(815e) The Toxin Antitoxin Persistence (TAP) Model for Revealing Insights Into Persister Cell Formation in Escherichia Coli

Toxin-antitoxin (TA) systems consist of a toxin which is always a protein and an antitoxin which is a protein or mRNA. Many TA systems are active in biofilms and serve to convert cells to a dormant state by reducing metabolism. While in this “persister” cell state, bacteria persevere through certain environmental stresses, such as those incurred by the presence of antibiotics, without attaining resistance through an altered genotype. The exact mechanisms behind persister cell formation remain unclear, so a computational model has been constructed in an attempt to reveal key insights behind this process. Two previously described TA systems (MqsR/MqsA and GhoT/GhoS) are included in the model to explore how they interact with each other and their roles in persistence.

The MqsR/MqsA TA system is of the type II class, in which a labile antitoxin protein (MqsA) binds to a more stable toxin protein (MqsR), blocking its toxic effect in the process (Appl. Microbiol. Biotechnol. 64:515, 2004; J. Bacteriol. 188: 305, 2006; PLoS Pathogens 5:e1000706, 2009). The toxin MqsR  is an endoribonuclease that specifically cleaves at 5’-GCU sites, resulting in enhanced degradation of the total cellular mRNA pool, with the exception  of 14 mRNA lacking 5’-GCU sites (Environ. Microbiol., 2012, on-line). While the antitoxin MqsA binds to MqsR to inhibit its activity, it also represses transcription from its own operon (mqsR and mqsA are co-transcribed). During conditions of stress, MqsA is degraded more rapidly from the enhanced action of certain proteases such as Lon (Nat. Chem. Biol. 7:359, 2011), thus derepressing the mqsRA promoter and increasing MqsR concentrations. Furthermore, the mqsR portion of the transcript has only one 5’-GCU site, while the mqsA portion has three, leading to the preferential cleavage of the mqsA portion by MqsR. The one 5’-GCU site found in the mqsRportion may give MqsR negative feedback on its own expression, somewhat decreasing the risk for attaining MqsR concentrations that lead to cell death.

Production of MqsR has been shown to increase the total percentage of persister cells in an Escherichia coli population, while deletion of mqsR  has the opposite effect (Biochem. Biophys. Res. Commun. 391:209, 2010); this was the first time that inactivation of a toxin reduced persistence. Production of a version of MqsR that was engineered to be more toxic, MqsR 2-1, increases persistence by decreasing cellular fitness (Microb. Biotech. 5:509, 2012). Hence, MqsR not only mediates persistence by enhancing the degradation of the total cellular mRNA pool, but also by not cleaving specific mRNA encoding other genes that are also needed for persistence. One of these mRNA’s is that encoding ghoT, whose translated product is another toxin whose exact role in persistence has been uncertain until now.

The GhoT/GhoS toxin-antitoxin system is the first example of the type V class, in which the antitoxin protein (GhoS) specifically cleaves the toxin-encoding mRNA (for ghoT) (Nat. Chem. Biol.8:855, 2012). Like MqsR, production of GhoT increases persistence (Nat. Chem. Biol.8:855, 2012). Unlike MqsR, GhoT is not an endoribonuclease, but is a small hydrophobic protein that damages the membrane, leading to the formation of “ghost” cells, when it is overproduced (Nat. Chem. Biol.8:855, 2012).  The ghoS portion of ghoST mRNA transcript has two 5’-GCU sites, while the ghoT portion has none; hence MqsR cleaves only the ghoS portion of the transcript, enriching the ghoT portion. GhoT concentrations then become heightened during times of elevated MqsR activity.  Hence, the GhoT/GhoS TA system is the first example of a TA system controlled by another TA system (MqsR/MqsA) (Environ. Microbiol., 2012, on-line).

The computational model constructed here to simulate the formation of persister cells via the action of the MqsR/MqsA and GhoT/GhoS TA systems is deterministic and includes over 20 ordinary differential equations. Changes in growth rate due to enhanced toxin activity is taken into account, and persistence caused by toxin-antitoxin activities can be seen as reduced metabolic activity. Previously, we demonstrated conclusively that persister cells have reduced metabolism (Antimicrob. Agents Chemother. 57:1468, 2013). The model assumes that the overall protein concentration within a cell is constant, so that cells manipulate their growth rate to ensure it stays constant. Heightened MqsR activity leads to a shorter half-life of the combined mRNA pool within the cell. By assuming that total cellular translation activity is proportional to the total mRNA concentration, MqsR reduces the growth rate. GhoT is modeled as reducing translation rates indirectly. It is assumed to have the same toxic effect as that of the TisB toxin, which acts as a pore specific for channeling small anions. By this mechanism, hydroxyl ions can exit the cell at increased rates and react with protons outside of the cell. They thus have the same virtual effect as allowing the diffusion of protons into the cell as the pH of the cytoplasm equilibrates to respond to the loss of hydroxyl anions. Toxins like these then act to decouple the electron transport chain, and at the same time, the cell must find a way to keep its cytoplasmic pH neutral if it is in a more acidic environment, while its buffering capacity is finite. It may be possible that cells use ATP synthases in reverse, along with a variety of other acid stress responses, to combat this problem. The drop in ATP levels leads to a drop in biosynthetic reactions, with the most major being that of translation. This whole effect is complicated by the fact that E. colimanipulates glycolytic (and possibly overall catabolic) activity depending on its demand for ATP. The cell does not take up more carbon to refill ATP that has been lost to increased GhoT activity since it has an upper limit for its catabolic activity, especially at the onset of persistence, when its biosynthetic capacities for increasing overall catabolism by regulation of transcription/translation are hindered. With this assumption, ATP concentrations stabilize at a lower level upon heightened activity of GhoT.

The ordinary differential equations presented in this model allow us to predict the concentration of chromosomal DNA bound or unbound by MqsA, of total and specific mRNA pools, of specific proteins, and of energy-related metabolites (NADH and ATP), with the production of each protein depending on the concentration of its mRNA transcripts and also on translation rates. Two other equations are incorporated to indicate changes in growth rate and translation rates. The metabolites simulated, NADH and ATP, are major molecules produced by cellular catabolism, and their concentrations exert an influence on translation rates. The effects of each toxin on each variable (including growth and translation rates) are visualized upon the onset of a stress condition, such as introducing enhanced proteolytic activity for the degradation of MqsA.

By including the effects of two toxin-antitoxin systems that are known to interact with each other, simulations performed by this model suggest why a cell would have more than one toxin-antitoxin system, if the action of just one is needed to send it into a state of persistence. Unlike the TisB/TisA TA system, which does not depend on other TA systems to enhance activity of the toxin, the GhoT/GhoS TA system relies on MqsR levels for enhancing GhoT concentrations under physiological conditions (Environ. Microbiol., 2012, on-line). Simulations have revealed that the action of GhoT is needed to lower translation rates by reducing ATP levels, decreasing the maximum concentration of MqsR reached and reducing the risk of too much MqsR being produced at the onset of persistence. The GhoT/GhoS system thus acts as another negative feedback mechanism for the MqsR/MqsA TA system, allowing the cell to eventually awaken from a persister state instead of being stuck in a comatose condition. Such a finding is non-intuitive, as the action of a toxin is needed to subdue the action of another toxin. Experimental data to verify the model include showing that, due to its enhanced stability, the effect of production of MqsR 2-1 on persistence is even more so dependent on the action of GhoT, compared to the production of MqsR.