions (179). Also, H₂O₂ may be formed during mixed-acid fermentation as a consequence of the activity of NADH oxidases (241). Superoxide anion as well as hydroxyl radicals are produced during the Fenton reaction (242, 243). Metal ions can catalyze certain reactions leading to the formation of ROS, which is in fact considered one of the underlying mechanisms of metal-induced stress. Iron, copper, and other redox-active metal ions exert their effects by stimulating the Fenton reaction.

(i) Cellular damage produced by ROS.

At the molecular level, ROS can react with macromolecules, such as proteins, lipids, and nucleic acids. Superoxide anions have a limited oxidizing potential and may attack compounds such as polyphenols, ascorbate, and catecholamines (239). H₂O₂ can inactivate proteins through the oxidation of cysteinyl residues (240) or can react with cations, such as Fe ^{and Cu, producing hydroxyl radicals by the Fenton reaction (}239, 244). The damage caused by the hydroxyl radical, which is the most destructive oxidative agent, includes strand breakage and chemical modification of DNA (239, 244).

(ii) ROS resistome.

Bacteria such as Lc. lactis, E. coli, and B. subtilis are able to cope with oxidative stress through a number of enzymes that counteract the negative effects of oxidation (245). Notably, induction of genes encoding ROS-protective enzymes is growth phase dependent and confers multistress resistance (184, 246). Among the main players in LAB that help to counteract the deleterious effects of oxidative stress are (i) NADH oxidase/NADH peroxidase, (ii) superoxide dismutase (SOD), (iii) cytochrome d oxidase, (iv) catalase in the presence of necessary cofactors, and (v) nonenzymatic dismutation of H₂O₂ by Mn^.

The most conserved oxidative resistance mechanism in LAB results from the coupling of NADH oxidase and NADH peroxidase (184). Oxygen is initially used for the oxidation of NADH to NAD ^{via the NADH oxidase, a reaction that produces H}₂O₂. H₂O₂ is further reduced to water by the NADH peroxidase. NADH peroxidase activity is rather low in Lc. lactis, and the detoxification of H₂O₂ is incomplete. It was recently proposed that metabolically synthetized pyruvate reacts nonenzymatically with residual H₂O₂ to produce water (247). Multiple enzymes with peroxidase activity have been described for LAB, such as AhpCF, Npr, and Tpx of E. faecalis and AhpCF and Tpx of S. mutans (248).

Two distinct NADH oxidases have been identified in S. mutans. They catalyze either the two-electron reduction of oxygen to H₂O₂ (performed by Nox1) or the four-electron reduction of oxygen to water (AhpCF) and thereby provide a very efficient enzymatic defense against oxidative stress (249). S. thermophilus, on the other hand, does not appear to encode a Nox1 enzyme but uses a single NADH oxidase to reduce oxygen directly to water (250).

Another key player in resistance to oxidative stress is SOD, which eliminates superoxide anion radicals. Bacteria possess different types of SOD enzymes depending on the bivalent metal cofactor with which the enzyme is coupled, i.e., Fe^{, Mn, and/or Cu-Zn. LAB produce mainly Mn-binding SODs (}251). Analysis of an Lc. lactissod mutant clearly established that SOD protects against superoxide anion radicals (251). An alternative strategy to counteract superoxide anion radicals is by the production of high levels of intracellular glutathione (252). Various enzymes that are known to counter oxidative stress, namely, MsrAB, AhpCF, Tpx, Npr, KatA, Sod, and HypR, have been characterized for E. faecalis (94, 248, 253,–255).

E. faecalis possesses a respiratory chain that produces extracellular superoxide, whose production level depends on the availability of hematin as a cofactor for cytochrome bd or on fumarate as a terminal electron acceptor (256, 257). When this pathway is impaired by restricting access to these nutrients, extracellular superoxide is generated by partially reduced demethylmenaquinones upon exposure to oxygen (257). Thus, E. faecalis necessarily expresses multiple antioxidant defense mechanisms, such as NADH peroxidase and manganese-containing SOD (MnSOD).

Bacteria can also cope with oxidative stress through catalase, an enzyme catalyzing the detoxification of H₂O₂. The following three classes of bacterial catalases are currently recognized: monofunctional catalases, catalase-peroxidases, and manganese catalases (pseudocatalases) (258). The monofunctional catalases and the catalase-peroxidases contain heme as a prosthetic group. The monofunctional catalase specified by the genome of E. faecalis has been shown to remove H₂O₂ (259).

In order to retain the low intracellular redox potential required to keep proteins in their reduced form, the presence of glutathione as well as the thioredoxin systems is crucial (260). Members of the genus Streptococcus contain glutathione. They either synthesize glutathione or import it from the medium (261). In addition, streptococci encode a glutathione reductase, which catalyzes the reduction of glutathione disulfide that is produced upon oxidation of glutathione by NADPH (262). The glutathione reductase gene in S. thermophilus is induced in the presence of oxygen (262). The S. thermophilus genome also contains genes encoding two thioredoxins, a thioredoxin reductase, and a single glutaredoxin. This relatively large genetic repertoire emphasizes the importance of such enzymes in oxidative stress resistance (for a review, see reference 250). E. faecalis is able to synthesize glutathione and expresses glutathione reductase (261, 263).

Adaptation is also part of the survival strategy of LAB to cope with oxidative stress. Upon exposure to sublethal concentrations of H₂O₂, Lc. lactis can survive in the presence of otherwise lethal concentrations of this molecule (264). Lb. plantarum Lp80 poxB, encoding pyruvate oxidase, is linked to oxidative stress resistance (265,–267). PoxB utilizes oxygen to convert pyruvate to acetyl-phosphate (Fig. 2), which results in the production of CO₂ and H₂O₂. Detoxification of the latter molecule is subsequently done by NADH peroxidase (268). Notably, PoxB activity is induced by oxygen or H₂O₂ and repressed by glucose (265). In silico analyses of Lb. plantarum Lp80 revealed a strongly conserved putative binding site in the promoter region of poxB for OhR, a regulator repressing the ohrA gene involved in peroxide detoxification. This is suggestive of regulation of poxB transcription in response to oxidative stress.

Other specific mechanisms involved in the adaptation of S. thermophilus to oxidative stress that support the aerotolerance of this bacterium imply the existence of a detoxifying mechanism against H₂O₂. Genome analysis of S. thermophilus revealed the existence of a thiol-peroxidase-encoding gene, which may explain this oxygen-tolerant phenotype (269).

(iii) Modulation of ROS-based stress.

The mechanisms that evolved in LAB to overcome oxidative stress seem to overlap, and thus cells subjected to a specific stress condition can trigger different stress responses (for a review, see reference 184).

The cAMP receptor protein, the fumarate and nitrate reduction regulator (FNR), and FNR-like proteins (Flp) are part of a superfamily of proteins that encompasses structurally related transcription factors implicated in the control of stress responses (270). As mentioned above, such proteins display structural features allowing transmission of environmental and metabolic signals to the cell and which may trigger general and specific responses to various (stressful) physiological conditions (270). In Lb. casei, FNP and Flp exert a key role in the prevention of oxidative injuries by interacting with the 4Fe-4S oxygen-labile cluster (271). The Lc. lactis genome contains the two Flp-encoding genes flpA and flpB. They are located in two paralogous operons in which the proximal genes are predicted to encode metal ion transport systems (272). Targeted mutagenesis experiments showed that the FlpA/FlpB proteins control Zn ^{uptake and various stress resistance mechanisms. It was postulated that} Lc. lactis utilizes Zn ^{to protect protein thiol groups from oxidation and that intracellular levels of this ion may act as a metabolite flux sensor for sensing and control of general stress responses (}273).

Many regulators in streptococci modulate the interplay between oxidative stress and iron homeostasis, virulence, and competence (250). Iron is considered a key element in regulatory proteins sensing redox modifications in the presence of oxygen (274). This is illustrated by an S. thermophilus CNRZ368 mutant lacking the genes encoding an iron ABC transporter; the consequent imbalance in intracellular iron is the cause of an increase in superoxide anion radical sensitivity (145). In fact, control of iron homeostasis is considered to be coupled to ROS protection in streptococci (250). Also, the S. pyogenes Rgg regulator modulates the expression of general stress response proteins (275). Its involvement in the regulation of proteins related to oxidative stress was revealed when the rgg gene was inactivated (275). Inactivation of an Rgg-like protein of S. thermophilus CNRZ368 produced a phenotype similar to that of S. pyogenesrgg (250).

Accumulation of compatible solutes.

Osmotically active compounds serve a dual function in osmoregulation. In addition to maintaining a stable cell turgor, they also exert protective effects on biomolecules in vitro during stressful challenges. Various osmotically active compounds, such as proline and betaine, protect soluble and peripherally membrane-bound proteins from the denaturing effects of modifications in intracellular ionic strength and water availability. Osmotically active compounds can interact with proteins in several different ways, such as preferential exclusion, water replacement, hydration force, and vitrification of sugars (276). Betaine transport activity in LAB, such as Lc. lactis or various lactobacilli, increases with increasing osmotic stress and as a consequence of thermal insults (277, 278). Other osmotically active compounds, such as the disaccharide sucrose, protect Lc. lactis cells against pressure-induced inactivation of vital cellular components (279). The protective effect of ionic solutes has been ascribed to the intracellular accumulation of compatible solutes as a response to osmotic stress. Thus, ionic solutes (e.g., glycine betaine) provide only asymmetric protection in Lc. lactis cells; baroprotection requires a higher concentration of ionic solutes than of disaccharides (279).

(Multi)drug resistance systems.

A large number of bacterial ABC-type drug export systems have been characterized (280). Most of these microbial drug extrusion systems are specific chemical or multidrug resistance (MDR) systems that remove various substances from the cytoplasm or membrane of a cell (281). Most of the currently known MDR systems use the PMF rather than ATP as the driving force and operate as drug/H ^{antiporters (}282). A representative member of the MDR systems is the LmrA transporter of Lc. lactis, which shares both functional and structural features with the human MDR1 P-glycoprotein (283). Notably, LmrA expressed in human or insect cells efficiently exports drugs out of these cells and complements a lack of expression of the human MDR1 P-glycoprotein (284). A homolog of lmrA with highly significant structural and functional similarities to its lactococcal equivalent has been identified in O. oeni (282).

Treating Damaged Macromolecules

Major molecular chaperones and the Clp family of proteins.

In bacteria, as in all living cells, the cellular proteome is in a constant state of flux. Protein quality control, including refolding or degradation of damaged proteins, plays an indispensable role under both stressed and nonstressed conditions. Similarly, the catalog of cellular proteins is continuously being modified to meet the challenges arising from changes in the environment. The synthesis of chaperones and proteases is quickly induced under stress conditions to combat the potentially deleterious aggregation of denatured proteins (Fig. 5). While the task of chaperones is to protect functional proteins and to refold misfolded ones, proteases provide the last line of defense by removing irreversibly damaged proteins, after which the amino acids are recycled. Under changing conditions, proteolysis also removes functional proteins for regulatory purposes or simply because they are no longer needed. It is obvious that the balance between refolding and proteolysis, including the highly specific recognition of substrates, is crucial to avoid wasteful destruction (285, 286).

FIG 5

Schematic representation of the main molecular mechanisms repairing damaged macromolecules in LAB.

Chaperone proteins recognize stretches of hydrophobic amino acids typically exposed by nonnative proteins and bind to these regions to prevent aggregation. They can also actively assist protein folding in an ATP-dependent manner. One well-conserved bacterial chaperone system that can efficiently refold misfolded proteins is that of DnaK (Hsp70). The system consists of DnaK, its cochaperone (DnaJ), and a nucleotide exchange factor (GrpE) (286). Expression analyses conducted at the transcript and/or protein level revealed that production of DnaK is induced in different LAB under diverse stress conditions (287,–290). A deletion in dnaK resulted in a thermosensitive phenotype for Lc. lactis (291) and Streptococcus intermedius (292). DnaK is essential in S. mutans (293). The inability of DnaK proteins of S. intermedius, Tetragenococcus halophilus, and Lc. lactis to complement the function of E. coli DnaK in vivo suggests that functional differences exist between the chaperone systems of LAB and the current models (292, 294). Since only a few reports deal with DnaK structure or biochemical function in LAB, several fundamental questions remain to be clarified. For example, to date, the cochaperone-dependent folding activity of DnaK has not been demonstrated for any LAB. A biochemical study using purified DnaK, DnaJ, and GrpE of halophilic T. halophilus did not reveal cooperation of DnaK with the cochaperones (294).

Another refolding system that is well characterized in model bacteria is the GroEL/GroES (Hsp60) chaperonin, comprised of the 14-mer GroEL protein and its heptameric cofactor GroES. In the absence of ATP, GroEL/GroES forms a macromolecular complex arranged as two stacked heptameric rings that can bind unfolded proteins via hydrophobic interactions (286). GroES is the most abundant protein in B. subtilis, with some 500,000 molecules per cell after 30 min of exposure to heat stress (295). The production of the highly conserved GroEL and GroES proteins is activated under various stress conditions in LAB (287,–290, 296).

The bacterial Clp family of proteins and their yeast homologs Hsp104 and mitochondrial Pim1 protease belong to the HSP100 family of HSPs (297). The Clp machinery is probably the main system for general protein turnover in LAB, as is the case in other low-GC Gram-positive bacteria (298). The Clp family is a subgroup of the AAA ^{ATPase superfamily, whose members are characterized by an α-helical core domain and one or two Walker-type ATP-binding domains. The number and spacing of the ATP-binding sites in the proteins are the foundation for their division into various Clp ATPase families, including ClpA, ClpB, ClpC, ClpE, ClpL, and ClpX (}299). The ATP-dependent ClpP protease is a two-component protease consisting of a clpP-encoded serine peptidase subunit and a Clp ATPase subunit. The central proteolytic core consists of 14 ClpP subunits stacked back to back in two heptameric rings, forming an internal chamber that encapsulates the active sites of the ClpP peptidases (300). The peptidase multimer associates with one or two hexameric rings of Clp ATPases to gain proteolytic activity. The Clp ATPases, together with the proteolytic core, are responsible for substrate recognition, unfolding, and translocation into the ClpP degradation chamber. In the absence of ClpP, threading of substrates through the Clp ATPase ring structures leads to disaggregation and refolding of protein substrates, which are activities typical of molecular chaperones (301). Some Clp ATPases are able to interact with ClpP, while others function as independent chaperones. ClpP-interacting Clp ATPases, which in LAB typically include the ClpX, ClpC, and ClpE proteins, can be distinguished by the presence of a ClpP recognition tripeptide (302). There are some distinctions between the catalogs of Clp ATPases in Gram-negative and Gram-positive bacteria, but ClpP and ClpX appear to be widely conserved. ClpE is absent in E. coli but present in many species of LAB, in conjunction with ClpC, ClpL, and ClpB (302). The ClpL and ClpB proteins lack the ClpP recognition tripeptide and are thus considered to be unable to interact with ClpP. With the exception of clpX, the genes encoding the Clp family of proteins are not essential for most bacteria, including LAB (303), which has facilitated the construction of knockout mutants and the study of the physiological roles of Clp proteins in a variety of LAB species. In contrast, ClpX is not essential in S. mutans (304). Phenotypic studies with mutant strains lacking functional ClpP, ClpL, ClpE, or ClpB revealed the central role of these proteins under stress conditions and, for pathogenic LAB species, in virulence (303, 305,–308). In contrast to the situation in B. subtilis, LAB clpC mutants usually do not exhibit stress-sensitive phenotypes (306, 309). Phenotypic characterization of an S. thermophilusclpC mutant revealed that ClpC plays a role in the development of genetic competence in this organism (310). The alternative sigma factor ComX, which drives the expression of competence genes, was shown in S. thermophilus and S. mutans to be under posttranslational regulation by ClpCP through active degradation (310,–313). This task is performed by ClpEP in S. pneumoniae (314).

Major molecular chaperones and the Clp family of proteins.

In bacteria, as in all living cells, the cellular proteome is in a constant state of flux. Protein quality control, including refolding or degradation of damaged proteins, plays an indispensable role under both stressed and nonstressed conditions. Similarly, the catalog of cellular proteins is continuously being modified to meet the challenges arising from changes in the environment. The synthesis of chaperones and proteases is quickly induced under stress conditions to combat the potentially deleterious aggregation of denatured proteins (Fig. 5). While the task of chaperones is to protect functional proteins and to refold misfolded ones, proteases provide the last line of defense by removing irreversibly damaged proteins, after which the amino acids are recycled. Under changing conditions, proteolysis also removes functional proteins for regulatory purposes or simply because they are no longer needed. It is obvious that the balance between refolding and proteolysis, including the highly specific recognition of substrates, is crucial to avoid wasteful destruction (285, 286).

FIG 5

Schematic representation of the main molecular mechanisms repairing damaged macromolecules in LAB.

Chaperone proteins recognize stretches of hydrophobic amino acids typically exposed by nonnative proteins and bind to these regions to prevent aggregation. They can also actively assist protein folding in an ATP-dependent manner. One well-conserved bacterial chaperone system that can efficiently refold misfolded proteins is that of DnaK (Hsp70). The system consists of DnaK, its cochaperone (DnaJ), and a nucleotide exchange factor (GrpE) (286). Expression analyses conducted at the transcript and/or protein level revealed that production of DnaK is induced in different LAB under diverse stress conditions (287,–290). A deletion in dnaK resulted in a thermosensitive phenotype for Lc. lactis (291) and Streptococcus intermedius (292). DnaK is essential in S. mutans (293). The inability of DnaK proteins of S. intermedius, Tetragenococcus halophilus, and Lc. lactis to complement the function of E. coli DnaK in vivo suggests that functional differences exist between the chaperone systems of LAB and the current models (292, 294). Since only a few reports deal with DnaK structure or biochemical function in LAB, several fundamental questions remain to be clarified. For example, to date, the cochaperone-dependent folding activity of DnaK has not been demonstrated for any LAB. A biochemical study using purified DnaK, DnaJ, and GrpE of halophilic T. halophilus did not reveal cooperation of DnaK with the cochaperones (294).

Another refolding system that is well characterized in model bacteria is the GroEL/GroES (Hsp60) chaperonin, comprised of the 14-mer GroEL protein and its heptameric cofactor GroES. In the absence of ATP, GroEL/GroES forms a macromolecular complex arranged as two stacked heptameric rings that can bind unfolded proteins via hydrophobic interactions (286). GroES is the most abundant protein in B. subtilis, with some 500,000 molecules per cell after 30 min of exposure to heat stress (295). The production of the highly conserved GroEL and GroES proteins is activated under various stress conditions in LAB (287,–290, 296).

The bacterial Clp family of proteins and their yeast homologs Hsp104 and mitochondrial Pim1 protease belong to the HSP100 family of HSPs (297). The Clp machinery is probably the main system for general protein turnover in LAB, as is the case in other low-GC Gram-positive bacteria (298). The Clp family is a subgroup of the AAA ^{ATPase superfamily, whose members are characterized by an α-helical core domain and one or two Walker-type ATP-binding domains. The number and spacing of the ATP-binding sites in the proteins are the foundation for their division into various Clp ATPase families, including ClpA, ClpB, ClpC, ClpE, ClpL, and ClpX (}299). The ATP-dependent ClpP protease is a two-component protease consisting of a clpP-encoded serine peptidase subunit and a Clp ATPase subunit. The central proteolytic core consists of 14 ClpP subunits stacked back to back in two heptameric rings, forming an internal chamber that encapsulates the active sites of the ClpP peptidases (300). The peptidase multimer associates with one or two hexameric rings of Clp ATPases to gain proteolytic activity. The Clp ATPases, together with the proteolytic core, are responsible for substrate recognition, unfolding, and translocation into the ClpP degradation chamber. In the absence of ClpP, threading of substrates through the Clp ATPase ring structures leads to disaggregation and refolding of protein substrates, which are activities typical of molecular chaperones (301). Some Clp ATPases are able to interact with ClpP, while others function as independent chaperones. ClpP-interacting Clp ATPases, which in LAB typically include the ClpX, ClpC, and ClpE proteins, can be distinguished by the presence of a ClpP recognition tripeptide (302). There are some distinctions between the catalogs of Clp ATPases in Gram-negative and Gram-positive bacteria, but ClpP and ClpX appear to be widely conserved. ClpE is absent in E. coli but present in many species of LAB, in conjunction with ClpC, ClpL, and ClpB (302). The ClpL and ClpB proteins lack the ClpP recognition tripeptide and are thus considered to be unable to interact with ClpP. With the exception of clpX, the genes encoding the Clp family of proteins are not essential for most bacteria, including LAB (303), which has facilitated the construction of knockout mutants and the study of the physiological roles of Clp proteins in a variety of LAB species. In contrast, ClpX is not essential in S. mutans (304). Phenotypic studies with mutant strains lacking functional ClpP, ClpL, ClpE, or ClpB revealed the central role of these proteins under stress conditions and, for pathogenic LAB species, in virulence (303, 305,–308). In contrast to the situation in B. subtilis, LAB clpC mutants usually do not exhibit stress-sensitive phenotypes (306, 309). Phenotypic characterization of an S. thermophilusclpC mutant revealed that ClpC plays a role in the development of genetic competence in this organism (310). The alternative sigma factor ComX, which drives the expression of competence genes, was shown in S. thermophilus and S. mutans to be under posttranslational regulation by ClpCP through active degradation (310,–313). This task is performed by ClpEP in S. pneumoniae (314).

Regulation of the major molecular chaperones and the Clp family of proteins.

LAB lack the alternative sigma factor σ^{, a key element in the regulation of the general stress response in low-GC Gram-positive organisms, such as} Staphylococcus aureus, Listeria monocytogenes, and B. subtilis. The primary stress induction of molecular chaperones and Clp proteins in LAB occurs at the level of transcription and is regulated mainly by the transcriptional repressors CtsR and HrcA. HrcA recognizes the highly conserved CIRCE (control of inverted repeat for chaperone expression) operator sequence, while CtsR binds to a heptanucleotide repeat (A/GGTCAAA/T) located in the promoter regions of target genes (315). The B. subtilis CtsR and HrcA regulons are distinct from each other, while in LAB they sometimes overlap. One of the repressors is absent in some LAB species, and a single regulator controls expression of both regulons (315). The expression of S. salivarius and S. thermophilusclpP is under dual regulation by CtsR and HrcA (316). The HrcA and CtsR regulons may partially overlap in several other members of the Streptococcaceae (316, 317). Sequence analysis of Lb. plantarum revealed two CIRCE elements and a CtsR-binding site upstream of hrcA-dnaK-grpE-dnaJ, indicating dual control of this operon by both regulators (318). O. oeni was the first example of a low-GC Gram-positive bacterial species that lacks the widely conserved CIRCE-HrcA regulatory circuit. Its CtsR regulon comprises genes that encode the classical chaperone proteins DnaK, GroEL, and GroES and Clp proteins (317). An example of reductive evolution of stress repressors with a different outcome is provided by Lb. acidophilus, Lb. delbrueckii subsp. bulgaricus, and Lactobacillus johnsonii, which all lack CtsR (201, 319, 320). The HrcA regulon in these species has expanded to include clp genes, in addition to genes encoding the groEL and dnaK operons. The molecular details behind the regulation of HrcA function have not been addressed for LAB. According to studies of other organisms, HrcA is prone to aggregation, a characteristic feature that is also central to its autoregulation (321). HrcA is thought to be released from ribosomes in an inactive form. To become an active DNA-binding protein, it must interact with the GroEL chaperone. Under stress conditions, GroEL is titrated away from HrcA by accumulating denatured proteins, which drives the equilibrium toward aggregated/inactive HrcA (321). Thus, protein stress directly affects HrcA activity. Molecular studies have revealed several differences in the mechanisms of regulation of CtsR function between LAB and other bacteria. The ctsR gene is cotranscribed with clpC in all low-GC Gram-positive bacteria. In members of the Bacillales order, the clpC operon is tetracistronic and consists of genes encoding two modulators of CtsR activity (McsA and McsB) in addition to ctsR and clpC. McsA and McsB are lacking in LAB, and ctsR forms a dicistronic operon with clpC (315). McsB is not involved in the regulation of CtsR activity during heat stress in vivo, and CtsR itself acts as an intrinsic thermosensor whose DNA-binding activity gradually decreases with increasing temperature. The CtsR regulon is also derepressed under several environmental stress conditions other than heat shock, such as acid stress and oxidative damage. This implies that alternative mechanisms for CtsR inactivation must exist. Interestingly, in low-GC Gram-positive bacteria lacking McsA/McsB, such as Lc. lactis, ClpE activity appears to partially substitute for the function of McsA and McsB. Lc. lactis ClpE has been reported to contain a zinc finger domain similar to that of McsA and to undergo a thiol-dependent ClpP-mediated cleavage similar to that of McsA (298, 322). Lc. lactisclpE can complement a B. subtilismcsB mutant in inactivating CtsR during oxidative stress (298). A recent report provided evidence that S. mutans ClpL interacts with CtsR both in vivo and in vitro and, by its chaperone activity, prevents CtsR from aggregating under ambient growth conditions (323).

FtsH, HtrA, and small HSPs (sHSPs).

FtsH is a membrane-bound HSP with dual chaperone-protease activity (Fig. 5). The FtsH protein contains transmembrane segments in the N-terminal region and a main cytosolic region consisting of an AAA ^{ATPase and a Zn metalloprotease domain (}324). The functional roles of FtsH in LAB have not been widely studied. FtsH plays roles in protein quality control during heat shock in LAB, such as Lc. lactis (325), O. oeni (326), and Lb. plantarum (327). The FtsH proteins of Lc. lactis and O. oeni are both able to complement the growth defect of an E. coliftsH mutant at 37°C, indicating that FtsH from these LAB can at least partially functionally replace E. coli FtsH. Inactivation of ftsH resulted in salt-, heat-, and cold-sensitive phenotypes of Lc. lactis (325), whereas reduced salt and heat tolerance as well as diminished biofilm formation were reported for an Lb. plantarumftsH mutant (176). Lb. plantarumftsH was shown to be a novel member of the CtsR regulon (327). The mechanisms controlling ftsH expression in other LAB have not been studied.

HtrA (high temperature requirement protein A), also known as DegP, is involved in degradation of extracellular polypeptides and misfolded proteins (Fig. 5). HtrA contains a chaperone and a proteolytic domain. An Lc. lactishtrA mutant is thermosensitive, and HtrA appears to play a fundamental role in the degradation of abnormal exported proteins (328). The htrA gene has been identified as a heat shock gene in Lactobacillus helveticus (329), and elevated HtrA levels have been detected in Lb. casei in response to phenolic compounds (330). However, the regulatory mechanism behind the stress-inducible expression of htrA in LAB remains to be elucidated.

The members of the sHSP family function as folding chaperones, assisting the protein folding process by stabilizing unfolded or partially folded proteins without actively promoting their remodeling (331). Furthermore, sHSPs appear to play an important role in maintaining membrane integrity, especially under stress conditions (332). The sHSP family is widely conserved, and its members have been identified in several LAB (302). In O. oeni, an sHSP named Lo18 was identified as a major stress protein whose synthesis is greatly increased under various conditions, including ethanol shock (333). Lo18 has been suggested to play a dual role in O. oeni by preventing protein aggregation and stabilizing the cytoplasmic membrane (334).

Proteins induced by cold shock.

A sudden temperature downshift induces a cold shock response, during which growth of cells is first arrested for a period before reinitiation at a new and much reduced rate. The cellular effects of cold shock include the following: (i) a decrease in membrane fluidity, (ii) a reduced efficiency of mRNA translation and transcription through stabilization of secondary structures of DNA and RNA, (iii) ineffective or slow protein folding, and (iv) hindered ribosome function (335). In cold-shocked E. coli, overall transcription and translation are slowed considerably, and a set of approximately 26 cold shock genes is preferentially and transiently expressed (336). Members of the CSP family enhance the expression of cold shock-inducible genes by destabilizing nonproductive secondary structures in mRNA at low temperatures and therefore are often called RNA chaperones (336). While cold and heat shocks are considered highly separate processes in E. coli and B. subtilis (337), recent evidence indicates that these responses are interconnected in LAB (338). Cold shock of S. thermophilus is accompanied by increased production of GroEL and ClpL (296, 307). Furthermore, the clpL knockout mutant of S. thermophilus is cold sensitive (307). Cold shock conditions increase the synthesis of DnaK and GroEL in Leuconostoc mesenteroides (288) and of sHSPs in Lb. plantarum (339). Transcriptional analyses have indicated that clpX is induced by both heat and cold in Lc. lactis (340). The proposed cross talk between cold and heat shock responses in LAB is further supported by the improved cryotolerance seen after heat treatment of Lb. johnsonii and Lc. lactis (341, 342). Similar findings may link acid and cold stress responses in LAB; for Lb. delbrueckii subsp. bulgaricus, an increase in freezing tolerance was observed after acidification of the growth medium (343).

Spx governs the response to oxidative damage of macromolecules.

Spx is a global transcriptional regulator of oxidative stress in several Gram-positive bacteria and plays important roles as both a positive and a negative modulator of gene expression (344). The LAB Spx homolog TrmA (now named SpxA2) was first identified via mutations that alleviate the heat-sensitive phenotype of clpP mutants and recA mutants of Lc. lactis (245, 345). Proteolysis of Spx is mediated by the ClpXP protease system, and Spx accumulates to high levels in clpX or clpP mutants (304). Studies of B. subtilis Spx activity revealed that it interacts directly with the α subunit of the RNA polymerase (RNAP) and thereby controls global transcription initiation, either negatively or positively, by a unique mechanism not involving initial contact between Spx and DNA (344). Negative regulation is mediated by binding of Spx to Tyr263 in the C terminus of the RNAP α subunit. This residue is highly conserved among RNAP α subunits of Gram-positive bacteria, and the binding blocks interaction between other transcriptional activators and RNAP (344). Lc. lactis encodes seven Spx paralogs, among which SpxB has been shown to be involved in the cell wall stress response. DNA microarray work with spx mutants confirmed the function of Spx as a major oxidative stress regulator (346, 347). Streptococci harbor two bona fide Spx regulators, whereas enterococci have only one (346,–349).

Repair of stress-induced DNA damage.

According to the current DNA damage paradigm, cells undergo an inducible process called the SOS response during genotoxic stress. This response triggers the expression of genes involved in DNA replication, repair, and mutagenesis (350). In the case of severe DNA damage, high-fidelity DNA polymerase cannot replicate over the incurred DNA lesions and is replaced by a low-fidelity DNA polymerase(s). The process, called translesion synthesis, is performed mainly by SOS-induced error-prone polymerases of the Y family (Pol IV and Pol V) or the C family (DnaE) (350). The SOS response is repressed under physiological conditions by binding of the LexA repressor protein to operator sites in the promoter regions of SOS genes, including its own gene, lexA (350). While the bacterial DNA damage response is one of the most extensively studied stress responses, it has not been studied widely in LAB. LexA regulons have been identified in some LAB species, such as Lb. plantarum (351) and E. faecalis (352). The uvrA gene, which encodes a nucleotide excision repair protein and is a typical member of LexA regulons, has been characterized for S. mutans (353) and Lb. helveticus (354). An S. mutansuvrA mutant, in addition to being sensitive to DNA damage, was also sensitive to acidic conditions, indicating that UvrA plays a role in the repair of acid-induced DNA damage (353). Other DNA repair activities that have been implicated in the resistance of S. mutans to acidic pH include those of the Smx nuclease (355, 356). The S. pyogenes C family DNA polymerase DnaE was shown to be highly error-prone and produced frameshift and point mutations during replication of both damaged and undamaged DNAs in vitro (357). In S. uberis, the Pol V subunit UmuC is SOS induced and essential for UV-induced mutagenesis (49).

PROTECTING THE CELL ENVELOPE