Mechanisms of bacterial immunity, protection, and survival during interbacterial warfare

Abstract

Most bacteria live in communities, often with closely related strains and species with whom they must compete for space and resources. Consequently, bacteria have acquired or evolved mechanisms to antagonize competitors through the production of antibacterial toxins. Similar to bacterial systems that combat phage infection and mechanisms to thwart antibiotics, bacteria have also acquired and evolved features to protect themselves from antibacterial toxins. Just as there is a large body of research identifying and characterizing antibacterial proteins and toxin delivery systems, studies of bacterial mechanisms to resist and survive assault from competitor’s weapons have also expanded tremendously. Emerging data are beginning to reveal protective processes and mechanisms that are as diverse as the toxins themselves. Protection against antibacterial toxins can be acquired by horizontal gene transfer, receptor or target alteration, induction of protective functions, physical barriers and other diverse processes. Here, we review recent studies in this rapidly expanding field.

Introduction

Just like the battles that occur when higher organisms compete for resources and habitable space, bacteria also produce an abundance of diverse weapons to inhibit, injure, and kill competitors. Also like higher organisms, the implementation of a new weapon is often met with the development, acquisition or evolution of defenses that protect the target population. Several review articles have focused on bacterial antagonism and the diverse arsenal of weapons that bacteria produce to inhibit competitors ^1-4. In this article, we review recent literature related to how bacteria subvert or defend against competitors armed with antibacterial peptides and proteins and from specialized toxin secretion systems.

The toxins that bacteria produce to inhibit competitors are extremely diverse. Some are one component peptides and proteins, which may be post-translationally modified, and are secreted extracellularly. Diffusion of these toxins allows for targeting of bacteria that may be somewhat distant from the toxin-producing cell. Other extracellularly secreted and diffusible toxins include complex multicomponent proteins such as phage-like tailocins. Diffusible toxins (those released from cells and not delivered via contact-dependent mechanisms) must bind the bacterial surface where they may oligomerize and form a pore, interact with an essential surface protein to inhibit its activity, be transported into the cell to reach its target, or puncture the cell membrane. Bacteria have also evolved elaborate, contact-dependent nanomachineries such as Type IV secretion systems (T4SS), Type VI secretions systems (T6SS) and Type VII secretion systems (T7SS, also known as ESX) that directly inject toxins into competitors. Both contact-dependent and contact-independent mechanisms of toxin delivery have their strengths and vulnerabilities depending on the ecological circumstances. For example, the T6SS directly delivers toxins into target cells bypassing the need for a surface receptor, however, these cells open themselves up to direct counterattacks, and these toxins cannot be effectively delivered over larger spatial distances.

Antibacterial toxins have an extensive range of cellular targets, affecting molecules in all major cellular compartments. The targets are often proteins that perform essential functions or are other critical macromolecules such as peptidoglycan, nucleic acids, and phospholipids ⁵.

The interest in and discovery of antibacterial toxins has increased tremendously in the last decade largely due to the search for next generation antibacterials to thwart the increasing rise in antibiotic resistant pathogens. Another area of interest has been to better understand how various antibacterial toxins and toxin secretion systems influence the ecology of bacterial communities, the ability of pathogens to invade into microbial communities to initiate disease, and their ability to allow members of a particular microbiota to resist such pathogen invasions. Equally important to each of these missions is an understanding of how bacteria acquire immunity, evolve resistance, or mount defenses that nullify or reduce the effectiveness of these toxins. In this review, we focus on the growing body of literature related to the various bacterial defenses that thwart the effectiveness of toxins produced by competitors. Some of these defense mechanisms provide complete immunity, while others provide partial protection and are induced when a threat is detected. As this field is vast, we do not cover all toxin systems and all mechanisms, but rather provide examples of various systems and the ecological consequences of the resistance phenotypes, which often have collateral effects. We divide these protective or defensive mechanisms into several categories: immunity due to gene acquisition, antagonistic silencing, alterations to receptors and targets, responses mounted upon sensing an attack, protection by physical barriers, and other protective mechanisms.

I. Acquisition of immunity genes

Just as bacteria acquire resistance to antibiotics by horizontal gene transfer, resistance to antibacterial toxins can similarly be acquired. In most described antibacterial toxin systems, from bacteriocins to T6SS toxic effectors, the toxin genes are present adjacent to a gene encoding its cognate immunity protein that protects the cell from self-intoxication (Fig. 1A). These immunity proteins typically either bind the toxin preventing it from engaging its target or inactivate the toxin. However, immunity may also involve active transport of the toxin out of the cell ⁶. There are many examples of bacterial genomes containing “orphan” immunity genes, i.e. an immunity gene without the cognate toxin gene. In this section, we provide examples of acquired immunity by horizonal transfer of immunity genes.

Figure 1. Acquisition of immunity genes to T6SS toxins or genes for T6SS transcriptional repression.

A. Vibrio cholerae, Bacteroidales, and other bacteria can acquire immunity genes and integrate them into their chromosome providing immunity to toxins of T6SS systems of other strains and species. Cognate effector-immunity genes are typically encoded adjacently and can be acquired together as demonstrated for V. cholerae, allowing for an increase in both antagonistic potential and protection (bottom left). Immunity proteins typically bind and inhibit toxins, as shown for periplasm acting toxins (top left inset). B. Horizontal gene transfer on MGEs can shut down transcription of the T6SS loci of both A. baumannii and B. fragilis. In A. baumannii, a TetR family transcriptional repressor is encoded on a multi-drug resistant mobile plasmid that represses transcription of the T6SS allowing conjugal transfer of the plasmid to other strains. In B. fragilis, the acquisition of an integrative and conjugative element containing the GA1 T6SS locus transcriptionally silences the GA3 T6SS locus via a TetR family repressor encoded on the ICE, preventing B. fragilis from antagonizing other Bacteroidales species with its GA3 T6SS and changing it to a strain that fires only the GA1 T6SS. Created with BioRender.com.

In many naturally competent bacteria such as Streptococcus spp. ^{7, 8} and V. cholerae ⁹, expression of bacteriocins or T6SSs is co-regulated with bacterial competence systems ^{7, 8, 10}. This co-regulation may allow the aggressor to benefit from the uptake of DNA released by the killed sensitive cells that, when recombined into the genome, can repair genes and allow the recipient to acquire new, potentially advantageous functions. In T6SS-producing V. cholerae, the uptake of DNA released from T6SS-killed cells can optimize T6SS efficacy through selection of optimized effector immunity gene pairs ¹¹, or increase the repertoire of effector/immunity pairs, which increases a strain’s antagonistic ability as well as its protection to a wider array of toxic effectors (Fig. 1A). Some V. cholerae genomes contain arrays of orphan immunity genes at the 3’ end of some T6SS loci that protect against toxic effectors beyond those of the strain’s own T6SS ¹² (Fig. 1A). Genome sequencing of aquatic isolates of V. cholerae further revealed the presence of orphan T6SS immunity genes in non-pathogenic environmental strains ¹³. Similarly, the plant pathogen Pantoea agglomerans pv. betae was shown to contain arrays of orphan T6SS immunity genes and effector immunity gene pairs outside of T6SS loci, allowing for competition with and protection from other plant-associated bacteria ¹⁴. A recent genomic analysis of 54 strains of E. coli, Salmonella enterica and Klebsiella pneumoniae identified numerous mechanisms of resistance to small peptide microcins, one of which was the acquisition of immunity genes predicted to be transferred between species by horizontal gene transfer ¹⁵.

In addition to uptake of immunity genes by transformation, orphan immunity genes of some bacteria are present on mobile genetic elements (MGEs) that are transferred between strains by conjugation. For example, some human gut Bacteroidales species contain arrays of immunity genes on mobile genetic elements termed acquired interbacterial defense (AID) systems ¹⁶ (Fig. 1A). Many of these T6SS immunity gene arrays are contained on integrative and conjugative elements (ICE) that are transferred between Bacteroidales species. One such element contains genes encoding 31 predicted immunity proteins to various antibacterial toxins. The transfer of an AID was shown to protect a B. fragilis strain from antagonism by another B. fragilis strain whose T6SS contained the cognate effectors to the AID encoded immunity proteins ¹⁶. Whether strains that acquire these AID regions receive a fitness benefit and therefore will dominate the population and eventually replace the non-transconjugants is likely dependent on the composition of the particular ecosystem, the presence of cognate toxins to the various immunity proteins in co-colonizing species, and any other fitness determinants - positive or negative - encoded on the MGEs.

Salmonella Typhi and Salmonella Paratyphi each have regions designated as SAIDI for Salmonella acquired interbacterial defense islet ¹⁷ (Fig. 1). In S. Typhi eight of the 14 genes of the islet are predicted to encode immunity genes. These immunity genes include those to toxins of T6SSs as well as to contact dependent growth inhibition (CDI) toxins ¹⁷. The SAIDI islets contain a terminal gene encoding an InsA IS1 family transposase, indicating these islets may constitute an MGE. Similarly, the genomes of some E. coli and Dickeya dadantii strains contain orphan CDI toxin/immunity gene pairs ¹⁸, again increasing both antagonistic potential as well as protection to a greater diversity of toxins. These gene pairs are similarly associated with transposase-like genes and may be components of MGEs.

Above we show examples of genetic material encoding protective functions transferred between bacteria by transformation or conjugation. Lysogenic phage also transfer genetic material between similar bacteria co-colonizing an ecosystem. Although prophage genomes have been shown to contain antibiotic resistance genes that can be transferred to new hosts ^19-22, to our knowledge, there are currently no reports of lysogenic phage genomes encoding immunity or protective proteins to antibacterial toxins, although perhaps these are yet to be identified.

II. Transcriptional silencing of antagonism by horizontal DNA transfer

One of the best means of protection from the antagonistic assault of competitors is to prevent the weapon from being deployed. There are two examples of T6SSs that are transcriptionally silenced by MGEs (Fig. 1B). Some strains of the nosocomial pathogen A. baumannii harbor a mobile conjugative multidrug resistance (MDR) plasmid. Many strains containing this plasmid also have a chromosomally encoded T6SS. It was shown that a TetR family transcriptional repressor encoded on the conjugative plasmid inhibits transcription of the T6SS locus ²³ (Fig. 1). As both T6SS antagonism and conjugal plasmid transfer require cell-cell contact, the silencing of the T6SS allows for transfer of the MDR plasmid into a new A. baumannii strain so that it is not antagonized by the donor’s T6SS during contact ²³. Although this mechanism allows for continued MDR plasmid transfer, it also protects sensitive strains from the plasmid containing strain’s T6SS. This mechanism, however, leaves the plasmid containing strain without its weapon and susceptible to T6SS attack by non-MDR plasmid-containing strains. Transfer of this MDR plasmid with its T6SS silencing property to other strains also depends on the frequency of two distinct A. baumannii strains coming into contact, for which epidemiological data are still lacking.

A frequent conjugal transfer event that mediates T6SS inactivation occurs between the gut symbiont B. fragilis and coresident Bacteroidales species. T6SSs are prevalent in gut Bacteroidales species, where three different genetic architectures of T6SS loci have been described (GA1, GA2 or GA3) ²⁴. The GA3 T6SS is present only in B. fragilis and antagonizes nearly all Bacteroides, Phocaeicola and Parabacteroides species that do not encode specific immunity proteins to the cognate toxic effectors ^{25, 26}. The GA1 and GA2 T6SS loci are contained on ICE ²⁴ and are among the most conjugally transferred elements in the human gut of industrialized populations ²⁷. Approximately 25% of B. fragilis strains with a GA3 T6SS also have a GA1 T6SS containing ICE in their genome ²⁷. In these strains, a TetR transcriptional regulator encoded on the GA1 ICE represses transcription of the GA3 T6SS, allowing only antagonism by the GA1 T6SS ²⁸ (Fig. 1B). Therefore, the transfer of the GA1 ICE into B. fragilis from Bacteroidales strains that are co-resident in the human gut microbiota prevents the B. fragilis strain from antagonizing these co-colonizing strains and changes the allegiance of the B. fragilis strain so that it now defends the ecosystem communally with the other GA1 T6SS containing Bacteroidales species ²⁸.

III. Alterations to receptors and targets

In addition to acquiring DNA to gain immunity or nullify antagonism, bacteria also adapt to interbacterial warfare with the selection and amplification of resistant mutants. Escape mutants have been noted since the early work on colicin biology ^{29, 30} and have been documented to arise to most diffusible antibacterial toxins. As diffusible toxins exert their effects by binding to the cell surface and either mediating their effects there, or hijacking internalization systems to gain access to internal cellular compartments, resistant mutations often occur in genes encoding surface receptors. In broth or plate assays, such mutations often arise readily, as do mutations in processes related to receptor regulation ³¹, toxin import ^{2, 32}, and internal target modification. As most escape mutants are typically selected in the lab, they are not subject to the selective pressures of their ecosystem where they often have severe fitness defects. Indeed, if resistant mutations do not incur a fitness cost, it would reason that the resistant variant would become dominant in natural ecosystems especially under toxin selection, and the toxin would soon be rendered ineffective.

The ecological dynamics and trade-offs supporting co-occurrence of mutants resistant to colicin toxins was demonstrated more than two decades ago and named for the game rock-paper-scissors ³³. E-colicins typically bind outer membrane transporters or porins which help facilitate their entry. For colicins E1 and E2, the btuB gene encoding the vitamin B₁₂ outer membrane transporter serves as the receptor ^{34, 35}. Studies showed that isogenic colicin producing, sensitive, and resistant strains are all able to co-exist in the mouse gut due to fitness defects and advantages of all three strains ³⁶. A mutation to btuB provides resistance to colicins, but the mutation renders it less fit; the colicin producing strain has an advantage over sensitive strains but incurs a fitness defect as it must lyse to release the toxin; the sensitive strain is killed by the colicin but does not incur the fitness disadvantages of autolysis or reduced vitamin acquisition. The fitness defect of the toxin producing strain, however, would not be expected for most diffusible toxins that are actively secreted from viable cells rather than released by lysis. Indeed, many of the mutations that arise in the lab are not detected in wild strains, even in the presence of the toxin. For example, the CDI system of E. coli EC93 requires the essential outer membrane protein BamA for cell-cell interactions ³⁷. The interaction with BamA lies in extracellular loops 6 and 7 and swapping E. coli BamA with those of other Enterobacteriaceae species that have differences in these regions confers resistance to CDI killing ³⁸ (Fig. 2). However, analysis of BamA sequences from hundreds of E. coli strains revealed that these extracellular loop regions are identical and do not contain resistant variants, suggesting that if such mutants do arise, they are less fit and not selected. A similar scenario was observed with the bacteriodetocin-A (Bd-A) toxin of Bacteroidaceae ³⁹ that also binds BamA and inhibits its function ⁴⁰. Several point mutations affecting single amino acids in the extracellular loops of BamA that prevent Bd-A interaction are readily selected in the lab (Fig. 2). The bacteroidetocins are unique peptide toxins in that they kill their producing strains due to lack of immunity proteins. Remarkably, Bd-A producing strains do not have altered BamA sequences, despite tremendous selective pressure for such mutations. In addition, no BamA alterations that resulted in Bd-A resistance were detected in strains of any species that are sensitive to Bd-A, including strains that were co-resident in the human gut with Bd-A producers ⁴⁰. Therefore, while resistant mutations to genes encoding conserved, essential proteins arise in the lab setting, they are rarely present in wild strains.

Figure 2. Changes to targets or receptors that prevent antagonism.

Changes to receptors or targets in the outer membrane, such as LOS/LPS, BamA, or other outer membrane proteins, that prevent toxin binding. Alterations in the extracellular loops of BamA can prevent intoxication by the CDI system of E. coli. Similarly, single point mutations in BamA of Bacteroidales species prevent binding and toxin activity of bacteroidetocin A.(Bd-A). Replacement of β-barrel OMPs with a functional equivalent, or alteration of the LOS/LPS glycan can render Bacteroidales strains resistant to BSAP pore-forming toxins. A single amino acid variant of the periplasmic PPIase renders B. fragilis strains resistant to the BfUbb toxin. Modification of the peptidoglycan of A. baumanni with the noncanonical d-amino acid d-lysine prevents degradation by PG targeting toxins. Created with BioRender.com.

For a minority of antibacterial toxins, resistant mutants may not be less fit in their natural communities. BfUbb is a peptide toxin similar in sequence to eukaryotic ubiquitin and is produced by a subset of B. fragilis strains ⁴¹. There are numerous B. fragilis strains that do not have the toxin gene but are naturally resistant. It was recently shown that the target of BfUbb is an essential periplasmic peptidyl-prolyl isomerase (PPIase) ⁴². BfUbb producing and resistant strains all have a single amino acid difference in the PPIase, a glutamic acid at amino acid 199 rather than a tyrosine, that renders strains resistant to the toxin (Fig. 2). Therefore, producing strains do not require an immunity protein for protection. Our analyses reveal that of the 641 B. fragilis genome sequences in the NCBI database, 390 have the resistant glutamate at amino acid position 119, of which, only 86 also contain the BfUbb encoding gene. Therefore, as the resistant PPIase variant is dominant in wild B. fragilis strains, it likely does not confer a fitness defect, although this has not been demonstrated experimentally. In such a scenario, it may be expected that the number of BfUbb resistant B. fragilis strains will continue to increase during toxin selection, potentially rendering the toxin ineffective over time.

Another form of resistance can occur when receptors or targets are replaced or modified. The BSAP membrane attack complex domain-containing toxins of the Bacteroidales target one of two distinct molecules: outer membrane β-barrel nutrient import proteins ^{43, 44} or the glycan component of the lipopolysaccharide/lipooligosaccharide (LPS/LOS) ^{43, 45} (Fig. 2). Present adjacent to the bsap genes are receptor replacement genes (functionally orthologous protein receptors or new glycosyltransferase genes) that alter the receptors, so they no longer bind the toxins. For the majority of BSAP toxins, only the toxin producing strains have the receptor replacement, with toxin gene negative strains encoding the sensitive variant, however, in the case of BSAP-4 of B. fragilis, resistant β-barrel outer membrane protein (OMP) receptor genes are found in some strains lacking the toxin gene ⁴⁴, making receptor replacement an additional mechanism for a sensitive stain to acquire resistance without incurring a fitness defect (Fig. 2).

In addition to target/receptor mutation and replacement, some bacteria protect themselves from antagonism by target modification. Some bacteria alter their peptidoglycan ⁴⁶, especially when entering stationary phase. Such modifications allow bacteria to avoid T6SS toxic effectors that specifically target this critical cellular structure. Species of Acetobacteria produce amidated muropeptides with unique peptide cross-linkages that are predicted to protect these bacteria from the T6SS toxic effectors of co-habitating environmental organisms such as P. aeruginosa and Acidovorax citrulli ⁴⁷ (Fig. 2). A. baumannii protects itself against PG-targeting T6SS effectors by modifying its PG with the noncanonical d-amino acid d-lysine ⁴⁸ (Fig. 2). While this protects A. baumannii from T6SS mediated killing, it also renders it less pathogenic, reflecting fitness trade-offs common to phenotypes that protect against antagonism.

IV. Mechanisms activated or induced upon attack

The mechanisms of immunity or resistance discussed above occur mostly at the DNA level by the acquisition or mutation of DNA. However, bacteria can also mount active responses to interbacterial attacks by sensing and mounting a counterattack, or by altering gene expression to mount a protective response, similar to processes described in response to antibiotics ⁴⁹. In an opinion by Cornforth and Foster, the authors suggest that bacteria sense competitors through detecting changes in nutrient availability and by detecting signs of cellular damage from antagonistic attacks in a process they call competition sensing ⁵⁰. This hypothesis posits that bacteria will mount a counterattack as part of a stress response to nutrient depletion or cellular damage, but that bacteria can also mount defensive responses without counterattack. LeRoux et al. later proposed a process of danger sensing whereby bacteria respond to specific external danger signals via receptors that lead to a signal transduction mediated response ⁵¹. In this model the danger signals themselves do not need to harm the bacteria that receive them. Westhoff et al. further expanded on the danger sensing hypothesis, showing that not only do bacteria respond to danger signals, but the type of signal (which corresponds to the distance between the cell and the impending threat) determines the type of response ⁵². Competition and danger sensing can lead to three different responses – preemptive actions, counterattack, or induction of a protective response.

IVa. Counterattack and Preemptive Actions-

Counterattack is a form of defense that has been demonstrated in several bacteria. A specific retaliation against aggression known as tit-for-tat was demonstrated in P. aeruginosa in response to an attack by the T6SS weapon of V. cholerae ⁵³ (Fig. 3, right). The V. cholerae T6SS stab initiates rapid P. aeruginosa T6SS assembly and firing, which can effectively retaliate with a more aggressive (multiple stabs) attack. The tit-for-tat response is induced by the V. cholerae TseL phospholipase effector ⁵⁴. Similar retaliation was shown to occur in response to T4SS-mediated DNA conjugation, likely due to membrane perturbation ^{53, 55, 56}. T6SS antagonism can also be stimulated when cells detect kin cell lysis ⁵⁷, or membrane disruption in general. A similar phenomenon has also been observed in Serratia marcescens, which fine tunes transcription of its T6SS genes to match the apparent degree of the threat it faces ⁵⁸.

Figure 3. Active responses to antagonism.

When facing direct competition, bacteria will sometimes retaliate by firing their T6SS or by secreting antimicrobial molecules (right). Bacteria can eavesdrop on potential competitors by sensing secreted molecules such as quorum sensing autoinducers, and mount a preemptive attack (left). P. aeruginosa can sense nearby lysed kin cells and activate a protective response involving three loci, Arc1-3. Arc-3A encodes for a membrane protein that protects against phospholipase toxins (right). Phocaeicola vulgatus senses outer membrane stress from toxins including BcpT and Bd-A through the anti-sigma factor Reo, releasing the EcfO sigma factor that transcribes its regulon leading to the synthesis of LPS with protective long O-antigens (left, top). Created with BioRender.com.

Bacteria that produce toxins such as colicins and tailocins, which are only released upon cell lysis, are examples where the population must strike a balance between the need to protect against competition while also minimizing the cost of autolysis ^59-61. E. coli have long been known to release colicins in response to the SOS stress response mediated by the cleavage of the transcriptional repressor LexA. Such a response is readily induced by DNA-targeting toxins and is therefore a form of competition or danger sensing. In addition, a small subset of a population of E. coli lyse to release colicins to preemptively protect the remaining bacteria against competitors. In such preemptive attacks, colicins were shown to be released from less fit cells, i.e. those unlikely to divide ⁶², thereby minimizing fitness costs.

Some bacteria are able to “eavesdrop” on neighbors, whereby they sense molecules secreted by other bacteria that they themselves do not produce. In some cases, eavesdropping leads to the production of toxins, consistent with the danger sensing principle (Fig. 3, middle). Chromobacterium violaceum detects the acyl-homoserine lactone quorum-sensing (QS) molecule of Burkholderia thailandensis which induces the production of antimicrobial molecules ⁶³. A similar behavior occurs in Streptococcus pneumoniae where strains have QS receptors for non-self QS peptides allowing for the induction of a preemptive response to competitor density ⁶⁴. Lactobacillus gasseri EV1461 and Lactobacillus plantarum secrete bacteriocins when co-cultured with other Gram-positive bacteria ^{65, 66}, although the inducing molecule of the co-colonizing bacteria to our knowledge has not been identified. Such preemptive mechanisms provide bacteria with an early competitive edge as they prepare for an upcoming assault from a competitor.

IVb. Induced protective responses -

Many bacteria do not have weapons to attack but are able to mount protective responses. whereas other bacteria mount protective responses in addition to a preemptive or counterattack, which, as discussed above, are also induced responses. It has long been recognized that Staphylococcus aureus is able to protect itself from low concentrations of the bacteriocin nisin by signaling through a two-component signal transduction system. This signaling induces the expression of downstream ABC transporter genes ^{67, 68} that likely transport the nisin from the cell. This signal transduction system was shown to respond to cell envelope stress ⁶⁹, likely induced by numerous diverse antibacterial stressors that affect the cell envelope.

Recent studies illustrate distinct protective responses induced by antagonism in several Gram-negative bacteria. The antibacterial toxin, BcpT, is produced mostly by the gut species P. vulgatus and Phocaeicola dorei ⁷⁰. This toxin initially interacts with the LipidA-core glycan of some Bacteroidales species (Fig. 2) resulting in growth inhibition and cell death. However, BcpT exposure induces a stress response likely triggered by outer membrane perturbation that results in the transcription of an unusual sigma factor/anti-sigma factor pair orthologous to the EcfO/Reo sigma factor/anti-sigma factor pair characterized in B. fragilis ⁷¹ (Fig. 3, left). The anti-sigma factor is unusual in that it has domains that span all cellular compartments with a large outer membrane β-barrel domain that likely senses outer membrane stress. When this regulon is activated by antibacterial toxins, the EcfO sigma factor is released from the Reo anti-sigma factor and transcribes several genes, some of which lead to the synthesis of LPS with long O-antigens ⁷⁰. Induction of this protective response confers P. vulgatus with 8-fold protection to BcpT and 100-fold protection to the pore forming toxin BSAP-3, for which the LPS also serves as receptor ⁴⁵.

Various T6SS effectors of V. cholerae induce general stress response pathways in both E. coli and P. aeruginosa. For example, E. coli is not killed when the V. cholerae peptidoglycan targeting effector TseH is delivered via the T6SS, however it is killed when this toxin is targeted to the periplasm from intracellular plasmid expression ⁷². Two systems involved in envelope stress response, the BaeSR two-component system and the Rcs system were shown to induce genes that protect E. coli from T6SS delivered toxin. The protective features include an extracellular polysaccharide and various factors predicted to mitigate cellular damage ^{54, 72}.

P. aeruginosa induces a general response upon sensing the lysis of kin cells through the Gac/Rsm global regulatory system, a form of danger sensing ⁵⁷. This response was shown to include activation of three different resistance gene clusters, termed Arc, which are involved in protection to toxins delivered by T6SSs ⁷³ (Fig. 3, left). The Arc1 cluster is predicted to play a general role in defense, whereas Arc2 and Arc3 protect against distinct toxic effectors. The Arc2 cluster protects against the toxic effector ColA of Burkholderia thailandensis, predicted to form pores in the inner membrane. The Arc3 response was shown to specifically protect against T6SS-delivered phospholipase effectors, including Tle3, that lead to the accumulation of toxic lysophospholipids. A protein with multiple membrane spanning domains encoded by the Arc3 cluster was shown to provide the majority of the protection to these toxins and was predicted to either sequester the lysophospholipids or somehow neutralize the phospholipase ⁷³. This example of defenses mounted to specific classes of toxic effectors suggests that such toxins are frequently encountered by P. aeruginosa in its natural environments.

V. Protection conferred by physical barriers

In addition to the responsive mechanisms described above, several studies show that bacteria produce physical barriers that defend against antagonism. The production of extracellular polysaccharide (EPS) by V. cholerae was shown to be a type of “protective armor” against contact-dependent T6SS attacks from numerous species of bacteria, including different strains of V. cholerae ⁷⁴ (Fig. 4A). Importantly, EPS production did not prevent the EPS-producing strain from antagonizing using its own T6SS, leading the authors to conclude that the EPS repels T6SS attacks by functioning as an impenetrable barrier rather than by simply creating physical distance between cells. A subsequent study including both modeling and experimental work expanded upon this finding. A community containing a T6SS-producing Acinetobacter baylyi and two variants of T6SS-sensitive E. coli (EPS producing and non-producing) demonstrated that EPS secretion not only protects individual cells, but it can also provide collective defense, as groups of EPS secretors protect each other and also protect vulnerable, non-EPS-producing cells ⁷⁵. An earlier study of P. aeruginosa demonstrated that the mere presence of competitors, as well as active assault from secreted tailocins known as pyocins, stimulates production of a protective biofilm ⁷⁶ (Fig 4B). These studies show that both contact-dependent and contact-independent modes of antagonism can be subverted to some extent by these barriers.

Figure 4. Other forms of bacterial defense or survival.

A. V. cholerae can produce a defensive exopolysaccharide layer that prevents T6SS antagonism by competitors but still allows aggression using its own T6SS. B. Biofilms protect bacteria from many stressors including attacks from pyocins, T6SS, and diffusible toxins. C. A boundary of dead bacterial bodies can act as a defensive barrier protecting kin cells on the other side from T6SS antagonism. D. Slow growth or persister cells (dark cell in population) can survive antagonistic attacks from pyocins and T6SSs. Created with BioRender.com.

A recent study showed the same protective property of physical barriers by a different molecular mechanism. The opportunistic food-borne pathogen Cronobacter malonaticus antagonizes E. coli using its T6SS ⁷⁷. A screen of the E. coli Keio library identified a fimE mutant that rendered E. coli four-log more resistant to T6SS killing ⁷⁸. FimE is a regulator of type 1 fimbriae expression and its deletion resulted in overexpression of fimbriae leading to the formation of microcolonies. These microbial aggregates protect E. coli against T6SS assault, similar to biofilm-mediated protection (Fig. 4).

Other studies using both experimental and modelling analyses showed that T6SS attacks can lead to a barrier of bacterial corpses that block the aggressor’s path ⁷⁹ (Fig. 4C). This method of protection is especially effective if the T6SS effectors lyse cells slowly so that the bodies of slain bacteria build up, preventing further attacks and shield sensitive bacteria on the opposite side of the barrier.

VI. Survival due to slow growth or persister cells

Bacterial populations, especially at stationary phase, typically include dormant or slow growing bacteria known as persisters. These bacteria are known to survive antibiotic exposure but have also been shown to survive attacks from antibacterial toxins of competitors (Fig. 4D). The phage-like tailocin of Pseudomonas syringae. pv. syringae B728a, a plant pathogen, antagonizes other P. syringae strains. Upon exposure of sensitive strains to tailocins, two surviving populations arose, those with genetic changes conferring complete resistance, and persisters without genetic changes, that exhibit transient survival to tailocins ⁸⁰. The mutants that gained full resistance due to mutation were 100-fold less fit for plant leaf infiltration whereas the persistent population did not suffer a fitness defect. Therefore, persistence allows bacteria to survive antibacterial assault without the fitness defect incurred from mutational resistance. A recent study added support for the protective properties of the slow growth phenotype. An analysis of E. coli mutants showed that those conferring slow growth were able to survive the T6SS effector, Tae1, of P. aeruginosa that targets the cell wall ⁸¹. Therefore, due to bacterial population heterogeneity for the slow growth phenotype, it is likely that there will be bacteria that are able to persist during antagonistic attacks.

Perspective

Here, we reviewed research addressing the various modes that bacteria use to protect themselves against bacterially-produced toxins including acquired immunity, defensive and protective mechanisms to antibacterial diffusible peptide and protein toxins or to those delivered by the T6SS. Bacteria often acquire immunity to specific antibacterial toxins via mobile genetic elements and in some instances, by acquisition of transcriptional repressors that can shut down antagonistic systems in would be attackers. Studies are beginning to uncover the complex response pathways that are activated either preemptively when competitors are in close proximity, or as a result of a direct assault. Continued research is needed to explore how these responses are triggered, and how the activated regulon protects and potentially repairs injured cells. Bacteria can also behave in a coordinated fashion to evade or protect themselves from predation. Biofilms, which allow for resources sharing and other cooperative functions such as protection from antibiotics, also protect against antibacterial toxins and toxin secretion systems. An important consideration in the effectiveness of both the weapons and the defenses is the spatial architecture of a given microbial community. Diffusible toxins can be effective over larger distances compared to toxins delivered via contact-dependent mechanisms. For both types of toxin delivery modes, the effectiveness will be greatest among bacteria that share a nutritional or spatial niche. Also critical to these interactions is the dispersal properties of diffusible toxins, which are not well studied in natural communities. Studies into both these areas will greatly augment and provide a more detailed ecological understanding of these antagonistic and protective functions. Aside from the ecological knowledge to be gained by studying antagonism and defense, there are also translational applications of these studies. Just as antibacterial toxins are being considered as next generation therapeutics in this age of increased antibiotic resistant bacteria, the immunity, protective, and defensive features of these bacteria also have translational utility. For example, numerous diverse live biotherapeutics (genetically modified probiotics) are in various stages of development ⁸² as are consortia of synthetic communities to treat various maladies and/or restore a healthy microbiota ⁸³. Genetic modification or selection of strains that are best able to survive the antagonistic weapons of competitors should help facilitate strain engraftment, a prerequisite to deliver the health promoting effects provided by live biotherapeutics.

In conclusion, the protective responses reviewed here are as diverse as the weapons they protect against and likely are only the beginning of discovery of the defensive side of interbacterial warfare.