ABSTRACT
Obesity and associated type 2 diabetes (T2D) are important risk factors for infection following orthopedic implant surgery. Staphylococcus aureus, the most common pathogen in bone infections, adapts to multiple environments to survive and evade host immune responses. Whether adaptation of S. aureus to the unique environment of the obese/T2D host accounts for its increased virulence and persistence in this population is unknown. Thus, we assessed implant-associated osteomyelitis in normal versus high-fat-diet obese/T2D mice and found that S. aureus infection was more severe, including increases in bone abscesses relative to nondiabetic controls. S. aureus isolated from bone of obese/T2D mice displayed marked upregulation of four adhesion genes (clfA, clfB, bbp, and sdrC), all with binding affinity for fibrin(ogen). Immunostaining of infected bone revealed increased fibrin deposition surrounding bacterial abscesses in obese/T2D mice. In vitro coagulation assays demonstrated a hypercoagulable state in obese/T2D mice that was comparable to that of diabetic patients. S. aureus with an inactivating mutation in clumping factor A (clfA) showed a reduction in bone infection severity that eliminated the effect of obesity/T2D, while infections in control mice were unchanged. In infected mice that overexpress plasminogen activator inhibitor-1 (PAI-1), S. aureusclfA expression and fibrin-encapsulated abscess communities in bone were also increased, further linking fibrin deposition to S. aureus expression of clfA and infection severity. Together, these results demonstrate an adaptation by S. aureus to obesity/T2D with increased expression of clfA that is associated with the hypercoagulable state of the host and increased virulence of S. aureus.
INTRODUCTION
Over one million total knee and hip arthroplasties are performed in the United States per year, and that number is projected to climb to 4 million by 2030 (1–3). Infection rates following these procedures are estimated to be around 1 to 3% (4, 5), with Staphylococcus aureus being the most frequent infectious organism in orthopedic surgery (6). Its impact is compounded by the emergence of methicillin-resistant S. aureus (MRSA) (7). Among the greatest risk factors for orthopedic infection are obesity and type 2 diabetes (T2D), as several studies have found that obese, diabetic patients are up to 5 times more likely to become infected following placement of an orthopedic device (8–10). The epidemic of obesity and diabetes globally is likely to exacerbate this orthopedic complication, making a reduction in infection rates among the obese/T2D population of the utmost importance.
Several groups have reported that defective immune responses contribute to increased infection rates in obesity/T2D. Defects in the innate immune response to infection involving impaired macrophages (11–13) and/or neutrophils (14, 15) have been described in this patient population. We have demonstrated in a mouse model of implant-associated osteomyelitis that obese/T2D mice have increased S. aureus infection severity compared to lean-fed control mice (16). This is associated with defects in humoral immunity involving suppressed antibody production. Weakened response to viral vaccines and viral challenge in obese patients also implicates defects in antibody production (16, 17) and T lymphocyte function (18). While these defects in immunity likely play a role in increased infection rates in obesity/T2D, little attention has been directed toward the S. aureus adaptive changes to the unique host environment of obesity/T2D, which is associated with multiple pathological factors and comorbidities. These include elevated blood glucose, insulin resistance, altered insulin production (19), hyperlipidemia, inflammation (20), altered microbiome (21), and increased clotting factors (22–24).
S. aureus is proficient in adapting to the host environment to increase survival and virulence. For example, in the acute stages of bone infection, S. aureus has been shown to upregulate genes associated with gluconeogenesis, iron acquisition, and evasion of host immune responses and stress responses compared to S. aureus grown under in vitro growth conditions (25). During infection, binding of S. aureus to host tissues via microbial surface components recognizing adhesive matrix molecules (MSCRAMMs) is paramount to survival (26). Therefore, S. aureus MSCRAMMs that bind to fibrinogen, elastin, and fibronectin are enriched in both the acute and chronic phases of infection compared to S. aureus grown in vitro (25). Among the proteins crucial for S. aureus persistence within a host is clumping factor A (ClfA) (27, 28), which is a cell wall-anchored surface protein that likely plays a role in S. aureus-mediated agglutination in the presence of fibrin(ogen) to evade host phagocytes (29). S. aureus binding to fibrin(ogen) via ClfA represents an early step in adaptation to host factors during infection and ultimately leads to the formation of abscesses (28). Abscess formation has been demonstrated to be crucial in both host immune responses and S. aureus survival. Abscesses contain an inflammatory exudate composed of both viable and nonviable neutrophils and phagocytes that engulf and kill bacteria. Moreover, a fibrous margin is formed on the periphery of the abscess that prevents bacterial dissemination. Thus, abscess formation is a mechanism used by the host to isolate and ultimately eliminate S. aureus from the host (30). However, abscess formation is also viewed as critical to S. aureus survival and virulence within a host (28). Abscesses contain one or more distinct colonies of S. aureus organisms, referred to as staphylococcal abscess communities (SACs). The necrotic inflammatory cells surrounding the SACs form a barrier to further infiltration by recruited phagocytic cells. The fibrin network encapsulating the abscess may also be viewed in this manner. Thus, abscesses appear to be sites of intersection between the host's immune response and the bacterium's countermeasures. Nonetheless, several laboratories have shown that ablation or mutation of S. aureus clfA reduces host mortality and bacterial abscess formation in sepsis models (31, 32). However, the impact of increased substrate availability for ClfA binding, namely, fibrin(ogen), and the potential effect on abscess formation and S. aureus survival have not been studied. Obese/type 2 diabetics are well known to exhibit a hypercoagulable state with elevated fibrinogen and plasminogen activator inhibitor-1 (PAI-1) (24, 33). Here, we test the hypothesis that the increased severity of implant-associated S. aureus osteomyelitis is associated with adaptive gene expression changes to the coagulation state of the obese/T2D host, which increases bacterial persistence and virulence.
RESULTS
To investigate S. aureus bone infection in an established obese/T2D model, C57BL/6J mice were fed a high-fat (60% kcal from fat) or matched control (10% kcal from fat) diet for 3 months and subsequently received a MRSA strain (USA300 JE2 lac::lux)-contaminated trans-tibial implant as described previously (16). Mice consuming the high-fat diet had greater body weight than those on the control diet at 3 months (see Fig. S1A in the supplemental material), as well as a higher body fat percentage (Fig. S1B) and higher fasting blood glucose levels (Fig. S1C). High-fat-fed mice demonstrated impaired glucose tolerance testing (Fig. S1D and E). At day 21 postinfection, mice were sacrificed and the infected tibiae and adjacent soft tissues were harvested to quantify the bacterial load. S. aureus CFU were elevated in both the bone (Fig. 1A) and surrounding infected soft tissue (Fig. 1B) from the obese/T2D mice. On average, 10-fold higher CFU of S. aureus were recovered from bone of the obese/T2D mice and 1,000-fold higher CFU from the soft tissue. Histological analysis of the infected tibiae showed sizable osteolysis in both obese/T2D and control mice (Fig. 1C). Bacterial abscesses were also present within the marrow space (Fig. 1C, black arrows) but were more abundant in marrow of obese/T2D mice (Fig. 1D). Consistent with CFU quantitation (Fig. 1A), 3-fold more S. aureus abscess communities (SACs) were observed within the abscesses of infected tibiae of obese/T2D mice than in control mice, as visualized by Brown-Brenn Gram staining (Fig. 1C and E). Together, these results demonstrate that the mice fed a high-fat diet display a phenotype that mirrors obesity and type 2 diabetes, and this phenotype is associated with more severe osteomyelitis.
Obesity/T2D is associated with increased S. aureus CFU, abscesses, and abscess communities. At 5 weeks of age, male C57BL/6J mice were fed either a high-fat diet (60% kcal from fat; Ob/T2D) or a lean control diet (10% kcal from fat; Cont.). At 3 months on diet, a pin, precoated with USA300 JE2 lac::lux S. aureus, was implanted into the right tibia of each mouse. CFU of S. aureus were isolated from bone (A) and soft tissue (B) at 21 days postinfection. A second cohort was sacrificed at 14 days postinfection, and tibiae were harvested and processed for histological analysis. (C) Representative images of alcian blue hematoxylin-orange G-stained sections at 1× (left; black arrows indicate abscesses) and Brown-Brenn Gram-stained sections at 2× (right; yellow arrows indicate S. aureus abscess communities). (D) Quantitation of bacterial abscesses from panel C. (E) Quantitation of S. aureus abscess communities (SACs) from panel C. Each point in panels D and E represents the average per tibia using at least three histologic sections each (100 μm apart). *, P < 0.05; **, P < 0.01; ****, P < 0.0001. Unpaired Student's t test was used for panels A and B. Mann-Whitney U test was used for panels D and E.
Differential S. aureus gene expression in infected bone from obese/T2D versus control mice.To determine if adaptive changes by S. aureus contribute to the increased virulence of S. aureus in the obese/T2D host, RNA was isolated from infected tibia (UAMS-1 S. aureus) at day 14 after infection and processed for transcriptome sequencing (RNA-seq) analysis. Expression of 4 S. aureus genes was prominently elevated in samples from obese/T2D mice (Table 1). All of the gene products are surface receptors involved in adhesion. Additionally, clumping factor A (clfA), clumping factor B (clfB), serine-aspartate repeat protein (sdrC), and bone sialoprotein-binding protein (bbp) have fibrin(ogen) as a known ligand. Clinical evidence indicates that obesity/T2D is associated with a hypercoagulable state (24, 33, 34). Further, protein products from two of the upregulated genes, bbp and sdrC, also bind to bone sialoprotein. This suggests adaptation to the bone environment. Importantly, there was no significant difference in expression of any other S. aureus genes known to be associated with virulence (35) in these isolates from control and obese/T2D mouse bone (Table S1). These results suggest that selective adaptive changes in surface receptor expression (clfA, clfB, sdrC, and bbp) are associated with the increase in S. aureus CFU (Fig. 1A) and abscesses (Fig. 1C) within the bone of obese/T2D mice.
S. aureus (UAMS-1) isolated from tibia of infected obese/T2D mice highly expresses genes associated with adhesion compared to lean mice
Increased fibrin deposition and shortened clotting time in obese/T2D mice with S. aureus infection.The selective upregulated expression by S. aureus in obese/T2D mice of four genes encoding surface proteins with fibrin(ogen) binding affinity suggests an adaptation to the host environment. One important component of this environment is the hypercoagulable state of obesity/T2D, which is associated with increased risk for cardiovascular pathologies (34). Contributing to this state are increased levels of fibrinogen (24), abnormal levels of coagulation factors upstream of thrombin (22), and increased levels of plasminogen activator inhibitor-1 (33). With this potential for increased fibrin deposition in obesity/T2D, it is noteworthy that others have reported that in S. aureus infection, fibrin encapsulation of abscesses can be a pathological process that protects SACs from the host immune system (28). To determine the abundance of fibrin in the infected tibia of obese/T2D and control mice, immunofluorescent staining for fibrin was performed on histologic sections. Considerably more fibrin deposition was observed surrounding abscesses in the obese/T2D mice (Fig. 2A). The organization of a fibrin ring surrounding abscess communities (USA300 JE2 lac::lux S. aureus) is in contrast to the disorganization of fibrinogen (likely nonclotted) that is detected throughout the marrow space in uninfected control and obese/T2D mice (Fig. 2A). Quantification of the fibrin staining indicated an increase in total fibrin area per infected tibia in obese/T2D mice (Fig. 2B). When normalized to the number of abscesses per tibia, the quantity of fibrin per abscess was also greater in obese/T2D mice (Fig. 2C).
Increased fibrin formation and shortened clotting time in obese/T2D mice. Control and obese/T2D mice were given either a sterile pin or a pin precoated with USA300 JE2 lac::lux S. aureus. At 14 days, mice were sacrificed and tibiae were isolated, sectioned, and stained for fibrin and/or counterstained with DAPI. (A) Representative sections of the marrow space in sterile and infected lean and obese/T2D mice at 14 days after pin placement. White arrows indicate fibrin deposition around abscesses, and the unstained area immediately adjacent to the staining is cortical bone (black area). These sections were then quantified using Visiopharm software for total fibrin area per section (B) and fibrin area per abscess/section (C). Whole blood was collected from obese/T2D and control mice and citrated. (D) Clotting time was determined by thromboelastography following activation with calcium and kaolin. Representative thromboelastograph plots from control and Ob/T2D mice are shown. Clotting time is represented by the R time (arrow). (E) Quantitation of thromboelastograph data. (F) Citrated whole blood was incubated overnight at 37°C with USA300 JE2 S. aureus. Clots formed only in Ob/T2D mice. Clots were recovered and weighed. (G) Fibrin formation in plasma from control and Ob/T2D mice following clot initiation by the addition of USA300 JE2 S. aureus was determined by absorbance at 600 nm. (H) AUC of fibrin formation from the 3-h time course. (I) At 3 h, the unclotted plasma fraction from panel G was isolated, and S. aureus CFU were measured. *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001. Unpaired Student's t test was used for panels H and I. Clotting time (E) and fluorescent fibrin measurements (B and C) were determined by Mann-Whitney U test. A contingency table using Fisher's exact test was used for the whole-blood clotting. Fibrin formation (G) was analyzed with a two-way ANOVA with Bonferroni's posttest.
The next objective was to determine if the obese/T2D mouse model embodies the hypercoagulable state of obese/T2D patients and if this state contributes to the fibrin deposition and more severe osteomyelitis in obesity/T2D. Using thromboelastography, clotting times in whole blood from obese/T2D mice were reduced relative to controls (Fig. 2D and E). As additional evidence of hypercoagulability, 4 of 7 whole-blood samples from obese/T2D mice formed and maintained clots at 12 h when initiated by the addition of S. aureus (USA300 JE2), while no clotting occurred under identical conditions using blood from control mice (Fig. 2F). To examine the contribution of fibrin formation to these effects, plasma from obese/T2D and control mice were tested in a modified coagulase assay with S. aureus initiating clotting. Fibrin formation was assessed by changes in optical density at 600 nm. This assay revealed significantly higher changes in absorbance in obese/T2D plasma than in control plasma, indicating greater fibrin formation (Fig. 2G and H). The supernatants (unclotted fraction) from these assays were collected at 3 h and plated for CFU determination. The results revealed no difference in CFU isolated from the plasma of control or obese/T2D mice (Fig. 2I), indicating the changes (Fig. 2G and H) were not a result of differences in bacterial proliferation.
Similar to the results described above using mouse plasma, plasma from diabetic patients (hemoglobin A1c [HbA1c] of >10.0) yielded greater fibrin deposition in an S. aureus (USA300 JE2) coagulation assay than plasma from nondiabetic patients (HbA1c of <6.0) (Fig. 3A and B). The time course and magnitude of difference between diabetic and nondiabetic samples were comparable to the differences between samples from obese/T2D and control mice. Additionally, bacterial proliferation at 5 h under assay conditions was not different between the two patient populations (Fig. 3C). To determine if the higher concentrations of glucose in the plasma samples from diabetic patients contributed to the differences in fibrin formation, plasma from nondiabetic patients was spiked with glucose to increase glucose concentrations by 400 mg/dl (+400 mg/dl) or 800 mg/dl (+800 mg/dl). Fibrin formation was unchanged by the supplemental glucose concentrations (Fig. 3D). Similarly, S. aureus proliferation under the assay conditions was unaffected by supplemental glucose (Fig. 3E). However, addition of 2.0 g/liter of fibrinogen to nondiabetic plasma (average fibrinogen level in nondiabetic adults is 3.0 g/liter, and that in type 2 diabetic adults is 6.5 g/liter) resulted in increased fibrin formation in the plasma coagulation assays (Fig. 3F). Furthermore, samples treated with fibrinogen also had an increased clot weight (Fig. 3G). These results suggest that the higher levels of fibrinogen in plasma from diabetics contribute to the greater fibrin formation in these samples.
Increased S. aureus-initiated fibrin formation in plasma from diabetic patients. Plasma was collected from diabetic (HbA1c of >10.0) and nondiabetic (HbA1c of <6.0) patients. Clotting of plasma samples was initiated by addition of USA300 JE2 S. aureus and incubation at 37°C. (A) Fibrin formation was measured over time by measuring absorbance at 600 nm. (B) Area under the curve of fibrin formation assay from panel A. (C) S. aureus CFU isolated from the unclotted fraction shown in panel A. (D) Fibrin clot formation in nondiabetic patient plasma (control) supplemented with 400 mg/dl (+400 Glu) or 800 mg/dl (+800 Glu) of glucose for 5 h. (E) S. aureus CFU isolated from the unclotted fraction shown in panel D. (F) Fibrin clot formation in nondiabetic plasma samples supplemented with excess fibrinogen (2 g/liter). (G) Weight of fibrin clots shown in panel F at 5 h. *, P < 0.05; **, P < 0.01. Two-way ANOVA and Bonferroni posttest were used for panel A. One-way ANOVA was used for panels D and E. An unpaired t test was used for panels B and C, and a paired t test was used for panels F and G. Each point represents one patient.
Mutation of S. aureusclfA ameliorates infection severity in obese/T2D mice.The increased fibrin deposition surrounding S. aureus abscesses and the hypercoagulable state in our obese/T2D mouse model provide increased opportunity for S. aureus to utilize its surface fibrin(ogen) receptors for survival advantage, particularly with clfA expression being 34-fold higher in the bones of the obese/T2D mice. ClfA has been shown to be crucial in the early stages of abscess formation (28). To assess the role of increased clfA expression in S. aureus bone infections in the obese/T2D host, bone infection severity was compared between a USA300-derived staphylococcal strain with an inactivating mutation in the clfA gene and the parental JE2 strain. In response to a parental USA300-coated metal pin inserted through the tibia, approximately 10-fold more S. aureus CFU were isolated from bone of obese/T2D mice than controls (Fig. 4A). Markedly higher CFU were also recovered from the soft tissue. Remarkably, the more severe bone infection in obese/T2D mice was absent from infections with the mutant ClfA strain. Recovered CFU were comparable to those of control mice infected with the parental strain of USA300 (Fig. 4A). Additionally, the mutant strain had no significant effect on infections in the control mice. Similarly, the more severe soft-tissue infection in obese/T2D was blunted in mutant strain infections (Fig. 4B). In agreement with CFU data, the number of SACs identified in tibia of obese/T2D mice infected with the ClfA mutant strain of USA300 was reduced relative to that of the parental control strain (Fig. 4C and D). No difference was observed in the number of abscesses in bone of control mice regardless of S. aureus strain (data not shown). Importantly, no differences in inoculum on the stainless steel pins were observed as assessed by CFU (Fig. 4E). Taken together, the increased infection severity in obese/T2D mice is essentially eliminated when clfA is mutated in S. aureus. Interestingly, no significant difference in infection severity was observed in lean-fed control mice regardless of S. aureus strain, indicating that the effect was specific to obese/T2D mice.
Decreased S. aureus CFU and SACs in the obese/T2D host infected with ClfA mutant S. aureus. Control and obese/T2D mice were infected with either wild-type USA300 JE2 S. aureus or USA300 JE2 clfA::Bursa, a ClfA mutant strain (ClfA Mut.). S. aureus-coated implants initiated tibial infections as described for Fig. 1. CFU of S. aureus were isolated from bone (A) and soft tissue (B) at 21 days postinfection. NS, not significant. (C) For histological analysis, samples were harvested at day 14, fixed, and stained. Representative Brown-Brenn Gram stains of infected tibiae are shown. Yellow arrows indicate S. aureus colonies. (D) Quantitation of S. aureus abscess communities shown in panel C. Each point represents the average of three sections per sample from 5 mice. (E) Inoculating doses of each S. aureus strain were isolated from pins prior to placement. *, P < 0.05; **, P < 0.01; ****, P < 0.0001. Two-way ANOVA with Bonferroni's posttest was utilized for panels A and B. An individual Mann-Whitney U test with Bonferroni correction was performed for panel D. Unpaired t test was used for panel E.
Impaired fibrin resorption increases S. aureus abscess formation and clfA expression.To confirm the association between increased fibrin clotting and increased S. aureus virulence and infection severity, a heterozygous plasminogen activator inhibitor-1 (PAI-1) transgenic mouse was utilized. This mouse is modestly hypercoagulable and has no known immune defects (36), in contrast to strains with either ablated or overexpressed fibrinogen that have immunologic impairments that are likely to confound results (37). This mouse model is also relevant because PAI-1 levels are known to be increased in diabetic patients (33). Mice were infected with S. aureus-coated (USA300 JE2) trans-tibial pins as described above and harvested for histological analysis 14 days after infection. Consistent with an impaired ability to resorb fibrin, there was a modest but significant increase in fibrin-encapsulated abscess formation in mice overexpressing PAI-1 (Fig. 5A and B). Moreover, there were, on average, 4-fold more SACs per histological section in PAI-1 mice than in control mice, as indicated by Brown-Brenn staining (Fig. 5A and C). Importantly, the increase in abscesses and SACs correlated with an increase in clfA gene expression in S. aureus isolated from tibia of PAI-1 mice (Fig. 5D) compared to controls. Taken together, these results further implicate an adaptive role for S. aureus clfA expression in hypercoagulable states, as exemplified by obesity/T2D.
More abundant abscess communities and clfA expression by S. aureus in mice overexpressing PAI-1. PAI-1 transgenic and control mice were infected with USA300 JE2 S. aureus-coated implants as described for Fig. 1. Samples were harvested at day 14, fixed, and stained for histological analysis or homogenized, and RNA was isolated. (A, left) Representative alcian blue hematoxylin-stained sections (yellow arrows indicate abscesses). (Center) Fibrin-encapsulated abscesses visualized by immunofluorescent staining (white arrows indicate abscesses). (Right) Representative Brown-Brenn Gram-stained sections indicating increased SACs in a PAI-1 transgenic mouse. (B) Quantitation of abscesses shown in panel A. (C) Quantitation of SACs from panel A. (D) Increased clfA gene expression by S. aureus isolated from tibia of PAI-1 mice compared to the control. *, P < 0.05. Mann-Whitney U test was used for panels B and C, and unpaired Student's t test was used for panel D.
DISCUSSION
In this study, S. aureus was shown to cause more severe bone infections in obese/T2D mice. Increased S. aureus CFU were isolated from bone of obese/T2D mice compared to lean controls following infection with an orthopedic implant. This result is consistent with our previous report (16) and clinical studies indicating increased orthopedic infection rates in obese and diabetic patients (8, 9). Further, increased infection severity in the obese/T2D mice was characterized by increased numbers of abscesses within the infected tibiae. Histologically, these abscesses likely represent Brodie abscesses, which are a common radiographic indication of medullary bone infection. Moreover, we are the first to show an increase in abscesses and abscess communities within an obese/T2D host, indicating worse infection. In the current study, SACs as well as abscesses were more abundant in infected tibia of obese/T2D mice, with several SACs being identified within individual bone marrow abscesses.
S. aureus gene expression analysis in bone of infected mice revealed important differences in the obese/T2D host versus lean controls. Marked increases were observed in expression of four genes associated with adhesion or matrix binding. Four genes (clfA, clfB, sdrC, and bbp) had previously been reported to mediate S. aureus binding to fibrin(ogen) (27, 38–41). By upregulating these specific genes, S. aureus is apparently selectively enhancing its ability to associate with fibrin(ogen), a protein that is increased in abundance in the obese/T2D host as a component of the hypercoagulable state that is a complication of this disease. An increase in fibrin deposition and a more robust fibrin ring encapsulating S. aureus abscesses in our obese/T2D mouse model of implant osteomyelitis may be a manifestation of this coagulopathy. The increase in fibrin deposition is likely mediated, at least in part, by an increase in fibrinogen levels. It has been reported by multiple groups that diabetics have elevated levels of fibrinogen (23, 24). We also demonstrated hypercoagulability in whole blood and plasma from obese/T2D mice and diabetic patients. Upon incubation with S. aureus, more fibrin was formed in whole blood and plasma from diabetic mice and humans than controls. This result was independent of glucose levels in vitro, although both inflammation and elevated glucose levels have been proposed to be main drivers of the hypercoagulable state of obesity/T2D (22, 34). Regardless, increased fibrin deposition is likely to provide a survival advantage to S. aureus and further protect bacteria within abscesses from host immune responses. Taking these findings together, we conclude that the more severe S. aureus infections in obesity/T2D are due in part to the organism's adaptation to the increase in host clotting tendency.
Adaptation of S. aureus to obesity/T2D appears to be mediated, at least in part, by ClfA. This conclusion is based both on RNA-seq results and experiments using an S. aureus strain with an inactivating mutation within the clfA gene. Obese/T2D mice infected with S. aureus expressing a mutant ClfA had reduced infection severity relative to obese/T2D mice infected with wild-type (WT) S. aureus. In fact, infection severity was reduced to the level of control mice, totally negating the impact of obesity/T2D on infection severity. It has been reported that S. aureus clfA expression is critical for S. aureus abscess formation (27, 31). Mutation of clfA has been reported to lead to reduced numbers of abscesses and increased host survival in an S. aureus sepsis model (31). Consistent with these results, we observed a reduction in S. aureus CFU and SACs within bone of obese/T2D mice infected with the mutant ClfA. In contrast to the earlier reports, however, implant-associated bone infection in lean, control mice expressing the mutant ClfA was indistinguishable from that of the wild-type ClfA strain. This argues that ClfA is not necessary for S. aureus bone infections in normal hosts. Perhaps in the absence of obesity/T2D, redundancy in the clumping factors and proteins that bind to available fibrin(ogen) or the multiple mechanisms of immune evasion by S. aureus are sufficient for efficient bone infections (i.e., sequestrum formation, biofilm, etc.). Consistent with this hypothesis, lean-fed control mice formed comparable abscesses when infected with either WT or ClfA mutant S. aureus, indicating a role for redundant surface adhesions, such as ClfB, SdrC, and Bbp. However, it is remarkable that despite the robust increase in four S. aureus genes associated with binding to fibrin(ogen) in obese/T2D mice (Table 1), abrogation of only ClfA was sufficient to reduce virulence of S. aureus in obese/T2D hosts to that in lean mice. The absence of an effect in the lean host suggests a unique or disproportionate contribution for ClfA in obesity/T2D. Possible explanations are that ClfA, but not ClfB, SdrC, or BbP, has been shown to be important in immune evasion through its role in the degradation of C3 (42). Additionally, ClfA has been shown to bind to the λ-chain of fibrinogen, whereas ClfB binds only to the Aα-chain of fibrinogen. Importantly, clot formation involving fibrinogen λ-chains facilitates more rapid clotting than α- or β-chains (43, 44). Thus, differences in substrate specificity and the subsequent clotting rate may also explain why abrogation of ClfA is sufficient to reduce S. aureus virulence in obese/T2D mice. Regardless, the robust increase in clfA gene expression in S. aureus isolated from bone of obese/T2D mice appears to be linked to increased virulence. Increased fibrin levels in the obese/T2D host are a strong candidate for promoting adaptation of S. aureus by upregulating clfA gene expression (Fig. 2 and 3). This interpretation is supported by results with mice overexpressing PAI-1. Increased expression of PAI-1 would be anticipated to antagonize the plasminogen-to-plasmin conversion. Less plasmin would impair fibrinolysis and increase the availability of fibrin, the ClfA binding partner. Importantly, there was a 2- to 3-fold increase in abundance of S. aureus abscess communities and abscesses in PAI-1 transgenic mice compared to those in wild-type controls, further implicating clot formation and impaired clot degradation in increased S. aureus virulence. S. aureus isolated from PAI-1 mice also had increased clfA expression, further supporting S. aureus adaptation to a fibrin(ogen)-rich environment. The more abundant S. aureus abscess communities observed in the PAI-1 model closely parallel results in the obese/T2D model (Fig. 1). Moreover, our preliminary data (not shown) and data from other groups (33) indicate an increase in PAI-1 in obese and diabetic individuals. This argues that both increased clot formation (elevated fibrinogen) and decreased clot resorption (elevated PAI-1) contribute to increased infection severity in obesity/T2D via a ClfA-dependent adaptive mechanism.
Adaptation by S. aureus to host conditions is not without precedent. For example, transcriptome studies of S. aureus isolated from the lungs of mice have demonstrated a downregulation of genes associated with gluconeogenesis and an upregulation in genes associated with the fermentation and tricarboxylic acid cycle pathways compared to S. aureus grown in culture (45). Moreover, a separate transcriptome study indicated an upregulation of S. aureus genes associated with binding to surface receptors in both the acute and chronic phases of infection compared to S. aureus grown in culture (25). Furthermore, S. aureus adaptation can be host species specific. S. aureus binding to human hemoglobin is more robust than binding to mouse hemoglobin, indicating adaptation of S. aureus to its preferential host (46). Therefore, it is not unexpected that S. aureus would adapt to altered states of a host, such as in obesity/T2D. This is particularly true because the altered metabolic state of obesity/T2D is complex, with many primary and secondary changes, including elevated fibrinogen and PAI-1, hyperglycemia, chronic inflammation, impaired immune function, and hyperinsulinemia. Interestingly, two of the four S. aureus genes encoding matrix-binding proteins share not only affinity for fibrin(ogen) but also bone matrix (sdrC and bbp). While the current study focuses primarily on fibrin and ClfA, each of these identified genes is likely to represent adaptations by S. aureus to the microenvironment of osteomyelitis in the obese/T2D host. Each warrants further investigation.
In conclusion, we demonstrate an increase in S. aureus infection severity in obesity/T2D. This effect is associated with increased abscess formation, increased fibrin deposition, and marked adaptive upregulation by S. aureus of surface fibrin(ogen) binding proteins, particularly ClfA. ClfA appears to be critical in mediating the more severe infections. The adaptive responses to obesity/T2D represent potential therapeutic targets to reduce the more severe orthopedic S. aureus infections in the obese, type 2 diabetic population.
MATERIALS AND METHODS
Animals.All handling of mice and associated experimental procedures were reviewed and approved by the University Committee on Animal Resources at the University of Rochester Medical Center. Male C57BL/6J mice were purchased from Jackson Laboratories (Bar Harbor, ME). Mice were housed five per cage in one-way housing on a 12-h light/dark cycle. To model obesity/T2D, mice were placed on a high-fat diet (60% kcal D12492), and control mice were on a matched low-fat diet (10% kcal D12450J) at 5 weeks of age (Open Source Diets, Research Diets Inc., New Brunswick, NJ). Prior to infection, overnight fasting blood glucose measurements were made using One Touch glucose meters (Lifescan Inc., Milpitas, CA). Glucose tolerance testing was performed on mice fasted overnight. Intraperitoneal injection of a glucose bolus (300 mg/kg of body weight) was followed by tail vein glucose measurements at 15, 30, 60, and 90 min. Percent fat mass was measured prior to infection using a Lunar PIXImus2 (GE Lunar Corp). Male mice overexpressing PAI-1 (PAI-1 OE) were purchased from Jackson Laboratories [B6.Cg-Tg(CMV-Serpine1)1DGI/J]. Heterozygotic male PAI-1 OE mice were crossed with female C57BL6/J mice (Jackson Laboratories). F1-generated male PAI-1 OE heterozygotic pups were used, as well as littermates homozygous for WT PAI-1 as controls. The presence of the PAI-1 transgene in individual mice was determined using PCR analysis of tail DNA specimens obtained at 8 weeks of age as previously described (36).
Orthopedic infection model.Mice were infected as previously described (16). In short, stainless steel surgical wire (0.02 by 0.5 mm) was cut to 4-mm lengths and bent at 1 mm to form L-shaped pins. Pins were then autoclaved and placed in an overnight culture of S. aureus for 20 min prior to infection. S. aureus strains were used as described in the figure legends. Control S. aureus (JE2) and mutant ClfA S. aureus were obtained from the Nebraska Transposon Mutant Library (47, 48). The mutant ClfA strain has an inactivating mutation that was generated by the insertion of a transposon from bursa aurealis at the 5′ end of the clfA gene (+70 bp from the start site). USA300 lac::lux was a kind gift from Barbara Broker (University of Greifswald, Germany). Mice were anesthetized with 60 mg/kg ketamine and 4 mg/kg xylazine and received preoperative buprenorphine. The right leg was then shaved and washed with 70% ethanol and surgical scrub. A 5-mm incision was made on the medial surface of the proximal tibia. The medial tibia was predrilled using successive 30- and 26-gauge needles. Pins were then inserted through the drill hole, and the surgical site was closed using 5-0 interrupted sutures. Mice were sacrificed at either day 14 or day 21 postinfection.
Quantitation of S. aureus tissue content. S. aureus cells were isolated from tissue of sacrificed mice. Infected and necrotic soft tissues adjacent to the pin insertion site were recovered and placed in phosphate-buffered saline (PBS) on ice. The remaining noninfected tissue was disarticulated from the tibia. Infected tibiae were separated and placed in sterile PBS on ice. Bone and soft-tissue samples were then homogenized using an IKA T-10 handheld homogenizer (Wilmington, NC). Tenfold serial dilutions were then prepared in PBS, and 100-μl aliquots of each dilution were streaked onto tryptic soy agar plates. Plates were then placed in a 37°C incubator for 24 h, followed by colony counting.
RNA extraction and sequencing.RNA sequencing was performed on samples harvested from mice at day 14 of infection with UAMS-1 S. aureus. Infected tibiae were placed in RNAprotect bacterial reagent (Qiagen) immediately after isolation and stored at −80°C. Prior to extraction of RNA, bone was pulverized at liquid nitrogen temperatures. RNA was then extracted from S. aureus by mechanical lysis in ice-cold acid phenol using the Fast-Prep-24 instrument (MP Biomedicals) at a setting of 5.5 for 40 s. Extracted RNA was purified using RNeasy mini columns (Life Technologies). Contaminating genomic DNA was removed using Turbo DNase (Ambion).
(i) Illumina library construction for RNA-seq.Libraries were prepared from rRNA-depleted RNA (Ribo-Zero; Epicenter) using the ScriptSeq v2 RNA-seq library preparation kit (Epicentre) according to the manufacturer's instructions. Libraries were sequenced on an Illumina HiSeq2500 platform with 4 to 5 libraries multiplexed per lane and ∼50 million 100-nucleotide (nt) single-direction reads for each sample. Initial sequence data from the HiSeq2500 were evaluated for quality. The low-quality sequence reads were removed prior to final analysis using Seqclean (http://sourceforge.net/projects/seqclean/ ). The remaining high-quality processed reads were then mapped or aligned to reference S. aureus genome COL (49) or MRSA252 (50) with SHRiMP version 2.2.3 (51) to match sequence reads to specific S. aureus genes. Differential expression levels of all genes were determined using Cufflinks (cuffDiff2) version 2.0.2 (52) and a false discovery rate (FDR) of 0.05, which calculates gene expression as the number of sequence reads that map to each gene. A greater than log2-fold increase or decrease in expression level was used to identify genes that were significantly different between control and obese/T2D mice. Three mice were used per group for RNA sequencing experiments. All RNA-seq data (raw sequence reads and primary analysis) were deposited at NCBI and SRA.
(ii) RT-PCR.Reverse transcription-PCR (RT-PCR) was performed by isolating S. aureus RNA as described before. Samples were DNase treated (Invitrogen, Carlsbad CA), and 15 μl of RNA then was added to iScript cDNA synthesis per the manufacturer's instructions (Bio-Rad, Hercules CA). An aliquot of 1.5 μl of cDNA was then added to 7 μl water, 10 μl of SYBR green (Quanta Biosciences, Gaithersburg, MD), and 1.5 μl of forward and reverse primer mix. PCR was carried out on a Rotor-Gene Q (Qiagen, Hilden, Germany) RT-PCR machine. The following primers were used: clfA forward, GGC TTC AGT TGT AGG TA; clfA reverse, GCT TTC GTT ACT TGC GCT ATC; gyrA forward, AAG GTG TTC GCT TAA TTC GC; gyrA reverse, ATT GCA TTT CCT GGT GTT TC.
Histological analysis.For histological analysis, samples were fixed for 3 days in 10% neutral buffered formalin at room temperature, followed by three rinses each in PBS and distilled water. Samples were then placed in EDTA for 7 days for decalcification. Following decalcification, samples were placed in 70% alcohol prior to processing and embedded in paraffin, and transverse sections were cut to 5 μm and placed on glass slides. Slides were stained with Brown-Brenn modified Gram stain according to the manufacturer's directions (Newcomers Supply, Middleton, WI). Alcian blue hematoxylin-orange G (ABH-OG) stain is a standard bright-field stain used to visualize bone: mature calcified bone stains orange to red, bone marrow stains dark blue, and cartilage stains blue.
Immunofluorescent staining for fibrin.Sodium citrate (Dako, Carpinteria, CA) was used for antigen retrieval at 95°C for 40 min, followed by 2 h of blocking at room temperature with 10% donkey serum in PBS. Following blocking, a previously validated antibody to fibrin(ogen) (53) which binds to both fibrinogen and cleaved fibrin in S. aureus abscesses (purified rabbit anti-mouse fibrinogen; Innovative Research, Novi, MI) was used at a concentration of 1:200 in 1% donkey serum overnight at 4°C. Anti-rabbit secondary antibody conjugated to Alexa Fluor 594 (ThermoFisher, Waltham, MA) was used at 1:200 for 1 h in Tris-buffered saline, followed by 4′,6-diamidino-2-phenylindole (DAPI) staining (ThermoFisher) and fixation using ProLong gold antifade mounting reagent (Life Technologies, Eugene, OR). All slides were visualized using an Olympus VS120 virtual slide scanning microscope and Olympus OlyVIA software. Fibrin quantification was performed using Visiopharm software (Broomfield, CO). Abscesses were identified and traced. Using thresholding, the area of positive fibrin staining above the background fluorescence was identified and quantified. Bacterial abscesses (Brodie's abscess) and S. aureus abscess communities were quantified by blinded reviewers counting each distinct colony as an abscess community and fibrin-stained encapsulated rings as abscesses. Three slides per sample separated by 100 μm in tissue depth were quantified and averaged as the number of abscesses per histological section.
Coagulation assays.All work with human samples and the protocol for specimen procurement were reviewed and approved by the Research Subjects Review Board at the University of Rochester. Following routine analysis of patient HbA1c levels in the University of Rochester Clinical Laboratory, samples with HbA1c of <6.0% (nondiabetic) and HbA1c of >10% (consistent with diabetes) were retrieved. Because of the exempt status of this study, diabetic status (type 1 versus type 2) and current blood glucose levels were unavailable to the investigators.
(i) Thromboelastography.Mouse blood was drawn via cardiac puncture. A 450-μl aliquot was immediately mixed with 50 μl of a 3.2% citrate solution in distilled water. Standard thromboelastography was performed with kaolin initiation of coagulation on a TEG5000 (Haemonetics Corp., Braintree, MA).
(ii) Whole-blood clotting assays.Blood was collected as described above and citrated. Whole blood (250 μl) then was incubated with 50 μl of S. aureus overnight at 37°C. Clotting was determined by gentle mixing with a pipette. Any clots that formed overnight were weighed.
(iii) Fibrin formation assays.Aliquots of overnight S. aureus cultures (10 or 20 μl) were added to 80- or 90-μl aliquots of human or mouse plasma for a final volume of 100 μl in 96-well plates. Plates were incubated without agitation at 37°C for 3 h with mouse plasma or 5 h with human plasma. The optical density at 600 nm was measured using a plate reader at the indicated times. Each absorbance value was corrected for time zero absorbance (after the addition of S. aureus) and plotted versus time. Where fibrinogen was supplemented into the fibrin formation assay, 40 μl of a 10-mg/ml solution of fibrinogen (F3879; Sigma-Aldrich) was added to 160 μl of nondiabetic human plasma prior to the addition of 20 μl of S. aureus washed in PBS from an overnight culture. Fibrin deposition was then analyzed as described above. Glucose supplementation was carried out by adding d-glucose in a concentration of either 400 mg/dl or 800 mg/dl to nondiabetic control samples.
Statistics.Multiple analyte comparisons were measured using two-way analysis of variance (ANOVA) and Bonferroni's posttest. Unpaired t test was used when two groups were compared, including measurements of the area under the concentration-time curve (AUC). Mann-Whitney U test was used when comparing scoring histology (abscesses and S. aureus abscess communities) and on nonnormally distributed data. Whole-blood clotting assay statistics were completed using a contingency table (clotted or not) and Fisher's exact test. All statistics were analyzed using GraphPad Prism.
Accession number(s).The raw sequence reads and primary analysis were deposited into NCBI GEO under the accession numbers GSM2527810 , GSM2527811 , GSM2527812 , GSM2527813 , GSM2527814 , and GSM2527815 .
ACKNOWLEDGMENTS
We thank Sarah Mack, Kathy Maltby, and Hannah McRae for their very capable technical skills and the staff of the Clinical Laboratory of the University of Rochester Medical Center for facilitating the procurement of human samples. The support of the Histology, Biochemistry, and Molecular Imaging Core in the Center for Musculoskeletal Research at the University of Rochester is greatly appreciated. RNA-seq and transcriptome analyses described in this study were completed by the University of Rochester Genomics Research Center (URGRC).
This work was supported by AOTrauma Research (CPP Bone Infection), NIH P30 AR069655, and NIH T32 AR053459.
FOOTNOTES
- Received 6 December 2016.
- Returned for modification 4 January 2017.
- Accepted 9 March 2017.
- Accepted manuscript posted online 20 March 2017.
Supplemental material for this article may be found at https://doi.org/10.1128/IAI.01005-16 .
- Copyright © 2017 American Society for Microbiology.
