Staphylococcus aureus Aggregates on Orthopedic Materials under Varying Levels of Shear Stress

Periprosthetic joint infection occurring after artificial joint replacement is a major clinical issue that require repeated surgeries and antibiotic interventions. Unfortunately, 26% of patients die within 5 years of developing these infections. Staphylococcus aureus is the bacterium most commonly responsible for this problem and can form biofilms to provide protection from antibiotics as well as the immune system. Although biofilms are evident on the infected implants, it is unclear how these are attached to the surface in the first place. Recent in vitro investigations have shown that staphylococcal strains rapidly form aggregates in the presence of synovial fluid and provide protection to bacteria, thus allowing them time to attach to the implant surface, leading to biofilm formation. In this study, we investigated the attachment kinetics of Staphylococcus aureus aggregates on different orthopedic materials. The information presented in this article will be useful in surgical management and implant design.

ABSTRACT

Periprosthetic joint infection (PJI) occurring after artificial joint replacement is a major clinical issue requiring multiple surgeries and antibiotic interventions. Staphylococcus aureus is the bacterium most commonly responsible for PJI. Recent in vitro research has shown that staphylococcal strains rapidly form aggregates in the presence of synovial fluid (SF). We hypothesize that these aggregates provide early protection to bacteria entering the wound site, allowing them time to attach to the implant surface, leading to biofilm formation. Thus, understanding the attachment kinetics of these aggregates is critical in understanding their adhesion to various biomaterial surfaces. In this study, the number, size, and surface area coverage of aggregates as well as of single cells of S. aureus were quantified under various conditions on different orthopedic materials relevant to orthopedic surgery: stainless steel (316L), titanium (Ti), hydroxyapatite (HA), and polyethylene (PE). It was observed that, regardless of the material type, SF-induced aggregation resulted in reduced aggregate surface attachment and greater aggregate size than the single-cell populations under various shear stresses. Additionally, the surface area coverage of bacterial aggregates on PE was relatively high compared to that on other materials, which could potentially be due to the rougher surface of PE. Furthermore, increasing shear stress to 78 mPa decreased aggregate attachment to Ti and HA while increasing the aggregates’ average size. Therefore, this study demonstrates that SF induced inhibition of aggregate attachment to all materials, suggesting that biofilm formation is initiated by lodging of aggregates on the surface features of implants and host tissues.

IMPORTANCE Periprosthetic joint infection occurring after artificial joint replacement is a major clinical issue that require repeated surgeries and antibiotic interventions. Unfortunately, 26% of patients die within 5 years of developing these infections. Staphylococcus aureus is the bacterium most commonly responsible for this problem and can form biofilms to provide protection from antibiotics as well as the immune system. Although biofilms are evident on the infected implants, it is unclear how these are attached to the surface in the first place. Recent in vitro investigations have shown that staphylococcal strains rapidly form aggregates in the presence of synovial fluid and provide protection to bacteria, thus allowing them time to attach to the implant surface, leading to biofilm formation. In this study, we investigated the attachment kinetics of Staphylococcus aureus aggregates on different orthopedic materials. The information presented in this article will be useful in surgical management and implant design.

INTRODUCTION

Adherence of bacteria to implanted medical devices and adjacent tissue can lead to biomaterial-associated infections (BAIs), often resulting in life-threatening diseases and implant failures (1). Typical BAIs in the health care setting include those on dental implants, prosthetic joints, catheters, cardiac pacemakers, and heart valves (2–6). Among these infections, infection of the prosthetic joint is one of the challenging complications after total joint arthroplasty (7). Despite methods implemented to prevent this complication, including antibiotic prophylaxis, patient risk stratification, detection and treatment of Staphylococcus aureus colonization, and maintaining a clean operating room environment, there is a 1% infection rate for total joint arthroplasty (7). Since the number of primary and revision total joint arthroplasties is rapidly increasing due to the growing size of the aging population, the number of periprosthetic joint infection (PJI) cases is expected to increase (8). In the United States, there were 332,000 total hip and 719,000 total knee arthroplasties in 2010, and those numbers are expected to reach 572,000 and 3.48 million by 2030 for hips and knees, respectively (9, 10). Therefore, PJIs are likely to become extremely problematic, because their treatment frequently requires implant removal and multiple antibiotic courses, and this process is often associated with increased patient morbidity and mortality (11). PJIs also lead to higher health care costs because of repeated surgeries, extended hospitalization, rehabilitation, and antibiotic therapy. For example, a single prosthetic joint BAI has an average estimated health care cost of at least $50,000 and up to $130,000 (12–14). Furthermore, these costs underestimate the true impact on the patient, as they do not include the long-term physical and social impairments that the patients potentially endure (15, 16).

Staphylococcus aureus is a Gram-positive facultatively anaerobic bacterium and is one of the most common causes of PJI (7, 17). It is also the pathogen most frequently associated with metal surfaces and with acute and chronic osteomyelitis (18). During PJIs, S. aureus biofilms and aggregates have been observed on the surfaces of implant prostheses and the surrounding tissue (19), though it is not clearly understood how S. aureus biofilms become established in PJIs. Biofilms are groups of bacteria encased in a matrix containing polysaccharides, proteins, and extracellular DNA (20). In vitro research shows that staphylococcal strains exhibit rapid aggregation in the presence of human and bovine synovial fluid (SF) and develop into macroscopic colonies within 24 h, which eventually develop into an antibiotic-resistant biofilm (21–23). Previous investigations have predominately analyzed the initial attachment of single cells to different biomaterial surfaces. The materials that are commonly implanted during orthopedic surgery include metals (stainless steel, chrome-cobalt, and titanium alloys), polymers (polymethylmethacrylate [PMMA]), high-density polyethylene, and bioglass (18). Joint implants are made from some of these materials and are typically cemented into place with hydroxyapatite. Barth et al. showed that S. aureus and Staphylococcus epidermidis colonize orthopedic implant materials such as titanium alloy discs, PMMA, and ultra-high-molecular-weight polyethylene (24). They found high numbers of S. aureus on metal and S. epidermidis on polymers. Furthermore, Oga et al. found greater S. epidermidis colonization on methylmethacrylate than stainless steel and aluminum (25). Overall, these studies indicated that certain materials may promote infection by favoring colonization by specific pathogens. Another study quantified differences in bacterial adherence due to polymer chemical properties, surface roughness, and flow conditions (26). That study demonstrated a decrease in adherent bacteria on polyvinyl chloride with an increase in the shear rate, which indicates that fluid shear rate plays a role in the attachment and detachment of bacteria to materials (26). The bacterial adhesion force calculated by atomic force microscopy in one study indicated that hydroxyapatite shows weak adhesion force, whereas stainless steel shows strong adhesion force, due to their surface energy and surface roughness (27).

Both host and bacterial factors can mediate bacterial aggregation. Several components of synovial fluid (a viscous substance in the joint space), such as fibrinogen, fibronectin, and free DNA, can induce bacterial aggregation, which has been hypothesized to play a role in PJI (28). The host protein fibronectin has also been shown to promote adhesion of S. aureus and S. epidermidis to stainless steel, pure titanium, and titanium-aluminum-niobium alloy (29). Indeed, the presence of fibronectin on bone-implanted metallic devices, such as those made of titanium, promoted attachment of S. aureus (30). Bacterial adhesion studies normally assess the interaction of single planktonic cells interacting with surfaces. Therefore, in this study we analyzed the attachment of S. aureus bovine synovial fluid-induced aggregates, compared to that of nonaggregated cells, to various orthopedic materials, including stainless steel (316L), hydroxyapatite (HA), polyethylene (PE), and titanium (Ti), under shear stresses of 15 and 78 mPa. Bacterial adherence was quantified by determining the number of attached particles, their surface area, and their average size. While the exact shear stress in the joint space is not known, we expect that a range of stresses are present depending on joint activity and the location within the joint. To replicate the conditions within the joint, the attachment in this study was examined under two different flow conditions with shear stresses of 15 and 78 mPa. Of the four materials, we limited testing of the effect of high shear stress (78 mPa) on bacterial attachment to Ti and HA because they are most used in clinical applications. Titanium is a material commonly used in orthopedics and dental implants (31, 32) on which biofilms can grow (33, 34), and hydroxyapatite has been used as a coating in the region where strong interface with the bone is required, such as femoral components for the knee joints (35). Hydroxyapatite coatings also play a role in accelerating the bone formation process in joint prosthesis (36). Therefore, studying bacterial attachment to HA is important, as this material covers a large surface area in prosthetic implants. We investigated the effects of synovial fluid on S. aureus surface attachment via fluorescence microscopy and found that it inhibits surface attachment to every material used in this study (Ti, 316L, HA, and PE) and at both shear stresses. We further concluded that materials with rougher surfaces, such as HA and PE, facilitate attachment of larger aggregates and greater surface area coverage than the metals, Ti and 316L.

RESULTS

Optical profilometry and contact angle measurement.

The topography of various surfaces was examined by measuring the surface roughness (Ra) and contact angle. The Ra value and contact angle of each material are listed in Table 1. The results indicate that there are 5- and 8-fold increases in the Ra values of HA and PE, respectively, compared with those of Ti and 316L. This demonstrates that HA and PE are significantly rougher than the Ti and 316L.

TABLE 1.

Average surface roughness (Ra) and water contact angle measurements on different materials^a

Material	Ra (μm)	Water contact angle (°)
Titanium (Ti)	0.195 ± 0.03	69 ± 3.67
Stainless steel (316L)	0.203 ± 0.02	64 ± 1
Hydroxyapatite (HA)	1.165 ± 0.15	33 ± 1.45
Polyethylene (PE)	1.634 ± 0.12	145 ± 4.84

^aValues are means ± standard deviations of three measurements.

The contact angles of Ti and 316L were similar and indicate that these are hydrophilic surfaces. With a contact angle of 33°, HA behaves as a hydrophilic material. PE, on the other hand, has the highest contact angle of the materials and thus is the most hydrophobic surface among these materials.

SF-induced aggregation inhibits bacterial attachment at 15 mPa shear stress on different orthopedic materials.

To determine how synovial fluid affects bacterial adherence within the joint, we measured cell surface attachment on orthopedic materials under two different shear stresses, 15 and 78 mPa. A schematic of the flow cell system used here is shown in Fig. 1. The attachment kinetics of single cells and aggregates on the different materials used are shown in Fig. 2. We quantified three parameters: the number of attached particles, surface area covered by particles, and the average size of the particles (Fig. 3). For all the materials tested, at 15 min, fewer aggregates (in the presence of SF [+SF]), i.e., 194, 121, 94, and 48, were attached to HA, Ti, 316L, and PE, whereas the single cells (in the absence of SF [−SF]) attached on the surface were 4494, 4441, 4629, 724 on HA, Ti, 316L, and PE, respectively, at 15 mPa shear stress. An analysis of variance (ANOVA) yielded significant variation (P < 0.05) between the +SF and −SF groups for each material at various attachment times (Fig. 3).

FIG 1.

Flow system for aggregate attachment assay.

FIG 2.

Bacterial attachment inhibition by SF on various orthopedic materials. S. aureus surface attachment to HA, Ti, 316L, and PE was quantified after 5, 10, and 15 min under a constant shear stress of 15 mPa. The black particles are the attached single cells and aggregates. Bar, 50 μm. +SF and −SF represent SF-induced aggregation and single cells in the absence of SF.

FIG 3.

Synovial fluid aggregation inhibits bacterial surface attachment. Attachment was quantified based on aggregate and single-cell numbers (53), surface area coverage, and sizes at 15 mPa shear stress on hydroxyapatite (HA), titanium (Ti), stainless steel (316L), and polyethylene (PE). +SF and −SF represent SF-induced aggregation and single cells in the absence of SF. Data are means and standard errors of the means. Statistical significance was determined by one-way ANOVA. *, P < 0.05; **, P < 0.01; ***, P < 0.001.

We also compared the total number, average size, and surface area of the attached particles on different materials under both +SF and −SF conditions at different attachment times. ANOVA of these numbers yielded significant variation within the materials for the number of particles attached in both +SF and −SF groups except for the results from +SF and a 15-min attachment time. A post hoc Tukey test within the +SF group at a 5-min attachment time results showed that the number of particles attached was significantly different for HA than Ti, PE, and 316L for the 5-min time. For the 10-min attachment time and +SF group, the number of particles was significantly different for HA than 316L and PE only, and for the 15-min and +SF group, it was significantly different for HA than PE. In the experiments without synovial fluid, significant differences were found only between HA and 316L as well as PE for the 5-min attachment time and only between HA and PE for the 10- and 15-min attachment times. For all other multiple comparisons, there were no significant differences (Tables 2 and 3).

TABLE 2.

One-way ANOVA multiple-comparison results^a

Variable	P values from ANOVA between groups at time (min)
	+SF			−SF
	5	10	15	5	10	15
No. of particles	0	0.008	0.064	0	0	0
Surface area coverage	0.274	0.128	0.077	0.003	0.001	0.003
Average size	0.282	0.219	0.17	0.003	0	0.002

^aThe total number, average size, and surface area coverage of the attached particles on various materials, in the presence and absence of synovial fluid, with different attachment durations and shear stresses were compared.

TABLE 3.

Tukey post hoc multiple-comparison results^a

Shear stress (mPa)	Material 1	Material 2	P value at time (min)
			Number of particles						Average size						Surface area coverage
			+SF			−SF			+SF			−SF			+SF			−SF
			5	10	15	5	10	15	5	10	15	5	10	15	5	10	15	5	10	15
15	HA	Ti	0.00	0.19	0.42	0.06	0.67	1.00	1.00	1.00	1.00	0.16	0.55	0.83	0.91	0.91	0.91	0.94	0.98	0.93
		316L	0.00	0.05	0.20	0.01	0.38	0.97	1.00	1.00	0.99	0.94	0.30	0.20	0.82	0.78	0.82	0.24	0.13	0.32
		PE	0.00	0.01	0.05	0.00	0.00	0.00	0.38	0.21	0.16	0.04	0.00	0.01	0.51	0.64	0.51	0.01	0.01	0.00
	Ti	316L	0.99	0.77	0.93	0.57	0.95	0.92	1.00	1.00	1.00	0.07	0.05	0.06	1.00	0.99	1.00	0.11	0.07	0.14
		PE	0.06	0.11	0.43	0.00	0.00	0.00	0.37	0.18	0.11	0.00	0.00	0.00	0.23	0.31	0.23	0.00	0.00	0.00
	316L	PE	0.09	0.41	0.75	0.00	0.00	0.00	0.33	0.17	0.10	0.09	0.05	0.19	0.18	0.22	0.18	0.11	0.02	0.04
78	HA	Ti	0.01	0.01	0.03	0.05	0.03	0.03	0.34	0.13	0.31	0.08	0.03	0.02	0.00	0.07	0.29	0.10	0.03	0.11

^aThe total number, average size, and surface area coverage of the attached particles on various materials, in the presence and absence of synovial fluid, with different attachment durations and shear stresses were compared.

Furthermore, we analyzed the rates of change in the number of particles, surface area coverage, and average size from 5 min to 15 min (Table 4). There was an initial rapid increase between 0 and 5 min, and then the rates generally appeared linear, as assessed by linear regression for 5 to 15 min (see Fig. S1 in the supplemental material). The rates of change in number of particles were about 11 to 80 times higher in the −SF group for HA, Ti, and PE and approximately 4,000 times higher for 316L than in the +SF group. Similarly, the rate of change of surface area coverage was found to be higher in −SF group than in +SF group. However, the rate of change of average size was higher in +SF group than in the −SF group (Table 4). Although the rate of change in attached number of particles was lower in SF, the average surface area was found to be higher. This could be due to the attachment of only larger aggregates in SF. On the other hand, without SF, numerous smaller particles are attached. Since a particle is counted as one irrespective of its size, we observed that the rate of change was higher without SF because of numerous smaller particles. Moreover, since the average size is a function of surface area over total number of particles, the rate of change of average size was higher with SF than without SF.

TABLE 4.

Rates of change in number of particles, surface area coverage, and average size on different materials and comparison between the rates mediated by the presence and absence of SF^a

Material	Rate of change in no. of particles		P value	Rate of change in surface area coverage		P value	Rate of change in avg size		P value
	+SF	−SF		+SF	−SF		+SF	−SF
HA	−7.2 ± 1.55	72.66 ± 22.5	0.02	0.04 ± 0.02	0.44 ± 0.18	0.09	1.45 ± 0.46	0.15 ± 0.07	0.05
Ti	2.23 ± 2.97	160.43 ± 25.01	0	0.02 ± 0.03	0.48 ± 0.02	0	−0.16 ± 0.23	0.07 ± 0.01	0.36
316L	0.06 ± 0.48	220.16 ± 25.85	0	0.01 ± 0.009	0.34 ± 0.09	0.02	0.21 ± 0.15	0.02 ± 0.01	0.27
PE	1.93 ± 1.31	41.13 ± 11.34	0.02	0.42 ± 0.21	0.03 ± 0.09	0.14	3.67 ± 6.94	0.02 ± 0.01	0.62

^aP values were determined by one-way ANOVA.

To determine the effect of synovial fluid on these rates of changes, an analysis of variance (ANOVA) of these rates yielded significant variation within the +SF and −SF groups (P < 0.05) for the rate of change in the number of particles and the surface area coverage. In the case of the rate of change of average size, significant differences were not found. This could be due to the dependency of this parameter on the total number of particles. In this study, the number of particles is the total sum of the number of distinctly separate particles irrespective of their sizes. This may yield variations in the calculated average size, and hence, significant difference could not be observed.

In addition, the total surface area covered by the aggregates was smaller on HA (3%), Ti (1%), and 316L (0.56%) than the surface coverage by single cells (10% [HA], 11% [Ti], and 7% [316L]). However, the opposite was observed on PE, where the surface area covered by aggregates (7%) was greater than that covered by single cells (0.48%), In addition, the average particle size of attached aggregates was greater than that of single cells for all the materials; in particular, PE had the largest aggregates. Thus, SF inhibited the attachment of bacterial aggregates on almost all materials compared to that of single cells. These data suggest that this phenotype is not specific to just one surface type and that synovial fluid seems to inhibit attachment to the implant during PJI.

Attachment of single cells and aggregates on HA and Ti under 78 mPa shear stress.

The range of shear stress present in the joint likely varies based on numerous factors. Therefore, we tested S. aureus attachment kinetics at a higher shear stress of 78 mPa on Ti and HA (Fig. 4 and 5). Particle number, surface coverage, and particle size were quantified for 5, 10, and 15 min of attachment duration at 78 mPa and 15 mPa. Through statistical analysis, significant differences were found in Ti and HA between 15 mPa and 78 mPa shear stress in terms of particles attached in aggregates (Fig. 5). Similar to the low shear stress condition of 15 mPa, the number of attached aggregates was reduced following synovial fluid exposure (Fig. 4 and 5). Interestingly, larger aggregates were observed on HA and Ti with higher shear stress than with the lower shear stress, similar to the results published by another group (37), where aggregates or clusters appeared with increased size. Their shear stress values range from 0.09 to 7.3 Pa. However, our results did not demonstrate an increase in bacterial attachment with higher shear stress, as shown in that study (37). Rather, the number of attached aggregates decreased as shear stress was increased. This could be because the accumulation rate as described by the mass balance equation (38, 39) (accumulation = attachment rate + growth rate − detachment rate − decay rate) is very low assuming zero growth and decay, since the durations of the experiments are relatively short. However, higher shear stresses tend to cause higher rates of detachment from smooth surfaces than from rougher surfaces, as the aggregates could be protected in the uneven features of the rougher surfaces.

FIG 4.

Bacterial attachment inhibition by SF on Ti and HA. S. aureus surface attachment to Ti and HA was quantified after 5, 10, and 15 min under a constant shear stress of 78 mPa. The black particles are the attached single cells and aggregates. Bar, 50 μm. +SF and −SF represent SF-induced aggregation and single cells in the absence of SF.

FIG 5.

Synovial fluid aggregation inhibits bacterial surface attachment. Attachment was quantified based on aggregate and single-cell numbers, surface area coverage, and sizes at 15 mPa and 78 mPa shear stress on titanium (Ti) and hydroxyapatite (HA). +SF and −SF represent SF-induced aggregation and single cells in the absence of SF. Data are means and standard errors of the means. Statistical significance was determined by one-way ANOVA. *, P < 0.05; **, P < 0.01; ***, P < 0.001.

SEM images of single cells and aggregates on different materials at 15 mPa and 78 mPa shear stress.

Figure 6 shows the scanning electron microscopy (SEM) images of aggregates and single cells on different materials under 15 mPa and 78 mPa shear stress. Many adherent S. aureus aggregates of various sizes could be observed on Ti and 316L coupons following flow under 15 mPa of shear stress. The single cells without SF exposure seem to be uniformly distributed over the surfaces of both Ti and 316L coupons. Similarly, in HA at 15 mPa, larger bacterial aggregates can be observed following treatment with SF. Along with larger aggregates, many smaller aggregates can also be found, especially on the HA coupon. The single cells on HA are distributed over the surface, and 3- to 5-cell clusters frequently formed at sites containing abnormal surface features. PE, which is the roughest of these materials, retained larger aggregates than the other three materials tested. There were fewer single cells on PE surfaces than on Ti, 316L, and HA coupons. Along with single cells, PE also contained smaller clusters of cells on the surfaces without SF. Although the number of aggregates at 78 mPa on Ti surfaces was lower than that at 15 mPa, the aggregates appeared larger than with 15 mPa shear stress. In contrast, there were fewer single cells on the Ti surfaces than at 78 mPa. On HA as well, the number of aggregates and single cells seems to be lower than at 15 mPa; however, the aggregates appeared to be larger and had an appearance similar to that of biofilm aggregates.

FIG 6.

(A) SEM images of aggregates and single cells on Ti, HA, 316L, and PE surfaces at 15 mPa and 78 mPa shear stress. +SF and −SF represent SF-induced aggregation and single cells in the absence of SF. Magnification, ×3,000. (B) Images of blank coupons, included to show different surface structures. Magnification, ×1,500.

DISCUSSION

Our group reported previously on bacterial aggregates during PJI (19). S. aureus forms aggregates in human and bovine synovial fluid which grow to a macroscopic size in 24 h (21–23). The formation of aggregates provides the bacteria with enhanced tolerance to antibiotics and promotes surface colonization and biofilm formation (40, 41). Therefore, we hypothesized that the aggregates that form in synovial fluid provide early protection to bacteria entering the surgical site, allowing them time to attach to the implant surface, leading to biofilm formation. Thus, understanding the kinetics of aggregate attachment to different materials under defined shear stresses is important for better understanding bacterial colonization within the artificial joint during infection.

Multiple investigations of bacterial adhesion on orthopedic materials have been conducted using planktonic bacteria. However, few studies have been published regarding SF-induced bacterial aggregates, and such studies may provide critical insight into the development of chronic infections. Thus, in this study, we tried to simulate in vivo human joint conditions by exploring the attachment of SF-induced aggregates on different surfaces. We found that the surface interaction of aggregates compared to that of single cells varied based on the surface chemistry of a given material. However, regardless of the material type, SF-induced aggregation resulted in less particle attachment and greater aggregate size than the single-cell populations in Ringer’s saline under both low and high shear stress (15 and 78 mPa, respectively). However, on PE, the surface area coverage by the aggregates was greater, which could possibly be due to the rougher surface and the aggregates may initially bind more tightly to the surface. In contrast, the number of single cells attaching to PE was less than the other materials. The reason for this could be that the increased surface roughness decreases the adhesion of Gram-positive bacteria, as explained in one study (42). It is possible that relatively smooth surfaces are infrequent sites of aggregate attachment and that aggregate-surface interaction is more prevalent after bacteria become lodged within large-scale surface features, such as edges and tapped holes, that are present on rougher surfaces. Furthermore, rougher surfaces promote bacterial adhesion due to increased surface area and depressions that provide more favorable and additional sites for colonization (43). Besides roughness, other factors responsible for bacterial adhesion are surface chemical composition, surface configuration, flow conditions, surface hydrophobicity or wettability, serum or tissue proteins such as fibronectin, fibrinogen, albumin, and laminin, and bacterial hydrophobicity (43, 44).

Furthermore, there was a high degree of correlation (R² = 0.92) between the roughness and the surface area covered by the aggregates but not with the hydrophobicity of the material (data not shown). It was determined that the total surface area covered by the aggregates increased as the material roughness increased. Rougher surfaces tend to provide more favorable conditions for bacterial attachment via the material irregularities (26). However, there was no correlation between the roughness and the number or the size of aggregates attached on different materials with different roughness. This is because total number of aggregates attached is counted as the sum of all the aggregates irrespective of their sizes. Therefore, a correlation coefficient cannot be precisely determined between the roughness and the total number of aggregates based on the sum of aggregates of various sizes. Similarly, the correlation between the material roughness and the average size of aggregates cannot be accurately determined, as there is no fixed range of aggregate sizes that can be used to calculate the average size of the aggregates.

SF inhibited the attachment of biomass on almost all materials (Fig. 2 and 4). We hypothesize that lodging within the surface features (i.e., screw holes, edges, corners, etc.) of implants and host tissue may be an important factor in the establishment of PJI, not recognized by studies looking at the interaction of single planktonic cells with nanoscale smooth flat surfaces, as is conventionally done. Even though these aggregates attach less frequently, there could be two potential pathways leading to biofilm formation. The first would be these aggregates becoming larger after recruiting other small aggregates or single cells and eventually forming biofilms (Fig. 2, images for HA and PE) provided that they remain attached on implant surfaces. Second, if the aggregates are detached, they may colonize other surfaces and follow the first pathway to form further aggregates or biofilms. In terms of host factors, our lab has also shown substantial aggregation in purified fibrinogen and fibronectin (components of synovial fluid), so these data (not shown) further provide evidence that S. aureus interaction with host factors has an important role in aggregate and biofilm formation during infection. As aggregates are seen attaching to the implant biomaterial within 5 min, this indicates that the process of biofilm formation during PJI may occur rapidly.

It has been demonstrated that bacterial adhesion decreases with higher flow rates, and flow conditions are known to be an important parameter that influences bacterial attachment and detachment during the initial attachment stage (45). The use of the higher shear stress of 78 mPa in our study resulted in a decreased bacterial attachment to both HA and Ti with the exception of the 10- and 15-min attachment times, which showed increased attachment areas on HA. Our results agree with another study (26), which also resulted in a decreased number of adherent bacteria except for diamond-like carbon (DLC) material, with an increase in shear rate from 150 s⁻¹ to 1,500 s⁻¹. Another study also showed decreased bacterial adhesion for several materials, as the shear rate increased from 50 to 500 or 1,000 s⁻¹ (46). Therefore, higher shear rates cause higher detachment forces, which result in a lower number of attached bacteria (46). Similarly, in our study, applying higher shear stress increased the average size of aggregates on both Ti and HA. This could be potentially due to the fact that when a higher shear stress is applied for longer than 12 min, the attached aggregates break away from the substrate and start to move in the direction of the flow (47). Thus, they continue circulating in the system until they stick to the surface or another aggregate that is strongly adhered to the surface. This phenomenon potentially decreases the total number of individual aggregates and increases the average size. The sticking of one aggregate to another could be mediated by different bacterial or host factors. We believe that it is possible that fibronectin or other factors in SF could be attached to the outside of the aggregates, which further stick to other aggregates via bridging. However, to demonstrate this, future studies could utilize staining of either fibronectin or bacterial poly N-acetylglucosamine (PNAG). The stronger adherence of attached aggregates could be due to their possible ability to regulate strength in response to environmental factors (48).

The SEM images depict the morphology of the aggregates on different surfaces and how they are attached to it (Fig. 6). The Ti and 316L have the same levels of aggregate attachment to the surface. In contrast to Ti and 316L, PE and HA contained larger attached aggregates on their surfaces. This is most likely because of the roughness of the material. As seen on the blank coupon images in Fig. 6B, Ti and 316L have smooth textures. In contrast, HA has small fissures and pits on the surface, and PE has the roughest surface, with a leaf-like topography. The increase in shear stress further causes aggregates to get bigger, as seen in Fig. 6A, with Ti and HA surfaces being similar to that mentioned in a previous study (49). The SEM images show the attached aggregates and single cells at the 15-min attachment time. In all the materials, different aggregate sizes can be seen with larger aggregates on PE. Further experiments are needed to confirm this; however, we believe that these aggregates potentially act as precursors to biofilm formation if they remain attached to the surfaces for longer times. On the other hand, if they are detached, they can further colonize other surfaces, resulting in the formation of more aggregates or biofilm. Therefore, this study focused on initial attachment and recruitment of aggregates to each other and did not address growth. However, in future studies, we are interested in studying longer periods where growth is also expected to play a significant role in biofilm accumulation. Therefore, we are currently undertaking studies involving time periods beyond 15 min that include both growth and attachment.

Conclusions.

This study demonstrates that SF-induced aggregation resulted in less particle attachment and greater aggregate size than observed with single-cell populations under 15 mPa and 78 mPa of shear stress. However, on PE, the surface area coverage by the aggregates was higher, which could be due to the rougher surface of PE. Furthermore, increasing shear stress to 78 mPa resulted in a decrease in bacterial attachment on Ti and HA but increased relative particle size. Therefore, this study shows the inhibition of the attachment of biomass in SF-induced aggregation on all materials, suggesting the initiation of biofilm formation by lodging on the surface features of implants and host tissues. It further provides additional knowledge to the orthopedic field that can promote an understanding of the kinetics of bacterial adhesion and highlights the necessity to develop strategies to disrupt or inhibit the aggregates before they become biofilms.

MATERIALS AND METHODS

Bacterial strain and culture conditions.

Bacterial stocks were maintained at –80°C in 20% glycerol. A green fluorescent protein (GFP)-expressing S. aureus strain (AH1726) (50) was stored and cultured in tryptic soy broth (TSB). A single bacterial colony from the streaked agar plate was used to inoculate 25 ml of TSB for each experiment. Culture tubes were incubated at 37°C for 17 to 18 h at 200 rpm.

Bacterial aggregate formation.

Five milliliters of the overnight culture of S. aureus (10⁸ CFU/ml) was centrifuged at 21,000 × g for 1 min. The supernatant was removed, and the pellet was washed in phosphate-buffered saline (PBS) and then resuspended in 10% synovial fluid (500 μl) in 4.5 ml Ringer’s solution (RS). The cells were then incubated under static conditions for 1 h to allow aggregate formation at room temperature. For preparation of single-cell suspensions, 5 ml of the overnight culture was centrifuged at 21,000 × g for 1 min, and after removal of the supernatant, the pellet was washed and resuspended in 5 ml RS. Both aggregates and single-cell suspensions were diluted individually in 15 ml of RS and recirculated through a flow cell (schematic shown in Fig. 1).

Biosurface flow cell.

Attachment of S. aureus single cells and aggregates on 316L, HA, PE, and Ti coupons (10-mm diameter and 2-mm thickness; Biosurface Technologies) was observed in a flow cell (FC 270-AL; BioSurface Technologies, Bozeman, MT, USA). The ends of this flow cell were connected by silicone tubes and a recirculating loop fed from a mixing vessel was created (Fig. 1). The circulation was run for 15 min using a peristaltic pump (IPC ISM932A; Cole-Parmer, Vernon Hills, IL, USA), and time-lapse video was recorded at 30 frames per s (fps) using a digital camera (QIClick charge-coupled device [CCD] camera; Teledyne QImaging, Surrey, BC, Canada) attached to a Leica DM2700 M upright material microscope with a 20× objective.

Image acquisition, processing, and analysis.

Time-lapse videos were recorded at 30 fps using the Micro-Manager software (μManager v1.4; Vale lab, University of California, San Francisco), and the acquisition was realized with a digital camera. Ten frames from near the center of each coupon from three independent experiments were analyzed for attachment after 5, 10, and 15 min of circulation using NIH ImageJ (51). The average intensity of the GFP signal was quantified across all frames. A low average intensity indicated that the bacteria did not adhere and were present in the liquid phase, resulting in gray particles. Therefore, low-intensity particles were excluded, and high-intensity attached particles that appeared black were quantified to determine the number of particles, surface area coverage, and the average particles size of the bacteria present on the coupon via ImageJ. The number of particles attached is representative of the total number of particles distinctly separated from each other regardless of their sizes, surface area coverage is the total surface area covered by these particles, and the average size was determined by dividing the total surface area by the number of particles.

Coupon modification and characterization.

Coupons were sanded using an aluminum oxide sanding sheet (P600 grit; no. 436A38; Grainger, USA) for 4 to 5 min. The roughness of these coupons was measured using a Zeta 20 Optical Profilometer, and the Ra value was calculated. An Ra value is the mean height calculated over the entire measured area and is given by the arithmetic average of deviations from the mean. Ra is used for measuring the roughness of machined surfaces, which reflects general variations in overall profile height characteristics (26). The contact angle was measured manually by placing a water droplet on the surface and measuring the angle with an online protractor. Contact angles greater than 90° represent hydrophobic surfaces, and values less than 90° indicate hydrophilic surfaces. The Ra and water contact angle values in Table 1 are the means and standard deviations of three measurements.

Qualitative SEM assessment of aggregates and single cells on different materials.

For SEM images, coupons with single cells and aggregates were fixed according to a procedure described previously (11, 52) with some modifications. The chemicals used for SEM were purchased from Thermo Fisher Scientific (Norwalk, IL, USA). Coupons were placed in a 24-well plate and soaked in a prefixing agent containing 2.5% glutaraldehyde in 0.2 M cacodylate buffer (pH 7.4) for 24 h at room temperature. The coupons were then rinsed with cacodylate buffer three times for 5 to 10 min. After the final rinse, the coupons were dehydrated by placing them in increasing concentrations of ethanol (70%, 90%, and 100%) three times each for 5 min. Finally, the coupons were dried with 100% hexamethyldisilazane (HMDS) twice each for 10 min, coated with gold-palladium, and then viewed under an SEM (Quanta 200l; FEI, Hillsboro, OR, USA) at an accelerating voltage of 10 kV.

Statistical analysis.

All experiments in this study were carried out three times. All the parameters analyzed in this study, i.e., number of particles, average particle size, and surface area coverage, were statistically analyzed for the effects of the presence or absence of synovial fluid, attachment times, shear stresses, and the materials used via SPSS (version 26; IBM, New York, NY, USA). The threshold for significance was set at a P value of <0.05. Statistical significance was determined by one-way ANOVA followed by Tukey’s post hoc multiple-comparison test. All error bars in the figures indicate standard errors of the means.