Hyperglycemia potentiates increased Staphylococcus aureus virulence and resistance to growth inhibition by Pseudomonas aeruginosa

ABSTRACT Diabetes is associated with several health consequences, including increased susceptibility to more frequent and severe infections. Bacterial infections associated with diabetes are typically polymicrobial, with Staphylococcus aureus and Pseudomonas aeruginosa frequently isolated from the same infection site. S. aureus and P. aeruginosa are frequently found in diabetic skin and soft tissue infections, in the lungs of people with cystic fibrosis, and in indwelling device infections. Numerous studies have investigated interactions between these two pathogens primarily using in vitro systems. These models have several limitations as they do not accurately reflect the complexities of an immune response nor the nutrient dynamics in a diabetic infection microenvironment. Here, we describe a novel murine indwelling device co-infection model that allows us to study the interactions between S. aureus and P. aeruginosa within the context of an immune response during both normal and diabetic infections. Our data shows that P. aeruginosa significantly inhibits S. aureus growth during co-infection in a normal mouse and that inhibition is not dependent on the P. aeruginosa PQS quorum sensing system. Conversely, in a diabetic co-infection, S. aureus overcomes inhibition by P. aeruginosa and this phenotype is reliant on S. aureus glycolysis. We also demonstrate that both organisms display increased virulence potential in a diabetic co-infection as we observe increased dissemination to peripheral tissues. This study revealed novel in vivo interactions between S. aureus and P. aeruginosa and advances our understanding of the complex interactions between microorganisms in polymicrobial infections in clinically relevant infection microenvironments. IMPORTANCE Individuals with diabetes are prone to more frequent and severe infections, with many of these infections being polymicrobial. Polymicrobial infections are frequently observed in skin infections and in individuals with cystic fibrosis, as well as in indwelling device infections. Two bacteria frequently co-isolated from infections are Staphylococcus aureus and Pseudomonas aeruginosa. Several studies have examined the interactions between these microorganisms. The majority of these studies use in vitro model systems that cannot accurately replicate the microenvironment of diabetic infections. We employed a novel murine indwelling device model to examine interactions between S. aureus and P. aeruginosa. Our data show that competition between these bacteria results in reduced growth in a normal infection. In a diabetic infection, we observe increased growth of both microbes and more severe infection as both bacteria invade surrounding tissues. Our results demonstrate that diabetes changes the interaction between bacteria resulting in poor infection outcomes.

Many labs use in vitro models to study the interactions between S. aureus and P. aeruginosa.However, a significant drawback to these studies is that they are in a closed system in artificial, nutrient-limiting conditions and outside of the context of the complex host immune response.Currently, few models exist that replicate host conditions during multispecies infections.Therefore, it is imperative to develop a model that accurately reflects the dynamics of S. aureus and P. aeruginosa in the context of diabetic infections that are not available in traditional in vitro systems.Unlike traditional in vitro culture conditions, our murine co-infection catheter model sustains chronic bacterial infections where nutrients are continually replenished by the host circula tory system.In addition to recapitulating infection of indwelling medical devices, this model allows us to observe these bacteria more generally within the complex infection microenvironment of diabetic tissues and in the context of a host immune response.
In addition to indwelling catheters, the interaction of the dominant pathogens, P. aeruginosa and S. aureus, is relevant to lung disease in patients with cystic fibrosis (CF) (21,(28)(29)(30).Typically, S. aureus first colonizes the lungs of children with CF and is later replaced by P. aeruginosa as the dominant pathogen during adolescence and adulthood (21,30,31).Another significant advantage to the catheter model is that we can inoculate the catheters sequentially with S. aureus and P. aeruginosa to mimic the dynamics of temporal infection typical of diabetic wounds and CFRD lung infections.A common co-morbidity of CF is the development of CF-related diabetes (CFRD).Interestingly, S. aureus and P. aeruginosa display an altered dynamic in the lungs of patients with CFRD, wherein S. aureus re-emerges as the dominant pathogen in the airways of CF patients after initially being replaced by P. aeruginosa (28,32).Furthermore, skin and soft tissue infections (SSTIs) with P. aeruginosa and S. aureus are a significant source of morbidity in individuals with diabetes (5-9, 11, 12, 33).These observations raise several important questions about the dynamics between S. aureus and P. aeruginosa and the metabolic interplay that results in altered interactions in diabetic infections.Although the pathogenicity and virulence mechanisms of S. aureus and P. aeruginosa are the subject of numerous studies, there remains a significant knowledge gap defining how S. aureus and P. aeruginosa interact in vivo and how infection dynamics change during diabetic co-infections.
In this study, we show that S. aureus growth is inhibited during co-infection with P. aeruginosa but significantly less inhibited during a diabetic infection.Additionally, we observe that both S. aureus and P. aeruginosa virulence is more severe in a diabetic host via increased dissemination into the tissue.We observed that P. aeruginosa growth is moderately inhibited during co-infection with S. aureus, which is mediated by both glucose-dependent and -independent mechanisms depending on the metabolic state of the host environment.We provide evidence that P. aeruginosa PqsR-regulated secreted factors are responsible for growth inhibition of S. aureus in vitro, and glucose availability and glycolytic function are required for S. aureus resistance to P. aeruginosa-mediated inhibition in vitro and in vivo.Interestingly, we demonstrate that P. aeruginosa inhibition of S. aureus growth during co-infection in vivo does not require PqsR-regulated factors and is instead mediated by unknown factors.Taken together, we demonstrate that our in vivo catheter model can reveal novel interactions between S. aureus and P. aeruginosa, as well as elucidate the virulence potential of these bacterial pathogens that are masked in traditional in vitro systems.

S. aureus proliferation during catheter infection is inhibited by P. aeruginosa
To determine the proliferative potential of S. aureus in the presence of P. aeruginosa during catheter co-infection, subcutaneous catheters were infected with S. aureus (1 × 10 5 CFU) alone or sequentially co-infected with S. aureus (1 × 10 5 CFU) and 4 days later with P. aeruginosa (1 × 10 5 CFU) (Fig. 1).During mono-infection, S. aureus burden increased 2-3.5 logs by day 11 post-infection.However, S. aureus burden within the catheter was significantly reduced (~20 fold on average, P < 0.0001) by day 11 when catheters were subsequently co-infected with P. aeruginosa.

The inhibitory effect of P. aeruginosa on S. aureus is lessened in a diabetic host
Co-infection of S. aureus and P. aeruginosa using a subcutaneous catheter was also performed in diabetic mice where diabetes was induced with streptozotocin at day four (Fig. 1).Diabetes was induced at day four to ensure that there was a similar amount of S. aureus in all catheters when P. aeruginosa was introduced.In mono-infection, S. aureus grew to a statistically higher burden in an infected catheter within diabetic mice than within normal mice (~3 fold, P < 0.0001).More importantly, S. aureus was able to grow to a significantly higher burden in the presence of P. aeruginosa in a diabetic host than in a normal host, >10 fold (P < 0.0001) the average observed in the catheter of normal mice during co-infection.These experiments establish that S. aureus is able to overcome growth inhibition by P. aeruginosa during co-infection in a diabetic host environment.

Inhibition of S. aureus growth by P. aeruginosa is ameliorated by glucose availability in vitro
Given the reduction of S. aureus growth observed in vivo when co-infecting with P. aeruginosa, we suspected that P. aeruginosa secreted factors may be responsible for S. aureus growth inhibition as indicated by previous studies (28,30,34,35).To model S. aureus inhibition by P. aeruginosa secreted factors in vitro, S. aureus was grown in purified spent P. aeruginosa culture supernatants, where the supernatant from an overnight culture of P. aeruginosa was filter sterilized and fortified with casamino acids (CAA) before the addition of S. aureus.S. aureus grown in P. aeruginosa culture supernatant showed minimal growth over 24-48 h incubation, significantly less than when grown in purified spent S. aureus culture supernatant (Fig. 2).We established in our in vivo model that more glucose was available in the catheter environment of diabetic mice than in the catheter environment of normal mice (Fig. S1).We, therefore, hypothesized that the hyperglycemic environment of the diabetic host was responsible for S. aureus resistance to P. aeruginosa growth inhibition.We used purified spent P. aeruginosa supernatants further fortified with glucose and an equimolar amount of CAA to grow S. aureus in vitro.
In our in vitro model, we observed significantly higher S. aureus growth in P. aeruginosa culture supernatant when the supernatant was supplemented with glucose and CAA compared to supernatant supplemented with only CAA.These experiments provided evidence that the increased glucose availability in a diabetic environment could confer resistance to S. aureus growth inhibition by P. aeruginosa secreted factors.

Glycolysis is required for resistance of S. aureus to P. aeruginosa soluble factors in vitro
Further mechanistic analysis of how S. aureus glucose metabolism confers resistance to growth inhibition by P. aeruginosa secreted factors was performed using two isogenic S. aureus mutant strains: (a) an S. aureus strain lacking four glucose transporters (ΔG4), which decreases, but does not inhibit glucose uptake altogether (15), and (b) an S. aureus strain lacking the gene for 6-phosphofructokinase (ΔpfkA) and, therefore, unable to run glycolysis after conversion of glucose 6-phosphate to fructose 6-phosphate (17).As observed with WT S. aureus, the ΔG4 mutant showed <1 log growth in P. aeruginosa culture supernatant over a 24-48 h period, and supplementation of the supernatant with glucose allowed significantly higher growth (Fig. 3A).The growth of the ΔpfkA S. aureus mutant, however, was significantly inhibited by P. aeruginosa culture supernatant, both with and without the supplementation of glucose (Fig. 3B).Interestingly, ΔpfkA growth was significantly inhibited by the presence of glucose, even when grown in spent S. aureus culture supernatant.Given that the ΔpfkA mutant has the machinery to import glucose but not fully metabolize it, we suspect that an accumulation of intracellular fructose-6-phospate induces cell death (36).From these observations, we concluded that slower glucose uptake by S. aureus did not impact resistance to P. aeruginosa soluble factors over a 24-48 h period, but that functional glycolysis was required for resistance.

Glycolysis supports S. aureus growth in catheter infection and resistance to P. aeruginosa-mediated inhibition
To further test our hypothesis that the hyperglycemic environment of the diabetic host confers resistance of S. aureus to P. aeruginosa-mediated growth inhibition, analysis of glucose uptake and glycolytic S. aureus mutants was performed in co-infection with P. aeruginosa in the catheter infection model in diabetic mice (Fig. 4).During mono-infec tion, the burden of the ΔG4 mutant within the catheter was not significantly different than WT S. aureus (Fig. 4A).In contrast, the burden of the ΔpfkA mutant was significantly less than WT S. aureus (P < 0.0001), ~2 log-fold on average.The magnitude of this difference was greatly increased in co-infection with P. aeruginosa, with the ΔpfkA S. aureus mutant displaying a significant ~5 log-fold reduction in burden compared to WT S. aureus (Fig. 4B).The ΔG4 S. aureus mutant showed significantly less burden compared to WT S. aureus during co-infection with P. aeruginosa (~5 fold).This difference was minimal compared to the difference in burden during mono-infection (~4 fold).The magnitude by which P. aeruginosa inhibited ΔG4 S. aureus mutant growth during co-infection was not significantly different from WT SA growth inhibition by P. aeruginosa (Fig. 4C).In contrast, the ΔpfkA S. aureus mutant was >1,000 fold more sensitive than WT S. aureus to growth inhibition by P. aeruginosa during co-infection.Importantly, the ΔpfkA S. aureus mutant burden during co-infection with P. aeruginosa was significantly below inoculum (~2 log-fold, P < 0.0001) and significantly less than the burden in mono-infection (~3 log-fold, P = 0.0005), suggesting substantial growth inhibition of the ΔpfkA S. aureus mutant by P. aeruginosa during co-infection.In normal mice, we again observed that the ΔpfkA S. aureus mutant had significantly less growth than WT S. aureus, both in mono-infection and during co-infection with P. aeruginosa (Fig. S2AB).However, we did not observe a significant sensitivity of the ΔpfkA S. aureus mutant to P. aeruginosa-mediated growth inhibition during co-infection in normal mice (Fig. S2C).From these in vivo experiments, glycolysis was determined to be a factor in both the success of S. aureus catheter infection and the resistance of S. aureus to growth inhibition by P. aeruginosa under diabetic conditions.

Dissemination of both S. aureus and P. aeruginosa from catheter infection into the tissue is heightened in a diabetic host
We observed that S. aureus mono-infection of the catheter in normal mice led to substantial dissemination of S. aureus into the tissue surrounding the catheter (Fig. 5A).S. aureus dissemination into the tissue was increased ~2 log-fold in diabetic mice.Co-infection in normal mice with P. aeruginosa significantly inhibited S. aureus dissemi nation into the tissues, though only moderately in either normal or diabetic mice.We concluded that, in addition to an effect on growth within the catheter, P. aeruginosa had an inhibitive effect on S. aureus dissemination into the tissue during catheter co-infec tion.However, while the diabetic environment enhanced S. aureus dissemination into the tissue, it did not affect the inhibitive ability of P. aeruginosa on S. aureus dissemination.
We also observed that the diabetic host environment supported modest, but significant dissemination of P. aeruginosa into the tissue during catheter infection (Fig. 5B).There also was a trend that co-infection with S. aureus moderately increased P. aeruginosa dissemination into the tissue during infection in normal mice, including in co-infections with ΔG4 and ΔpfkA mutant S. aureus (Fig. S3).Ultimately, it was concluded that the diabetic host environment increased both S. aureus and P. aeruginosa dissemina tion into the tissue during catheter infection in both mono-and co-infection.

S. aureus glycolysis is required for increased dissemination into tissues during diabetic infection
To determine if the glucose-rich environment of the diabetic host was also contributing to increased S. aureus dissemination into the tissue, we analyzed infections with the ΔG4 and ΔpkfA mutant S. aureus strains in diabetic mice (Fig. 6).The ΔpfkA mutant was significantly inhibited in its ability to disseminate into the tissues during mono-infec tion, with the ΔG4 mutant displaying a strong trend for decreased dissemination (P < 0.06) compared to WT S. aureus (Fig. 6A).In co-infection with P. aeruginosa, signifi cantly less ΔG4 and ΔpfkA mutant colonies were recovered from the tissue than WT S. aureus (Fig. 6B).Colonies were detected in tissue from only 2 of 10 mice infected with the ΔpfkA mutant.These trends of ΔG4 and ΔpfkA mutant S. aureus displaying decreased dissemination into the tissue compared to WT S. aureus were also observed to a lesser extent in normal mouse infections, including co-infection with P. aeruginosa (Fig. S4).Indeed, only the difference in dissemination between the ΔpfkA mutant and WT S. aureus during co-infection with P. aeruginosa reached the level of statistical significance.These experiments revealed that, while both glucose uptake and glycolysis likely support dissemination of S. aureus into the tissue, these processes are required for the increased dissemination afforded by the hyperglycemic environment of the diabetic host, especially in the presence of P. aeruginosa.

S. aureus inhibits P. aeruginosa growth in co-infection by a glucose-depend ent mechanism in a normal host, but a glucose-independent mechanism in a diabetic host
While S. aureus growth in catheter infections was inhibited during co-infection with P. aeruginosa, we also observed that P. aeruginosa growth was moderately inhibited during co-infection as well (Fig. 7A).The amount of P. aeruginosa recovered from the catheter was not significantly different between a normal or diabetic host, and this remained true in both mono-infection and in co-infection with S. aureus.Thus, co-infection with S. aureus inhibited P. aeruginosa growth in the catheter independently of the diabetic host environment.
During co-infection of P. aeruginosa with the ΔG4 and ΔpfkA mutant S. aureus strains, the P. aeruginosa burden was significantly higher than in co-infection with WT S. aureus.(Fig. 7B).This provided evidence that the inhibition of P. aeruginosa growth by S. aureus observed in the catheter was a process that required both maximal glucose uptake potential and functional glycolysis by S. aureus.Surprisingly, however, P. aeruginosa burden within the catheter was not significantly different in co-infection with S. aureus mutants in a diabetic infection despite the higher glucose availability (Fig. 7C).This result suggested that S. aureus inhibition of P. aeruginosa growth during co-infection in a diabetic host was independent of S. aureus glucose uptake and metabolism.

P. aeruginosa-secreted soluble factors inhibit S. aureus growth in vitro but not in vivo
The inhibition of S. aureus growth by spent P. aeruginosa supernatant (Fig. 2) was consistent with numerous studies showing that P. aeruginosa secreted factors inhibit S. aureus growth (28,30,34,35).These antistaphylococcal factors, which include pyocyanin, HQNO, and rhamnolipids, are positively regulated by the PQS QS system (20)(21)(22)(23).We used a P. aeruginosa ΔpqsR mutant, which does not produce a majority of P. aeruginosa antistaphylococcal factors, to test our hypothesis that P. aeruginosa secreted factors were primarily responsible for S. aureus growth inhibition.Over the span 48 h, S. aureus was able to grow well (~2 log-fold) in spent supernatant from the P. aeruginosa ΔpqsR mutant both with and without supplemented glucose (Fig. 8A).Though S. aureus displayed marginally higher growth after 24 h when supplemented with glucose, no significant difference in S. aureus culture density was observed by 48 h.These results indicated that S. aureus growth inhibition when grown in P. aeruginosa spent supernatant was due to PQS-regulated secreted factors.
We hypothesized that the P. aeruginosa ΔpqsR mutant would not inhibit S. aureus growth during co-infection in the catheter model based on our in vitro results.Contrary to our hypothesis, the P. aeruginosa ΔpqsR mutant inhibited S. aureus growth in the catheter lumen and dissemination to surrounding tissues in both normal and diabetic mice (Fig. 8BC).These results suggest that although P. aeruginosa secreted factors can potently inhibit S. aureus growth in vitro, these secreted factors were not responsible for S. aureus growth inhibition in the catheter.We observed a trend of growth inhibition by S. aureus on the P. aeruginosa ΔpqsR mutant within the catheter, but it did not reach the level of statistical significance (Fig. 8D).There was no statistically significant effect on P. aeruginosa ΔpqsR mutant dissemination into the tissue by either the diabetic environment or co-infection with S. aureus (Fig. 8E).

DISCUSSION
In this study, we employed a novel murine indwelling device co-infection model to examine the interactions between and virulence potential of S. aureus and P. aeruginosa in normal and diabetic infection microenvironments.We previously showed that S. aureus burden during SSTI in a diabetic host is significantly increased (10).Here, we show that S. aureus burden within the catheter was significantly increased (P < 0.0001) in a diabetic host (Fig. 1).Furthermore, the S. aureus burden was significantly increased in the tissue surrounding the catheter for diabetic animals (Fig. 5A).This is consistent with the increased dissemination and lesion size that is associated with S. aureus in diabetic SSTI (10).The greater dissemination of S. aureus into diabetic tissues observed here was associated with glucose uptake and glycolysis (Fig. 6).S. aureus dissemination in diabetic SSTIs is associated with both toxin and protease expression, which are under the control of the accessory gene regulator (Agr) system, with Agr activity being upregulated in response to increased glucose levels (10).It is likely that Agr-associated increases in S. aureus virulence factor production are similarly involved in the greater tissue dissemina tion observed here in the catheter model.
P. aeruginosa dissemination into the tissue during catheter infection was greater in diabetic animals (Fig. 5B).P. aeruginosa can use glucose as a sole carbon source (37,38) but will preferentially use other sources (39).Therefore, while S. aureus glucose metabolism was required for increased tissue dissemination in the diabetic host, P. aeruginosa may respond to the hyperglycemic environment independently of its glucose metabolism.Despite the metabolic preferences of P. aeruginosa, there is evidence that diabetic hyperglycemia can contribute to P. aeruginosa virulence.It has been shown that high levels of glucose can increase P. aeruginosa biofilm formation (40), and it has been recently shown that P. aeruginosa isolated from patients with diabetes displays enhanced virulence potential in the presence of glucose (41).Another possible explanation for the increase in tissue dissemination is that immune suppression in the diabetic host allows for greater P. aeruginosa and S. aureus proliferation in the absence of immune pressure.In particular, P. aeruginosa is sensitive to free radical-mediated killing by phagocytes (42), a process which is suppressed in a diabetic environment (10).Indeed, others have shown in a mouse model of S. aureus and P. aeruginosa acute lung co-infections that S. aureus can prevent phagosome acidification through the production of α-toxin, thereby potentiating P. aeruginosa growth and virulence (43).Whether the increased glucose availability in the diabetic host contributes to increased P. aeruginosa virulence or how diabetic-induced immune suppression influences the polymicrobial infection of S. aureus and P. aeruginosa is yet to be elucidated.
The glucose concentration used in the in vitro experiments described here was 25 mM (450 mg/dL), representing a moderately hyperglycemic diabetic host.Mice in our in vivo experiments had blood glucose concentrations ranging from 300 to 600 mg/dL (16.7-33.3mM).Additionally, glucose was not replenished in vitro, eventually being depleted by the cultured bacteria.However, glucose and other nutrients are constantly replen ished in an in vivo environment by the host circulatory system.Glucose availability may be scarce in the catheter due to high bacterial concentrations in an environment with limited space and may also act as a physical barrier from host nutrient replenishment.However, it was determined that ~4-fold more glucose was available in the diabetic catheter environment (Fig. S1).As both glucose uptake and glycolysis were implicated in increased S. aureus dissemination during diabetic catheter infection (Fig. 4), S. aureus dissemination is most likely dependent on constant glucose availability in the subcuta neous tissue, which has previously been determined to be significantly increased during S. aureus infection in diabetic mice (10).
S. aureus glycolysis was crucial to its survival in the host, even in mono-infection, as the ΔpfkA mutant was attenuated in mono-infection (Fig. 4A; Fig. S2A) and displayed growth defects in culture (Fig. 3B).It is plausible that glycolysis does not confer unique resistance to S. aureus against P. aeruginosa-mediated growth inhibition but instead is required for continued growth potential despite the presence of P. aeruginosa when glucose is readily available.This explanation is supported by the fact that an S. aureus mutant lacking dedicated glucose transporters, but still able to perform glycolysis, did not show a defect in resistance to P. aeruginosa-mediated inhibition (Fig. 4C).An important consideration is that S. aureus lacking its four dedicated glucose transporters can still import glucose but at a slower rate (15).Presumably, this import is through other promiscuous sugar transporters.Additionally, S. aureus has transporters for other glycolytic substrates that may be available in the host environment (15).Thus, an S. aureus mutant deficient in the glycolytic enzyme that represents a bottleneck in the glycolytic process (the ΔpfkA mutant used here) serves as an important tool to study the role of S. aureus glycolysis during co-infection with P. aeruginosa.Interestingly, the ΔpfkA mutant showed decreased growth in the presence of a glucose-rich environment, mostly notably when grown in supernatants supplemented with glucose, either with or without the presence of P. aeruginosa-secreted factors (Fig. 3B).This is most likely caused by the accumulation of intracellular fructose-6-phosphate (36).The ΔpfkA mutant was also significantly killed during co-infection with P. aeruginosa in vivo (Fig. 4).These results imply that the S. aureus response to a glucose-rich environment is detrimental if glycolysis is inhibited and strongly indicate that glycolysis is required for resistance to P. aeruginosa-mediated growth inhibition.
In both in vivo and in vitro experiments with WT S. aureus and WT P. aeruginosa, diabetes or addition of glucose to the environment only conferred partial resistance of S. aureus to growth inhibition by P. aeruginosa, with S. aureus growth never completely reaching the level of S. aureus in the absence of P. aeruginosa or P. aeruginosa-related factors (Fig. 1 and 2).These data suggest that there are additional P. aeruginosa-associ ated inhibitory factors other than the ones that can be ameliorated by increasing S. aureus glycolysis.
Our study provides insight into the glycolysis-dependent mechanism by which S. aureus resists P. aeruginosa-mediated inhibition in a diabetic environment, but the specific mechanisms by which glycolysis confers resistance to P. aeruginosa-mediated factors are not fully understood.The in vitro assay results from WT and mutant S. aureus grown in P. aeruginosa culture supernatant (Fig. 2 and 3) followed a close trend with the results from the in vivo experiments (Fig. 1 and 4; Fig. S2) in terms of P. aeruginosa inhibition of S. aureus growth during co-infection in the catheter model.PQS-regulated secreted factors are known to inhibit S. aureus growth in vitro (30,34,35,(44)(45)(46).This is why we initially suspected that P. aeruginosa secreted factors could be responsible for inhibition of S. aureus growth in our in vivo model.S. aureus requires glycolysis and subsequent fermentation to overcome oxidative stress (17).When glucose was abundant in our in vitro assays, presumably S. aureus was able to overcome respiratory inhibitors produced by P. aeruginosa.However, P. aeruginosa ΔpqsR was still readily able to inhibit S. aureus growth in the catheter model (Fig. 8A and B).This suggests that P. aeruginosa factors other than secreted antistaphylococcal agents are responsible for the growth inhibition observed in the catheter model.One possibility is that P. aeruginosa secretes other non-PQS-regulated antistaphylococcal toxins in vivo that are not secreted in vitro.Another possibility is that P. aeruginosa outcompetes S. aureus for limited nutrients in a normal environment, but this pressure is alleviated in a diabetic microenvironment that is replete with glucose.Yet another potential explanation is that P. aeruginosa alters the immune response during normal infection in a way that influences S. aureus survival within the host.We recently demonstrated that the innate immune response is repressed in a diabetic infection (10).We predict that it is likely a combination of these factors.Future work will aim to establish the exact mechanism for P. aeruginosa suppression of S. aureus in normal infection and how S. aureus resists this suppression in diabetic infection.
In summary, we describe an in vivo model that provides useful mechanistic exami nation of S. aureus and P. aeruginosa co-infections.Importantly, the described model shares multiple similarities to chronic infections observed in the clinic.Emergence of P. aeruginosa infection often occurs in existing chronic infections in patients with CF or diabetic wounds where S. aureus already resides and P. aeruginosa can establish as a major (or even the dominant) infecting microbe (31,47,48).Relevant to this fact, our co-infection model mimics the emergence of a P. aeruginosa infection during an established S. aureus infection (Fig. 1).These co-infections also showed that P. aerugi nosa was able to negatively influence S. aureus burden and dissemination (Fig. 1 and  5).Additionally, this model showed that in vitro interactions between S. aureus and P. aeruginosa shown by us and others do not always translate to an in vivo infection.Our in vitro experiments revealed that secreted factors from P. aeruginosa can inhibit S. aureus growth (Fig. 2), as many others have previously shown (34,35,(44)(45)(46).We were also able to recapitulate a glucose-dependent method by which S. aureus was able to inhibit P. aeruginosa growth in vivo (Fig. 7), which has been demonstrated by others in vitro (49).However, we provide evidence that these well-established in vitro mechanisms do not provide an explanation for in vivo inhibition within the catheter model.While others have been able to model S. aureus and P. aeruginosa co-infections in rodent models (29,43), the model described here allowed the study of S. aureus and P. aeruginosa co-infection both in a diabetic environment and using easily accessible, ordinary laboratory mice.These key distinctions led to the finding that the hyperglycemic environment of the diabetic host led to increased S. aureus survival and virulence despite P. aeruginosa-medi ated inhibition, which may explain why S. aureus and P. aeruginosa co-infections are more common in patients with CFRD (28).Future efforts will apply this model to further validate in vitro results as well as continue to elucidate the mechanisms of polymicrobial interactions in vivo.

Animals
Six-to eight-week-old male C57BL/6 mice were obtained from the Jackson Labora tory.Mice were kept at the University of North Carolina and used for experiments in accordance with an IACUC-approved protocol.

Bacterial strains and mutants
All S. aureus strains used here were on the LAC background, including two previously described S. aureus mutants: a quadruple mutant lacking the four glucose transporters glcA, glcU, glcB, and glcC (ΔG4) (10,15) and an upper glycolysis mutant lacking pfkA, which encodes phosphofructokinase (ΔpfkA) (10,17).The P. aeruginosa ΔpqsR mutant was generated in the MPAO1 genetic background used in this study by allelic exchange as previously described (50) using the previously published deletion vector, pΔpqsR-suc (51).

In vivo S. aureus and P. aeruginosa co-infection model
Subcutaneous catheter insertions were performed similar to what has been previously described (27).A 1-cm section of a 14G catheter was subcutaneously inserted into either flank of each mouse.Immediately following insertion, 1 × 10 5 CFU of S. aureus (LAC strain) were injected into the catheter in 20-µL PBS.S. aureus infection was allowed to establish for 4 days before co-infection with 1 × 10 5 CFU of P. aeruginosa (MPAO1 strain) in 20-µL PBS.P. aeruginosa mono-infections were also established at this time in catheters not previously inoculated with S. aureus.On day 11 following catheter insertion (7 days following co-infection with P. aeruginosa), catheters and immediately surround ing fibrotic tissue were removed, separated, and homogenized in 500-µL PBS.Homoge nates were plated on mannitol salt agar (Sigma-Aldrich) and Pseudomonas isolation agar (Thermo Fisher Scientific) to isolate S. aureus and P. aeruginosa colonies, respectively.Glucose measurements from filtered catheter homogenates were made using a Glucose (GO) Assay Kit (Sigma Aldrich).

Generation of diabetic mice
Immediately following P. aeruginosa infection on day 4, mice were made diabetic by intraperitoneal injection of streptozotocin (Sigma-Aldrich; 200-250 mg/kg) as previously described (10).Blood glucose was monitored starting on day 7. Injected mice with blood glucose <300 mg/dL by the end of the study were not considered diabetic and were removed from analysis.

In vitro S. aureus growth in P. aeruginosa culture supernatant
Overnight cultures of P. aeruginosa or wild-type S. aureus were grown in 5 mL Luria-Ber tani (LB) broth or tryptic soy broth (TSB), respectively, at 37°C, 225 rpm for 20 h.Next, 100 mL TSB without dextrose was inoculated with 1 mL of overnight broth culture in a pre-sterilized 500-mL Erlenmeyer flask.The Erlenmeyer flask opening was sealed with a Breathe-EASIER sealing membrane (Diversified Biotech) and incubated at 37°C, 270 rpm for 24 h.After incubation, the broth culture was decanted into 50-mL conical tubes and centrifuged at 3,200 × g for 30 min.This process of decanting and centrifugation was repeated with fresh 50-mL conical tubes, and then, supernatant was filter sterilized using a 0.45 µm filter.Filtered supernatant was evenly split into new containers.To one portion, glucose was added to 25 mM and Bacto casamino acids (Thermo Fisher Scientific) were added to 1% (wt/vol).The other portion was carbon-balanced with the addition of casamino acids alone to 1.385% (wt/vol).S. aureus supernatants were titrated to pH 7.15 using 10 M sodium hydroxide.Supernatants were then filter sterilized again and stored at −80°C for future use.
An overnight culture of S. aureus was prepared in TSB without dextrose.Ten microliters of the S. aureus overnight culture was added to 3 mL of P. aeruginosa or S. aureus supernatant supplemented with glucose and casamino acids or 3 mL P. aerugi nosa or S. aureus supernatant supplemented with casamino acids only.Cultures were incubated at 37°C, 225 rpm, and bacteria were enumerated by measuring colony forming units (CFU) per mL after 0, 24, and 48 h post-inoculation using Brain Heart Infusion (BHI) agar drip plates.

Statistical analysis
Power calculation on an initial pilot experiment of WT S. aureus mono-infection in normal mice established that a group size of n = 4 had statistical power of 0.8 to determine at least a twofold difference in CFU.Statistical comparisons between groups were made using ANOVA and performed using GraphPad Prism software.Corresponding P-values were calculated using Tukey's or Dunnett's post-test for multiple comparisons, where appropriate.Data expressed in CFU were log-transformed for statistical analyses.

FIG 1
FIG 1 Inhibitive effects of P. aeruginosa on S. aureus during co-infection are lessened in a diabetic host.(A) A schematic of the diabetic catheter infection model used throughout this study.(B) S. aureus colony-forming units (CFU) recovered from the catheter 11 days after initial infection, either alone or in co-infection with P. aeruginosa.Infections were carried out in mice treated with streptozotocin (diabetic) or untreated mice (normal).Bars represent geometric mean and standard error.Dotted line represents 10 5 CFU catheter inoculum.ANOVA with Tukey's test for multiple comparisons: ****P < 0.0001.

FIG 2
FIG 2 The S. aureus growth inhibition by P. aeruginosa secreted factors is ameliorated by glucose availability.S. aureus was cultured in vitro in the presence of P. aeruginosa or S. aureus culture supernatant supplemented with or without 25 mM glucose.All supernatants were fortified with 1% casamino acids before inoculation with S. aureus.Supernatants without glucose were supplemented with an additional carbon-equivalent of casamino acids.Bars represent geometric mean and standard error.Dotted line represents 10 7 CFU/ml culture inoculum.ANOVA with Tukey's test for multiple comparisons: **P < 0.01, ***P < 0.001, ****P < 0.0001.

FIG 3 FIG 4
FIG 3 Glycolysis is required for resistance of S. aureus to P. aeruginosa soluble factors.Mutant S. aureus lacking gene(s) for (A) four glucose transporters (ΔG4) or (B) 6-phosphofructokinase (ΔpfkA) was cultured in vitro in the presence of P. aeruginosa or wild-type S. aureus culture supernatant supplemented with or without 25 mM glucose.All supernatants were fortified with 1% casamino acids (CAA) before inoculation with S. aureus.Supernatants without glucose were supplemented with an additional carbon-equivalent of casamino acids.Bars represent geometric mean and standard error.Dotted line represents 10 7 CFU/mL culture inoculum.ANOVA with Tukey's test for multiple comparisons: ****P < 0.0001.

FIG 5
FIG 5 Diabetes supports both S. aureus and P. aeruginosa dissemination into tissues.(A) S. aureus or (B) P. aeruginosa colonies recovered from the surrounding tissue 11 days after catheter insertion and infection, either alone or in co-infection.Infections were carried out in normal or diabetic mice.Bars represent geometric mean and standard error.ANOVA with Tukey's test for multiple comparisons: *P < 0.05, **P < 0.001, ****P < 0.0001.

FIG 7 S
FIG 7 S. aureus inhibits P. aeruginosa growth in co-infection by a glucose-dependent mechanism in a normal host but a glucose-independent mechanism in a diabetic host.(A) P. aeruginosa colonies recovered from the catheter 11 days after insertion, either alone or in co-infection with wild-type S. aureus.Infections were carried out in normal and diabetic mice.(B and C) P. aeruginosa colonies were also recovered from the catheter after co-infection with ΔG4 or ΔpfkA mutant S. aureus.Bars represent geometric mean and standard error.Dotted line represents 10 5 CFU catheter inoculum.ANOVA with (A) Tukey's or (B and C) Dunnett's test for multiple comparisons (to WT): *P < 0.05, **P < 0.001, ***P < 0.001.

FIG 8 P
FIG 8 P. aeruginosa secreted factors under pqsR regulation inhibit S. aureus growth in vitro but not in the catheter model.(A) S. aureus was cultured in vitro in the presence of ΔpqsR P. aeruginosa culture supernatant supplemented with or without 25 mM glucose.All supernatants were fortified with 1% casamino acids before inoculation with S. aureus.Supernatants without glucose were supplemented with an additional carbon-equivalent of casamino acids.(B and C) S. aureus and (D and E) ΔpqsR P. aeruginosa CFU recovered from the catheter and surrounding tissue 11 days after initial infection, either alone or in co-infection.Infections were carried out in mice treated with streptozotocin (diabetic) or untreated mice (normal).Bars represent geometric mean and standard error.Dotted line represents inoculum.ANOVA with (A) Bonferroni's or (B-E) the Tukey's test for multiple comparisons: **P < 0.01, ***P < 0.001, ****P < 0.0001.